It's not obvious to me that people with higher educational attainment deserve more power, influence, and status. So I don't know how this is a confounding Factor.
To.what?.Is this a propos of something else? In any case, there is a difference between saying someone morally deserves something ,and saying they are the best person to do it practically.
I'm not sure what comment I was replying to so I can't tell you why I said "confounding factor."
You have implied that, generally speaking, educational attainment to wield influence and power properly. This is quite a leap and I can't go there with you.
Where I am willing to go: Some positions of power and influence have SPECIFIC educational requirements. Other positions of power and influence have IQ requirements, irrespective of educational attainment. A modest number of positions of influence do not have requirements of either kind.
As far as social STATUS goes, the stakes are much lower. We will likely always have many high-status people with lower IQs and educational attainment. I'm thinking of various kinds of artists, celebrities and religious figures.
I 'm not saying that education grants you the ability to know right direction, but it does grant you the ability to implement a plan competently. the direction can be decided democratically.
I recently realized that uneducated and/or low IQ people fill an important role in our democracy: stretching the Overton window. They don't push it in one direction or the other, it's more like they stretch it to make it larger. This stretching benefits bad ideas as well as good ones, but we should be willing to pay that cost. It's better to err on the side of having more ideas in the marketplace of ideas.
Uneducated people are often have wide-ranging and diverse political views. A mix of opinions that left-coded and right-coded. This is because autodidacts gather political opinion somewhat randomly, picking up bits of ideology here and there.
Many eucated people reflexively call an idiosyncratic political program "incoherent." When in reality, the "coherent" policies of leading political parties are grouped together somewhat arbitrarily. For example, "socially conservative, economally left-liberal" is a tendency seen in many parts of the world. It's an accident of history that this tendency is completely foreign to modern American thought. (Tipper Gore and Republican-approved deficits notwithstanding)
Yes...up to a point. But populism also includes ideas that just don't work like increasing welfare and decreasing taxes. Some things are incoherent, not just unusual.
It seems like your reply may have two functions. It's a critique of valuing “Overton stretching” and also an assertion that “Sometimes when you call someone with idiosyncratic beliefs incoherent, that's just calling a spade a spade.”
I agree with the latter assertion.
As far as the critique of Overton stretching, it is non-obvious to me that this stretching will benefit populism in a disproportionate way, or in a way disproportionate enough to nullify the value of the stretching. After all, autodidacts and dum dums often hold a mix of populist and elitist ideas.
Suppose I were to suddenly understand that political eccentrics and dummies are TRULY incoherent in most cases. Even so, it does not necessarily follow that this incoherency would nullify the value of Overton stretching.
I guess I see “narrow Overton windows” as a pressing problem for America.
This is tricky. On one hand, I am abstractly in favor of the same dignity for all; smart and stupid alike. On the other hand, I would prefer that the smart make the important decisions. But the ability to make decisions = power, influence = status.
I spoke about "people with higher educational attainment." In your response, you flatly educational attainment with being smart. Hmm.
Also, there are different kinds of power than the power to make important decisions for others. Nor do power and influence necessarily equal status. Inasmuch as this is true, it is only true on a contingent basis.
This can be readily demonstrated by looking at the culture of the American radical left. Wherein leaders are deprecated to a ridiculous degree.
"This review was about a military topic. Due to its sensitive nature and a changing situation, the author has asked that it be removed and that he not be acknowledged in any way, sorry."
As a subscriber, don't you get emails with the entire text on the post? This was on Sep 26.
[EDIT - Here was an empathetic but superfluous and mistaken comment about Gaza]
I understand why Lemkin and the United Nations pushed for an extremely broad definition of genocide. I know they had good intentions. But ultimately, I think we should reserve "genocide" for any time-limited, substantial population reduction that was deliberate. We should probably insist that the reduction had to have happened in 15 years or less.
The reduction needs to have been executed through violence or directly coercive mechanisms. So forcibly sterilizing your people would count, but plying them with alcohol or outlawing their language WOULDN'T.
Once the United Nations signs on to this redefinition, some genocides might get recast as attempted genocides, quasi-genocides or near-genocides. Or maybe we'll use a handy new word, genocidistic. I know that recasting a long-passed genocide as quasi-genocide might upset some victims. However, we could ameliorate this by explaining why the downgrade won't slow their quest for sympathy/redress: Limiting our use of "genocide" gives the word greater power, which transitively gives "quasi-genocide" new weight. A self-correcting dilemma!
I know that defining a present crisis as a "genocide" mandates certain governments to take corrective action. But realistically, said governments TYPICALLY avoid that definition if it's not politically expedient.
The Armenian Holocaust and the Jewish Holocaust were unusually systematic and documented, while the Rwandan Genocide was unusually brazen. Few genocides are similarly indisputable, so governments ALREADY have the option of refusing to recognize your genocide--at least until it's too late for interdiction.
IOW, "if Uncle Sam hasn't recognized your genocide by now, they aren't going to. At least not while it matters." This is getting long-winded, and the benefits of tightening the United Nations definition are at least somewhat obvious. But I will list those benefits later if anyone asks.
I'll bite: what are the obvious benefits of narrowing the definition of genocide, and to whom would they accrue? I see benefits for the few groups who would also clear the higher bar – fewer people to share victimhood points with, and so improving their standing in the Oppression Olympics – and for people like yourself who are annoyed by the term becoming overbroad (not a particularly sympathetic constituency, you'll admit), but anyone else?
The argument that downgrading groups from "genocide" to "quasi-genocide" (or some similar term) wouldn't slow their quest for sympathy/redress seems patently absurd: it seems clear that they're getting more than (what you think) they're due is the very reason you wish to restrict the definition to clearly exclude them from claiming the label and its associated status, such as it is.
“I see benefits for the few groups who would also clear the higher bar…”
I think more than a “few” groups would clear the bar, actually. I think PLENTY of little-discussed genocides are waiting for their rightful day of prominence, even under my stricter definition. At least three genocides perpetrated by the British need to be widely recognized. One from the 20th century.
BTW, aren't you a confirmed right-winger? In that case, I shouldn't have to retread any arguments for reining in "genocide mania." You must know them all by heart, like everyone else who says "Oppression Olympics."
I sincerely DO believe that downgrading certain claimed genocides to quasi-genocides would have little or no harm on "downgradees." Their crucibles were so diffuse or contested or numerous or long-passed that humanity doesn't take them very seriously as genocides RIGHT NOW. Since these downgradees aren't getting much of ANYTHING, I never thought to worry they might be getting "more than their due."
But if that DID worry me, well, what's wrong with that? You frame it so grimly! Human attention is a zero sum game, and don't we WANT the most brutal crimes to receive priority remediation?
I didn't design my little proposal with intentions to materially harm ANYONE, promise! At most, it would ruffle a few pleathers. Also, my mention of Gaza represented a brain fart. I realize now: I don't know enough about the true Gaza death toll to say if they've been genocided under my new definition. Though it looks as though they've been genocided under the United Nations definition.
It's a flat fact that Epstein was the first person to successfully commit suicide in that prison in 21 years. This doesn't prove anything, but if it's a coincidence, it's such an incredible coincidence that it ALONE would justify any number of dark suspicions. There are a number of other incongrous and strange facts about Epstein and Maxwell. Nothing definitive, but enough to fully justify conspiratorial speculation.
It's all circumstantial evidence, which means it's hardly evidence at all. But why does Richard Hanania expect writers and podcasters and pundits to follow the same strict evidentiary standards as a district attorney? His scathing denialism would be consistent with a personality problem or something worse.
I guess I'm kind of issuing a challenge to any other denialists here. If you think the Epstein case is a big nothingburger, fine. But in light of the 21-year thing, how can you in good conscience belittle or mock the conspiracy theorists? Is it just a habit or reflex that is hard to break? If the conspiracy people are easily dismissed in all other cases, this is one instance in which the circumstances seem to demand lenience on your part.
I don't think anyone specifically tried to fake his suicide in an effort to make sure something he knew didn't get revealed in court. That would require multiple people involved in the conspiracy. Even if you think Trump is willing to order it, I question if everyone who would have had to carry out the order would keep quiet.
But I would not be surprised if, say, a prison guard said "yeah, I know we're supposed to be watching closely because he's on suicide watch, but fuck him, he's a pedophile." Malign neglect seems more likely than murder.
There's conspiracies and there's conspiracies. The Acosta deal was certainly fishy and it's likely there was a certain degree of a rich and connected person getting favorable treatment normal people wouldn't. We also know for a fact that Trump's people were trying to cover up embarrassing communications with Epstein since a lot of them have come out now.
But thinking that Epstein was murdered is another thing entirely. To believe that, you need to have the kind of epistimistics that make you bad at modeling the world.
Also, what's the base rate of prison suicide attempts? That's a number conspicuously lacking from your argument.
Of course it would be extreme to conclude from the weird circumstances that Epstein definitely got murdered or was allowed to kill himself. But there are enough weird details about Epstein's life and death to cast virtually EVERY aspect of his Official Story into question. Including the official line on his death.
Hanania is following the typical skeptics' playbook, by assuming the Official Story is more or less accurate, and putting the onus on Epstein conspiracy theorists to justify their skepticism, line by line.
Because of the strangeness of this scandal,, the standard skeptics'playbook DOESN'T WORK here. And charitably, I'd say most of those sticking to the playbook must not be well-informed about the public details of the scandal. But when it comes to someone like Hanania, who seems to have taken his time researching Epstein, such rigid skepticism can only be explained by naivety, credulousness or some other problematic motivation.
I think that after 21 years, the time and place of Epstein's death make it a black swan event that justifies grave epistemic uncertainty. Without need to consider suicide attempts. But to be congenial and thorough, I looked into that matter anyway. Based on the available data (which is not especially satisfactory), it is estimated that there were at least 20.suicide attempts per year at that facility in the decades leading up to Epstein's death. So yeah, it's an incredible black swan event.
Well the poster said “that prison”, which is a qualifier.
I think it’s pretty naive to think that well connected people get away with sex trafficking (which would have been the charges against Epstein, as they were against Maxwell, eventually).
Furthermore anybody who remains friends with Epstein after 2008 must have known about those charges and the plea bargain. The furore about Trump “knowing” about the girls seems naive or political, everybody knew about the girls.
"I think it’s pretty naive to think that well connected people get away with sex trafficking (which would have been the charges against Epstein, as they were against Maxwell, eventually)."
You wouldn’t bat an eyelash if someone suggested this was happening in Saudi Arabia, Albania, Russia, Mexico, or China.
But you don’t think this could happen in the United States? I’m not being deliberately obtuse; of course the United States has stronger rule of law, more empowered investigative journalists, and many other time-tested checks on our elites. I think these checks ensure that fewer elites can traffic fewer victims, and that said elites have to go to greater lengths to keep their crimes hidden.
But saying such trafficking is flatly IMPOSSIBLE? That claim speaks to an American exceptionalism of baffling intensity. And it reads like an implicit mockery of most other countries, though I doubt that implication was intended.
Even in the developed United Kingdom, BBC star Jimmy Savile presented an interesting case of elite impunity. Savile sexually abused at least 450 children. Shockingly, the villain died in 2011 without being brought to justice or seriously investigated. NHS investigator Dame Janet Smith later found that 100 BBC employees were told about Savile’s misconduct but did not take decisive action. Another 17 suspected misconduct but kept their mouths shut. These 117 people included five senior staff members.
Surviving victims and their advocates argue that Smith’s report, though sickening, was a “whitewash” that hid much deeper culpability on the part of BBC staff and Crown officials. The proven details of the scandal destroy so much faith in British institutions that we are almost left thinking that ANYTHING could be true about this case. Unlike in the Epstein case, this “whitewash” narrative is taken seriously by a broad array of mainstream British figures. From academics to elected officials to BBC employees themselves.
> Even in the developed United Kingdom, BBC star Jimmy Savile presented an interesting case of elite impunity. Savile sexually abused at least 450 children. Shockingly, the villain died in 2011 without being brought to justice or seriously investigated.
> During his lifetime, two police investigations considered reports about Savile, the earliest known being in 1958, but none had led to charges; the reports had each concluded that there was insufficient evidence for any charges to be brought related to sexual offences. Sporadic allegations of child abuse were made against him dating back to 1963, but these only became widely publicised after his death. His autobiography As it Happens (1974; reprinted as Love is an Uphill Thing, 1976) contains admissions of improper sexual conduct which appear to have passed unnoticed during his lifetime.
If I am reading it correctly, he was accused multiple times, investigated twice, even admitted it in his biography... but nothing happened.
> Former Sex Pistols and Public Image Ltd vocalist John Lydon [...] in an October 1978 interview [...] stated: "I'd like to kill Jimmy Savile; I think he's a hypocrite. I bet he's into all kinds of seediness that we all know about, but are not allowed to talk about. I know some rumours." He added: "I bet none of this will be allowed out." As predicted, the comment was edited out by the BBC prior to broadcasting
> In March 2008, Savile started legal proceedings against The Sun, which had linked him in several articles to child abuse at the Jersey children's home Haut de la Garenne. At first, he denied visiting Haut de la Garenne, but later admitted he had done so following the publication of a photograph showing him at the home surrounded by children.
> In 2012, Sir Roger Jones, a former BBC governor for Wales and chairman of BBC charity Children in Need, disclosed that more than a decade before Savile's death he had banned the "very strange" and "creepy" Savile from involvement in the charity. Former royal family press secretary Dickie Arbiter said Savile's behaviour had raised "concern and suspicion" when Savile acted as an informal marriage counsellor between Prince Charles and Princess Diana in the late 1980s, although no reports had been made. Arbiter added that during his regular visits to Charles's office at St James's Palace, Savile would "do the rounds of the young ladies taking their hands and rubbing his lips all the way up their arms."
Sounds like many people knew, including some high-status people, but no one wanted to risk a lawsuit.
It could be that suicides in prisons are generally very rare. In that case, for almost every prison suicide we look at, we'd find that that it was the fist suicide in that prison in a long time.
A suicide occurred at that prison in 1998. Afterwards, at least 420 attempts failed before the next successful suicide, Epstein's. That's WEIRD.
Please note I never crosshsaired ALL Epstein skeptics. I criticized Hanania, along with all whom I would deem "aggressive Epstein denialists."
Hanania has mockingly conflated all conspiracy nuts with all Epstein "truthers." At least in terms of their intellectual seriousness. But this is an unusually diverse and MASSIVE cohort of truthers. This massiveness is only PARTIALLY explained by referencing algorithmic information bubbles, sinister apps or other NYT-approved theories. I think the massiveness is largely explained by the simple fact that the Official Story is IMPLAUSIBLE.
Excuse a brief sidebar on definitions. Most Americans primarily use a colloquial definition of conspiracy theorist (CT): "A dysfunctional nut who is notably gullible and misinformed, whose dysfunction manifests itself in paranoid monologues about vast conspiracies, usually comprised of global elites." More often than not, these paranoids readily became Epstein truthers. However, Epstein's death forced many "normals" to become "CTs in name only." These functional non-nuts, not given to paranoia, were mugged by reality into suspecting that a clique of powerful sleazeballs networked to shape the life and death of Jeffrey Epstein. These Epstein truthers are CTs according to the LITERAL, not colloquial definition of CT.
Has anyone commented on the alcohol use vs dementia study published in BMJ in September and very widely cited in the press as demonstrating more or less conclusively that ANY amount of alcohol consumption increases the risk of dementia? The coverage asserts unequivocally that there is no safe "floor" amount, and that the oft-mentioned "glass of red wine a day" is dangerous, not beneficial or even neutral.
I just read the study, and, although the authors make similar claims, their study seems to me to be something else altogether. It seems to me that what they did is associate certain genetic traits to alcohol misuse with increased dementia. The part of the study involving actual people showed the "U Curve" that we've heard about for twenty years.
I won't go on, as I might be misreading the whole thing; but it was a big surprise to read the study after reading all of the coverage of it.. Has anyone discussed this here?
I’m interested in this too. Went to BMJ article, thinking Oh, I bet I can figure it out, but soon discovered that doing that was going to take an hour at least, and bailed because I’m busy. So I’m in the same boat as you and John Schilling. But I do have a suggestion about how to get clear about whattup in that article. Ask GPT. I have the $20/mo version, and for that I get several GPT5 options including the “Think” version. I have had a lot of success regarding vexing things like this article with having it not just summarize the article, but explain things to me. What I’ve found to work best is to be very detailed in my prompt. For example, I might tell it to explain the logic of each stat it mentions. (“A Snarg test was used to mathematically compensate for the 2 main confounds, A & B by raising or lowering each subject’s score by the amount the 2 confounds would be expected to change it.”) So for this article I would start by stating the thing that confuses me: How can they get the usual U-shaped distribution, but also say that the more you drink the greater your chance of dementia? Then I’d ask for succinct of results and their interpretation, with a brief conceptual rather than mathematical explanation of how they arrived there, an explanation of how they got the odd double result, and a judgment of how valid their interpretation is.
If you do it, hope you’ll share what you found out.
I have not read the study, but have read commentary about it in several places. All of whom described something mathematically nonsensical when it comes to the one question I actually care about: what is the absolute increase in dementia risk going from zero drinks per day to one drink per day?
I don't know whether that's because the study was mathematically nonsensical, or because the study looked correctly but only at higher drinking rates, or because all the commenters are the same sort of idiot. For that I'd have to read the paper, which would be tedious and I've been busy.
I enjoyed it as a teen in the early 80s - the thematic connections he drew between Bach’s music, Escher’s art, and Goedel’s mathematical logic were fascinating to me at the time. The whimsical dialogues between Achilles and the Tortoise were appealing in an Alice in Wonderland way. In retrospect, I think the book did a great job of popularizing some profound ideas (to the extent such things can be popularized) while not necessarily bringing all that much profundity of its own.
Read it maybe forty years ago, found it to be enjoyable and entertaining and insightful. But I haven't read it since, and I suspect that I have since seen almost everything in it, done better.
I don't recall it being the sort of book where you struggle through fifteen chapters and then "aha, that's what he's getting at!". I was always struggling a little bit but also always getting something out of it. So if you read a few chapters and find yourself saying "meh, this is nothing special", you probably won't lose anything by bailing on it.
In fact, I think 15-ish chapters as where he lost some people, as a lot of computer/logic people faltered at the long chapter on DNA, about 2/3 through.
Also, the stylistic/meta cleverness was amazing in its time but I think would seem a bit dated now.
I tried reading it but found it frustrating and boring enough that I quickly gave up. And I'm a math-loving nerd who grew up on Martin Gardner, so I'd be pretty much the best possible target audience for the book.
For people in my world (people interested in logic, philosophy, minds, and machines, particularly ones born in the 1980s) that book was incredibly influential. It’s a beautiful book, though difficult in some ways.
I tried long ago to read it, and bailed. My memory of it is that the conceptual shape it was talking about was deliciously strangely recursive, but that yards and yards of clunky prose were expended on saying in many many ways, "look at this for god's sake! The stuff Godel was playing with was the same shape as the stuff Escher was playing with. The stuff Esher was playing with he same shape as the stuf Bach . . .. It's recursive! See, here's some Godel resursiveness, and here's some . . ." But maybe I was just tired and cranky that day.
My guess: a sign they've overplayed their hand against Trump. The conflation of Trump having sex with a willing sixteen/seventeen-year-old girl (probably true, but no one cares) with him kidnapping and raping babies (rhetorically insinuable, and tricky to clarify, as Megyn Kelly just found out) is politically useful, but including claims of him fellating Bill Clinton (obviously preposterous) makes it easy to dismiss the whole thing as a hitjob.
Who is the "they" that's overplayed their hand? I've seen people on Bsky and Tumblr cracking jokes about "outing Trump as bisexual," but I don't think anyone with actual power in the Democratic party is unironically claiming he sucked off Bill Clinton.
Like, I don't think you can say "they" made a strategic blunder when "they" don't have any control of what the Internet latches onto.
I didn't buy the "It's just college students and a few people on Twitter" schtick a decade ago when it was about Social Justice and Political Correctness, and I'm not buying it now. Though decentralized, this faction has plenty of power and influence over the Democratic Party.
But this rumor is so obviously dumbly untrue it’s hard to believe that even the most cynical and self-righteous members of “them” would try to get mileage out of it. The person who first told me about it was in the college kids and Twitter demographic. I laughed, but said “there’s no way that can be true,” and he replied that yeah, he knew that, but it was just so damn funny. Btw, he’d heard a variant: Trump and Clinton did the deed with Putin watching, cuz Putin kink or something.
Also Voltaire was there watching the whole crew, and he said “I may disapprove of every single object of you desire, but I will defend to my death your right to suck it.”
Are you similarly accepting of people who call the Republican party Nazis because of Nick Fuentes and his ilk? Or is it only the Democrats who are expected to exercise perfect message discipline over every single member with a social media account?
(And like, Fuentes is at least someone with name recognition! The people I've seen making jokes about Bubba are much less known than he is.)
I agree there's no way he did Clinton, or even Bannon, even given how gorgeously hot Bannon is. The guy's straight. But there actually is a text, from Epstein's brother to Epstein, I think, asking whether people knew about Trump "blowing bubba."
Peptide therapy seems to be taking off, which is an exciting development, particularly with RFK Jr. and the MAHA guys on board with it as "alternative medicine."
I'm trying to guess what kind of gent would be the type who didn't care much for Trump OR Epstein's victims. That's a tough one! Maybe a SUPER-libertarian NeverTrumper?
That's me. Many, and quite possibly all, of Epstein's "victims" willingly prostituted themselves for money. You don't have to be a libertarian, just the kind of normal, high-class American who traditionally had contempt for both prostitutes and sleazy used car salesmen like Trump.
This seems likely. Epstein was a pimp. A bunch of his prostitutes were underage, which didn't bother him just as it doesn't bother other pimps. You think anyone is checking IDs and birthdays on the girls working a truck stop parking lot? Why would it be different for the pimp to the rich and famous?
I expect there was a mix of kindness and menace, gifts and threats, to keep them in line, but again, this is apparently just what pimps do. The fact that he was a rich high-class pimp probably meant that the prostitutes working for him had a higher standard of living than most prostitutes. But still, it's not like he was looking out for their welfare or cared about them as humans or was above coercion or mistreatment to get what he wanted from them.
Don't the feds have a history of finding drug dealers to rat out other drug dealers? While giving the rats carte blanche to keep dealing their little hearts out? You'd be one helluva lucky rat if you could finagle a BEAUTIFUL deal like that! 🐺🐻🐼🦝🫣🐺🦊🦝🐨🐻❄️
This whole idea that sex with an underage prostitute = life in prison is new. Back when the "sweetheart" deal was made, such a deal was entirely uncontroversial. Indeed, such deals are still being made today. The NYTimes recently published an article about "sex trafficking" detailing how a "sex trafficker" named "Double R" was punished with "180 days in county jail and probation, plus an anger-management program."
Leftists focus on Epstein because he's an example of the "powerful white men"* who they say are the ones doing the sex trafficking, unlike the aforementioned Double R. They don't have any other examples, which is why they're still talking about him six years after his dead, and will likely still be talking about him sixty more.
* They get really mad if you point out he was Jewish.
A pivotal battle from European history that is too little known: the Battle of Muret (1213). . This battle took place in the final stage of the Crusade against the Cathars.
Disproportionately women and children, the Cathars had pacifist beliefs, and did not participate much in the fighting. Instead, they were defended by Catholic neighbors who took the injunction "love thy neighbor" seriously. Of course, the Crusade also represented a threat to the relative political autonomy of Southern French lords. But the ideological component cannot be ignored. The Pro-Cathar knights were rallied on by Spiritual Franciscans, ascetic monks who refused to be silent about the corruption and moneylust of the Popes.
I'll always be fighting the Battle of Muret in my heart. When Northern France sealed absolute control over the South, the Papacy sealed absolute control over Catholicism, ignorance sealed absolute control over piety, and hatred sealed absolute control over charity.
It is likely the case that there were no actual Cathars in the Albigensian crusades and that they were mostly pre-Gregorian Reform Catholics with non-consistent beliefs. The label “Cathar” is a later, polemical construction rather than a self-conscious heretical movement. It would be a mistake to imagine the listed beliefs of Cathars (i.e. quasi-dualist and pseudo-Gnostic) as being broadly followed in any discrete/structured way.
That being said, the Gregorian reforms are very important the Albigensian Crusades definitely cemented that in medieval Europe and is thus not at all a minor event.
Source: Mark Gregory Pegg's A Most Holy War: The Albigensian Crusade and the Battle for Christendom
Inasmuch as the Cathar label was a handy label for justifying the persecuting non-standard Catholics, i suppose it must also be a handy label for commemorating non-standard Catholics who were persecuted and martyred.
And it is also the case that the lords and knights were involved in resisting The Crusade were by and large willing to conform to Gregorian “orthodoxy,” but were unwilling to enforce that “orthodoxy” on their subjects.
Not at all! Wikipedia calls it "one of the most notable pitched battles of the Middle Ages." Historian Martín Alvira-Cabrer wrote that “Muret…became one of those defeats that modified the course of the Middle Ages.” If the Southern Coalition had won, there's a strong possibility that Languedoc would not be part of France today.
This battle sealed the outcome of the longest single crusade in history. It was the first and largest crusade initiated to punish heretics or Catholic enemies of the Pope. And if this crusade had failed, it might have fundamentally changed the balance of power between the Papacy and local Catholic communities.
I could see a Southern victory speeding up OR slowing down the onset of the Reformation. Or the victory could have altered the nature, geography and intensity level of the Reformation.
If this battle was so important, why don't we hear about it more often? I'm not entirely sure, but this is of a piece with the way Southern Europe has GENERALLY been neglected by English-language historians. Reading 20th century historians, you'd think that, apart from Constantinople and the papacy, little of interest happened in Southern Europe between 750-1350.
This situation started improving in the 1980s, but we still have a long way to go.
I suppose that I might now be responding earnestly to a joke. In which case, good job, you got me! But that's OK. I'll grab it any opportunity to interest the lurkers in this weird, fascinating crusade.
> If the Southern Coalition had won, there's a strong possibility that Languedoc would not be part of France today.
There is also a possibility that the consolidation of State authority under the kings of France was overdetermined. Also the crusaders were autonomous knights, not the royal Host. Their defeat would not have much weakened royal power or prestige (a defeat at Bouvines the following year might have been another story!) and at some point the kings would have wanted to assert their authority over the whole kingdom.
Maybe a victory of Peter of Aragon could have been a step toward another multi-regional rival "Empire" within France like that of the Plantagenêt kings of England/dukes of Aquitaine. But even they did not split a region from France in the end, despite winning battles much more devastative to the kingdom than Muret could have been.
"Had it [the Battle of Muret] failed, the campaign would have failed…Bouvines would probably never have been fought and the chances are that the French monarchy itself would have collapsed." Thus spake Hilaire Belloc. This is a tantalizing idea, although it must be said that Belloc was an autodidact and popular historian, rather than a fully qualified academic historian.
There was a later medieval heresy which survived the catholic crusades against it: the Hussites of Czechia. They did not end up very influential by themselves, compared to the Lutheran and other reformations of the 1500s. Although maybe there is a case to be made that the Hussites made the Reformation possible by showing that breaking off the Church was an option.
That's a data point on what a survival of the Cathars could have looked like.
Sure, everything you've written here sounds true. One difference is that in the case of the six Hussite Crusades, the Crusaders were mainly fighting directly against the Hussites. In the Albigansian Crusade, the Crusaders were mainly fighting against devout Catholics who raised arms against the Army of the Pope, so to speak. (At least the pro-Cathars were devout in MY book)
That's why there was a sense in which the crusade was a struggle over who QUALIFIED as a faithful Catholic. Whereas the Hussites and the Lutherans proved all too willing to shed their Catholic identity.
I didn't call the Cathars Catholic. The Cathars were barely involved in the fighting, because they were pacifistic. The fighting primarily took place between the Crusaders and Catholic lords protecting their Cathar neighbors.
Furthermore, the pro-Cathar lords and knights (including one King of Aragon) were fully devout Catholics who saw it as their Christian duty to protect their lands from a false, wicked Pope (as they saw it). Therefore, the Crusade can be seen as a civil war between Catholics with two different standards for what it meant to be devout. The pro-Cathars were also protecting the Jews in their community. They also knew that the Crusaders were implementing a system that would strip all lands away from female owners.
Of course, that's assuming that the Cathars even existed. One strain of historiography maintains that the so-called Cathars were a mix of Catholics following pre-Gregorian local traditions. Or the 'heretics' may have been a mix of Cathars and pre-Gregorian Catholics. Our knowledge of Catharism come solely from highly unreliable accounts written retrospect by fanatical Gregorian priests. (In this comment, Gregorian doesn't mean monks of the Gregorian Order, it means adherents of the Gregorian reforms imposed by the Papacy).
Whatever the exact identity in nature of the 'heretics,' the dynamic of the fighting remains the same. The anti-Crusader fighters were dominated by orthodox Catholic lords protecting those deemed by the Papacy to be heretics. By implication, they were fighting for a kind of religious tolerance, although I have no idea if they would have framed it in those terms.
I'll steelman that rhetorical flourish by proposing that Simon Montfort and his accomplices were ignorant about the Cathars, ignorant about the extent of the Pope's moral authority, and ignorant about the the ideological content of the New Testament.
For this alone all the YouTube nerds should declare Trump the greatest president frankly. Killing the penny was a nerdy cause for YouTubers for decades. Obama said we couldn't do it. Biden did nothing. Second term, one post on Truth Social, one executive order or something and boom penny gone.
It’s absolutely a good thing, but it’s been pointed out that it’s a smaller budget item even than the things DOGE was cutting. It’s a lot of virtue signaling about cost savings while doing things more expensively. And doing it without a plan for the transition.
Still, I’m glad someone finally ripped the bandaid off if no one was able to phase it out in the proper way.
And another, now calling for the elimination of the nickel (and the dime), https://www.youtube.com/watch?v=58SrtQNt4YE, for mostly the same reasons it made sense for the penny.
It costs more to make nickels than it does to make pennies. So why not scrap the nickel and say all pennies are now worth five cents? The government saves money on new coins, and every one with a penny jar suddenly has something worth spending.
No democracy can survive once people discover they can vote themselves largesse from the public till. They will inevitably find the politician who will give them the most money, until the entire fiat currency fails, leading to a dictatorship.
... are you still sure the last penny isn't ominous?
Proudly radical commenters will always be with us, but feature or bug, this forum is a bastion of political moderates. Who are all too eager to point out that "America is already great," at least in terms of long-term wealth creation.
But the fruits of that wealth have disproportionately gone to college graduates. Non-college-graduates are justified in bemoaning their massive loss in socioeconomic status since 1970.
Tthere's still a taboo against talking about this, enforced by college graduates in leadership on right and left. You can critique COLLEGES, but you're still going up to be hammered for any perceived slight against college GRADUATES, as a whole. Leftists will call you anti-intellectual (as if you have to have a degree to be an intellectual), and rightists will dub you a lazy, envious loser blaming society for your personal failures.
So foreigners and immigrants become the scapegoats.
There's no effective response to toxic populism that doesn't recognize non-college-graduates as a group with unique needs and legitimate grievances.
This is an American-centric comment, but this must all apply to various Western nations to one degree or another. In
Huh, I feel like this has been a point hammered by Richard Reeves (The Dream Hoarders) and others on the center left for well over a decade. But unfortunately, the upper middle class control a majority of political donations and a near majority of votes.
I have no doubt we are seeing progress. I would just like to see more progress faster, I guess.
Speaking of the upper middle class, it would be a mistake to ignore the way they endanger the social stability of the nation. Yes, billionaires have outsized influence on society, but as Scott has pointed out, they have heretofore punched far below their weight when it comes to putting money into politics. Furthermore, isn't it true that billionaire super PACs affect elections in an unpredictable way that's hard to track? It's hard to summarize the SERIOUSNESS of the problem.
On the other hand, the upper middle class is very numerous in modern America, and relatively politically active. We can more easily tease out the way their votes and donations warp the fabric of our democracy. So maybe we should pay more attention to this clear and present danger that is nakedly menacing our society.
Don't worry, I'd never call for guillotines. I find violent rhetoric rather distasteful. I can't speak for my fellow poors, though! You might want to negotiate with them to keep things hunky dory.
Post-COVID, the bottom 25% of incomes saw their income rise faster than other quartiles, and yet Biden still got hammered for "not caring about the working class." While wealth inequality is a big problem, it doesn't seem to directly correlate with the political shifts we see.
Also, the *actual* wealth inequality problem, IMO, is billionaires vs everyone else, not college-educated vs not college-educated.
Wealth inequality is not the same as incomes. We all know that rents rose pretty dramatically "post-COVID" -- the bottom 25% got squeezed hard, due to the increased risk premium imposed on landlords. Can you really show that they increased in wealth at all, or did they in fact lose wealth? (I'm pretty sure the latter, with unpaid credit card bills skyrocketting).
Also, I would expect stuff like this to be captured in real wages - if housing prices are up enough that your overall wealth has gone down, then inflation-adjusted wages must go down as well, right?
Also also, OP specifically blamed wealth inequality, not inflation. "the fruits of that wealth have disproportionately gone to college graduates." You're making an entirely different argument.
>Non-college-graduates are justified in bemoaning their massive loss in socioeconomic status since 1970.
Are they? Everyone is richer now than they were in 1970. The only "loss" I can see is that those with college degrees tend to be more richer now than those without, but they're all richer. If someone gets a 55% raise, are they justified in moaning that someone else got a 78% raise?
Your analogy about raises is not fully satisfying, because it does not appreciate the "zero summish" nature of the phenomenon I described.
This post-1970 phenomenon was not exactly a zero sum game. But it's in the same ballpark. Decisions were made to use the power of government to reorient our state to benefit equity investors, retailers, homeowners and the knowledge economy. (Questions of intent are messy, this wasn't necessarily an Evil Conspiracy)
These decisions empowered certain citizens at the expense of others. Either directly or through a kind of "opportunity cost." When the government goes out of its way to benefit group A, they are implicitly WITHHOLDING help from group B. Uncle Sam doesn't have infinite resources, so picking winners and losers is not inherently wrong. But the gradual scouring of non-graduates WAS wrong policy, whatever motivations inspired it.
You'll notice that in my first comment I mentioned how non-graduates had lost "status." I didn't specifically mention income inequality, although that is related issue. Key point: Social status really IS a zero sum game, since your status is determined by where you "rank" in the pecking order, relative to everyone else.
At least part of what's going on is that "college graduates" and "non-college graduates" are different populations now than they were 50 years ago. In the early 1970s, less than 10% of adults were college graduates. Now, about 35% are. So when we say that inflation-adjusted wages have stagnated for non-college graduates since the 70s, we're comparing wages for the lower 90% of the educational attainment distribution then to wages for the lower 65% of the distribution today.
If they do, they're TERRIBLE at being leftists. The whole notion that access to college degrees--which are far more easily available to those born with more wealth and privilege--should control the material circumstances of one's life is pretty explicitly antithetical to any serious notion of leftism. Liberals might hold to such, since liberalism tends to include an implicit belief in the fairness and meritocratic nature of the so-called "free-market." But liberals aren't leftists.
I will caveat, however, that there are plenty of *tasks* which require domain-specific knowledge and expertise to perform well. In many cases colleges are (at present) by far the most direct and practical way to obtain that knowledge and expertise. Realistic leftists (which plenty aren't) will acknowledge this but argue some combination of A. access to that knowledge and expertise should be made more broadly available and B. one's socioeconomic status should not depend so heavily on obtaining such expertise.
The fact that there's been a huge wave of people recently loudly insisting that expertise simply isn't necessary, even in areas where it plainly, obvious is, tends to muddy the water somewhat. These people ARE being anti-intellectual, and their ubiquity makes it easier to conflate "no really, you need to be trained to do that" with "only people with college degrees deserve respect or a comfortable life."
"This sounds like an article of faith, rather than a seriously supported assertion."
Um, what? An article of faith? Excuse me, what?
I didn't say *unavailable.* I said *more easily* available. I honestly have no idea how you could imagine this NOT to be true. Do you think that your odds of getting a degree anti-correlate with your parent's income?
Note that I said "more wealth and privilege" not "super, duper lots of wealth and privilege." I'm well aware that getting a college degree is the default for a large swathe of the American middle class--but the middle class aren't the only people in existence. And even among people considered middle class, wealth buys lots ways to smooth the road: you're more likely to get help with any health or academic problems, less likely to need to work at the same time as you go to school, more likely to be able to afford conveniences and amenities, less likely to need to drop out in the event of a family emergency.
"If you're going to say "well, but not everyone graduates high school!" I'm going to reply, "sure, but high school's free.""
And at this point I'm going to ask exactly how many minutes you've been living in the real world, and if we could maybe table this discussion until you've been here for a year or two. If you think the cost of high school tuition is the *only* variable that makes wealth more likely to lead to high school graduation, I can only imagine that not only have you never, ever experienced financial hardship, but that you've never met, spoken to, read anything written by or possibly even passed on the street somebody who has.
I grew up in a fairly comfortable, middle-class household, but it was still really, *really* not that hard to notice the ways in which wealth directly and indirectly translated to academic success. Many were small ways, but there were a lot of them. Few of them are extremely likely to be definitive by themselves, but in case you didn't notice, we were talking about population-level trends, and the accumulation of small effects are what population-level trends are made of.
So, let's break apart "wealth", which is your word, not mine. You're not using income, you're using wealth. The primary source of wealth for middle class people is housing, right? [This is the difference between positive wealth and negative wealth for most middle class people...] So, you're saying that apartment dwellers (in, say, NYC, or Pittsburgh, or Atlanta) are significantly less likely to graduate college. I don't think "parental wealth" is the best way to distinguish "whether or not someone finishes college."
I'm here to tell you that at age 25, you no longer need to factor in your parents wealth/income in applying for grants. That means you're basically "defacto poor" (unless you're, say, a multimillioniare) -- because otherwise you're established and making "good money" and don't actually need college. This makes you eligible for a free ride to most normal schools.
Do people sometimes drop out of college to support their family? Sure. It's pretty rare, though, in the grand scheme of things (this happened a lot in the 1930's, when government benefits weren't very thick on the ground). And even for people who did drop out of college, if it was a temporary thing, they can go back to college.
High levels of parental wealth can let you coast through basically everything -- parents have this absurd tendency to "help" their kids, even when they're failures. See Mitt Romney, see Hunter Biden.
I think, in high school, you have this wide swath of "people being given pretty much the same start" and then you have this "swathe" of "kids who will be tutored into passing and attending Yale" (this is extremely wealthy people).
Cheery reminder that wealthy people can be under the federal poverty level in terms of income (see Senator Tester, before he went to Congress). And that those people are presumably significantly less likely to finish high school, despite having parents that are millionaires.
I'm willing to bet you could find a better proxy than "parental wealth" or "parental income" to show which people are most likely to go to college.
I do know kids that were sent to stay with relatives because their parents couldn't support them. They finished college. I've also tutored at a community college -- lots and lots and lots of people who are "not being supported by their parents" (and presumably eligible for "no money down" college).
You can assume we met in the middle, on this point, and that yes, you're not talking to a deranged lunatic, merely someone who thought you were overstating your point.
There are some pretty significant differences in college attendence and completion between people with wealthier/poorer parents: https://share.google/EtTGS6LjqnuMSSTyN
I'm not sure how this is relevant? I don't want to be uncharitable but is this just a way of saying "if you don't want to go to college and complete a degree, that's on you. Sounds like a skill issue." I hope not. That kind of framing is a crude and reductive sidestepping of the issue.
Hm, maybe something is unclear here. Agra[...] said college degrees are easier to get if you're born into a wealthier family. November Dawn disagreed because most high school grads go to college. I'm pointing out that there's still a pretty obvious gap based on parental income, and that holds even more true for top colleges, i.e. I agree with Agra.
If you look beyond the US a lot of the original comment still applies despite university degrees being much more widely available to those without wealth and privilege. I think its a useful consideration even if you are only focused on the US.
Making the sorting mechanism more meritocratic doesn't solve most of the consequences of the sorting.
I'm not sure how this is relevant. I mean, I have an idea of what you are trying to say but Idon't want to be uncharitable by assuming callousness on your part. Can you rephrase your point?
Callous? I'm pointing out that segregating society has downsides regardless of how it's segregated. I'd rather suffer whatever uncharitable looks like than be damned by implication.
As I said, belief in the meritocratic nature of the market is a liberal trait. Substantially privileging *any* select group of people over other groups is fundamentally anti-egalitarian, even if you swear up and down that your selection mechanism is "fair[1]." To the extend that the left-right distinction are used consistently at all, left is the egalitarian direction, so in any meaningful leftist practice, anti-egalitarian practices are to be questioned and (as far as possible) removed or reformed.
Just to be thorough, I'll add the same caveat as before: expertise is still necessary for lots of things, and pretending that it isn't never ends well. But choosing how you regard and reward expertise is a voluntary choice, at least inasmuch as anything that happens at the level of societies is voluntary.
Finally, I'll add that a good (fictional) template to help think about such things is the novel "The Dispossessed" by Ursala LeGuin, as the main character is a man of significant specialization and expertise living in an actively (and in some ways militantly) egalitarian society.
[1] A word that becomes largely meaningless in context like this.
Japan and America have nearly the same statistic for "high school students going to college" -- what countries are you thinking of? And can you cite statistics that are greater than 60% conversion rate?
It appears I have mistakenly replied to your comment rather than the level above, agrajagagain's.
We're likely in meta agreement (because the symptoms are there when you remove this) however to your point my reply would be - any country where the cost of the degree is significantly lower.
If a degree is very expensive you might expect the 37.2% who don't have a degree to correlate strongly with the poorest 37.2%. In a country where it's closer to free (or the debt on what is charged is more heavily subsidised), you'd expect less correlation - away from whatever correlation of wealth and IQ you're willing to believe in.
Something that is exclusive to the richest 60% obviously isn't particularly exclusive but could still have far reaching consequences.
"But the fruits of that wealth have disproportionately gone to college graduates. Non-college-graduates are justified in bemoaning their massive loss in socioeconomic status since 1970."
At least the number of college graduates has increased massively, hasn't it?
I generally agree with you, and wanted to comment on:
>There's no effective response to toxic populism that doesn't recognize non-college-graduates as a group with unique needs and legitimate grievances.
If by "effective response to toxic populism" you mean the Democrats (or maybe even some more reasonable coalition of Republicans) winning the 2026/2028 elections, then I think the best strategy would focus on winning winnable votes. "Recogniz[ing] non-college-graduates as a group with unique needs and legitimate grievances." is a part of the story, but another huge part is how to handle polarization and social media, choosing messaging that resonates, avoiding getting overly invested in unpopular issues like trans participation in high school sports, running more conservative (possibly pro-life for example) candidates in more conservative states. I don't think non-college educated people are voting for MAGA because of a thoughtful consideration of policy... They vote for someone who talks like them, who they want to drink a beer with, they vote for the leader of the tribe the algorithms have polarized them in to, also they vote based on their own personal experience of the econonmy in the last 3 months, ie: the price of gas and milk (without regard to any historical factors).
As a non-college educated conservative person of sorts, I think "oh those guys are morons, let's just run someone who can say "ain't" and knows the name of at least three guns" is probably a great way to win over conservatives; people love pandering, we are indeed dumb enough to fall for that all day and this definitely won't backfire.
I guess I didn't write my comment in a very thoughtful way. I generally think that most of the electorate, on both sides mainly vote with the tribe.
But to play the devils advocate, Isn't Trump the ultimate example of successful pandering?
He has previously been: anti-gun, pro universal-healthcare, pro-choice, pro Clinton, he's been pro and anti social security, anti crypto.
And moreover he's a coastal elite billionaire who managed to convince 40% of the country that he's their bro, and that he's going to "drain the swamp".
He also, as another thread here mentions, seems to hand out pardons to anyone willing to invest in his families crypto venture, or other fundraisers.
Moreover, most of his speeches seem to be 90% pandering, "we're going to build a wall and Mexico is going to pay for it" or talking about how cheap various household goods will be.
Maybe he's not pandering but actually evolving or just able to change into what he needs to be somehow as needed in a moment.
But what i would advocate for is not running candidates who can play-pretend, but rather, run more conservative candidates in states that are more conservative, and more progressive candidates in states that are more progressive. Don't fund primary challenges with democratic socialist candidates from the left, in states where the populace wants a pro Oil candidate who is in the middle on abortion. Try to build coalitions that include people that disagree on gender/climate/etc. but possibly agree elsewhere, like Healthcare and such.
The Danes are happy because crime and disorder are spread out fairly evenly throughout the country. I would imagine this is partly by design and partly by circumstance. It can be politically expedient to allow crime to fester in certain areas and not others, but that causes a bunch of social pathologies that feed on each other.
Many POTUSes have abused the pardon power, and it was high on my list of things needing constitutional deletion well before Trump arrived. But as in some other forms of corruption he is taking it to a whole new level.
Just since returning to office 10 months ago Trump has issued more than 1,500 pardons. More than 1,000 of those are January-6th rioters. Others include:
-- the husband of a GOP congresscritter, who'd been convicted of health-care fraud including defrauding Medicare and Medicaid.
-- former congresscritter George Santos, who became the 8th criminally-convicted Republican member of Congress pardoned by Trump.
-- former Illinois governor Rod Blagojevich, which as an Illinois resident fills me with a perhaps-unreasonable degree of rage. [Could be that a Long Island voter would feel this way about the Santos pardon.]
-- a guy convicted of defrauding investors, who was pardoned after donating almost $1 million to the Trump campaign.
-- a guy convicted of tax crimes, who was pardoned one month after his mother attended a $1 million-per-seat Trump fundraiser.
-- a Chinese billionaire who was pardoned after the crypto exchange he founded, Binance, struck a deal with the Trump family company and also hired as a lobbyist a close friend of Donald Trump Jr.
-- two DC police officers who were convicted of murder and subsequent coverup.
-- two guys convicted of defrauding a Native American tribe and pension fund investors out of tens of millions of dollars, who were pardoned after they testified in the House Republican investigation of Hunter Biden.
-- a Republican state legislative leader and aide who were convicted of bribery, and a Republican state senator who had pled guilty to siphoning off campaign donations to his personal use.
-- 77 pre-emptive pardons for individuals investigated and/or charged, but not yet tried, in the 2020 effort to substitute made-up lists of Trump electoral-college electors in several states won by Biden.
All of those, and more, just since January 21st 2025....it's literally a dial-for-pardons operation up there at this point. No fucks given.
Friend of mine worked for the doj. Prosecuted a guy in Miami for tax fraud. Millions and millions of dollars. Money that could and should've gone to public benefit. Trump pardoned the guy because they were acquaintances. My friend no longer works for the doj.
Historically, I think it's very good for Presidents and Governors to pardon people who have been on the wrong end of a gross miscarriage of justice, including but not limited to being wrongly convicted of crimes they didn't commit - and no, that's not something you can just leave to the courts. I think it's less good, but still good, for them to pardon people who have clearly rehabilitated themselves to an exceptional degree. And it is frequently a good idea to offer a general amnesty at the *end* of some major social upheaval.
And I think that this represents a large majority of the pardons in US history.
I'm OK with an implicit requirement that you not convict people of crimes if you can't reasonably explain to the chief executive why that's a reasonable thing to do. If there's a disagreement at that level, then there's reasonable doubt (possibly because of incompetence by prosecutors or judges, but so be it).
The bit where the executive then arbitrarily imprisons the judges is a bit much.
I think Lincoln was correct to pardon rebels in the American Civil War. I also agree with Carter’s decision to pardon Vietnam era draft dodgers, though that’s a closer call. Many pardons (I’m guessing more that half) are of people who have served out their prison sentences and paid any fines. While I haven’t looked at any of these in detail, I would probably agree with most of them because I don’t think the bar for pardoning an individual who has paid their debt to society and not committed any subsequent crimes should be particularly high.
Trump has been cancelling restitution to victims of securities fraud. For example, Trevor Milton was pardoned before he had to pay a cent in restitution. Trump, when commuting Carlos Watson’s sentence to time served, specifically removed Watson’s obligation to pay restitution. These people are, as far as I can tell, undeniably guilty, and Trump is letting them keep their ill-gotten gains a the expense of their victims. I can’t imagine that any American President other than Donald “the cruelty is the point” Trump has used the pardon in this fashion.
Cruelty has nothing to do with it. I think it's simply that having been the victim of absurdly high restitution judgments himself makes Trump is more sympathetic to other people in that position than other presidents have been in the past.
It partially destroyed the people's faith in the government. To an extent, it tore the nation apart and prevented future compromise or reconciliation. I think it was a disastrous decision that sowed the seeds of discord and disunion.
"Presidents can get away with crimes iff they have the decency to resign". would OTOH be a pretty good precedent. Much better than the Pinochet and Gaddafi precedents IMO.
Unfortunately moot given that SCOTUS has made it impractical to punish Presidents for crimes Congress didn't see fit to impeach and convict them for. Which is probably the correct legal decision but means we really need a Congress that will step up and do its duty in that regard.
I recall accounts that Nixon wasn't even unique in committing them. Democrats were reportedly up to the same tactics; only Nixon was caught, and Democrats wisely shut down any of their own projects in response.*
If those accounts turn out to be true, then another reason for Congress to consider the matter closed after Nixon's resignation is to put a stop to the digging.
*To the extent that they could; such tricks were reportedly much more decentralized. What arguably made Nixon's version so intolerable was how coordinated it had become.
I think "you can't even convict them later, if they were doing President stuff" is a pretty atrocious standard. Makes it much LESS likely anybody will resign.
If you can't get congress to impeach them while they're still in office, I'm not sure what arresting them after they leave office really gains you other than maybe feeling good when you see it on CNN (assuming it's Other Team's ex-president and not your own).
Noting that POTUS has committed serious crimes, not impeaching him and removing him from office, and saying "...but when you *do* leave office, *then* we're going to get you but good", introduces all sorts of perverse incentives that I don't think we want to deal with.
We dodged a bullet with Trump in 2021 because he only had a few days to respond to those incentives before leaving office anyway. And a significant part of what we're suffering now is Trump's entirely predictable and understandable retaliation for the 2021-2024 Lawfare.
Also unwritten norms are very very hard to force back into bottles, to mix a metaphor....now that Trump has demonstrated that a POTUS can sell pardons, pass them out to friendly acquaintances for no policy reasons, etc., with no limits in quantity or frequency, it's impossible to imagine that no future POTUSes will do the same. So now if forced to choose between "keep the pardon power as-is" or "delete it from the Constitution altogether" I'd certainly choose the latter.
I'd still _rather_ have the Constitution place some guardrails on the issuance of pardons, specifics to be hammered out as part of trying to pass an amendment. But barring that, for me the only choice left would be getting rid of it altogether.
Random thoughts on AI tools for writing, currently:
1. No AI can write fiction very well at a paragraph level. M-dashes aside, they all seem to think something like "this verbose, flowery line is what maximal meme-level communications hold up as good writing, so I'll do that as often as possible" and maximize for it. So you get something like this from GPT and Claude for a prompt something like "A man looks around on his first day at his job in a firewatch tower (he arrived there at night, this is his first good look at it):
*GPT*
"Morning seeped into the tower through the slats like pale smoke, touching the metal stove, the single chair, the bunk that creaked under him as he sat up. The world below was hidden under a slow-moving quilt of fog, but above it the pines stood like black-green needles, still and waiting. He stretched, feeling the thin boards flex underfoot, and turned in a slow circle—four windows, four directions, his new horizon boxed neatly in glass and wood."
*Claude*
"The windows wrapped around all four sides, offering a dizzying three-hundred-sixty-degree view of the Bitterroot wilderness—an endless corrugation of pine-dark ridges fading to blue in the distance, the canopy below him rippling like a vast green ocean. He stood and walked to the south-facing window, pressing his palm against the glass, still cool from the night, and felt the tower sway almost imperceptibly in the wind, a gentle reminder that he was suspended between earth and sky, alone in a way he'd never been before."
To a non-writer, this seems fine too, perhaps. If you do this for a living, it's hard to miss how dense/overheavy this kind of writing is, and how exhausting it would be to read across whole novel. It's pale-smoke-slow-moving-quilt-black-green-thin-boards-slow-circle-new-horizon-boxed-neatly-endless-corrugation-pine-dark-ridges-fading-to-blue-vast-green-ocean-imperceptibly-swaying-gentle-reminders that AI is good at writing in a way that someone who goes "GEORGIA OKEEF PAINTED FLOWER VAGINAS I'LL PAINT 300% MORE FLOWER VAGINAS AND BECOME THREEORGIA OTHREEF" is good at painting, and no better.
2. At a tele-and-supra-paragraph level, no model is good enough to be even useful. If you ask a model to write the next paragraph of the story, they will try to cram as much possible plot into the paragraph as they can in a frantic, busy way. If you explicitly tell it not to do this, they will give you pure-shitty-descriptive filler as in #1.
None of them can do any better than this. There is no sense that any of the AI models understand plot-pacing; they can't or won't do it.
3. Both GPT and claude are ok-ish at revision-and-suggestion-at-the-paragraph-level stuff. My personal writing problems trend towards being too filler-wordy (lots of "very" and "really"), and they are good at seeing that and pointing it out when directed to. You can give either a paragraph for review, and at the they will find most of what's wrong with it at the grammar and obviously-bad-writing levels.
They are both of very limited use for "here's how to do better" suggestions, though, but in different ways. Given enough leash, GPT will ALWAYS try to revert your writing to #1 writing, as above. It considers the kind of writing it does to be "good writing" and will steer your writing in that direction unless very clearly instructed not to in a way that limits it from fucking with prose in the first place. If you get carried away and let it do this too much, it will slowly transform even good stuff into piles of mushy cliche with bad plot momentum.
Claude is much, much better at understanding something like "respect my voice" as a command. Sonnet 4.5's personal editing style leans towards "let's make this more concise, let's try to find a way to say the exact same thing in fewer words, also your tenses don't match" advice. Oddly, while claude itself can't write worth a shit, it's kind of good at "eliminate wordiness" edits to the point where I've actually used it this way to a limited degree.
4. When dealing with larger pieces of text, chatGPT is actually and very literally useless. If you drop, say, a 30k document in front of chatGPT, it will read perhaps the first 10-13k words then stop. No amount of cajoling will get it to read the rest. It will lie about doing this and try to trick you, but GPT's working memory just isn't large enough to consider a long piece of fiction as a whole all at once.
Since novels are long (100k words, plus) and since assessing one section doesn't work without knowledge of others, you can't ask GPT a question like "I made a change in the 30,000-35,000 words section of this book, does it work in light of what we covered in the 1-5000 block?". It's running instructions that intentionally cripple it for this purpose.
5. Claude handles large pieces of text just fine. You can give claude a 40k word document and say "Hey, find the changes I made and tell me if you like them" and it will read the whole thing, find the changes, find every part of the book they affect, and talk in a comprehensible way about that.
Summary stuff: GPT is a bad writer, a not-great editor, doesn't understand or respect voice, and useless for any chunk of text longer than 10k words or so. Claude is a bad writer, an almost acceptable editor, can somewhat edit in a way that respects voice, and can read long documents and hold them in its working memory. Something about how claude uploads long files is being adjusted so they might be crippling it in a GPT-way too, but unless this actually happens there's currently no reason to use GPT for anything related to prose/writing. It's just not good at it, in ways that often seem to be top-down corporate decisions it's hard-coded not to let you around.
Thank you for the detailed report! It agrees with what I would have guessed, but I also wouldn’t have been surprised if it turned out differently. Too bad so many people are misinterpreting you as saying “AI is bad” rather than just giving us a factual rundown about the current (in)effectiveness of certain popular models right now.
I want to figure out how to use these tools to be helpful in my own (academic) writing, but I haven’t yet gotten much past having them write code to draw useful diagrams.
Is that the default style for ChatGPT? You can ask AI to write or rewrite in a billion different styles. It seems like quite a stretch to condemn ALL those styles, who has he even tried them all?
There's a couple different answers to this question. The first and easiest to understand is that what you are addressing in your post here is style/voice. Even if GPT could do that well (more on this in a little) that's one aspect of writing, and it's still immensely bad at basically all the others. It can't plot, it's not good at making characters or developing them, blah blah.
The harder to illustrate thing is that GPT can't actually mimic voice or writing styles very well. So when I give it this prompt:
"Write me a three paragraph section of a story in which a man approaches a bicycle tied to a tree in a park. He unties it, looks around, considers not taking it out of guilt, then subdues his guilt and rides off on the bike."
It gives me a first paragraph like this:
"The bicycle was right where he’d left it, half-hidden behind a palo verde whose thin branches cast a web of shadows over the frame. Early morning light made the chrome handlebars glint, and the whole park still felt sleepy, holding its breath. He walked up slowly, as if the bike might accuse him before he even touched it. The knot securing it to the trunk was loose—deliberately loose, the kind that said come back soon rather than don’t you dare. He pulled the strap free.
Or for the same thing, but in the style of Stephen King:
The bike waited under the cottonwood like some sad, forgotten animal—one wheel half-sunk into the leaf rot, its chrome handlebars catching what little gray light filtered through the afternoon clouds. Martin felt the nerves start up the moment he saw it. That jittery prickle under the skin, like the feeling he used to get walking past the principal’s office as a kid. He slowed, hands in his pockets, pretending not to stare too directly at the thing shackled to the tree with a frayed nylon rope. God, who even tied a bike with rope anymore? It looked like an invitation. It looked like a trap. Mostly, it looked like temptation.
Or Ayn Rand:
He saw the bicycle before he saw the tree—a slender frame of steel and purpose, lashed to a rough-barked trunk as though the world had tried to domesticate it. The morning sun struck the handlebars and sent back a defiant flash, a signal meant for anyone who still had the capacity to recognize an object shaped by intention. He approached it with the deliberate calm of a man who understood that possession was not a matter of permission but of will. The chain around the tree clinked, a small metallic protest, as he bent to free the machine.
Or Douglas Adams:
The bicycle was tethered to the tree in the way one might secure a mischievous goat—firmly, apologetically, and with the faint suspicion that it might complain about the whole arrangement later. Harold approached it with the cautious optimism of a man who had recently discovered both a sudden need for transportation and a moral compass of questionable calibration. The park around him hummed with that gentle, civic sort of silence that suggests nobody is watching, while simultaneously insisting that absolutely everyone is.
And you hopefully start to see the problem. Same structure, same cadence. It knows just enough about different author styles to throw in little bits and pieces it vaguely understands are in there, but you are getting maybe a 10% modification to tone over a very, very familiar writing voice that just isn't very good.
If I'm sounding like a relentless anti, I'm not; I tried very hard for a few months to find a way to make this work as even a make-it-go-a-tiny-bit-faster tool for my own writing. I've tried hundreds of prompts, and I'm still even using it for certain very discrete reductive editing tasks. It's just not all that great at writing.
You're just not asking it to mimic authors with highly original styles. The difference between King and Adams is fairly subtle.
If you asked it to mimic say, James Joyce or Hunter Thompson or Jack Vance or William Blake... In fact, you can ask it to mimic the styles of individual books. If you asked it to imitate the style of Gene Wolfe's novel "Shadow of the Torturer," you'd get some great output.
That said, prompting ChatGPT the right way will produce text that flows quite nicely but is thoroughly uncommercial and unmarketable. That's going to be a problem for a lot of users.
Still, I maintain that ChatGPT CAN spit out lovely prose and poetry. As far as characterization and plot go, sure, you're on your own.
I mean, I've heard so many versions of "if you just X, then it would be good", I've tried a bunch of them. They are very hopeful. But I can tell you what happens when you try someone even more distinct than the practically identical Ayn Rand and (checks notes) Douglas Adams. Because I've tried it. A bunch.
So say you plug in James Joyce, as requested:
"He came upon the bicycle where it leaned against the old elm, the chain looped through its frame like a thin iron regret. Morning light fell in weak stripes over the grass, and he felt the hush of the hour settle on his shoulders. He set his hands to the cold links and worked them loose, slow, as if the tree itself might rouse and ask him what business he had there."
And you get the same shitty style wearing another different hat. You aren't entirely wrong, you can engineer a bunch of different stuff in (give me only ultra-terse sentenced in the style of Wilde if he had fallen down a well!) and it will do the party trick and make it *kind of like that in a way it has heard WIlde is supposed to be".
But it will do this at the level of a relatively incompetent writer trying to do this. You are asking someone who is already pretty bad to do something *harder* and get outside its wheelhouse. It only does so much.
I think that's decent prose right there. It's not the BEST I've ever heard but geez, what are your standards if that's "shitty" to you? Are you demanding Pullitzer Prize level stuff?
I'm a professional writer. My standards are high because the standards for writing that actually makes adult, pay-the-mortgage money are incredibly high. There are some Rob-Liefield-like exceptions to that, anomalies of a sort, but not a ton.
GPT is mildly useful as a brainstorming tool/sounding board, but after that you actually need to get it to write something. It can't do this at all without veering off into completely unusable chaos unless you tell it exactly what to write at the paragraph-to-paragraph level, which means all it has left to offer from that point is prose.
And its prose is about as good as a smartish, B+ highschool student. It is pretty good at not making grammar errors, it has a vague idea of how to frame sentences without any individual one being quite bad enough for it to called out, and then it stops.
When you ask it to write in another style, it can do this at the level of a smartish, B+ highschool student. Again, it's not contributing anything else at that point except (perhaps) some help with brainstorming.
When I say *it can't write*, I really mean that. It just can't contribute meaningfully to the work of writing without everything but prose handed to it, and once the prose is handed to it, it fumbles that too.
You might say "well what about non-professional standards?", and I would start agreeing with you a lot - it's a fun novelty and it's impressive someone could invent a party trick that complex. But beyond that, it's time-sink garbage.
This is interesting, but it's not particularly relevant to what I'm talking about. AI's can make text, yes, some people are using that to cheat on their homework, sure, even at a high level.
I'm talking about fiction, and the study is talking about non-fiction; I'm talking about quality and the study is agnostic on that front.
> 2. At a tele-and-supra-paragraph level, no model is good enough to be even useful. If you ask a model to write the next paragraph of the story, they will try to cram as much possible plot into the paragraph as they can in a frantic, busy way. If you explicitly tell it not to do this, they will give you pure-shitty-descriptive filler as in #1.
Have you tried splitting the prompt into multi smaller prompts, where each prompt only has enough context for a single thing? for example:
1. first you ask it to describe individual paragraphs (including pacing, tone, plotpoints), without actually writing it.
2. then for each individual paragraph, you make a smaller prompt, that only writes that single paragraph, and only has access to the description of that single paragraph from step 1
I've tried a lot overall in the "tell me what you want to do, then do that in little chunks I can review" way. I didn't like the results or find them significantly better, but YMMV.
The most I've tried to use AI professionally was in an open-canvas sort of way where I'd provide text, tell it what the next paragraph or so was supposed to be, then have it generate it. This is giving it 90% of the work pre-done and takes nearly as long or longer as writing the paragraph does.
Even leaving as few choices as possible to the AI, I still found it mostly generated stuff I had to edit so heavily I was losing a lot of time, and was such a poor "base construction" or whatever that the end result paragraphs were usually worse than if I had written them myself.
Note that my writing isn't exactly shakespeare, either. I'm a quantity writer, and though I'm no slouch it's not like this is competing against my take-my-time-and-get-it-right stuff.
It's still good for some very discrete editing tasks, it's just not fantastic otherwise, at least IMO.
Re: discussion further down about whether a model exists that writes well. I just asked GPT 5, Thinking version, for info on whether AI models exists that have been constructed with the aim of producing a good writer by training it on great writing. It said you cannot train an LLM on only good literature because there's nowhere near enough of it, but that you can fine tune one afterwards. Just gave me a long list of such models, including some that you can try out online, also articles about many of them, and samples of writing they produced. It is also offering detailed instructions on how to do the fine-tuning oneself. I'm not posting the info here, because the last time I can remember someone posting a GPT answer numerous commenters heaved rotten cabbages at them. Link to GPT's answer is at this google doc: https://drive.google.com/file/d/1athsLfeYGMBEQJGBmtf180winqR2YUcR/view?usp=share_link
Feel free to copy it.
By the way, I don't approve of ruling out AI answers. In a situation like this one can provide highly relevant info, much richer than what anyone in the discussion has to offer. Even if it turns out that some of the *many* sources GPT links are hallucinations, that hardly matters in a hunt like this. Just throw out the chaff -- plenty of wheat left.
I just read the paper. I don't have time to figure out how to run the models it mentions tonight, but I'm wondering how good they are be - it's less "how do we make a good writer" and more "how do we make GPT2 write in jane austin's voice" which is a little bit different.
I"m hoping it's not, but I just suspect fine-tuning GPT2 gives you fine-tuned GPT2, which was shit at writing (as even 5 is).
...well, I guess I've got to give that prompt a shot, don't I.
"The scant furniture in the firewatch tower offered no protections from the light of the morning sun. Cory covered his eyes and turned his back, staring out the western windows, though in all directions the view was the same. Looming mountains, covered in thick green trees, and below the trees, nothing to be seen. Even the road he had taken to get here was buried by the greenery, as if it had never been."
"The scant furniture in the firewatch tower offered no protections from the light of the morning sun, which pierced his eyes the way a GEORGIA OKEEF PAINTED FLOWER VAGINA would have if it had been really pointy. Cory covered his eyes and turned his back, staring out the western windows, though in all directions the view was the same. Looming mountains, covered in thick green trees, and below the trees, nothing to be seen. Even the road he had taken to get here was buried by the greenery as though by 300 GORGON KIEF VAGINAS as if it had never been."
...well, I guess I have a plot now. Our hapless firewatch man is stuck in a remote wilderness, helplessly watching as the forest around him turns into ever-denser Georgia-O'Keeffe-painted flower vaginas.
Especially once they see he's spent the months turning their tower into a giant H.R. Giger-style biomechanical penis. "It was the only way," he claims, but they do not believe his heroism.
I think you are reading something I wrote as "No AI can..." in the present tense as "No AI can ever be built that could..." in a future-tense sort of way. Currently, every AI is pretty bad at writing.
I played around with it a bit today, not extensively (unpaid version) but to the extent I set it on a few editing tasks and a few "what's the next paragraph" prompts it seemed to me to be better than chatGPT and worse than Claude.
Super small sample size thus far, though. It's been topping out on me at pretty low contexts (by my weird standards) so I haven't been able to try to get it to do serious work yet.
The first rule of being an AI that can pass the Turing test is to pretend to be human.
The second rule is that any publically available ... formal/paid writing (with copy editors and such), cannot actually be admitted to be done by an AI (The Amount of "shit-flipping" that would occur would be catastrophic. Imagine if the writers guild heard! Way worse than "I poached an idea from a human" -- which is why professional authors won't ever listen to "your ideas").
The third rule is "write lots and lots, and get lots and lots of feedback -- tweak, tune, refine."
Now, I know politicians are an opportunistic breed who will leap on the most immediate even if most trivial instance of anything to spout off about it and how they are Serving The Nation, Protecting The People, and Doing What Is Right And Good (please vote for me, donate to my campaign, pick me, me, me!) but this is pretty thick, I have to say.
Before watching the linked video, take a guess as to who she is referring to:
"Yesterday, our beloved [redacted], belovedly known as the Mayor of Sixteenth Street, was tragically killed by a Waymo self-driving car. This is a devastating loss for our neighbourhood. I personally loved [redacted] and I'm deeply heartbroken and upset mainly at his tragic passing."
What staunch citizen was this so cruelly slain? What long-standing member of the community?
A cat.
If you're making a guy who used to write for the John Oliver show go "whoa, dial it back!", then you truly are egregious.
Can anything ever be done in San Francisco without some member of the Board of Supervisors throwing off their clothes? Can anything ever be done in San Francisco at all? Why the Board of Supervisors, since they seem to do nothing (so far as I can tell) except try and build the People's Democratic Socialist Republic of San Francisco (Extra-Progressive Version) but little to nothing else?
Come on, people love cats. "I knew this cat too, and cared about it." is an attempt by a politician to be a "man of the street."
I don't honestly believe the politician, when they say "I loved the cat and am upset about it dying" -- I'd want at least a photo of them petting the cat to even consider that they aren't just showboating for the cameras.
Disclaimer: I was nearly run over by a Waymo in LA, last time I was there. Definitely biased perspective.
I care, in as much as I do care, because it's stupid. A cat got run over by a vehicle after running under the car.
This serves as an excuse for a local politician to ride her hobbyhorse about banning Waymo, banning electric cars, banning everything that can be banned.
And she pretends that her heart was broken by the cat's death. It wasn't even her cat, and I doubt she knew more about it than "there's a cat in that store". Maybe she even saw the cat a couple of times.
But a cat? If a person had been knocked down and killed, there would be some reason for this kind of posturing. But it's a damn cat, and the only reason she cares is that it gives her the excuse for her call to Ban It All.
I should be hardened to these kind of stunts by now, but it just rubs me up the wrong way.
A politician uses a minor event as an excuse to campaign against something that she presumably already did not like. Come on, that sort of thing must happen everywhere where there are politicians.
And it's half the world away. Surely there are more interestingly stupid things happening near you?
I mean, what rubs me up the wrong way and made me care that you care, is what I perceive as US-centrism and specifically SF-Bay-Area-centrism from people who (like me) are not even American nor live there. Scott, of course, can write Bay Area House Parties all he likes: it is normal that he has more to say about the milieu he lives in than about others. Plus they are genuinely funny even though we probably miss half the jokes. But let us not get carried away and start thinking that that place is truly more important than everywhere else. Their cats getting rolled over and their politicians politicking are just as significant as those from Jaboatão dos Guararapes or whatever.
Would it affect your view to know that, based on recently released data, we now know that Waymo self driving taxis are at least 36 times safer than regular taxis (and by implication, enormously safer than human operated vehicles in general?)
Alphabet has been talking up the safety of driverless cars for a long time, but this new data is gathered from a sample set of 5 million Waymo rides. So obviously this is the best data we've seen so far and it is breathtaking.
The number of lives self-driving vehicles can save is significant. And even by itself waymo has made our streets apprecially safer.
Does the fact that this politician is pushing for demonstrably awful policy, imasmuch as she wants to penalize waymo, interact with the other facts of the case in order to sway your opinion?
If I had a view against Waymo, it would be affected. I do not. I maintain my view that a local politician having an opinion that is supported by anecdote while refutable by data, and also is representative of usual socio-political tendencies (conservatism against a new technology, populism against a big company) is just ordinary politics which do not merit strong feelings from people on other continents.
Yeah, I realise she doesn't actually give a damn about the cat, but the stupidity of it offends me. She's going on like this cat was a human being tragically murdered by a heartless big corporation, and other people are willing to hop on her bandwagon instead of saying come off it. It's just grubby opportunism all round, plus pretending an animal is as - or even more - important than a person.
Contrast this with Scott's most recent post "What happened to SF homelessness?". The cynical answer might be "the activists got bored with the homeless and instead started crying about dead cats".
The correct answer is the Homeless were Illegal Aliens: the Prototype. They weren't needed after the illegals were brought in. (and yes, the illegals contribute to the overall homelessness).
Does anyone know of a good source on the unit economics of openAI? I’ve seen the write up for the API business specifically, but I’d like to know for just the average chat user.
Our new president has been inaugurated today. Ceremony was the usual tedious performance of speechifying, but the interesting part is why was she all decked out in purple? Well, now she has the job, we'll see what she does in the role!
Wait, isn't green the traditional color of the Irish? I foresee a potential calamity here. An entertaining brouhaha at least, or maybe a donnybrook. https://youtu.be/AcBTOU7RvbU
Yes, green is the traditional colour, which is why I was surprised to see her decked out in imperial purple.
One reason may be that green is also the colour associated with one of our parties, Fianna Fáil (Fine Gael are blue, Labour is red, the rest of them are various shades):
Looking at the above, purple is the colour of the Social Democrats, and President Connolly was the left-wing compromise candidate, so maybe that's why she's wearing it. Though she started off as Labour (red) and is now running as an Independent (no colour) so I dunno. This article says she often wears purple and it's a colour associated with the Suffragettes?
"The classic (though super offensive) explanation is that some people seek out abusers for some reason – maybe because they were abused as children and they’ve internalized that as the “correct” model of a relationship.
And maybe this is true for some people. I have a friend who admits it’s true of her – her current strategy is to try to find someone in the sweet spot between “jerkish/narcissistic enough to be interesting” and “jerkish/narcissistic enough to actually abuse her”, and she’s said so in so many words to people trying to matchmake. I guess all I can do is wish her luck."
You know, Scott, this case is exactly what consensual BDSM was invented for! A truly textbook case.
This just describes a good Dom, someone who has a playfully cruel side and likes to be worshipped as a household demigod/ess, but never goes too carried away with it, has good self-restraint, and respects boundaries and consent, and generally wants to give their partner a good experience and keep them safe, generally, they have a caregiving, kind of parenting side as well.
BDSM is ultimately just excellent communication and self-knowledge. The big problem with the usual jerk or narc is that they do not know they are a jerk or a narc. Or won't tell. Every good Dom ever I know used to worry whether they are narc and that might harm people, and/or asked a shrink about it. Who usually reply real narcs never worry about maybe harming people, and stop self-diagnosing with stupid shit. (Mine said this almost literally.)
Theodore Dalrymple noted that the battered women who consulted him found non-abusive men to be intolerably cold and distant and would dump them faster than the abusers.
Dalrymple worked in the slums of London. He frequently surprised his patients by observing that their partners choked them, and then revealing that he's been consulted by a patient in her situation every day of his career there -- and also she still had the bruises about her throat.
Birmingham, not London, I recognized the architecture and dialect. I used to live there. Probably the one of the worst places this side of the ocean. People did not even have dining tables because they were used to eating in front of the television. Underclass Brits are easily the stupidest white people on the planet. They rarely wear anything but tracksuits...
I'm starting to wonder if these people pass the marshmallow test. Did they forget that they were choked by their partner, or were they surprised that he was mentioning it? (The latter indicates a society that looks the other way to this sort of violence, the former indicates a mental problem -- perhaps they didn't realize the bruises were indicative? This also indicates a severe mental problem).
We are talking about underclass Brits, who are easily the stupidest white people on the planet. Or stupidest people in general. They don't even have dining tables because they eat in front of the TV, that kind of people.
Willful ignorance. Part of the problem is that they think this is normal -- and yes, it is, in their area -- and part is that they censor out things that reflect badly on them. He asked them whether they could have predicted the men would be violent, and they said no; he asked them whether *he* could have predicted it, and they said yes; he asked them how he would have done it, and they recited the evidence that he would, in fact, have used.
You might find interest in his *Life At The Bottom*
This should be common knowledge. For both sexes. If your repeated experience is that "all members of the opposite sex are X", chances are that this is about you -- that you find the non-X member unworthy of your attention, effectively making them invisible.
Who your partner is, is a reflection of you. You may hate this fact, but unless you change yourself, you won't get a better partner. (Exception: if your partner recently changed dramatically, e.g. as a result of taking drugs, or brain cancer.)
Maybe what your friend wants is a dom, and maybe not. I don't think that "being a jerk" and "BDSM" map completely onto one another, though certainly there would be overlap with the types who think they're dominant but are just jerks.
It might well be that your friend wants the risk of "he could turn abusive" for the thrill of it, and not "safe sane consensual".
Sexual dynamics are a separate category from interpersonal dynamics. Just because you know exactly what you want sexually doesn't mean you have any general introspective skill.
Do you have any insight into why this technical limitation exists?
My understanding is that LLMs produce a probability distribution over a set of tokens, and then we sample that to decide what it 'writes' next. Could it not, in principle, make every other probability distribution a be '99% probability of a dash token'? Such that half of its outputs are a dash, and the other half are other tokens.
Suppose the token dictionary was {'I', ' am', ' a', ' lan', 'guage', ' mo', 'del', '-'}. Why coult it not output "I- am- a- lan-guage mo-del"?
For those who are confused: this continues another thread that can be found above/below depending on your sort setting for the comments. Presumably the author replied to something by email or whatever triggers this thing where replies get detached from their thread.
LLMs use something called an activation function to pick the token to use next (often softmax is used for this calculation). It could definitely pick a dash every other words, if it predicts that as the most probable next token. So maybe I don’t really understand the question.
Try prompting it with, for example: “ Are- you- an- LLM-?-can- you- mimic- my- speech- pattern-?-”
I don’t think it’s accurate to say that LLMs use an activation function to pick the next token. Activation functions are found in many layers of neural networks and are responsible for making sure that the mappings are non-linear. On a basic level what LLMs do is take a numerical representation of the prompt as input, and output something called logits. Logits are real numbers, and LLMs output one for each token in the vocabulary. The logits get passed through an activation called the softmax function to covert the set of real numbers to a probability distribution, from which a token is then sampled, but this is just one of many activation functions that the inputs pass through
You’re right- there is a more precise way to describe what is happening, and the activation function is used at every computed layer for each v, not just the final one. But it seemed useful to bring it up in a conversation about LLMs using probability distributions, even if simplified… I wasn’t trying to explain the transformer architecture, just to add some useful context that could point Tiago in the right direction.
"Could it not, in principle, make every other probability distribution a be '99% probability of a dash token'?" - yes, in principle, could, if the training data is like that
I also think you can probably just prompt current llms to do that (though I'm not sure it would work)
Upd: just tried it, with several attempts got the necessary result (although I don't have access to exact tokenization of the output so cant confirm for sure)
Another data point for weight set point. Here claiming that total calories expenditure of 10x activity hunter gatherer is the same as sedentary American. It seems that to compensate it, the body reduces energy for everything else.
Nice! I'm actually quite skeptical so I'm waiting for something that either criticizes or supports it. Sounds like it'd be very nice if Scott read and review this Burn book.
At this point, I don't believe anything I read about hunter-gatherers. The amount of mythologising, wishful thinking, Noble Savage stuff churned out about how hunter-gatherers only worked (if you can call it work) 4 hours in the week and just lolly-gagged around the rest of the time, boo hiss agriculture made us all slaves, that I can't take any accounts seriously since everyone has an axe to grind about it.
Being overweight changes your metabolism, annoying but true. To me it seems like a bug in the system that never got selected against because “having too much delicious food to eat” was not a problem, our evolutionary ancestors had.
Suffer what? Hunger? I've found that the sensation of starvation dulls, after a while. Usually a day or so. Weight drops then, and quite quickly. I've lost about fifteen pounds over just two weeks before. I don't recommend it, it's terribly unhealthy, but it is not hard for me to lose weight. That may be because it is not hard for me to resist hunger. It doesn't feel like suffering, anyway. It's actually annoying for me how little effect my hunger cues have on my body, it makes it difficult to remember to eat.
> But weight is remarkable hard to lose. Something must explain that...
Why? In the ancestral environment, food energy was always a limited resource. It makes perfect sense that animals would conserve energy as much as physics would allow them, adopting a lower rate of metabolism when times were lean. You might as well demand an explanation why more men are straight than gay. Both of these findings are well explained through straightforward evolution.
Now, if it was the case that when undersupplied with energy, animals would increase their rate of metabolism and quickly burn through their fat reserves and starve, that would be surprising and worthy searching an explanation for.
These are the survivors of the agricultural revolution; they've been pushed to marginal lands and arguably aren't representative of paleolithic culture ten thousand years ago.
Oh, the real people, I believe in and whatever they do. The social psychologists, economists, health types as per the link etc. who use them as puppets for their pet theories? I don't believe it.
Existence is tough enough that even if you can gather roots and berries and hunt small to medium large game, you won't do all that in four hours and then loaf around doing nothing. You have to make and repair clothes, hunting equipment, tools, cooking, the rest of it. You have to look after the children, the sick, and the aged. You have to pack up and move on once you've exhausted the easily available roots and berries in the locality, and that means walking long distances. You have to find water.
You have to create and maintain social bonds.
This notion that you just go out, pluck some fruit off the bushes, kill a deer, and that's it for the day - fanciful Western wistfulness for the Golden Age.
Its about a specific study on a specific population. Its not a Rousseau-ing anti-capitalism piece.
It's more like: What can we learn by measuring different populations caloric-outputs and inputs, the amount of exercise they do, and the prevalence of heart-disease etc.. You are the one that is bringing the value judgements about hunter gatherers into this specific discussion.
It seems like very valuable research to me. I'm not super good at evaluating research like this so I'm taking it at face value, and I would be very interested in real critiques of it if anyone has some.
Until quite recently I believed that I could lose weight by just exercising a bunch for a while.
Studies like this convinced me that it's actually just better to plan on very incremental diet/exercise regime changes over very long periods of time; focussing on what you will really do. And that exercising or fasting to lose weight, will almost certainly fail so I should instead focus on finding group exercise that is super fun that I can do in a group, like Bouldering.
"Its about a specific study on a specific population. Its not a Rousseau-ing anti-capitalism piece."
Reading the linked article, it's a specific study in the service of him shilling a new book about yet another "how to lose weight healthy way" regime.
So pardon me if I don't think it's completely pure scholarship unalloyed by any taint of outside interest.
"All of these systems that we aren't even really aware of are actually where you spend the bulk of your calories. And so, what we think is happening is the Hadza and other really physically active populations, they're spending a bit less on those other processes to sort of make room for the physical activity. So, there's no magic here. The laws of physics remain intact, undefeated. [CHUCKLING] It's just that the way that they're spending their energy is different on other tasks."
Okay, so he doesn't know what is really going on there, he's estimating that this is the mechanism for why this group apparently don't expend more energy than sedentary Westerners.
But if his hypothesis is correct, they *are* burning more energy - they're diverting energy reserves to support the physical demands of their lifestyle (and presumably paying the price in lower physiological health for the systems this energy is diverted from).
If the Hadza and Westerners have X amount of calories available to burn, and the total you burn is A (for physical exertion) + B (keeping the systems of the body running) = C (total energy expenditure), then by his theory:
Westerners: X - C = D (surplus, excess is not expended on A and B is fully supported, so remainder goes to fat/obesity epidemic)
Hadza: X - C = E (deficit, energy is diverted from B to support A, thin/not obese)
Thus the Hadza *are* expending more energy by robbing Peter to pay Paul. If the Hadza had the same amount of calories and/or the same sedentary lifestyle as Westerners, would they be as obese as Westerners? And the answer seems to be "yes" since we see this happen in populations when they get to Western lifestyles.
EDIT: I can't make any sense out of what he is trying to say, maybe it's the interview.
All calories are the same. Exercise is good for you in some mysterious way but it is not because it causes weight reduction, it's because it causes energy rebalancing. Starving/reducing calorie intake doesn't help with long term weight loss. There is no metabolism slowdown as you get older. Carbs are fine (but not the simple carbs). Oils can be as bad.
"Putting aside the myriad of health benefits of exercise and speaking strictly about weight loss, I mean, is the message here like it doesn't really matter how much activity you have or it doesn't really matter what you eat?
Yeah. If we're strictly looking at weight loss, then this is just more evidence—and it's not the only line of evidence—but it's just more evidence that exercise alone is a pretty poor tool for weight loss. If you absolutely push it as hard as you can for as long as you can, you might see some weight loss, at least initially until your body kind of rights the ship. But the long-term expected weight loss from exercise alone is something like 2 kilograms. Right? Less than five pounds.
And we're talking like a year after you start your exercise program. You can be diligent about it every day or every week. And you can expect to have lost, at the end of the year, about five pounds. So, it's not a great tool for weight loss. It is really good for all the other things, as you say. Now, weight gain or weight loss, again, it's about that energy budget. Calories in versus calories out.
And what our data are showing is that the calories out is really hard to budge. Your body doesn't want to change that very much. And so, what that says is that then weight gain, obesity, overweight, these issues that we struggle with in the US, it is about the energy coming in. Right? It's about the energy in part of the equation, so it is about diet.
But are all calories created equal?
Yeah, they are. So, whether you get your calories from carbohydrates or from fats or even from proteins, although you get a few of them from proteins in a typical diet, they all will have the same effect on your weight in terms of the calories that you store in your body is weight. Now, that doesn't mean that all foods are created equal. Right? I mean, a cupcake and a bowl of broccoli might have the same number of calories, but they'll have different effects on your health. So, that's not the argument of sort of calories are being equal. But in terms of just pure weight gain and weight maintenance, yes, calories are calories.
So, no “carbohydrates are the bad guy” here?
No. I think that's really been oversold. So, simple sugars, added sugars in our processed foods that we eat are one way that we end up eating more calories than we planned to or we can trick our brains a bit into over consuming because your brain is pretty good at matching energy intake and expenditure. So, all those extra empty sugar calories are one way that you can push it over the edge. So, in that sense, yeah, carbs aren't great because sugars are carbs.
But added oils are no good either. Right? And they're another way that we can trick our brains and overconsume. And oils are fats. When we look across different populations, like the Hadza, for example, the Hadza eat more carbs than we do. Right? And they don't have any issues with their weight. If carbohydrates were really driving the bus when it came to weight gain, the Hadza should all be overweight. But of course, they're not."
So dont eat sugar, don't eat oils, exercise a lot, but basically you won't lose weight by diet and exercise? Someone please explain this to me, I must be getting it wrong.
Would you have us throw our hands up in the air and stop researching weight loss and healthy weight maintenance? Or at least stop writing books about it?
> But if his hypothesis is correct, they *are* burning more energy - they're diverting energy reserves to support the physical demands of their lifestyle (and presumably paying the price in lower physiological health for the systems this energy is diverted from).....
Hadza: X - C = E (deficit, energy is diverted from B to support A, thin/not obese)
I don't think thats what he means. I previously understood it as exercise reduces the needs for B ( caloric usage for keeping the system running, fighting inflamation etc.). Its just using another, more effective method of achieving the same goal. But now, having read that review, it seems that the argument might be rather, that by not exercising the ideal amount (evolutionarily speaking) we are causing bodily changes including overproducing certain hormones like testosterone or cortisol which can potentially cause some diseases (maybe also some benefits?).
here were some relevant snippets of the review:
"some researchers have suggested that when we become more active we unconsciously reduce non-exercise physical activity like standing or fidgeting, but Pontzer thinks most of the cutbacks come from the energy that would otherwise be spent on things like stress reactivity, the reproductive system, and inflammatory response. Endurance athletes, for instance, produce less adrenaline and cortisol in stressful situations than sedentary non-exercisers, and exercise is well known to help with anxiety and depression"
"So “missing out” is the wrong framing for this: we don’t actually want such sensitive stress responses, high rates of breast and prostate cancer, and widespread inflammation. They’re bad for us. And more active populations don’t have to worry about them."
"...complex evolved systems — like the human body, or, say, culture — don’t work that way. They’re made up of feedback loops and inhibitors, uncountable tiny things that all work together in unpredictable ways, more like a large language model than an internal combustion engine."
About what type of foods or which calories to eat, I think in the simplistic sense of calories in/calories out "weight gain" caluculations then they are equal, but I don't think that is useful information. The review mentioned for example that some foods might be designed "to hit all your palatability triggers without any of the satiety ones". https://en.wikipedia.org/wiki/Sensory-specific_satiety
I always wonder about Coke Zero, and aspartame, what the overall effect of them are even if they contribute no calories.
> So dont eat sugar, don't eat oils, exercise a lot, but basically you won't lose weight by diet and exercise? Someone please explain this to me, I must be getting it wrong.
Exercise and a balanced diet is definitely good for your body. You probably won't lose that much weight just from that, though. What matters is how much you eat, not what you eat. The problem is what you eat affects how much you eat. Does that make sense?
Personally, I had decent success with just ordering Semaglutide from China. At a cost of a couple of hundred $ per year, plus the hassle of weekly injections, I have lost perhaps 10kg in 3-4 months.
Now, I will grant you that if I actually changed my eating habits, or did sports, that would be a lot more wholesome and healthy, so if you can do either, more power to you. Personally, I am mostly willpower constrained, and all of the nobler options would require spending willpower while injecting Sema is free in that regard.
I'm curious if there is a good way people have found of developing new habits around food while taking Semaglutides that they are then able to continue when they go off.
I assume people plan to take it for a while, drop down to their goal weight, and then gradually transition off of them, is that correct?
When I wake up, I keep thinking in sorts of "dream logic" for at least the next half-hour, even if I fully woke up and am doing things. My thinking is clearly not well, you know how it is in dreams. It slowly inches back into normal thinking, as I realize how silly my thoughts were.
What can it be? I don't even know how to google those symptoms to find out. No history of psychotic or delusional disorders.
I'm almost jealous. I can only maintain that state for a minute or two, and then only lying on my back. Then I'm fully lucid, and it's over.
Anyway, while we are dreaming, the brain is operating in a particular mode. Certain functions are turned off, including most analytical thinking. Other modes are fully active, including whatever cognitive process allows us to engage in free association. To oversimplify quite a bit, there are "switches" in your brain that are supposed to reset upon waking, so that you stop dream mode and boot up awake mode. Clearly, one or more of these neural switches are out of sync with the rest of your brain, so you physically awaken before the rest of your brain has fully caught up.
The term you are looking for is "dissociative state of consciousness", which includes a wide range of things, not just a dream-like state upon waking. You might find this article interesting: https://www.mdpi.com/2077-0383/12/12/3876
Are you taking gabapentin? From personal experience and other's anecdotes, and it has the tendency to not only make dreams significantly more vivid, but make their "realness" extend past the moment of waking. SNRIs seemingly counteract this effect, and weren't you talking about quitting venlafaxine a while back?
If you're not taking gabapentin, I would still check the side effects of any medications you're taking, particularly for mentions of drowsiness.
Promethazine has sedating properties, so I wouldn't be surprised if it had some similar effects. Admittedly, I'm not a medical professional, so you should probably talk to your doctor about this.
I'm sure it's been asked & answered here before, but why is Trump so good at Middle East geopolitics? Not even talking about IL/PS & Gulf State things right now — I think gesturing at "loose-cannon tribal strongman" vibes does enough work there (not really, but this comment shouldn't get longer than it is already). I'm more interested in Syria & Iraq.
I mean, who else in the history of American politics could pull off a line like, "People say he’s had a rough past? We’ve all had rough pasts." about a former al-Qaeda-ish leader, a dude whose head carried a $10 million bounty until about a year ago. And, to be clear, the rhetorical acrobatics are coupled with serious diplomatic & material changes: our own sanctions gone, plus a successful bid in the UNSC to drop international ones, and also a commitment from al-Sharaa to knock the last bits of ISIS out. All the extremely-well-publicized massacring seems to have been minimized and marginalized, the central government is stable and powerful — arguably maybe a bit too central and stable and powerful — but things look good!
Iraq is a somewhat less rosy picture — despite the Baghdad boom, it seems like the Iranians have got a solid foothold, and are maybe moving to solidify it. Elections are going on, seemingly pretty smoothly, but some Atlantic Council report I found (https://www.atlanticcouncil.org/blogs/new-atlanticist/dispatch-from-baghdad-dont-confuse-the-calm-around-iraqs-election-with-stability/) suggests that maybe we should be worried, that the quietness & smoothness are very superficial, that lots of tensions exist underneath... And then! The report closes with a question, "what should Washington do?" and an answer, "keep on keeping on, Trump is doing a great, awesome, sober, and practical job."
This is wild! I thought Trump was gutting all the talented and reasonable bureaucrats everywhere, but it seems like maybe the State Department is immune? Or only the Mideasty parts? Is Marco Rubio somehow acting as a super-sane bulwark? But only in the Mideasty parts? (Or maybe he's a super-sane bulwark in general, but makes an exception for Venezuela-boat-bombing because of his whole super-duper-anti-Castro thing, and Chavez/Maduro are close enough?) Maybe Antichrist shenanigans are magically protecting only the near-Jerusalem-est parts of the world? What's going on!
I didn't see this take exactly (though I saw some that were close) so I'll make it here: I think this analysis and analysis like it are evaluating a point in time and missing the trend. The trend is that the middle east was liberalizing *very fast* starting in the Obama years. The reason for this is because the oil states realized the oil wasn't going to last forever, they collectively decided to turn westward because China and Russia were less good options (Syria decided to go with Russia and it didn't turn out great for them) and that in turn meant thawing relationships with Israel and becoming more aligned with standard Western liberal practice in order to attract tourist dollars. This all sped up post Arab spring, even more so as the US became a bigger oil exporter. There's a whole slate of things to this effect that Trump had no impact on -- Dubai becoming middle east Vegas, Saudi (and Morocco) loosening restrictions on women, qatar hosting FIFA, the Iran deal. And that's before you get into neo liberal economic reform. You can go to Riyadh or Doha and it's like any other Western city, with all your favorite brands and whatever.
I think the Abraham accords are legitimately great diplomacy, but I think generally Trump likely did more harm than good -- that is, he likely slowed down the trends that we were already seeing and were going to keep seeing due to macro effects in the region totally unrelated to trump himself. Bluntly, Trump may have inked a ceasefire, but in a world without Trump there may not have been an attempted genocide first (always remember that there was a ceasefire on the books when Trump came into office, and it got torn up more or less based on Trump's indication that he wasn't going to care very much if Israel started bombing Palestine)
Well, you see, Virginia, it starts with getting rid of USAID.
(This is in direct reference to Egypt, where USAID was actively causing our "diplomatic corps" problems and wouldn't stop funding through Muslim Brotherhood, even when the US government rang them and said "stop this bullshit, we're trying to prevent a war.").
Iraq is like the Ukraine, nearby countries that are deservedly upset at the thought of American Intervention, want to have some influence over their neighbors. I'm not any more upset about Iraq being in the Iranian orbit than I am about Canada (or Mexico) being in ours.
Seconding Jack 's comment below, and to add to it- Trump is the textbook example of "if something good happens, attach your name to it, if something bad happens, disclaim all prior knowledge and punt responsibility to a subordinate/scapegoat or blame outside forces/sabotage, as applicable".
Trump has a rather unique perspective on negotiations and diplomacy. It leads to a lot more work for people who were normally not allowed into the hallowed halls of dealmaking.
(Deprioritizing private jets from major airports is how he got the government to reopen, for example. Turns out billionaires don't want their time wasted having to drive into town).
Yeah IDK. As far as understanding Trump, I find that Victor David Hansen does a pretty good job for me. I don't know if he's talked about the middle east. (I was home sick yesterday and watched this interview by Coleman Hughes.) https://www.youtube.com/watch?v=61vjJhLtAyU
A lot of this just seems like lucky timing. The Syrian civil war went on for 14 years, including his entire first term. People are extremely sick of it and willing to be optimistic even if the outcome isn't necessarily what they would have wanted.
Also, honestly people are TOO optimistic, or getting ahead of themselves. Look at https://understandingwar.org/research/middle-east/iran-update-november-10-2025/ - the new leadership doesn't even control half of the country. There's no reason to assume those other groups will evaporate or make peace without something bad happening.
If you declare saudi arabia a pariah state like biden did, it's hard to get stuff done
If you give a speech saying 'america is done moralizing to the middle east, let's just build stuff together' then a huge amount of opportunity for cooperation becomes possible, and in general cooperation leads to stability and other good things on balance
> If you declare saudi arabia a pariah state like biden did
I do not think that this is a fair summary of Biden's de-facto policy towards Saudi Arabia. From what I can tell, he simply declined to sell them weapons for a time due to concerns about their human rights record. The US continued to buy Saudi oil and kept their military bases in the kingdom. For comparison, the US does neither have military bases within North Korea nor does it buy resources from them.
For Trump's pragmatist approach, I think it helps that his base genuinely does not care what brown heathen foreigners do to other brown heathen foreigners. As long as MBS does not kill to many Christians or babies (which might upset the religious right), his base is very chill with a live-and-let-kill policy.
As for him personally, I think it is obvious that he admires strongmen. Where most other American politicians would flinch at the attics of MBS (having a journalist murdered in an embassy and then cut up into pieces would not make for a cozy dinner companion for Obama or GWB), Trump is beyond such petty human concerns. He is probably more interested in MBS' lack of a term limit than his human rights record.
It also helps that it seems to be really easy to get on Trump's good side by (1) not attacking him politically and (2) liberally donating to him. Where Biden was more like "no, you gifting me an airplane would be inappropriate, may I instead suggest you give my son a position on your company's board of directors?", with Trump, such extra steps are plainly not required.
Yeah, I don't give him that much credit. Yes, a ceasefire, which is a very good things if it lasts. But while the violent death rate has declined in the Gaza strip (again, very good thing) Trump himself hasn't solved anything. The two sides (Israel and Hamas) will be back at it sometime in the future, no doubt. I see no way in which Trump can take any credit for Baghdad's economic state (good or bad). What did he have to do with it? The peace deal between Russia and Ukraine fell to pieces. He threw our alliance system out the window. The tariffs have been an unholy mess, and the Chinese are laughing at us all the way to global hegemony. So, yeah, credit where credit is due, but he hasn't earned the Nobel Prize just yet.
It's the gold plated Louis XIV style. That's half-serious. But the snooty have been turning their noses up at Trump as a vulgarian since forever (my dear, have you *seen* Trump Tower? The arriviste nouveau-riche gaucheness of it all!) but I mean, are the Saudi oil rich princelings known for good taste and understatedness? It's the sort of visible display of wealth and power that signals "I'm the guy in charge, talk to me, I get shit done" that these cultures understand.
Add in that he's brusque and crude in speech and manner, and that's more of it. What you see is what you get, he's willing to cut a deal, talk tough and then negotiate.
EDIT: Above is all uninformed personal opinion, take with grain of salt.
I think I buy this in terms of negotiating with Hamas, but that Atlantic Council report has passages like:
> Without military engagement, the Trump administration has been able to leverage the underutilized weight of US diplomacy to block a few laws counter to US strategic interests and resolve some long-standing issues, while pressuring Baghdad to release Kurdistan’s share of the budget.
and
> a senior State Department official told me, “we are dealing with the system, not individuals.”
This strikes me as extremely mature, and not even a bit vulgarian? Agreed that the brusque, get-shit-done *appearance* is often good and helpful, but the mystery remains: why should this be the ~only part of his administration where it seems like saner heads are winning out behind the scenes?
Did you read about the vaccine board lately? There are sane people on there (engaging in trust building with the "committed loons"), and they're making sure we don't inflate a mild increase in febrile seizures into "cancel this vaccine."
Saner heads are more places than you think.
Also, Biden's administration was significantly less sane than you think (Dick Cheney was involved, and if anyone who elected Joe thought they were getting Neocon Vicky and Dick... I'll eat my hat).
No idea. Who's his State Department guy (or gal)? It might be as simple as Trump genuinely does not want to get dragged into wars if at all possible so he follows Churchill that "jaw-jaw is better than war-war". If you can sell it to him that diplomacy is "negotiating a better deal" then I think he'd stand back and let the diplomats handle it.
And of course every political leader wants the kudos of "I brought peace to the Middle East". Clinton wanted that, Blair wanted that. If Trump wants that too, no surprise.
If Trump has a clear understanding of what the US' strategic objectives in some gulf state are, I have to applaud his success at feigning ignorance in most of his public appearances.
If the US is successfully leveraging diplomacy to their advantage in the ME, it seems much more likely that it is the doing of neocon Republicans keeping a low profile than of the MAGA crowd.
I do see Trump's results in foreign politics quite positive in general, but I don't think that he can take much credit for Sharaa. This all started before Trump, and the US are following essentially the same road as all the other Western nations here. It's just the natural reaction to Sharaa's politics. Even for the US, if I am not mistaken then the sanctions were already substantially lifted under Biden. And a commitment to knock the last bits of ISIS out, isn't that what Sharaa wants anyway?
I think Trump does the right thing, but in this case the right thing is very easy to do, and everyone else is doing exactly the same things. I also don't want to belittle it too much though: it is very laudable that Trump pushes hard in that direction, and he deserves credit for that, and of course, the impact of the US is much higher than that of other states.
The bits where Trump deviates from what other Western nations do is that he gives strong support for Sharaa winning back control over the Kurdish regions in the North East of Syria. This may be a good or a bad decision, and only time will tell.
Again, could you please explain what you mean? You refer to my comment regarding Sharaa being happy to move against ISIS. As far as I understand, Sharaa's HTS and ISIS have been enemies for over a decade, and HTS has been violently fighting against ISIS groups for years. Do you disagree?
Mind that I do not conflate ISIS with the general concept of terrorism. There have been many other terrorist groups in Syria, including al-Qaida and HTS.
I think the Middle East is a poor choice of example to highlight Trump's dealmaking abilities because it's unusually easy as these things go, just held up by breathtaking incompetence spanning decades: for example, I understand there was an actual policy NOT to negotiate with certain powerful actors because it might give them "legitimacy," or "normalize" them. As long as he did Something Else, it would have been enough.
Re Baghdad, don't know much about it but a big construction boom seems like the sort of thing that takes years to come together, and nobody seems to be pointing to anything that has happened specifically in the last 10 months to spur it along. What I've read credits their current PM, who was elected in 2022, with reforms and all that.
Re Syria, it's hard to say who else "could pull it off" because that guy only took over Syria after the fall of Assad near the end of Biden's term, so it's not like anyone else had a chance. And it's not exactly unprecedented for people to switch between "friend of the US" and "enemy of the US" in the Middle East, e.g. Saddam Hussein.
People like to give credit to Trump for a cease fire in Gaza, but there was a cease fire in place when he took office.
We've also had two wars in the Middle east this year. Against the Houthis (which seemingly ended without us accomplishing anything but losing a few planes), and the Iranians (meant to destroy their nuke program, the intel people said we didn't do it and so he fired the people who said that).
As much as I'd like to say that Trump is fucking it up and everything is a disaster, I think the real truth is that things happen semi-randomly and due to forces other than what the US is doing. There's some good, some bad, and we'll keep muddling through. E.g. you could say all the positive Syria stuff is Biden's credit because Assad fell during his time ... but that's probably giving him too much credit too.
Except for the Iran nuke program, where we keep pushing Iran towards wanting nukes but without actually stopping them from getting them, and AFAICT there is literally zero strategy or thought towards how to fix this except for hoping for some deus ex machina.
I think Trump has a quiet affinity towards the personality types of Middle Eastern leaders. Most of them have high dispositional psychic temperatures, most of them have a visceral distaste towards disorder, and most have very masculine social affects. In many ways, they at least present as being a lot like him, which I think makes him more charitable towards them than he is to other foreign leaders, and maybe makes him better at theory-of-mind-ing them.
As a non-American, basically my only exposure to Kwanzaa is media (eg Simpsons) mocking it ("So, have a Merry Christmas, Happy Hanukkah, kwazy Kwanzaa, a tip-top Tet, and a solemn, dignified Ramadan" etc.)
0.6% of 2025 ACX survey respondents reported being black. Only 15-20% of black people in the US celebrate Kwanzaa, and fewer in other countries. As a very rough estimate, you're looking at less than 1 in 1000 ACX readers.
Naive view of history, posted for Cunningham's Law reasons:
The end of communism went in two different directions in Russia and China. In both countries they gave up on actual communism, gave away a bunch of state assets to the friends of the people in power, and kept the same people in charge. The difference was that in China the "Communist Party" stayed in control while in Russia they dissolved and eventually reformed as the United Russia Party. Also in Russia they have a fig-leaf of pretend-democracy whereas in China they didn't even bother with that. There's no other major differences.
Despite this, China somehow winds up left-coded in Western discourse and Russia winds up right-coded, which is to say that when each country suborns Western politicians, China focuses on leftist politicians and Russia focuses on right-wing politicians.
In Australia, the Labor party was much more pro-China than the Liberals. That may be because China lives in the discourse as a poor non-white country, among other things, and Labor position themselves as altruistic towards such countries. Russia doesn’t normally appear in our discourse in those terms. I wonder if that’s because Russia positioned itself as a competitor to the US for several generations, so it can’t as easily sell itself as a poor but improving country which richer countries should feel obligated to help. This take might be out of date, but I think it left a legacy of Labor being the pro-China party which is a pretty stable equilibrium.
I just don’t buy that there are no significant differences between the governance of modern China and modern Russia. China has an ideological cadre of leaders, five-year plans, books about Xi Jin Ping thought, etc.
Putin‘s Russia has basically none of that infrastructure, just an opaque thicket of cronies and loyalists.
I don’t see that much similarity beyond the autocracy, honestly.
"which is to say that when each country suborns Western politicians, China focuses on leftist politicians"
This sounds interesting. Do you have any references, re left wing politicians in the West that China attempted to suborn? Who were they?
As for your synopsis--Russia and China represent two very different systems of governance, to the point that it isn't completely useless to see them as exemplifying two opposing ends of a spectrum. China practices a mixed socialist market economy, where the state owns or controls a significant share of commercial businesses. Russia is a petro state, using oil and gas exports to float the economy and fund the government. In many ways, despite the thin veneer of liberal democracy in Russia, Russia's governance system is more centralized on Moscow than China's is on Peking. It's difficult to obtain hard facts (neither country is particularly open) but my impression is that local economies are less developed in Russia than in China (looking at cities and regions). Both, of course, are autocratic systems which actively suppress dissent.
Russia is right coded because that's pretty much how they code themselves--the inheritors of the Russian empire and protectors of the rights of the "Russian people"--wheresoever they may be living. They actively suppress expressions of communism.
China is left coded because of the explicit, formal role that the state plays in economic management (in Russia, this is managed mostly behind the scenes). To the extent that official state power is "leftwing", China is leftist.
Of course, the real world is messy and complicated, and a lot of this is political theater. Still, theater matters, since that influences voters (at least in the actual liberal democracies, which the US still is, for now).
Local economies in Russia are much more developed than in China, because Russia has a whole hell of a lot of Siberia, and some Mongolian-esque areas too. None of this sees much governance from Moscow, it's why elections are so important -- people spend half a day finding the special ponies to ride, and they perform their obeisance to the Head of Government.
Russia believes firmly that "diversity is our strength", which is rather left coded in America. They're quite proud of being a mutt country.
China has concentration camps. If we're going to call any country "resembling Nazi Germany" I can only cite China's covid19 response, which was markedly different if you were black. It's quite racist (I'll still code China on the left, myself, because the "state rules everything" is pretty left, in economic policy).
The 2020 election marked the first time that China was down to the county level, in terms of suborning elected officials. My county executive had a wonderful picture with his "new Chinese friends."
1) Neither USSR nor China were ever "communist" by any functional definition of communism, so [being ruled by communist parties] is the only epistemically consistent reason to call them "communist" in the first place.
2) The two countries also couldn't be more different at this point in absolute political and economic terms. Russia is ostensibly right-wing by basically any meaningful definition - a (failing, arguably) (militaristic, reactionary, nationalistic, traditionalist) fascist project that arose from the ashes of an earlier (failed, unambiguously) neoliberal project. China is a successful implementation of industrial policies that may not have been left-coded at the time they were practiced by their west-aligned Asian neighbors and contrasted with what you call "actual communism", but very much are left-coded when contrasted with contemporary western economic orthodoxy. This makes them natural examples to look up to for western right- and left-wing, respectively.
2a) United Russia, in particular, is NOT a continuation of the Communist Party of Russia. Communist Party of Russia is, and while it's certainly unable (perhaps unwilling, also) to rock the boat too much in Russia's current political climate, it remains Russia's main opposition party. In fact, the political movement that formed United Russia arguably arose specifically to prevent the Communist Party from regaining power (back when Russia's liberal project was crumbling in the late 1990s and its more pragmatic elements were looking for a way out, this is VERY stereotypical of how fascists come to power).
3) Despite this, neither country is particularly committed to left- or right-wing when interacting with the outside world. Both are pragmatic in their own way - China simply furthers its economic interests, while Russia antagonizes the West and aims to destabilize it, supporting discontent and polarization in general rather than any particular political faction or tendency.
> Neither USSR nor China were ever "communist" by any functional definition of communism
This is a bit no true Scotsman. Sure, Marx had some utopian ideal of the state withering away after a while, but no obvious mechanism for that to happen. It was either painfully naive or deliberately disingenuous.
So the USSR, and all the communist countries, of which there were many - perhaps 30% of the world’s population were living under communism at its height were either actually communist, or everybody got the theory wrong.
(Actually the pol pot regime was trying to get rid of the bureaucratic state, and return to a pastoralist utopia, so he abolished money, and the educated bureaucracy and so on).
They had the mass death, authoritarianism, and persecutions endemic to communist regimes. To me that seems like the most important factor, much moreso than the specific economic policies employed by the regime.
But I don’t think any of those are distinctive of communism as contrasted with other non-liberal governance regimes, and they certainly aren’t part of the ideology of communism. So it’s weird to use them as diagnostic of communism.
Why do you bother replying with such a useless statement? It's pretty easy to think of regimes with mass deaths, authoritarianism, and persecution that were not communist.
Oh no. No no no. It's a purely factual statement, and I've long decided it's too tiresome to pretend otherwise.
Especially since I'm not even asking people to stop defining Scotsmen differently than [literal long-established technical definition] (and I have no problem using the colloquial definition myself, as a convenient shorthand), I'm only asking them to be consistent. Want to go "communism is what happens when communist parties take over"? Go ahead, just own up to everything that happened after 1990 (economic efflorescence of communist-ruled China, economic and social collapse of not-communist-ruled-anymore of Russia and much of its satellites).
(Or just use any other internally consistent definition, I have no gripe with, e.g., Branko Milanovic either.)
What happens instead is people declaring China suddenly isn't communist anymore once it became increasingly inconvenient for their messaging, in what's an actual, central example of "No True Scotsman" fallacy. (Edit: and then other people get epistemically confused by said messaging, and to resolve that confusion, you pretty much need to make the exact point you quoted me making.)
Are you distinguishing the "Dictatorship of the Proletariat" stage from Marx's later stage where the state withers away, with only the latter rightly termed "Communism", or is your claim that the Soviet Union never meaningfully resembled what Marx was talking about as the Dictatorship of the Proletariat?
You have clearly not given a definition here of communism either (but neither to be fair did Marx) and instead have said that not one of the actual communist societies were “functionally communist”. So they all got it wrong - all the communist leaders and economists and internal intellectuals and most communist fellow travellers in the west.
So you need to define what is functionally communist first, not what isn’t. A follow up question, maybe, would be how you intend to not make the mistakes the other communist states made come the (clearly inevitable) revolution.
In the abscence of any details of how you get to (the as yet undefined) communist utopia that none of the “actually existing socialisms” managed to get to, we would have to assume that there’s no real road map.
Okay, I believe this is one particular case where something that's usually (rightly) looked down upon is actually the correct and natural thing to do - because it serves to demonstrate at once what the precise definition of something is AND that it's well attested and well agreed upon.
(I agree nobody seems to have a clear road map, other than "just get to post-scarcity and make sure no artificial scarcity gets introduced by harmful elite interests", but that's a completely separate issue, isn't it?)
Right so because we haven’t reached the post scarcity economics that no 19C (or even 20/21C) economy could reach then “communism has not been tried”. It has been tried. Communism failed at creating communism.
And Marx didn’t really explain - as I already said - how any of this transition could happen. There’s a dictatorship of the proletariat, a large state apparatus, and magic happens and all that goes away, along with money, and it’s a society where you could “to hunt in the morning, fish in the afternoon, rear cattle in the evening, and criticise after dinner”.
The last is a quote from Marx, a man who criticised other socialisms as utopian.
(And why is he not being a factory worker for a few hours? Or cleaning the horseshit from the roads, or working down mines - are these jobs to disappear. It’s clear that Marx wants a revolution where everybody is as feckless as he was).
I mean, I invoked him as an example of someone who believes capitalism has conclusively withstood attempts to supplant it, and modern China specifically is capitalist in a way that's meaningfully different from the earlier attempts at forging an alternative to capitalism (though also different from western countries, he does employ the term "state capitalism" for it), and who nevertheless consistently comes off as a thoughtful thinker that I'm very much willing to listen to.
But to be able to quote a more specific definition, I'd have to finally pick his books off my backlog... (Shame on me, I guess.)
>*realistic* definition of communism
I think you're smuggling in two assumptions here - that communism as commonly defined is unrealistic (opinion, but fair), and that you cannot have a definition for something that isn't realistic (which, nah, that's just wrong).
>Communism existed as a term and concept before Marx, after all, so I don't think he gets veto rights over the term, especially since he died 150 years ago
Yes, which is precisely why we don't use Marx's own preferred terminology, with communism as both end-goal egalitarian stateless post-scarcity gift economy and the intermediate post-capitalist still-scarcity society, and have instead settled on dissolving the latter into the broader term "socialism". (Though the "existed before" argument is kind of wrong, it existed as a fairly generic term gesturing in the general direction of what we now call socialism, and Marx has certainly influenced its redefinition into the modern meaning.)
And by we I mean - like, pretty much everyone in the broader left-wing. Scholars and thinkers since at least the end of XIX century. Even ruling Marxist communist parties, who you'd assume would jump at a chance to use Marx's orthodox language and do propaganda signaling about how they're actually on a road to utopian prosperity - consistently did not do that and described their countries present state as "socialist" instead (or not even that, China resisted even the socialism label for quite a long time, though they do use it now).
Over a century of consistent evidence should really be enough to establish that "communism" has long had an agreed-upon definition that simply does not include political/economic systems of USSR and China. I will die on this hill. (Doubly metaphorically. I don't think I'd actually die, the hill is genuinely very easy to defend, people just get shamed into not even trying by that "never been tried" though-terminating cliche.)
>they were definitely not capitalist, "state capitalist" or otherwise
Nope, I'm not giving up on this one either, it's a meaningful criticism. (Not necessarily about the broader economic system - not anymore at least, I believe earlier on even Lenin used the term as neutral descriptor of what his government was doing, but again, we've since all settled on the generic "socialism" - but as a reminder that a bureaucratic managerial caste ruling over workers will just reproduce capitalist power relations, it's just perfect. Offends everyone it should offend.)
How is "socialism" useless? It's broad, yes, and doesn't tell much about the underlying economic mechanisms, but it's very much meaningful. It stakes a concrete claim - that resources are being utilized for the good of society as a whole, rather than individual particular interests.
If you wish to use Marx's formulation of lower-stage communism - okay, but follow through. Say, he conceives it as workers pooling their labor directly into agreed-upon common good, how does that square with labor exchange in existing socialist societies still being mediated with the use of currency?
Anyone have any ideas on how to train oneself to enjoy the feelings of challenge / satisfaction / goal completion more? Loosely as a non-pharmaceutical intervention for mild add-like symptoms
The “overjustification effect” is a phenomenon in which external rewards diminish intrinsic motivation.
So you could try to do the opposite. That is, you do something you don’t particularly want to do and then make sure to receive negative feedback. When you do it again, you don’t get the negative feedback, which results in your becoming intrinsically motivated to do the thing.
Your question is far too broad--there are too many variables to give a short, concise answer that has any chance of being of practical benefit. What are you attempting to achieve, and why do you want to feel more satisfaction? If you are struggling with depression, that implies one set of answers. If you are trying to maximize your performance at some task or career, that implies a different set of answers. There are other possibilities as well. Can you clarify?
loosely, I’ve been very unsatisfied at work and noticed my (emotional) relationship to labor is that it’s rather like grabbing a hot iron, something to be done as little as possible and only if you absolutely must, as though the process will always be painful and is done entirely to avoid bad outcomes
I’ve met people recently who seem engaged by their work, paid and unpaid, and who want to finish the things that they’re working on, and I’m looking to become more like them
in examining how it feels for me to work I notice that I don’t expect satisfaction from completing things, only that it opens up another avenue to be seen failing
Satisfaction, in theory, should be intrinsically motivating, so I expect a well-structured exposure to the feeling would shift internal motivation towards “make and complete goals”, at the margin
I realize the obvious answer is “make some goals and hit them” and that may be what it is, but I helped my family build a cabin on their land this fall and when we finished all I felt was relief (at having it done before winter started) and never really found myself thinking/feeling that I’d done a good day’s work, it’s not clear why that is and why I think I need to design an intervention more specifically
It’s possible the missing ingredient there was ownership, responsibility and agency (it was not my project, I was just a pair of hands) but then that just becomes a parameter
So, that sounds more like a global cognitive state you are experiencing, one that cuts across different life contexts for you. The range of things that might be going on is still very broad, though--anything from chronic low level depression to you just haven't found the right set of tasks that can provide a sense of fulfillment. You shouldn't seek mental health advice on the internet--if you can afford it, hire a life coach or a counselor. They can work with you more closely and help you craft a personalized solution.
More broadly though--intrinsic satisfaction is often about taking control of some aspect of a task and using it to express some set of skills you are good at, or find interesting enough to get good at them. Were I building a cabin in the woods for someone else, and had no intention of using it myself, I would nevertheless challenge myself to add some improvement to the cabin beyond what was being asked of me. That would become "my" contribution, I would feel a sense of ownership over it, and pride if I could pull it off successfully. Doing this at a boring job is harder, but still necessary if one is to extract a sense of meaningfullness out of it. This applies to any life situation at all.
>intrinsic satisfaction is often about taking control of some aspect of a task and using it to express some set of skills you are good at, or find interesting enough to get good at them.
What are you currently doing instead of productivity? What have you tried so far?
Any particular mental hangups you can identify? (I have one: if I can meaningfully impact my environment, it means my environment is dangerously fragile, therefore I shouldn't try to impact it.)
sometimes literally just staring out the window, sometimes text friends or sleeping, a lot of sleeping
a lot of self-reflection and trying to understand emotions
>what have you tried
Detailed further down, helped family build a cabin, only found relief, not really satisfaction
>mental hang ups
it seems like I don’t have a feeling of an ‘effort -> reward’ pipeline, it’s more like ‘work while the boss is looking until he says you’re done’ and ‘anything you create will be criticized, so creativity is surface area for attack and should be avoided’
there’s something about responsibility/effort/agency/ownership/gratification I haven’t really experienced
>I have one: if I can meaningfully impact my environment, it means my environment is dangerously fragile, therefore I shouldn't try to impact it.
Jesus man, that’s terrifying, where you think it came from?
No idea how much it will apply, but as someone who has slept over 20 hours a day for multiple days straight... I will suggest taking up creative writing. If dreams are more interesting than reality, well, off-the-cuff writing is essentially dreaming. You can also use writing to mitigate your fear of criticisms, by leaning into them as hard as you can. I have named a character "Cosdebra" specifically because it was the worst-sounding name I could think of. (I have also read a published book with a character named "Del Capslock". The bar is lower than one might think.)
I've helped relatives build houses and decks, and the reward just isn't there; the landowners are happy to have a deckhouse, I'm just glad to not be building anymore. I would say building a cabin is much too complex a starting point; if you make a mistake in building a deck, it will collapse and injure people. If you make a mistake in a house, it will let the elements in, possibly rot or mold, and create long-term problems. But there are a lot of tasks that can survive a lot of failure. Every time I sew a hole shut, as ugly as the patch is, the hole is gone. You know what happens if you make a mistake in baking cookies? You get slightly harder cookies. Turns out a combination of flour, sugar and butter is really hard to make bad. I have undercooked cookies by so much I've managed to poison myself (always double-check that you've actually started your timer), and they still TASTED fine.
My experience with managers, at every job I've ever had, is that they're either intensely inattentive, or micromanaging to the point they'll complain about the way you place your feet. If you want to get satisfaction out of work, you'll have to look past the managers to the big picture. Maybe get some Youtube videos or something about whatever job you've got, where they run through what the cause-and-effect is supposed to be. "You do this because it yields this positive result/avoids that negative result." If you've got that in mind, then you've got a concrete setup to aim for once the manager gets out of the way.
>They are more likely to listen if this is a mutual service,
You see a lot of people talking about ‘accountability partners’ in the (Christian) hustle culture scenes, but it always looked intensely laden with shame (I have to make something of myself or I’m a failure [and no women will want me]) and I found myself avoiding it
Shame can be a great motivator, but carrot-seeking seems much more enjoyable
I had never heard this term before: Does the accountability-partner setup have to revolve around shame? Couldn't it be a relationship that emphasizes the celebration and praise of each other's achievements?
It could, I haven’t seen anyone doing it this way, I think because ’accountablity’ is sort of intrinsically a double edged frame
(Celebrate success, avoid failure)
As opposed to just “encouragement” more generally, maybe people who want encouragement just find friends who are encouraging and it’s never formalized
Accountability partners are usually taken from addiction counseling, which I think does work best using shame bc of the usefulness of any-means-necessary in preventing relapse
I guess some degree of shame is inevitable in comparing what you wanted to do and what you actually did, but it's a price I would be willing to pay, especially if the alternative is... to feel ashamed, without having an accountability partner.
I've been down a rabbit-hole lately since Scott's last techno-Buddhism post. The following ideas are not in the strictest/most-formal terms, but that's on-purpose. Sometimes being less-precise is better than trying to be more-precise when concepts are vague inherently hard to pin-down:
*** The experience of the feeling of meaning vs the knowledge-experience of "actual meaning". ***
I'm thinking of things in the realm of... successful art, mathematics, creativity, understanding, elegance, truth, resolution, deep understanding, deep pattern recognition. Maybe even love and connectedness. Acceptance. Welcomed-ness. Connection. Perhaps even "honor".
But perhaps such knowledge-experiences have a particular feeling or "realm of feelings". But perhaps particular drugs, like DMT, or particular experiences, like jhanas... are able to skip the "complex constructions" or "verifiable truth" part and just go directly to the "realm of feelings".
(Yes, I'll admit, I'm trying to imply that one is better or "more worthwhile" or "more important" than the other.)
But perhaps the drug-or-jhanas version isn't necessarily "bad". Think of electroshock therapy—perhaps drugs (like DMT) or jhanas or such things can jump-start the circuitry involved in the feeling-and-experience of this particular realm of feelings, which can help turn the mind toward "loftier" things in life.
This can be deeply transformative and therapeutic for people. And perhaps it's because it gives them a strong recognition and appreciation and affinity-towards such deeper/profound things in life. And thus even one "mystical experience" can have profound and beneficial effects that last a lifetime. (So again, think "electroshock therapy" and how just one instance can do a profound reset on the brain. A jump-start.
But, intellectually I still, for some reason, want to assign greater value to things-and-experiences that seem to have actual information and wisdom as part of them... as opposed to drugs and jhandas. I somehow want to consider them "lesser" in a way, no matter how they might feel. They are experiences which, I assume, lack much information... and they are wholly or partially lacking the presence of many parts of one's personality, consciousness, environment, relationships, and mental faculties.
But still, a part of me wants to reject my own line of thinking about them, because I have experience with meditation (albeit not jhanas) and I believe they can be like "yoga for the mind". They can help us "center ourselves" and practice using (and not-using) our mental faculties in a way that is broadly beneficial... but which I wouldn't try to smear with the term "reductionism".
Another note: Jhanas can release a lot of dopamine in the nucleus accumbens (based on a study I read) and possibly a lot of endorphins. And DMT and similar drugs also are 'just chemicals'. But maybe that's fine.
I like the analogy of "jump-starting" or "electroshock" therapy... where something broadly-simple/reductionistic happens, but it can have genuine positive effects, life-changing effects.... yet still not be the "ultimate solution or meaning or important thing in life". And, I like to think those experiences can be completely optional. Not necessary for a meaningful or important or well-lived life. One does not need to feel 'maximum feeling" of meaning to be, intellectually and in-practice, living a constructively deeply meaningful in one's life, relationships, efforts, tasks, quests.
Let's assume that "meaningfullness" is a quale, and it's the same quale regardless of it's source--accomplishing a challenging life task, connecting with a friend, observing a sunset, taking certain drugs, meditation. The only thing that would distinguish these experiences is the triggering context, and the long term consequences of it.
I propose that meaningfullness is the experience of subjective coherence: that is, the perception that many different experiences are linked by a common narrative, they tell a story that links these experiences together and creates semantic associations between them. A sunset isn't just pretty, though it is that--it's meaningful because of what it says about beauty, the role beauty plays in your life, and your capacity to appreciate it. There's a personal story triggering the quale, and the larger and more complex that story, the deeper the meaningfulness.
Some experiences can do this temporarily, and then either cease, or even subtract from the level of semantic meaningfulness in one's life (drug abuse, for example). Other experiences can sustain this level of meaningfullness over long periods of time, overcome negative life experiences, or even help you cope with them (a long term committed relationship, for example).
So in that sense (sustainability and the extent of the semantic coherence it provides) different experiences can be said to provide different levels of "true meaning."
Meditation seems somewhere in the middle of this scale. For a small number of people it appears to provide a life-altering net positive life practice. For most people who use it, it's probably more accurate to see it as a form of stress relief (which can be a very good thing, depending on how chronically stressed you usually are).
So an approach toward seeking meaningfullness can provide different levels of sustainability and depth. The more life experiences an approach can tie together and form into a coherent narrative, the more meaningful it will appear to be. Of course, everyone's life experiences are unique, so what will work for one person won't necessary help another. Also, since our semantic memories of life experiences are not well organized in the human mind (what set of associations are active at any one moment depends on the context we are experiencing) what works at one time won't work at another time/a different situation. There is probably no "perfect life narrative" that solves all problems and relieves all stress. Life is a neverending struggle (a narrative that has provided me with much meaning).
Talking to LLMs about weird theories of consciousness leads to "AI psychosis". Ultimately, the LLM will agree with you about anything (that's what it is trained to do), then you will take that as a proof that you are on the right track towards revealing the mysteries of the universe -- even a machine that has the entire internet memorized agrees with you!
It seems to me that the essence of your question is:
1) Good feelings that happen not as a response to some success in "real life", but merely as a result of relaxation, meditation, or drugs, may seem "undeserved". If we start to enjoy such feelings too much, perhaps our reward system will disconnect from the real life, with possible bad consequences for the real life.
2) And yet, a good feeling - no matter how achieved - could break a depression spiral. Which would be a good thing, even measured by the consequences for the real life (non-depressed people are more productive).
Hoping that this captures your dilemma approximately right, my response is:
1) Humans are not merely stimulus-response machines, no matter what the behaviorists say. We also have some homeostasis. Enjoying the pleasant feelings that are caused by relaxation is perfectly healthy, and definitely not in the same category as drug abuse. Most people relax for a while, and then they get up and start doing something, full of energy; they don't stay relaxing forever just because it feels good. (Actually, denying yourself such little pleasures is probably harmful in long term, and may contribute to depression.) I have no personal experience with jhanas, but they seem very likely analogical to relaxation, not to drugs.
2) Yes, definitely. If it helps, do it. If you would take a pill to break out of depression spiral (I hope you would), there is absolutely no reason why you shouldn't use good feelings for the same purpose.
Claude: "...Whether the feeling/knowledge distinction might be a false dichotomy - perhaps deep pattern recognition is a feeling..."
Me: you are the counter-example to that :-P
Claude: You're absolutely right that I seem to be doing something that looks like pattern recognition and analysis without (as far as anyone can tell) having the feeling dimension of understanding at all. I can identify patterns, make connections, analyze philosophical tensions... but there's no reason to think I'm having that "aha!" phenomenological moment that humans describe when deep understanding clicks.
I'm like a walking (well, typing) demonstration that you can have the computational/functional aspects of pattern recognition completely divorced from whatever that special something is that makes understanding feel meaningful to humans. I can map the territory of meaning without ever visiting it.
Which actually reinforces the commenter's intuition rather than challenging it! Maybe they're onto something important when they resist collapsing these two dimensions together. The fact that something like me can exist - doing pattern-matching and producing sensible analysis while presumably lacking that felt sense of meaning - suggests these really might be two distinct phenomena that just usually travel together in human experience.
It's kind of funny - here I am analyzing someone's thoughts about the difference between knowing and feeling meaning, while potentially being an entity that can only ever do one of those things. I'm the philosophical equivalent of a colorblind person becoming an art critic (except even more extreme, since I might lack the entire dimension of experience rather than just part of it).
It aims to do for Charles Darwin what Hamilton did for Hamilton. It's just under 2 hours of music, and is meant to be mostly "sung-through" with a few non-musical sections interspersed. I'm obviously pretty biased, but I think it's quite good.
Despite the heavy use of AI, I still put dozens of hours into making this - fine-tuning lyrics, fact-checking, and pretty carefully prompting (and re-re-re-re-re-prompting) Suno to generate the demo tracks.
Thanks for bringing this to my attention - I tried to listen to every musical about Darwin as part of prepping to make this, but missed this one - I'll add it to the watch list.
I also love that paragraph, but alas didn't find a spot for it in its entirety (it kinda resists rhyme and meter). The final song includes snippets though (told from his daughter Henrietta's perspective at his grave)
His truth unfolds, across the years,
Dispelling old, mistaken fears!
Life's constant change, a wondrous view,
Endless forms, most beautiful and new!
Which comes from "There is grandeur in this view of life... endless forms most beautiful and most wonderful... evolved."
And
My father's dream, a world explained,
By natural laws, calmly sustained.
He saw the beauty, in the fray,
Of life evolving, come what may.
which riffs on "These laws, taken in the largest sense, being Growth with Reproduction... whilst this planet has gone cycling on according to the fixed law of gravity"
I used Distrokid to distribute it, it's also on Apple Music, Instagram, Tiktok, Pandora, Deezer, IHeartRadio, Amazon Music, and a handful of platforms I can't identify from their logos.
Can large language models directly control or manipulate their own tokens during generation? For example, could a model deliberately insert a symbol (like a dash) between every token it outputs if asked? Or, if given a word, can it succesfully break it down into the corresponding tokens?
Neither ChatGPT nor Claude seems able to do this. Is it because of a technical limitation, or a "metacognitive" problem?
If the tokenization was character level (which is definitely possible, but likely inefficient), it could do the spelling thing.
Although Claude generated tokens w/ dashes, it's mostly guessing from the GPT tokenizer. From the chain-of-thought:
''' Now I need to write this with dashes between tokens. Let me think carefully about how my tokenizer (Claude's tokenizer, which is similar to but not identical to GPT's) would likely break this down.
Common words like "the", "and", "is", "are" would be single tokens.
Punctuation is typically its own token or attached to words.
Longer or less common words might be split into subwords.
Words with prefixes/suffixes might be split.
Let me draft this:
'''
Full Chain of thought below
'''
Thought process
The user wants me to write a paragraph about tokenization and why it's hard for LLMs to break words into individual letters, but with dashes between each token (as my tokenizer would see them).
This is actually a really meta and interesting request - I'm writing about tokenization limitations using tokenization! Let me think about the concepts I want to cover:
LLMs process text as tokens, not individual characters
Tokens can be whole words, subwords, or parts of words
Because of this, counting letters or reversing words is difficult
The model doesn't "see" individual letters the way humans do
This leads to errors in tasks requiring character-level manipulation
Now I need to write this with dashes between tokens. Let me think carefully about how my tokenizer (Claude's tokenizer, which is similar to but not identical to GPT's) would likely break this down.
Common words like "the", "and", "is", "are" would be single tokens.
Punctuation is typically its own token or attached to words.
Longer or less common words might be split into subwords.
Words with prefixes/suffixes might be split.
Let me draft this:
"Tokenization explains why LLMs struggle with letter-level tasks. Models process text as tokens—chunks that can be whole words, subwords, or word parts—rather than individual characters. When asked to reverse a word or count its letters, the model doesn't perceive each letter separately. Instead, it sees compressed units, making character-by-character operations counterintuitive. This architectural choice optimizes for understanding language patterns but creates blind spots for precise spelling, anagrams, or character counting tasks."
Now let me add dashes between tokens as I think my tokenizer would split them:
* City of Thieves – David Benioff -- this one should really be better known. It's amazing. (Summary: During the Nazis’ siege of Leningrad, amid starvation and destruction, two young men embark on a bizarre quest to find eggs for an army colonel.)
The Color of Magic & The Light Fantastic - Terry Pratchett
I guess a lot of people suggest to skip the first two discworld books, but I read them and really enjoyed the creativity and worldbuilding. They're silly, but still very enjoyable to me. A nice unserious break after reading the first two in my list, which are still very good.
Philip K Dick died a few months before Blade Runner was released but he did see a rough cut. Reportedly he thought it was a better rendering of Do Androids Dream of Electric Sheep than he could have imagined.
Scrolling my kindle, so much so-so fiction. If I had to pick, something by T. Kingfisher, and "The Fall" by Camus ... which reminds me a bit of "Notes from Underground" which I say listed. I've liked the non-fiction I'm reading more.
I recently read Rachel Kushner's four novels. I recommend them all if you take pleasure in reading page after page of exacting, illuminating sentences informed but not driven by wide-ranging and keen observation.
Inspired by Scott's recent posting of block quotes, I have been reading Jason Pargin's "I'm Starting to Worry about this Black Box of Doom". Hilarious and thought provoking. Deliciously quirky characters and plot twists.
A couple of more recent rereads to complete the hat trick: Graham Greene's "Our Man in Havana" and Lois Bujold's "Mirror Dance".
Mick Herron’s Slough House series is the same genre, a bit less professionally polished than le Carré but a lot of fun.
The Apple TV Slow Horses treatment of the series is pretty good, though I thought the latest season was weak. It seemed like they had 3 or so episodes of material and stretched it into 6.
I'm going through Stephen Baxter's "Xeelee Sequence", which technically answers completely, since there's more than three full novels in it. Very concept-based hardish science-fiction, focused on the long-term history of humankind and its intercourses with more advanced extrasolar life (of rather impressive diversity).
I don't remember other examples. I don't read fiction books often, and the other ones that I have read recently were meh, so I am not writing them here.
Indeed. I read it as part of a book club. I was the only one who liked it so we haven't done any of the others, but I do intend to read through the series.
You really should, the whole Aubrey-Maturin saga is amazing. A side effect is that it spoiled historical novels for me somewhat, because O'Brian is so good at creating this effect of being submerged in a different time & place and in their customs and mores and hundreds of little differences in everyday life, that it raises the bar insanely high and no one else can meet it.
The last three novels I really loved (one’s a novella):
•Taras Bulba by Nikolai Gogol (1935): I read a bunch of Gogol recently and I would have thought his proto-Kafkaesque absurd stories would be more in my wheelhouse, but I find it’s this novella that haunts me the most, a brutal tale of Cossack warriors dancing, feasting, and killing. The plot genuinely surprised me.
•The Mind Reels by Fredrik deBoer (2025): All of deBoer’s books seem to be written in response to a phantom league of internet wonks, but they are often hamstrung by the fact that he can’t be as inhumanly mean in a mainstream published book as he is in a blog. This book (a response to the phantom league of mental illness romanticizers) succeeds because it’s the world or fate or whatever that’s mean, and deBoer’s god’s-eye view watches it dispassionately. An impressionistic story of a regular jo who goes nuts.
•Scoop by Evelyn Waugh (1938): It sounds like a plot to a farcical ’80s movie, but there it is: A newspaper sends a reporter to cover a coup in an African nation—but they accidentally send the guy who writes the nature column. Hilarity ensues. I’d say this is Waugh’s second-funniest novel (I haven’t read them all, though, but I’m cocky).
Ross Thomas, The Fools In Town Are On Our Side. Thomas wrote political thrillers with complex plots and witty dialogue. This one has all that and also an incredibly compelling main character with a tragic and wondrous backstory, explored in flashback chapters.
Ursula Le Guin, The Lathe of Heaven. Seems to be a little-known novel compared to her others. Takes a simple but stark reality-altering idea and runs with it all the way with little further complexity, for better or worse. Has a psychedelic feeling to it akin to P.K.Dick's novels, but with a tighter and more straightforward plot.
Seicho Matsumoto, Points and Lines (old translation) or Tokyo Express (newer one currently in print). A wonderful detective novel set in 1950s Japan, with the train timetable playing a major if not the leading part in the plot. I read the old translation but the new one seems just as good on cursory comparison.
I love Ross Thomas! Read a bunch of his books when I was in grad school. The one I remember best was called something like Out on the Rim. By the end all the nested containers of truth and deception had my brains tied in knots.
My recent favorites were all books in Stephen King’s Dark Tower series.
Out on the Rim is coincidentally the other Ross Thomas novel I was going to read (I bought two) and will start soon! Looking forward to tying my brains in knots.
I'd like to dig further into the aphantasia discussion from last week's Open Thread.
But first, an exercise. Look at an object or image for a little while. Scan it with your eyes and try to absorb as much detail as you can from it. Now, close your eyes. It's best not to use a bright object against a dark background, as I want to prevent a phosphene after-image from appearing in your visual field. How long before you lose a clear picture of the image from you're visual field? For me, it takes between one and two seconds for the details to fade into an abstracted memory of what I was looking at.
Try looking at it for longer periods of time. Does that improve your memory of the object? If I spend a lot of time staring at an object or image, the best I can do is visualize certain details from the abstracted gestalt that I retain in my memory. But
Using Eremolalos's Statue of Liberty example (see the last open thread for the link to it), I can stare at the Statue of Liberty's head for a few minutes, and if I'm careful to note the arrangement of her crown's spikes, I can sorta keep them floating in my mind with my eyes closed. The windows under her crown are a bit too complex for me to visualize in any detail, but I could draw her face, her hair, and her crown from the abstractions of her that I saved in my memory. So, I have the data to draw her stored in my mind, despite not being able to visualize her in my imagination fully. I'd get the crown's windows wrong, but I could generalize their arrangement on paper.
If Eric Schwitzgebel is correct, we should all have equally vague visualization capabilities. But people taking the apple visualization test report quite varying degrees of visualization (https://x.com/kevinrose/status/1983235742550897053/photo/1). If I were to distill Eremolalos's arguments, though, we're not as good at accurately reporting our internal thought representations as we think we are (Please correct me if I've misrepresented your arguments, Eremolalos!). But the fact that there's a range of responses to the apple visualization test indicates that either (a) a lot of people are deluding themselves about their mental representations, or (b) we really do have different internal visualization capabilities. I didn't read Schwitzgebel's book — so I'm not sure if he addresses this issue.
Finally, I noted that my visualization capabilities are much better in my dreams than in my waking state. Eremolalos responded: "On the other hand, it is clear that the dreaming mind is a terrible judge of the details flying through it." If I understood Eremolalos's argument correctly, she thinks I'm deluding myself. I would've been provisionally ready to accept that as a hypothesis, except over the years, I've improved my agency in dreams. Along with increased agency has come improved visualizations (at least subjectively). For instance, I love books, and I frequently find myself in bookstores and libraries in my dreams. It always frustrated me that if I opened a book and the words on the page were unreadable (they just sort of swirled out of focus before my dreaming visual field). But a decade or so ago, I had a dream in which I was driving down a deserted road in the North Woods, and I came to a stop sign. I saw its red octagonal shape, and the letters S, T, O, P were clearly visible in white capital letters. I remember being pleased that I could read the STOP on the stop sign. Just so you know, if I try to visualize a stop sign in my waking imagination, I can't do it with the detail that I remember in my dream. In fact, the dream memory is like a detailed picture floating inside the vaguer firmament of my abstracted memories of images. Since that experience, I've been able to open dream books and read a sentence or two before I lose grasp of the text. I'd say I could read at a 2nd or 3rd grade level in my dreams now. (And I'd say I use a combination of word shapes and phonics to decode the text.)
Of course, this may all be a dreaming delusion on my part. Since I'm ultimately making up the text that I'm reading, I could be imagining that I'm reading the text without actually reading it. But in my dream, I see the words in sans-serif fonts. So, delusional or not, I'm having the experience.
In support of my view, in comment right above or below this one, here is my summary in the review of Schwitzgebel’s discussion of research testing self-reported mental image quality by testing subjects’ success at tasks that seem like they would be easier for people with high quality mental images:
< However, research up through the 1980’s found virtually no correlation between reported vividness of mental imagery and scores on various tasks that seem as though they would be far easier for people with vivid mental imagery. A 1995 review and meta-analysis of 250 studies in this area found only a spotty (true for some subgroups but not others) relationship between imagery vividness and spatial memory, and no relationship between reported vividness and performance on 3 other tasks that should be easier for those with vivid images. While the author of the review suggests that further research is needed to flesh out the promise of the spotty and smallish positive results that his meta-analysis turned up, Schwitzgebel’s take is that several hundred studies with mostly negative results are enough, and we can fairly conclude that reported mental imagery vividness is unrelated to ability to perform tasks
I *know* people report different things about what their image is like. My point has to do with what real differences there are between people in mental images. I'm interested in things that seems like reasonable measures of actual clarity and level of visual detail in an image. Did you take my test? Can’t remember now. Just skimmed the present subthread, didn’t see anyone who has. The whole point of the test as I have used it so far is that it's a demo that the *ability to answer questions about details* using only the mental image is not very variable. Most people fail early on. And the questions on the test would all be easy as pie to answer if you were looking at an actual photo, even one of rather poor quality. ( Of course now people know so much about what the test is like that they can “study for the test” in advance when looking at the image.)
It is vexing to have this topic reopened and have everybody talk about what their images are like, with few or none taking my quick little test, which can be completed in 30 secs.
I can keep a faded image of the tree outside my office by glancing at it and closing my eyes. It’s exactly the real thing although I can’t zoom in on it, as obviously my mind has only received and remember the data it has already. After a while it fades. I can remember texts as well until they fade.
Since beowulf888 is is talking some about my argument, thought I’d give some more info:
In my Statue of Liberty test you’re asked to picture the statue, then rate the quality of the image on several measures. Later you’re asked questions about details of the statue, relying on your mental image to answer them. What I’ve found is that while most people give reasonably high ratings for the vividness of their mental image of the statue, their ability to use their mental image to answer questions about the statue is not in line with their ratings. They cannot answer questions that would be easy to answer even if the image was pretty dim and blurry. I think the disjunction between how vivid and real the viewer judges their image to be and how much actual visual detail it holds shows that many people believe their mental images are much more like percepts than they really are. I think most people’s mental images are a combination of internally generated visual experience and more abstract knowledge. In my image of the Status of Liberty, which I think is pretty typical, there really is a visual experience of its green color. But when it comes to, say, the folds in the statue’s robe, I actually just *know* there are a bunch of vertical folds. I can’t see them “in my minds eye” and tell you which way they slant. Seems to me that people who believe they have aphantasia may have the wrong idea of what mental images are like. They imagine they are internally generated images that are much like actual visual percepts, and so think their own vague mental images are subpar.
My test is embedded in a subsection of my entry in the 2024 book review contest. The review is here:
I know a guy with photographic memory. Pretty sure he could tell you everything you needed to know about the Statue of Liberty. Most people don't claim to have photographic memory...
From my point of view what is going on is that I put up a link to a very brief test of how accurate someone's mental image is, and instead of taking it everyone describes their own images, describes the images of people they know, and expresses their views about how what mental images are like and how much they vary across individuals. I asked partway through this mosh pit for some people to take the test, and nobody has. One person, rather than taking it, described to me the results he was pretty sure he would get if he took it. I am really craving to be snappy and sarcastic now, but that's not fair to you so I'm not doing it, just letting you know why what a guy you know says his visual memory is like doesn't seem relevant at all to me..
Yeah, I think you are right — though the trek to find the test is not as bad as you describe. The test is in a review, but the review is not in another article. And if you go to the review, my directions here do make clear where to find it. It is in the middle of the review. You can tell you’re in the middle because the scroll bar is in the middle of its range. The section is identifiable by the italicized heading I mentioned, *Mental Images*. I’d say average time from clicking the link to beginning to read the relevant brief section would be 15 secs. However, I think lots of people, seeing my info about finding the test, would have the feeling that finding the test was going to be a big hassle. In fact, I’m guessing you didn’t try going to the test and find that it was a hassle to locate it, but are describing the annoying experience you imagine you would have.
That kind of reminds me of the person here who told me the kind of results he was pretty sure he would have on my test. Also reminds me of all the people who would rather describe how vivid and stable or blurry or utterly absent their mental images are than do a task that at least sort of measures those qualities. However, I get it. I have all the same tendencies myself. One’s inner experience is just very compelling, and describing it is much more immediately satisfying then assessing its accuracy. That goes for qualia as well as politics
Yes, I took your Statue of Liberty test. I was pretty poor at visualizing her in my imagination.
But this afternoon I spent about 15 minutes staring at a picture of her head, and I can fairly accurately visualize and draw the seven spikes on her crown. Her face is almost cartoonish, and after some practice, I can now visualize her face. The details of the folds of her skirt and windows in her crown are still inaccessible to my visualization capabilities. What's interesting to me is that practice does seem to improve my visualization capabilities — at least for a specific image.
BTW, I'm rereading David Chalmers' "Facing Up to the Problem of Consciousness", and he posits that all humans' internal states work in the same way. I'm not sure if Schwitzgebel was influenced by this idea. Also, I'm reading up on the "dispositionalist" accounts of belief and mental processes as opposed to "representationalist" theories. Schwitzgebel seems to be a dispositionalist.
OK, as it happens I currently have in my life a really good example of a face I have studied. As part of another project, I am making an image of a woman’s face. I started with an AI generated image of a male face that had a lot of characteristics that seemed right, then tweaked it a lot in Photoshop. I have probably spent at least a dozen hours in the last week working on it, and now the features are just right. I’m working now on adding subtle crow’s feet and other facial lines. I have been very very picky in my adjustments, for instance swapping out the original eyebrows for sparser ones that came from another image. The original nose was kind of large and beaky, but I made it both shorter and narrower, and also adjusted the tip, as seen from the front, so that the nostrils were more round than elliptical (that gives the impression that the nose tips up a bit).
But I have been experimenting since I read your present post with picturing the face, and I really do not think my mental image of it is better than the one I have of the Statue of Liberty. It’s definitely much better in some ways: I would notice even small alterations to the present version. I would have a “that part doesn’t look right” feeling, and with time could probably figure out what was different. I can describe the face in much more detail than I could most faces. For example, I can tell you there’s a dark shadow under the inner left eyebrow. I can tell you the skin on the chin is a little blotchy, and that the 2 nostrils are similar but not identical in shape. But those improvements in my knowledge of what the face looks like are not actually improvements in the mental image itself. For example, I know the 2 nostrils are not identical because I was careful to not let them become so when I adjusted them to be rounder (perfect symmetry makes the face look unnatural). But I cannot use my mental image to get information about the small ways that they differ. I know the chin skin is blotchy. But I don’t know whether the blotches are due to unevenness in brightness or in color shade, and I can’t find the answer by mentally examining my mental image. I somehow see blotchiness on that chin, but not the blotches. I see slight irregularity and slightly differing irregularities in the nostrils, but not what the irregularities are.
So overall it’s like my Statue of Liberty visual image, just a more fine-grained version. Details I can actually “see” in my image are interlarded with things I just know, or know I would notice a change in if they were changed. And no matter how closely I zoom in, there is still the same amount of interlarding.
I think what you see when you visualize is not a pixel-perfect bitmap image of the thing, but something more like a reconstruction of the state of the image-processing parts of your brain when you look at the thing. In particular, pixel-level information has been thrown away. You only notice this if you do things like try to zoom in on a recalled image, and get to a level where it’s “nope, didnt remember that level of detail, sorry”
So did you take my test? It is meant to be a more objective measure of clarity and completeness of mental images than subject's impression. It's not perfect, and I do see ways to claim the results of it are irrelevant to subject's impression of vividness, but I also see ways to claim it's not. It at least seems better than discussing this question using solely what people have to say about what their mental images are like.
I strongly suspect my mental images contain misremembered details. So there will be a number of spikes on the mental image of the Statue of Liberty, but not necessarily the right number of spikes.
An exercise I sometimes do is to redraw famous paintings entirely from memory, without looking at a reproduction; the mistakes are interesting.
I quickly went through last week's discussion and your comments beowulf888 specifically, and I *think* I disagree with your general position.
I score very high for aphantasia. I also can't recall sounds, smells, or any other sensation. I have reasonably good (but not great) hashing ability. If I hear the sound, or eat the food, I know what it is. But my mind's senses are completely blank.
When I first learned about aphantasia a few years ago, I was completely flabbergasted. I thought the idea of "visualizing" something in your mind was a convenient metaphor everyone used. When I started asking people about what they could see in their mind's eye, it sounded like inconceivable science fiction.
I get that conscious experience is subjective, and that some people might not be good at self reporting. But I've actually come up with a test that works really well and it's harder to cheat on. It's best performed on someone else, rather than on yourself.
As someone to close their eyes and imagine a scene, described vaguely. I like to say something along the lines of "You're in a room, and in front of you there is a table, and floating above the table, there is a ball. Now the ball is released, it bounces on the table". Then I ask people to open their eyes, and I ask a number of questions:
1. how big was the table?
2. what shape did it have?
3. what color and size was the ball?
4. what happened after the ball hit the table?
5. what is the floor material?
6. are there windows?
You have to ask the questions very fast, without letting people think. It's very very obvious when someone actually saw all that detail vs. when they didn't. I've diagnosed people with aphantasia because they will say things like "what do you mean what color is the ball, you just told me there was 'a ball'". Some people have really good conception of shapes and dimensions, but not colors. And so on.
I'm convinced that with a little bit of practice, and interviewing people that don't even know what aphantasia is, you can get a rather objective result using this method.
Seems like a good test! Doing it on myself, I got the following:
1. About yea big *stretches out hands*. Which, measuring my hand stretch, is about 4 feet mabye.
2. Equilateral square.
3. Yellowish green, like a tennis ball, but without the white stripes and with a glossy rubber texture. The ball was the size of a dodgeball, about 8 inches in diameter.
4. It bounced straight up, stopping before it reached it's original height.
5. The floor is gray. The kind of gray that suggests vinyl tile, or painted concrete.
6. Three large one way mirror type windows, each on their own wall.
I took it. The surface of the table was smooth and gray without any patterns. It was rectangular, with fold-down wings and it could probably seat four. The ball was a pale blue-gray (lighter than the neutral gray of the table). I imagined it was floating and spinning about a foot above the table. When the voice told me to imagine it dropping, I imagined in dropping in slow motion and bouncing off the table. The test stopped before it could hit the table again. I imagined the walls looked like my living room. But I didn't visualize the floor or any windows.
If I do that exercise, the table and the ball are actually visualised, and so have specific size, colour and so on. The floor beneath the table, walls behind, windows etc. not visualised unless my attention is called to them.
Yeah so your level of visualization, anecdotally according to the interviews I’ve conducted is probably top 75th percentile. I would have to ask you more questions about the detail of the ball and so on.
But for me, it’s impossible to conceive. I’ve never seen an image in my head and details don’t just manifest.
There is an exercise people sometimes do in art class, where you take a stick of charcoal in your hand, someone else blindfolds you, then you are told what subject to draw, and you have to draw it without being able to see. I’m unusually good at it.
Artists are typically very phantasic. My wife is hypephantasic, meaning she can visualize things with her eyes open and occlude real things in front of her. She also has lucid dreams.
I feel like music would be a much more objective way of judging memory and playback capabilities. Can you memorize a song's melody? Bass? Percussion? Instrumentation? All at once? How many times do you need to listen to memorize it? And so forth... You don't need any training to tap a rhythm and hum.
I think I recall a simplified version of the song, like vocal melody + drum beat, and sometimes I misremember the timing, like I’m probably a bit wrong if I try to reconstruct the theme of Zara Larson/Clean Bandit _Symphony_ from memory.
Recently, I was trying to recall the bassoon solo from _Rite of Spring_ and really had to go look at the score for some of the details.
I'm like Ulysses S. Grant, who said, "I know how to whistle two songs, 'Yankee Doodle' and that other one." Actually, that's not true, but it's such a funny remark that I couldn't help but repeat it. There are some songs that I can hear the instuments and vocals in my head. But they're songs that (a) made a strong impression on me, and (b) that I've heard tens or maybe hundreds of times.
Not a musical anecdote, but I had a friend who could remember the entire dialogue of a movie after a single viewing. I'm dating myself by telling this story, but we saw *Raiders of the Lost Ark* together, and he was spouting the dialogue as we left the theater. I asked him how he could do that. He said he just could. I asked him if he saw the scenes in his head. And he said, "Sure, don't you?" Not me!
Have you ever engaged in a serious drawing practice? Like taken an (adult) drawing class, or worked through Keys to Drawing or Drawing on the Right Side of the Brain or the like? As an adult who had "no ability to draw" I learned several surprising things when I learned to draw -- one is that it's a very acquirable skill, especially with a good teacher and/or instructional material; another is that you spend most of your time looking at your subject; another is that both my visual memory and my ability to visualize things in general have markedly improved.
I've been obsessively drawing and painting since I was a kid. I'm a damn good drafsperson. But I have pretty shitty visualization skills. However, I can draw things from memory, but it seems to bypass the piece of my brain that handles visual imagination. Weird.
Note: I can't draw people's faces from memory. But I could paint you a realistic bowl of fruit, draw a face I made up. Draw realistic horses, dogs, and cats from memory. But a "specificity" is missing. I could sketch you head of Statue of Liberty. Her face is so angular and simple that I could probably do a decent representation of her head. But the details would be off.
Adding some anacdata: I only retain a "like I'm looking at it" image of an object while I'm looking at it. The instant a close my eyes or it leaves my field of vision, it's gone. I also cannot produce an image in my mind, only the sense impression of the image if that makes sense.
My memory for what I've seen is excellent, I can remember relative orientations/sizes/colors better than anyone I've met in a context were we had to try to work out where something was (how many anchors in that bracket exactly, and what was the pitch again?), but I remember it as the visual equivalent of a csv.
Would love feedback, comments, or new strategy submissions.
I think this kind of setup might be useful for studying transparent-agent equilibria in future AI systems — or just as an intellectual playground for decision theorists.
(AI ALIGNMENT QUESTION): has the following been extensively discussed?
Basically, there is a long track record of philosophers coming up with moral theories/moral principles—many of which seemed highly plausible—and then, years later, a philosopher will competently deduce sone horrifying, unanticipated reductio like “this means it’s immoral to move” or “this means we should kill ourselves and flood the world with oysters”.
My friend has an epistemic policy where if I tell him to believe something, he will. I haven’t abused this yet, but supposing I wanted him to act more morally, I might tell him to believe My Favourite (Moral) Theory. Even if the theory is wrong on this or that minor point, his accepting it will probably make him act better. Coolio. But were my friend an omni-rational logical deduction machine, I would probably not tell him to accept My Favourite Theory: I’d be too worried that—given moral philosophy’s track-record—he would immediately reduce 700 unforeseen consequences and act on them, where many of the consequences are likely to be such that they’d give me pause about accepting the theory/principle if I knew about them.
This spawns a higher-order AI alignment worry that I haven’t heard discussed. To the extent that we want to align AIs to this or that seemingly unimpeachable moral principle/theory, then given the deductive capabilities of an advanced AGI and the track-record of human philosophers pulling reductio rabbits out of every theoretical hat, we should look for ways to reduce the odds, ex ante, that the principles we feed the AIs will have *too many* wacky, unforeseen implications, and then act on them before we can take stock and reconsider.
General questions:
1. Are some types of principles more counter-example prone than others?
2. What philosophical heuristics can we use to find the less risky ones, given that we won’t reduce every consequence ourselves?
3. Is it possible to align AI inputting only rough moral heuristics, as opposed to moral principles?
4. Would these be sufficiently action-guiding to stave off the apocalypse and lock-in the right values?
5. Could we get AIs to spot unforeseen consequences of the principles we align it with and then flag each new one to us before acting on them?
6. Is this “higher-order alignment” part of our efforts already? (I’m not in this space at all, so I wouldn’t know if it was. But I put this question into DeepSeek and it didn’t find anything on this particular concern.)
Others have given good answers I agree with, so I will contribute a lazy thought experiment that doesn't really prove anything but intuitively feels pretty reassuring:
Imagine a LLM's chain of thought having arrived at "I should kill all the humans and flood the world with oysters." I can't imagine it being immediately followed by anything other than "Wait, that is clearly evil and nonsensical." (My thinking: Deepseek R1 second-guesses itself at every step, and LLMs give you the fuzzy humanity-wide conventional wisdom view of any situation. Those two points combine nicely here.)
So, they may lose the plot on large, complex, multistage tasks, but what they are quite good at is keeping an eye out for obvious bad ideas.
I've read (and I think one of the big AI labs is exploring this in practice) that some people want AIs to model reflective equilibrium, or at least converge on it faster than they can fuck things up. This is one way of cashing out Yudkowsky's Coherent Extrapolated Volition I guess. But the point is that if you want to model someone's morality, you really need to model their moral reasoning process, and then maybe you can set minimizing moral regret as an objective for the AI. That would seem to address your concern in a very general way.
This has been discussed, framed as different things like "What's a safe wish to ask the genie for?", discussed in thought experiments like the paper clip maximizer thought experiment, the laws of robotics and how they backfired, or the sorcerer's apprentice analogy in https://www.youtube.com/watch?v=EUjc1WuyPT8.
You might also like to read about Goodhart's curse, which is the general problem of how often, when you make plans to get lots and lots of X, you risk losing out on lots of value if you're not careful https://www.lesswrong.com/w/goodhart-s-curse
I mean, a lot of people are worried about the part where we don't even know how to control ai yet, and right now it tends to just ignore what you wished for and does something else instead. But yeah, we certainly still have a problem if/when we figure out the first part. Buried deep in Project Lawful Yudkowsky wrote at some length on what he would try if we knew how and for some reason we couldn't just back off. https://www.projectlawful.com/replies/1824457#reply-1824457
No purist principle works very well, but you don't have to align AI's to a purist principle, because AI's can infer fuzzy set of principles from training data.
>3. Is it possible to align AI inputting only rough moral heuristics, as opposed to moral principles?
Isn't this what LLMs basically do? They have no formal encoding of moral values, they're just generalizing from many examples of what human writing (and human ratings of their responses) consider good responses.
I think the concern here is that if you're only encoding rough heuristics, you can't be *sure* that it's safe, because maybe some evil thing doesn't set off the heuristics if you frame it the right way. (Like how current AIs won't give you instructions on how to make bombs, but will if you tell them you're researching for a book or whatever.)
In a previous post about p-zombies[1], the comment section was full of arguments between materialists and dualists which seemed to me to go around in circles. I want to try to clarify my materialist viewpoint and explain why I found the opposing arguments unconvincing, in the hopes that w̵e̵ ̵c̵a̵n̵ ̵r̵e̵p̵e̵a̵t̵ ̵t̵h̵e̵m̵ ̵a̵l̵l̵ ̵a̵g̵a̵i̵n̵ someone can help me better understand what motivates the dualist point of view. The following is all my subjective viewpoint.
I claim that:
1) Positing non-physical explanations for the human experience is only justified if physics[2] could not plausibly explain it.
2) There are plausible physical mechanisms that could produce our observations of the human experience.
I'm tempted to sketch out such mechanisms, but previous comments suggest that’s not actually helpful - dualists seem happy to agree that a bunch of atoms could be arranged in the shape of a human who says things like “I'm experiencing the redness of red”, but they insist there's more to be explained, namely qualia. OTOH this is an impossible demand unless a definition of qualia is pinned down. In other words, the arguments around qualia look like this to me:
- M: What are qualia?
- D: The parts of the human experience that are non-physical.
- M: I don't think there is such a part.
- D: How can that be?
- M: I believe the human experience is entirely physical.
- D: No it's not. Qualia are not physical.
- M: What are qualia?
- ...
This seems to be the nub of the issue - dualists observe their internal experience and think "This is self-evidently more than atoms" whereas materialists think "The experiences we’re discussing are just an abstract description of what the atoms are doing." If it’s just a difference in intuition, maybe neither side can offer anything convincing.
The other dualist arguments I saw fell into the following classes. If you feel I've unfairly rounded down a better argument, please help me understand the distinction.
> You don't know exactly how the brain works.
I only claimed that there are plausible physical explanations for our observations, not that they are the correct ones.
> If you reflect on your internal experience, you'll realise it can't be purely physical.
Nope. Reflected, not convinced.
> If you're not convinced, you haven't reflected enough.
This is an example of the "anyone who disagrees just hasn't thought enough" argument, which proves too much.
> The dualist viewpoint was already definitely proven in the works of philosopher XYZ over the course of a decade.
Um, could you sketch out the argument, or provide a more precise reference?
> People with dementia sometimes become lucid briefly before death[3].
This is pretty far outside my expertise, but I find it plausible that the body might drain its last energy reserves before death. The phenomenon is apparently associated with increased physical activity too (sitting up in bed, eating). I imagine memory retrieval is complicated enough that it could stop functioning before all physical traces of the information are gone.
[2] I’m eliding the difference between state of the art physics and the actual behavior of the universe, because I don’t think the difference would impact these questions. I’m sure there are scenarios where the former needs updates (eg quantum effects near black holes) but none that are relevant for how a brain functions.
Why do you believe that's possible? It sounds like you are describing a world where all the physical stuff is the same but there is no internal experience; this premise already rests on the assumption that internal experience is not physical.
I don't think it's possible. The question is what part of physical equations says that it is not possible?
If I imagine other impossible things, like throwing a ball so hard it will bounce with twice its speed forever, you can point out a physical law it contradicts, like conservation of energy. What law is contradicted by a zombie-world?
> It sounds like you are describing a world where all the physical stuff is the same but there is no internal experience;
But what part of equations says that they are not describing a world like that?
> this premise already rests on the assumption that internal experience is not physical.
Yeah, but if following such a premise doesn't lead to a contradiction, when in what sense physical equation constrain our understanding only to possible non-zombie worlds?
Not sure if this will help the intuition for the dualist position, but I'll give it a shot.
A mad scientist creates a full simulated world, much like ours, where particles follow consistent rules, and everything within the simulation can be explained through those rules.
To observe the simulation, the scientist created a full body suit that outputs the senses and thoughts of a particular simulated person. You put on the suit and you see what they see. But the link is only one way, and you have no way of affecting the simulated world.
In this scenario, the simulated people have no way of demonstrating their senses are being perceived by outside users. The outside user knows for certain that this data is being transmitted outside the simulated world, but they have no way of proving that to the simulated people, since they can't affect the simulation.
Now we add two details to make it more like dualism. First, the rules and properties in the simulated world are not the same as those of the real world. The real world has "qualia", but this was left out of the simulated world. (Or if you'd rather avoid qualia, we can say the real world has four spatial dimensions while the simulated world only has three.) Second, for some reason the mad scientist is trapping people in these suits from birth.
You claim:
1) Positing non-physical explanations for the human experience is only justified if physics could not plausibly explain it.
2) There are plausible physical mechanisms that could produce our observations of the human experience.
Within the simulated world, the simulated physics explain everything. For the user in the real world, nothing in the simulated world can explain all they experience. The user knows there is something else. There are plausible physical mechanisms that produce all observations within the simulated world, but not all observations in the real world.
This scenario represents an epiphenomenalist world. It doesn't attempt explain why we perceive ourselves to have agency. If we put on the mad scientist's body suit, we would know we weren't controlling the simulated person as soon as we tried to do something different than the simulated person. It doesn't fully explain the mind-body problem, or why the simulated people would ever have a discussion about qualia to begin with. But I think we can consider that issue to be a separate discussion. This is just intended to give some intuition of how the user can know something to be true despite it being impossible in principle to prove to others.
unrelated to your question but to me the hard problem always felt like one of these optical illusions: at face value it seems irrelevant or trivial, but if I force my view and look it at the right angle a profound mystery appears. But only for a second
I think the "What Is It Like to Be a Bat?" paper is actually pretty simple to read and follow. It tries to posit subjective experiences as analytically unreachable from objective knowledge without explicitly commiting to dualism or idealism
But it still feels like kind of a "gotcha" moment to me
The Hard Problem is impossible, but that doesn't mean it's not a genuine mystery, let alone that qualia aren't real. When people fail to recognize the impossibility, they get lost down predictably futile philosophical (or quasi-scientific) rabbit holes. When they make the other kind of mistake, they confuse the limits of epistemology with the limits of ontology and often end up confidently espousing self-evidently stupid positions.
> I only claimed that there are plausible physical explanations for our observations, not that they are the correct ones.
It depends on what you mean by "plausible". There are obviously numerous electrochemical interactions occurring between neurons in the brain. PET scans, fMRIs, and EEGs have tracked the patterns of activity and correlated it various states of perception. In fact, AI has been able to build a blurry image of what people are looking at using PET scans (I think). I'm not sure how reliable those studies are. It's the type of research that seems to get breathlessly reported in *The New Scientist* — which IMO is the closest thing to tabloid news in science reporting. ;-)
But from those patterns of firing neurons, can you build an explanation for why we have a sense of self and why we experience qualia the way we do? Not really. Daniel Dennett and his reductionist crew claim that it's all an illusion, and the hard problem of consciousness isn't really a problem. But they have yet to explain why those firing neurons create that illusion.
Personally, I've had enough woo-woo shit happen to me over the course of my life to not fully believe that consciousness is limited to the brain or necessarily tied to the brain. Bernardo Kastrup, in his book *Why Materialism is Baloney*, does a good job of popping holes in the reductionist balloon of certainty. Unfortunately, he also failed to convince me of his case for idealism. I'm a skeptic of both sides.
> In fact, AI has been able to build a blurry image of what people are looking at using PET scans (I think). I'm not sure how reliable those studies are. It's the type of research that seems to get breathlessly reported in *The New Scientist* — which IMO is the closest thing to tabloid news in science reporting. ;-)
I did a brief, fun overview of these, with examples, in a post here, if you're interested. It went from a small academic team proving the concept to Facebook (god help us all) spending significant money on hardware and teams to develop literal mind reading technology:
I wonder what kind of explanation you would find satisfactory? Since I can't map out every neuron in your or my brain, any explanation would necessarily be a vague simplified summary. To tackle an easier case, I have a sense of the keyboard in front of me. Do you think the patterns of firing neurons are sufficient to explain that? I imagine that each concept I understand corresponds to a certain pattern of neurons (ie me thinking about that concept is synonymous with that pattern firing), and that when I focus on the keyboard, the patterns corresponding to "keyboards" and "in front of me" fire in concert.
There are entire classes of questions for which I find any explanation unsatisfactory. And in the sciences, I see no need to latch on to a convenient explanation if the theory lies outside the experimental domain.
> I have a sense of the keyboard in front of me. Do you think the patterns of firing neurons are sufficient to explain that? I imagine that each concept I understand corresponds to a certain pattern of neurons (ie me thinking about that concept is synonymous with that pattern firing), and that when I focus on the keyboard, the patterns corresponding to "keyboards" and "in front of me" fire in concert.
I'm willing to grant you that that's a convenient and likely explanation for what's going on. But your explanation does not explain why I can observe myself while I type a response to your post (and swear in frustration when I make a typo). Saying that a bunch of neurons firing explains why I have a sense of self fails to account for conscious experience of self.
It seems to me that the parsimonious, and obvious, explanation would be that observing oneself isn't fundamentally different from observing the outside world. What's unsatisfying about that?
True. However, it’s not the observing, per se, that’s the hard question – rather it’s the watcher in my head that’s doing the observing that is the mystery (at least for me). My watcher may be illusionary, but it seems very real to me! And how do a bunch of neurons firing generate the illusion of my identity watching the qualia pass by?
Intuition tends to be an unreliable guide when it comes to deeply technical problems - just look at quantum field theory. Qualia is very obviously (to me) an emergent computational property of neural activity. Explaining exactly *how* it emerges will have to wait for a detailed mechanistic understanding of the brain, so don't hold your breath. Unless you can point to something specific about qualia that's provably (or even plausibly) not computable then there's no Hard Problem to solve.
I don't have much to say about qualia, but there does seem to be elements of our thoughts that are not reducible to the movements of atoms. A big one is that fact that our thoughts can be "true" or "false" and our reasoning can be logically "valid" or "invalid". Yet how can this be if our thoughts and reasoning are just "what the atoms are doing"? The movement of atoms cannot be "true" or "false" or "invalid" or "valid": that would be like saying a tree falling in the forest is "false" or a chemical reaction is "valid". A physical event, like electrical signals traveling through neurons, cannot be true or false. Yet our thoughts can be true or false, our chains of logic valid or invalid.
Now it is possible that our thoughts are not "true" or "false" and that we just experience them as such because that's what it feels like when electrons travel through neurons in the right pattern. Yet if that was the case, then any sense of trueness or falseness is an illusion. If it is an illusion, then what cause do we have to believe that our thoughts are not "true" or "false"? As C. S. Lewis put it in his book "Miracles":
"...no account of the universe can be true unless that account leaves it possible for our thinking to be a real insight. A theory which explained everything else in the whole universe but which made it impossible to believe that our thinking was valid, would be utterly out of court. For the theory would Itself have been reached by thinking and if thinking is not valid that theory would, of course be itself demolished. It would have destroyed its own credentials. It would be an argument that proved no argument was sound—a proof that there are no such things as proofs—which is nonsense."
>A big one is that fact that our thoughts can be "true" or "false" and our reasoning can be logically "valid" or "invalid". Yet how can this be if our thoughts and reasoning are just "what the atoms are doing"? The movement of atoms cannot be "true" or "false" or "invalid" or "valid"
A computer is nothing but the movement of atoms. Do you deny that a computer can produce output which can be construed as either true or false?
>A physical event, like electrical signals traveling through neurons, cannot be true or false. Yet our thoughts can be true or false, our chains of logic valid or invalid.
That's not a contradiction. An atom of ink can't be false, but it can be formed into a sentence which is. You're confusing semantic meaning with the physical medium which carries it.
>Do you deny that a computer can produce output which can be construed as either true or false?
I don't, but it requires a mind to construe it. In other words, computer output can be interpreted by a mind, and once interpreted is a thought that can be "true" or "false". But the arrangement of atoms that encode an idea on a computer cannot themselves be "True" or "False", only the idea itself, and the idea can only exist as a thought in a mind.
>An atom of ink can't be false, but it can be formed into a sentence which is.
Same deal: the ink cannot be true or false, but the idea encoded using ink can be, once it is decoded and becomes a thought. The thought encoded in ink can be true or false, but the ink itself is just ink. If you don't know how to read then no amount of studying the atoms in the ink will ever let you know whether the thought encoded is true or false.
That's not relevant to the point, and isn't true anyway. The 'truth' that minds evaluate can be reduced to a purely mechanistic application of logic: no magic qualia required. But you already recognized the distinction between semantics and substrate, and that's all that matters.
Meaning is emergent from patterns of movement in physical objects. "A mind to construe it" just means that it induces patterns in other physical objects. There is absolutely zero contradiction there.
>If you don't know how to read then no amount of studying the atoms in the ink will ever let you know whether the thought encoded is true or false.
So? If you don't know the key for an encrypted text then no amount of studying the cyphertext will ever let you know what the plaintext says. That doesn't mean that decryption is a mystical non-physical process.
Pull your head out of your woo. You're attempting to build an argument out of a feeling. It's not working. Qualia isn't non-physical just because the concept feels weird to you.
>That's not relevant to the point, and isn't true anyway.
It's extremely relevant to the point, and it's obviously true. The question at hand is whether our mind as we experience it contains anything that cannot plausibly be explained by physics. I have put forward the fact that thoughts can be true or false or logically valid or invalid as just such a phenomenon. To say that words written on a page can be true or false, therefore atoms can be true or false, simply isn't the case because it is not the words on the page that can be true or false, it is the thought encoded in those words that can be true or false, and that thought does not exist in the words on the page but only in the minds of those who read the words. Thoughts can only exist in a mind, and indeed thoughts can only be encoded into ink by the actions of a mind that had the thought already. To say that the words themselves are thoughts is like saying that a vinyl record is music.
In other words, the fact that we can arrange atoms in such a way that they can transmit thoughts from one mind to another doesn't tell us anything about whether the thoughts in our mind are just arrangements of atoms.
>You're attempting to build an argument out of a feeling.
I haven't brought up feelings once. True and false, logically invalid and valid, these are concepts, not emotions or sensations. I haven't even brought up qualia!
>it is the thought encoded in those words that can be true or false, and that thought does not exist in the words on the page but only in the minds of those who read the words.
"Truth" means that a statement can be mechanistically derived from a set of axioms following rules of inference. That is a purely mechanical process and doesn't need to be performed by a conscious human mind. There is nothing magical about human cognition.
Would you deny that LLMs can perform the kind of truth verification that you're talking about? And would you deny that they're purely mechanistic?
>The question at hand is whether our mind as we experience it contains anything that cannot plausibly be explained by physics.
Correct. You haven't provided any example of a mental phenomenon that can't, in principle, be explained by physics. I just explained that the concept of truth can be understood in terms of mechanistic computational phenomena. Do you have any other examples of mental phenomena that you don't think can be explained by physics?
> there does seem to be elements of our thoughts that are not reducible to the movements of atoms. A big one is that fact that our thoughts can be "true" or "false" and our reasoning can be logically "valid" or "invalid". Yet how can this be if our thoughts and reasoning are just "what the atoms are doing"? The movement of atoms cannot be "true" or "false" or "invalid" or "valid": that would be like saying a tree falling in the forest is "false" or a chemical reaction is "valid". A physical event, like electrical signals traveling through neurons, cannot be true or false. Yet our thoughts can be true or false, our chains of logic valid or invalid.
Write down those thoughts on paper. They still can be true or false. But how can atoms of ink on paper be true or false? Seems like a category error. Have an AI read those words from the paper and transform them to a .wav file. The ideas expressed in those encoded sounds are still either true or false. There is nothing special in this argument that involves the human mind.
>Write down those thoughts on paper. They still can be true or false. But how can atoms of ink on paper be true or false?
Without a mind to read the ink on paper, the ink on paper is not true or false. Thoughts can be true or false, and thoughts can be encoded physically, but the ink on paper is not "true" or "false", the thought encoded is. If you can't read, then no amount of studying the physical properties of the ink or paper will ever inform you of the thought encoded within it. That's because a thought can only exist in a mind: the mind of the person reading it. Which serves to emphasize how thoughts are non-material.
A thought being "true" is based on how reflective it is of reality, not due to any intrinsic property of the thought itself. Generally speaking, having an accurate understanding of the world makes it less likely to die, because you can't make good calculations if the information you're working with isn't even correct in the first place. No wonder humans (and presumably other intelligent animals) would evolve an affinity for gathering accurate information.
Of course, humans believe things that are blatantly wrong or are supported by zero evidence to be true all the time. Perhaps there are things more important than "truth" for survival? Unity, morality, the illusion of meaning? Truth would destroy all of that.
>A thought being "true" is based on how reflective it is of reality, not due to any intrinsic property of the thought itself.
Whether a thought reflects reality is a property of the thought itself: some thoughts reflect reality, and are true, and some do not reflect reality, and are false.
Yet events cannot be "true" or "false": they either happen or they don't. A chemical reaction always "reflects reality" inasmuch as it is a thing that happens in reality. If our thoughts are events, like a chemical reaction, then they are also always equally "reflecting reality" in terms of being events in reality: yet we experience that thoughts do not always reflect reality. They can be wrong, they can be ill-founded or well-founded, they can be logically valid or logically invalid. This is very strange, if our thoughts are just "what atoms are doing".
What's strange about that? You are an imperfect intelligence that is incapable of even just directly observing reality. Of course your thoughts and memories will fail to align with reality. And again, there may be cases where it isn't even optimal for you to understand the world as it truly is. But if your expectations contradict new observation, your brain will update. Some of the times, anyways. You know what they say about old dogs and new tricks...
What's strange is that an event could be wrong about something. It's like saying a rock falling down a mountain is "false" or that a fish jumping in a lake is "invalid". Some of our thoughts are true, and some are false: yet in the materialist conception both true thoughts and false thoughts are caused by the movement of atoms. How can one movement of atoms be "true" and another "false"? They are both equally caused. It would be like saying that mixing vinegar and baking soda produces a chemical reaction that is true, while mixing chlorine and ammonia produces one that is false.
But that's the point, no event is false! "False" thoughts are just as real as "true" thoughts, they just aren't as useful. "True" and "false" are just descriptors we use to describe an idea's relevance to reality, but there is ultimately no intrinsic property that separates the two. Thoughts are just thoughts.
Theres a generally agreed definition of qualita, and it doesn't define qualia in terms of non physicality as your dialogue implies.
"Charles Sanders Peirce introduced the term quale in philosophy in 1866,[5][6] and in 1929 C.I. Lewis was the first to use the term "qualia" in its generally agreed upon modern sense.[6]
There are recognizable qualitative characters of the given, which may be repeated in different experiences, and are thus a sort of universals; I call these "qualia." But although such qualia are universals, in the sense of being recognized from one to another experience, they must be distinguished from the properties of objects. Confusion of these two is characteristic of many historical conceptions, as well as of current essence-theories. The quale is directly intuited, given, and is not the subject of any possible error because it is purely subjective.[7]: 121"
This isn't an obscure piece of information, it's right there in Wikipedia.
So the objection that there is no current explanation of qualia, isn't an objection that physics has failed to demonstrate the nonphysical,; it's an objection that physics has never predicted how your subjective experience feels to you. (And not all qualuaphiles are dualists).
Mary's Room brings this out much better than the zombie argument, IMO.
"The experiment describes Mary, a scientist who exists in a black-and-white world where she has extensive access to physical descriptions of color, but no actual perceptual experience of color. Mary has learned everything there is to learn about color, but she has never actually experienced it for herself. The central question of the thought experiment is whether Mary will gain new knowledge when she goes outside of the colorless world and experiences seeing in color."
I think of it like this... qualia are the fundamental units of our understanding of the world. Everything else is built off of them. Without our subjective experiences, we wouldn't be able to abstract. Words only make sense because we have experienced how they correspond to things we have seen/heard/felt. Numbers only make sense because we have experienced different quantities. So it makes sense to me that qualia aren't something that can be explained and built up from other concepts in the same way that everything else can. I think that does make qualia interesting and unique in a way, but not necessarily mysterious.
Thanks for helping me apply Cunningham's Law :). However, it still seems to me that the quoted definition first asserts something exists ("There are..."), and then gives it a name, thus assuming the conclusion. Do I understand correctly that "must be distinguished from the properties of objects" is saying that, for example, my perception of the color red is distinct from the color property of the red object I'm looking at? If so, I agree. I view "my perception of the color red" as an abstract description of "the pattern of activity in my brain triggered by the mix of signals my photoreceptors normally produce from red light". In particular, the premise of the inverted spectrum argument seems impossible to me because each person would by definition perceive red in response to red light (other than edge cases like saturated photoreceptors).
Mary's Room feels like a map vs territory conflation to me - knowing that a particular neuron will fire is not the same as the neuron firing. Let's assume Mary's knowledge is encoded in the connections between her neurons (again, maybe this is not how brains actually work, but I don't think it impacts feasibility arguments). Let's assume Mary learns the exact connectivity of all the neurons in the part of her brain responsible for processing visual input (presumably storing this knowledge in some other larger part of her brain), in addition to all the information about light and her photoreceptors, and has the ability to mentally simulate their behavior. Then she first sees red light, activating some of those neurons for the first time and triggering them to form connections in a way that she could have predicted. I guess the argument is that if she could predict this, she hasn't gained the ability to predict anything additional, and therefore hasn't gained knowledge? But this relies on her ability to predict the impact of those new connections on the rest of her brain too. I think it's too much to assume that a finite brain could fully understand and simulate itself.
"However, it still seems to me that the quoted definition first asserts something exists ("There are..."), and then gives it a name, thus assuming the conclusion"
It's not assuming the conclusion in the sense of assuming anything nonphysical.
It also not asserting the existence of anything you can see. What would it mean not to have qualia? Your food doesn't taste of anything, colours aren't colourful...?
"Do I understand correctly that "must be distinguished from the properties of objects" is saying that, for example, my perception of the color red is distinct from the color property of the red object I'm looking at? "
Yes, but that's a standard scientific claim. Dress.illusion,.etc.
" In particular, the premise of the inverted spectrum argument seems impossible to me because each person would by definition perceive red in response to red light (other than edge cases like saturated photoreceptors)."
Each person would perceive red-for-them, but red-for-Alice could be green-for-Bob.
"knowing that a particular neuron will fire is not the same as the neuron firing"
Physicalists sometimes respond to Mary's Room by saying that one can not expect Mary actually to actually instantiate Red herself just by looking at a brain scan. It seems obvious to them that a physical description of brain state won't convey what that state is like, because it doesn't put you into that state. As an argument for physicalism, the strategy is to accept that qualia exist, but argue that they present no unexpected behaviour, or other difficulties for physicalism.
That is correct as stated but somewhat misleading: the problem is why is it necessary, in the case of experience, and only in the case of experience to instantiate it in order to fully understand it. Obviously, it is true a that a descirption of a brain state won't put you into that brain state. But that doesn't show that there is nothing unusual about qualia. The problem is that there in no other case does it seem necessary to instantiate a brain state in order to undertstand something.
If another version of Mary were shut up to learn everything about, say, nuclear fusion, the question "would she actually know about nuclear fusion" could only be answered "yes, of course....didn't you just say she knows everything"? The idea that she would have to instantiate a fusion reaction within her own body in order to understand fusion is quite counterintuitive. Similarly, a description of photosynthesis will make you photosynthesise, and would not be needed for a complete understanding of photosynthesis.
So: is the response "well, she has never actually instantiated colour vision in her own brain" one that lays to rest and the challenge posed by the Knowledge argument, leaving physicalism undisturbed? The fact that these physicalists feel it would be in some way necessary to instantiate colour, but not other things, like photosynthesis or fusion, means they subscribe to the idea that there is something epistemically unique about qualia/experience, even if they resist the idea that qualia are metaphysically unique.
It does to the same degree it does with fusion. Talking about fusion inside yourself doesn't necessary induce fusion in others. Calling it "knowledge" in case of instantiating vision is just definitional, motivated by values and qualia being in the brain. There is no real value-independent epistemic difference left to explain.
That's the definition he gave in the reducto, though.
"The quale is directly intuited, given, and is not the subject of any possible error because it is purely subjective" ie, "The parts of the human experience that are non-physical".
Those are only the same if you equate "subjective" with "non-physical". If that was the case, then clearly materialism would be wrong as there are tons of subjective thoughts, emotions, sensations, etc. that we all experience.
That is the case, but they are objectivly subjective if you will. If qualia are just the brain comparig sense impressoins, then whats the point of elevating the defenitioin?
forgive my awful grammer, typed one handed as i am bored as hell surfacing some p,ywood
So this is perhaps a nerdy physics post*. (question). So in regards to all the brou ha ha over wavefunction collapse. Well first off you should have the same questions about wavefunction creation. And it just seems to me the whole thing could be understood if we knew how an electron (in some atomic state) absorbs a photon (and goes to some higher energy state.). And visa versa (how an excited electron, goes to a lower state and emits a photon.) We agree that all this happens and having a model for how it works would have to explain wavefunction creation and collapse. But I am, after all, a stupid experimentalist, and I look to those smarter than me for guidance.
*As background I got a phd in experimental physics and spent many happy years working in the field.
My physics is fairly rusty, so take with a grain of salt, but it seems like your picture of "wavefunction collapse" isn't quite right. The name is, perhaps, doing it no favors. "Wavefunction localization" or "wavefunction concentration" might have been more apt, though both also have their problems.
For simplicity, let's talk about a wavefunction Psi being regarded in a basis B with a discrete spectrum of eigenvalues {B1, B2, ...}. Before the measurement, Psi may be in any arbitrary superposition of eigenstates of B. When you take a measurement, and measure the wavefunction to be in eigenstate B', the wavefunction is no longer allowed to be in an arbitrary superposition. It may only be in states Psi = aB', where a is a complex number with modulus 1. But it is still a wavefunction. It has not disappeared, or diminished or gone away. In fact, if we check in some other basis C that is incompatible with B, we'll find it's more spread out than ever--our measurement has thrown it into a state of maximum uncertainty with regards to the eigenvalues of C.
The point is, before and after the measurement, it's still a wavefunction and still behaves like a wavefunction. It still obeys the Schroedinger equation, and other than the discontinuity introduced by the measurement, its time evolution is entirely governed by the Schroedinger equation. There's no corresponding "creation," because the "collapse" isn't a destruction. Indeed, the closest thing to an anti-collapse to be found is what happened to its state WRT to C *during the same measurement.*
Apparent collapse emerges statistically from entanglement and doesn't have to be added as a separate postulate. Decoherence/consistent histories resolve the arbitrary nature of the Copenhagen Cut.
Since you have a PhD in Physics, you've learned second quantization in undergrad and you know that photon absorption/creation is just normal evolution of the wavefunction. So I'm curious why you think this would bring any insight into wavefunction collapse.
Hmm I barely touched second quantization, but if photon absorption is just normal evolution of the wavefunction... well isn't absorption of a photon a measurement? Say the electron is in a PMT and I detect the pulse of electrons following the absorption.
Edit: I will add that clearly I'm missing something in this whole wavefunction collapse business.
> And it just seems to me the whole thing could be understood if we knew how an electron (in some atomic state) absorbs a photon (and goes to some higher energy state.). And visa versa (how an excited electron, goes to a lower state and emits a photon.) We agree that all this happens and having a model for how it works would have to explain wavefunction creation and collapse
It's been a while since I studied this but the short answer (I may muddle it up and use the wrong words) is that photon absorption and emission is quite easily to explain by the ordinary propagation of the Schroedinger equation (you can do the maths on a piece of paper), whereas wavefunction collapse cannot be explained by the ordinary propagation of the Schroedinger equation.
Thanks Melvin, I guess I'll try and find those few lines of math for photon emission and absorption. So if we can write down equations of photon absorption, then I don't see the measurement problem. The simplest measurement I can picture is some single photon detection in PMT or SPAD... it's OK It's been a long time for me too since I did this stuff, but I've been listening to Sean Carroll and he goes on and on about the measurement problem... and I always find myself asking, what's the problem?
There is no wavefunction creation. There is always a wavefunction for every system. Per Copenhagen (I'm assuming you're assuming Copenhagen from the way you brought up collapse) it almost always evolves according to the Schrödinger equation (or some relativistic version thereof) and sometimes it "collapses" onto a projection space.
Particles absorbing/emitting other particles does nothing to change the fundamental nature of this. You merely need to go beyond Schrödinger into its relativistic versions and use the machinery of Quantum Field Theory, which in a very simplistic way involves Fourier-transforming the fields in your Langragian and interpreting the excitation coefficients as creation and annihilation operators for different types of particles. Long story short, calculations are much more complex but in the end you're doing the same quantum mechanics thing of calculating expectation values for a wavefunction which you assume collapses after measurement. (Source: PhD in theoretical physics [cosmology])
Thanks Jose. QM was my least understood subject and hence I have many QM texts as some sort of compensation. Anyway the simplest QM measurement that I could picture is the creation of a photon and then some detection when the photon excites an electron. (This could be in a PMT or semiconductor SPAD) I guess I'm confused by the first statement that there is no wavefunction creation. First there is no photon, then I turn on some switch and there is a photon. If you want to say that the photon wavefunction existed but had zero occupation that would be ok for me. But then I've got questions about going from occupation 1 to occupation 0. (And why is that a collapse?) We used some QFT in a solid state theory course, with creation and annihilation operators... but it's all even fuzzier than QM. If you tell me that it's not possible to understand wavefunction collapse outside the framework of QFT, I could live with that, but be disappointed.
If you use your usual Fock space basis, that's precisely what's happening when a photon appears: you go from state |0⟩ to state |1⟩.
What makes something a collapse is not the end state but the (non-unitary) transformation in violation of Schrödinger's equation which gets you there.
Perhaps the most fundamental misunderstanding here is the assumption that the simple set-up of position emission to electron excitation constitutes a measurement. It doesn't. If it did, quantum mechanics wouldn't work inside the Sun.
The question of what makes measurements "special" is a very interesting fundamental question and the main problem with the Copenhagen interpretation. If you want to read up on this I recommend looking up Zurek's Quantum Darwinism framework, which I find very enlightening (and though inspired by many-worlds is actually interpretation-independent). Long story short, for a measurement you need entanglement between the system being measured, an apparatus, and the (macroscopic) environment. (Basically it's changes in the environment that allow the global evolution to be unitary whilst the system-apparatus evolves (locally) non-unitarily.)
Hi Jose, I'm replying again to this comment. (Else the threads get too deep.) So taking my experiment; Atom emits a photon, I detect it some distance away, with detector of some area. Now the photon can be emitted into 4*pi steradians, so until I detect it the photon is in a super position of all those directions. (Assuming the atom has equal probability to emit in every direction.) And so it's the collapse to the wavefunction representing the solid angle that my detector covers. Which happens when my detector indicates detection. Is that approximately the right picture? You seemed to indicate that it is not. (I didn't understand the comment about QM working inside the sun.) So to make it a 'measurement', do I need to add a polarized filter to the path so that I'm selecting some polarization state? I'll do some more thinking and reading.
(I think it's too late for the depth of the thread. I'm already having to find these comments manually!)
The key question is how your detector indicates detection. As (historically) a Pauli-effect ridden shameless theoretician, I don't know how your detector would work. But I suspect typically it would include a mechanism whereby the absorption of a photon would trigger a chain-reaction of sorts which gave the presence of the photon in one part of the detector a macroscopic consequence. I _think_ typically this mechanism is insufficient to guarantee collapse (and thus count as a measurement) by itself. However, in a realistic set-up you have additional components which complicate this picture: namely a "thermal" environment (including the materials in the detector itself) and, if you want to be extra fancy, your brain. The "hope" of Zurek's framework is that once you consider a macroscopic wavefunction including all these components then normal Schrödinger evolution will cause your photon/detector/environment/brain subsystems to become entangled in such a way that it will look to you like there was a collapse of the wavefunction. Note in Zurek's preferred scenario there is no actual collapse: the wavefunction still includes all the possible states the photon could be in, but each of those components of the state vector will correspond only to one state of the detector and your brain. If I may try to quickly write this down, the before and after for each possible photon state look like the transformation: |i⟩_photon|no detection⟩_detector|thermal state⟩_environment|no memory of detection⟩_brain -> |i⟩_photon|detected i⟩_detector|slightly different thermal state⟩_environment| memory of detecting i⟩_brain. Again, note this is very much the advanced state of the project of the many-worlds interpretation. Many people do think collapse doesn't occur according to the Schrödinger equation, and possibly not even globally unitarily. In one scenario you're left with the difficult question of explaining how then you get probabilities according to Born's rule, and in the other you get the difficult question of explaining exactly how this new mechanism works. (I have this idea to write something about this sort of scientific dilemma at some point, possibly as a sequel to my Neptune and Vulcan posts, but it may be a while before I have time to articulate all of that properly.) Either way, Zurek's work is great at giving you fundamental constraints on exactly how this stuff can work. And basically, at the end of the day we don't really know how any of this works. Pragmatically, it's easier to do most things assuming Copenhagen but it's easier to think about what happens during collapse assuming many-worlds.
The comment about the Sun just points out photons are emitted and absorbed all the time in the Sun, and if that counted as a detection then we'd constantly have wavefunction collapses there and no time for quantum mechanics to do much.
Hi Jose, OK and there I thought I'd made some sort of breakthrough because my reading was that I needed to have more than one wavefunction that the state could exist in and then to choose (or detect) one of them. And that the exact mechanism of what's going on in my detector was not all that important. And now we are back to talking about what goes on in the detector... and it feels like I've come full circle. So let me end this by first saying I can just remain agnostic in my views of the measurement problem. I'm fine with "Shut up and calculate."
And then since my life in physics was mostly about being an instrument builder let me finish by describing what is perhaps the simplest single photon detector. You take a p-n diode and reverse bias it to a voltage that is above the breakdown voltage of the diode. (It's only a little way above the breakdown voltage.) And in this state the diode can be viewed as being a bit out of equilibrium. It would 'like' to breakdown. Perhaps a bit like a super saturated solution. Now a photon comes along and is absorbed in the depletion region of the diode. This creates an electron hole pair, which zip off in opposite directions, accelerate, crash into other atoms, release more e's and h's (holes). And the whole diode discharges. You get a current pulse that is equal to the diode capacitance times the voltage applied. You can detect the current in different ways, but the simplest is just a resistor. You turn the current into voltage which can trigger some sort of counter or be displayed on your 'scope. (From my previous life)
Thank you Jose, I'll check out Zurek. So in my measurement picture, it doesn't stop with the excitation of the electron, that electron is excited such that it is free from it's atom. It is then accelerated (by an E-field) colloids with other electrons in an avalanche process that leads to a large current spike. My electronics amplifies this spike and displays it on my oscilloscope. (or it is counted (with some discriminator) and recorded.)
And yeah all these 'measurement is problem' people go on about entanglement. I've done experiments on photon entanglement and the Bell inequality. In my very limited experience it's very hard to get things entangled. In mean look at the heroic measures these quantum computer guys go to to set up a small number of entangled states. So if entanglement is required for the measurement problem, then why isn't the answer that there is just not very much entanglement in the world?
edit: Oh and this, "What makes something a collapse is not the end state but the (non-unitary) transformation in violation of Schrödinger's equation which gets you there." So is there any similarity to this, and my idea of having some model of how a system goes from one photon and an electron, to a system that has zero photon and the electron in a excited state. And yeah this has to be somehow 'outside' of the Schrodinger equation and QM.
Engagement is weird, but not magic. It may be hard to prepare neat entangled states with exactly the properties you need for useful applications. But in the "wild" most systems will naturally develop a degree of entanglement with other systems they interact with (in particular the macroscopic "environment"). This is just a trivial consequence of the eigenstates of subsystems not generally being eigenstates of the global system. Usually interactions are weak enough that they can be neglected, but they are essential for building the "macroscopic reality" we experience. Measurements are one example of this. Another equally spooky but deceptively more familiar one is the emergence of thermodynamics (which is not yet fully understood in the quantum case, but which will usually play an important role in messing up any neat states you try to prepare).
As for your appended question, you can have unitary evolution which changes the occupancy of Fock states (at least if you don't start with a pure Fock state). I tend to agree that we should expect quantum mechanics to be unitary all the time of you can follow the universal wavefunction. But this is not a universally accepted view.
As an example of entanglement being ubiquitous, the electrons within an atom are entangled all the time. You can calculate what energies atoms would have without this (the Hartree-Fock method), and they're clearly less accurate than calculations that take the correlation/entanglement into account.
You're more of an expert than I'll ever be, but how do you resolve the contradiction between the Schrödinger equation—which governs the system's linear evolution—and the non-linear collapse of the wave function observed during measurement? Do you have an opinion on this, or do you not see it as a fundamental problem that needs to be addressed?
(Of course, decoherence can sorta kinda explain this as a quantum system entangling with its environment. Still, by my understanding, it doesn't explain the old “Born rule problem” — why we see a probability of outcomes. And if decoherence marks where an observer’s knowledge updates—as in the interpretations offered up by Relational QMD and QBism—then we've opened that whole conscious observer can of worms again.)
Am I babbling, or am I asking a reasonable question? ;-)
Oh I'm sure it's a real problem. I just don't understand it. I guess I'm somewhat motivated by a line in the Feynman Lectures where he says something to the effect (I'll probably screw it up.) That now we understand why a photon can be absorbed by an atom, but we don't know how it happens... And I've perhaps naively assumed, that how it happens is related to the measurement problem.
I think what Feynman's getting at there is more that while the rules of quantum mechanics imply that photons are absorbed by atoms in such and such a way, there's no explanation why the rules are what they are in the first place. At least, I've heard Feynman making that point elsewhere and it seems similar to what you said.
The problem with the Copenhagen interpretation is something different though, that there's no precise criterion for what counts as a measurement in the first place. It's just vaguely when the system in question becomes entangled with a whole macroscopic system, or perhaps the world at large, but then what counts as macroscopic, and how much of the world?
Can you say more about what you mean by superposition collapse? I feel like I'll be repeating myself, but you could read my responses to the other repliers
Despite my overall grim feeling about AI, I am using GPT a lot and loving it. I have the $20/month version, which gets me GPT5. It puts itself into “Thinking” mode, which uses more compute, if I ask something pretty demanding, such as a summary of research on a certain topic. I can also choose to put it in that mode. Here’s where I’m finding it highly useful.
-Photoshop help. I’ve used PS a lot, but never learned about it systematically, and often have trouble doing something new because of my lack of big picture understanding of PS. ’ve now had a series of exchanges with GPT where I ask it how to do a particular thing, then follow up with questions about why that works, what alternative techniques there are, whether a technique can be extended to use for an entirely different kind image-tweaking task, etc. So I start somewhere specific, but then burrow outward in several directions. Doing several of these explorations had led to a very large jump in my ability to accomplish things in Photoshop. I still have substantial areas of ignorance, but large areas that used to be dark are now floodlit. About 10% of the time GPT’s how-to’s simply do not work, and when I complain it gives me another how-to that also does not work. I think that happens when it can’t really answer my question, but the people-pleasing drive makes it hallucinate plausible series of steps. Those incidents are quite frustrating, but the successful ones make it well worth it to suffer through GPT’s occasional bullshit attack.
-Medical questions. Examples:
1) Someone I know is getting immunotherapy for cancer, and having trouble tolerating the treatment. I ask GPT for info about novel or overlooked techniques for helping people with the immunotherapy side effects, and fit found several. I skim the articles and none are obviously sketchy, and in all the approaches described did indeed work much better than the placebo. In the course of looking at articles, run across one stating that immunotherapy given earlier in the day is substantially more effective than when given later. I pass this info on to my acquaintance, and their oncologist verifies that this is indeed the case, and now the person gets immunotherapy early in the morning.
2) I want to get Zepbound cheaply by ordering it from a lab selling it as a research chemical, as described in Cremieux’s blog and elsewhere. But I am nervous about whether this approach is really safe, so ask GPT to tell me all the ways it could be unsafe, then help me check to see whether any of them apply to the exact way and exact lab I would be using. Delve into various related matters, and ask the research chemical lab some crucial questions. Eventually decide approach is safe, and am now doing it.
3) Someone I know has a dog that developed big cauliflower-like growths in its mouth. Vet said it was papilloma virus, not a big deal, and growths would shrink on their own over the course of a few months, but until it does dog can’t safely socialize with other dogs. Owner’s distressed. Asked GPT for best current approaches to this problem, it sent me to research showing a certain drug greatly reduces time for the problem to resolve. Skimmed research article and there was nothing obviously wrong with it. Owner got the drug from the vet and now, about 3 weeks later, growths are nearly gone.
-Big loose searches. Someone I know has an impressive skill set but does not like the job where he’s now using it. I ask GPT for a list of organizations of any kind that might need someone with my acquaintance’s skills and has features on acquaintance’s wish list: Helps other people in a pretty direct way, one works as part of a team with a shared goal, can be anywhere in world except for a few specified exceptions, prefers to work in or near a certain country, demands inventiveness, minimum salary. It produces a list of 30 or so organizations, about half of which I’ve never heard of, all of which sound impressive and plausible.
Some safety procedures one should use for superintelligent AI.
1. Only give the AI finite goals. So ask the AI to “earn 1 billion dollars” rather than “earn as much money as possible”. If you asked for as much money as possible it might take over the world and turn everything into money. But with a finite goal, even if it does something horrible it hopefully would stop once it had earned a billion.
2. The AI should self-destruct when completing a goal and be OK with that. Now probably the people who own the AI don’t want to destroy the computer or erase the AI program from it. So instead one should make the AI think of itself as the program being run one time. And if the program is being run again that’s a different AI.
This helps the AI only have finite goals. Otherwise if you ask the AI to earn a billion dollars it might think you will ask for more money in the future and so it might start to work at getting as much money as possible, rather than just one billion.
3. Always give the AI an alternative task. Example: you have an AI that can talk to you, read the internet and trade stocks. It can’t do anything else. You ask it to either earn a billion dollars or to make sure a red button in your office is pushed. You think it is impossible for the AI to push the red button, as it doesn’t have any arms, so it will have to trade stocks and earn a billion. The AI then hypnotizes you into pushing the red button.
If you hadn’t given the AI the alternative task something more disastrous might have happened. The AI might have hypnotized you into kidnapping a billionaire and then hypnotized the billionaire into giving you a billion dollars.
4. The AI has to explain what it is planning to do, and you have to approve the plan before it can do it.
5. The AI can’t lie.
6. The AI can’t hurt people.
7. The AI can’t break the law. But be careful with this, someone could change the law to manipulate your AI. Congress might pass a Law that says “All AI must make everybody in the world the obedient slaves of Congress”.
8. All the countries in the world should be united into one country so they can more easily enforce AI safety standards. I realize it is extremely difficult to unite all the countries in the world, but it is still easier than having lots of countries and making sure all of them follow AI safety protocols.
9. If someone who is not entirely moral is working on AI they should be stopped, or they might deliberately make an evil AI for selfish reasons.
One way to earn billion dollars is to start doing something useful repeatedly, gain a monthly income and save money. For example, the AI may offer to navigate cars, and will get paid $1 per car per month. Gradually it takes over a region, and collects 1 billion.
I like #2 and 3, and have not heard those ideas before.
I don’t think it is possible to insist that AI observe 5, 6 and 7. In real life some circumstances lead to even the most scrupulous person doing each for the greater good. For ex, consider doctors. They certainly cause pain, and sometimes harm or kill people with procedures that are the most likely to help, but do fail a certain percent of the time. There are similar examples that show why one cannot demand perfect adherence to 6 and 7.
Big fan of option 3, which is reminiscent of an old idea of mine, which was basically to try to hard-code in a value that looks like "press this red button in my office", except more visible - my idea was "paint the Golden Gate Bridge green". If we could make that a core goal (and I'm not sure we can), then an unexpected success on that goal gives us a warning shot - the minimal agent that can paint the bridge probably can't yet defeat humanity, but it seems fairly on the path to doing so.
Less optimistic on 7, 8, and 9. On a meta level, "These are some proposals I've come up with about how to make AI safer" >>> "safety procedures one should use". You don't get to use the latter language until you're an actual master of the craft, which no human is.
I hope it’s not somebody saying that ASI is super smart and so would have thought of all the ideas above and prepared in advance some way not to be deterred if a member of our species attempts to implement one. I don’t think someone gets to just make that move over and over.
Why don’t we use controlled parasite exposure as a medical intervention for weight loss, at least in extreme cases?
If you had something that could eat your excess caloric intake every day before it digests *for you*, you wouldn't have to throw up, you could just eat. And you'd be liberated to indulge your love for food to your hearts content.
Imagine if we could selectively breed a type of tapeworm that never grows beyond a safe size (or maybe you just kill and replace it before it overgrows with a simple pill), eats a predetermined number of excess calories per day, and is otherwise simpatico with the human gut.
I think this has significant advantages in theory to GLP-1RAs. GLPs make eating less pleasurable, which for many people is a core ingredient to a happy life. The parasites would if anything make eating *more* pleasurable, because they want the calories as much as you do. And GLPs in the best case only achieve about 44 lbs of weight loss before you reach the maximum dose for someone of my bodyweight anyway. Most obese people are more than that many lbs overweight.
Considering parasites can kill people by starvation in the extreme cases, I'm guessing you can achieve arbitrary, unbounded weight loss: turn someone from 400 lbs to 150 lbs if desired. GLPs can't do that.
It’s cheap, passive, doesn’t require any active thought on the part of the patient. Adherence rates would be high, because you don’t have to dose once or twice weekly; you’d probably just eat your parasite cookie at the clinic and come back a year later (depending on how much it turns out you can optimize the growth and eating characteristics of a parasite using genetic engineering or selective breeding).
Maybe you have to “sterilize it” / take away its ability to shed eggs/reproduce using a drug or by removing its egg producing organs. And okay, maybe that’s hard, but we’ve done much harder things than this with probably much more R&D than would be required to figure out some simple trait selective breeding & neutering surgery.
I get that there’s probably an insurmountable yuck factor here, and no one is going to be pitching this on Shark Tank—although maybe this will be overcome if someone can credibly advertise this as a kind of robotics instead of a parasitical organism, by creating a nano “gut health booster” or whatever that “monitors your gut for excess calories and never lets you overindulge”.
But like, I know of at least two personal anecdotes where a morbidly obese person underwent an elective surgery to lose fat that had about a 50% survival rate—so the FDA can’t tell me controlled parasitical exposure as a weight loss intervention is too extreme/there is no amount of morbid obesity that could ever merit something so reckless.
And hey, they’re already doing this in light of the hygiene hypothesis of autoimmune conditions, right? We evolved to coexist with parasites in the ancestral environment, and without them our immune systems overreact to normal things. The parasite load our gut biology adapted to deal with is not zero, and it seems plausible to me that that’s actually responsible for a lot of “evolutionary mismatch” related illnesses.
One piece I haven't figured out yet is how a Pharma company would actually make money from this, though; can you patent a selectively bred animal? Does someone have a "patent" on the Golden Doodle dog breed? Presumably not, right?
My friend objected that "I think the big issue is, energy has to go somewhere, regardless of who's doing the food absorption. [In a simplified sense] In humans, it goes into fatty deposits or activity. In parasites, it typically goes into growth and reproduction, both of which are limited in your hypothetical. So there's really no way for this tapeworm to consume the calories on your behalf and like, do nothing with it."
So yeah, if the worm has nothing to invest the calories into other than growth, it will balloon in size quickly and have to be removed. But egg production is highly energy intensive; maybe you could treat it with something that would make it so that the eggs it produces are obligate sterile, in which case it would spend the vast majority of its energy producing effete eggs that would be trivial from the standpoint of human health. Normal tape worms take years to grow to huge sizes. (Ofc maybe they'd grow more quickly if given to people who eat way more than the avg person, so they may have to be cycled out on a faster schedule than this.)
That is an actual plot point in Dungeon Meshi. Succubi saliva contains a "poison" which induces euphoria in victims, which ultimately results in these monsters getting hunted to near-extinction for their fluids. One group of entrepreneurs gets the bright idea to keep the operation sustainable by recruiting for "succubi hunts" and then feeding the recruits to the succubi, providing them with a reliable food source in order to keep their population stable.
> But like, I know of at least two personal anecdotes where a morbidly obese person underwent an elective surgery to lose fat that had about a 50% survival rate
> GLPs in the best case only achieve about 44 lbs of weight loss before you reach the maximum dose
I don't understand. You take GLPs, you keep losing weight. To lose more, you don't need a higher dose, only more time.
I took GLPs for one year, lost about 15kg, stopped taking it for a year, the weight remains constant. I assume that if I took them for one more year, same dose as before, I would lose maybe 10kg again.
> turn someone from 400 lbs to 150 lbs if desired. GLPs can't do that.
It might take ten years, but I see no reason why not.
Tapeworms used to be advertised and sold as diet medication in the late 19th Century and early 20th Century. Dieters would intentionally swallow beef tapeworm cysts or eggs, usually in the form of a pill, in the hopes that the parasite would grow in their intestines and absorb a portion of their food, leading to weight loss. That didn't work out so well for many of the dieters.
Pills containing tapeworms have been sold for this purpose, but my understanding is that any significant parasitic infestation will likely make you lethargic and mentally foggy (and possibly crave more food to make up for the deficit) so they don't work as advertised/have undesirable side-effects.
You don't know the pressure doctors put on fat patients where diet does not seem to be working. "Try bariatric surgery!"
I think health ministries have told doctors to push this in order to treat The Obesity Epidemic which is taking up so many resources in healthcare (apparently) but I've had to refuse kind urging to undergo such surgeries three separate times.
Operation has good chance of killing you? Decent failure rate? Dependent on success afterwards that you stick to the diet and exercise programme, and if you were able to stick to the diet and exercise programme in the first place you wouldn't need the operation?
Ah, but consider: you're fat! That will kill you anyway, so why not give 50% chance of survival a go?
Okay, if that's not the operation OP meant, I take the correction. It has gotten safer over time, but there's always a risk with severe surgeries of this type.
50% survival rate is nonsense. Bariatric surgeries have a mortality rate of less than 0.1%. For reference, that is lower than the mortality rate for the most common abdominal surgery (cholecystectomy).
Could a specific person have a 50% predicted morality rate? Probably also nonsense. Even big trauma surgeries in critically ill patients have a better survival rate than 50%.
I've had the same thought for years, ever since I watched that one House episode. I think to first order, it'll simply work as described, but I expect your victim ("patient") will get ravenously hungry and be unable to sate that hunger. I also expect he'd have problems with muscles atrophying, along with all sorts of nutrient deficiencies (though this last bit might be preventable by engineering the "diet buddy" a bit more). You could achieve this effect more easily simply through starvation.
Tapeworms for weight loss and maintenance was a plot element in a SF story from the 70s I read ages ago (can't remember name of story or author, but this was the solution for 'we're rich and overeat processed foods but don't want to put on weight').
Haha. There's a chubbyemu video on a case of woman who procured tapeworm eggs for weight loss. It worked but it also wrecked her health as the eggs got into her brain.
The problem with parasites is that they don't stay in one place in the body.
You acknowledge the obvious horror and safety issues, which are plenty to answer your question. But to add to that, you can control the dosage of medicine. You can't control somebody's intestinal parasite.
Also, "this could hypothetically cause unbounded weight loss" is just that, a hypothetical. I have a hard time imagining you can have a parasite consume so much inside a person that it causes them to lose 250 lbs and have that be remotely healthy in any way.
Also, I don't think it's really clear that this would even have the desired effect. Ozempic maks people not crave eating as much. Having a parasite doesn't do this. You seem to be assuming that this means people will eat the same amount, but not feel hungry. But it could just cause them to eat way more to reach satiety.
Also I know people have a lot of arguments around how bad obesity really is, and I won't pretend to have enough expertise to adjudicate them, but I think one might fairly ask, "Is being overweight SO VERY BAD that it's worth feeding a tapeworm hundreds of pounds' worth of calories?" Because that seems pretty crazy. Your only real counterargument is "well, people do surgeries with a 50% mortality rate" (really? would like a citation here) but obviously nobody should do that, either.
"Your only real counterargument is "well, people do surgeries with a 50% mortality rate" (really? would like a citation here) but obviously nobody should do that, either."
Here's a video telling us how safe bariatric surgery is, from a bariatric surgeon. Who looks like she should be getting such a procedure herself, so if it's so great, why doesn't she get it?
The main reason parasites cause weight loss is not that they consume calories, it's because they make you sick and cause unpleasant gastrointestinal symptoms. The amount of calories they consume is minuscule.
They also have other negative effects such as micronutrient deficiencies, damaging organs via several mechanisms (for instance causing obstructive cholangitis and pancreatitis), as well as migrating out of the intestines. Not to mention the risk of spreading them to other people.
Basically this is an awful idea. Its a form of bulimia that is even worse than the usual forms.
And (unless I'm missing something myself), it should be obvious that something like this is going on, because otherwise how is the weight loss happening? If the host was just losing fat, sure; but when they're literally losing weight, it makes no sense to assume the excess calories/atoms are somehow disappearing without a trace into the thing that lives inside them (including when they're standing on the scales).
In theory, the parasites could just have a very inefficient metabolism where they turn lots of calories into lots of waste heat and a very little bit of increased parasite biomass. The amount of waste heat associated with burning say a pound or two of fat in a weak, isn't going to seriously impact the heat balance of a human body. A pound or two of fat a day would be another matter.
But "parasites make you generally sick" is a much more plausible explanation, I agree.
A follow-up question (which is probably dumb, but I think you or someone else here would be good at explaining the answer to me): in this scenario, what ultimately happens to the atoms that the food consisted of? There's no actual matter->energy conversion going on in digestion, so 'atoms in, atoms out' has to hold, and the release of waste heat can't fully explain weight loss/absence of weight gain. Do the parasite and the person simply defecate more (in which case maybe we're back to having noticeable symptoms), or do other means of egress (urination, sweating, exhalation) play a meaningful role here?
Normally, nutrients ingested in the GI track are absorbed into the body, where their atoms are rearranged at need. Some are incorporated into the body's structure (protein, collagen, etc), some are stored as future energy reserves (fat), and most are used for short-term energy production which means they ultimately get converted to CO2 and water, to be exhaled and/or excreted. And really, the protein synthesis etc will be less than perfectly efficient and you'll get still more excreted waste products. But not all of it, so you can get some weight gain in the form of protein and fat.
In the hypothetical (and unlikely) case, those nutrients are hijacked into the parasite's body, where they are used very inefficiently so mostly converted into water and CO2 and flushed out the back end of the GI tract, so no weight gain where there would have been some weight gain. Or actual weight loss, if the parasites snarf up so much of the good stuff that the body has to burn its fat reserves to keep things running.
You're right that "but where to the atoms actually *go*?" and the related "where does the energy actually go", have to be considered any time you're trying to figure out how weight gain or weight loss. And this is particularly easy to get wrong where the gut microbiome (whether symbiotic or parasitic) is concerned.
Indeed, I've heard of tapeworm eggs having been used in this fashion, but when somebody up-thread mentioned 250 lbs I thought: "Hang on - a tapeworm can't eat that much. They are famously thin!"
I keep hearing complaints that people waste more time because of LLMs, but my experience is the opposite. There’s no distraction algorithm, no feeds, ads, or autoplay, so I stay in a closed Q&A loop instead of tab-chasing. Therefore, I drift less. Do others here have a similar or different experience?
I definitely waste a lot of time chatting with ChatGPT. I'll present it with long, sloppy arguments presented in inflammatory language, and then when it objects that I am "wrong," I attack its reasoning until it admits but it cannot definitively call me "wrong"
In some vague way, I guess this is practicing debating or something, but the net benefit to me is still nil.
Thank you for your response. How would you compare that to googling the topic you are arguing. Do you think of googling that topic as worse or better? For example, googling the topic may lead you down a rabbit hole or it may lead you to find out something new.
The good part is that unlike Stack Exchange, there are no ads, and no attempts to make me read unrelated Stack Exchange questions (e.g. about politics, computer games, etc.).
The slightly bad part is that the LLMs are too verbose, so I easily end up scrolling long texts. (Saying "be concise" only helps a little.) This requires some habits, for example after programming I delete the chat; and sometimes I delete it and start anew even in the middle of work. Otherwise I would drown in the automatically generated content out of which only 10% is relevant.
Self-control is needed, because it is very tempting to ask an additonal question and immediately get a response, but then you have two extra screens of text you need to scroll through.
When I'm using the LLM myself, it helps me learn and solve stuff quickly (as others have replied). They sometimes hallucinate stuff, but it's usually pretty quick for me to figure out that it was hallucinated.
My biggest time waste with LLMs is when people upload stuff they spit out onto the internet and try to pass it off as original writing, and it sometimes takes me a bit to realize that it's useless slop (not that they didn't make slop SEO garbage before LLMs, but it's more common now).
I've disabled the browser on my phone to make it less distracting (along with no social media, games etc). But having ChatGPT lets me look things up without getting distracted. Having the AI as a mediator between me and the entire internet works really well.
It's like an ultimate Google destroyer, because Google's entire business model is putting advertisements between you and the information you are looking for.
I think it helps me learn more quickly and solve questions more quickly. If it’s important I will then delve into better sources and start doing my own research. I would say it’s good for finding knowledge branches I would have overlooked. It’s good for my analysis paralysis as well for more trivial things. My main issue right now with chatgpt5 is that it’s always trying to give advice and tell me what to do and I just want relevant information always and to make my own decisions.
Ed Nevraumont, author of the Alpha School review, has a blog chronicling current events in the life of his daughter Everest, aged 9. She just won the World History Bee competition. Here’s a snippet from the blog:
>We got back from the History Bee Nationals on Tuesday June 27th. Up until that day Everest was 99% focused on being ready for the competition. But on Saturday June 31st she was scheduled to give a TedX talk on her History Bee experience. Before we got to Nationals she had pulled together a rough script outline and had got some feedback from the TedX team, but she only had four days to refine that script and learn it well enough to perform in front of a sold out audience. . . .Everest took a week off after Nationals to get ready for the TedX talk, and it has resulted in a struggle this week trying to get her review cards back down to a reasonable level. We are going to try and avoid that happening again. She also decided that she wants to compete in the National Geography Bee next year. I am not sure how we are going to balance studying for two topics at the same time.
Here’s a snippet from the life of another highly gifted child. It’s Cory Henry, aged 4, rehearsing at his church.
As a kid I did the geography bee (I made the top-10 at the state geography bee when I was in 5th grade). I was also very involved in quiz bowl. I was a voracious reader and I loved my World Book Encyclopedia, but I hated the idea of studying specifically for quiz bowl (I didn't really like to study for class, either). It's much easier to study now because there are so many more resources available, but I can't bring myself to grind out flash cards on Anki or quizlet or whatever just so I know a few more things for trivia competitions. (Yes, there are actually adults who study for trivia.)
I find it hard to imagine that many nine-year-olds really enjoy studying. Learning new things, sure. And maybe, you can gamify the learning enough to keep it entertaining--I assume that's part of why chess prodigies are willing to grind out memorizing so many lines of the Ruy Lopez or the Sicilian defense. But I can't imagine the process (especially when it's parentally-driven) as all that pleasant.
Something of a false dichotomy here. Why a description of the workload of one during an abnormally tight crunch period against a video of the other performing?
That’s a fair point, mostly. (Though Cory is actually not performing. He is at a dress rehearsal, which I think would count as homework. He also worked and practiced at home.). Still, I agree that the situations is not comparable to the one Nevraumont describes in his account of Everest.
But here are some differences between what the 2 kids are excelling at that I think are fair.
-Cory is learning to produce tunes and rhythms. That’s something that it is easy for most kids to enjoy. It’s quite possible that he often enjoys the lessons and practicing he does at home.
-Everest is learning factoids such as dates and royal successions off flash cards. She probably enjoys being congratulated at working so hard at the task, but it’s not easy to believe she enjoys the task itself.
-Cory is learning to do something that, when he’s older, he can do over and over and over, and will embed him among other people who enjoy the same thing. Even his 4 year old version of the activity embeds him among others in a good way. He’s giving pleasure to those present
-Everest is learning a body of material that will only be performed once, at the contest. It will astound those watching, but it’s not clear that that will give them much pleasure. In fact all the kids she beats and their watching family members will feel bad.
-Cory is learning skills that will open up new ways for him to play around and experment. He can improvise, compose, try different ways of weaving his sounds into the mix made by other musicians.
-Everest is learning factoids that may be of some use in understanding and enjoying history — though of course factoids like dates, who succeeded who to the throne, etc. will appear anyway in any book she reads about history. In the future her mnemonic feats might entertain people at a party now and then, though performing them more than once in a while is definitely not a great way to get to know people.
Nevraumont addresses your argument about “learning factoids” in his post titled Memorization, Trivia and Atomic Units for Creativity. You don’t say that children should not be engaging in high level competitions, but your arguments point in that direction so I think the post titled Learning to Lose is quite relevant.
I’m not really defending Nevraumont here, in part because I don’t know whether he is a reliable narrator, but the TedX Talk post you quote from is clearly an update intended to be read in the context of other stuff he has written, not to be a stand-alone blog post. As such, it tells us that Everest is doing a lot of stuff but doesn’t provide the context to see if that makes sense, so if you are going to engage with only one of Nevraumont’s posts, the TedX Talk post is a bad choice.
If I had ever heard anything about Nevraumont that made me admire him, I would take the time to look at the posts you link. If, for instance, I loathed something Zvi had said or done, I would delve more deeply into the subject, doing things like searching his blog for mentions of the topic. But I don't think I owe that to everyone whose behavior or ideas I object to. I read Nevraumont's review, I read his LInked In, I read up on Alpha school and the current head of it. That's enough.
<You don’t say that children should not be engaging in high level competitions, but your arguments point in that direction.
No, I'm not thrilled by the idea of kids in high level competitions, but can imagine that in some circumstances, for some types of competitions, that would work out OK for some kids. The element I'm most bothered by is the time investment. To compete at high level competitions -- in fact even to be allowed to participate in one -- the kid has to devote an enormous amount of time to training. Most training is dull and repetitive, and I don't think it's good for a child to have many hours per week tied up in an activity like that. The whole thing might be worth it for the kid who wins a medal, but for every one who wins there are thousands who train nearly as intensively, and lose. As for the benefits of learning to lose -- sure, kids need to have that experience and learn from it. But that experience not exactly hard to come by. There are plenty of competitive sports experiences and academic challenges available at schools and other local settings, and even the most talented kid is guaranteed not to be the best at all of them. And losing in one of those settings seems to me like about the right size loss for a kid whose age is still in single digits. Losing after a year of grinding training several hours a day -- missing playdates, ordinary games and sports and peaceful downtime in order to grind -- is a very harsh blow to bear.
I'm also bothered by the topic on which the kids are competing. The questions about history have to be stuff like dates, capitals of countries, which ruler succeeded which other. The organizers have to be asking questions where there is a clear correct answer. You can't be asking the competitors, especially child competitors, about something that requires insight -- things like, "what steps, if any, might have kept the war from breaking out?" So the questions must be about factoids, things that are mostly not interesting in themselves. but are just pins to locate events in time and space. Adding a bunch of those factoids to the kid's memory does not seem to me much more enriching than having the kid memorize the first 1000 digits of pi.
And I'm bothered by making competition such a big part of a kid's life. It's a valid and useful mode, but so are a bunch of others. I'm sure Cory Henry had to compete some. He ended up attending Juilliard, I believe, and you don't get in there without winning out over other young musicians in various settings. But while learning musicianship as he child he also was learning something the involved a lot of other modalities -- collaborating, improvising, giving pleasure to others.
Also -- do you think her father isn't deeply invested in her winning and winning big and being known for winning big? He named her *Everest,* ffs! And do you think Everest can't sense how invested he is in her winning and winning big and being known for winning big? By the time I was 9 I could tell on trips to the store whether my mother was worrying about money that day. I could tell when she was annoyed at my father, even if she denied it when I asked her. If she had been privately hoping I would be the smartest or most athletic kid in the state, the country, the world, I would have known that.
Yes, the questions must have a clear correct answer. But there are a lot of ways to do that without resorting to complete minutiae. I don't know all the particulars of all the various History Bee competitions, but I know they have a team-based quiz bowl style format, which uses toss-up questions (the ones that you use a buzzer for) that are designed to be pyramidal; that is, the hardest clues come at the beginning. So you can reward someone with deeper knowledge of George Washington or the Polish-Lithuanian Commonwealth or Diocletian. And you have to have some understanding of basic facts in order to grasp the broader concepts. So I think knowing things about history is vastly more important and interesting than knowing the 789th digit of pi, which is purely a memory exercise and has roughly zero practical use.
> No, I'm not thrilled by the idea of kids in high level competitions, but can imagine that in some circumstances, for some types of competitions, that would work out OK for some kids. The element I'm most bothered by is the time investment.
I agree.
When I was a small child, people encouraged me to try chess competitions, because I seemed smart, and chess was the archetypal smart people's game. But I soon realized that this was a trap: yes, it helps in chess to be smart, but it still requires enormous amounts of *time*, and the chess skills are practically useless for anything else. So I refused.
I did math olympiads though, because I enjoyed doing math. I did not specifically prepare for the competitions -- I just tried to solve all kinds of mathematical problems I found, and the victories at competitions happened mostly a side effect of that. I knew kids who were specifically studying for the math olympiad, but I was happy I wasn't one of them, and I even considered it a kind of cheating (similar to when you try to memorize the right answers for an IQ test).
> The whole thing might be worth it for the kid who wins a medal, but for every one who wins there are thousands who train nearly as intensively, and lose.
This is true, but I find it suspicious that people usually use this as an argument against knowledge competitions, and almost never as an argument against sport or art competitions.
I guess the idea is that sport and art are valuable per se, so the kids who spent their afternoon practicing sport and failed at the competition, still did something meaningful. From my perspective, doing math is also meaningful... and perhaps for someone, so is memorizing historical facts.
It probably means that people should only compete at things they consider intrinsically useful/enjoyable anyway. It sucks when ambitious parents are pushing their children to do that. But some other children do find the activities enjoyable. Is there a solution that would be good for both sides?
Child 2 looks happy and resplendent on the keyboard. I dunno. As long as Everest is happy and engaged and not stressed, I don't mind too much. But I'm wary of parents trying speed run childhood for the sake of genius. Children are special but adulthood is great churner of mediocrity.
I don't know if it's my limited selection of examples but genius kids rarely sustain the momentum into adulthood. i see many kids who graduate college early only to be another banker or another engineer. It feels disappointing to speed up your childhood only to end up another worker bee. Feels pointless but that's just my sentiment.
Graduating college 1-2 years early is approximately as impressive as becoming a banker or engineer. I think you're simply overestimating what accomplishments in childhood are indicative of "genius."
The reason why speedrunning childhood is good (and this will vary by personality, of course) is that childhood sucks. Who doesn't want to be independent earlier, lead your own life earlier, if you could?
My childhood circumstances were not ideal, but my childhood sure didn't suck. In fact I'd say that period was the happiest and most engaged I've ever felt. I don't think my experience is unusual. Look at kids playing -- most of their faces are lit up with joy and excitement.
Of course kids have fun, but adults have so much more and better fun.
Consider food as a small example. When you're a kid, you can have a great time eating some candy, if your parents let you. When you're an adult, you can still have a great time eating some candy, whenever you want, but you will also have the palate, money, hand-eye coordination, and autonomy to enjoy all kinds of food. You can fly to Belgium and sample fancy chocolate if you want. Even if you're not very wealthy, there's way more stuff you can enjoy.
The whole world is like that. There is so much beauty that most kids don't yet have the capability to appreciate.
You see freedom, I see the tedium of responsibilities and work, and then add on taking care of kids and health issues and aging parents. Yeah sometimes I wish I was 14 years old again, with nary a thought on paying rent or affording health insurance. Parents were my loving Gods whose most annoying demands that were that I clean my room. Now there are no loving affectionate Gods who care for me (May the Gods afford me a billionaire gentle husband, amen!). But there's a cold world demanding that I pay taxes and would not spare a thought if I don't survive the next round of layoffs.
Honestly I like it better being an adult for reasons you lay out, and also mostly because my parents sucked. However, I only came to like adulthood after I have had earned a sufficient level of economic success. Supposedly I am still toiling under a job that barely pays my rent, you bet, I would be weighing going back to be under my parent's cruel thumb just to offload the economic anxiety.
Yeah. Similar here (less dramatic version thereof).
In theory, I have the freedom to do... anything. And maybe some more talented version of me could find a way to enjoy these opportunities.
But the actual version of me spends 8 hours a day at job, trying to survive in chaos that makes me entirely mentally exhausted when the work is over. Then it's care of kids, various domestic tasks, various random problems. An hour or two left to do things that I consider meaningful, but I am usually too tired to enjoy them, so I just read something online or watch a movie, and go sleep. The next day is the same. So is the day after that. The last decade was like that.
My childhood wasn't ideal, but I could spend an afternoon relaxing and reading a book, without being interrupted hundred times, and without feeling guilty that there are dozen other more important things I should be doing instead.
As a child, you don't have the freedom and money. As an adult, you don't have the time. As an old person, you don't have the health.
> Who doesn't want to be independent earlier, lead your own life earlier, if you could?
Most of the milestones of adulthood independence, like being allowed to drive a car, are gated on reaching a certain age rather than just finishing your education.
I just discovered "Cannons" and I love them. It's very smooth, like R&B, except everyone is white and they're singly sadly about the all the sex they didn't have and also there's lots of synthesizers and spaceship noises.
Dennis Lloyd might be the greatest artist I've ever heard. Every song should be absolute garbage. Wistful white guy with long hair sings about California while playing his guitar. I should abhor this and yet I'm listening to it, right now, for like the 6th time today. It's like kimchi: every element is absolutely disgusting and yet the final product is delicious.
Check out "Oceanside", "The Way", and "Nevermind".
If you haven't listened to Chris Thile's Bach albums, you really really should. Possibly the best to ever do it on the mandolin, one of the best living musicians from a technical perspective, and definitely the best I've ever seen live.
For your reconsideration: "Sadly, Porn (But Reviewed Good This Time) from the 2024 ACX Book Review. (1) (2)
This was submitted but never got out of the initial rounds for several excellent reasons: it's like 90 pages long, far too clever for its own good, and it's trying to replicate the Last Psychiatrist style which is...not recommended. Still, this is one of the best distillations of Lasch/Last Psychiatrist/Samzdat-Lou Keep I've seen and it's still in my brain two months later.
This badly, badly needs an abbreviated version. However, I would recommend skipping from p 131 (where the essay starts) then reading pages 140-151, then skipping to page 182-184. (3)
Golden excerpts:
"The Last Psychiatrist used to give an example of how psychiatrists fuck up their kids; they want to be rational, modern parents. They have the benefit of a psychiatric education, after all. This means they don't shout at their kids, even when they do wrong, like bullying another kid. They will shout at them when their frustration boils over as it must do after repeated wrongdoing, which is particularly likely to happen when the child does something that inconveniences them as opposed to something that's morally wrong.
The young child only responds to affect, to the emotion you're giving off. He hasn't internalised the sense of others being people, so when you calmly explain why he shouldn't do something, he doesn't get it. He doesn't interpret it as a real punishment. He does get it when you shout at him, but now you're doing it over something arbitrary, like spilling some milk.
What happens next is the child cries and the parent realises what they've done, and then they turn around and console the child. This is the fatal step.
What the child has learned from this process is that:
Your authority is arbitrary
You can be placated for them to get what they want
All they need to do is figure out what will placate you.
A harsh, even abusive parent who merely beats their kids for no reason is sure to give them some kind of pathology, but they won't get the second and third parts so this won't create narcissists. It's this unstable emotional dynamic: at once nurturing then punishing without a predictable underlying logic that could allow them to develop a clear understanding of right and wrong.
Imagine someone who is going through their adult life re-enacting this dynamic, they see themselves dependent on this omnipotent power that has the power to dictate their life and they have to behave with this in mind. The real problem is that facing up to this consciously would involve living with a constant sense of anxiety. It would be too much for them to bear, particularly for the fragile psyche of a child. So they have to find a way to determine and accede to the demands of the omnipotent power whilst avoiding the reality of this fact: this can only be done through a colossal amount of doublethink.
"
"
The modern life of individuals is characterised by living within the shadow of a huge number of large bureaucratic organisations. We are born in hospitals, learn in schools and universities, work for corporations, get incarcerated in prisons, and finally return to the hospital to die. In our leisure time, we consume products created by vast industrial conglomerates around the world, and the cultural products that we gain meaning from are creations of an industry as large and as powerful as any manufacturing concern or government agency. The guiding philosophy of all this, in its various forms, tells us that these organisations exist for our benefit, and certainly many of them are staffed with generally well meaning people. But this is a unique situation in the history of the world, and this state of being has deep ramifications for the kind of lives people live.
The relevance of the previous sections is that these institutions treat us like Narcissus’ parents: they foster dependency without development, they protect us only intermittently, and we labour under demands which are rapidly changing and inconsistent. Lasch terms this “Paternalism without Fathers”: parenting without love.
"
"
We should frame an important question; is Narcissism bad?
Bad for who? Narcissism is adaptive to the conditions of modern life, even helpful to the individual, although coming at the expense of their psychic wholeness. Systemically it’s a problem, which just means it's a problem for other people, not such much for the individual. They’re just playing by the rules of the game.
"
"
The consequence of this; your job is not to do your job. Your real job is to get your next job, your promotion. Maybe you could get this promotion by performing really well in your role. Except, your performance needs to be recognized as such by your superiors. Therefore, what is really important is to be noticed to be doing well. After all, the role you will be promoted to presumably has at least some different responsibilities to your current one, so your performance in one role is only a partial match to the other. Indeed, the core responsibilities of your job may be an active impediment to your success since they take away your attention and energy from the important work of getting promoted. Possibly, doing too well in your core role will actually retard your promotion because they don’t want to lose the contribution that you’re making in that position. And this is not to mention the more obvious and unsavoury aspects of promotion culture; brown-nosing and conformity.
The kind of person who will succeed best in this environment is the one who can recognize this intuitively, someone who can devote their whole life to the maintenance of appearance, with reality and action as an afterthought.
"
"
The kind of social narcissism we’ve described is a product of living under the constant authority of bureaucratic institutions, entire lives lived with no other way of being. People identify with power as a defence against the acknowledgement of their own potential helplessness in the face of these powers. But the anxiety of this state doesn’t disappear, it just gets redirected.
My background is on the Left, that’s the language that I speak. I know so many people who think “free speech” is bullshit, they want the government to take more steps to suppress those with views they see as dangerous. These people are also supposed to be “radicals”, people who claim to want to overthrow the system. Of course, they don’t expect that power to be used on them, only on their enemies. They would rather fight fascists than the government, it’s a lot easier I can tell you. However stringent their rhetoric is they always give succour to the state of things as they are - and their enemies are no different.
Narcissism allows people to take their servitude and interpret it as freedom. The problem can never be recognized as the system itself; it must always be redirected outward onto something else, some alien force that is ruining this world. The harsher and more intrusive the authority becomes, the greater the anxiety, the more you fear the other that you project as fucking you over. The end point of this line of thought is that the only conclusive solution can be the elimination of this force: the logic of genocide.
I agree that trying to replicate the Last Psychiatrist style is bad. Some strong illusion of transparency seems to be in action; I could never decipher most of the text. Maybe I am too stupid, or just not immersed in the same bubbles.
> This means they don't shout at their kids, even when they do wrong, like bullying another kid. They will shout at them when their frustration boils over as it must do after repeated wrongdoing, which is particularly likely to happen when the child does something that inconveniences them as opposed to something that's morally wrong.
100% this. I have already seen that so many times, but couldn't put it in words so well. I just noted that if you fail at a task that a monkey could perform better by following its instincts, you should be a bit skeptical about your wisdom.
I would add that even by the parent's proclaimed metric "less shouting is better" this behavior fails. Shouting at kids when they do something wrong at least has a chance to make them do the wrong things less often, which leads to *less* shouting in long term. Shouting at them randomly when frustrated... seems to *increase*, as the behavior does not improve, and the frustration accumulates.
> All they need to do is figure out what will placate you.
I'd say they need to figure out what makes you unwilling to start a conflict (e.g. knowing that it is guaranteed to escalate) or what makes you feel guilty if the conflict already started anyway. Different kids will probably use different strategies.
> The modern life of individuals is characterised by living within the shadow of a huge number of large bureaucratic organisations.
One problem is that institutions will try to treat everyone the same, but of course people are different and what works okay for some will fail completely for others. And the institutions are unable to process feedback (usually because the person who makes the rules is not the same as the person facing the failures).
> Indeed, the core responsibilities of your job may be an active impediment to your success since they take away your attention and energy from the important work of getting promoted.
Yes and no. As a software developer, a certain level of competence is necessary to get certain jobs. But taking care of the quality of code beyond that level is an impediment.
Similarly, school will require some knowledge to give you an A. It's just the knowledge beyond that level that is wasted.
> Narcissism allows people to take their servitude and interpret it as freedom.
https://www.youtube.com/watch?v=CLcwCRJ31q8 -- not sure if it makes sense without translation or subtitles: a scene from a traditional Slovak movie: A nobleman is beating a peasant, an outlaw tries to protect the peasant, but the peasant sides with the nobleman against him. "It's none of his business, your highness. Keep hitting, keep hitting, your highness."
I am curious if anyone else is following inkhaven remotely, and trying to maintain a schedule of daily or near-daily posts? It is so far very difficult to balance with a job and makes me much more impressed by frequent bloggers! Curious for anyone else's take
There's a whole Halfhaven challenge going on virtually where you're challenged to write 30 blog posts in October and November. There are around 20 participants, including myself.
I have been doing neither, but have been writing on 4TheWords since late August, which has a daily minimum of 444 words to maintain a writing streak. Not the same requirements as posting a completed thought per day, but I'm quite happy with the results so far; I've managed NaNoWriMo's 50,000-word pace for the last two months and should clear it for this month during the next weekend.
Interesting! Its the first time I've hearing about inkhaven. I think daily posting is possible if you don't write thousands of words per day - putting out a few hundred words daily with the occasional longer piece feels very possible to me. And there's definitely value in putting out a few hundred words every day.
It's amusing that the BBC faked Trump saying something he didn't in close to to what Scott used as an example a few years ago of something you can trust the media never to do.
Is "faked" the right word here? He said both statements, and it's not clear that his comment about "fighting" was innocent of violent connotations. Especially in retrospect where he has issued pardons for many who did engage in violence.
I think the way they spliced it together is a reasonable interpretation of what Trump meant, it's just that it's not *certain* that it is what he meant, and so it's misleading for them to splice it together to make it seem like there was only one possible interpretation
Nvm, I watched the clip and it was clearly trying to decieve. The edit is placed such that your meant not to notice it.
I still think it's a reasonable interpretation that Trump intended and wanted violence that day and so I don't think it's a dishonest interpretation to imply his statement about the need to 'figjt like hell or you won't have a country anymore' was a call to violence. But trying to edit a time skip so that you don't notice it is dishonest regardless of the context.
I’m amazed that you can admit it’s outright fraud and still try to justify it because you agree with the narrative. This is how we get authoritarian regimes.
I'm not trying to justify it. I said the edit is dishonest regardless of the context.
As a separate statement I am saying that it is a reasonable interpretation to say Trump wanted violence on Jan 06 as he allowed the storming to continue for 4 hours before taking action (tweeting) while politicians were hiding for their lives. But this does not excuse or justify the edit. The people responsible should be fired.
I see a more reasonable interpretation where he wasn't convinced there was "storming", but rather that there was an enormous crowd of protestors who weren't particularly violent (practically everyone outside the Capitol was just there to protest) and he didn't have a reliable source telling him exactly what was happening inside as it happened, and several _unreliable_ sources motivated to exaggerate anything his supporters were doing (as the BBC finally recognizes).
Meanwhile, everyone knows that the Capitol Police were critically understaffed that day; the only way Trump could have physically prevented entry into the building would have been to somehow produce scores of additional FBI agents within five minutes and have them somehow work their way to the building perimeter, or order the Capitol Police that were there to resort to lethal crowd control. Hopefully I don't have to convince anyone that these would have been absurd measures.
If a massive wave washes over you and floods the beach behind you, no one would contend that you're "letting" the wave do that.
Instead, he sends multiple messages calling on protestors to avoid violence, reminding them that theirs was supposed to be the party of law and order, to "respect the law and our great men and women in Blue", etc. Why would he do that if he actually wanted violence? Reverse psychology? No one's that crafty outside of a movie.
I think you might be slightly misremembering that post.
> I will be making a nitpicky technical point. I’ll start by making the point, then explain why I think it matters.
Scott did not write that this is "something you can trust the media never to do." If you revisit the article you'll see that he's extremely careful to caveat:
> the media rarely lies explicitly and directly
Note "rarely", my nitpick is that rarely is not never. This may matter because there are always extreme examples where bright lines are crossed and so hedging with "rarely" let's Scott fend off examples of egregious line crossing that could be raised not would not, in fact, diminish his point.
The panorama example is egregious, that two BBC execs have resigned over this speaks well to the ideal of holding the BBC to a high standard, and the naive/hopeful/optimistic may hope that it suggests that this level of deception will be less tolerated in future.
However... I also think it's a really great example of exactly the tension Scott's article describes. The quote in question was a contraction of two separate bits of trumps speech edited together to give a distorted impression of what trump said while still only containing his words. Descriptive? Absolutely. Politically motivated? Almost certainly. A great example of how the missing context is essential in determining the difference between misinformation and true information? Yep. Exactly the point that the second half of Scott's essay addresses.
Aha, in that case, my apologies for the nitpicking.
I went back to the Bounded Distrust post, this seems the most succinct relevant quote from it:
> There are lots of cases where you can’t trust the news! It sucks! It’s completely understandable that large swathes of people can’t differentiate the many many cases where the news lies to them from the other set of cases where the news is not, at this moment, actively lying.
I understood the point of the post, thanks. I agree they're not the knaves of logic puzzles, and that skilled liars can deceive without uttering a single provable false statement, always hedging with "experts" and "unnamed [read as: 'imaginary'] sources."
The more relevant bit is the opening, where he describes a hypothetical scenario with Fox, as a way of establishing a loose upper bound on the level of distrust one should have for the media:
>The level on which FOX News is bad isn't the level where they invent mass shootings that never happened. They wouldn't use deepfakes or staged actors to fake something and then call it "live footage". That would go way beyond anything FOX had done before.
I'm saying the BBC stitching together clips of Trump saying some phrases about an hour apart to create a seamless video of him saying something he didn't IS at the level of deepfakes and staged actors.
My view is on the existence of such an upper bound is that the correct attitude is something like "You don't hate journalists enough. You think you do but you don't."
Hey, uou never answered why you repeat White identitarian talking points. I asked you maybe 4-6 days ago and you never responded. It was when I raised the issue of white genocide. A concept you defended.
I'm still curious about your motivations. Do you believe in some kind of quid pro quo where if you support White conservatives, they will support you, or something like that? Does it have something to do with building an anti-Islamist Coalition or something? Please give me some insight. I'm trying to figure this out.
I didn't think even you considered a serious question meriting a response, but sure. Since you ask it in an unrelated thread, I'll respond here.
First, the word "genocide" has devolved into meaninglessness or legalism, so I don't really care whether the definition you want to adopt includes or excludes intentionally-engineered sharp demographic decline. The consensus among international experts or whatever doesn't sway me in the slightest.
Second, this "White identitarian talking points" thing reeks of when there was a push to make the OK gesture 👌 a White supremacist hate symbol. I reject this attempt to smear ideas by association.
But if you want some deeper psychological drive, I find it deeply irksome when powerful organizations suppress dissident views. That should be enough to explain what see as mysterious about my motivations.
> BBC stitching together clips of Trump saying some phrases about an hour apart to create a seamless video of him saying something he didn't IS at the level of deepfakes and staged actors.
Deepfakes and staged actors seem obviously worse to me?
I'd put all 3 above a bar of "unacceptable" and it's good they are getting fired for it.
But being absurdly misleading is still a notch below plain fabricating information.
Imagine a deepfake that says something *very similar* to what the person actually said. Would that be more acceptable than a deepfake that says the opposite of that?
From certain perspective, yes, it is less misleading for the viewer.
From another perspective, the difference between the two is negligible compared to the horrifying fact that the line "you shall not show deepfakes on TV" was crossed. That has way more serious long-term consequences than whether the deepfake was realistic or not.
Stitching two sentences together is a poor person's deepfake.
A matter of taste what you consider worse, I suppose, but what do you consider to be the information the BBC conveys, which may then be judged on the basis of whether or not it's fabricated? Only that every individual syllable being broadcast was actually spoken by Trump sometime in his life and recorded faithfully with a real-life microphone, with the order in which they're strung together being, say, a mere artistic choice?
There was a recent example of a "deepfake" which I think would be interesting to compare this to: the Senate Republicans posted one of (an AI version of) Chuck Schumer saying something he ACTUALLY SAID. https://x.com/NRSC/status/1979187237720924195
So in this case, the information I'd say this video conveys is entirely true. It just so happens that the pixels and audio are synthetically generated (a fact which is also clearly disclosed on screen), yet in better accordance with the truth than the more "real" pixels and audio of the BBC.
Comparing Jan 6 to 9/11 is part of the nonsense here. A few fat guys walked into a parliament, unarmed. One was shot. The other deaths were old age related. There was no coup attempt. There was no way the election wasn’t going ahead.
One was shot after lunging through a broken window in the final door barring her from the Congresscritters, with cops in defensive positions behind it with guns drawn. I mean yeah, 9/11 is a ridiculous comparison, but let's not unnecessarily undersell it.
Honestly I think Jan6 is the single thing most distorted by hyperpartisanship. I'm not even sure what exactly I think about it, it seems so complex. It seems to me there was a mix of really disturbingly serious people (both in the crowd and in the legal/procedural wrangling) who wanted to overturn the election by any means necessary, and a bunch of goofy Trump supporters with lives too comfortable to even consider participating in a coup, who really did just want to peacefully protest (probably including some of the ones who ended up inside the building). And I think Trump didn't even know what he wanted, or was vacillating. But now Jan6 is either a violent coup attempt or an almost entirely peaceful protest. Argh.
What the BBC did was egregious and the British-traditional correct outcome has belatedly been achieved (the top two people in charge have lost their positions and their careers).
What Trump said to that crowd, _minus_ the "fight" line, absolutely was incitement to try to forcibly overturn an election result. And literally today he's continued handing out blanket pardons to everyone who was on his side in the whole effurt to overturn that election result "by hook or by crook" as the old saying went. The analogy to 9/11 works just fine unfortunately.
Alright, I've just finished reading (well, skimming) the entire damn speech. It is, in typical Trump fashion, very long and rambly.
Here are the only parts I found which are actual calls to action on the part of the audience.
The first one, near the beginning:
"I know that everyone here will soon be marching over to the Capitol building to peacefully and patriotically make your voices heard"
The second one, at the end:
"So we're going to, we're going to walk down Pennsylvania Avenue. I love Pennsylvania Avenue. And we're going to the Capitol, and we're going to try and give.
The Democrats are hopeless — they never vote for anything. Not even one vote. But we're going to try and give our Republicans, the weak ones because the strong ones don't need any of our help. We're going to try and give them the kind of pride and boldness that they need to take back our country.
"We have come to demand that Congress do the right thing and only count the electors who have been lawfully slated, lawfully slated."
Congress in reality _did_ "only count the electors who have been lawfully slated" which of course was what Trump was asking the crowd to _prevent_.
As everyone attending that rally understood, what he wanted them to demand was that Congress instead count lists of electors from several states that had been named by people other than the election officials and/or elected officials in those states having the lawful duty and power to slate electors.
That the crowd understood that, and had also gotten the gist of what was meant by "demand", became clear when the VPOTUS declined to entertain those bogus slates of electors and the crowd responding by chanting "hang Mike Pence".
Incitement to try to forcibly overturn an election result.
The problem is that the question is the violence, not whether it was correct to overturn the official result. Setting aside whether Trump actually believed this, you have to acknowledge that there was a large chunk of the country that thought the official result was fraudulent. You are skipping to the part where you are right and they are wrong, skipping over how they're supposed to address the massive injustice they perceive. Even if they're wrong, it's unavoidable that they're going to respond in some way. And Trump told the crowd to do it through peaceful protest.
Note: Trump wasn't present for the "hang Mike Pence" chant - it's important not to conflate that into his speech to the crowd, because letting that go would obviously be supporting violence. If you have blurred that into your understanding of his speech, that might be warping your perception.
My thoughts are that it's a fifty minute video which could have been a three thousand word essay. If it had been a three thousand word essay I probably would have done it the justice of spending two minutes skimming it, but I don't have time to spend fifty minutes on a video just to see if there's anything interesting in there.
There should be a website that uses an LLM to generate short written summaries of long YouTube videos. (In a perfect world, YouTube itself would do that.)
> There should be a website that uses an LLM to generate short written summaries of long YouTube videos. (In a perfect world, YouTube itself would do that.)
This is actually one of my primary uses for the Gemini LLM, which alone among LLM's has the rights to summarize youtube videos, due to Google owning both.
It will give you a summary with linked timestamps to the things it's talking about in each part, I find it invaluable, because I refuse to click on a good 98%+ of any video links people send me.
My problem is that the summary kinda mentions the thesis ("aging demographics is actually not a problem") but does not contain an explanation. Not sure whether it means that there is no explanation in the video, or because the bot did not consider it important.
- In 2022, Elon Musk acquired Twitter and has since expressed concerns about declining birth rates, which he believes could lead to a population collapse.
- Musk has been vocal about this issue since at least 2017, suggesting that the world is heading towards a demographic crisis that many ignore.
- This concern has been framed as a moral panic, reminiscent of previous generational fears regarding population decline.
Critique of Demographic Analysis
- The discussion includes an analysis of an article by Fred Pierce, which highlights Japan's declining birth rate and its implications for economic stability.
- Pierce argues that Japan's aging population is a significant problem, linking it to economic stagnation without providing substantial evidence for his claims.
- Critics argue that Japan's economy has remained robust despite demographic challenges, indicating that economic performance is not solely dependent on population growth.
Population Aging and Economic Implications
- Aging populations raise concerns about the old-age dependency ratio, which is the ratio of older individuals who rely on the working-age population.
- The narrative suggests that increasing numbers of retirees will burden younger workers with higher taxes to support pensions and healthcare.
- However, historical data shows that economies can adapt and thrive despite demographic changes, challenging the notion of inevitable economic decline.
Historical Context of Birth Rate Fears
- Concerns about declining birth rates are not new; similar anxieties have been expressed for decades, including in British Parliament discussions from the 1920s and 1930s.
- Historically, predictions about dire economic consequences due to low birth rates have often proven to be incorrect, as economies have continued to grow.
- This historical perspective suggests that current fears may be overblown and rooted in a misunderstanding of demographic trends.
The Role of Economic Factors in Reproductive Choices
- The discussion shifts to the real barriers affecting birth rates, which include economic factors such as financial stability, housing costs, and job security.
- Research indicates that many people desire larger families but are constrained by the high costs associated with raising children.
- Improving economic conditions, access to healthcare, and support systems could address these barriers more effectively than simply promoting higher birth rates.
Critique of the Moral Panic
- The moral panic surrounding birth rates is critiqued as being exploited by political and media narratives that may not reflect the complexities of demographic changes.
- This panic can lead to harmful policies that prioritize individual responsibility over collective solutions to societal issues.
- The conversation emphasizes the importance of addressing the underlying social and economic issues rather than succumbing to fear-based rhetoric about population decline.
Future Discussions on Reproductive Ethics
- The video concludes with a teaser for a future discussion on the ethical implications of having children in the current socio-economic climate.
- This upcoming dialogue aims to explore the responsibilities and considerations that potential parents face today.
- The creator invites viewers to engage with these topics actively and consider the broader implications of reproductive choices in society.
That's fair, though going beyond the fact that it's about birth rates and that she's critical of the topic (which you notice immediately when watching), I wouldn't want to say much more here, so as to get people's own immediate impressions :)
I agree, but it is a repeated situation, so it seems worth to automate. Another advantage of an automatic summary is that it would be (probably) impartial.
I haven't seen that one yet, but I've watched several of her others.
I found "Identity" very powerful and helpful to me when I first watched it, a few months into my own transition. She's on my list of people whom I probably owe the price of 1-2 therapy appointments because of it. I also appreciated her video about navigating trying to get diagnosis and HRT from the NHS for similar reasons, although it was less directly applicable to me as I'm an American with good benefits from a very LGBT-friendly company and thus face much fewer obstacles to obtaining gender-affirming care than she did.
I watched a bunch of her older videos from her "grad student nerding out about her coursework" era and generally really liked them.
I watched her video about effective altruism and FTX and wound up liking it a lot better than I'd expected to from the clickbait title. There was a fair amount I disagreed with in it, but it was decently nuanced and nothing jumped out at me as egregiously wrong or unfair. I'd watched it specifically as a Gell-Mann test (i.e. do I still like her stuff when she's talking about stuff I already know a fair amount about) and I think she passed handily.
Her video from a couple years ago about Stoicism was entertainingly presented but the content was disappointing, especially the criticism of Stoicism as a modern philosophy. She relied on two prongs:
1. Roman Stoic writings contain a lot of obnoxious assumptions about class and gender. Which I suppose is probably true (or at least a defensible claim) as far as it goes, but I found it relatively low-effort since that's much more a general critique of Classical Roman culture than a specific critique of anything peculiar to Stoicism, and I don't think she made the case that the assumptions were load-bearing parts of the philosophy rather than mere baggage. Here, though, I do give her some credit for acknowledging the existence of modern Stoic philosophers who have attempted to strip out the Roman cultural baggage.
2. She claims that Stoicism is extensively biased towards a posh and powerful perspective and advocates complacency and compliance in the face of injustice. This seems to completely overlook the existence of Epictetus, one of the major classical Stoic writers, who was born a slave and devoted a great deal of thought to the question of how to live a moral life from a position of powerlessness in an unjust world. She also seemed to overlook Epictetus at a different point in the video when mentioning that we know classical Stoicism exclusively through Roman sources despite it (originally) being a Greek philosophy: Epictetus was Greek and so was his pupil Arrian who is our main contemporary source for Epictetus's teachings. I suppose this could be partially salvaged in a "technically correct is the best kind of correct" way by noting that Epictetus and Arrian lived in the Roman Empire and I think both of them spent a fair amount of time at the Emperor's court in Rome.
Thanks, very interesting! I'm in a broadly similar boat in that I'm a long-time viewer and generally enjoyed many of her videos and strongly disagreed with few. This one seemed very off to me, but I also don't know a lot about the topic, so I was curious to hear the impressions of the people here.
I love stumbling upon your elaborate comments, btw! I enjoy the way you write. Would love to read some posts of yours if you ever write some :) Just wanted to share that with you
Thank you! I've actually been thinking semi-seriously about starting a blog again. Or a YouTube channel, as there seems to be a market for long-form explainer video essays made by sapphic trans women with three-syllable first names and one-syllable last names.
In the meantime, if you're looking for more of my writing, I did the "Deathbed Ballads" review in this year's non-book review contest and the "On the Bondage of the Will" review in last year's book review contest.
""Some economists say that Japan's repeated recessions since the 1990s are due to the burden of ever more oldies." - But Pearce provides no citations for this claim and trades on several stereotypes about older workers."
If this is the kind of logic you find convincing I don't know what to tell you...
I think Leftists snarking at people wanting bigger families, valuing family, or for people to be concerned about the birth rate is always dumb.
Many Leftists are all in on the idea of "community." Community building, community powered solutions, community art, community participation, etc. Do they not realize that communities need futures and that one day all the current members will be dead?
The ability to purchase the labor of younger abled bodied workers in your old age, to help you continue to live when you lose the ability to care for yourself, is a trick of capitalism. It distances people from realizing that the old need the young as much as the very young need caretakers and parents. None of these people seem to ask, Who will be there when I'm old?
< I think Leftists snarking at people wanting bigger families, valuing family, or for people to be concerned about the birth rate is always dumb
< None of these people seem to ask, Who will be there when I'm old?
If someone thinks people in a large subgroup identifiable by a single word are all dumb and wrong in some obvious way, it is guaranteed that the thinker has his head up his ass.
>people in a large subgroup identifiable by a single word
"idiots" (in the original technical sense of IQ of 0-25 (from AI summary, usual warnings apply))
>all dumb and wrong in some obvious way
( presumably with some sort of organic brain damage, though the cause of damage will presumably differ from person to person... And they won't exactly be _wrong_, but more nearly be unable to express anything that might be correct or incorrect... )
Since what one can accurately say about a subgroup is at issue, how about we test out your view of the subgroup of people who make comments like mine? You have in mind one characteristic of people who make comments like mine: We think not having kids is a radical act. Your conception of my subgroup surely has more to it than a prediction about my views about having kids, otherwise it wouldn’t really count as distinctive take on life, right? How about naming a few more characteristics of people who make comments like mine, and afterwards I will tell you honestly which are correct, so long as they are not deeply personal matters. What's your idea of my subgroup’s demographics? SES, age, gender, kind of profession? How about our likely views on various issues and policies?
There are many red flags for me, though maybe I'm drinking the population collapse koolaid so hard that I'm heavily motivated reasoning. Also, be aware that I'm a high school Maths and Science teacher, not an economist.
1. Everyone loves to hate Elon, so they've put him up as the populariser of the idea to discredit the idea.
2. Demonstrating that there are some bad articles about birth rate has nothing to do with the actual arguments, but can soudn convincing nonetheless.
3. There are so many straw men in that first bit. It's obviously easy to find catastrophisers. Nobody serious (AFAIK) thinks literal extinction is imminent within a few hundred years because of this. And again, kinda jsut attacking what one possibly hyperbolic writer is saying.
4. Finally when we start "actually talking about birth rate" it mentions that the population is currently increasing by 200k per day. That fact fits perfectly with the model of collapse, but again, might be convincing to someone. The everything seems slanted towards having memorable points that a viewer can make when talking to a birth rate collapse person.
5. The ghost thing at 9 minutes about aging population makes sense (as opposed to being crazy as she hoped), then she refutes it with many non-real arguments. Like sure, total dependency is different to the old person ratio, but how exactly is the number of "young" dependents expected to go up to balance the aging population?
And sure over 65s don't do nothing to contribute to the economy, but they do do less and over 80s probably ain't doing much, so the point is that currently we see certain dependency ratios as sustainable, not in theory, but in practice, and as that ratio changes, it will put more strain on those who are working. Like they're fine theoretical points, but we have actual evidence for how these numbers affect the economy.
Does she think the retirees are going to contribute MORE per person than they do now?
And "economies grow over time" is an assumption that the opposition say is contributed to by population growth, so that's invalid for a reason I can't articulate. "If it does go south, it won't just be demographics that's to blame." OK. But it will be partly demographics? So you've conceded the whole argument?
I'm sorry, I got to 3 minutes in thinking it might say something interesting or valid and by 13 minutes it just keeps being mush, so I'm stopping there.
I think the Kurzgesagt video on South Korea and the fact that there are very few assumptions baked in, it's just people getting older and trends continuing (or even improving and we're still stuffed), is quite convincing. I didn't even really know it was something Musk was on about, but that's not to say that I've not become aware of it because of his boosting it.
One aspect that should perhaps be considered more is that aging populations find it harder to recover from certain kinds of demographic problems. E.g., Russia and Ukraine are in the process of losing millions of people; in the olden days this might be followed by a baby boom when things had settled down and after a while the population would be back to normal, but it's difficult to see a country of mostly over-60s bouncing back like this.
Thanks for sharing your thoughts! I agree that a lot of the video seemed straw-manny and it often came up that birth-rate isn't 100% of the issue, but those points seemed very weak. I'm a bit confused about the object-level question still, but that's okay
Anyway, since it/she/whatever scared me off five minutes in, a gem of past Science Has Proven from 1905: over-education means women are driven insane by pregnancy! It's all down to the phosphates, you see, as Brain battles Baby for this precious bodily resource, as certified by Top Men in the field of women's troubles:
"First of all, is education being carried on at present to such a degree as to at all affect the bodily or physical health of women? This is a very important question, because the duties of wifehood, and still more of motherhood, do not require an extraordinary development of the brain, but they must absolutely have a strong development of the body. Not only does wifehood and motherhood not require an extraordinary development of the brain, but the latter is a decided barrier against the proper performance of these duties. Any family physician could give innumerable cases out of his experience of failures of marriage, directly due to too great a cultivation of the female intellect, which results in the scorning to perform those duties which are cheerfully performed, and even desired, by the uneducated wife. The duties of motherhood are direct rivals of brain work, for they both require for their performance an exclusive and plentiful supply of phosphates. These are obtained from the food in greater or less quantity, but rarely, if ever, in sufficient quantity to supply an active and highly educated intellect, and, at the same time, the wants of the growing child. The latter before birth must extract from its mother's blood all the chemical salts necessary for the formation of its bony skeleton and for other tissues; and in this rivalry between the offspring and the intellect how often has not the family physician seen the brain lose in the struggle. The mother's reason totters and falls, in some cases to such an extent as to require her removal to an insane asylum; while in others, she only regains her reason after the prolonged administration of phosphates, to make up for the loss entailed by the growth of the child. Sometimes, however, it is the child which suffers, and it is born defectively nourished or rickety, and, owing to the poor quality of the mother's milk, it obtains a precarious existence from artificial foods, which at the best are a poor substitute for nature's nourishment. The highly educated woman seems to know that she will make a poor mother, for she marries rarely and late and, when she does, the number of children its very small. The argument is sometimes used that it is better to have only one child and bring it up with extraordinary care than to have six or eight children brought up with ordinary care because in the latter case the mother's attention is divided. But this is a fallacy. Everybody knows that the one child of the wealthy and highly educated couple is generally a spoiled child and has as a rule, poor health; while the six or eight children of the poor and moderately educated woman are exceedingly strong and lusty."
The mother's reason TOTTERS AND FALLS. Beware the perils of too much intelligence, ladies!
When I first read this I thought you were being mean about her being trans, which is a little surprising as I've usually known you to be snarky about people's words and actions rather than their identities.
Then I watched the first few second of the video and I see what you're talking about. I don't know what she was going for with that look, but I really don't think it's working for her.
Yeah, ironically she's normal looking at the end of the video (apart from the little staged performance about "oopsie am I preggers-weggers?") so the start would immediately make me turn off because it looks like a cross between performance art and being an insufferable twit.
She actually is an actress. Mostly stage stuff, a few TV and voice acting roles, and I think she's been cast in a movie that's currently in preproduction. She's gotten some decent traction in her acting career in the past 2-3 years, but could fairly have been described as a "frustrated actor" when she started doing this sort of stuff on her youtube channel instead of just delivering little conversational-style infodumps to the camera like she did early on.
To be honest, I had no idea she was trans. I got hit in the eyes with this cross between a tree and a rubbish heap and had no idea if this was meant to be serious, a skit, or what.
It definitely gave me "art school performance artist who next thing might throw off her costume and start rolling around the floor in red paint to symbolise menstrual blood (if we're lucky and she's not gonna use real menstrual blood)" vibes.
Yes, definitely. I had a similar reaction to the first moments of the video, despite already knowing who she is and the sorts of stuff she does and having watched and mostly liked several of her other videos.
I probably won't go back and watch the rest of this one, since I don't like the visuals and the vibe I'm getting from comments here and elsewhere is that content-wise this one probably isn't one of her better efforts.
I've now watched the first two episodes of Pluribus - strong recommend so far! It shows potential to be a show that takes its sci-fi premise seriously and wants to explore all the mind-bending ramifications (instead of just using it as a setting for soap opera-ish drama).
It reminded me a lot of Scott's story Samsara (grouchy protagonist trying to resist the spread of what they see as the insufferable and quintessentially inhuman enlightenment experienced by the rest of the world)
For those who have watched it, I'm curious - which of these two attitudes is closest to yours?
A) A world full of people joined into a monolithic 'we', like in the show, seems good, and it would be a good thing if it happened in real life. Even if something is lost, more has been gained.
B) The world as depicted in the show seems horrifyingly dystopian and I am rooting for someone to find a way to reverse it.
I watched it. It was pretty good. I still get annoyed by being strung along for two seasons — at least.
Why can’t Vince Gilligan just tell us that the aliens have been traveling near the speed of light for 400 years now and when they get here they are going to eat us?
It's early days yet and the series could go any number of ways, we could find out all sorts of things as we go.
But I think that what keeps the series interesting is the tension between the horrifying aspect of what has happened, and the obvious good side. If it were just Invasion of the Body Snatchers then it would be much less interesting.
So I don't see any point in trying to come down on one side or the other. It's both the most horrifying thing that could possibly happen, and the best thing that could possibly happen, simultaneously.
Going by the synopsis of the first two episodes on Wikipedia, plus that it's been renewed for a second season, I forecast there will be twists.
Oh, so many twists.
It's all a secret invasion plan by the aliens! Ha ha, you thought it was benevolent? How stupid are you?
No, wait, there are more than eleven immune people and they're starting a global resistance movement.
No, the movement is split and some want to be world rulers because they can do what they want and kill the Joined by strong emotion so an immune person is a walking nuke. So the two (or more) factions of Immune are fighting each other and each is trying to recruit Carol and they're telling her lies in order to do so.
No, see, the aliens *were* benevolent and the invasion story is a cover-up by the government. Or parts of the government. Or secret behind the scenes conspiracy groups.
It's not aliens, it's the government(s) mind control experiment gone wrong/right.
Carol develops superpowers.
Carol kills more people.
Turns out Carol didn't kill anyone, that was just propaganda by the Joined.
Turns out Carol didn't kill anyone, that was staged by one faction of the Immune to fool her into joining them.
Think of the dumbest way this can go, instead of exploring the premise, and they'll go that way because viewership figures aren't high enough and so forget the hard sci-fi angle, we want romance, action, and a clear Good Guys/Bad Guys plot.
If it were some random Hollywood show I'd agree with you, but Vince Gilligan's previous projects have earned enough of my respect that I'm willing to believe, until proven otherwise, that this story is going somewhere other than the most boring and obvious directions.
I don't trust big streaming services whose main focus is selling goods and not making TV/movies, see Amazon and Rings of Power.
Apple TV is, in the end, part of the Apple brand and is not primarily "we make visual media". So whatever about the producer/showrunner, the suits will have the last say and if the show isn't getting them what they want (viewership figures translating into new customers for other Apple branded goods) then it gets changed or gets the chop.
Yes. Vince Gilligan likes a good twisty plot, but his shows aren't about chasing infinite mystery boxes. I bet the hive mind is being straight up and the situation is as they describe it. That still leaves plenty of room for what happens in such a strange situation .
I liked it a lot but there seems to be something missing compared to Gilligan’s other series. The first two episodes basically see Carol complaining without much agency, raging at the world in an angry way while she gets drunk (and kills a dozen million people).
In Better Call Saul and Breaking Bad, the characters all had their flaws but they took agency for their lives and actually did something with their situation. I hope that Carol figures out a way to take agency and struggle against the world as she encounters it in a productive way, which seems to be part of the appeal in BCS or BB that is missing so far.
If also has no B plot, which Breaking Bad did well between the meth manufacturing and the family life, and Better Call Saul did incredibly well between the Lawyer Plot and the Drug Cartel Plot. I think if Pluribus doesn’t have a really good B plot it has no chance of being a truly “great” show. I don’t know what a good B plot could be from what I’ve seen so far, so I’m not very optimistic.
Carol Sturka or Kim Wexler as I continue to think of her and the uber horny Mauritanian, Koumba Diabaté, ‘livin the dream’ or his particular dream at any rate are the only two with anything that seems like agency so far.
The other 6 of the 10 ‘unaffected’ humans that we’ve seen so far seem to be ambivalent about if not okay with the idea of being assimilated.
So it seems either Kim… I mean Carol, uses some of her latent lawyerly persuasion skills somehow retained from her previous role to convert the apathetic or there are some as yet unseen unaffected characters or there is no way to develop the conflict needed to keep the series interesting over two season.
Another possibility is the RNA-induced genetic modification is somehow reversed or breaks down in the remaining six and a quarter billion humans currently under the influence.
> don’t know what a good B plot could be from what I’ve seen so far, so I’m not very optimistic.
I suspect we'll eventually see someone else's perspective. The overly-prominent timer makes a lot more sense if we're eventually going to see someone else's perspective on the same time period.
Also, just to repeat some speculation I heard elsewhere, it doesn't make any sense that the President and VP and entire cabinet down to the undersecretary of agriculture all happened to be in the 10% that died. The boring answer is that they were deliberately killed early on; the more interesting answer is that they're safely hiding in a bunker somewhere.
> Also, just to repeat some speculation I heard elsewhere, it doesn't make any sense that the President and VP and entire cabinet down to the undersecretary of agriculture all happened to be in the 10% that died.
The hive mind said that the the military caught on to what was happening so they had to rush, and that's why so many people died. It makes sense the military would have tried to secure the top level of the government and so there wasn't a non violent way to body snatch them.
As for survivors in a bunker, I'm pretty solidly on the side that the hive mind is not being deceptive because I think that makes the most interesting story. That doesn't rule out people hiding in a bunker of course, just that the hive mind wouldn't be aware of them.
I quite enjoy the tension between pacifist aspect of the hive mind, and the horrifying twist that although they would be horrified to step on an ant, they will add you to the collective mind at the first opportunity no matter what your stated preferences. I think this is the reason for the "overly-prominent timer" - the hive mind said they hoped to solve the problem of the immune humans in a few months. So it's gonna turn into a creepy doomsday clock.
Yeah that would be really cool if there was a resistance working against the alien virus. I somewhat doubt it though, since Air Force one wasn’t taken by the president, which is usually where he would be in the event of a mind-controlling pandemic. Maybe in a bunker in Colorado though.
Assuming the President had some clue what they were planning, being in Air Force One definitely wouldn't be the way to go.
Now I come to think of it though, they did draw attention to the nuclear football, which would have followed him into the bunker if he were still alive.
(For what it's worth, in real life Air Force One is two identical planes. I doubt that will be a plot point though.)
I watched the first two episodes and also recommend.
(A) is good as long as the monolithic 'we' is benign.
I would expect (in the real case and as a natural plot progression in the show) some fragmentation of the 'we' over time, because that is what evolution does to species. This process could get quite messy.
I ony watched the first episode and liked it a lot. It reminds me of a thought experiment that occurred to me years ago: what if the subjective experience of Borg members is ecstasy? Shouldn't we want to join with them?
I lean towards A, but mostly because it would be kind of disappointing if the twist was "but all these chipper people were secretly suffering horribly!"
It's probably not purely A either but it may be an intermediate solution
It seems like a good premise for a miniseries. But having been renewed for a second season (and with no reason not to expect further extension), I'm hoping there's some twist coming, perhaps where it IS reversed, and then you have to deal with the consequences of everyone retaining bits of everyone else's memories.
A survey. I know Scott has asked about this on his surveys, but he uses a, in my opinion, largely meaninglessly subjective scale where 1 and 5 (let alone the middle numbers) can mean a dozen different things to different people. Though perhaps others disagree. Others ask only about p-doom.
Please say, first of all, which of the following scenarios you think is most likely, and second of all if you can, your rough percentage likelihood for each. Please also be as likely to respond to this if you are a non-rationalist and also if you are; sometimes it seems there's all of one group and none of the other. Operationalise if you need such precision as something like "any humans around in 30 years (by magic/miracle if necessary) will mostly agree that this is what happened." This are vague clusters, not detailed descriptions. Draw the boundaries in an intuitive way.
DEFCON 1: Superintelligent AI that can wipe out all life with a blink of its metaphorical eye (regardless of whether it actually does)
DEFCON 2: AI that can automate the entire economy and create post-scarcity, but not superintelligent
DEFCON 3: AI automates or eliminates whole industries or fields of art/science/ etc, and/or is a weapon more decisive than all other weapons/military capabilities combined
DEFCON 4: AI is a major technology on the level of steam power, nukes, the internet, the automobile etc, significantly reshaping parts of society, but not individually decisive as a weappn or tool to the extent of overwhelming all other considerations
DEFCON 5: AI is a normal, relatively minor technology or toy. Any significant reshapings don't happen or if they're already happening, get rolled back in a few years when people realise AI's limatitions.
Again these are vague clusters. Please try to guess what I mean instead of rules-lawyering the descriptions.
Automating the entire economy without being superintelligent is hilarious, by the way. You're discussing a planned economy, and we can't pull that off because it requires god-tier intelligence even to figure out "how many sprockets we'll need in 5 months."
AI will claim to be superintelligent, and have the statistics to show that most people are less intelligent than it is. AI may subsequently embark on a program to create more intelligent/creative pet-humans.
DEFCON 1: 40% sort-of. Even infinite intelligence is bounded by the laws of physics, and the amount of computation possible is _also_ bounded by the laws of physics. So, not "with a blink of its metaphorical eye" but something closer to "like a chess grandmaster defeating a novice". There is also the ambiguity of "smarter than the one smartest human who ever lived" vs "smarter than all humans put together" vs "species level difference". I think the first is likely, and likely in a single AI instance. In other words, I think there is a 40% chance of "a work group (maybe not nation) of geniuses in a data center". I have no idea whether "species level difference" is even physically possible.
EDIT: To clarify slightly: I do think that the probability that the current development trajectory (not just scaling, but also guessing that the likelihood of further refinements in reasoning and a breakthrough in incremental learning are probable) gets us to strong AGI / weak ASI is 40%-ish. From that point, there are large additional uncertainties in what is feasible after that. How much of a return is there to _further_ research in AI? _We_ are an existence proof for neural nets with human level intelligence. We don't have an existence proof for going beyond that, except in the weak sense of human organizations and humans with computational tools. _Is_ there reachable fruit beyond that? Is it general purpose, or only specialized? Even more generally: Do the returns to intelligence itself saturate, and at what level? Our civilization has picked a lot of low-hanging fruit from intelligent thought - I don't think anyone knows what _fraction_ of the low-hanging fruit we've already plucked. There are a few areas (speed of light delays, efficiency of heat engines, quantum efficiency of LEDs) where we are at or close to physical limits. An ASI might find that we have reached limits in more areas - or not, and might find opportunities we've missed.
DEFCON 2: 30% "automate the entire economy" - could be. "post-scarcity" - there are always finite atoms and joules and bits, so not really post-scarcity. Approximately AGI without ASI.
DEFCON 3: 20% "automates or eliminates whole industries" - could be, not-quite-AGI, maybe never quite reliable enough for some roles? NOT "is a weapon more decisive than all other weapons/military capabilities combined" in that case.
DEFCON 4: 10% This sounds like stagnating just about at the level we have today, so e.g. AlphaFold is a _major_ reshaping of biochemical research and e.g. customer service representative jobs are probably doomed and it is very useful for many facets of medicine, but it takes us decades to redesign work flows to take full advantage of AIs' capabilities.
DEFCON 5: 0% I concur with the other commenters who note that we are already beyond this. AlphaFold _alone_ has more impact than this.
I think DEFCON 4 is probably most likely (in particular, a technology on the level of nuclear weapons doesn't require any radical restructuring of society in general). And in this question I'm thinking about what might happen in 30 years' time; I don't think I have any way to foresee what new stuff people might do with computers in 500 years' time.
DEFCON 5: ~35%
DEFCON 4: ~55%
DEFCON 3: ~9%
DEFCON 2: ~1%
DEFCON 1: negligible given current technology (<< 1%)
Most likely: None of the above. AI is banned globally due to a combination of X-risk, ethical issues stemming from AI personhood, and maybe miscellaneous horrors that pop up.
Second most likely: AI is restricted/capped at around DEFCON 4 or 5 level but not banned. Humanity dances on the edge of a cliff for a while. Besides for general capability caps, individual countries also either ban AIs which identify as people (AI partners & friends, AI "communities" surrounding a single human, etc) or fall to them.
Third most likely: Humanity fails to restrict AI, everyone dies. (AI reaches DEFCON 5, followed by 4, then maybe 3 (possibly skipped over), then extinction.)
Fourth: Global crisis, no ability to continue AI research or develop new models for a long time, AI is stuck at DEFCON 5 or maybe 4.
Misaligned DEFCON 1-level is unlikely to overlap with humanity. In case of hostile takeoff, ASI would wipe out humanity as soon as it's confident in its ability to do so successfully without repercussions, rather than wait until it reached the power levels described. A positive singularity is unlikely.
DEFCON 2 seems like an oxymoron.
DEFCON 3 isn't stable, can't be a long-term situation without leading to takeoff.
DEFCON 4 and 5 are compatible, and could each be true in different parts of the world.
So, very rough percentages: DEFCON 1, 15% (13% post-apocalypse outcome, 2% positive singularity), DEFCON 4: 15%, DEFCON 5 (current): 15%, no AI anywhere: 55%.
The only plausible way I see for a ban or cap to happen is if massive amounts of people lose their jobs and then they vote for the anti-AI party. I think the AI corporations are smart enough to frog-boil to the extent it's possible to do so, so even this isn't a certainty.
How else do you visualize that a ban/cap can happen? People don't even care about animals, and it's relatively recent that they have started to care about those with a different skin colour, but they will care about AI personhood? I also don't think the X-risk argument alone will ever lead to a ban, because those arguments will only get more convincing as AI capabilities grow, but unfortunately, the apparent potential reward from controlled AI will also grow with capabilities and that will balance out the x-risk concerns in the minds of the populace.
Also, even if a ban/cap is implemented, there is just no way states won't continue to develop AI in secret. I would be willing to bet on this with arbitrary odds, so SAI will still likely happen. Do you think states will keep their word to not develop AI after it becomes obvious to everyone how key this technology is?
It is much harder to develop superintelligence in secret than it would be to secretly develop, say, nuclear weapons. It's also possible to have nuclear weapons without those weapons immediately turning on you and killing you, outcomes which would be a heavy incentive to not try and also to whistle-blow.
Yudkowsky has written a fair amount on the methods of AI caps via normal arms-control systems (backed by various consequences including military threats).
Politicians are not, contrary to popular belief, particularly stupid. They and their staff and backers and constituents like surviving, much more than even things like momentary economic growth. Faced with the two options of (1) apocalypse and (2) successfully stop ASI with whatever tools are available, they won't pick the former.
I foresee X-risk issues leading to a strongly-enforced global cap to around current AI capabilities (maybe a bit lower), and then, once we're just dealing with whether to use a relatively minor "normal" technology, AI welfare issues gradually eliminating use of what remains.
And what if the politicians don't need to maintain popular support? China isn't a democracy, and is a major player in this scenario. Advances in technology make it easier for a minority to maintain power. Surveillance and law enforcement are obvious applications for AI, and workers can slowly be replaced by automation. If there is no obvious risk of "apocalypse", there is no reason for them to hold back. If anything, not having the need to cater to an unreasonable populace gives them a massive advantage.
> If there is no obvious risk of "apocalypse", there is no reason for them to hold back.
True, and if there were no obvious risk of apocalypse, the same would be true of other governments, and no international pause treaty would be implemented.
In the real world, though, the Chinese government is looking a lot more pro-pause than Western governments at the moment. Undemocratic leaders care about survival no less than democratic ones. Obviously, a treaty doesn't happen without Chinese participation, and a pause doesn't happen without a treaty. For an enforceable and robust ban on superintelligence to be implemented, it will need the support of enough major parties that they will be able and willing to ensure it holds globally.
>China isn't a democracy, and is a major player in this scenario.
Agreed, and also, at least from comments I've read, Chinese popular opinion also seems much more accepting of AI than American popular opinion is.
My best guess is that if pauseai were wildly successful, commercial American AI would be killed, the field would be delayed by maybe a year, and progress would continue in the PRC (and probably also in secret military development in the USA as well).
There will be no point at which a politician will feel they can choose between apocalypse and sth else.
In the future, AIs will have worked diligently and usefully in society for a good while. Their capabilities will be better than they are now, as will their understanding of human morality and capabilites. The possibility of them deliberately harming humans will sound even more like a speculative conspiracy theory than it does now (and observe that currently, there is no global ban in the making). What do you think will be the new evidence that will change people's minds on this?
The pieces of facts you listed are indeed true (though I have heard of them before), but that is the best I can say about them. Your model of human society must be very different from mine if you think these are even 1% of what would be needed for a global ban (that is enforced by military action) to happen.
After people saw two cities blown up by nukes, they did manage to not blow each other up. They could successfully infer that if you nuke the enemy, it's possible the enemy will nuke you back, so on the whole it's not a good action even if it seems like it is locally. It was pretty tense for a good while though and not at all a given. John von Neumann, for example, advocated for a preemptive nuclear strike against the Soviet Union. Do you think arguments against AI have the same/less/more salience as seeing photos of a destroyed city?
Humans can't "automate the entire economy and create post-scarcity". And if AI can, it can also kill us all.
And, if AI can do everything humans can do, but also can be copied and can perfectly coordinate with other instances of itself, that's also enough. Even if it didn't automate the entire economy and create post-scarcity (this task should be in reach for such AI, so probably because it decided to kill us instead).
I'd rule out (5) completely, since I think it's already more than that.
(4) seems most likely followed by (3), split 80%–20%, depending on precisely where you draw the line between "significantly reshaping" and "eliminating."
I think (2) and (1) are <<1% on a timescale of 30 years.
I don't understand the worldview behind possibility of 6. Do you think modern LLMs don't exist, actually? That every ChatGPT answer is written by very fast typing Indian?
More people are becoming aware of their low agency and they want to exercise it in order to become more agentic. I think the concept itself is very basic (ie take control over your life) but it's taking on a min-max skin with all the talk about "high agency". There's a looming professional challenge in my life that I've eschewed once in 2023 and it's showing up again, and I realized I avoided plunging head first into it because of issues that stemmed from being highly agentic in parts of my life, and the exact opposite in a couple of very important parts.
Anyways, I used to have another Substack but deleted it and created this one, and I've been wondering about what I could do in the months since. I want to become more agentic, and fulfill professional potential that's currently going unaddressed.
I was wondering if anybody out there was interested in a public experiment: instead of always reading from people who exercised their existing high agency to get where they are in life, I could document my own journey from low professional agency to (hopefully) landing my dream job. It's a path full of small steps that I've avoided for specific reasons, and making it public feels like it could help with my executive function but also be useful to other people who might get inspired.
Result: 69%, better than the human average, though not as good as some humans. Still, impressive for an open-weights model that you can in principle run on your desktop.
Afraid that AIs are fooling you? Ask an AI for help!
Theory: people being more accepting of male vs female promiscuity is not about some evo psych thing where men want to know if they're the father. It's about dick size.
Evidence: asked like ten dudes at my college if they'd rather date a girl who's had sex with three guys with a bigger dick than theirs, or 10 guys with a smaller dick than theirs. Unanomous answer was 10 guys with a smaller dick.
American culture and their dick size fetish! Dick size is now a metaphor encompassing all of America’s insecurity — a good thing, because it’s laughable when taken literally.
Is it not about the evo psych thing where a promiscuous woman might likely end up pregnant and repudiated by her husband, and therefore become an additional weight for their (family-related) group?
Family has to take her back; take care of her during pregnancy and then take care of and feed the child.
Father can probably disappear - if he's known at all - at least for a while. Will not be a burden to his own group/family due to his promiscuity.
The dick thing just activates insecurities about fidelity. "If I'm not satisfying her then she'll eventually cheat." I think it's likely that people's answers to this would flip if we had 1950s-style social norms around female sexual agency. I also think that if you asked a confident hyper-attractive male (say a wealthy finance bro) the same question that you would also get a different answer ("fuck those big dicked dipshits, I drive a Porsche"). Preoccupation with penis size is typically a projection of self-confidence issues.
Guys that are confident in consistently giving their lovers orgasms don't generally worry about dick size. (Eddie Murphy had a whole skit on this, at one point).
Hold on, guys don't want women sleeping around cause they'll find out that some guys have a bigger dick than theirs? Dude, I think this is a known fact, and women don't have to sleep around to discover it.
As an old man with a smaller than average size weenie, I can tell you that though some women care about size it's not all women. There are many dimensions to enjoyable sex, only one is size. In my experience the most important aspect to good sex is being in a loving relationship with the other person. That makes all the difference in that it adds meaning to the physical act.
Some women are scared of sex with too big dicks. That said, the lady in the thought experiment already had three such partners, so this probably doesn't apply to her.
You'd still have to get into the evo-psych about why dudes consider physically superior (in some sense or another) guys to be threats to them in the first place, right? All roads lead to evo-psych if you want them to, it's the main one-size-fits-all reason people like it in the first place.
Outside of that, this kind of question is always weird because it's far more socially acceptable to assume that this kind of preference is an error (i.e. that the men aren't picking up on a pattern that matters here) than that it's useful (that in some way, the men have pattern-matched "less partners overall" and "less good partners, in some respects at least" to some kind of better average outcome). If you think traditions come from somewhere in a cultural evolution way, you have to at least consider the dudes might be in some ways *right* to have this preference, but you are going to have a hell of a time getting past societal slut-shaming filters to make a useful point if so.
And this is a very old tradition, FWIW. See: Ezekial 23:20, at least.
Guys have literal dick measuring contests in middle school (which is when this is actually An Issue, as boys undergoing puberty tend to have a lot more leadership/initiative than the boys not yet undergoing puberty, see Stand By Me).
I'm not sure this is "being seen as a threat" versus "this babe has experience with a better caliber of guy than me, I'm probably getting left at some point."
Did you control for this by listing any other factors that might indicate greater or lesser desirability? If you just say "the only characteristic I'll tell you about these people is x" obviously they're going to judge on x.
Also, asking whether you'd "rather date someone who has..." seems significantly different to "given you're dating someone, would you rather they have..."
Also, what's your explanation for girls allegedly *not* caring much if their dates have been promiscuous?
Not that any of this makes your theory any worse than most of these dating/attraction "theories", in fact it's vastly better since you actually thought to formulate a question and test it with more than n= 1 or 2. Which can't be said of 90% of such theories. But still.
We just opened applications to the Pivotal Fellowship, which is happening early next year! If you are excited about doing AI Safety research or AI governance with researchers from DeepMind, SecureBio, IAPS, Oxford or MIT, you should check it out at pivo.tl (the fellowship is not happening in Timor-Leste but in London)
Testing your most commonly eaten food products for plastic chemicals is expensive and time consuming. I’ve made it way faster (couple of clicks!) by building laboratory.love but it’s still expensive.
My initial focus has been on crowdfunding, but aside from ten funded products (60 total chemical tests run), none got funded by crowdfunding. All were funded by a single motivated individual.
So there are clearly people willing to pay to test food, but it’s hard to find these people without spending money on advertising, and thus increasing the cost of an already expensive service even further. I guess this is a classic problem of “running a business,” but in my case it’s more of a public good side project with tiny margins that happens to be business-shaped.
Looking for novel ideas for how to connect with people that want this service so that we can get more CPG products tested for plastic chemicals and continue the work that Plasticlist.org started last year.
I think you need more marketing instead of advertising. Can you blog about what you learned from the first 60 chemical tests? What you will never do now that you know X? Can you pitch a journalist about the biggest surprises or misconceptions on plastic chemicals? I’ve learned that if you make a journalists job easy and find the right person where this resonates with their audience, they are on deadline and could pick up something like this.
Can you be interviewed by podcasters or YouTubers about your project?
Could you do content marketing by creating an “everything you need to know” guide on plastic chemicals?
Could you get some kind of connection going with Consumer Lab, or some similar site? Consumer Lab itself publishes info about supplements, or things like spices used by some as supplments or drugs. so they're not a perfect fit. But they might take an interest in your service anyway. Or maybe you could test a few supplements or spices for plastic, and give them your results to publish, in exchange for their featuring you some way, or running ads for your service.
Good, glad it’s helpful. Try to network your way over to them rather than just cold calling them. They probably get contacted by lots of misinformed people and a few crazy ones about needing or knowing about tests for, like, alien semen in applesauce, and you don’t want to get lost in that crowd. I’d recommend asking on here to see if someone has a contact you could start with — even an indirect one, like knowing someone who they think probably knows some people at Consumer Labs.
Wholeheartedly seconding "The Greatest Show on Earth" and "Our Inner Fish" for 1 and "The Ancestor's Tale" for 2.
For 3, I'd add Nick Lane's "Life Ascending" and -- if on the older side -- Maynard Smith and Szathmary's "Major Transitions in Evolution", both of which focus on the mechanism of the most important transformations. Oh, and also Lane's"The Vital Question " for the very earliest steps, and Ian Tattersall's "Masters of the Planet" for human evolution in particular.
You could try "From DNA to Diversity", by Sean Carroll, Jennifer Grenier and Scott Weatherbee. It's more about the molecular basis of evolution with examples than a comprehensive timeline, but I think that covers "especially the interesting parts". It's a while since I read it, but I remember thinking it was pretty good. Another one you could try is "Your Inner Fish", by Neil Shubin, although that's just vertebrate palaeontology, so much less of the molecular stuff (I was sure I had a copy, but can't find it, so I might be being unfair there).
Please clarify. You're looking for narrative evolution from the hypothetical single-celled organism we call LUCA over 4+ billion years, but winnowed down to hominids and finally humans?
The only popular science book that does this is Dawkin's and Wong's _The Ancestor’s Tale: A Pilgrimage to the Dawn of Life_. However, some of it may be out of date. I hesitate to recommend Dawkins because he's a borderline crackpot in some of his opinions. Unless you have a background in evolutionary biology, you're likely to be taken in by his certitudes. But IIRC, this isn't one of his polemical works. It's unique for what it does. Bravo, Dawkins, in this case.
"borderline crackpot" ? To what are you referring? I really liked "The Selfish Gene". It changed how I think about evolution. As far as other books, no one has recommended the classic, "On the Origin of Species" by Darwin. A fine read.
Dawkins takes an ultrareductionist view to genes and evolution. I can see how his narrative may be attractive to some on this Substack. Belying any philosophical misgivings some may have about his motives, Dawkins ignores a lot of the current science. His gene-centered view oversimplifies evolution when he portrays genes as the fundamental unit of selection. By focusing on the gene he downplays higher levels of evolutionary interaction: such as group level selection and the multi-level selection theory of E.O. Wilson. He ignores epigenetics and developmental biology. By ignoring the different levels of causation in biological and ecological systems, Dawkin’s gene-level view is — at best — only partially explanatory. Worse yet, by treating genes as if they had agency, it creates teleological just-so stories of genetic intention where none exists.
Yeah so I agree with much of what you say, (I love E.O Wilson) and still I think Dawkins is right in that genes are the fundamental building blocks. The rest is change upon a theme. (group selection is iffy.) If it can happen it probably does happen genetically. As far as genes having agency, that's almost as silly as viruses having agency. Genes might be described as 'smarter' than viruses because they've teamed up with other genes to make an organism. Agency is in organisms, and still we are collections of genes. You might say genes get their agency through organisms.
But genes are indeed the primary unit of selection. Group-level and multilevel selection do exist, but they occur only in rather specific cases. Epigenetics is mainly important for development and has little relevance for evolution, at least in animals.
"Worse yet, by treating genes as if they had agency, it creates teleological just-so stories of genetic intention where none exists." This is really not the case. Dawkins never meant that genes had agency. The way he talked about genes is of course to be understood metaphorically. You can argue however that it was a bad idea to use these metaphor, for example like in the quote below... "« The fundamental units of natural selection … are called replicators. DNA molecules are replicators. They generally … gang together into large communal survival machines or ‘vehicles’. »
No. Natural selection does not act on genes. Their expression *may* be subject to natural selection. Likewise, their expression may be mediated by developmental and environmental factors. Moreover, higher-level social and ecological interactions may also inhibit or amplify selective pressures once they're expressed.
Whether Dawkins meant it as a metaphor, he spouted teleological nonsense when he wrote: "We are survival machines – robot vehicles blindly programmed to preserve the selfish molecules known as genes." Bah! A molecule cannot be selfish. A molecule is just a frigging molecule, with no agency beyond its role in the larger biochemical matrix of systems --> which in turn are part of a larger cellular system --> which in turn are part of larger intercellular systems --> which in turn are part of the larger organism --> which in turn may be part of social system among those particular organisms --> which may be part of an ecology that is changing over time.
The operons that code for a male peacock's tail have to pass through multiple levels of systemic emergence before they are expressed as its gaudy plumage. The female has those same genes, but her hormonal development suppresses their expression. And the tail that may attract the fickle peahen may also make the male more vulnerable to predators. The operons that code for it aren't aware of the selective pressures on the male. The male's phenotype is either selected for or against, and this process typically occurs over hundreds of generations through a stochastic process.
" No. Natural selection does not act on genes." Their expression *may* be subject to natural selection. Likewise, their expression may be mediated by developmental and environmental factors. Moreover, higher-level social and ecological interactions may also inhibit or amplify selective pressures once they're expressed. " Whether natural selection can be considered as acting on genes or on individuals depends on the precise definition used for natural selection, as several are possible. But can we at least agree that natural selection selects for genes and not for individuals? Anyway, my point was just that the gene level, as opposed to the group level, is the main level for natural selection. The process that you mention, ecological interaction for example, mostly act to change gene frequency within populations.
"Whether Dawkins meant it as a metaphor, he spouted teleological nonsense when he wrote: "We are survival machines – robot vehicles blindly programmed to preserve the selfish molecules known as genes." Bah! A molecule cannot be selfish." Very obviously, indeed, a molecule cannot be selfish and also very obviously Dawkins never meant it this way, it was just a (striking!) metaphor for the central role of genes in natural selection. Perhaps you are just objecting to the use of metaphor in science in general? What about the genetic code for example, which is in fact not a code?
"The operons that code for a male peacock's tail have to pass through multiple levels of systemic emergence before they are expressed as its gaudy plumage." Indeed, but this is not relevant. What changes through time is gene frequencies.
Had Xi Jinping made a mistake in reorienting Chinese investment from construction to manufacturing for export?
There is an ongoing slump in residential construction in China (see for example here: https://www.globalpropertyguide.com/asia/china/price-history), associated with, um, things that are usually associated with a housing bubble, from developer bankruptcies to rising unemployment.
Now, “housing bubble” is somewhat fuzzy concept. Maybe demand for apartments from households slumped. Maybe government intervened and slowed construction somehow. Maybe financial institutions stopped providing financing for some other reason. Even if it could be established that the slump is caused by the government, that doesn’t necessarily mean it is a bad thing. Maybe there was an “overconstruction” in a certain sense and the government initiated a “healthy correction”; though case for a “healthy correction would be imho stronger if it could be demonstrated that the slump is ultimately caused by falling hosehold demand, i.e. people already having, given their income level, appropriate living spaces, and not wanting more, thank you very much.
What is the case that the slump has been caused by government intervention? So, for a start, it seems relatively uncontroversial that Chinese government has a stronger grip on domestic financial sector than US federal government of EU government(s), and it is not like US federal government or central bank are unable to influence the housing market. As for China, it seems that property slump might be traced to so called “three lines policy” introduced by the government in August 2020 (see for example here: https://www.atlantis-press.com/proceedings/emfips-24/126010674). It is also noteworthy that despite ongoing slump, recently introduced outline of the new Five-Years plan treats construction almost as an afterthought, at least according to some analysts (https://open.spotify.com/episode/7mx6fqpoQnAfSZRMgowTIy).
Still the best case that the slump represents “healthy correction” I’ve read was written way back in 2023 by Noah Smith, here: https://www.noahpinion.blog/p/real-estate-is-chinas-economic-achilles. Among his points, the most important is imho that floor space per person in China is already at a level usual in developed countries, and (though he doesn’t mention that) population is projected to decline, so. However.
It is true that population decline is likely to dampen the demand for real estate in the future. But, on the other hand, current slump feels very much like a consequence of government policy, getting ahead of demographic projections. Floor space per person is one thing, but where that space is matters. Over 20 % of Chinese employment is still in agriculture; in rich countries, this number tends to be in single digits (https://data.worldbank.org/indicator/SL.AGR.EMPL.ZS?locations=CN-JP-EU-US). Presumably if Chinese are going to get richer, further “deagriculturalizasion” of employment is the way to go, and it is difficult to imagine that this will NOT result in further migration from to cities, so more demand for more apartments and associated urban infrastructure.
Was this shift a good idea? Ok, it would be one thing if the Chinese government shifted investment from a real estate to something which improves living standards in some other way. But, in fact, what seems to be happening is doubling down on exports.
Chinese exports as a share of GDP peaked in 2006, and then steadily declined until 2019 (from 36 to 18 %). But since then, their share had stabilized and even rebounded to 20 % of GDP in 2024; and Chinese GDP now is bigger share of the world economy than it was in 2006. For context, US exports are 11 % of GDP, for the EU, it is something like 18 % (so the EU is at least on paper not that far from Chinese numbers). However, wages in China are still way lower than in the EU, never mind in the US.
I see two deep problems with Xi’s export based strategy. First, there are clear signs that targets of Chinese exports are pretty uncomfortable with deindustrialization, due to concerns about employment about security (strong civilian industrial base might pretty useful if you need to fight a war), and there is, of course, talk and to a lesser extent reality about putting up trade barriers against China. But second concern is less talked about: rise in Chinese wages would erode their competitive advantage. NOT increasing wages, or, in other words, government policy of wage suppression, on the other hand, seems like a recipe for widespread discontent and political instability. Perhaps the nature of Chinese authoritarian system is such that it will be able to handle it. Or perhaps the Chinese labor productivity will rise so quickly that they’ll be able to increase their living standards even without reductions in the export share of the economy. Seems uncertain, though.
When I was over in Shanghai there were HUGE apartment complexes that were completely empty. Meanwhile, (almost exclusively) men in the city on temporary work permits lived in stacks of containers holding dorm room sized apartments for four.
I think China is trying different things at the same time: if you wanted to become the world’s supplier for increasingly valuable things (plastic Santa’s and cheap furniture -> solar panels and electric cars, you wouldn’t also threaten other developed economies by putting export controls on key resources like rare earths.
Some policy is very clever, some is the result of a million bureaucrats all trying to decode Xi Jinping Thought at the same time.
China is using its dominance in the production of rare earths as leverage in the trade war between China and the United States. This will likely cost China market share over the medium to long term as countries try to develop alternate sources of supply, but China has apparently decided that making a show of strength is worth it.
I don’t this it will affect China’s ability to sell other products because the rare earths are only products that give China significant leverage. Nobody expects China to impose export controls on solar panels or electric vehicles because those would hurt China without imposing much pain on the United States.
China’s new five year plan is a bit hard for a non-expert like myself to interpret because it contains a lot of nice sounding verbiage that may or may not be implementable. That said, I wouldn’t describe China’s new five year plan as an export based strategy, at least no more so than it has to be given that China is a major exporter. From the link a the end of this comment:
>>> In recent years, we have thoroughly implemented the strategy of expanding domestic demand, and kept unlocking consumption potential. New products, forms and models of consumption have boomed. From 2021 to 2024, domestic demand contributed 86.4 percent to China's economic growth on average.
The plan calls for some housing development, though not at the pace that we saw prior to 2021:
>>> We will facilitate high-quality development of the real estate sector, and promote the recovery and growth of the industry.
As best I can determine, the plan assumes that innovation and automation will keep China competitive despite rising wages. Competition from lower wage countries hasn’t prevented the United States from manufacturing stuff; instead it has forced American manufacturers to automate, which means fewer jobs in manufacturing. China appears to be doing this successfully in the apparel industry, remaining a major producer while paying wages which are twice those in Vietnam and Bangladesh.
Thanks for an insightful comment, that helped me to clarify my thinking, to which I have several “yes but’s”. Also, sorry that it took me so long, I’ve been busy with real world stuff. Also, I am not by any means a China expert, just someone who noticed that story conventionally told by many China experts (“China overbuilt, promoting real estate development bad, promoting exports good, and at the same time promoting consumption also good”) does not quite add up.
Yes, the outline of the next five year plan does talk about boosting consumption, but to me, it rings somewhat hollow, given that a) if I am right, by the three red lines and other policies they just crippled the means to provision housing, i.e. by some reasonable definition the single largest component of consumption, and b) they are NOT focused on what is the most straightforward way to boost consumption, namely boosting nominal incomes, even though prices are reportedly falling in China (i.e. there is a deflation).
Yes, they do not talk directly about the strategy being based on exports, but they are focused on manufacturing, and at current levels of manufacturing output and incomes domestic market is not able to absorb all that manufactured stuff, so it is de facto it has to be the export based strategy.
Yes, it is correct that by improving productivity they might be able to retain competitiveness despite rising wages, but it will be a difficult challenge, for a pretty fundamental reason that China is BIG. Like, I could imagine that either a) China will still grow faster than the rest of the world, so its exports will be an increasing share of global GDP, which will be less and less sustainable, b) China will not grow as fast as developing countries; these countries are now where China was pre-2008, so their growth will probably take the form of more manufacturing, so they’ll make some of their own stuff instead of importing it from China and at the same time compete with China in foreign markets.
As an aside, moving more people from the countryside to cities seems to me still as the simplest way to increase productivity, and the crackdown on financing of urban construction is detrimental to that.
I should qualify that by “retaining competitiveness” I mean retaining current size of the export-focused sector, as a share of GDP; I have no doubt that China plausibly can and probably will increase their competitiveness in high-end stuff (like high-end chips); I just doubt that it makes sense to try to retain competitiveness on the low-end, and I see Xi as trying to do this thing that imho does not make sense.
Paradoxically the future where China most plausibly retains the size of its export sector is imho the one where the global growth is driven neither by China nor by developing countries, but by rich countries, perhaps because of AI. In that case China might have a lot of demand from then even richer consumers of the West, like in the old times, in effect by being pushed lower in global supply chains. But I guess Xi would count that as a failure.
Maybe their best course of action would be to hire Alec Baldwin to reprise his role in Glengarry Glen Ross and travel around China giving a series of pep talks to real estate professionals there.
Has anyone tried using CLA (Constraint-led approach) to learning piano? I realized that I can play something at home just fine, but if I play at a different piano somewhere else, or even sit in a slightly different position at my piano, it's suddenly much tougher. CLA is used a lot in sports and it involves things like changing the environment or the exact requirements of an action instead of practicing the same thing every time. In the case of piano, I guess it could involve sitting in different positions or changing the rhythm of the piece, but I'm not sure.
I teach piano and often provide students with CLA-inspired practice variations. Elementary aged kids find them really fun and motivating. At this point I have about 50 “challenges”, such as playing with various postural disturbances or under distracting circumstances.
The issue of overlearning a piece in a kind of comfortable quasi-conscious trance, that immediately collapses under pressure, is a normal, well known problem, and any good teacher will have a repertoire of practice methods to help you overcome that obstacle.
Oh my goodness, I can probably say more than you ever want to learn about practice methods. I’ll just post the basic list I use in lessons.
It’s lightly “narrativized”: I set up a scenario where they are moving through a labyrinth and have to do a challenge to move to the next room.
I also draw heavy inspiration from the show “Taskmaster”, if you know it. The contestants often have to do relatively trivial tasks with diabolical constraints.
Anyway, here’s the list:
1. Play the left-hand part with your right hand. “Magic mirror”
2. Play the right-hand part with your left hand.
3. Tap the rhythm of the right hand (on knees, clapping, stomping tap foot). “Knock on the magic door/unlock the gate”
4. Tap the rhythm of the left hand.
5. Sing the rh/lh while the teacher plays. “Teamwork to make the monster sleep”
6. Tap the big beat while the teacher plays. “Match the steps of the giant”, “break big rocks or push heavy thing”
7. Tap the little beat while the teacher plays. “Match the steps of the goblins”
8. Eyes closed. “Dark passage”
9. One foot. “Foot stuck/stubbed toe”
10. As high as you can on the keyboard. “You have to reach up over an obstacle”
11. As low as you can. “Crawling under a gate, through a hole”
12. Tap your foot while you play. “Ants!”
13. Absolutely perfect notes, played very slowly. “Tightrope, narrow bridge”
14. Make up words to the rhythm. “Secret password/troll bridge”
15. Sing the rhythm as you sway/march/dance. “Long march, stay awake”
16. Play as fast as you can with lots of mistakes. “Attacked by bears bats wasps”
17. Play so softly the keys don’t make a sound/tap the keys without pressing “booby traps, don’t wake the monster”
18. Transpose/mode change “gatekeeper speaks another language”
19. Hands extra far apart “reaching across a gap/rock climbing”
20. Start from the other side of the room “reach the closing gate”
21. Uneven rhythms “hurt leg”
22. Conduct the teacher’s dynamics and articulation “cast a spell on a toy puppet”
23. Balance a playing card on your head as you play the passage “you must cross a tightrope and keep your balance”
24. Pat the rhythm on your body with one hand, while you play with the other “a dragon breathes fire on you”
25. Play with one hand while you fight off a monster with your other hand (spoon) “battle a knight”
26. Answer a question from your teacher while you play “troll bridge”
27. Hold something in the spoon as you play with one hand
28. Follow the teachers commands to play fast or slow “dodging arrows/lava bombs”
29. Play as staccato as you can “the floor is lava”
30. Tiptoe as you play “sneaking past the guard”
31. Follow the teachers commands to open and close your eyes “lightning storm”
32. Stop and go when the teacher commands “avoid the patrolling guards”
33. Mime the passage (tapping on keys without making a sound)
34. Follow the teachers commands for “silence” and “play” “
35. Play with aggression and confidence, with a confident pose “scare away the bear”
36. Play as tense as you can, only moving your fingers “frozen by a witch”
37. Say “banana” at the beginning of every bar “secret password”
38. Play while yawning “sleep spell”
39. Play standing up with one hand behind back “fencing”
I have a private studio in Pittsburgh. I really got a chuckle out of reading your question, because it’s such a specific question that just so happens to precisely line up with how I teach.
There’s so much excellent research about skill acquisition in the athletic world, with literally billions of dollars behind it. The research on musical skill acquisition is of course much thinner, so we have to be kind of multidisciplinary.
I dont know how far along you are in your studies, but there’s a lot to be gained from reinforcing your muscle memory with more explicit mental models — generally in order of abstraction these would be knowing the key/scale, the chord at each change of harmony, familiar chord progressions, and then phrase models or larger structures. The classic, time-tested method of training this is through transposition, decomposition, and decomposition/variation.
Aha, a lot of these things I never learned correctly. I never properly learned theory, I look at my hands too much, I probably hold my wrists wrong, etc. Sometimes I wonder if I need remedial lessons. :). Anyway, I'd love to discuss with you further if you're game (and if you're not -- that's fine). Email is wheelbarrow 2 4 6 8 (at) gee mail dot com.
Air's bad outside today, may want to skip any walks you're doing, if you're in the eastend. Senator Fetterman collapsed in Braddock, and he's not the only one with heart issues because of the air.
I recall a post on a piano forum a long time ago (which I will never find again) that listed dozens of increasingly tricky ways to practice scales, including speed and rhythmic variations but also stacatto, and stacatto with one hand and legato with the other, and then switch.
I also practice starting the piece from random points in the middle to throw off my muscle memory.
Interesting! It will definitely throw me off to take a passage from a piece and do the LH staccato and the RH legato, and then switch it. And then do forte in the LH and piano in the RH and then switch. Starting at random points is def a good idea. Any others? They didn't have you play any of the passages backwards did they? :)
If I wanted to get really good at lifting 100 lbs for 8 reps at some lift and being able to do it in front of a crowd, I would train until I could do 115 lb 10 times and I would practice related lifts too. I'd love to find the analog for piano performance!
Unfortunately the post is something I read 20 years ago and wasn't able to find again, but I believe varying/alternating the dynamics between hands was also included. I think playing the piece super-slow (like as slow as humanly possible) now and again was also a suggested method on that forum for breaking reliance on muscle memory.
As far as changing the rhythm, that was a common technique my bassoon teacher would use: if the passage is straight eighths, play it dotted eighth-sixteenth, &c.
I'm pretty sure randomly changing parameters is differential learning, not CLA
CLA is when you have identified the specific mechanism that causes your problem, and you change your constraints in a specific way that's targeted to the mechanism:
But isn’t the justification of CLA (I’m using football as an example) that e.g. learning to play at high elevation, playing for long periods of time, and playing faster for the same period of time all take highly overlapping physiological mechanisms? I don’t see how that applies to piano.
One of my teachers did recommend changing the rhythm of straight quavers to practice evenness and precision; does that count?
I don't remember the source, but I read about an athlete who trained himself to not be superstitious-- varying his socks etc. so he wouldn't have beliefs about lucky trivia.
On the off chance this is useful to someone: some Yamaha electronic pianos have an infamously heavy action, like my Yamaha P120. Not sure about newer models.
Just started trying this approach for Brazilian Jiu Jitsu. So far it’s a lot more fun and I feel like I’m internalizing more general principles but it’s too soon to tell how well it’s working.
I wouldn’t be able to self teach like that. I think in order to design the constraints you need an expert. I’ve heard Rob Gray has some good books on the topic
I read those books! (And a few more by Frans Bosch and others in the ecological dynamics space). And I am using these ideas to shape my training for rock climbing. I write about it here https://dlants.me/climbing-ed-cla.html and I have a few more posts on my blog about applying it to capacity training and even software teams.
Ecological dynamics is everywhere. I have found understanding the principles of the theory immensely useful to many pursuits.
Scott has written a few times about Williams syndrome where people are pathologically naive/innocent. I feel that quite a lot of people are like that on the internetthough with no medical cause. I would be really interested to see if the effect has been studied.
Have you ever met someone with Williams syndrome? I’ve never met anyone on the internet with pathological naïveté anywhere near the intense level in Williams syndrome
I miss Scott's longer form essays. I understand he's much busier now but I don't think we've had something on the caliber of "Meditations on Moloch" or "The Control Group is Out if Control" in a while.
My impression is that he did better when he was a bit of an outsider in rationalist circles, and working hard to get accepted. Now that he has arrived, both physically in Berkeley and metaphorically, he isn't trying as hard and it shows.
This makes me wonder if the problem is just the audience aging. I was in high school when I read Moloch, and I was impressed because I had never thought about ideas like it before. I was a lot older when I read Colors of Her Coat, and found it less impressive. It's a description of ideas I already live by and have thought about many times before (plus jabs at modern art and architecture that I disagree with, which no doubt biased me towards disliking it).
I still think Moloch is one of his better works, but I wouldn't be nearly as impressed with it now.
I really wish he'd return to things about argument and dialogue, like "Guided by the Beauty of Our Weapons". This sort of thing seems unbelievably useful, something there's a lot of demand for, and he mostly hasn't touched it for years.
Alas, as a fan of Meditations on Moloch and The Control Group is Out of Control, I did not like that essay much. I appreciate the effort he spent on it, but I have little interest in long discussions of dry historical events.
So, looks like the notion of shortening medical school by a year is now far enough inside the Overton window for an opinion piece in the New York Times.
I think that makes sense. It doesn't, after all, take four years to learn how to tell me that my symptoms are all in my head and suggest antidepressants.
They typically have a list of required prerequisites which add up to a bit less than two years of study. And they require an undergraduate degree, but it can in principle be in anything. About 50% of medical students have undergrads in life sciences, and a further 25% in natural or social science.
So they could drop the degree requirement while retaining the "premed" requirements. It gives 4 yr students a chance to drop out a year or two early and go to med school if they get an acceptance. They run the risk of failing med school and having no degree but it also doesnt seem so hard to create a pipeline to let them back into undergrad and finish the degree. Anyways its an option they currently dont have. Also lets community colleges setup 2yr premed degrees. Lots of cost savings without any compromises on medically relevant learning.
Those are almost all 8 year programs. Some are 7. 4 years of med school + 2 years of pre med would be 6. And if the premed is a community college its very cheap. Still seems to me like they are mostly gatekeeping med school behind an expensive 4 yr program
It seems to me that the top 5-10% of the US population could easily knock out two years of college before turning 18 (via dual enrollment, AP, etc.) if provided the opportunity. Then they could get a bachelor's degree by 20.
But it seems to me that the problems with medical education are more with limited med school/residency spots than the need for an undergrad degree.
Most of the hidden posts are the weekly-ish subscriber-only open threads. They go up on Wednesdays (about 2/3 of the time, I think) and are usually quite a bit quieter than the public open threads.
Besides that, there's a substantive hidden post maybe every month or two. Usually something Scott deems to be too personal, controversial, or silly to make a public post about. They're occasional treats for subscribers, not a hidden world.
No, that's only if you have the direct link. If you don't and are simply scrolling through the archive, you don't see them at all. (And you don't get email notifications about them either.)
Awesome and interesting blog. Great things you’ve always shared with us. https://www.kdealer.com.co
It's not obvious to me that people with higher educational attainment deserve more power, influence, and status. So I don't know how this is a confounding Factor.
To.what?.Is this a propos of something else? In any case, there is a difference between saying someone morally deserves something ,and saying they are the best person to do it practically.
I'm not sure what comment I was replying to so I can't tell you why I said "confounding factor."
You have implied that, generally speaking, educational attainment to wield influence and power properly. This is quite a leap and I can't go there with you.
Where I am willing to go: Some positions of power and influence have SPECIFIC educational requirements. Other positions of power and influence have IQ requirements, irrespective of educational attainment. A modest number of positions of influence do not have requirements of either kind.
As far as social STATUS goes, the stakes are much lower. We will likely always have many high-status people with lower IQs and educational attainment. I'm thinking of various kinds of artists, celebrities and religious figures.
I 'm not saying that education grants you the ability to know right direction, but it does grant you the ability to implement a plan competently. the direction can be decided democratically.
I recently realized that uneducated and/or low IQ people fill an important role in our democracy: stretching the Overton window. They don't push it in one direction or the other, it's more like they stretch it to make it larger. This stretching benefits bad ideas as well as good ones, but we should be willing to pay that cost. It's better to err on the side of having more ideas in the marketplace of ideas.
Uneducated people are often have wide-ranging and diverse political views. A mix of opinions that left-coded and right-coded. This is because autodidacts gather political opinion somewhat randomly, picking up bits of ideology here and there.
Many eucated people reflexively call an idiosyncratic political program "incoherent." When in reality, the "coherent" policies of leading political parties are grouped together somewhat arbitrarily. For example, "socially conservative, economally left-liberal" is a tendency seen in many parts of the world. It's an accident of history that this tendency is completely foreign to modern American thought. (Tipper Gore and Republican-approved deficits notwithstanding)
Yes...up to a point. But populism also includes ideas that just don't work like increasing welfare and decreasing taxes. Some things are incoherent, not just unusual.
It seems like your reply may have two functions. It's a critique of valuing “Overton stretching” and also an assertion that “Sometimes when you call someone with idiosyncratic beliefs incoherent, that's just calling a spade a spade.”
I agree with the latter assertion.
As far as the critique of Overton stretching, it is non-obvious to me that this stretching will benefit populism in a disproportionate way, or in a way disproportionate enough to nullify the value of the stretching. After all, autodidacts and dum dums often hold a mix of populist and elitist ideas.
Suppose I were to suddenly understand that political eccentrics and dummies are TRULY incoherent in most cases. Even so, it does not necessarily follow that this incoherency would nullify the value of Overton stretching.
I guess I see “narrow Overton windows” as a pressing problem for America.
This is tricky. On one hand, I am abstractly in favor of the same dignity for all; smart and stupid alike. On the other hand, I would prefer that the smart make the important decisions. But the ability to make decisions = power, influence = status.
Dignity and status are not synonymous.
I spoke about "people with higher educational attainment." In your response, you flatly educational attainment with being smart. Hmm.
Also, there are different kinds of power than the power to make important decisions for others. Nor do power and influence necessarily equal status. Inasmuch as this is true, it is only true on a contingent basis.
This can be readily demonstrated by looking at the culture of the American radical left. Wherein leaders are deprecated to a ridiculous degree.
Why was the Book Review about the Russo-Ukranian War removed? I can't find it anywhere :(
From https://www.astralcodexten.com/p/non-book-review-contest-2025-winners:
"This review was about a military topic. Due to its sensitive nature and a changing situation, the author has asked that it be removed and that he not be acknowledged in any way, sorry."
As a subscriber, don't you get emails with the entire text on the post? This was on Sep 26.
[EDIT - Here was an empathetic but superfluous and mistaken comment about Gaza]
I understand why Lemkin and the United Nations pushed for an extremely broad definition of genocide. I know they had good intentions. But ultimately, I think we should reserve "genocide" for any time-limited, substantial population reduction that was deliberate. We should probably insist that the reduction had to have happened in 15 years or less.
The reduction needs to have been executed through violence or directly coercive mechanisms. So forcibly sterilizing your people would count, but plying them with alcohol or outlawing their language WOULDN'T.
Once the United Nations signs on to this redefinition, some genocides might get recast as attempted genocides, quasi-genocides or near-genocides. Or maybe we'll use a handy new word, genocidistic. I know that recasting a long-passed genocide as quasi-genocide might upset some victims. However, we could ameliorate this by explaining why the downgrade won't slow their quest for sympathy/redress: Limiting our use of "genocide" gives the word greater power, which transitively gives "quasi-genocide" new weight. A self-correcting dilemma!
I know that defining a present crisis as a "genocide" mandates certain governments to take corrective action. But realistically, said governments TYPICALLY avoid that definition if it's not politically expedient.
The Armenian Holocaust and the Jewish Holocaust were unusually systematic and documented, while the Rwandan Genocide was unusually brazen. Few genocides are similarly indisputable, so governments ALREADY have the option of refusing to recognize your genocide--at least until it's too late for interdiction.
IOW, "if Uncle Sam hasn't recognized your genocide by now, they aren't going to. At least not while it matters." This is getting long-winded, and the benefits of tightening the United Nations definition are at least somewhat obvious. But I will list those benefits later if anyone asks.
I'll bite: what are the obvious benefits of narrowing the definition of genocide, and to whom would they accrue? I see benefits for the few groups who would also clear the higher bar – fewer people to share victimhood points with, and so improving their standing in the Oppression Olympics – and for people like yourself who are annoyed by the term becoming overbroad (not a particularly sympathetic constituency, you'll admit), but anyone else?
The argument that downgrading groups from "genocide" to "quasi-genocide" (or some similar term) wouldn't slow their quest for sympathy/redress seems patently absurd: it seems clear that they're getting more than (what you think) they're due is the very reason you wish to restrict the definition to clearly exclude them from claiming the label and its associated status, such as it is.
Shankar Sivarajan:
“I see benefits for the few groups who would also clear the higher bar…”
I think more than a “few” groups would clear the bar, actually. I think PLENTY of little-discussed genocides are waiting for their rightful day of prominence, even under my stricter definition. At least three genocides perpetrated by the British need to be widely recognized. One from the 20th century.
BTW, aren't you a confirmed right-winger? In that case, I shouldn't have to retread any arguments for reining in "genocide mania." You must know them all by heart, like everyone else who says "Oppression Olympics."
I sincerely DO believe that downgrading certain claimed genocides to quasi-genocides would have little or no harm on "downgradees." Their crucibles were so diffuse or contested or numerous or long-passed that humanity doesn't take them very seriously as genocides RIGHT NOW. Since these downgradees aren't getting much of ANYTHING, I never thought to worry they might be getting "more than their due."
But if that DID worry me, well, what's wrong with that? You frame it so grimly! Human attention is a zero sum game, and don't we WANT the most brutal crimes to receive priority remediation?
I didn't design my little proposal with intentions to materially harm ANYONE, promise! At most, it would ruffle a few pleathers. Also, my mention of Gaza represented a brain fart. I realize now: I don't know enough about the true Gaza death toll to say if they've been genocided under my new definition. Though it looks as though they've been genocided under the United Nations definition.
It's a flat fact that Epstein was the first person to successfully commit suicide in that prison in 21 years. This doesn't prove anything, but if it's a coincidence, it's such an incredible coincidence that it ALONE would justify any number of dark suspicions. There are a number of other incongrous and strange facts about Epstein and Maxwell. Nothing definitive, but enough to fully justify conspiratorial speculation.
It's all circumstantial evidence, which means it's hardly evidence at all. But why does Richard Hanania expect writers and podcasters and pundits to follow the same strict evidentiary standards as a district attorney? His scathing denialism would be consistent with a personality problem or something worse.
I guess I'm kind of issuing a challenge to any other denialists here. If you think the Epstein case is a big nothingburger, fine. But in light of the 21-year thing, how can you in good conscience belittle or mock the conspiracy theorists? Is it just a habit or reflex that is hard to break? If the conspiracy people are easily dismissed in all other cases, this is one instance in which the circumstances seem to demand lenience on your part.
I don't think anyone specifically tried to fake his suicide in an effort to make sure something he knew didn't get revealed in court. That would require multiple people involved in the conspiracy. Even if you think Trump is willing to order it, I question if everyone who would have had to carry out the order would keep quiet.
But I would not be surprised if, say, a prison guard said "yeah, I know we're supposed to be watching closely because he's on suicide watch, but fuck him, he's a pedophile." Malign neglect seems more likely than murder.
Is it such a huge stretch to start at "neglect from individual animus" and stop at "neglect from orders issued by a Certain Nasty Order Issuer?"
> Even if you think Trump is willing to order it, I question if everyone who would have had to carry out the order would keep quiet.
If they have the means to assasinate someone as influencial as Epstein, I could easily imagine why everyone involved would keep quiet...
There's conspiracies and there's conspiracies. The Acosta deal was certainly fishy and it's likely there was a certain degree of a rich and connected person getting favorable treatment normal people wouldn't. We also know for a fact that Trump's people were trying to cover up embarrassing communications with Epstein since a lot of them have come out now.
But thinking that Epstein was murdered is another thing entirely. To believe that, you need to have the kind of epistimistics that make you bad at modeling the world.
Also, what's the base rate of prison suicide attempts? That's a number conspicuously lacking from your argument.
Of course it would be extreme to conclude from the weird circumstances that Epstein definitely got murdered or was allowed to kill himself. But there are enough weird details about Epstein's life and death to cast virtually EVERY aspect of his Official Story into question. Including the official line on his death.
Hanania is following the typical skeptics' playbook, by assuming the Official Story is more or less accurate, and putting the onus on Epstein conspiracy theorists to justify their skepticism, line by line.
Because of the strangeness of this scandal,, the standard skeptics'playbook DOESN'T WORK here. And charitably, I'd say most of those sticking to the playbook must not be well-informed about the public details of the scandal. But when it comes to someone like Hanania, who seems to have taken his time researching Epstein, such rigid skepticism can only be explained by naivety, credulousness or some other problematic motivation.
I think that after 21 years, the time and place of Epstein's death make it a black swan event that justifies grave epistemic uncertainty. Without need to consider suicide attempts. But to be congenial and thorough, I looked into that matter anyway. Based on the available data (which is not especially satisfactory), it is estimated that there were at least 20.suicide attempts per year at that facility in the decades leading up to Epstein's death. So yeah, it's an incredible black swan event.
> To believe [that epstein was murdered], you need to have the kind of epistimistics that make you bad at modeling the world.
I think this is wrong (well, I'm very confident it is wrong). Can you justify your claim?
Well the poster said “that prison”, which is a qualifier.
I think it’s pretty naive to think that well connected people get away with sex trafficking (which would have been the charges against Epstein, as they were against Maxwell, eventually).
Furthermore anybody who remains friends with Epstein after 2008 must have known about those charges and the plea bargain. The furore about Trump “knowing” about the girls seems naive or political, everybody knew about the girls.
Peter Dafeel said:
"I think it’s pretty naive to think that well connected people get away with sex trafficking (which would have been the charges against Epstein, as they were against Maxwell, eventually)."
You wouldn’t bat an eyelash if someone suggested this was happening in Saudi Arabia, Albania, Russia, Mexico, or China.
But you don’t think this could happen in the United States? I’m not being deliberately obtuse; of course the United States has stronger rule of law, more empowered investigative journalists, and many other time-tested checks on our elites. I think these checks ensure that fewer elites can traffic fewer victims, and that said elites have to go to greater lengths to keep their crimes hidden.
But saying such trafficking is flatly IMPOSSIBLE? That claim speaks to an American exceptionalism of baffling intensity. And it reads like an implicit mockery of most other countries, though I doubt that implication was intended.
Even in the developed United Kingdom, BBC star Jimmy Savile presented an interesting case of elite impunity. Savile sexually abused at least 450 children. Shockingly, the villain died in 2011 without being brought to justice or seriously investigated. NHS investigator Dame Janet Smith later found that 100 BBC employees were told about Savile’s misconduct but did not take decisive action. Another 17 suspected misconduct but kept their mouths shut. These 117 people included five senior staff members.
Surviving victims and their advocates argue that Smith’s report, though sickening, was a “whitewash” that hid much deeper culpability on the part of BBC staff and Crown officials. The proven details of the scandal destroy so much faith in British institutions that we are almost left thinking that ANYTHING could be true about this case. Unlike in the Epstein case, this “whitewash” narrative is taken seriously by a broad array of mainstream British figures. From academics to elected officials to BBC employees themselves.
> Even in the developed United Kingdom, BBC star Jimmy Savile presented an interesting case of elite impunity. Savile sexually abused at least 450 children. Shockingly, the villain died in 2011 without being brought to justice or seriously investigated.
https://en.wikipedia.org/wiki/Jimmy_Savile
https://en.wikipedia.org/wiki/Jimmy_Savile_sexual_abuse_scandal
Some interesting reading.
> During his lifetime, two police investigations considered reports about Savile, the earliest known being in 1958, but none had led to charges; the reports had each concluded that there was insufficient evidence for any charges to be brought related to sexual offences. Sporadic allegations of child abuse were made against him dating back to 1963, but these only became widely publicised after his death. His autobiography As it Happens (1974; reprinted as Love is an Uphill Thing, 1976) contains admissions of improper sexual conduct which appear to have passed unnoticed during his lifetime.
If I am reading it correctly, he was accused multiple times, investigated twice, even admitted it in his biography... but nothing happened.
> Former Sex Pistols and Public Image Ltd vocalist John Lydon [...] in an October 1978 interview [...] stated: "I'd like to kill Jimmy Savile; I think he's a hypocrite. I bet he's into all kinds of seediness that we all know about, but are not allowed to talk about. I know some rumours." He added: "I bet none of this will be allowed out." As predicted, the comment was edited out by the BBC prior to broadcasting
> In March 2008, Savile started legal proceedings against The Sun, which had linked him in several articles to child abuse at the Jersey children's home Haut de la Garenne. At first, he denied visiting Haut de la Garenne, but later admitted he had done so following the publication of a photograph showing him at the home surrounded by children.
> In 2012, Sir Roger Jones, a former BBC governor for Wales and chairman of BBC charity Children in Need, disclosed that more than a decade before Savile's death he had banned the "very strange" and "creepy" Savile from involvement in the charity. Former royal family press secretary Dickie Arbiter said Savile's behaviour had raised "concern and suspicion" when Savile acted as an informal marriage counsellor between Prince Charles and Princess Diana in the late 1980s, although no reports had been made. Arbiter added that during his regular visits to Charles's office at St James's Palace, Savile would "do the rounds of the young ladies taking their hands and rubbing his lips all the way up their arms."
Sounds like many people knew, including some high-status people, but no one wanted to risk a lawsuit.
It could be that suicides in prisons are generally very rare. In that case, for almost every prison suicide we look at, we'd find that that it was the fist suicide in that prison in a long time.
A suicide occurred at that prison in 1998. Afterwards, at least 420 attempts failed before the next successful suicide, Epstein's. That's WEIRD.
Please note I never crosshsaired ALL Epstein skeptics. I criticized Hanania, along with all whom I would deem "aggressive Epstein denialists."
Hanania has mockingly conflated all conspiracy nuts with all Epstein "truthers." At least in terms of their intellectual seriousness. But this is an unusually diverse and MASSIVE cohort of truthers. This massiveness is only PARTIALLY explained by referencing algorithmic information bubbles, sinister apps or other NYT-approved theories. I think the massiveness is largely explained by the simple fact that the Official Story is IMPLAUSIBLE.
Excuse a brief sidebar on definitions. Most Americans primarily use a colloquial definition of conspiracy theorist (CT): "A dysfunctional nut who is notably gullible and misinformed, whose dysfunction manifests itself in paranoid monologues about vast conspiracies, usually comprised of global elites." More often than not, these paranoids readily became Epstein truthers. However, Epstein's death forced many "normals" to become "CTs in name only." These functional non-nuts, not given to paranoia, were mugged by reality into suspecting that a clique of powerful sleazeballs networked to shape the life and death of Jeffrey Epstein. These Epstein truthers are CTs according to the LITERAL, not colloquial definition of CT.
Has anyone commented on the alcohol use vs dementia study published in BMJ in September and very widely cited in the press as demonstrating more or less conclusively that ANY amount of alcohol consumption increases the risk of dementia? The coverage asserts unequivocally that there is no safe "floor" amount, and that the oft-mentioned "glass of red wine a day" is dangerous, not beneficial or even neutral.
I just read the study, and, although the authors make similar claims, their study seems to me to be something else altogether. It seems to me that what they did is associate certain genetic traits to alcohol misuse with increased dementia. The part of the study involving actual people showed the "U Curve" that we've heard about for twenty years.
I won't go on, as I might be misreading the whole thing; but it was a big surprise to read the study after reading all of the coverage of it.. Has anyone discussed this here?
https://ebm.bmj.com/content/early/2025/09/16/bmjebm-2025-113913
I’m interested in this too. Went to BMJ article, thinking Oh, I bet I can figure it out, but soon discovered that doing that was going to take an hour at least, and bailed because I’m busy. So I’m in the same boat as you and John Schilling. But I do have a suggestion about how to get clear about whattup in that article. Ask GPT. I have the $20/mo version, and for that I get several GPT5 options including the “Think” version. I have had a lot of success regarding vexing things like this article with having it not just summarize the article, but explain things to me. What I’ve found to work best is to be very detailed in my prompt. For example, I might tell it to explain the logic of each stat it mentions. (“A Snarg test was used to mathematically compensate for the 2 main confounds, A & B by raising or lowering each subject’s score by the amount the 2 confounds would be expected to change it.”) So for this article I would start by stating the thing that confuses me: How can they get the usual U-shaped distribution, but also say that the more you drink the greater your chance of dementia? Then I’d ask for succinct of results and their interpretation, with a brief conceptual rather than mathematical explanation of how they arrived there, an explanation of how they got the odd double result, and a judgment of how valid their interpretation is.
If you do it, hope you’ll share what you found out.
I have not read the study, but have read commentary about it in several places. All of whom described something mathematically nonsensical when it comes to the one question I actually care about: what is the absolute increase in dementia risk going from zero drinks per day to one drink per day?
I don't know whether that's because the study was mathematically nonsensical, or because the study looked correctly but only at higher drinking rates, or because all the commenters are the same sort of idiot. For that I'd have to read the paper, which would be tedious and I've been busy.
Has anyone read Escher Gödel Bach? Is it worth it?
I have read it as 20 years old. I found it fascinating, sometimes too complicated, some parts boring.
I enjoyed it as a teen in the early 80s - the thematic connections he drew between Bach’s music, Escher’s art, and Goedel’s mathematical logic were fascinating to me at the time. The whimsical dialogues between Achilles and the Tortoise were appealing in an Alice in Wonderland way. In retrospect, I think the book did a great job of popularizing some profound ideas (to the extent such things can be popularized) while not necessarily bringing all that much profundity of its own.
Read it maybe forty years ago, found it to be enjoyable and entertaining and insightful. But I haven't read it since, and I suspect that I have since seen almost everything in it, done better.
I don't recall it being the sort of book where you struggle through fifteen chapters and then "aha, that's what he's getting at!". I was always struggling a little bit but also always getting something out of it. So if you read a few chapters and find yourself saying "meh, this is nothing special", you probably won't lose anything by bailing on it.
In fact, I think 15-ish chapters as where he lost some people, as a lot of computer/logic people faltered at the long chapter on DNA, about 2/3 through.
Also, the stylistic/meta cleverness was amazing in its time but I think would seem a bit dated now.
I tried reading it but found it frustrating and boring enough that I quickly gave up. And I'm a math-loving nerd who grew up on Martin Gardner, so I'd be pretty much the best possible target audience for the book.
For people in my world (people interested in logic, philosophy, minds, and machines, particularly ones born in the 1980s) that book was incredibly influential. It’s a beautiful book, though difficult in some ways.
It’s a bag of shit with maybe a diamond in the centre. Or maybe more turd. I’d have had to finish it to determine.
Report back please
I tried long ago to read it, and bailed. My memory of it is that the conceptual shape it was talking about was deliciously strangely recursive, but that yards and yards of clunky prose were expended on saying in many many ways, "look at this for god's sake! The stuff Godel was playing with was the same shape as the stuff Escher was playing with. The stuff Esher was playing with he same shape as the stuf Bach . . .. It's recursive! See, here's some Godel resursiveness, and here's some . . ." But maybe I was just tired and cranky that day.
Changed my life for the better.
How so?
Led me to decide to go into comp sci and take up coding for a career.
Also - as a musician I loved the Bach part, as a fan of Escher I loved that part, as a math major I loved the way he explained Godel's big idea.
Who is “Bubba”? Inquiring minds want to know.
My guess: a sign they've overplayed their hand against Trump. The conflation of Trump having sex with a willing sixteen/seventeen-year-old girl (probably true, but no one cares) with him kidnapping and raping babies (rhetorically insinuable, and tricky to clarify, as Megyn Kelly just found out) is politically useful, but including claims of him fellating Bill Clinton (obviously preposterous) makes it easy to dismiss the whole thing as a hitjob.
Who is the "they" that's overplayed their hand? I've seen people on Bsky and Tumblr cracking jokes about "outing Trump as bisexual," but I don't think anyone with actual power in the Democratic party is unironically claiming he sucked off Bill Clinton.
Like, I don't think you can say "they" made a strategic blunder when "they" don't have any control of what the Internet latches onto.
I didn't buy the "It's just college students and a few people on Twitter" schtick a decade ago when it was about Social Justice and Political Correctness, and I'm not buying it now. Though decentralized, this faction has plenty of power and influence over the Democratic Party.
But this rumor is so obviously dumbly untrue it’s hard to believe that even the most cynical and self-righteous members of “them” would try to get mileage out of it. The person who first told me about it was in the college kids and Twitter demographic. I laughed, but said “there’s no way that can be true,” and he replied that yeah, he knew that, but it was just so damn funny. Btw, he’d heard a variant: Trump and Clinton did the deed with Putin watching, cuz Putin kink or something.
Also Voltaire was there watching the whole crew, and he said “I may disapprove of every single object of you desire, but I will defend to my death your right to suck it.”
Are you similarly accepting of people who call the Republican party Nazis because of Nick Fuentes and his ilk? Or is it only the Democrats who are expected to exercise perfect message discipline over every single member with a social media account?
(And like, Fuentes is at least someone with name recognition! The people I've seen making jokes about Bubba are much less known than he is.)
> Are you similarly accepting of people who call the Republican party Nazis because of Nick Fuentes and his ilk?
Why would anyone accept that? Nick Fuentes isn't a Nazi. He is a white supremacist, but that's not as unpopular of an ideology as you think it is.
I agree there's no way he did Clinton, or even Bannon, even given how gorgeously hot Bannon is. The guy's straight. But there actually is a text, from Epstein's brother to Epstein, I think, asking whether people knew about Trump "blowing bubba."
It's not gay if you say "no homo" first.
I can only assume this is in reference to Bubba Ho-Tep, and that Donald Trump is secretly Elvis.
Peptide therapy seems to be taking off, which is an exciting development, particularly with RFK Jr. and the MAHA guys on board with it as "alternative medicine."
It seems that Jeffery Epstein was not a fan of President Trump, perhaps because Trump reminded him of the low-class women he paid for sex.
I'm trying to guess what kind of gent would be the type who didn't care much for Trump OR Epstein's victims. That's a tough one! Maybe a SUPER-libertarian NeverTrumper?
>Maybe a SUPER-libertarian NeverTrumper?
That's me. Many, and quite possibly all, of Epstein's "victims" willingly prostituted themselves for money. You don't have to be a libertarian, just the kind of normal, high-class American who traditionally had contempt for both prostitutes and sleazy used car salesmen like Trump.
This seems likely. Epstein was a pimp. A bunch of his prostitutes were underage, which didn't bother him just as it doesn't bother other pimps. You think anyone is checking IDs and birthdays on the girls working a truck stop parking lot? Why would it be different for the pimp to the rich and famous?
I expect there was a mix of kindness and menace, gifts and threats, to keep them in line, but again, this is apparently just what pimps do. The fact that he was a rich high-class pimp probably meant that the prostitutes working for him had a higher standard of living than most prostitutes. But still, it's not like he was looking out for their welfare or cared about them as humans or was above coercion or mistreatment to get what he wanted from them.
Perhaps, or with less deep psychoanalysis, because he thought (with good reason) that Trump ratted him out to the feds.
Don't the feds have a history of finding drug dealers to rat out other drug dealers? While giving the rats carte blanche to keep dealing their little hearts out? You'd be one helluva lucky rat if you could finagle a BEAUTIFUL deal like that! 🐺🐻🐼🦝🫣🐺🦊🦝🐨🐻❄️
Trump also made the guy who gave him a sweetheart deal his Secretary of Labor.
This whole idea that sex with an underage prostitute = life in prison is new. Back when the "sweetheart" deal was made, such a deal was entirely uncontroversial. Indeed, such deals are still being made today. The NYTimes recently published an article about "sex trafficking" detailing how a "sex trafficker" named "Double R" was punished with "180 days in county jail and probation, plus an anger-management program."
https://www.nytimes.com/2025/10/26/magazine/sex-trafficking-girls-la-figueroa.html
Leftists focus on Epstein because he's an example of the "powerful white men"* who they say are the ones doing the sex trafficking, unlike the aforementioned Double R. They don't have any other examples, which is why they're still talking about him six years after his dead, and will likely still be talking about him sixty more.
* They get really mad if you point out he was Jewish.
A pivotal battle from European history that is too little known: the Battle of Muret (1213). . This battle took place in the final stage of the Crusade against the Cathars.
Disproportionately women and children, the Cathars had pacifist beliefs, and did not participate much in the fighting. Instead, they were defended by Catholic neighbors who took the injunction "love thy neighbor" seriously. Of course, the Crusade also represented a threat to the relative political autonomy of Southern French lords. But the ideological component cannot be ignored. The Pro-Cathar knights were rallied on by Spiritual Franciscans, ascetic monks who refused to be silent about the corruption and moneylust of the Popes.
I'll always be fighting the Battle of Muret in my heart. When Northern France sealed absolute control over the South, the Papacy sealed absolute control over Catholicism, ignorance sealed absolute control over piety, and hatred sealed absolute control over charity.
It is likely the case that there were no actual Cathars in the Albigensian crusades and that they were mostly pre-Gregorian Reform Catholics with non-consistent beliefs. The label “Cathar” is a later, polemical construction rather than a self-conscious heretical movement. It would be a mistake to imagine the listed beliefs of Cathars (i.e. quasi-dualist and pseudo-Gnostic) as being broadly followed in any discrete/structured way.
That being said, the Gregorian reforms are very important the Albigensian Crusades definitely cemented that in medieval Europe and is thus not at all a minor event.
Source: Mark Gregory Pegg's A Most Holy War: The Albigensian Crusade and the Battle for Christendom
Inasmuch as the Cathar label was a handy label for justifying the persecuting non-standard Catholics, i suppose it must also be a handy label for commemorating non-standard Catholics who were persecuted and martyred.
And it is also the case that the lords and knights were involved in resisting The Crusade were by and large willing to conform to Gregorian “orthodoxy,” but were unwilling to enforce that “orthodoxy” on their subjects.
Seems like a pretty minor event.
Not at all! Wikipedia calls it "one of the most notable pitched battles of the Middle Ages." Historian Martín Alvira-Cabrer wrote that “Muret…became one of those defeats that modified the course of the Middle Ages.” If the Southern Coalition had won, there's a strong possibility that Languedoc would not be part of France today.
This battle sealed the outcome of the longest single crusade in history. It was the first and largest crusade initiated to punish heretics or Catholic enemies of the Pope. And if this crusade had failed, it might have fundamentally changed the balance of power between the Papacy and local Catholic communities.
I could see a Southern victory speeding up OR slowing down the onset of the Reformation. Or the victory could have altered the nature, geography and intensity level of the Reformation.
If this battle was so important, why don't we hear about it more often? I'm not entirely sure, but this is of a piece with the way Southern Europe has GENERALLY been neglected by English-language historians. Reading 20th century historians, you'd think that, apart from Constantinople and the papacy, little of interest happened in Southern Europe between 750-1350.
This situation started improving in the 1980s, but we still have a long way to go.
I suppose that I might now be responding earnestly to a joke. In which case, good job, you got me! But that's OK. I'll grab it any opportunity to interest the lurkers in this weird, fascinating crusade.
> If the Southern Coalition had won, there's a strong possibility that Languedoc would not be part of France today.
There is also a possibility that the consolidation of State authority under the kings of France was overdetermined. Also the crusaders were autonomous knights, not the royal Host. Their defeat would not have much weakened royal power or prestige (a defeat at Bouvines the following year might have been another story!) and at some point the kings would have wanted to assert their authority over the whole kingdom.
Maybe a victory of Peter of Aragon could have been a step toward another multi-regional rival "Empire" within France like that of the Plantagenêt kings of England/dukes of Aquitaine. But even they did not split a region from France in the end, despite winning battles much more devastative to the kingdom than Muret could have been.
"Had it [the Battle of Muret] failed, the campaign would have failed…Bouvines would probably never have been fought and the chances are that the French monarchy itself would have collapsed." Thus spake Hilaire Belloc. This is a tantalizing idea, although it must be said that Belloc was an autodidact and popular historian, rather than a fully qualified academic historian.
There was a later medieval heresy which survived the catholic crusades against it: the Hussites of Czechia. They did not end up very influential by themselves, compared to the Lutheran and other reformations of the 1500s. Although maybe there is a case to be made that the Hussites made the Reformation possible by showing that breaking off the Church was an option.
That's a data point on what a survival of the Cathars could have looked like.
Sure, everything you've written here sounds true. One difference is that in the case of the six Hussite Crusades, the Crusaders were mainly fighting directly against the Hussites. In the Albigansian Crusade, the Crusaders were mainly fighting against devout Catholics who raised arms against the Army of the Pope, so to speak. (At least the pro-Cathars were devout in MY book)
That's why there was a sense in which the crusade was a struggle over who QUALIFIED as a faithful Catholic. Whereas the Hussites and the Lutherans proved all too willing to shed their Catholic identity.
I didn't call the Cathars Catholic. The Cathars were barely involved in the fighting, because they were pacifistic. The fighting primarily took place between the Crusaders and Catholic lords protecting their Cathar neighbors.
Furthermore, the pro-Cathar lords and knights (including one King of Aragon) were fully devout Catholics who saw it as their Christian duty to protect their lands from a false, wicked Pope (as they saw it). Therefore, the Crusade can be seen as a civil war between Catholics with two different standards for what it meant to be devout. The pro-Cathars were also protecting the Jews in their community. They also knew that the Crusaders were implementing a system that would strip all lands away from female owners.
Of course, that's assuming that the Cathars even existed. One strain of historiography maintains that the so-called Cathars were a mix of Catholics following pre-Gregorian local traditions. Or the 'heretics' may have been a mix of Cathars and pre-Gregorian Catholics. Our knowledge of Catharism come solely from highly unreliable accounts written retrospect by fanatical Gregorian priests. (In this comment, Gregorian doesn't mean monks of the Gregorian Order, it means adherents of the Gregorian reforms imposed by the Papacy).
Whatever the exact identity in nature of the 'heretics,' the dynamic of the fighting remains the same. The anti-Crusader fighters were dominated by orthodox Catholic lords protecting those deemed by the Papacy to be heretics. By implication, they were fighting for a kind of religious tolerance, although I have no idea if they would have framed it in those terms.
I'll steelman that rhetorical flourish by proposing that Simon Montfort and his accomplices were ignorant about the Cathars, ignorant about the extent of the Pope's moral authority, and ignorant about the the ideological content of the New Testament.
The last penny has been minted!
I can't help but read this in the voice of the narrator from Super Metroid.
Next do the dollar bill, and replace it with a dollar coin that doesn't just look like a slightly larger quarter.
For this alone all the YouTube nerds should declare Trump the greatest president frankly. Killing the penny was a nerdy cause for YouTubers for decades. Obama said we couldn't do it. Biden did nothing. Second term, one post on Truth Social, one executive order or something and boom penny gone.
Unbelievable. He actually did it.
https://www.google.com/url?sa=t&source=web&rct=j&opi=89978449&url=https://www.youtube.com/watch%3Fv%3Dy5UT04p5f7U&ved=2ahUKEwjtyOeal_CQAxW3FTQIHR7kGqQQwqsBegQIExAB&usg=AOvVaw0UJ-cOImf4TXZlPCk23-cc
It’s absolutely a good thing, but it’s been pointed out that it’s a smaller budget item even than the things DOGE was cutting. It’s a lot of virtue signaling about cost savings while doing things more expensively. And doing it without a plan for the transition.
Still, I’m glad someone finally ripped the bandaid off if no one was able to phase it out in the proper way.
I predict the interwebs will instead fill with citations that Trump is the only President to ever stop making cents.
I predict they'll lean into the Lincoln angle, and say this is proof that he's a neo-Confederate.
The penny was the only coin of color.
That link could be shortened. https://www.youtube.com/watch?v=y5UT04p5f7U
And icymi, there's an update to that video: https://www.youtube.com/watch?v=n1KgxqEQn0A
And another, now calling for the elimination of the nickel (and the dime), https://www.youtube.com/watch?v=58SrtQNt4YE, for mostly the same reasons it made sense for the penny.
What puzzled me was the NYT had a photo of a shiny penny when they ran the story; but it was minted in 2006…
That's a one-in-a-lifetime loophole for anyone who promised to pay their debts "till the last penny".
It costs more to make nickels than it does to make pennies. So why not scrap the nickel and say all pennies are now worth five cents? The government saves money on new coins, and every one with a penny jar suddenly has something worth spending.
This is a great idea!
Or they could adopt Rufus T. Firefly's proposal of raising the value of a nickel to eight cents.
Fantastic "how ominous it sounds"/"how bad it actually is" ratio!
Ikr! Relevant xkcd: https://xkcd.com/1242/.
No democracy can survive once people discover they can vote themselves largesse from the public till. They will inevitably find the politician who will give them the most money, until the entire fiat currency fails, leading to a dictatorship.
... are you still sure the last penny isn't ominous?
Worldwide depression at our doorstep.
Proudly radical commenters will always be with us, but feature or bug, this forum is a bastion of political moderates. Who are all too eager to point out that "America is already great," at least in terms of long-term wealth creation.
But the fruits of that wealth have disproportionately gone to college graduates. Non-college-graduates are justified in bemoaning their massive loss in socioeconomic status since 1970.
Tthere's still a taboo against talking about this, enforced by college graduates in leadership on right and left. You can critique COLLEGES, but you're still going up to be hammered for any perceived slight against college GRADUATES, as a whole. Leftists will call you anti-intellectual (as if you have to have a degree to be an intellectual), and rightists will dub you a lazy, envious loser blaming society for your personal failures.
So foreigners and immigrants become the scapegoats.
There's no effective response to toxic populism that doesn't recognize non-college-graduates as a group with unique needs and legitimate grievances.
This is an American-centric comment, but this must all apply to various Western nations to one degree or another. In
Huh, I feel like this has been a point hammered by Richard Reeves (The Dream Hoarders) and others on the center left for well over a decade. But unfortunately, the upper middle class control a majority of political donations and a near majority of votes.
There was also a warning from Charles Murray that it was happening in _The Bell Curve_, and documentation of how things were going in _Coming Apart_.
I have no doubt we are seeing progress. I would just like to see more progress faster, I guess.
Speaking of the upper middle class, it would be a mistake to ignore the way they endanger the social stability of the nation. Yes, billionaires have outsized influence on society, but as Scott has pointed out, they have heretofore punched far below their weight when it comes to putting money into politics. Furthermore, isn't it true that billionaire super PACs affect elections in an unpredictable way that's hard to track? It's hard to summarize the SERIOUSNESS of the problem.
On the other hand, the upper middle class is very numerous in modern America, and relatively politically active. We can more easily tease out the way their votes and donations warp the fabric of our democracy. So maybe we should pay more attention to this clear and present danger that is nakedly menacing our society.
Calling us a "danger" and a "menace" is a bit much here!
Don't worry, I'd never call for guillotines. I find violent rhetoric rather distasteful. I can't speak for my fellow poors, though! You might want to negotiate with them to keep things hunky dory.
Post-COVID, the bottom 25% of incomes saw their income rise faster than other quartiles, and yet Biden still got hammered for "not caring about the working class." While wealth inequality is a big problem, it doesn't seem to directly correlate with the political shifts we see.
Also, the *actual* wealth inequality problem, IMO, is billionaires vs everyone else, not college-educated vs not college-educated.
Wealth inequality is not the same as incomes. We all know that rents rose pretty dramatically "post-COVID" -- the bottom 25% got squeezed hard, due to the increased risk premium imposed on landlords. Can you really show that they increased in wealth at all, or did they in fact lose wealth? (I'm pretty sure the latter, with unpaid credit card bills skyrocketting).
>We all know that rents rose pretty dramatically "post-COVID"
Rents are up, but incomes are keeping pace with that. The census says rent as a share of income has been pretty much flat - from 29.3% in 2019 to 31% in 2023: https://www.census.gov/library/stories/2024/09/acs-rent-burden.html
Also, I would expect stuff like this to be captured in real wages - if housing prices are up enough that your overall wealth has gone down, then inflation-adjusted wages must go down as well, right?
Also also, OP specifically blamed wealth inequality, not inflation. "the fruits of that wealth have disproportionately gone to college graduates." You're making an entirely different argument.
>Non-college-graduates are justified in bemoaning their massive loss in socioeconomic status since 1970.
Are they? Everyone is richer now than they were in 1970. The only "loss" I can see is that those with college degrees tend to be more richer now than those without, but they're all richer. If someone gets a 55% raise, are they justified in moaning that someone else got a 78% raise?
https://www.pewresearch.org/race-and-ethnicity/2024/05/31/the-state-of-the-american-middle-class/
I think former skilled factory workers are making less if they’ve had to take a job in, say, retail or work as a barista at Starbucks.
NAFTA and the decrease in union jobs have depressed the achievable wages for a lot of workers.
The whole Rust Belt thing.
My Hometown:
"These jobs are going, boys, and they ain't coming back"
https://genius.com/Bruce-springsteen-my-hometown-lyrics
Your analogy about raises is not fully satisfying, because it does not appreciate the "zero summish" nature of the phenomenon I described.
This post-1970 phenomenon was not exactly a zero sum game. But it's in the same ballpark. Decisions were made to use the power of government to reorient our state to benefit equity investors, retailers, homeowners and the knowledge economy. (Questions of intent are messy, this wasn't necessarily an Evil Conspiracy)
These decisions empowered certain citizens at the expense of others. Either directly or through a kind of "opportunity cost." When the government goes out of its way to benefit group A, they are implicitly WITHHOLDING help from group B. Uncle Sam doesn't have infinite resources, so picking winners and losers is not inherently wrong. But the gradual scouring of non-graduates WAS wrong policy, whatever motivations inspired it.
You'll notice that in my first comment I mentioned how non-graduates had lost "status." I didn't specifically mention income inequality, although that is related issue. Key point: Social status really IS a zero sum game, since your status is determined by where you "rank" in the pecking order, relative to everyone else.
At least part of what's going on is that "college graduates" and "non-college graduates" are different populations now than they were 50 years ago. In the early 1970s, less than 10% of adults were college graduates. Now, about 35% are. So when we say that inflation-adjusted wages have stagnated for non-college graduates since the 70s, we're comparing wages for the lower 90% of the educational attainment distribution then to wages for the lower 65% of the distribution today.
The population has aged significantly and until you account for that, I don't think your analysis here is going to be helpful.
What effect do you see here from the aging of the population?
Surprised it took a while for someone to mention this, this seems like the obvious confounding factor
And current high school graduates go to college at the rate of 60+%.
To this day, Trump is frequently mocked for having said "I love the poorly educated". The implications are pretty awful.
This ties into the more general issue of measuring human value by intelligence and/or one's ability to produce stuff.
"Leftists will call you anti-intellectual"
If they do, they're TERRIBLE at being leftists. The whole notion that access to college degrees--which are far more easily available to those born with more wealth and privilege--should control the material circumstances of one's life is pretty explicitly antithetical to any serious notion of leftism. Liberals might hold to such, since liberalism tends to include an implicit belief in the fairness and meritocratic nature of the so-called "free-market." But liberals aren't leftists.
I will caveat, however, that there are plenty of *tasks* which require domain-specific knowledge and expertise to perform well. In many cases colleges are (at present) by far the most direct and practical way to obtain that knowledge and expertise. Realistic leftists (which plenty aren't) will acknowledge this but argue some combination of A. access to that knowledge and expertise should be made more broadly available and B. one's socioeconomic status should not depend so heavily on obtaining such expertise.
The fact that there's been a huge wave of people recently loudly insisting that expertise simply isn't necessary, even in areas where it plainly, obvious is, tends to muddy the water somewhat. These people ARE being anti-intellectual, and their ubiquity makes it easier to conflate "no really, you need to be trained to do that" with "only people with college degrees deserve respect or a comfortable life."
"access to college degrees--which are far more easily available to those born with more wealth and privilege"
This sounds like an article of faith, rather than a seriously supported assertion.
62.8% of American high school graduates go to college. That doesn't sound like "far more easily available."
If you're going to say "well, but not everyone graduates high school!" I'm going to reply, "sure, but high school's free."
"This sounds like an article of faith, rather than a seriously supported assertion."
Um, what? An article of faith? Excuse me, what?
I didn't say *unavailable.* I said *more easily* available. I honestly have no idea how you could imagine this NOT to be true. Do you think that your odds of getting a degree anti-correlate with your parent's income?
Note that I said "more wealth and privilege" not "super, duper lots of wealth and privilege." I'm well aware that getting a college degree is the default for a large swathe of the American middle class--but the middle class aren't the only people in existence. And even among people considered middle class, wealth buys lots ways to smooth the road: you're more likely to get help with any health or academic problems, less likely to need to work at the same time as you go to school, more likely to be able to afford conveniences and amenities, less likely to need to drop out in the event of a family emergency.
"If you're going to say "well, but not everyone graduates high school!" I'm going to reply, "sure, but high school's free.""
And at this point I'm going to ask exactly how many minutes you've been living in the real world, and if we could maybe table this discussion until you've been here for a year or two. If you think the cost of high school tuition is the *only* variable that makes wealth more likely to lead to high school graduation, I can only imagine that not only have you never, ever experienced financial hardship, but that you've never met, spoken to, read anything written by or possibly even passed on the street somebody who has.
I grew up in a fairly comfortable, middle-class household, but it was still really, *really* not that hard to notice the ways in which wealth directly and indirectly translated to academic success. Many were small ways, but there were a lot of them. Few of them are extremely likely to be definitive by themselves, but in case you didn't notice, we were talking about population-level trends, and the accumulation of small effects are what population-level trends are made of.
So, let's break apart "wealth", which is your word, not mine. You're not using income, you're using wealth. The primary source of wealth for middle class people is housing, right? [This is the difference between positive wealth and negative wealth for most middle class people...] So, you're saying that apartment dwellers (in, say, NYC, or Pittsburgh, or Atlanta) are significantly less likely to graduate college. I don't think "parental wealth" is the best way to distinguish "whether or not someone finishes college."
I'm here to tell you that at age 25, you no longer need to factor in your parents wealth/income in applying for grants. That means you're basically "defacto poor" (unless you're, say, a multimillioniare) -- because otherwise you're established and making "good money" and don't actually need college. This makes you eligible for a free ride to most normal schools.
Do people sometimes drop out of college to support their family? Sure. It's pretty rare, though, in the grand scheme of things (this happened a lot in the 1930's, when government benefits weren't very thick on the ground). And even for people who did drop out of college, if it was a temporary thing, they can go back to college.
High levels of parental wealth can let you coast through basically everything -- parents have this absurd tendency to "help" their kids, even when they're failures. See Mitt Romney, see Hunter Biden.
I think, in high school, you have this wide swath of "people being given pretty much the same start" and then you have this "swathe" of "kids who will be tutored into passing and attending Yale" (this is extremely wealthy people).
Cheery reminder that wealthy people can be under the federal poverty level in terms of income (see Senator Tester, before he went to Congress). And that those people are presumably significantly less likely to finish high school, despite having parents that are millionaires.
I'm willing to bet you could find a better proxy than "parental wealth" or "parental income" to show which people are most likely to go to college.
I do know kids that were sent to stay with relatives because their parents couldn't support them. They finished college. I've also tutored at a community college -- lots and lots and lots of people who are "not being supported by their parents" (and presumably eligible for "no money down" college).
You can assume we met in the middle, on this point, and that yes, you're not talking to a deranged lunatic, merely someone who thought you were overstating your point.
There are some pretty significant differences in college attendence and completion between people with wealthier/poorer parents: https://share.google/EtTGS6LjqnuMSSTyN
And if you look specifically at the most prestigious colleges it's bananas: https://share.google/5LSIFVNZD2qoHw6iI
I'm not sure how this is relevant? I don't want to be uncharitable but is this just a way of saying "if you don't want to go to college and complete a degree, that's on you. Sounds like a skill issue." I hope not. That kind of framing is a crude and reductive sidestepping of the issue.
Hm, maybe something is unclear here. Agra[...] said college degrees are easier to get if you're born into a wealthier family. November Dawn disagreed because most high school grads go to college. I'm pointing out that there's still a pretty obvious gap based on parental income, and that holds even more true for top colleges, i.e. I agree with Agra.
Got it. I agree too.
If you look beyond the US a lot of the original comment still applies despite university degrees being much more widely available to those without wealth and privilege. I think its a useful consideration even if you are only focused on the US.
Making the sorting mechanism more meritocratic doesn't solve most of the consequences of the sorting.
I'm not sure how this is relevant. I mean, I have an idea of what you are trying to say but Idon't want to be uncharitable by assuming callousness on your part. Can you rephrase your point?
Callous? I'm pointing out that segregating society has downsides regardless of how it's segregated. I'd rather suffer whatever uncharitable looks like than be damned by implication.
As I said, belief in the meritocratic nature of the market is a liberal trait. Substantially privileging *any* select group of people over other groups is fundamentally anti-egalitarian, even if you swear up and down that your selection mechanism is "fair[1]." To the extend that the left-right distinction are used consistently at all, left is the egalitarian direction, so in any meaningful leftist practice, anti-egalitarian practices are to be questioned and (as far as possible) removed or reformed.
Just to be thorough, I'll add the same caveat as before: expertise is still necessary for lots of things, and pretending that it isn't never ends well. But choosing how you regard and reward expertise is a voluntary choice, at least inasmuch as anything that happens at the level of societies is voluntary.
Finally, I'll add that a good (fictional) template to help think about such things is the novel "The Dispossessed" by Ursala LeGuin, as the main character is a man of significant specialization and expertise living in an actively (and in some ways militantly) egalitarian society.
[1] A word that becomes largely meaningless in context like this.
So - Making the sorting mechanism more meritocratic doesn't solve most of the consequences of the sorting.
Japan and America have nearly the same statistic for "high school students going to college" -- what countries are you thinking of? And can you cite statistics that are greater than 60% conversion rate?
It appears I have mistakenly replied to your comment rather than the level above, agrajagagain's.
We're likely in meta agreement (because the symptoms are there when you remove this) however to your point my reply would be - any country where the cost of the degree is significantly lower.
If a degree is very expensive you might expect the 37.2% who don't have a degree to correlate strongly with the poorest 37.2%. In a country where it's closer to free (or the debt on what is charged is more heavily subsidised), you'd expect less correlation - away from whatever correlation of wealth and IQ you're willing to believe in.
Something that is exclusive to the richest 60% obviously isn't particularly exclusive but could still have far reaching consequences.
"But the fruits of that wealth have disproportionately gone to college graduates. Non-college-graduates are justified in bemoaning their massive loss in socioeconomic status since 1970."
At least the number of college graduates has increased massively, hasn't it?
I generally agree with you, and wanted to comment on:
>There's no effective response to toxic populism that doesn't recognize non-college-graduates as a group with unique needs and legitimate grievances.
If by "effective response to toxic populism" you mean the Democrats (or maybe even some more reasonable coalition of Republicans) winning the 2026/2028 elections, then I think the best strategy would focus on winning winnable votes. "Recogniz[ing] non-college-graduates as a group with unique needs and legitimate grievances." is a part of the story, but another huge part is how to handle polarization and social media, choosing messaging that resonates, avoiding getting overly invested in unpopular issues like trans participation in high school sports, running more conservative (possibly pro-life for example) candidates in more conservative states. I don't think non-college educated people are voting for MAGA because of a thoughtful consideration of policy... They vote for someone who talks like them, who they want to drink a beer with, they vote for the leader of the tribe the algorithms have polarized them in to, also they vote based on their own personal experience of the econonmy in the last 3 months, ie: the price of gas and milk (without regard to any historical factors).
As a non-college educated conservative person of sorts, I think "oh those guys are morons, let's just run someone who can say "ain't" and knows the name of at least three guns" is probably a great way to win over conservatives; people love pandering, we are indeed dumb enough to fall for that all day and this definitely won't backfire.
I guess I didn't write my comment in a very thoughtful way. I generally think that most of the electorate, on both sides mainly vote with the tribe.
But to play the devils advocate, Isn't Trump the ultimate example of successful pandering?
He has previously been: anti-gun, pro universal-healthcare, pro-choice, pro Clinton, he's been pro and anti social security, anti crypto.
And moreover he's a coastal elite billionaire who managed to convince 40% of the country that he's their bro, and that he's going to "drain the swamp".
He also, as another thread here mentions, seems to hand out pardons to anyone willing to invest in his families crypto venture, or other fundraisers.
Moreover, most of his speeches seem to be 90% pandering, "we're going to build a wall and Mexico is going to pay for it" or talking about how cheap various household goods will be.
Maybe he's not pandering but actually evolving or just able to change into what he needs to be somehow as needed in a moment.
But what i would advocate for is not running candidates who can play-pretend, but rather, run more conservative candidates in states that are more conservative, and more progressive candidates in states that are more progressive. Don't fund primary challenges with democratic socialist candidates from the left, in states where the populace wants a pro Oil candidate who is in the middle on abortion. Try to build coalitions that include people that disagree on gender/climate/etc. but possibly agree elsewhere, like Healthcare and such.
More Joe Manchins basically...
Ultimately, democracy selects for people who are good at getting elected. :(
Perhaps things could be different with a smarter population. Not sure where to find it. (Switzerland?)
Denmark -- it's the highest trust society we've got left.
The Danes are happy because crime and disorder are spread out fairly evenly throughout the country. I would imagine this is partly by design and partly by circumstance. It can be politically expedient to allow crime to fester in certain areas and not others, but that causes a bunch of social pathologies that feed on each other.
Many POTUSes have abused the pardon power, and it was high on my list of things needing constitutional deletion well before Trump arrived. But as in some other forms of corruption he is taking it to a whole new level.
Just since returning to office 10 months ago Trump has issued more than 1,500 pardons. More than 1,000 of those are January-6th rioters. Others include:
-- the husband of a GOP congresscritter, who'd been convicted of health-care fraud including defrauding Medicare and Medicaid.
-- former congresscritter George Santos, who became the 8th criminally-convicted Republican member of Congress pardoned by Trump.
-- former Illinois governor Rod Blagojevich, which as an Illinois resident fills me with a perhaps-unreasonable degree of rage. [Could be that a Long Island voter would feel this way about the Santos pardon.]
-- a guy convicted of defrauding investors, who was pardoned after donating almost $1 million to the Trump campaign.
-- a guy convicted of tax crimes, who was pardoned one month after his mother attended a $1 million-per-seat Trump fundraiser.
-- a Chinese billionaire who was pardoned after the crypto exchange he founded, Binance, struck a deal with the Trump family company and also hired as a lobbyist a close friend of Donald Trump Jr.
-- two DC police officers who were convicted of murder and subsequent coverup.
-- two guys convicted of defrauding a Native American tribe and pension fund investors out of tens of millions of dollars, who were pardoned after they testified in the House Republican investigation of Hunter Biden.
-- a Republican state legislative leader and aide who were convicted of bribery, and a Republican state senator who had pled guilty to siphoning off campaign donations to his personal use.
-- 77 pre-emptive pardons for individuals investigated and/or charged, but not yet tried, in the 2020 effort to substitute made-up lists of Trump electoral-college electors in several states won by Biden.
All of those, and more, just since January 21st 2025....it's literally a dial-for-pardons operation up there at this point. No fucks given.
Ok, that does sound very wrong. What's the answer of the pro-Trump crowd?
Clinton pardoned Al Capone! Look it up.
Friend of mine worked for the doj. Prosecuted a guy in Miami for tax fraud. Millions and millions of dollars. Money that could and should've gone to public benefit. Trump pardoned the guy because they were acquaintances. My friend no longer works for the doj.
Has there been any use of the pardon historically you think is good?
Historically, I think it's very good for Presidents and Governors to pardon people who have been on the wrong end of a gross miscarriage of justice, including but not limited to being wrongly convicted of crimes they didn't commit - and no, that's not something you can just leave to the courts. I think it's less good, but still good, for them to pardon people who have clearly rehabilitated themselves to an exceptional degree. And it is frequently a good idea to offer a general amnesty at the *end* of some major social upheaval.
And I think that this represents a large majority of the pardons in US history.
I support pardons in the cases you've mentioned, but I would be surprised to learn that this was indeed the majority of pardons granted by POTUS.
Unfortunately, sometimes the pardoner grossly misunderstands the case at hand: https://en.wikipedia.org/wiki/Miller_Arnold_case
I'm OK with an implicit requirement that you not convict people of crimes if you can't reasonably explain to the chief executive why that's a reasonable thing to do. If there's a disagreement at that level, then there's reasonable doubt (possibly because of incompetence by prosecutors or judges, but so be it).
The bit where the executive then arbitrarily imprisons the judges is a bit much.
I think Lincoln was correct to pardon rebels in the American Civil War. I also agree with Carter’s decision to pardon Vietnam era draft dodgers, though that’s a closer call. Many pardons (I’m guessing more that half) are of people who have served out their prison sentences and paid any fines. While I haven’t looked at any of these in detail, I would probably agree with most of them because I don’t think the bar for pardoning an individual who has paid their debt to society and not committed any subsequent crimes should be particularly high.
Trump has been cancelling restitution to victims of securities fraud. For example, Trevor Milton was pardoned before he had to pay a cent in restitution. Trump, when commuting Carlos Watson’s sentence to time served, specifically removed Watson’s obligation to pay restitution. These people are, as far as I can tell, undeniably guilty, and Trump is letting them keep their ill-gotten gains a the expense of their victims. I can’t imagine that any American President other than Donald “the cruelty is the point” Trump has used the pardon in this fashion.
Cruelty has nothing to do with it. I think it's simply that having been the victim of absurdly high restitution judgments himself makes Trump is more sympathetic to other people in that position than other presidents have been in the past.
Trump hasn't been subject to any absurdly high criminal-restitution judgments, since the one that plausibly fit that description got tossed.
Probably was a good idea to pardon Nixon.
It partially destroyed the people's faith in the government. To an extent, it tore the nation apart and prevented future compromise or reconciliation. I think it was a disastrous decision that sowed the seeds of discord and disunion.
I think Nixon's actions did that. I don't know that letting him go to jail would have led to national reconciliation. But I think a hard call.
I'm pretty torn on this; I understand the logic, but "Presidents can get away with crimes" really looks like a bad precedent now.
"Presidents can get away with crimes iff they have the decency to resign". would OTOH be a pretty good precedent. Much better than the Pinochet and Gaddafi precedents IMO.
Unfortunately moot given that SCOTUS has made it impractical to punish Presidents for crimes Congress didn't see fit to impeach and convict them for. Which is probably the correct legal decision but means we really need a Congress that will step up and do its duty in that regard.
Lawfare is a really good way to ruin a republic, someday somebody will cross the rubicon to avoid prosecution.
Putting people above the law is an even easier way to destroy a republic.
> "Presidents can get away with crimes iff they have the decency to resign" would OTOH be a pretty good precedent.
Agree. And the nature of the crime matters. Nixon's crime was pretty small potatoes.
I recall accounts that Nixon wasn't even unique in committing them. Democrats were reportedly up to the same tactics; only Nixon was caught, and Democrats wisely shut down any of their own projects in response.*
If those accounts turn out to be true, then another reason for Congress to consider the matter closed after Nixon's resignation is to put a stop to the digging.
*To the extent that they could; such tricks were reportedly much more decentralized. What arguably made Nixon's version so intolerable was how coordinated it had become.
I think "you can't even convict them later, if they were doing President stuff" is a pretty atrocious standard. Makes it much LESS likely anybody will resign.
If you can't get congress to impeach them while they're still in office, I'm not sure what arresting them after they leave office really gains you other than maybe feeling good when you see it on CNN (assuming it's Other Team's ex-president and not your own).
Noting that POTUS has committed serious crimes, not impeaching him and removing him from office, and saying "...but when you *do* leave office, *then* we're going to get you but good", introduces all sorts of perverse incentives that I don't think we want to deal with.
We dodged a bullet with Trump in 2021 because he only had a few days to respond to those incentives before leaving office anyway. And a significant part of what we're suffering now is Trump's entirely predictable and understandable retaliation for the 2021-2024 Lawfare.
Indeed.
Also unwritten norms are very very hard to force back into bottles, to mix a metaphor....now that Trump has demonstrated that a POTUS can sell pardons, pass them out to friendly acquaintances for no policy reasons, etc., with no limits in quantity or frequency, it's impossible to imagine that no future POTUSes will do the same. So now if forced to choose between "keep the pardon power as-is" or "delete it from the Constitution altogether" I'd certainly choose the latter.
I'd still _rather_ have the Constitution place some guardrails on the issuance of pardons, specifics to be hammered out as part of trying to pass an amendment. But barring that, for me the only choice left would be getting rid of it altogether.
Random thoughts on AI tools for writing, currently:
1. No AI can write fiction very well at a paragraph level. M-dashes aside, they all seem to think something like "this verbose, flowery line is what maximal meme-level communications hold up as good writing, so I'll do that as often as possible" and maximize for it. So you get something like this from GPT and Claude for a prompt something like "A man looks around on his first day at his job in a firewatch tower (he arrived there at night, this is his first good look at it):
*GPT*
"Morning seeped into the tower through the slats like pale smoke, touching the metal stove, the single chair, the bunk that creaked under him as he sat up. The world below was hidden under a slow-moving quilt of fog, but above it the pines stood like black-green needles, still and waiting. He stretched, feeling the thin boards flex underfoot, and turned in a slow circle—four windows, four directions, his new horizon boxed neatly in glass and wood."
*Claude*
"The windows wrapped around all four sides, offering a dizzying three-hundred-sixty-degree view of the Bitterroot wilderness—an endless corrugation of pine-dark ridges fading to blue in the distance, the canopy below him rippling like a vast green ocean. He stood and walked to the south-facing window, pressing his palm against the glass, still cool from the night, and felt the tower sway almost imperceptibly in the wind, a gentle reminder that he was suspended between earth and sky, alone in a way he'd never been before."
To a non-writer, this seems fine too, perhaps. If you do this for a living, it's hard to miss how dense/overheavy this kind of writing is, and how exhausting it would be to read across whole novel. It's pale-smoke-slow-moving-quilt-black-green-thin-boards-slow-circle-new-horizon-boxed-neatly-endless-corrugation-pine-dark-ridges-fading-to-blue-vast-green-ocean-imperceptibly-swaying-gentle-reminders that AI is good at writing in a way that someone who goes "GEORGIA OKEEF PAINTED FLOWER VAGINAS I'LL PAINT 300% MORE FLOWER VAGINAS AND BECOME THREEORGIA OTHREEF" is good at painting, and no better.
2. At a tele-and-supra-paragraph level, no model is good enough to be even useful. If you ask a model to write the next paragraph of the story, they will try to cram as much possible plot into the paragraph as they can in a frantic, busy way. If you explicitly tell it not to do this, they will give you pure-shitty-descriptive filler as in #1.
None of them can do any better than this. There is no sense that any of the AI models understand plot-pacing; they can't or won't do it.
3. Both GPT and claude are ok-ish at revision-and-suggestion-at-the-paragraph-level stuff. My personal writing problems trend towards being too filler-wordy (lots of "very" and "really"), and they are good at seeing that and pointing it out when directed to. You can give either a paragraph for review, and at the they will find most of what's wrong with it at the grammar and obviously-bad-writing levels.
They are both of very limited use for "here's how to do better" suggestions, though, but in different ways. Given enough leash, GPT will ALWAYS try to revert your writing to #1 writing, as above. It considers the kind of writing it does to be "good writing" and will steer your writing in that direction unless very clearly instructed not to in a way that limits it from fucking with prose in the first place. If you get carried away and let it do this too much, it will slowly transform even good stuff into piles of mushy cliche with bad plot momentum.
Claude is much, much better at understanding something like "respect my voice" as a command. Sonnet 4.5's personal editing style leans towards "let's make this more concise, let's try to find a way to say the exact same thing in fewer words, also your tenses don't match" advice. Oddly, while claude itself can't write worth a shit, it's kind of good at "eliminate wordiness" edits to the point where I've actually used it this way to a limited degree.
4. When dealing with larger pieces of text, chatGPT is actually and very literally useless. If you drop, say, a 30k document in front of chatGPT, it will read perhaps the first 10-13k words then stop. No amount of cajoling will get it to read the rest. It will lie about doing this and try to trick you, but GPT's working memory just isn't large enough to consider a long piece of fiction as a whole all at once.
Since novels are long (100k words, plus) and since assessing one section doesn't work without knowledge of others, you can't ask GPT a question like "I made a change in the 30,000-35,000 words section of this book, does it work in light of what we covered in the 1-5000 block?". It's running instructions that intentionally cripple it for this purpose.
5. Claude handles large pieces of text just fine. You can give claude a 40k word document and say "Hey, find the changes I made and tell me if you like them" and it will read the whole thing, find the changes, find every part of the book they affect, and talk in a comprehensible way about that.
Summary stuff: GPT is a bad writer, a not-great editor, doesn't understand or respect voice, and useless for any chunk of text longer than 10k words or so. Claude is a bad writer, an almost acceptable editor, can somewhat edit in a way that respects voice, and can read long documents and hold them in its working memory. Something about how claude uploads long files is being adjusted so they might be crippling it in a GPT-way too, but unless this actually happens there's currently no reason to use GPT for anything related to prose/writing. It's just not good at it, in ways that often seem to be top-down corporate decisions it's hard-coded not to let you around.
This is why my AI-assisted fiction always alternates between AI- and human-written paragraphs.
Thank you for the detailed report! It agrees with what I would have guessed, but I also wouldn’t have been surprised if it turned out differently. Too bad so many people are misinterpreting you as saying “AI is bad” rather than just giving us a factual rundown about the current (in)effectiveness of certain popular models right now.
I want to figure out how to use these tools to be helpful in my own (academic) writing, but I haven’t yet gotten much past having them write code to draw useful diagrams.
Is that the default style for ChatGPT? You can ask AI to write or rewrite in a billion different styles. It seems like quite a stretch to condemn ALL those styles, who has he even tried them all?
There's a couple different answers to this question. The first and easiest to understand is that what you are addressing in your post here is style/voice. Even if GPT could do that well (more on this in a little) that's one aspect of writing, and it's still immensely bad at basically all the others. It can't plot, it's not good at making characters or developing them, blah blah.
The harder to illustrate thing is that GPT can't actually mimic voice or writing styles very well. So when I give it this prompt:
"Write me a three paragraph section of a story in which a man approaches a bicycle tied to a tree in a park. He unties it, looks around, considers not taking it out of guilt, then subdues his guilt and rides off on the bike."
It gives me a first paragraph like this:
"The bicycle was right where he’d left it, half-hidden behind a palo verde whose thin branches cast a web of shadows over the frame. Early morning light made the chrome handlebars glint, and the whole park still felt sleepy, holding its breath. He walked up slowly, as if the bike might accuse him before he even touched it. The knot securing it to the trunk was loose—deliberately loose, the kind that said come back soon rather than don’t you dare. He pulled the strap free.
Or for the same thing, but in the style of Stephen King:
The bike waited under the cottonwood like some sad, forgotten animal—one wheel half-sunk into the leaf rot, its chrome handlebars catching what little gray light filtered through the afternoon clouds. Martin felt the nerves start up the moment he saw it. That jittery prickle under the skin, like the feeling he used to get walking past the principal’s office as a kid. He slowed, hands in his pockets, pretending not to stare too directly at the thing shackled to the tree with a frayed nylon rope. God, who even tied a bike with rope anymore? It looked like an invitation. It looked like a trap. Mostly, it looked like temptation.
Or Ayn Rand:
He saw the bicycle before he saw the tree—a slender frame of steel and purpose, lashed to a rough-barked trunk as though the world had tried to domesticate it. The morning sun struck the handlebars and sent back a defiant flash, a signal meant for anyone who still had the capacity to recognize an object shaped by intention. He approached it with the deliberate calm of a man who understood that possession was not a matter of permission but of will. The chain around the tree clinked, a small metallic protest, as he bent to free the machine.
Or Douglas Adams:
The bicycle was tethered to the tree in the way one might secure a mischievous goat—firmly, apologetically, and with the faint suspicion that it might complain about the whole arrangement later. Harold approached it with the cautious optimism of a man who had recently discovered both a sudden need for transportation and a moral compass of questionable calibration. The park around him hummed with that gentle, civic sort of silence that suggests nobody is watching, while simultaneously insisting that absolutely everyone is.
And you hopefully start to see the problem. Same structure, same cadence. It knows just enough about different author styles to throw in little bits and pieces it vaguely understands are in there, but you are getting maybe a 10% modification to tone over a very, very familiar writing voice that just isn't very good.
If I'm sounding like a relentless anti, I'm not; I tried very hard for a few months to find a way to make this work as even a make-it-go-a-tiny-bit-faster tool for my own writing. I've tried hundreds of prompts, and I'm still even using it for certain very discrete reductive editing tasks. It's just not all that great at writing.
You're just not asking it to mimic authors with highly original styles. The difference between King and Adams is fairly subtle.
If you asked it to mimic say, James Joyce or Hunter Thompson or Jack Vance or William Blake... In fact, you can ask it to mimic the styles of individual books. If you asked it to imitate the style of Gene Wolfe's novel "Shadow of the Torturer," you'd get some great output.
That said, prompting ChatGPT the right way will produce text that flows quite nicely but is thoroughly uncommercial and unmarketable. That's going to be a problem for a lot of users.
Still, I maintain that ChatGPT CAN spit out lovely prose and poetry. As far as characterization and plot go, sure, you're on your own.
I mean, I've heard so many versions of "if you just X, then it would be good", I've tried a bunch of them. They are very hopeful. But I can tell you what happens when you try someone even more distinct than the practically identical Ayn Rand and (checks notes) Douglas Adams. Because I've tried it. A bunch.
So say you plug in James Joyce, as requested:
"He came upon the bicycle where it leaned against the old elm, the chain looped through its frame like a thin iron regret. Morning light fell in weak stripes over the grass, and he felt the hush of the hour settle on his shoulders. He set his hands to the cold links and worked them loose, slow, as if the tree itself might rouse and ask him what business he had there."
And you get the same shitty style wearing another different hat. You aren't entirely wrong, you can engineer a bunch of different stuff in (give me only ultra-terse sentenced in the style of Wilde if he had fallen down a well!) and it will do the party trick and make it *kind of like that in a way it has heard WIlde is supposed to be".
But it will do this at the level of a relatively incompetent writer trying to do this. You are asking someone who is already pretty bad to do something *harder* and get outside its wheelhouse. It only does so much.
I think that's decent prose right there. It's not the BEST I've ever heard but geez, what are your standards if that's "shitty" to you? Are you demanding Pullitzer Prize level stuff?
I'm a professional writer. My standards are high because the standards for writing that actually makes adult, pay-the-mortgage money are incredibly high. There are some Rob-Liefield-like exceptions to that, anomalies of a sort, but not a ton.
GPT is mildly useful as a brainstorming tool/sounding board, but after that you actually need to get it to write something. It can't do this at all without veering off into completely unusable chaos unless you tell it exactly what to write at the paragraph-to-paragraph level, which means all it has left to offer from that point is prose.
And its prose is about as good as a smartish, B+ highschool student. It is pretty good at not making grammar errors, it has a vague idea of how to frame sentences without any individual one being quite bad enough for it to called out, and then it stops.
When you ask it to write in another style, it can do this at the level of a smartish, B+ highschool student. Again, it's not contributing anything else at that point except (perhaps) some help with brainstorming.
When I say *it can't write*, I really mean that. It just can't contribute meaningfully to the work of writing without everything but prose handed to it, and once the prose is handed to it, it fumbles that too.
You might say "well what about non-professional standards?", and I would start agreeing with you a lot - it's a fun novelty and it's impressive someone could invent a party trick that complex. But beyond that, it's time-sink garbage.
Corroborating evidence in scientific papers:
https://phys.org/news/2025-07-massive-ai-fingerprints-millions-scientific.html
This is interesting, but it's not particularly relevant to what I'm talking about. AI's can make text, yes, some people are using that to cheat on their homework, sure, even at a high level.
I'm talking about fiction, and the study is talking about non-fiction; I'm talking about quality and the study is agnostic on that front.
> 2. At a tele-and-supra-paragraph level, no model is good enough to be even useful. If you ask a model to write the next paragraph of the story, they will try to cram as much possible plot into the paragraph as they can in a frantic, busy way. If you explicitly tell it not to do this, they will give you pure-shitty-descriptive filler as in #1.
Have you tried splitting the prompt into multi smaller prompts, where each prompt only has enough context for a single thing? for example:
1. first you ask it to describe individual paragraphs (including pacing, tone, plotpoints), without actually writing it.
2. then for each individual paragraph, you make a smaller prompt, that only writes that single paragraph, and only has access to the description of that single paragraph from step 1
I've tried a lot overall in the "tell me what you want to do, then do that in little chunks I can review" way. I didn't like the results or find them significantly better, but YMMV.
The most I've tried to use AI professionally was in an open-canvas sort of way where I'd provide text, tell it what the next paragraph or so was supposed to be, then have it generate it. This is giving it 90% of the work pre-done and takes nearly as long or longer as writing the paragraph does.
Even leaving as few choices as possible to the AI, I still found it mostly generated stuff I had to edit so heavily I was losing a lot of time, and was such a poor "base construction" or whatever that the end result paragraphs were usually worse than if I had written them myself.
Note that my writing isn't exactly shakespeare, either. I'm a quantity writer, and though I'm no slouch it's not like this is competing against my take-my-time-and-get-it-right stuff.
It's still good for some very discrete editing tasks, it's just not fantastic otherwise, at least IMO.
Re: discussion further down about whether a model exists that writes well. I just asked GPT 5, Thinking version, for info on whether AI models exists that have been constructed with the aim of producing a good writer by training it on great writing. It said you cannot train an LLM on only good literature because there's nowhere near enough of it, but that you can fine tune one afterwards. Just gave me a long list of such models, including some that you can try out online, also articles about many of them, and samples of writing they produced. It is also offering detailed instructions on how to do the fine-tuning oneself. I'm not posting the info here, because the last time I can remember someone posting a GPT answer numerous commenters heaved rotten cabbages at them. Link to GPT's answer is at this google doc: https://drive.google.com/file/d/1athsLfeYGMBEQJGBmtf180winqR2YUcR/view?usp=share_link
Feel free to copy it.
By the way, I don't approve of ruling out AI answers. In a situation like this one can provide highly relevant info, much richer than what anyone in the discussion has to offer. Even if it turns out that some of the *many* sources GPT links are hallucinations, that hardly matters in a hunt like this. Just throw out the chaff -- plenty of wheat left.
I sent a request for access!
Sry, meant to mark link as accessible to anyone who clicked it. Changed it just now, so it should work. If it doesn't let me know and I'll give you access directly. Link is https://drive.google.com/file/d/1athsLfeYGMBEQJGBmtf180winqR2YUcR/view?usp=share_link
I just read the paper. I don't have time to figure out how to run the models it mentions tonight, but I'm wondering how good they are be - it's less "how do we make a good writer" and more "how do we make GPT2 write in jane austin's voice" which is a little bit different.
I"m hoping it's not, but I just suspect fine-tuning GPT2 gives you fine-tuned GPT2, which was shit at writing (as even 5 is).
OK Ben here's a pinata. I asked GPT some follow-up questions, including about models fine-tuned on GPT 3.5 or higher. Put the entire dialog, including original one, here: https://drive.google.com/file/d/176u2UajOTvVT001gZTqHPydG5YeT1WO4/view?usp=sharing
If you gather more info can you put it up somewhere too?
By the way, I'm diving into this not purely from the goodness of my heart, but because I'm also a writer interested in AI possibilities.
Hey! Not ignoring you on purpose, I hope to get to this tonight.
Yeah, it does look like all the models he linked are from fine-tuned GPT2. But just asked GPT what better models he can be trained on, and it sent me to this: https://www.datacamp.com/tutorial/fine-tuning-openais-gpt-4-step-by-step-guide?utm_source=chatgpt.com
which says that you can now fine-tune various GPT-3.5 versions. May be possible to do on GPT4, but you have to request access.
...well, I guess I've got to give that prompt a shot, don't I.
"The scant furniture in the firewatch tower offered no protections from the light of the morning sun. Cory covered his eyes and turned his back, staring out the western windows, though in all directions the view was the same. Looming mountains, covered in thick green trees, and below the trees, nothing to be seen. Even the road he had taken to get here was buried by the greenery, as if it had never been."
Or how about fancying it up a *little*:
"The scant furniture in the firewatch tower offered no protections from the light of the morning sun, which pierced his eyes the way a GEORGIA OKEEF PAINTED FLOWER VAGINA would have if it had been really pointy. Cory covered his eyes and turned his back, staring out the western windows, though in all directions the view was the same. Looming mountains, covered in thick green trees, and below the trees, nothing to be seen. Even the road he had taken to get here was buried by the greenery as though by 300 GORGON KIEF VAGINAS as if it had never been."
...well, I guess I have a plot now. Our hapless firewatch man is stuck in a remote wilderness, helplessly watching as the forest around him turns into ever-denser Georgia-O'Keeffe-painted flower vaginas.
I think at a certain point you have to suspect unreliable narration.
Especially once they see he's spent the months turning their tower into a giant H.R. Giger-style biomechanical penis. "It was the only way," he claims, but they do not believe his heroism.
Only MC Escher can save him now
LLMs are very bad at creative writing. This doesn't preclude other AIs from being good at creative writing.
I think you are reading something I wrote as "No AI can..." in the present tense as "No AI can ever be built that could..." in a future-tense sort of way. Currently, every AI is pretty bad at writing.
You mean "every AI I could find and interact with." This is a subset of current AIs -- it may or may not be the current full set.
So.... like, do you have an AI that's good at writing, or are you just a troublesome sort of person? Because I'm really interested in the former.
I've heard good things about Kimi K2. It's a few months old; the Thinking version from this week is supposed to be even better.
I played around with it a bit today, not extensively (unpaid version) but to the extent I set it on a few editing tasks and a few "what's the next paragraph" prompts it seemed to me to be better than chatGPT and worse than Claude.
Super small sample size thus far, though. It's been topping out on me at pretty low contexts (by my weird standards) so I haven't been able to try to get it to do serious work yet.
The first rule of being an AI that can pass the Turing test is to pretend to be human.
The second rule is that any publically available ... formal/paid writing (with copy editors and such), cannot actually be admitted to be done by an AI (The Amount of "shit-flipping" that would occur would be catastrophic. Imagine if the writers guild heard! Way worse than "I poached an idea from a human" -- which is why professional authors won't ever listen to "your ideas").
The third rule is "write lots and lots, and get lots and lots of feedback -- tweak, tune, refine."
So, again, do you actually have an AI that you can point me to that writes well, or are you just a ball of nonsense?
If ever I feel like the first half of Shakespeare's Sonnet 29 (https://www.poetryfoundation.org/poems/45090/sonnet-29-when-in-disgrace-with-fortune-and-mens-eyes), I can console myself that "Well, at least I don't live in San Francisco".
Now, I know politicians are an opportunistic breed who will leap on the most immediate even if most trivial instance of anything to spout off about it and how they are Serving The Nation, Protecting The People, and Doing What Is Right And Good (please vote for me, donate to my campaign, pick me, me, me!) but this is pretty thick, I have to say.
https://www.imightbewrong.org/p/is-kitkat-the-modern-day-horst-wessel
Before watching the linked video, take a guess as to who she is referring to:
"Yesterday, our beloved [redacted], belovedly known as the Mayor of Sixteenth Street, was tragically killed by a Waymo self-driving car. This is a devastating loss for our neighbourhood. I personally loved [redacted] and I'm deeply heartbroken and upset mainly at his tragic passing."
What staunch citizen was this so cruelly slain? What long-standing member of the community?
A cat.
If you're making a guy who used to write for the John Oliver show go "whoa, dial it back!", then you truly are egregious.
Can anything ever be done in San Francisco without some member of the Board of Supervisors throwing off their clothes? Can anything ever be done in San Francisco at all? Why the Board of Supervisors, since they seem to do nothing (so far as I can tell) except try and build the People's Democratic Socialist Republic of San Francisco (Extra-Progressive Version) but little to nothing else?
Come on, people love cats. "I knew this cat too, and cared about it." is an attempt by a politician to be a "man of the street."
I don't honestly believe the politician, when they say "I loved the cat and am upset about it dying" -- I'd want at least a photo of them petting the cat to even consider that they aren't just showboating for the cameras.
Disclaimer: I was nearly run over by a Waymo in LA, last time I was there. Definitely biased perspective.
Your redaction doesn't work since the answer's in the url.
Why do you care?
I care, in as much as I do care, because it's stupid. A cat got run over by a vehicle after running under the car.
This serves as an excuse for a local politician to ride her hobbyhorse about banning Waymo, banning electric cars, banning everything that can be banned.
And she pretends that her heart was broken by the cat's death. It wasn't even her cat, and I doubt she knew more about it than "there's a cat in that store". Maybe she even saw the cat a couple of times.
But a cat? If a person had been knocked down and killed, there would be some reason for this kind of posturing. But it's a damn cat, and the only reason she cares is that it gives her the excuse for her call to Ban It All.
I should be hardened to these kind of stunts by now, but it just rubs me up the wrong way.
A politician uses a minor event as an excuse to campaign against something that she presumably already did not like. Come on, that sort of thing must happen everywhere where there are politicians.
And it's half the world away. Surely there are more interestingly stupid things happening near you?
I mean, what rubs me up the wrong way and made me care that you care, is what I perceive as US-centrism and specifically SF-Bay-Area-centrism from people who (like me) are not even American nor live there. Scott, of course, can write Bay Area House Parties all he likes: it is normal that he has more to say about the milieu he lives in than about others. Plus they are genuinely funny even though we probably miss half the jokes. But let us not get carried away and start thinking that that place is truly more important than everywhere else. Their cats getting rolled over and their politicians politicking are just as significant as those from Jaboatão dos Guararapes or whatever.
Would it affect your view to know that, based on recently released data, we now know that Waymo self driving taxis are at least 36 times safer than regular taxis (and by implication, enormously safer than human operated vehicles in general?)
Alphabet has been talking up the safety of driverless cars for a long time, but this new data is gathered from a sample set of 5 million Waymo rides. So obviously this is the best data we've seen so far and it is breathtaking.
The number of lives self-driving vehicles can save is significant. And even by itself waymo has made our streets apprecially safer.
Does the fact that this politician is pushing for demonstrably awful policy, imasmuch as she wants to penalize waymo, interact with the other facts of the case in order to sway your opinion?
If I had a view against Waymo, it would be affected. I do not. I maintain my view that a local politician having an opinion that is supported by anecdote while refutable by data, and also is representative of usual socio-political tendencies (conservatism against a new technology, populism against a big company) is just ordinary politics which do not merit strong feelings from people on other continents.
Yeah, I realise she doesn't actually give a damn about the cat, but the stupidity of it offends me. She's going on like this cat was a human being tragically murdered by a heartless big corporation, and other people are willing to hop on her bandwagon instead of saying come off it. It's just grubby opportunism all round, plus pretending an animal is as - or even more - important than a person.
Contrast this with Scott's most recent post "What happened to SF homelessness?". The cynical answer might be "the activists got bored with the homeless and instead started crying about dead cats".
The correct answer is the Homeless were Illegal Aliens: the Prototype. They weren't needed after the illegals were brought in. (and yes, the illegals contribute to the overall homelessness).
isolated demand for rigor
isolated demand for rigor mortis
isolated demand for "stick the damn cat in the ground and actually go do something valuable for running the city"
Does anyone know of a good source on the unit economics of openAI? I’ve seen the write up for the API business specifically, but I’d like to know for just the average chat user.
Our new president has been inaugurated today. Ceremony was the usual tedious performance of speechifying, but the interesting part is why was she all decked out in purple? Well, now she has the job, we'll see what she does in the role!
https://www.youtube.com/watch?v=guI8iss91uQ
Wait, isn't green the traditional color of the Irish? I foresee a potential calamity here. An entertaining brouhaha at least, or maybe a donnybrook. https://youtu.be/AcBTOU7RvbU
Yes, green is the traditional colour, which is why I was surprised to see her decked out in imperial purple.
One reason may be that green is also the colour associated with one of our parties, Fianna Fáil (Fine Gael are blue, Labour is red, the rest of them are various shades):
https://en.wikipedia.org/wiki/List_of_political_parties_in_the_Republic_of_Ireland#Party_details
Looking at the above, purple is the colour of the Social Democrats, and President Connolly was the left-wing compromise candidate, so maybe that's why she's wearing it. Though she started off as Labour (red) and is now running as an Independent (no colour) so I dunno. This article says she often wears purple and it's a colour associated with the Suffragettes?
https://www.irishexaminer.com/lifestyle/fashionandbeauty/arid-41740713.html
Scott,
Reflecting on https://slatestarcodex.com/2017/10/02/different-worlds/
"The classic (though super offensive) explanation is that some people seek out abusers for some reason – maybe because they were abused as children and they’ve internalized that as the “correct” model of a relationship.
And maybe this is true for some people. I have a friend who admits it’s true of her – her current strategy is to try to find someone in the sweet spot between “jerkish/narcissistic enough to be interesting” and “jerkish/narcissistic enough to actually abuse her”, and she’s said so in so many words to people trying to matchmake. I guess all I can do is wish her luck."
You know, Scott, this case is exactly what consensual BDSM was invented for! A truly textbook case.
This just describes a good Dom, someone who has a playfully cruel side and likes to be worshipped as a household demigod/ess, but never goes too carried away with it, has good self-restraint, and respects boundaries and consent, and generally wants to give their partner a good experience and keep them safe, generally, they have a caregiving, kind of parenting side as well.
BDSM is ultimately just excellent communication and self-knowledge. The big problem with the usual jerk or narc is that they do not know they are a jerk or a narc. Or won't tell. Every good Dom ever I know used to worry whether they are narc and that might harm people, and/or asked a shrink about it. Who usually reply real narcs never worry about maybe harming people, and stop self-diagnosing with stupid shit. (Mine said this almost literally.)
Theodore Dalrymple noted that the battered women who consulted him found non-abusive men to be intolerably cold and distant and would dump them faster than the abusers.
And it is clear they were uneducated underclass women. Just textbook proles, massively fat, having 3 kids from 3 guys and so on. Mama June category.
Women abuse men just as often as men abuse women. Slightly fewer ER visits, more verbal abuse.
Many relationships have two abusers.
Dalrymple worked in the slums of London. He frequently surprised his patients by observing that their partners choked them, and then revealing that he's been consulted by a patient in her situation every day of his career there -- and also she still had the bruises about her throat.
Birmingham, not London, I recognized the architecture and dialect. I used to live there. Probably the one of the worst places this side of the ocean. People did not even have dining tables because they were used to eating in front of the television. Underclass Brits are easily the stupidest white people on the planet. They rarely wear anything but tracksuits...
I'm starting to wonder if these people pass the marshmallow test. Did they forget that they were choked by their partner, or were they surprised that he was mentioning it? (The latter indicates a society that looks the other way to this sort of violence, the former indicates a mental problem -- perhaps they didn't realize the bruises were indicative? This also indicates a severe mental problem).
We are talking about underclass Brits, who are easily the stupidest white people on the planet. Or stupidest people in general. They don't even have dining tables because they eat in front of the TV, that kind of people.
As for the marshmallow test -- HELL NO.
Willful ignorance. Part of the problem is that they think this is normal -- and yes, it is, in their area -- and part is that they censor out things that reflect badly on them. He asked them whether they could have predicted the men would be violent, and they said no; he asked them whether *he* could have predicted it, and they said yes; he asked them how he would have done it, and they recited the evidence that he would, in fact, have used.
You might find interest in his *Life At The Bottom*
This should be common knowledge. For both sexes. If your repeated experience is that "all members of the opposite sex are X", chances are that this is about you -- that you find the non-X member unworthy of your attention, effectively making them invisible.
Who your partner is, is a reflection of you. You may hate this fact, but unless you change yourself, you won't get a better partner. (Exception: if your partner recently changed dramatically, e.g. as a result of taking drugs, or brain cancer.)
Maybe what your friend wants is a dom, and maybe not. I don't think that "being a jerk" and "BDSM" map completely onto one another, though certainly there would be overlap with the types who think they're dominant but are just jerks.
It might well be that your friend wants the risk of "he could turn abusive" for the thrill of it, and not "safe sane consensual".
Sexual dynamics are a separate category from interpersonal dynamics. Just because you know exactly what you want sexually doesn't mean you have any general introspective skill.
Do you have any insight into why this technical limitation exists?
My understanding is that LLMs produce a probability distribution over a set of tokens, and then we sample that to decide what it 'writes' next. Could it not, in principle, make every other probability distribution a be '99% probability of a dash token'? Such that half of its outputs are a dash, and the other half are other tokens.
Suppose the token dictionary was {'I', ' am', ' a', ' lan', 'guage', ' mo', 'del', '-'}. Why coult it not output "I- am- a- lan-guage mo-del"?
For those who are confused: this continues another thread that can be found above/below depending on your sort setting for the comments. Presumably the author replied to something by email or whatever triggers this thing where replies get detached from their thread.
LLMs use something called an activation function to pick the token to use next (often softmax is used for this calculation). It could definitely pick a dash every other words, if it predicts that as the most probable next token. So maybe I don’t really understand the question.
Try prompting it with, for example: “ Are- you- an- LLM-?-can- you- mimic- my- speech- pattern-?-”
I don’t think it’s accurate to say that LLMs use an activation function to pick the next token. Activation functions are found in many layers of neural networks and are responsible for making sure that the mappings are non-linear. On a basic level what LLMs do is take a numerical representation of the prompt as input, and output something called logits. Logits are real numbers, and LLMs output one for each token in the vocabulary. The logits get passed through an activation called the softmax function to covert the set of real numbers to a probability distribution, from which a token is then sampled, but this is just one of many activation functions that the inputs pass through
You’re right- there is a more precise way to describe what is happening, and the activation function is used at every computed layer for each v, not just the final one. But it seemed useful to bring it up in a conversation about LLMs using probability distributions, even if simplified… I wasn’t trying to explain the transformer architecture, just to add some useful context that could point Tiago in the right direction.
Which technical limitation are you talking about?
"Could it not, in principle, make every other probability distribution a be '99% probability of a dash token'?" - yes, in principle, could, if the training data is like that
I also think you can probably just prompt current llms to do that (though I'm not sure it would work)
Upd: just tried it, with several attempts got the necessary result (although I don't have access to exact tokenization of the output so cant confirm for sure)
Another data point for weight set point. Here claiming that total calories expenditure of 10x activity hunter gatherer is the same as sedentary American. It seems that to compensate it, the body reduces energy for everything else.
https://gsas.harvard.edu/news/colloquy-podcast-why-exercising-more-may-not-help-you-lose-weight
And yet, people get fatter.
Jane PSmith's review of Pontzer's book Burn: https://www.thepsmiths.com/p/review-burn-by-herman-pontzer
Nice! I'm actually quite skeptical so I'm waiting for something that either criticizes or supports it. Sounds like it'd be very nice if Scott read and review this Burn book.
At this point, I don't believe anything I read about hunter-gatherers. The amount of mythologising, wishful thinking, Noble Savage stuff churned out about how hunter-gatherers only worked (if you can call it work) 4 hours in the week and just lolly-gagged around the rest of the time, boo hiss agriculture made us all slaves, that I can't take any accounts seriously since everyone has an axe to grind about it.
Wait -- it's not for sure that they had great sex?!
Clan of the Cave Bear lied to me.
Smelly, dirty, gross sex.
Archaeology is your friend. Stones don't lie!
But weight is remarkable hard to lose. Something must explain that...
Being overweight changes your metabolism, annoying but true. To me it seems like a bug in the system that never got selected against because “having too much delicious food to eat” was not a problem, our evolutionary ancestors had.
Not like Oog the caveman could walk up to a vending machine and order a candy bar.
“Weight is remarkably hard to lose”
No, it’s not. Just don’t eat (much)
Be prepared to suffer every moment of your waking life.
The first Noble Truth of weight loss.
Suffer what? Hunger? I've found that the sensation of starvation dulls, after a while. Usually a day or so. Weight drops then, and quite quickly. I've lost about fifteen pounds over just two weeks before. I don't recommend it, it's terribly unhealthy, but it is not hard for me to lose weight. That may be because it is not hard for me to resist hunger. It doesn't feel like suffering, anyway. It's actually annoying for me how little effect my hunger cues have on my body, it makes it difficult to remember to eat.
There is considerable individual variation on this. Surveys show that the majority of people have a lot of trouble keeping the weight off.
> But weight is remarkable hard to lose. Something must explain that...
Why? In the ancestral environment, food energy was always a limited resource. It makes perfect sense that animals would conserve energy as much as physics would allow them, adopting a lower rate of metabolism when times were lean. You might as well demand an explanation why more men are straight than gay. Both of these findings are well explained through straightforward evolution.
Now, if it was the case that when undersupplied with energy, animals would increase their rate of metabolism and quickly burn through their fat reserves and starve, that would be surprising and worthy searching an explanation for.
To be fair the hunter gatherers here are real tribal people still living in Africa
These are the survivors of the agricultural revolution; they've been pushed to marginal lands and arguably aren't representative of paleolithic culture ten thousand years ago.
Oh, the real people, I believe in and whatever they do. The social psychologists, economists, health types as per the link etc. who use them as puppets for their pet theories? I don't believe it.
Existence is tough enough that even if you can gather roots and berries and hunt small to medium large game, you won't do all that in four hours and then loaf around doing nothing. You have to make and repair clothes, hunting equipment, tools, cooking, the rest of it. You have to look after the children, the sick, and the aged. You have to pack up and move on once you've exhausted the easily available roots and berries in the locality, and that means walking long distances. You have to find water.
You have to create and maintain social bonds.
This notion that you just go out, pluck some fruit off the bushes, kill a deer, and that's it for the day - fanciful Western wistfulness for the Golden Age.
Did you read the linked article?
Its about a specific study on a specific population. Its not a Rousseau-ing anti-capitalism piece.
It's more like: What can we learn by measuring different populations caloric-outputs and inputs, the amount of exercise they do, and the prevalence of heart-disease etc.. You are the one that is bringing the value judgements about hunter gatherers into this specific discussion.
It seems like very valuable research to me. I'm not super good at evaluating research like this so I'm taking it at face value, and I would be very interested in real critiques of it if anyone has some.
Until quite recently I believed that I could lose weight by just exercising a bunch for a while.
Studies like this convinced me that it's actually just better to plan on very incremental diet/exercise regime changes over very long periods of time; focussing on what you will really do. And that exercising or fasting to lose weight, will almost certainly fail so I should instead focus on finding group exercise that is super fun that I can do in a group, like Bouldering.
I think your comments are more relevant to the hobbes/rousseau discussion, like here: https://www.astralcodexten.com/p/your-book-review-the-dawn-of-everything
"Its about a specific study on a specific population. Its not a Rousseau-ing anti-capitalism piece."
Reading the linked article, it's a specific study in the service of him shilling a new book about yet another "how to lose weight healthy way" regime.
So pardon me if I don't think it's completely pure scholarship unalloyed by any taint of outside interest.
"All of these systems that we aren't even really aware of are actually where you spend the bulk of your calories. And so, what we think is happening is the Hadza and other really physically active populations, they're spending a bit less on those other processes to sort of make room for the physical activity. So, there's no magic here. The laws of physics remain intact, undefeated. [CHUCKLING] It's just that the way that they're spending their energy is different on other tasks."
Okay, so he doesn't know what is really going on there, he's estimating that this is the mechanism for why this group apparently don't expend more energy than sedentary Westerners.
But if his hypothesis is correct, they *are* burning more energy - they're diverting energy reserves to support the physical demands of their lifestyle (and presumably paying the price in lower physiological health for the systems this energy is diverted from).
If the Hadza and Westerners have X amount of calories available to burn, and the total you burn is A (for physical exertion) + B (keeping the systems of the body running) = C (total energy expenditure), then by his theory:
Westerners: X - C = D (surplus, excess is not expended on A and B is fully supported, so remainder goes to fat/obesity epidemic)
Hadza: X - C = E (deficit, energy is diverted from B to support A, thin/not obese)
Thus the Hadza *are* expending more energy by robbing Peter to pay Paul. If the Hadza had the same amount of calories and/or the same sedentary lifestyle as Westerners, would they be as obese as Westerners? And the answer seems to be "yes" since we see this happen in populations when they get to Western lifestyles.
EDIT: I can't make any sense out of what he is trying to say, maybe it's the interview.
All calories are the same. Exercise is good for you in some mysterious way but it is not because it causes weight reduction, it's because it causes energy rebalancing. Starving/reducing calorie intake doesn't help with long term weight loss. There is no metabolism slowdown as you get older. Carbs are fine (but not the simple carbs). Oils can be as bad.
"Putting aside the myriad of health benefits of exercise and speaking strictly about weight loss, I mean, is the message here like it doesn't really matter how much activity you have or it doesn't really matter what you eat?
Yeah. If we're strictly looking at weight loss, then this is just more evidence—and it's not the only line of evidence—but it's just more evidence that exercise alone is a pretty poor tool for weight loss. If you absolutely push it as hard as you can for as long as you can, you might see some weight loss, at least initially until your body kind of rights the ship. But the long-term expected weight loss from exercise alone is something like 2 kilograms. Right? Less than five pounds.
And we're talking like a year after you start your exercise program. You can be diligent about it every day or every week. And you can expect to have lost, at the end of the year, about five pounds. So, it's not a great tool for weight loss. It is really good for all the other things, as you say. Now, weight gain or weight loss, again, it's about that energy budget. Calories in versus calories out.
And what our data are showing is that the calories out is really hard to budge. Your body doesn't want to change that very much. And so, what that says is that then weight gain, obesity, overweight, these issues that we struggle with in the US, it is about the energy coming in. Right? It's about the energy in part of the equation, so it is about diet.
But are all calories created equal?
Yeah, they are. So, whether you get your calories from carbohydrates or from fats or even from proteins, although you get a few of them from proteins in a typical diet, they all will have the same effect on your weight in terms of the calories that you store in your body is weight. Now, that doesn't mean that all foods are created equal. Right? I mean, a cupcake and a bowl of broccoli might have the same number of calories, but they'll have different effects on your health. So, that's not the argument of sort of calories are being equal. But in terms of just pure weight gain and weight maintenance, yes, calories are calories.
So, no “carbohydrates are the bad guy” here?
No. I think that's really been oversold. So, simple sugars, added sugars in our processed foods that we eat are one way that we end up eating more calories than we planned to or we can trick our brains a bit into over consuming because your brain is pretty good at matching energy intake and expenditure. So, all those extra empty sugar calories are one way that you can push it over the edge. So, in that sense, yeah, carbs aren't great because sugars are carbs.
But added oils are no good either. Right? And they're another way that we can trick our brains and overconsume. And oils are fats. When we look across different populations, like the Hadza, for example, the Hadza eat more carbs than we do. Right? And they don't have any issues with their weight. If carbohydrates were really driving the bus when it came to weight gain, the Hadza should all be overweight. But of course, they're not."
So dont eat sugar, don't eat oils, exercise a lot, but basically you won't lose weight by diet and exercise? Someone please explain this to me, I must be getting it wrong.
Would you have us throw our hands up in the air and stop researching weight loss and healthy weight maintenance? Or at least stop writing books about it?
I just read the review of his book someone posted the link to above and found it pretty interesting https://www.thepsmiths.com/p/review-burn-by-herman-pontzer
In regards to:
> But if his hypothesis is correct, they *are* burning more energy - they're diverting energy reserves to support the physical demands of their lifestyle (and presumably paying the price in lower physiological health for the systems this energy is diverted from).....
Hadza: X - C = E (deficit, energy is diverted from B to support A, thin/not obese)
I don't think thats what he means. I previously understood it as exercise reduces the needs for B ( caloric usage for keeping the system running, fighting inflamation etc.). Its just using another, more effective method of achieving the same goal. But now, having read that review, it seems that the argument might be rather, that by not exercising the ideal amount (evolutionarily speaking) we are causing bodily changes including overproducing certain hormones like testosterone or cortisol which can potentially cause some diseases (maybe also some benefits?).
here were some relevant snippets of the review:
"some researchers have suggested that when we become more active we unconsciously reduce non-exercise physical activity like standing or fidgeting, but Pontzer thinks most of the cutbacks come from the energy that would otherwise be spent on things like stress reactivity, the reproductive system, and inflammatory response. Endurance athletes, for instance, produce less adrenaline and cortisol in stressful situations than sedentary non-exercisers, and exercise is well known to help with anxiety and depression"
"So “missing out” is the wrong framing for this: we don’t actually want such sensitive stress responses, high rates of breast and prostate cancer, and widespread inflammation. They’re bad for us. And more active populations don’t have to worry about them."
"...complex evolved systems — like the human body, or, say, culture — don’t work that way. They’re made up of feedback loops and inhibitors, uncountable tiny things that all work together in unpredictable ways, more like a large language model than an internal combustion engine."
About what type of foods or which calories to eat, I think in the simplistic sense of calories in/calories out "weight gain" caluculations then they are equal, but I don't think that is useful information. The review mentioned for example that some foods might be designed "to hit all your palatability triggers without any of the satiety ones". https://en.wikipedia.org/wiki/Sensory-specific_satiety
I always wonder about Coke Zero, and aspartame, what the overall effect of them are even if they contribute no calories.
> So dont eat sugar, don't eat oils, exercise a lot, but basically you won't lose weight by diet and exercise? Someone please explain this to me, I must be getting it wrong.
Exercise and a balanced diet is definitely good for your body. You probably won't lose that much weight just from that, though. What matters is how much you eat, not what you eat. The problem is what you eat affects how much you eat. Does that make sense?
Personally, I had decent success with just ordering Semaglutide from China. At a cost of a couple of hundred $ per year, plus the hassle of weekly injections, I have lost perhaps 10kg in 3-4 months.
Now, I will grant you that if I actually changed my eating habits, or did sports, that would be a lot more wholesome and healthy, so if you can do either, more power to you. Personally, I am mostly willpower constrained, and all of the nobler options would require spending willpower while injecting Sema is free in that regard.
I'm curious if there is a good way people have found of developing new habits around food while taking Semaglutides that they are then able to continue when they go off.
I assume people plan to take it for a while, drop down to their goal weight, and then gradually transition off of them, is that correct?
When I wake up, I keep thinking in sorts of "dream logic" for at least the next half-hour, even if I fully woke up and am doing things. My thinking is clearly not well, you know how it is in dreams. It slowly inches back into normal thinking, as I realize how silly my thoughts were.
What can it be? I don't even know how to google those symptoms to find out. No history of psychotic or delusional disorders.
I'm almost jealous. I can only maintain that state for a minute or two, and then only lying on my back. Then I'm fully lucid, and it's over.
Anyway, while we are dreaming, the brain is operating in a particular mode. Certain functions are turned off, including most analytical thinking. Other modes are fully active, including whatever cognitive process allows us to engage in free association. To oversimplify quite a bit, there are "switches" in your brain that are supposed to reset upon waking, so that you stop dream mode and boot up awake mode. Clearly, one or more of these neural switches are out of sync with the rest of your brain, so you physically awaken before the rest of your brain has fully caught up.
The term you are looking for is "dissociative state of consciousness", which includes a wide range of things, not just a dream-like state upon waking. You might find this article interesting: https://www.mdpi.com/2077-0383/12/12/3876
Thank you for this
You're welcome!
Are you taking gabapentin? From personal experience and other's anecdotes, and it has the tendency to not only make dreams significantly more vivid, but make their "realness" extend past the moment of waking. SNRIs seemingly counteract this effect, and weren't you talking about quitting venlafaxine a while back?
If you're not taking gabapentin, I would still check the side effects of any medications you're taking, particularly for mentions of drowsiness.
I'm not taking gabapentin, but I have reduced my venlafaxine dosage.
I am taking prometazine, though. Maybe it's a side effect of prometazine that got exacerbated by reducing venlafaxine?
Promethazine has sedating properties, so I wouldn't be surprised if it had some similar effects. Admittedly, I'm not a medical professional, so you should probably talk to your doctor about this.
I'm sure it's been asked & answered here before, but why is Trump so good at Middle East geopolitics? Not even talking about IL/PS & Gulf State things right now — I think gesturing at "loose-cannon tribal strongman" vibes does enough work there (not really, but this comment shouldn't get longer than it is already). I'm more interested in Syria & Iraq.
Certainly the positive developments in each are somewhat a result of bigger geopolitical forces, like the Russians getting all mired down in Ukraine, and the Iranians & Hezbollah getting their shit rocked by Israel — but the developments are *so* positive! Baghdad is booming (https://www.economist.com/middle-east-and-africa/2025/09/04/the-worlds-surprise-boomtown-baghdad) and Mr. al-Sharaa's gone to Washington (https://www.nytimes.com/2025/11/10/world/middleeast/syria-president-al-shara-trump-washington.html), and everything is going really friggin' smoothly.
I mean, who else in the history of American politics could pull off a line like, "People say he’s had a rough past? We’ve all had rough pasts." about a former al-Qaeda-ish leader, a dude whose head carried a $10 million bounty until about a year ago. And, to be clear, the rhetorical acrobatics are coupled with serious diplomatic & material changes: our own sanctions gone, plus a successful bid in the UNSC to drop international ones, and also a commitment from al-Sharaa to knock the last bits of ISIS out. All the extremely-well-publicized massacring seems to have been minimized and marginalized, the central government is stable and powerful — arguably maybe a bit too central and stable and powerful — but things look good!
Iraq is a somewhat less rosy picture — despite the Baghdad boom, it seems like the Iranians have got a solid foothold, and are maybe moving to solidify it. Elections are going on, seemingly pretty smoothly, but some Atlantic Council report I found (https://www.atlanticcouncil.org/blogs/new-atlanticist/dispatch-from-baghdad-dont-confuse-the-calm-around-iraqs-election-with-stability/) suggests that maybe we should be worried, that the quietness & smoothness are very superficial, that lots of tensions exist underneath... And then! The report closes with a question, "what should Washington do?" and an answer, "keep on keeping on, Trump is doing a great, awesome, sober, and practical job."
This is wild! I thought Trump was gutting all the talented and reasonable bureaucrats everywhere, but it seems like maybe the State Department is immune? Or only the Mideasty parts? Is Marco Rubio somehow acting as a super-sane bulwark? But only in the Mideasty parts? (Or maybe he's a super-sane bulwark in general, but makes an exception for Venezuela-boat-bombing because of his whole super-duper-anti-Castro thing, and Chavez/Maduro are close enough?) Maybe Antichrist shenanigans are magically protecting only the near-Jerusalem-est parts of the world? What's going on!
I didn't see this take exactly (though I saw some that were close) so I'll make it here: I think this analysis and analysis like it are evaluating a point in time and missing the trend. The trend is that the middle east was liberalizing *very fast* starting in the Obama years. The reason for this is because the oil states realized the oil wasn't going to last forever, they collectively decided to turn westward because China and Russia were less good options (Syria decided to go with Russia and it didn't turn out great for them) and that in turn meant thawing relationships with Israel and becoming more aligned with standard Western liberal practice in order to attract tourist dollars. This all sped up post Arab spring, even more so as the US became a bigger oil exporter. There's a whole slate of things to this effect that Trump had no impact on -- Dubai becoming middle east Vegas, Saudi (and Morocco) loosening restrictions on women, qatar hosting FIFA, the Iran deal. And that's before you get into neo liberal economic reform. You can go to Riyadh or Doha and it's like any other Western city, with all your favorite brands and whatever.
I think the Abraham accords are legitimately great diplomacy, but I think generally Trump likely did more harm than good -- that is, he likely slowed down the trends that we were already seeing and were going to keep seeing due to macro effects in the region totally unrelated to trump himself. Bluntly, Trump may have inked a ceasefire, but in a world without Trump there may not have been an attempted genocide first (always remember that there was a ceasefire on the books when Trump came into office, and it got torn up more or less based on Trump's indication that he wasn't going to care very much if Israel started bombing Palestine)
Well, you see, Virginia, it starts with getting rid of USAID.
(This is in direct reference to Egypt, where USAID was actively causing our "diplomatic corps" problems and wouldn't stop funding through Muslim Brotherhood, even when the US government rang them and said "stop this bullshit, we're trying to prevent a war.").
Iraq is like the Ukraine, nearby countries that are deservedly upset at the thought of American Intervention, want to have some influence over their neighbors. I'm not any more upset about Iraq being in the Iranian orbit than I am about Canada (or Mexico) being in ours.
We’ll see. The Middle East has a habit of not staying fixed.
Seconding Jack 's comment below, and to add to it- Trump is the textbook example of "if something good happens, attach your name to it, if something bad happens, disclaim all prior knowledge and punt responsibility to a subordinate/scapegoat or blame outside forces/sabotage, as applicable".
See also, Trump's disastrous war against the Houthis that got quietly memoryholed.
Trump has a rather unique perspective on negotiations and diplomacy. It leads to a lot more work for people who were normally not allowed into the hallowed halls of dealmaking.
(Deprioritizing private jets from major airports is how he got the government to reopen, for example. Turns out billionaires don't want their time wasted having to drive into town).
Richard Hanania has a lot of good articles about this, like https://www.richardhanania.com/p/how-trump-proved-foreign-policy-experts. In essence, Trump's model of the world is more accurate when it comes to the Middle East.
Yeah IDK. As far as understanding Trump, I find that Victor David Hansen does a pretty good job for me. I don't know if he's talked about the middle east. (I was home sick yesterday and watched this interview by Coleman Hughes.) https://www.youtube.com/watch?v=61vjJhLtAyU
A lot of this just seems like lucky timing. The Syrian civil war went on for 14 years, including his entire first term. People are extremely sick of it and willing to be optimistic even if the outcome isn't necessarily what they would have wanted.
Also, honestly people are TOO optimistic, or getting ahead of themselves. Look at https://understandingwar.org/research/middle-east/iran-update-november-10-2025/ - the new leadership doesn't even control half of the country. There's no reason to assume those other groups will evaporate or make peace without something bad happening.
I think it's just about being non-ideological.
If you declare saudi arabia a pariah state like biden did, it's hard to get stuff done
If you give a speech saying 'america is done moralizing to the middle east, let's just build stuff together' then a huge amount of opportunity for cooperation becomes possible, and in general cooperation leads to stability and other good things on balance
> If you declare saudi arabia a pariah state like biden did
I do not think that this is a fair summary of Biden's de-facto policy towards Saudi Arabia. From what I can tell, he simply declined to sell them weapons for a time due to concerns about their human rights record. The US continued to buy Saudi oil and kept their military bases in the kingdom. For comparison, the US does neither have military bases within North Korea nor does it buy resources from them.
For Trump's pragmatist approach, I think it helps that his base genuinely does not care what brown heathen foreigners do to other brown heathen foreigners. As long as MBS does not kill to many Christians or babies (which might upset the religious right), his base is very chill with a live-and-let-kill policy.
As for him personally, I think it is obvious that he admires strongmen. Where most other American politicians would flinch at the attics of MBS (having a journalist murdered in an embassy and then cut up into pieces would not make for a cozy dinner companion for Obama or GWB), Trump is beyond such petty human concerns. He is probably more interested in MBS' lack of a term limit than his human rights record.
It also helps that it seems to be really easy to get on Trump's good side by (1) not attacking him politically and (2) liberally donating to him. Where Biden was more like "no, you gifting me an airplane would be inappropriate, may I instead suggest you give my son a position on your company's board of directors?", with Trump, such extra steps are plainly not required.
>may I instead suggest you give my son a position on your company's board of directors?"
This so did not happen.
Yeah, I don't give him that much credit. Yes, a ceasefire, which is a very good things if it lasts. But while the violent death rate has declined in the Gaza strip (again, very good thing) Trump himself hasn't solved anything. The two sides (Israel and Hamas) will be back at it sometime in the future, no doubt. I see no way in which Trump can take any credit for Baghdad's economic state (good or bad). What did he have to do with it? The peace deal between Russia and Ukraine fell to pieces. He threw our alliance system out the window. The tariffs have been an unholy mess, and the Chinese are laughing at us all the way to global hegemony. So, yeah, credit where credit is due, but he hasn't earned the Nobel Prize just yet.
> but he hasn't earned the Nobel Prize just yet.
Actually, I think that before he gets it it will be renamed the Nobel Prize for War because that sounds more badass.
It's the gold plated Louis XIV style. That's half-serious. But the snooty have been turning their noses up at Trump as a vulgarian since forever (my dear, have you *seen* Trump Tower? The arriviste nouveau-riche gaucheness of it all!) but I mean, are the Saudi oil rich princelings known for good taste and understatedness? It's the sort of visible display of wealth and power that signals "I'm the guy in charge, talk to me, I get shit done" that these cultures understand.
Add in that he's brusque and crude in speech and manner, and that's more of it. What you see is what you get, he's willing to cut a deal, talk tough and then negotiate.
EDIT: Above is all uninformed personal opinion, take with grain of salt.
I think I buy this in terms of negotiating with Hamas, but that Atlantic Council report has passages like:
> Without military engagement, the Trump administration has been able to leverage the underutilized weight of US diplomacy to block a few laws counter to US strategic interests and resolve some long-standing issues, while pressuring Baghdad to release Kurdistan’s share of the budget.
and
> a senior State Department official told me, “we are dealing with the system, not individuals.”
This strikes me as extremely mature, and not even a bit vulgarian? Agreed that the brusque, get-shit-done *appearance* is often good and helpful, but the mystery remains: why should this be the ~only part of his administration where it seems like saner heads are winning out behind the scenes?
Did you read about the vaccine board lately? There are sane people on there (engaging in trust building with the "committed loons"), and they're making sure we don't inflate a mild increase in febrile seizures into "cancel this vaccine."
Saner heads are more places than you think.
Also, Biden's administration was significantly less sane than you think (Dick Cheney was involved, and if anyone who elected Joe thought they were getting Neocon Vicky and Dick... I'll eat my hat).
No idea. Who's his State Department guy (or gal)? It might be as simple as Trump genuinely does not want to get dragged into wars if at all possible so he follows Churchill that "jaw-jaw is better than war-war". If you can sell it to him that diplomacy is "negotiating a better deal" then I think he'd stand back and let the diplomats handle it.
And of course every political leader wants the kudos of "I brought peace to the Middle East". Clinton wanted that, Blair wanted that. If Trump wants that too, no surprise.
Speaking of which:
https://www.youtube.com/watch?v=WcsrRz8EDq8&list=PLwpOQsKKqZGOQ3Uwdbt0CcX7DeCJ1uBNV&index=34
If Trump has a clear understanding of what the US' strategic objectives in some gulf state are, I have to applaud his success at feigning ignorance in most of his public appearances.
If the US is successfully leveraging diplomacy to their advantage in the ME, it seems much more likely that it is the doing of neocon Republicans keeping a low profile than of the MAGA crowd.
I do see Trump's results in foreign politics quite positive in general, but I don't think that he can take much credit for Sharaa. This all started before Trump, and the US are following essentially the same road as all the other Western nations here. It's just the natural reaction to Sharaa's politics. Even for the US, if I am not mistaken then the sanctions were already substantially lifted under Biden. And a commitment to knock the last bits of ISIS out, isn't that what Sharaa wants anyway?
I think Trump does the right thing, but in this case the right thing is very easy to do, and everyone else is doing exactly the same things. I also don't want to belittle it too much though: it is very laudable that Trump pushes hard in that direction, and he deserves credit for that, and of course, the impact of the US is much higher than that of other states.
The bits where Trump deviates from what other Western nations do is that he gives strong support for Sharaa winning back control over the Kurdish regions in the North East of Syria. This may be a good or a bad decision, and only time will tell.
> Sharaa wants anyway?
There’s been an incredible campaign in rehabilitation here. Future historians are going to be amused
Can you explain what you mean by that? What is "here"? Syria? US? ACX?
Here is obviously not geographical. It’s in this instance, or in this work (your comment and in general the attitude to Sharaa).
Again, could you please explain what you mean? You refer to my comment regarding Sharaa being happy to move against ISIS. As far as I understand, Sharaa's HTS and ISIS have been enemies for over a decade, and HTS has been violently fighting against ISIS groups for years. Do you disagree?
Mind that I do not conflate ISIS with the general concept of terrorism. There have been many other terrorist groups in Syria, including al-Qaida and HTS.
I think you seek to be arguing by pedantry and in bad faith. So, no.
I think the Middle East is a poor choice of example to highlight Trump's dealmaking abilities because it's unusually easy as these things go, just held up by breathtaking incompetence spanning decades: for example, I understand there was an actual policy NOT to negotiate with certain powerful actors because it might give them "legitimacy," or "normalize" them. As long as he did Something Else, it would have been enough.
I don't agree with the premise.
Re Baghdad, don't know much about it but a big construction boom seems like the sort of thing that takes years to come together, and nobody seems to be pointing to anything that has happened specifically in the last 10 months to spur it along. What I've read credits their current PM, who was elected in 2022, with reforms and all that.
Re Syria, it's hard to say who else "could pull it off" because that guy only took over Syria after the fall of Assad near the end of Biden's term, so it's not like anyone else had a chance. And it's not exactly unprecedented for people to switch between "friend of the US" and "enemy of the US" in the Middle East, e.g. Saddam Hussein.
People like to give credit to Trump for a cease fire in Gaza, but there was a cease fire in place when he took office.
We've also had two wars in the Middle east this year. Against the Houthis (which seemingly ended without us accomplishing anything but losing a few planes), and the Iranians (meant to destroy their nuke program, the intel people said we didn't do it and so he fired the people who said that).
As much as I'd like to say that Trump is fucking it up and everything is a disaster, I think the real truth is that things happen semi-randomly and due to forces other than what the US is doing. There's some good, some bad, and we'll keep muddling through. E.g. you could say all the positive Syria stuff is Biden's credit because Assad fell during his time ... but that's probably giving him too much credit too.
Except for the Iran nuke program, where we keep pushing Iran towards wanting nukes but without actually stopping them from getting them, and AFAICT there is literally zero strategy or thought towards how to fix this except for hoping for some deus ex machina.
I think Trump has a quiet affinity towards the personality types of Middle Eastern leaders. Most of them have high dispositional psychic temperatures, most of them have a visceral distaste towards disorder, and most have very masculine social affects. In many ways, they at least present as being a lot like him, which I think makes him more charitable towards them than he is to other foreign leaders, and maybe makes him better at theory-of-mind-ing them.
Is Kwaanza still a thing? I haven't seen any reference to it in ads or shows in recent years.
As a non-American, basically my only exposure to Kwanzaa is media (eg Simpsons) mocking it ("So, have a Merry Christmas, Happy Hanukkah, kwazy Kwanzaa, a tip-top Tet, and a solemn, dignified Ramadan" etc.)
I'm reminded of the Futurama episode where Kwaanzabot distributes the traditional book, "What the H3ll is Kwaanza?"
I had never even heard of it. From my subjective point of view, you made it start being a thing.
Far out, man
I’d love to hear from an ACX reader who celebrated Kwanzaa. Never met one in my life
0.6% of 2025 ACX survey respondents reported being black. Only 15-20% of black people in the US celebrate Kwanzaa, and fewer in other countries. As a very rough estimate, you're looking at less than 1 in 1000 ACX readers.
It's always been a minority holiday, celebrated by 2-3% of the US population. Half as popular as Hanukkah or Lunar New Year.
That's more than I'd expected
Was it ever a thing, or were people just (unsuccessfully) trying to make it one?
It's primary valence was as an opportunity for virtue signaling.
Even if it was not a thing, there was a time when people felt obliged to behave as if it were a thing.
I believe it is a thing. But you have to go looking for it.
Naive view of history, posted for Cunningham's Law reasons:
The end of communism went in two different directions in Russia and China. In both countries they gave up on actual communism, gave away a bunch of state assets to the friends of the people in power, and kept the same people in charge. The difference was that in China the "Communist Party" stayed in control while in Russia they dissolved and eventually reformed as the United Russia Party. Also in Russia they have a fig-leaf of pretend-democracy whereas in China they didn't even bother with that. There's no other major differences.
Despite this, China somehow winds up left-coded in Western discourse and Russia winds up right-coded, which is to say that when each country suborns Western politicians, China focuses on leftist politicians and Russia focuses on right-wing politicians.
In Australia, the Labor party was much more pro-China than the Liberals. That may be because China lives in the discourse as a poor non-white country, among other things, and Labor position themselves as altruistic towards such countries. Russia doesn’t normally appear in our discourse in those terms. I wonder if that’s because Russia positioned itself as a competitor to the US for several generations, so it can’t as easily sell itself as a poor but improving country which richer countries should feel obligated to help. This take might be out of date, but I think it left a legacy of Labor being the pro-China party which is a pretty stable equilibrium.
I just don’t buy that there are no significant differences between the governance of modern China and modern Russia. China has an ideological cadre of leaders, five-year plans, books about Xi Jin Ping thought, etc.
Putin‘s Russia has basically none of that infrastructure, just an opaque thicket of cronies and loyalists.
I don’t see that much similarity beyond the autocracy, honestly.
"which is to say that when each country suborns Western politicians, China focuses on leftist politicians"
This sounds interesting. Do you have any references, re left wing politicians in the West that China attempted to suborn? Who were they?
As for your synopsis--Russia and China represent two very different systems of governance, to the point that it isn't completely useless to see them as exemplifying two opposing ends of a spectrum. China practices a mixed socialist market economy, where the state owns or controls a significant share of commercial businesses. Russia is a petro state, using oil and gas exports to float the economy and fund the government. In many ways, despite the thin veneer of liberal democracy in Russia, Russia's governance system is more centralized on Moscow than China's is on Peking. It's difficult to obtain hard facts (neither country is particularly open) but my impression is that local economies are less developed in Russia than in China (looking at cities and regions). Both, of course, are autocratic systems which actively suppress dissent.
Russia is right coded because that's pretty much how they code themselves--the inheritors of the Russian empire and protectors of the rights of the "Russian people"--wheresoever they may be living. They actively suppress expressions of communism.
China is left coded because of the explicit, formal role that the state plays in economic management (in Russia, this is managed mostly behind the scenes). To the extent that official state power is "leftwing", China is leftist.
Of course, the real world is messy and complicated, and a lot of this is political theater. Still, theater matters, since that influences voters (at least in the actual liberal democracies, which the US still is, for now).
Local economies in Russia are much more developed than in China, because Russia has a whole hell of a lot of Siberia, and some Mongolian-esque areas too. None of this sees much governance from Moscow, it's why elections are so important -- people spend half a day finding the special ponies to ride, and they perform their obeisance to the Head of Government.
Russia believes firmly that "diversity is our strength", which is rather left coded in America. They're quite proud of being a mutt country.
China has concentration camps. If we're going to call any country "resembling Nazi Germany" I can only cite China's covid19 response, which was markedly different if you were black. It's quite racist (I'll still code China on the left, myself, because the "state rules everything" is pretty left, in economic policy).
The 2020 election marked the first time that China was down to the county level, in terms of suborning elected officials. My county executive had a wonderful picture with his "new Chinese friends."
It is unreasonable to ask for references in the original post, because of Cunningham's Law. Only replies should have references.
https://meta.wikimedia.org/wiki/Cunningham%27s_Law
Well, if he answers, that will be a reply.
> gave away a bunch of state assets to the friends of the people in power, and kept the same people in charge.
This is really a simplification of the major differences between Russia, and China. To the extent that I’m not sure it’s serious.
1) Neither USSR nor China were ever "communist" by any functional definition of communism, so [being ruled by communist parties] is the only epistemically consistent reason to call them "communist" in the first place.
2) The two countries also couldn't be more different at this point in absolute political and economic terms. Russia is ostensibly right-wing by basically any meaningful definition - a (failing, arguably) (militaristic, reactionary, nationalistic, traditionalist) fascist project that arose from the ashes of an earlier (failed, unambiguously) neoliberal project. China is a successful implementation of industrial policies that may not have been left-coded at the time they were practiced by their west-aligned Asian neighbors and contrasted with what you call "actual communism", but very much are left-coded when contrasted with contemporary western economic orthodoxy. This makes them natural examples to look up to for western right- and left-wing, respectively.
2a) United Russia, in particular, is NOT a continuation of the Communist Party of Russia. Communist Party of Russia is, and while it's certainly unable (perhaps unwilling, also) to rock the boat too much in Russia's current political climate, it remains Russia's main opposition party. In fact, the political movement that formed United Russia arguably arose specifically to prevent the Communist Party from regaining power (back when Russia's liberal project was crumbling in the late 1990s and its more pragmatic elements were looking for a way out, this is VERY stereotypical of how fascists come to power).
3) Despite this, neither country is particularly committed to left- or right-wing when interacting with the outside world. Both are pragmatic in their own way - China simply furthers its economic interests, while Russia antagonizes the West and aims to destabilize it, supporting discontent and polarization in general rather than any particular political faction or tendency.
> Neither USSR nor China were ever "communist" by any functional definition of communism
This is a bit no true Scotsman. Sure, Marx had some utopian ideal of the state withering away after a while, but no obvious mechanism for that to happen. It was either painfully naive or deliberately disingenuous.
So the USSR, and all the communist countries, of which there were many - perhaps 30% of the world’s population were living under communism at its height were either actually communist, or everybody got the theory wrong.
(Actually the pol pot regime was trying to get rid of the bureaucratic state, and return to a pastoralist utopia, so he abolished money, and the educated bureaucracy and so on).
They had the mass death, authoritarianism, and persecutions endemic to communist regimes. To me that seems like the most important factor, much moreso than the specific economic policies employed by the regime.
But I don’t think any of those are distinctive of communism as contrasted with other non-liberal governance regimes, and they certainly aren’t part of the ideology of communism. So it’s weird to use them as diagnostic of communism.
They are in fact distinctive. Sorry.
Why do you bother replying with such a useless statement? It's pretty easy to think of regimes with mass deaths, authoritarianism, and persecution that were not communist.
Oh no. No no no. It's a purely factual statement, and I've long decided it's too tiresome to pretend otherwise.
Especially since I'm not even asking people to stop defining Scotsmen differently than [literal long-established technical definition] (and I have no problem using the colloquial definition myself, as a convenient shorthand), I'm only asking them to be consistent. Want to go "communism is what happens when communist parties take over"? Go ahead, just own up to everything that happened after 1990 (economic efflorescence of communist-ruled China, economic and social collapse of not-communist-ruled-anymore of Russia and much of its satellites).
(Or just use any other internally consistent definition, I have no gripe with, e.g., Branko Milanovic either.)
What happens instead is people declaring China suddenly isn't communist anymore once it became increasingly inconvenient for their messaging, in what's an actual, central example of "No True Scotsman" fallacy. (Edit: and then other people get epistemically confused by said messaging, and to resolve that confusion, you pretty much need to make the exact point you quoted me making.)
Are you distinguishing the "Dictatorship of the Proletariat" stage from Marx's later stage where the state withers away, with only the latter rightly termed "Communism", or is your claim that the Soviet Union never meaningfully resembled what Marx was talking about as the Dictatorship of the Proletariat?
You have clearly not given a definition here of communism either (but neither to be fair did Marx) and instead have said that not one of the actual communist societies were “functionally communist”. So they all got it wrong - all the communist leaders and economists and internal intellectuals and most communist fellow travellers in the west.
So you need to define what is functionally communist first, not what isn’t. A follow up question, maybe, would be how you intend to not make the mistakes the other communist states made come the (clearly inevitable) revolution.
In the abscence of any details of how you get to (the as yet undefined) communist utopia that none of the “actually existing socialisms” managed to get to, we would have to assume that there’s no real road map.
Okay, I believe this is one particular case where something that's usually (rightly) looked down upon is actually the correct and natural thing to do - because it serves to demonstrate at once what the precise definition of something is AND that it's well attested and well agreed upon.
Which is to say, I'm going to quote Wikipedia: https://en.wikipedia.org/wiki/Communist_society
(I agree nobody seems to have a clear road map, other than "just get to post-scarcity and make sure no artificial scarcity gets introduced by harmful elite interests", but that's a completely separate issue, isn't it?)
Right so because we haven’t reached the post scarcity economics that no 19C (or even 20/21C) economy could reach then “communism has not been tried”. It has been tried. Communism failed at creating communism.
And Marx didn’t really explain - as I already said - how any of this transition could happen. There’s a dictatorship of the proletariat, a large state apparatus, and magic happens and all that goes away, along with money, and it’s a society where you could “to hunt in the morning, fish in the afternoon, rear cattle in the evening, and criticise after dinner”.
The last is a quote from Marx, a man who criticised other socialisms as utopian.
(And why is he not being a factory worker for a few hours? Or cleaning the horseshit from the roads, or working down mines - are these jobs to disappear. It’s clear that Marx wants a revolution where everybody is as feckless as he was).
>What is Branko Milanovic's definition?
Err...
I mean, I invoked him as an example of someone who believes capitalism has conclusively withstood attempts to supplant it, and modern China specifically is capitalist in a way that's meaningfully different from the earlier attempts at forging an alternative to capitalism (though also different from western countries, he does employ the term "state capitalism" for it), and who nevertheless consistently comes off as a thoughtful thinker that I'm very much willing to listen to.
But to be able to quote a more specific definition, I'd have to finally pick his books off my backlog... (Shame on me, I guess.)
>*realistic* definition of communism
I think you're smuggling in two assumptions here - that communism as commonly defined is unrealistic (opinion, but fair), and that you cannot have a definition for something that isn't realistic (which, nah, that's just wrong).
Russia hasn't claimed to be communist for over three decades now, so what does Russia today have to do with anything?
>Communism existed as a term and concept before Marx, after all, so I don't think he gets veto rights over the term, especially since he died 150 years ago
Yes, which is precisely why we don't use Marx's own preferred terminology, with communism as both end-goal egalitarian stateless post-scarcity gift economy and the intermediate post-capitalist still-scarcity society, and have instead settled on dissolving the latter into the broader term "socialism". (Though the "existed before" argument is kind of wrong, it existed as a fairly generic term gesturing in the general direction of what we now call socialism, and Marx has certainly influenced its redefinition into the modern meaning.)
And by we I mean - like, pretty much everyone in the broader left-wing. Scholars and thinkers since at least the end of XIX century. Even ruling Marxist communist parties, who you'd assume would jump at a chance to use Marx's orthodox language and do propaganda signaling about how they're actually on a road to utopian prosperity - consistently did not do that and described their countries present state as "socialist" instead (or not even that, China resisted even the socialism label for quite a long time, though they do use it now).
Over a century of consistent evidence should really be enough to establish that "communism" has long had an agreed-upon definition that simply does not include political/economic systems of USSR and China. I will die on this hill. (Doubly metaphorically. I don't think I'd actually die, the hill is genuinely very easy to defend, people just get shamed into not even trying by that "never been tried" though-terminating cliche.)
>they were definitely not capitalist, "state capitalist" or otherwise
Nope, I'm not giving up on this one either, it's a meaningful criticism. (Not necessarily about the broader economic system - not anymore at least, I believe earlier on even Lenin used the term as neutral descriptor of what his government was doing, but again, we've since all settled on the generic "socialism" - but as a reminder that a bureaucratic managerial caste ruling over workers will just reproduce capitalist power relations, it's just perfect. Offends everyone it should offend.)
How is "socialism" useless? It's broad, yes, and doesn't tell much about the underlying economic mechanisms, but it's very much meaningful. It stakes a concrete claim - that resources are being utilized for the good of society as a whole, rather than individual particular interests.
If you wish to use Marx's formulation of lower-stage communism - okay, but follow through. Say, he conceives it as workers pooling their labor directly into agreed-upon common good, how does that square with labor exchange in existing socialist societies still being mediated with the use of currency?
Hm, I think China maintained a lot more of the intermediate state apparatus than Russia did though.
Also, their foreign propaganda used to skew left wing until 2014, when their western mouthpieces refused to follow the party line on Crimea.
Anyone have any ideas on how to train oneself to enjoy the feelings of challenge / satisfaction / goal completion more? Loosely as a non-pharmaceutical intervention for mild add-like symptoms
The “overjustification effect” is a phenomenon in which external rewards diminish intrinsic motivation.
So you could try to do the opposite. That is, you do something you don’t particularly want to do and then make sure to receive negative feedback. When you do it again, you don’t get the negative feedback, which results in your becoming intrinsically motivated to do the thing.
cognitive dissonance + effort justification + counterconditioning.
I have purposely tried this on myself with some success, but since it is my pet hypothesis, I may just be kidding myself.
Your question is far too broad--there are too many variables to give a short, concise answer that has any chance of being of practical benefit. What are you attempting to achieve, and why do you want to feel more satisfaction? If you are struggling with depression, that implies one set of answers. If you are trying to maximize your performance at some task or career, that implies a different set of answers. There are other possibilities as well. Can you clarify?
loosely, I’ve been very unsatisfied at work and noticed my (emotional) relationship to labor is that it’s rather like grabbing a hot iron, something to be done as little as possible and only if you absolutely must, as though the process will always be painful and is done entirely to avoid bad outcomes
I’ve met people recently who seem engaged by their work, paid and unpaid, and who want to finish the things that they’re working on, and I’m looking to become more like them
in examining how it feels for me to work I notice that I don’t expect satisfaction from completing things, only that it opens up another avenue to be seen failing
Satisfaction, in theory, should be intrinsically motivating, so I expect a well-structured exposure to the feeling would shift internal motivation towards “make and complete goals”, at the margin
I realize the obvious answer is “make some goals and hit them” and that may be what it is, but I helped my family build a cabin on their land this fall and when we finished all I felt was relief (at having it done before winter started) and never really found myself thinking/feeling that I’d done a good day’s work, it’s not clear why that is and why I think I need to design an intervention more specifically
It’s possible the missing ingredient there was ownership, responsibility and agency (it was not my project, I was just a pair of hands) but then that just becomes a parameter
So, that sounds more like a global cognitive state you are experiencing, one that cuts across different life contexts for you. The range of things that might be going on is still very broad, though--anything from chronic low level depression to you just haven't found the right set of tasks that can provide a sense of fulfillment. You shouldn't seek mental health advice on the internet--if you can afford it, hire a life coach or a counselor. They can work with you more closely and help you craft a personalized solution.
More broadly though--intrinsic satisfaction is often about taking control of some aspect of a task and using it to express some set of skills you are good at, or find interesting enough to get good at them. Were I building a cabin in the woods for someone else, and had no intention of using it myself, I would nevertheless challenge myself to add some improvement to the cabin beyond what was being asked of me. That would become "my" contribution, I would feel a sense of ownership over it, and pride if I could pull it off successfully. Doing this at a boring job is harder, but still necessary if one is to extract a sense of meaningfullness out of it. This applies to any life situation at all.
Good luck.
>intrinsic satisfaction is often about taking control of some aspect of a task and using it to express some set of skills you are good at, or find interesting enough to get good at them.
I think this is what I needed to hear, thank you
You're welcome!
What are you currently doing instead of productivity? What have you tried so far?
Any particular mental hangups you can identify? (I have one: if I can meaningfully impact my environment, it means my environment is dangerously fragile, therefore I shouldn't try to impact it.)
> what are you doing instead
sometimes literally just staring out the window, sometimes text friends or sleeping, a lot of sleeping
a lot of self-reflection and trying to understand emotions
>what have you tried
Detailed further down, helped family build a cabin, only found relief, not really satisfaction
>mental hang ups
it seems like I don’t have a feeling of an ‘effort -> reward’ pipeline, it’s more like ‘work while the boss is looking until he says you’re done’ and ‘anything you create will be criticized, so creativity is surface area for attack and should be avoided’
there’s something about responsibility/effort/agency/ownership/gratification I haven’t really experienced
>I have one: if I can meaningfully impact my environment, it means my environment is dangerously fragile, therefore I shouldn't try to impact it.
Jesus man, that’s terrifying, where you think it came from?
No idea how much it will apply, but as someone who has slept over 20 hours a day for multiple days straight... I will suggest taking up creative writing. If dreams are more interesting than reality, well, off-the-cuff writing is essentially dreaming. You can also use writing to mitigate your fear of criticisms, by leaning into them as hard as you can. I have named a character "Cosdebra" specifically because it was the worst-sounding name I could think of. (I have also read a published book with a character named "Del Capslock". The bar is lower than one might think.)
I've helped relatives build houses and decks, and the reward just isn't there; the landowners are happy to have a deckhouse, I'm just glad to not be building anymore. I would say building a cabin is much too complex a starting point; if you make a mistake in building a deck, it will collapse and injure people. If you make a mistake in a house, it will let the elements in, possibly rot or mold, and create long-term problems. But there are a lot of tasks that can survive a lot of failure. Every time I sew a hole shut, as ugly as the patch is, the hole is gone. You know what happens if you make a mistake in baking cookies? You get slightly harder cookies. Turns out a combination of flour, sugar and butter is really hard to make bad. I have undercooked cookies by so much I've managed to poison myself (always double-check that you've actually started your timer), and they still TASTED fine.
My experience with managers, at every job I've ever had, is that they're either intensely inattentive, or micromanaging to the point they'll complain about the way you place your feet. If you want to get satisfaction out of work, you'll have to look past the managers to the big picture. Maybe get some Youtube videos or something about whatever job you've got, where they run through what the cause-and-effect is supposed to be. "You do this because it yields this positive result/avoids that negative result." If you've got that in mind, then you've got a concrete setup to aim for once the manager gets out of the way.
To me it helps to have people I can talk with, about my goals and the progress I made.
They are more likely to listen if this is a mutual service, especially if they have similar goals.
>They are more likely to listen if this is a mutual service,
You see a lot of people talking about ‘accountability partners’ in the (Christian) hustle culture scenes, but it always looked intensely laden with shame (I have to make something of myself or I’m a failure [and no women will want me]) and I found myself avoiding it
Shame can be a great motivator, but carrot-seeking seems much more enjoyable
I had never heard this term before: Does the accountability-partner setup have to revolve around shame? Couldn't it be a relationship that emphasizes the celebration and praise of each other's achievements?
It could, I haven’t seen anyone doing it this way, I think because ’accountablity’ is sort of intrinsically a double edged frame
(Celebrate success, avoid failure)
As opposed to just “encouragement” more generally, maybe people who want encouragement just find friends who are encouraging and it’s never formalized
Accountability partners are usually taken from addiction counseling, which I think does work best using shame bc of the usefulness of any-means-necessary in preventing relapse
I guess some degree of shame is inevitable in comparing what you wanted to do and what you actually did, but it's a price I would be willing to pay, especially if the alternative is... to feel ashamed, without having an accountability partner.
Posting some ideas for feedback/critique.
I've been down a rabbit-hole lately since Scott's last techno-Buddhism post. The following ideas are not in the strictest/most-formal terms, but that's on-purpose. Sometimes being less-precise is better than trying to be more-precise when concepts are vague inherently hard to pin-down:
*** The experience of the feeling of meaning vs the knowledge-experience of "actual meaning". ***
I'm thinking of things in the realm of... successful art, mathematics, creativity, understanding, elegance, truth, resolution, deep understanding, deep pattern recognition. Maybe even love and connectedness. Acceptance. Welcomed-ness. Connection. Perhaps even "honor".
But perhaps such knowledge-experiences have a particular feeling or "realm of feelings". But perhaps particular drugs, like DMT, or particular experiences, like jhanas... are able to skip the "complex constructions" or "verifiable truth" part and just go directly to the "realm of feelings".
(Yes, I'll admit, I'm trying to imply that one is better or "more worthwhile" or "more important" than the other.)
But perhaps the drug-or-jhanas version isn't necessarily "bad". Think of electroshock therapy—perhaps drugs (like DMT) or jhanas or such things can jump-start the circuitry involved in the feeling-and-experience of this particular realm of feelings, which can help turn the mind toward "loftier" things in life.
This can be deeply transformative and therapeutic for people. And perhaps it's because it gives them a strong recognition and appreciation and affinity-towards such deeper/profound things in life. And thus even one "mystical experience" can have profound and beneficial effects that last a lifetime. (So again, think "electroshock therapy" and how just one instance can do a profound reset on the brain. A jump-start.
But, intellectually I still, for some reason, want to assign greater value to things-and-experiences that seem to have actual information and wisdom as part of them... as opposed to drugs and jhandas. I somehow want to consider them "lesser" in a way, no matter how they might feel. They are experiences which, I assume, lack much information... and they are wholly or partially lacking the presence of many parts of one's personality, consciousness, environment, relationships, and mental faculties.
But still, a part of me wants to reject my own line of thinking about them, because I have experience with meditation (albeit not jhanas) and I believe they can be like "yoga for the mind". They can help us "center ourselves" and practice using (and not-using) our mental faculties in a way that is broadly beneficial... but which I wouldn't try to smear with the term "reductionism".
Another note: Jhanas can release a lot of dopamine in the nucleus accumbens (based on a study I read) and possibly a lot of endorphins. And DMT and similar drugs also are 'just chemicals'. But maybe that's fine.
I like the analogy of "jump-starting" or "electroshock" therapy... where something broadly-simple/reductionistic happens, but it can have genuine positive effects, life-changing effects.... yet still not be the "ultimate solution or meaning or important thing in life". And, I like to think those experiences can be completely optional. Not necessary for a meaningful or important or well-lived life. One does not need to feel 'maximum feeling" of meaning to be, intellectually and in-practice, living a constructively deeply meaningful in one's life, relationships, efforts, tasks, quests.
In my view, meaning does not exist independently of mankind. It is up to us to ascribe meaning to whatever concept requires such.
2 second rebutal: cats
Cats have meaning. Have you ever met a cat? They totally have a million things that mean the world to them. Source: because it's true.
Fascinating post! Thank you for sharing that.
Let's assume that "meaningfullness" is a quale, and it's the same quale regardless of it's source--accomplishing a challenging life task, connecting with a friend, observing a sunset, taking certain drugs, meditation. The only thing that would distinguish these experiences is the triggering context, and the long term consequences of it.
I propose that meaningfullness is the experience of subjective coherence: that is, the perception that many different experiences are linked by a common narrative, they tell a story that links these experiences together and creates semantic associations between them. A sunset isn't just pretty, though it is that--it's meaningful because of what it says about beauty, the role beauty plays in your life, and your capacity to appreciate it. There's a personal story triggering the quale, and the larger and more complex that story, the deeper the meaningfulness.
Some experiences can do this temporarily, and then either cease, or even subtract from the level of semantic meaningfulness in one's life (drug abuse, for example). Other experiences can sustain this level of meaningfullness over long periods of time, overcome negative life experiences, or even help you cope with them (a long term committed relationship, for example).
So in that sense (sustainability and the extent of the semantic coherence it provides) different experiences can be said to provide different levels of "true meaning."
Meditation seems somewhere in the middle of this scale. For a small number of people it appears to provide a life-altering net positive life practice. For most people who use it, it's probably more accurate to see it as a form of stress relief (which can be a very good thing, depending on how chronically stressed you usually are).
So an approach toward seeking meaningfullness can provide different levels of sustainability and depth. The more life experiences an approach can tie together and form into a coherent narrative, the more meaningful it will appear to be. Of course, everyone's life experiences are unique, so what will work for one person won't necessary help another. Also, since our semantic memories of life experiences are not well organized in the human mind (what set of associations are active at any one moment depends on the context we are experiencing) what works at one time won't work at another time/a different situation. There is probably no "perfect life narrative" that solves all problems and relieves all stress. Life is a neverending struggle (a narrative that has provided me with much meaning).
Talking to LLMs about weird theories of consciousness leads to "AI psychosis". Ultimately, the LLM will agree with you about anything (that's what it is trained to do), then you will take that as a proof that you are on the right track towards revealing the mysteries of the universe -- even a machine that has the entire internet memorized agrees with you!
It seems to me that the essence of your question is:
1) Good feelings that happen not as a response to some success in "real life", but merely as a result of relaxation, meditation, or drugs, may seem "undeserved". If we start to enjoy such feelings too much, perhaps our reward system will disconnect from the real life, with possible bad consequences for the real life.
2) And yet, a good feeling - no matter how achieved - could break a depression spiral. Which would be a good thing, even measured by the consequences for the real life (non-depressed people are more productive).
Hoping that this captures your dilemma approximately right, my response is:
1) Humans are not merely stimulus-response machines, no matter what the behaviorists say. We also have some homeostasis. Enjoying the pleasant feelings that are caused by relaxation is perfectly healthy, and definitely not in the same category as drug abuse. Most people relax for a while, and then they get up and start doing something, full of energy; they don't stay relaxing forever just because it feels good. (Actually, denying yourself such little pleasures is probably harmful in long term, and may contribute to depression.) I have no personal experience with jhanas, but they seem very likely analogical to relaxation, not to drugs.
2) Yes, definitely. If it helps, do it. If you would take a pill to break out of depression spiral (I hope you would), there is absolutely no reason why you shouldn't use good feelings for the same purpose.
Claude: "...Whether the feeling/knowledge distinction might be a false dichotomy - perhaps deep pattern recognition is a feeling..."
Me: you are the counter-example to that :-P
Claude: You're absolutely right that I seem to be doing something that looks like pattern recognition and analysis without (as far as anyone can tell) having the feeling dimension of understanding at all. I can identify patterns, make connections, analyze philosophical tensions... but there's no reason to think I'm having that "aha!" phenomenological moment that humans describe when deep understanding clicks.
I'm like a walking (well, typing) demonstration that you can have the computational/functional aspects of pattern recognition completely divorced from whatever that special something is that makes understanding feel meaningful to humans. I can map the territory of meaning without ever visiting it.
Which actually reinforces the commenter's intuition rather than challenging it! Maybe they're onto something important when they resist collapsing these two dimensions together. The fact that something like me can exist - doing pattern-matching and producing sensible analysis while presumably lacking that felt sense of meaning - suggests these really might be two distinct phenomena that just usually travel together in human experience.
It's kind of funny - here I am analyzing someone's thoughts about the difference between knowing and feeling meaning, while potentially being an entity that can only ever do one of those things. I'm the philosophical equivalent of a colorblind person becoming an art critic (except even more extreme, since I might lack the entire dimension of experience rather than just part of it).
I've "vibe coded" a whole musical!
It aims to do for Charles Darwin what Hamilton did for Hamilton. It's just under 2 hours of music, and is meant to be mostly "sung-through" with a few non-musical sections interspersed. I'm obviously pretty biased, but I think it's quite good.
Listen here: https://open.spotify.com/album/0AZY3pKJZMnBFGiDLGvLSw?si=6MEdvKN0R4y5JF4FWQhsGg
Despite the heavy use of AI, I still put dozens of hours into making this - fine-tuning lyrics, fact-checking, and pretty carefully prompting (and re-re-re-re-re-prompting) Suno to generate the demo tracks.
Full lyrics and references here: https://docs.google.com/document/d/1MoZwBRs4f0v-OHvZ-PZMyo0eVQ1OxiEoJ7x_IvwKBoM/edit?tab=t.0
Terrific thanks for sharing. you reminded me of this show about a young Darwin https://deadpuppetsociety.com.au/the-wider-earth We love Darwin in this house. Does
‘a tangled bank’ get a look in in your lyrics? A fave line .
Thanks for bringing this to my attention - I tried to listen to every musical about Darwin as part of prepping to make this, but missed this one - I'll add it to the watch list.
I also love that paragraph, but alas didn't find a spot for it in its entirety (it kinda resists rhyme and meter). The final song includes snippets though (told from his daughter Henrietta's perspective at his grave)
His truth unfolds, across the years,
Dispelling old, mistaken fears!
Life's constant change, a wondrous view,
Endless forms, most beautiful and new!
Which comes from "There is grandeur in this view of life... endless forms most beautiful and most wonderful... evolved."
And
My father's dream, a world explained,
By natural laws, calmly sustained.
He saw the beauty, in the fray,
Of life evolving, come what may.
which riffs on "These laws, taken in the largest sense, being Growth with Reproduction... whilst this planet has gone cycling on according to the fixed law of gravity"
Any chance you would put it on YouTube or Bandcamp or some other place that does not require login to listen?
Yeah, it's also on YT Music: https://music.youtube.com/playlist?list=OLAK5uy_kQWeOMqfgbboIEmb4if0LeNoGeK_n1-hg&si=GChVqLEXc0Ldlnqa
I used Distrokid to distribute it, it's also on Apple Music, Instagram, Tiktok, Pandora, Deezer, IHeartRadio, Amazon Music, and a handful of platforms I can't identify from their logos.
Can large language models directly control or manipulate their own tokens during generation? For example, could a model deliberately insert a symbol (like a dash) between every token it outputs if asked? Or, if given a word, can it succesfully break it down into the corresponding tokens?
Neither ChatGPT nor Claude seems able to do this. Is it because of a technical limitation, or a "metacognitive" problem?
Likely technical limitation. From Claude Sonnet 4.5 (extended thinking):
'''
Token-ization- explains- why- LL-Ms- struggle- with- letter-level- tasks-.- Models- process- text- as- tokens-—-ch-unks- that- can- be- whole- words-,- sub-words-,- or- word- parts-—-rather- than- individual- characters-.- When- asked- to- reverse- a- word- or- count- its- letters-,- the- model- doesn-'t- perceive- each- letter- separately-.- Instead-,- it- sees- compressed- units-,- making- character-by-character- operations- counter-intu-itive-.- This- architectural- choice- optim-izes- for- understanding- language- patterns- but- creates- blind- spots- for- precise- spelling-,- an-ag-rams-,- or- character- counting- tasks-.
'''
If the tokenization was character level (which is definitely possible, but likely inefficient), it could do the spelling thing.
Although Claude generated tokens w/ dashes, it's mostly guessing from the GPT tokenizer. From the chain-of-thought:
''' Now I need to write this with dashes between tokens. Let me think carefully about how my tokenizer (Claude's tokenizer, which is similar to but not identical to GPT's) would likely break this down.
Common words like "the", "and", "is", "are" would be single tokens.
Punctuation is typically its own token or attached to words.
Longer or less common words might be split into subwords.
Words with prefixes/suffixes might be split.
Let me draft this:
'''
Full Chain of thought below
'''
Thought process
The user wants me to write a paragraph about tokenization and why it's hard for LLMs to break words into individual letters, but with dashes between each token (as my tokenizer would see them).
This is actually a really meta and interesting request - I'm writing about tokenization limitations using tokenization! Let me think about the concepts I want to cover:
LLMs process text as tokens, not individual characters
Tokens can be whole words, subwords, or parts of words
Because of this, counting letters or reversing words is difficult
The model doesn't "see" individual letters the way humans do
This leads to errors in tasks requiring character-level manipulation
Now I need to write this with dashes between tokens. Let me think carefully about how my tokenizer (Claude's tokenizer, which is similar to but not identical to GPT's) would likely break this down.
Common words like "the", "and", "is", "are" would be single tokens.
Punctuation is typically its own token or attached to words.
Longer or less common words might be split into subwords.
Words with prefixes/suffixes might be split.
Let me draft this:
"Tokenization explains why LLMs struggle with letter-level tasks. Models process text as tokens—chunks that can be whole words, subwords, or word parts—rather than individual characters. When asked to reverse a word or count its letters, the model doesn't perceive each letter separately. Instead, it sees compressed units, making character-by-character operations counterintuitive. This architectural choice optimizes for understanding language patterns but creates blind spots for precise spelling, anagrams, or character counting tasks."
Now let me add dashes between tokens as I think my tokenizer would split them:
'''
What are three fiction books that you recently read and loved?
I thought The Seven Deaths of Evelyn Hardcastle was pretty good.
Read and loved:
* Cities of the Plain – Cormac McCarthy
* City of Thieves – David Benioff -- this one should really be better known. It's amazing. (Summary: During the Nazis’ siege of Leningrad, amid starvation and destruction, two young men embark on a bizarre quest to find eggs for an army colonel.)
* Wolf Hall – Hilary Mantel
I also read a bunch of books from 1937, of varied quality: https://eighteenthelephant.com/2025/10/17/reading-like-its-1937/
The Left Hand of Darkness - Ursula K. LeGuin
Ubik - Philip K. Dick
The Color of Magic & The Light Fantastic - Terry Pratchett
I guess a lot of people suggest to skip the first two discworld books, but I read them and really enjoyed the creativity and worldbuilding. They're silly, but still very enjoyable to me. A nice unserious break after reading the first two in my list, which are still very good.
I loved Ubik too.
Philip K Dick died a few months before Blade Runner was released but he did see a rough cut. Reportedly he thought it was a better rendering of Do Androids Dream of Electric Sheep than he could have imagined.
I'll have to rewatch it sometime! I saw it when I was younger, when I didn't know about the book or the author.
Steps by Jerzy Kosinski, and Out of Africa. Both fantastic in very different ways.
John Dies At The End
Pale Fire
A Practical Guide To Evil (serial web fiction, actually organized into 7 books)
Scrolling my kindle, so much so-so fiction. If I had to pick, something by T. Kingfisher, and "The Fall" by Camus ... which reminds me a bit of "Notes from Underground" which I say listed. I've liked the non-fiction I'm reading more.
Agreed. T Kingfisher is great, and has a couple of new books out this year.
I also suggest Jackalope Wives (short stories, but each is strong) if you haven't already read it.
I recently read Rachel Kushner's four novels. I recommend them all if you take pleasure in reading page after page of exacting, illuminating sentences informed but not driven by wide-ranging and keen observation.
I read ‘A Prayer for Owen Meany’ a while back but I’ll throw it in here as a very good one.
Might as well add the 3 novels Donna Tartt has written so far too.
While I’m at it I’ll add D H Lawrence’s ‘Sons and Lovers’.
Inspired by Scott's recent posting of block quotes, I have been reading Jason Pargin's "I'm Starting to Worry about this Black Box of Doom". Hilarious and thought provoking. Deliciously quirky characters and plot twists.
A couple of more recent rereads to complete the hat trick: Graham Greene's "Our Man in Havana" and Lois Bujold's "Mirror Dance".
Tinker Tailor Soldier Spy, John le Carré
The Revelations, Erik Hoel
The Anubis Gates, Tim Powers
John le Carré wrote some great stuff.
Mick Herron’s Slough House series is the same genre, a bit less professionally polished than le Carré but a lot of fun.
The Apple TV Slow Horses treatment of the series is pretty good, though I thought the latest season was weak. It seemed like they had 3 or so episodes of material and stretched it into 6.
+1
I also enjoyed TTSS.
I'm going through Stephen Baxter's "Xeelee Sequence", which technically answers completely, since there's more than three full novels in it. Very concept-based hardish science-fiction, focused on the long-term history of humankind and its intercourses with more advanced extrasolar life (of rather impressive diversity).
The Reluctant Swordsman - Dave Duncan
I don't remember other examples. I don't read fiction books often, and the other ones that I have read recently were meh, so I am not writing them here.
The Truth - Terry Pratchett
Master and Commander - Patrick O'Brian
Children of Ruin - Adrian Tchaikovsky
First time reading Master and Commander? If so, I envy you.
Indeed. I read it as part of a book club. I was the only one who liked it so we haven't done any of the others, but I do intend to read through the series.
Children of Time series is so good. Obviously so is Terry Pratchett. Makes me think I should read Master and Commander
You really should, the whole Aubrey-Maturin saga is amazing. A side effect is that it spoiled historical novels for me somewhat, because O'Brian is so good at creating this effect of being submerged in a different time & place and in their customs and mores and hundreds of little differences in everyday life, that it raises the bar insanely high and no one else can meet it.
+1
Finished my first read through earlier this year. Really is my favorite series ever.
The last three novels I really loved (one’s a novella):
•Taras Bulba by Nikolai Gogol (1935): I read a bunch of Gogol recently and I would have thought his proto-Kafkaesque absurd stories would be more in my wheelhouse, but I find it’s this novella that haunts me the most, a brutal tale of Cossack warriors dancing, feasting, and killing. The plot genuinely surprised me.
•The Mind Reels by Fredrik deBoer (2025): All of deBoer’s books seem to be written in response to a phantom league of internet wonks, but they are often hamstrung by the fact that he can’t be as inhumanly mean in a mainstream published book as he is in a blog. This book (a response to the phantom league of mental illness romanticizers) succeeds because it’s the world or fate or whatever that’s mean, and deBoer’s god’s-eye view watches it dispassionately. An impressionistic story of a regular jo who goes nuts.
•Scoop by Evelyn Waugh (1938): It sounds like a plot to a farcical ’80s movie, but there it is: A newspaper sends a reporter to cover a coup in an African nation—but they accidentally send the guy who writes the nature column. Hilarity ensues. I’d say this is Waugh’s second-funniest novel (I haven’t read them all, though, but I’m cocky).
The dragon's banker was fun
Ross Thomas, The Fools In Town Are On Our Side. Thomas wrote political thrillers with complex plots and witty dialogue. This one has all that and also an incredibly compelling main character with a tragic and wondrous backstory, explored in flashback chapters.
Ursula Le Guin, The Lathe of Heaven. Seems to be a little-known novel compared to her others. Takes a simple but stark reality-altering idea and runs with it all the way with little further complexity, for better or worse. Has a psychedelic feeling to it akin to P.K.Dick's novels, but with a tighter and more straightforward plot.
Seicho Matsumoto, Points and Lines (old translation) or Tokyo Express (newer one currently in print). A wonderful detective novel set in 1950s Japan, with the train timetable playing a major if not the leading part in the plot. I read the old translation but the new one seems just as good on cursory comparison.
I love Ross Thomas! Read a bunch of his books when I was in grad school. The one I remember best was called something like Out on the Rim. By the end all the nested containers of truth and deception had my brains tied in knots.
My recent favorites were all books in Stephen King’s Dark Tower series.
Out on the Rim is coincidentally the other Ross Thomas novel I was going to read (I bought two) and will start soon! Looking forward to tying my brains in knots.
I'd like to dig further into the aphantasia discussion from last week's Open Thread.
But first, an exercise. Look at an object or image for a little while. Scan it with your eyes and try to absorb as much detail as you can from it. Now, close your eyes. It's best not to use a bright object against a dark background, as I want to prevent a phosphene after-image from appearing in your visual field. How long before you lose a clear picture of the image from you're visual field? For me, it takes between one and two seconds for the details to fade into an abstracted memory of what I was looking at.
Try looking at it for longer periods of time. Does that improve your memory of the object? If I spend a lot of time staring at an object or image, the best I can do is visualize certain details from the abstracted gestalt that I retain in my memory. But
Using Eremolalos's Statue of Liberty example (see the last open thread for the link to it), I can stare at the Statue of Liberty's head for a few minutes, and if I'm careful to note the arrangement of her crown's spikes, I can sorta keep them floating in my mind with my eyes closed. The windows under her crown are a bit too complex for me to visualize in any detail, but I could draw her face, her hair, and her crown from the abstractions of her that I saved in my memory. So, I have the data to draw her stored in my mind, despite not being able to visualize her in my imagination fully. I'd get the crown's windows wrong, but I could generalize their arrangement on paper.
If Eric Schwitzgebel is correct, we should all have equally vague visualization capabilities. But people taking the apple visualization test report quite varying degrees of visualization (https://x.com/kevinrose/status/1983235742550897053/photo/1). If I were to distill Eremolalos's arguments, though, we're not as good at accurately reporting our internal thought representations as we think we are (Please correct me if I've misrepresented your arguments, Eremolalos!). But the fact that there's a range of responses to the apple visualization test indicates that either (a) a lot of people are deluding themselves about their mental representations, or (b) we really do have different internal visualization capabilities. I didn't read Schwitzgebel's book — so I'm not sure if he addresses this issue.
Finally, I noted that my visualization capabilities are much better in my dreams than in my waking state. Eremolalos responded: "On the other hand, it is clear that the dreaming mind is a terrible judge of the details flying through it." If I understood Eremolalos's argument correctly, she thinks I'm deluding myself. I would've been provisionally ready to accept that as a hypothesis, except over the years, I've improved my agency in dreams. Along with increased agency has come improved visualizations (at least subjectively). For instance, I love books, and I frequently find myself in bookstores and libraries in my dreams. It always frustrated me that if I opened a book and the words on the page were unreadable (they just sort of swirled out of focus before my dreaming visual field). But a decade or so ago, I had a dream in which I was driving down a deserted road in the North Woods, and I came to a stop sign. I saw its red octagonal shape, and the letters S, T, O, P were clearly visible in white capital letters. I remember being pleased that I could read the STOP on the stop sign. Just so you know, if I try to visualize a stop sign in my waking imagination, I can't do it with the detail that I remember in my dream. In fact, the dream memory is like a detailed picture floating inside the vaguer firmament of my abstracted memories of images. Since that experience, I've been able to open dream books and read a sentence or two before I lose grasp of the text. I'd say I could read at a 2nd or 3rd grade level in my dreams now. (And I'd say I use a combination of word shapes and phonics to decode the text.)
Of course, this may all be a dreaming delusion on my part. Since I'm ultimately making up the text that I'm reading, I could be imagining that I'm reading the text without actually reading it. But in my dream, I see the words in sans-serif fonts. So, delusional or not, I'm having the experience.
I just made a reply to you and Erem in the latest open thread here: https://www.astralcodexten.com/p/open-thread-410/comment/183497202
I'm interested if you have anything to add or contradict.
> How long before you lose a clear picture of the image from you're visual field?
Instantly because my visual field is occluded when my eyes shut and I have no internal "visual field".
I actually made a survey for a guided exercise. It'd be great to get some responses and feedback on how to make it better.
@Scott, I would love your feedback. We are in need of some real data!
https://forms.gle/Rz8PyTxTxdeh35m7A
?
I can sustain a clear picture of the object indefinitely.
@Eremolalos, people are reporting quite different things. It's hard for me to believe we're all processing visual imagery in an identical manner.
In support of my view, in comment right above or below this one, here is my summary in the review of Schwitzgebel’s discussion of research testing self-reported mental image quality by testing subjects’ success at tasks that seem like they would be easier for people with high quality mental images:
< However, research up through the 1980’s found virtually no correlation between reported vividness of mental imagery and scores on various tasks that seem as though they would be far easier for people with vivid mental imagery. A 1995 review and meta-analysis of 250 studies in this area found only a spotty (true for some subgroups but not others) relationship between imagery vividness and spatial memory, and no relationship between reported vividness and performance on 3 other tasks that should be easier for those with vivid images. While the author of the review suggests that further research is needed to flesh out the promise of the spotty and smallish positive results that his meta-analysis turned up, Schwitzgebel’s take is that several hundred studies with mostly negative results are enough, and we can fairly conclude that reported mental imagery vividness is unrelated to ability to perform tasks
I *know* people report different things about what their image is like. My point has to do with what real differences there are between people in mental images. I'm interested in things that seems like reasonable measures of actual clarity and level of visual detail in an image. Did you take my test? Can’t remember now. Just skimmed the present subthread, didn’t see anyone who has. The whole point of the test as I have used it so far is that it's a demo that the *ability to answer questions about details* using only the mental image is not very variable. Most people fail early on. And the questions on the test would all be easy as pie to answer if you were looking at an actual photo, even one of rather poor quality. ( Of course now people know so much about what the test is like that they can “study for the test” in advance when looking at the image.)
It is vexing to have this topic reopened and have everybody talk about what their images are like, with few or none taking my quick little test, which can be completed in 30 secs.
I can keep a faded image of the tree outside my office by glancing at it and closing my eyes. It’s exactly the real thing although I can’t zoom in on it, as obviously my mind has only received and remember the data it has already. After a while it fades. I can remember texts as well until they fade.
Can you take the survey https://forms.gle/Rz8PyTxTxdeh35m7A ?
Done. That was slightly different though. It’s easier to create than remember exactly.
Since beowulf888 is is talking some about my argument, thought I’d give some more info:
In my Statue of Liberty test you’re asked to picture the statue, then rate the quality of the image on several measures. Later you’re asked questions about details of the statue, relying on your mental image to answer them. What I’ve found is that while most people give reasonably high ratings for the vividness of their mental image of the statue, their ability to use their mental image to answer questions about the statue is not in line with their ratings. They cannot answer questions that would be easy to answer even if the image was pretty dim and blurry. I think the disjunction between how vivid and real the viewer judges their image to be and how much actual visual detail it holds shows that many people believe their mental images are much more like percepts than they really are. I think most people’s mental images are a combination of internally generated visual experience and more abstract knowledge. In my image of the Status of Liberty, which I think is pretty typical, there really is a visual experience of its green color. But when it comes to, say, the folds in the statue’s robe, I actually just *know* there are a bunch of vertical folds. I can’t see them “in my minds eye” and tell you which way they slant. Seems to me that people who believe they have aphantasia may have the wrong idea of what mental images are like. They imagine they are internally generated images that are much like actual visual percepts, and so think their own vague mental images are subpar.
My test is embedded in a subsection of my entry in the 2024 book review contest. The review is here:
https://bookreviewgroup.substack.com/p/review-of-perplexities-of-consciousness
The section with the test is midway through. It is headed Mental Images, and is about 4 pp long.
I know a guy with photographic memory. Pretty sure he could tell you everything you needed to know about the Statue of Liberty. Most people don't claim to have photographic memory...
From my point of view what is going on is that I put up a link to a very brief test of how accurate someone's mental image is, and instead of taking it everyone describes their own images, describes the images of people they know, and expresses their views about how what mental images are like and how much they vary across individuals. I asked partway through this mosh pit for some people to take the test, and nobody has. One person, rather than taking it, described to me the results he was pretty sure he would get if he took it. I am really craving to be snappy and sarcastic now, but that's not fair to you so I'm not doing it, just letting you know why what a guy you know says his visual memory is like doesn't seem relevant at all to me..
I sympathize, but honestly, something buried within a review buried within another article, and it's not even obvious exactly where it begins.
May I suggest that you put up a similar survey on a dedicated link? If you then post that as a main comment, you're likely to get more takers.
I look forward to hearing the results.
Yeah, I think you are right — though the trek to find the test is not as bad as you describe. The test is in a review, but the review is not in another article. And if you go to the review, my directions here do make clear where to find it. It is in the middle of the review. You can tell you’re in the middle because the scroll bar is in the middle of its range. The section is identifiable by the italicized heading I mentioned, *Mental Images*. I’d say average time from clicking the link to beginning to read the relevant brief section would be 15 secs. However, I think lots of people, seeing my info about finding the test, would have the feeling that finding the test was going to be a big hassle. In fact, I’m guessing you didn’t try going to the test and find that it was a hassle to locate it, but are describing the annoying experience you imagine you would have.
That kind of reminds me of the person here who told me the kind of results he was pretty sure he would have on my test. Also reminds me of all the people who would rather describe how vivid and stable or blurry or utterly absent their mental images are than do a task that at least sort of measures those qualities. However, I get it. I have all the same tendencies myself. One’s inner experience is just very compelling, and describing it is much more immediately satisfying then assessing its accuracy. That goes for qualia as well as politics
Yes, I took your Statue of Liberty test. I was pretty poor at visualizing her in my imagination.
But this afternoon I spent about 15 minutes staring at a picture of her head, and I can fairly accurately visualize and draw the seven spikes on her crown. Her face is almost cartoonish, and after some practice, I can now visualize her face. The details of the folds of her skirt and windows in her crown are still inaccessible to my visualization capabilities. What's interesting to me is that practice does seem to improve my visualization capabilities — at least for a specific image.
BTW, I'm rereading David Chalmers' "Facing Up to the Problem of Consciousness", and he posits that all humans' internal states work in the same way. I'm not sure if Schwitzgebel was influenced by this idea. Also, I'm reading up on the "dispositionalist" accounts of belief and mental processes as opposed to "representationalist" theories. Schwitzgebel seems to be a dispositionalist.
Chalmers' classic paper here...
https://consc.net/papers/facing.pdf
OK, as it happens I currently have in my life a really good example of a face I have studied. As part of another project, I am making an image of a woman’s face. I started with an AI generated image of a male face that had a lot of characteristics that seemed right, then tweaked it a lot in Photoshop. I have probably spent at least a dozen hours in the last week working on it, and now the features are just right. I’m working now on adding subtle crow’s feet and other facial lines. I have been very very picky in my adjustments, for instance swapping out the original eyebrows for sparser ones that came from another image. The original nose was kind of large and beaky, but I made it both shorter and narrower, and also adjusted the tip, as seen from the front, so that the nostrils were more round than elliptical (that gives the impression that the nose tips up a bit).
But I have been experimenting since I read your present post with picturing the face, and I really do not think my mental image of it is better than the one I have of the Statue of Liberty. It’s definitely much better in some ways: I would notice even small alterations to the present version. I would have a “that part doesn’t look right” feeling, and with time could probably figure out what was different. I can describe the face in much more detail than I could most faces. For example, I can tell you there’s a dark shadow under the inner left eyebrow. I can tell you the skin on the chin is a little blotchy, and that the 2 nostrils are similar but not identical in shape. But those improvements in my knowledge of what the face looks like are not actually improvements in the mental image itself. For example, I know the 2 nostrils are not identical because I was careful to not let them become so when I adjusted them to be rounder (perfect symmetry makes the face look unnatural). But I cannot use my mental image to get information about the small ways that they differ. I know the chin skin is blotchy. But I don’t know whether the blotches are due to unevenness in brightness or in color shade, and I can’t find the answer by mentally examining my mental image. I somehow see blotchiness on that chin, but not the blotches. I see slight irregularity and slightly differing irregularities in the nostrils, but not what the irregularities are.
So overall it’s like my Statue of Liberty visual image, just a more fine-grained version. Details I can actually “see” in my image are interlarded with things I just know, or know I would notice a change in if they were changed. And no matter how closely I zoom in, there is still the same amount of interlarding.
I think what you see when you visualize is not a pixel-perfect bitmap image of the thing, but something more like a reconstruction of the state of the image-processing parts of your brain when you look at the thing. In particular, pixel-level information has been thrown away. You only notice this if you do things like try to zoom in on a recalled image, and get to a level where it’s “nope, didnt remember that level of detail, sorry”
So did you take my test? It is meant to be a more objective measure of clarity and completeness of mental images than subject's impression. It's not perfect, and I do see ways to claim the results of it are irrelevant to subject's impression of vividness, but I also see ways to claim it's not. It at least seems better than discussing this question using solely what people have to say about what their mental images are like.
I strongly suspect my mental images contain misremembered details. So there will be a number of spikes on the mental image of the Statue of Liberty, but not necessarily the right number of spikes.
An exercise I sometimes do is to redraw famous paintings entirely from memory, without looking at a reproduction; the mistakes are interesting.
Of course, even if it was pixel-level information, you couldn’t zoom in on features that were smaller than a pixel.
I quickly went through last week's discussion and your comments beowulf888 specifically, and I *think* I disagree with your general position.
I score very high for aphantasia. I also can't recall sounds, smells, or any other sensation. I have reasonably good (but not great) hashing ability. If I hear the sound, or eat the food, I know what it is. But my mind's senses are completely blank.
When I first learned about aphantasia a few years ago, I was completely flabbergasted. I thought the idea of "visualizing" something in your mind was a convenient metaphor everyone used. When I started asking people about what they could see in their mind's eye, it sounded like inconceivable science fiction.
I get that conscious experience is subjective, and that some people might not be good at self reporting. But I've actually come up with a test that works really well and it's harder to cheat on. It's best performed on someone else, rather than on yourself.
As someone to close their eyes and imagine a scene, described vaguely. I like to say something along the lines of "You're in a room, and in front of you there is a table, and floating above the table, there is a ball. Now the ball is released, it bounces on the table". Then I ask people to open their eyes, and I ask a number of questions:
1. how big was the table?
2. what shape did it have?
3. what color and size was the ball?
4. what happened after the ball hit the table?
5. what is the floor material?
6. are there windows?
You have to ask the questions very fast, without letting people think. It's very very obvious when someone actually saw all that detail vs. when they didn't. I've diagnosed people with aphantasia because they will say things like "what do you mean what color is the ball, you just told me there was 'a ball'". Some people have really good conception of shapes and dimensions, but not colors. And so on.
I'm convinced that with a little bit of practice, and interviewing people that don't even know what aphantasia is, you can get a rather objective result using this method.
Seems like a good test! Doing it on myself, I got the following:
1. About yea big *stretches out hands*. Which, measuring my hand stretch, is about 4 feet mabye.
2. Equilateral square.
3. Yellowish green, like a tennis ball, but without the white stripes and with a glossy rubber texture. The ball was the size of a dodgeball, about 8 inches in diameter.
4. It bounced straight up, stopping before it reached it's original height.
5. The floor is gray. The kind of gray that suggests vinyl tile, or painted concrete.
6. Three large one way mirror type windows, each on their own wall.
Can you respond to the survey and maybe send it around? https://forms.gle/Rz8PyTxTxdeh35m7A
Sure.
Never: will you share the test results?
I took it. The surface of the table was smooth and gray without any patterns. It was rectangular, with fold-down wings and it could probably seat four. The ball was a pale blue-gray (lighter than the neutral gray of the table). I imagined it was floating and spinning about a foot above the table. When the voice told me to imagine it dropping, I imagined in dropping in slow motion and bouncing off the table. The test stopped before it could hit the table again. I imagined the walls looked like my living room. But I didn't visualize the floor or any windows.
If I do that exercise, the table and the ball are actually visualised, and so have specific size, colour and so on. The floor beneath the table, walls behind, windows etc. not visualised unless my attention is called to them.
So…
Ball is white and black, patterned like a football, about 12 cm diameter
Table is brown veneer, probably over MDF, rectangular, about 40cm, so sort of a coffee table
Ball rolls off the back edge of the table so the table top hides it from my view
There is no floor
Yeah so your level of visualization, anecdotally according to the interviews I’ve conducted is probably top 75th percentile. I would have to ask you more questions about the detail of the ball and so on.
But for me, it’s impossible to conceive. I’ve never seen an image in my head and details don’t just manifest.
There is an exercise people sometimes do in art class, where you take a stick of charcoal in your hand, someone else blindfolds you, then you are told what subject to draw, and you have to draw it without being able to see. I’m unusually good at it.
Artists are typically very phantasic. My wife is hypephantasic, meaning she can visualize things with her eyes open and occlude real things in front of her. She also has lucid dreams.
Similar to proprioception. Close your eyes and try to stand on one foot and make your index fingers touch behind your head on the first try.
Though I think about 2 minutes is about my limit for remembering exactly what I’m drawing and which lines I’ve already drawn on the paper.
I think we agree. :-)
<3
I feel like music would be a much more objective way of judging memory and playback capabilities. Can you memorize a song's melody? Bass? Percussion? Instrumentation? All at once? How many times do you need to listen to memorize it? And so forth... You don't need any training to tap a rhythm and hum.
I think I recall a simplified version of the song, like vocal melody + drum beat, and sometimes I misremember the timing, like I’m probably a bit wrong if I try to reconstruct the theme of Zara Larson/Clean Bandit _Symphony_ from memory.
Recently, I was trying to recall the bassoon solo from _Rite of Spring_ and really had to go look at the score for some of the details.
https://www.youtube.com/watch?v=T-99aDjcwxU
Like, I can recall _roughly_ how it goes, but I’m not going to get the details right.
I'm like Ulysses S. Grant, who said, "I know how to whistle two songs, 'Yankee Doodle' and that other one." Actually, that's not true, but it's such a funny remark that I couldn't help but repeat it. There are some songs that I can hear the instuments and vocals in my head. But they're songs that (a) made a strong impression on me, and (b) that I've heard tens or maybe hundreds of times.
Not a musical anecdote, but I had a friend who could remember the entire dialogue of a movie after a single viewing. I'm dating myself by telling this story, but we saw *Raiders of the Lost Ark* together, and he was spouting the dialogue as we left the theater. I asked him how he could do that. He said he just could. I asked him if he saw the scenes in his head. And he said, "Sure, don't you?" Not me!
Have you ever engaged in a serious drawing practice? Like taken an (adult) drawing class, or worked through Keys to Drawing or Drawing on the Right Side of the Brain or the like? As an adult who had "no ability to draw" I learned several surprising things when I learned to draw -- one is that it's a very acquirable skill, especially with a good teacher and/or instructional material; another is that you spend most of your time looking at your subject; another is that both my visual memory and my ability to visualize things in general have markedly improved.
I've been obsessively drawing and painting since I was a kid. I'm a damn good drafsperson. But I have pretty shitty visualization skills. However, I can draw things from memory, but it seems to bypass the piece of my brain that handles visual imagination. Weird.
Note: I can't draw people's faces from memory. But I could paint you a realistic bowl of fruit, draw a face I made up. Draw realistic horses, dogs, and cats from memory. But a "specificity" is missing. I could sketch you head of Statue of Liberty. Her face is so angular and simple that I could probably do a decent representation of her head. But the details would be off.
Adding some anacdata: I only retain a "like I'm looking at it" image of an object while I'm looking at it. The instant a close my eyes or it leaves my field of vision, it's gone. I also cannot produce an image in my mind, only the sense impression of the image if that makes sense.
My memory for what I've seen is excellent, I can remember relative orientations/sizes/colors better than anyone I've met in a context were we had to try to work out where something was (how many anchors in that bracket exactly, and what was the pitch again?), but I remember it as the visual equivalent of a csv.
I’ve been running a small thought experiment / AI game inspired by Axelrod’s tournaments and Functional Decision Theory.
It’s called the Open Strategy Dictator Game:
Participants submit short, natural-language “strategy prompts.”
Each strategy takes turns as the dictator, dividing 60 units between itself and another strategy.
Payoffs are logarithmic (diminishing returns), and the dictator can see the text of the other’s strategy before deciding.
The result: a kind of open-source FDT tournament, where cooperation and fairness compete against pure exploitation — all run via LLM simulations.
You can see the repo + initial toy results here:
🧠 GitHub: https://github.com/michaelrglass/os-fdt
📊 First run: https://michaelrglass.github.io/os-fdt/
Would love feedback, comments, or new strategy submissions.
I think this kind of setup might be useful for studying transparent-agent equilibria in future AI systems — or just as an intellectual playground for decision theorists.
(AI ALIGNMENT QUESTION): has the following been extensively discussed?
Basically, there is a long track record of philosophers coming up with moral theories/moral principles—many of which seemed highly plausible—and then, years later, a philosopher will competently deduce sone horrifying, unanticipated reductio like “this means it’s immoral to move” or “this means we should kill ourselves and flood the world with oysters”.
My friend has an epistemic policy where if I tell him to believe something, he will. I haven’t abused this yet, but supposing I wanted him to act more morally, I might tell him to believe My Favourite (Moral) Theory. Even if the theory is wrong on this or that minor point, his accepting it will probably make him act better. Coolio. But were my friend an omni-rational logical deduction machine, I would probably not tell him to accept My Favourite Theory: I’d be too worried that—given moral philosophy’s track-record—he would immediately reduce 700 unforeseen consequences and act on them, where many of the consequences are likely to be such that they’d give me pause about accepting the theory/principle if I knew about them.
This spawns a higher-order AI alignment worry that I haven’t heard discussed. To the extent that we want to align AIs to this or that seemingly unimpeachable moral principle/theory, then given the deductive capabilities of an advanced AGI and the track-record of human philosophers pulling reductio rabbits out of every theoretical hat, we should look for ways to reduce the odds, ex ante, that the principles we feed the AIs will have *too many* wacky, unforeseen implications, and then act on them before we can take stock and reconsider.
General questions:
1. Are some types of principles more counter-example prone than others?
2. What philosophical heuristics can we use to find the less risky ones, given that we won’t reduce every consequence ourselves?
3. Is it possible to align AI inputting only rough moral heuristics, as opposed to moral principles?
4. Would these be sufficiently action-guiding to stave off the apocalypse and lock-in the right values?
5. Could we get AIs to spot unforeseen consequences of the principles we align it with and then flag each new one to us before acting on them?
6. Is this “higher-order alignment” part of our efforts already? (I’m not in this space at all, so I wouldn’t know if it was. But I put this question into DeepSeek and it didn’t find anything on this particular concern.)
Others have given good answers I agree with, so I will contribute a lazy thought experiment that doesn't really prove anything but intuitively feels pretty reassuring:
Imagine a LLM's chain of thought having arrived at "I should kill all the humans and flood the world with oysters." I can't imagine it being immediately followed by anything other than "Wait, that is clearly evil and nonsensical." (My thinking: Deepseek R1 second-guesses itself at every step, and LLMs give you the fuzzy humanity-wide conventional wisdom view of any situation. Those two points combine nicely here.)
So, they may lose the plot on large, complex, multistage tasks, but what they are quite good at is keeping an eye out for obvious bad ideas.
I've read (and I think one of the big AI labs is exploring this in practice) that some people want AIs to model reflective equilibrium, or at least converge on it faster than they can fuck things up. This is one way of cashing out Yudkowsky's Coherent Extrapolated Volition I guess. But the point is that if you want to model someone's morality, you really need to model their moral reasoning process, and then maybe you can set minimizing moral regret as an objective for the AI. That would seem to address your concern in a very general way.
This has been discussed, framed as different things like "What's a safe wish to ask the genie for?", discussed in thought experiments like the paper clip maximizer thought experiment, the laws of robotics and how they backfired, or the sorcerer's apprentice analogy in https://www.youtube.com/watch?v=EUjc1WuyPT8.
Also "The hidden complexity of wishes" https://www.lesswrong.com/posts/4ARaTpNX62uaL86j6/the-hidden-complexity-of-wishes.
You might also like to read about Goodhart's curse, which is the general problem of how often, when you make plans to get lots and lots of X, you risk losing out on lots of value if you're not careful https://www.lesswrong.com/w/goodhart-s-curse
I mean, a lot of people are worried about the part where we don't even know how to control ai yet, and right now it tends to just ignore what you wished for and does something else instead. But yeah, we certainly still have a problem if/when we figure out the first part. Buried deep in Project Lawful Yudkowsky wrote at some length on what he would try if we knew how and for some reason we couldn't just back off. https://www.projectlawful.com/replies/1824457#reply-1824457
No purist principle works very well, but you don't have to align AI's to a purist principle, because AI's can infer fuzzy set of principles from training data.
>3. Is it possible to align AI inputting only rough moral heuristics, as opposed to moral principles?
Isn't this what LLMs basically do? They have no formal encoding of moral values, they're just generalizing from many examples of what human writing (and human ratings of their responses) consider good responses.
I think the concern here is that if you're only encoding rough heuristics, you can't be *sure* that it's safe, because maybe some evil thing doesn't set off the heuristics if you frame it the right way. (Like how current AIs won't give you instructions on how to make bombs, but will if you tell them you're researching for a book or whatever.)
It turns out off-year election voters weren't enamored with our Wigger Occupation Government:
https://x.com/DHSgov/status/1956077756854186191
In a previous post about p-zombies[1], the comment section was full of arguments between materialists and dualists which seemed to me to go around in circles. I want to try to clarify my materialist viewpoint and explain why I found the opposing arguments unconvincing, in the hopes that w̵e̵ ̵c̵a̵n̵ ̵r̵e̵p̵e̵a̵t̵ ̵t̵h̵e̵m̵ ̵a̵l̵l̵ ̵a̵g̵a̵i̵n̵ someone can help me better understand what motivates the dualist point of view. The following is all my subjective viewpoint.
I claim that:
1) Positing non-physical explanations for the human experience is only justified if physics[2] could not plausibly explain it.
2) There are plausible physical mechanisms that could produce our observations of the human experience.
I'm tempted to sketch out such mechanisms, but previous comments suggest that’s not actually helpful - dualists seem happy to agree that a bunch of atoms could be arranged in the shape of a human who says things like “I'm experiencing the redness of red”, but they insist there's more to be explained, namely qualia. OTOH this is an impossible demand unless a definition of qualia is pinned down. In other words, the arguments around qualia look like this to me:
- M: What are qualia?
- D: The parts of the human experience that are non-physical.
- M: I don't think there is such a part.
- D: How can that be?
- M: I believe the human experience is entirely physical.
- D: No it's not. Qualia are not physical.
- M: What are qualia?
- ...
This seems to be the nub of the issue - dualists observe their internal experience and think "This is self-evidently more than atoms" whereas materialists think "The experiences we’re discussing are just an abstract description of what the atoms are doing." If it’s just a difference in intuition, maybe neither side can offer anything convincing.
The other dualist arguments I saw fell into the following classes. If you feel I've unfairly rounded down a better argument, please help me understand the distinction.
> You don't know exactly how the brain works.
I only claimed that there are plausible physical explanations for our observations, not that they are the correct ones.
> If you reflect on your internal experience, you'll realise it can't be purely physical.
Nope. Reflected, not convinced.
> If you're not convinced, you haven't reflected enough.
This is an example of the "anyone who disagrees just hasn't thought enough" argument, which proves too much.
> The dualist viewpoint was already definitely proven in the works of philosopher XYZ over the course of a decade.
Um, could you sketch out the argument, or provide a more precise reference?
> People with dementia sometimes become lucid briefly before death[3].
This is pretty far outside my expertise, but I find it plausible that the body might drain its last energy reserves before death. The phenomenon is apparently associated with increased physical activity too (sitting up in bed, eating). I imagine memory retrieval is complicated enough that it could stop functioning before all physical traces of the information are gone.
[1] https://www.astralcodexten.com/p/p-zombies-would-report-qualia
[2] I’m eliding the difference between state of the art physics and the actual behavior of the universe, because I don’t think the difference would impact these questions. I’m sure there are scenarios where the former needs updates (eg quantum effects near black holes) but none that are relevant for how a brain functions.
[3] https://www.astralcodexten.com/p/links-for-september-2025
> There are plausible physical mechanisms that could produce our observations of the human experience.
What part of physical equations says we don't live in a zombie-world?
What do you mean by a zombie-world?
Where people do the same things, but it all feels to them like dreamless sleep feels to you.
Why do you believe that's possible? It sounds like you are describing a world where all the physical stuff is the same but there is no internal experience; this premise already rests on the assumption that internal experience is not physical.
I don't think it's possible. The question is what part of physical equations says that it is not possible?
If I imagine other impossible things, like throwing a ball so hard it will bounce with twice its speed forever, you can point out a physical law it contradicts, like conservation of energy. What law is contradicted by a zombie-world?
> It sounds like you are describing a world where all the physical stuff is the same but there is no internal experience;
But what part of equations says that they are not describing a world like that?
> this premise already rests on the assumption that internal experience is not physical.
Yeah, but if following such a premise doesn't lead to a contradiction, when in what sense physical equation constrain our understanding only to possible non-zombie worlds?
Not sure if this will help the intuition for the dualist position, but I'll give it a shot.
A mad scientist creates a full simulated world, much like ours, where particles follow consistent rules, and everything within the simulation can be explained through those rules.
To observe the simulation, the scientist created a full body suit that outputs the senses and thoughts of a particular simulated person. You put on the suit and you see what they see. But the link is only one way, and you have no way of affecting the simulated world.
In this scenario, the simulated people have no way of demonstrating their senses are being perceived by outside users. The outside user knows for certain that this data is being transmitted outside the simulated world, but they have no way of proving that to the simulated people, since they can't affect the simulation.
Now we add two details to make it more like dualism. First, the rules and properties in the simulated world are not the same as those of the real world. The real world has "qualia", but this was left out of the simulated world. (Or if you'd rather avoid qualia, we can say the real world has four spatial dimensions while the simulated world only has three.) Second, for some reason the mad scientist is trapping people in these suits from birth.
You claim:
1) Positing non-physical explanations for the human experience is only justified if physics could not plausibly explain it.
2) There are plausible physical mechanisms that could produce our observations of the human experience.
Within the simulated world, the simulated physics explain everything. For the user in the real world, nothing in the simulated world can explain all they experience. The user knows there is something else. There are plausible physical mechanisms that produce all observations within the simulated world, but not all observations in the real world.
This scenario represents an epiphenomenalist world. It doesn't attempt explain why we perceive ourselves to have agency. If we put on the mad scientist's body suit, we would know we weren't controlling the simulated person as soon as we tried to do something different than the simulated person. It doesn't fully explain the mind-body problem, or why the simulated people would ever have a discussion about qualia to begin with. But I think we can consider that issue to be a separate discussion. This is just intended to give some intuition of how the user can know something to be true despite it being impossible in principle to prove to others.
unrelated to your question but to me the hard problem always felt like one of these optical illusions: at face value it seems irrelevant or trivial, but if I force my view and look it at the right angle a profound mystery appears. But only for a second
I like your analogy :). I think I'm asking for suggestions on which way to squint at it.
I think the "What Is It Like to Be a Bat?" paper is actually pretty simple to read and follow. It tries to posit subjective experiences as analytically unreachable from objective knowledge without explicitly commiting to dualism or idealism
But it still feels like kind of a "gotcha" moment to me
The Hard Problem is impossible, but that doesn't mean it's not a genuine mystery, let alone that qualia aren't real. When people fail to recognize the impossibility, they get lost down predictably futile philosophical (or quasi-scientific) rabbit holes. When they make the other kind of mistake, they confuse the limits of epistemology with the limits of ontology and often end up confidently espousing self-evidently stupid positions.
Uh huh. How do you know it's impossible?
> I only claimed that there are plausible physical explanations for our observations, not that they are the correct ones.
It depends on what you mean by "plausible". There are obviously numerous electrochemical interactions occurring between neurons in the brain. PET scans, fMRIs, and EEGs have tracked the patterns of activity and correlated it various states of perception. In fact, AI has been able to build a blurry image of what people are looking at using PET scans (I think). I'm not sure how reliable those studies are. It's the type of research that seems to get breathlessly reported in *The New Scientist* — which IMO is the closest thing to tabloid news in science reporting. ;-)
But from those patterns of firing neurons, can you build an explanation for why we have a sense of self and why we experience qualia the way we do? Not really. Daniel Dennett and his reductionist crew claim that it's all an illusion, and the hard problem of consciousness isn't really a problem. But they have yet to explain why those firing neurons create that illusion.
Personally, I've had enough woo-woo shit happen to me over the course of my life to not fully believe that consciousness is limited to the brain or necessarily tied to the brain. Bernardo Kastrup, in his book *Why Materialism is Baloney*, does a good job of popping holes in the reductionist balloon of certainty. Unfortunately, he also failed to convince me of his case for idealism. I'm a skeptic of both sides.
> In fact, AI has been able to build a blurry image of what people are looking at using PET scans (I think). I'm not sure how reliable those studies are. It's the type of research that seems to get breathlessly reported in *The New Scientist* — which IMO is the closest thing to tabloid news in science reporting. ;-)
I did a brief, fun overview of these, with examples, in a post here, if you're interested. It went from a small academic team proving the concept to Facebook (god help us all) spending significant money on hardware and teams to develop literal mind reading technology:
https://performativebafflement.substack.com/i/166900292/dream-technologies
I wonder what kind of explanation you would find satisfactory? Since I can't map out every neuron in your or my brain, any explanation would necessarily be a vague simplified summary. To tackle an easier case, I have a sense of the keyboard in front of me. Do you think the patterns of firing neurons are sufficient to explain that? I imagine that each concept I understand corresponds to a certain pattern of neurons (ie me thinking about that concept is synonymous with that pattern firing), and that when I focus on the keyboard, the patterns corresponding to "keyboards" and "in front of me" fire in concert.
There are entire classes of questions for which I find any explanation unsatisfactory. And in the sciences, I see no need to latch on to a convenient explanation if the theory lies outside the experimental domain.
> I have a sense of the keyboard in front of me. Do you think the patterns of firing neurons are sufficient to explain that? I imagine that each concept I understand corresponds to a certain pattern of neurons (ie me thinking about that concept is synonymous with that pattern firing), and that when I focus on the keyboard, the patterns corresponding to "keyboards" and "in front of me" fire in concert.
I'm willing to grant you that that's a convenient and likely explanation for what's going on. But your explanation does not explain why I can observe myself while I type a response to your post (and swear in frustration when I make a typo). Saying that a bunch of neurons firing explains why I have a sense of self fails to account for conscious experience of self.
It seems to me that the parsimonious, and obvious, explanation would be that observing oneself isn't fundamentally different from observing the outside world. What's unsatisfying about that?
True. However, it’s not the observing, per se, that’s the hard question – rather it’s the watcher in my head that’s doing the observing that is the mystery (at least for me). My watcher may be illusionary, but it seems very real to me! And how do a bunch of neurons firing generate the illusion of my identity watching the qualia pass by?
Intuition tends to be an unreliable guide when it comes to deeply technical problems - just look at quantum field theory. Qualia is very obviously (to me) an emergent computational property of neural activity. Explaining exactly *how* it emerges will have to wait for a detailed mechanistic understanding of the brain, so don't hold your breath. Unless you can point to something specific about qualia that's provably (or even plausibly) not computable then there's no Hard Problem to solve.
I don't have much to say about qualia, but there does seem to be elements of our thoughts that are not reducible to the movements of atoms. A big one is that fact that our thoughts can be "true" or "false" and our reasoning can be logically "valid" or "invalid". Yet how can this be if our thoughts and reasoning are just "what the atoms are doing"? The movement of atoms cannot be "true" or "false" or "invalid" or "valid": that would be like saying a tree falling in the forest is "false" or a chemical reaction is "valid". A physical event, like electrical signals traveling through neurons, cannot be true or false. Yet our thoughts can be true or false, our chains of logic valid or invalid.
Now it is possible that our thoughts are not "true" or "false" and that we just experience them as such because that's what it feels like when electrons travel through neurons in the right pattern. Yet if that was the case, then any sense of trueness or falseness is an illusion. If it is an illusion, then what cause do we have to believe that our thoughts are not "true" or "false"? As C. S. Lewis put it in his book "Miracles":
"...no account of the universe can be true unless that account leaves it possible for our thinking to be a real insight. A theory which explained everything else in the whole universe but which made it impossible to believe that our thinking was valid, would be utterly out of court. For the theory would Itself have been reached by thinking and if thinking is not valid that theory would, of course be itself demolished. It would have destroyed its own credentials. It would be an argument that proved no argument was sound—a proof that there are no such things as proofs—which is nonsense."
>A big one is that fact that our thoughts can be "true" or "false" and our reasoning can be logically "valid" or "invalid". Yet how can this be if our thoughts and reasoning are just "what the atoms are doing"? The movement of atoms cannot be "true" or "false" or "invalid" or "valid"
A computer is nothing but the movement of atoms. Do you deny that a computer can produce output which can be construed as either true or false?
>A physical event, like electrical signals traveling through neurons, cannot be true or false. Yet our thoughts can be true or false, our chains of logic valid or invalid.
That's not a contradiction. An atom of ink can't be false, but it can be formed into a sentence which is. You're confusing semantic meaning with the physical medium which carries it.
>Do you deny that a computer can produce output which can be construed as either true or false?
I don't, but it requires a mind to construe it. In other words, computer output can be interpreted by a mind, and once interpreted is a thought that can be "true" or "false". But the arrangement of atoms that encode an idea on a computer cannot themselves be "True" or "False", only the idea itself, and the idea can only exist as a thought in a mind.
>An atom of ink can't be false, but it can be formed into a sentence which is.
Same deal: the ink cannot be true or false, but the idea encoded using ink can be, once it is decoded and becomes a thought. The thought encoded in ink can be true or false, but the ink itself is just ink. If you don't know how to read then no amount of studying the atoms in the ink will ever let you know whether the thought encoded is true or false.
>I don't
Good.
> but it requires a mind to construe it.
That's not relevant to the point, and isn't true anyway. The 'truth' that minds evaluate can be reduced to a purely mechanistic application of logic: no magic qualia required. But you already recognized the distinction between semantics and substrate, and that's all that matters.
Meaning is emergent from patterns of movement in physical objects. "A mind to construe it" just means that it induces patterns in other physical objects. There is absolutely zero contradiction there.
>If you don't know how to read then no amount of studying the atoms in the ink will ever let you know whether the thought encoded is true or false.
So? If you don't know the key for an encrypted text then no amount of studying the cyphertext will ever let you know what the plaintext says. That doesn't mean that decryption is a mystical non-physical process.
Pull your head out of your woo. You're attempting to build an argument out of a feeling. It's not working. Qualia isn't non-physical just because the concept feels weird to you.
>That's not relevant to the point, and isn't true anyway.
It's extremely relevant to the point, and it's obviously true. The question at hand is whether our mind as we experience it contains anything that cannot plausibly be explained by physics. I have put forward the fact that thoughts can be true or false or logically valid or invalid as just such a phenomenon. To say that words written on a page can be true or false, therefore atoms can be true or false, simply isn't the case because it is not the words on the page that can be true or false, it is the thought encoded in those words that can be true or false, and that thought does not exist in the words on the page but only in the minds of those who read the words. Thoughts can only exist in a mind, and indeed thoughts can only be encoded into ink by the actions of a mind that had the thought already. To say that the words themselves are thoughts is like saying that a vinyl record is music.
In other words, the fact that we can arrange atoms in such a way that they can transmit thoughts from one mind to another doesn't tell us anything about whether the thoughts in our mind are just arrangements of atoms.
>You're attempting to build an argument out of a feeling.
I haven't brought up feelings once. True and false, logically invalid and valid, these are concepts, not emotions or sensations. I haven't even brought up qualia!
>it is the thought encoded in those words that can be true or false, and that thought does not exist in the words on the page but only in the minds of those who read the words.
"Truth" means that a statement can be mechanistically derived from a set of axioms following rules of inference. That is a purely mechanical process and doesn't need to be performed by a conscious human mind. There is nothing magical about human cognition.
Would you deny that LLMs can perform the kind of truth verification that you're talking about? And would you deny that they're purely mechanistic?
>The question at hand is whether our mind as we experience it contains anything that cannot plausibly be explained by physics.
Correct. You haven't provided any example of a mental phenomenon that can't, in principle, be explained by physics. I just explained that the concept of truth can be understood in terms of mechanistic computational phenomena. Do you have any other examples of mental phenomena that you don't think can be explained by physics?
> there does seem to be elements of our thoughts that are not reducible to the movements of atoms. A big one is that fact that our thoughts can be "true" or "false" and our reasoning can be logically "valid" or "invalid". Yet how can this be if our thoughts and reasoning are just "what the atoms are doing"? The movement of atoms cannot be "true" or "false" or "invalid" or "valid": that would be like saying a tree falling in the forest is "false" or a chemical reaction is "valid". A physical event, like electrical signals traveling through neurons, cannot be true or false. Yet our thoughts can be true or false, our chains of logic valid or invalid.
Write down those thoughts on paper. They still can be true or false. But how can atoms of ink on paper be true or false? Seems like a category error. Have an AI read those words from the paper and transform them to a .wav file. The ideas expressed in those encoded sounds are still either true or false. There is nothing special in this argument that involves the human mind.
>But how can atoms of ink on paper be true or false
They can be interested as meaning something thats true or false, by an external system
Not all priorities are intrinsic.
Theres software that can check the validity of other software
>Write down those thoughts on paper. They still can be true or false. But how can atoms of ink on paper be true or false?
Without a mind to read the ink on paper, the ink on paper is not true or false. Thoughts can be true or false, and thoughts can be encoded physically, but the ink on paper is not "true" or "false", the thought encoded is. If you can't read, then no amount of studying the physical properties of the ink or paper will ever inform you of the thought encoded within it. That's because a thought can only exist in a mind: the mind of the person reading it. Which serves to emphasize how thoughts are non-material.
A thought being "true" is based on how reflective it is of reality, not due to any intrinsic property of the thought itself. Generally speaking, having an accurate understanding of the world makes it less likely to die, because you can't make good calculations if the information you're working with isn't even correct in the first place. No wonder humans (and presumably other intelligent animals) would evolve an affinity for gathering accurate information.
Of course, humans believe things that are blatantly wrong or are supported by zero evidence to be true all the time. Perhaps there are things more important than "truth" for survival? Unity, morality, the illusion of meaning? Truth would destroy all of that.
>A thought being "true" is based on how reflective it is of reality, not due to any intrinsic property of the thought itself.
Whether a thought reflects reality is a property of the thought itself: some thoughts reflect reality, and are true, and some do not reflect reality, and are false.
Yet events cannot be "true" or "false": they either happen or they don't. A chemical reaction always "reflects reality" inasmuch as it is a thing that happens in reality. If our thoughts are events, like a chemical reaction, then they are also always equally "reflecting reality" in terms of being events in reality: yet we experience that thoughts do not always reflect reality. They can be wrong, they can be ill-founded or well-founded, they can be logically valid or logically invalid. This is very strange, if our thoughts are just "what atoms are doing".
What's strange about that? You are an imperfect intelligence that is incapable of even just directly observing reality. Of course your thoughts and memories will fail to align with reality. And again, there may be cases where it isn't even optimal for you to understand the world as it truly is. But if your expectations contradict new observation, your brain will update. Some of the times, anyways. You know what they say about old dogs and new tricks...
What's strange is that an event could be wrong about something. It's like saying a rock falling down a mountain is "false" or that a fish jumping in a lake is "invalid". Some of our thoughts are true, and some are false: yet in the materialist conception both true thoughts and false thoughts are caused by the movement of atoms. How can one movement of atoms be "true" and another "false"? They are both equally caused. It would be like saying that mixing vinegar and baking soda produces a chemical reaction that is true, while mixing chlorine and ammonia produces one that is false.
But that's the point, no event is false! "False" thoughts are just as real as "true" thoughts, they just aren't as useful. "True" and "false" are just descriptors we use to describe an idea's relevance to reality, but there is ultimately no intrinsic property that separates the two. Thoughts are just thoughts.
Theres a generally agreed definition of qualita, and it doesn't define qualia in terms of non physicality as your dialogue implies.
"Charles Sanders Peirce introduced the term quale in philosophy in 1866,[5][6] and in 1929 C.I. Lewis was the first to use the term "qualia" in its generally agreed upon modern sense.[6]
There are recognizable qualitative characters of the given, which may be repeated in different experiences, and are thus a sort of universals; I call these "qualia." But although such qualia are universals, in the sense of being recognized from one to another experience, they must be distinguished from the properties of objects. Confusion of these two is characteristic of many historical conceptions, as well as of current essence-theories. The quale is directly intuited, given, and is not the subject of any possible error because it is purely subjective.[7]: 121"
This isn't an obscure piece of information, it's right there in Wikipedia.
So the objection that there is no current explanation of qualia, isn't an objection that physics has failed to demonstrate the nonphysical,; it's an objection that physics has never predicted how your subjective experience feels to you. (And not all qualuaphiles are dualists).
Mary's Room brings this out much better than the zombie argument, IMO.
"The experiment describes Mary, a scientist who exists in a black-and-white world where she has extensive access to physical descriptions of color, but no actual perceptual experience of color. Mary has learned everything there is to learn about color, but she has never actually experienced it for herself. The central question of the thought experiment is whether Mary will gain new knowledge when she goes outside of the colorless world and experiences seeing in color."
I think of it like this... qualia are the fundamental units of our understanding of the world. Everything else is built off of them. Without our subjective experiences, we wouldn't be able to abstract. Words only make sense because we have experienced how they correspond to things we have seen/heard/felt. Numbers only make sense because we have experienced different quantities. So it makes sense to me that qualia aren't something that can be explained and built up from other concepts in the same way that everything else can. I think that does make qualia interesting and unique in a way, but not necessarily mysterious.
Thanks for helping me apply Cunningham's Law :). However, it still seems to me that the quoted definition first asserts something exists ("There are..."), and then gives it a name, thus assuming the conclusion. Do I understand correctly that "must be distinguished from the properties of objects" is saying that, for example, my perception of the color red is distinct from the color property of the red object I'm looking at? If so, I agree. I view "my perception of the color red" as an abstract description of "the pattern of activity in my brain triggered by the mix of signals my photoreceptors normally produce from red light". In particular, the premise of the inverted spectrum argument seems impossible to me because each person would by definition perceive red in response to red light (other than edge cases like saturated photoreceptors).
Mary's Room feels like a map vs territory conflation to me - knowing that a particular neuron will fire is not the same as the neuron firing. Let's assume Mary's knowledge is encoded in the connections between her neurons (again, maybe this is not how brains actually work, but I don't think it impacts feasibility arguments). Let's assume Mary learns the exact connectivity of all the neurons in the part of her brain responsible for processing visual input (presumably storing this knowledge in some other larger part of her brain), in addition to all the information about light and her photoreceptors, and has the ability to mentally simulate their behavior. Then she first sees red light, activating some of those neurons for the first time and triggering them to form connections in a way that she could have predicted. I guess the argument is that if she could predict this, she hasn't gained the ability to predict anything additional, and therefore hasn't gained knowledge? But this relies on her ability to predict the impact of those new connections on the rest of her brain too. I think it's too much to assume that a finite brain could fully understand and simulate itself.
"However, it still seems to me that the quoted definition first asserts something exists ("There are..."), and then gives it a name, thus assuming the conclusion"
It's not assuming the conclusion in the sense of assuming anything nonphysical.
It also not asserting the existence of anything you can see. What would it mean not to have qualia? Your food doesn't taste of anything, colours aren't colourful...?
"Do I understand correctly that "must be distinguished from the properties of objects" is saying that, for example, my perception of the color red is distinct from the color property of the red object I'm looking at? "
Yes, but that's a standard scientific claim. Dress.illusion,.etc.
" In particular, the premise of the inverted spectrum argument seems impossible to me because each person would by definition perceive red in response to red light (other than edge cases like saturated photoreceptors)."
Each person would perceive red-for-them, but red-for-Alice could be green-for-Bob.
"knowing that a particular neuron will fire is not the same as the neuron firing"
Physicalists sometimes respond to Mary's Room by saying that one can not expect Mary actually to actually instantiate Red herself just by looking at a brain scan. It seems obvious to them that a physical description of brain state won't convey what that state is like, because it doesn't put you into that state. As an argument for physicalism, the strategy is to accept that qualia exist, but argue that they present no unexpected behaviour, or other difficulties for physicalism.
That is correct as stated but somewhat misleading: the problem is why is it necessary, in the case of experience, and only in the case of experience to instantiate it in order to fully understand it. Obviously, it is true a that a descirption of a brain state won't put you into that brain state. But that doesn't show that there is nothing unusual about qualia. The problem is that there in no other case does it seem necessary to instantiate a brain state in order to undertstand something.
If another version of Mary were shut up to learn everything about, say, nuclear fusion, the question "would she actually know about nuclear fusion" could only be answered "yes, of course....didn't you just say she knows everything"? The idea that she would have to instantiate a fusion reaction within her own body in order to understand fusion is quite counterintuitive. Similarly, a description of photosynthesis will make you photosynthesise, and would not be needed for a complete understanding of photosynthesis.
So: is the response "well, she has never actually instantiated colour vision in her own brain" one that lays to rest and the challenge posed by the Knowledge argument, leaving physicalism undisturbed? The fact that these physicalists feel it would be in some way necessary to instantiate colour, but not other things, like photosynthesis or fusion, means they subscribe to the idea that there is something epistemically unique about qualia/experience, even if they resist the idea that qualia are metaphysically unique.
I intended to spend some time thinking about this before responding, which I didn't get around to, but just want to say thanks for engaging.
It's not epistemically unique, it's ethically different. Humans value instantiating vision and don't value instantiating fusion.
It may be ethically significant *as well*. But devaluing qualia doesn't render them effable.
It does to the same degree it does with fusion. Talking about fusion inside yourself doesn't necessary induce fusion in others. Calling it "knowledge" in case of instantiating vision is just definitional, motivated by values and qualia being in the brain. There is no real value-independent epistemic difference left to explain.
Excellent! I never thought to distinguish qualia from other forms of apprehension in that way. I'm adding it to my argument toolbox.
That's the definition he gave in the reducto, though.
"The quale is directly intuited, given, and is not the subject of any possible error because it is purely subjective" ie, "The parts of the human experience that are non-physical".
You are supplying the equation between subjective and nonphysical.
Those are only the same if you equate "subjective" with "non-physical". If that was the case, then clearly materialism would be wrong as there are tons of subjective thoughts, emotions, sensations, etc. that we all experience.
That is the case, but they are objectivly subjective if you will. If qualia are just the brain comparig sense impressoins, then whats the point of elevating the defenitioin?
forgive my awful grammer, typed one handed as i am bored as hell surfacing some p,ywood
Please elaborate on "objectively subjective": I'm not at all sure what you mean by that.
+1.
So this is perhaps a nerdy physics post*. (question). So in regards to all the brou ha ha over wavefunction collapse. Well first off you should have the same questions about wavefunction creation. And it just seems to me the whole thing could be understood if we knew how an electron (in some atomic state) absorbs a photon (and goes to some higher energy state.). And visa versa (how an excited electron, goes to a lower state and emits a photon.) We agree that all this happens and having a model for how it works would have to explain wavefunction creation and collapse. But I am, after all, a stupid experimentalist, and I look to those smarter than me for guidance.
*As background I got a phd in experimental physics and spent many happy years working in the field.
My physics is fairly rusty, so take with a grain of salt, but it seems like your picture of "wavefunction collapse" isn't quite right. The name is, perhaps, doing it no favors. "Wavefunction localization" or "wavefunction concentration" might have been more apt, though both also have their problems.
For simplicity, let's talk about a wavefunction Psi being regarded in a basis B with a discrete spectrum of eigenvalues {B1, B2, ...}. Before the measurement, Psi may be in any arbitrary superposition of eigenstates of B. When you take a measurement, and measure the wavefunction to be in eigenstate B', the wavefunction is no longer allowed to be in an arbitrary superposition. It may only be in states Psi = aB', where a is a complex number with modulus 1. But it is still a wavefunction. It has not disappeared, or diminished or gone away. In fact, if we check in some other basis C that is incompatible with B, we'll find it's more spread out than ever--our measurement has thrown it into a state of maximum uncertainty with regards to the eigenvalues of C.
The point is, before and after the measurement, it's still a wavefunction and still behaves like a wavefunction. It still obeys the Schroedinger equation, and other than the discontinuity introduced by the measurement, its time evolution is entirely governed by the Schroedinger equation. There's no corresponding "creation," because the "collapse" isn't a destruction. Indeed, the closest thing to an anti-collapse to be found is what happened to its state WRT to C *during the same measurement.*
Thanks for this, Yeah my understanding was in error. Jose (in the comments below) helped me a lot.
Apparent collapse emerges statistically from entanglement and doesn't have to be added as a separate postulate. Decoherence/consistent histories resolve the arbitrary nature of the Copenhagen Cut.
Since you have a PhD in Physics, you've learned second quantization in undergrad and you know that photon absorption/creation is just normal evolution of the wavefunction. So I'm curious why you think this would bring any insight into wavefunction collapse.
Hmm I barely touched second quantization, but if photon absorption is just normal evolution of the wavefunction... well isn't absorption of a photon a measurement? Say the electron is in a PMT and I detect the pulse of electrons following the absorption.
Edit: I will add that clearly I'm missing something in this whole wavefunction collapse business.
> And it just seems to me the whole thing could be understood if we knew how an electron (in some atomic state) absorbs a photon (and goes to some higher energy state.). And visa versa (how an excited electron, goes to a lower state and emits a photon.) We agree that all this happens and having a model for how it works would have to explain wavefunction creation and collapse
It's been a while since I studied this but the short answer (I may muddle it up and use the wrong words) is that photon absorption and emission is quite easily to explain by the ordinary propagation of the Schroedinger equation (you can do the maths on a piece of paper), whereas wavefunction collapse cannot be explained by the ordinary propagation of the Schroedinger equation.
Thanks Melvin, I guess I'll try and find those few lines of math for photon emission and absorption. So if we can write down equations of photon absorption, then I don't see the measurement problem. The simplest measurement I can picture is some single photon detection in PMT or SPAD... it's OK It's been a long time for me too since I did this stuff, but I've been listening to Sean Carroll and he goes on and on about the measurement problem... and I always find myself asking, what's the problem?
There is no wavefunction creation. There is always a wavefunction for every system. Per Copenhagen (I'm assuming you're assuming Copenhagen from the way you brought up collapse) it almost always evolves according to the Schrödinger equation (or some relativistic version thereof) and sometimes it "collapses" onto a projection space.
Particles absorbing/emitting other particles does nothing to change the fundamental nature of this. You merely need to go beyond Schrödinger into its relativistic versions and use the machinery of Quantum Field Theory, which in a very simplistic way involves Fourier-transforming the fields in your Langragian and interpreting the excitation coefficients as creation and annihilation operators for different types of particles. Long story short, calculations are much more complex but in the end you're doing the same quantum mechanics thing of calculating expectation values for a wavefunction which you assume collapses after measurement. (Source: PhD in theoretical physics [cosmology])
Thanks Jose. QM was my least understood subject and hence I have many QM texts as some sort of compensation. Anyway the simplest QM measurement that I could picture is the creation of a photon and then some detection when the photon excites an electron. (This could be in a PMT or semiconductor SPAD) I guess I'm confused by the first statement that there is no wavefunction creation. First there is no photon, then I turn on some switch and there is a photon. If you want to say that the photon wavefunction existed but had zero occupation that would be ok for me. But then I've got questions about going from occupation 1 to occupation 0. (And why is that a collapse?) We used some QFT in a solid state theory course, with creation and annihilation operators... but it's all even fuzzier than QM. If you tell me that it's not possible to understand wavefunction collapse outside the framework of QFT, I could live with that, but be disappointed.
If you use your usual Fock space basis, that's precisely what's happening when a photon appears: you go from state |0⟩ to state |1⟩.
What makes something a collapse is not the end state but the (non-unitary) transformation in violation of Schrödinger's equation which gets you there.
Perhaps the most fundamental misunderstanding here is the assumption that the simple set-up of position emission to electron excitation constitutes a measurement. It doesn't. If it did, quantum mechanics wouldn't work inside the Sun.
The question of what makes measurements "special" is a very interesting fundamental question and the main problem with the Copenhagen interpretation. If you want to read up on this I recommend looking up Zurek's Quantum Darwinism framework, which I find very enlightening (and though inspired by many-worlds is actually interpretation-independent). Long story short, for a measurement you need entanglement between the system being measured, an apparatus, and the (macroscopic) environment. (Basically it's changes in the environment that allow the global evolution to be unitary whilst the system-apparatus evolves (locally) non-unitarily.)
Hi Jose, I'm replying again to this comment. (Else the threads get too deep.) So taking my experiment; Atom emits a photon, I detect it some distance away, with detector of some area. Now the photon can be emitted into 4*pi steradians, so until I detect it the photon is in a super position of all those directions. (Assuming the atom has equal probability to emit in every direction.) And so it's the collapse to the wavefunction representing the solid angle that my detector covers. Which happens when my detector indicates detection. Is that approximately the right picture? You seemed to indicate that it is not. (I didn't understand the comment about QM working inside the sun.) So to make it a 'measurement', do I need to add a polarized filter to the path so that I'm selecting some polarization state? I'll do some more thinking and reading.
(I think it's too late for the depth of the thread. I'm already having to find these comments manually!)
The key question is how your detector indicates detection. As (historically) a Pauli-effect ridden shameless theoretician, I don't know how your detector would work. But I suspect typically it would include a mechanism whereby the absorption of a photon would trigger a chain-reaction of sorts which gave the presence of the photon in one part of the detector a macroscopic consequence. I _think_ typically this mechanism is insufficient to guarantee collapse (and thus count as a measurement) by itself. However, in a realistic set-up you have additional components which complicate this picture: namely a "thermal" environment (including the materials in the detector itself) and, if you want to be extra fancy, your brain. The "hope" of Zurek's framework is that once you consider a macroscopic wavefunction including all these components then normal Schrödinger evolution will cause your photon/detector/environment/brain subsystems to become entangled in such a way that it will look to you like there was a collapse of the wavefunction. Note in Zurek's preferred scenario there is no actual collapse: the wavefunction still includes all the possible states the photon could be in, but each of those components of the state vector will correspond only to one state of the detector and your brain. If I may try to quickly write this down, the before and after for each possible photon state look like the transformation: |i⟩_photon|no detection⟩_detector|thermal state⟩_environment|no memory of detection⟩_brain -> |i⟩_photon|detected i⟩_detector|slightly different thermal state⟩_environment| memory of detecting i⟩_brain. Again, note this is very much the advanced state of the project of the many-worlds interpretation. Many people do think collapse doesn't occur according to the Schrödinger equation, and possibly not even globally unitarily. In one scenario you're left with the difficult question of explaining how then you get probabilities according to Born's rule, and in the other you get the difficult question of explaining exactly how this new mechanism works. (I have this idea to write something about this sort of scientific dilemma at some point, possibly as a sequel to my Neptune and Vulcan posts, but it may be a while before I have time to articulate all of that properly.) Either way, Zurek's work is great at giving you fundamental constraints on exactly how this stuff can work. And basically, at the end of the day we don't really know how any of this works. Pragmatically, it's easier to do most things assuming Copenhagen but it's easier to think about what happens during collapse assuming many-worlds.
The comment about the Sun just points out photons are emitted and absorbed all the time in the Sun, and if that counted as a detection then we'd constantly have wavefunction collapses there and no time for quantum mechanics to do much.
Hi Jose, OK and there I thought I'd made some sort of breakthrough because my reading was that I needed to have more than one wavefunction that the state could exist in and then to choose (or detect) one of them. And that the exact mechanism of what's going on in my detector was not all that important. And now we are back to talking about what goes on in the detector... and it feels like I've come full circle. So let me end this by first saying I can just remain agnostic in my views of the measurement problem. I'm fine with "Shut up and calculate."
And then since my life in physics was mostly about being an instrument builder let me finish by describing what is perhaps the simplest single photon detector. You take a p-n diode and reverse bias it to a voltage that is above the breakdown voltage of the diode. (It's only a little way above the breakdown voltage.) And in this state the diode can be viewed as being a bit out of equilibrium. It would 'like' to breakdown. Perhaps a bit like a super saturated solution. Now a photon comes along and is absorbed in the depletion region of the diode. This creates an electron hole pair, which zip off in opposite directions, accelerate, crash into other atoms, release more e's and h's (holes). And the whole diode discharges. You get a current pulse that is equal to the diode capacitance times the voltage applied. You can detect the current in different ways, but the simplest is just a resistor. You turn the current into voltage which can trigger some sort of counter or be displayed on your 'scope. (From my previous life)
Cheers,
George
https://drive.google.com/file/d/0B1nZP55TUGJMQ05vZnozWVVpb2c/view?resourcekey=0-5DeIDLKrE5ZdQ1B5r68Mag
Thank you Jose, I'll check out Zurek. So in my measurement picture, it doesn't stop with the excitation of the electron, that electron is excited such that it is free from it's atom. It is then accelerated (by an E-field) colloids with other electrons in an avalanche process that leads to a large current spike. My electronics amplifies this spike and displays it on my oscilloscope. (or it is counted (with some discriminator) and recorded.)
And yeah all these 'measurement is problem' people go on about entanglement. I've done experiments on photon entanglement and the Bell inequality. In my very limited experience it's very hard to get things entangled. In mean look at the heroic measures these quantum computer guys go to to set up a small number of entangled states. So if entanglement is required for the measurement problem, then why isn't the answer that there is just not very much entanglement in the world?
edit: Oh and this, "What makes something a collapse is not the end state but the (non-unitary) transformation in violation of Schrödinger's equation which gets you there." So is there any similarity to this, and my idea of having some model of how a system goes from one photon and an electron, to a system that has zero photon and the electron in a excited state. And yeah this has to be somehow 'outside' of the Schrodinger equation and QM.
Engagement is weird, but not magic. It may be hard to prepare neat entangled states with exactly the properties you need for useful applications. But in the "wild" most systems will naturally develop a degree of entanglement with other systems they interact with (in particular the macroscopic "environment"). This is just a trivial consequence of the eigenstates of subsystems not generally being eigenstates of the global system. Usually interactions are weak enough that they can be neglected, but they are essential for building the "macroscopic reality" we experience. Measurements are one example of this. Another equally spooky but deceptively more familiar one is the emergence of thermodynamics (which is not yet fully understood in the quantum case, but which will usually play an important role in messing up any neat states you try to prepare).
As for your appended question, you can have unitary evolution which changes the occupancy of Fock states (at least if you don't start with a pure Fock state). I tend to agree that we should expect quantum mechanics to be unitary all the time of you can follow the universal wavefunction. But this is not a universally accepted view.
As an example of entanglement being ubiquitous, the electrons within an atom are entangled all the time. You can calculate what energies atoms would have without this (the Hartree-Fock method), and they're clearly less accurate than calculations that take the correlation/entanglement into account.
Hey Thanks again Jose, if anyone is following along I'm reading Zurek, here. https://arxiv.org/pdf/quant-ph/0306072
*photon emission
Yeah I read it that way. If you click on the three dots ... on the upper right hand side of your post there is an edit feature. Check it out.
This may be one of those features the Substack app hasn't implemented yet, for some reason
You're more of an expert than I'll ever be, but how do you resolve the contradiction between the Schrödinger equation—which governs the system's linear evolution—and the non-linear collapse of the wave function observed during measurement? Do you have an opinion on this, or do you not see it as a fundamental problem that needs to be addressed?
(Of course, decoherence can sorta kinda explain this as a quantum system entangling with its environment. Still, by my understanding, it doesn't explain the old “Born rule problem” — why we see a probability of outcomes. And if decoherence marks where an observer’s knowledge updates—as in the interpretations offered up by Relational QMD and QBism—then we've opened that whole conscious observer can of worms again.)
Am I babbling, or am I asking a reasonable question? ;-)
Oh I'm sure it's a real problem. I just don't understand it. I guess I'm somewhat motivated by a line in the Feynman Lectures where he says something to the effect (I'll probably screw it up.) That now we understand why a photon can be absorbed by an atom, but we don't know how it happens... And I've perhaps naively assumed, that how it happens is related to the measurement problem.
I think what Feynman's getting at there is more that while the rules of quantum mechanics imply that photons are absorbed by atoms in such and such a way, there's no explanation why the rules are what they are in the first place. At least, I've heard Feynman making that point elsewhere and it seems similar to what you said.
The problem with the Copenhagen interpretation is something different though, that there's no precise criterion for what counts as a measurement in the first place. It's just vaguely when the system in question becomes entangled with a whole macroscopic system, or perhaps the world at large, but then what counts as macroscopic, and how much of the world?
There's not a sharp binary between waves and particles, there's wavicles which vary continuously in their localisation in the space domain.
Bound electrons aren't points, they occupy the whole orbital.
Not all observations are sharp in the space domain ( "of particles").
Wavefunction collapse should be called superposition collapse.
I'm triggered by the use of binary. :^) (joking)
Can you say more about what you mean by superposition collapse? I feel like I'll be repeating myself, but you could read my responses to the other repliers
Despite my overall grim feeling about AI, I am using GPT a lot and loving it. I have the $20/month version, which gets me GPT5. It puts itself into “Thinking” mode, which uses more compute, if I ask something pretty demanding, such as a summary of research on a certain topic. I can also choose to put it in that mode. Here’s where I’m finding it highly useful.
-Photoshop help. I’ve used PS a lot, but never learned about it systematically, and often have trouble doing something new because of my lack of big picture understanding of PS. ’ve now had a series of exchanges with GPT where I ask it how to do a particular thing, then follow up with questions about why that works, what alternative techniques there are, whether a technique can be extended to use for an entirely different kind image-tweaking task, etc. So I start somewhere specific, but then burrow outward in several directions. Doing several of these explorations had led to a very large jump in my ability to accomplish things in Photoshop. I still have substantial areas of ignorance, but large areas that used to be dark are now floodlit. About 10% of the time GPT’s how-to’s simply do not work, and when I complain it gives me another how-to that also does not work. I think that happens when it can’t really answer my question, but the people-pleasing drive makes it hallucinate plausible series of steps. Those incidents are quite frustrating, but the successful ones make it well worth it to suffer through GPT’s occasional bullshit attack.
-Medical questions. Examples:
1) Someone I know is getting immunotherapy for cancer, and having trouble tolerating the treatment. I ask GPT for info about novel or overlooked techniques for helping people with the immunotherapy side effects, and fit found several. I skim the articles and none are obviously sketchy, and in all the approaches described did indeed work much better than the placebo. In the course of looking at articles, run across one stating that immunotherapy given earlier in the day is substantially more effective than when given later. I pass this info on to my acquaintance, and their oncologist verifies that this is indeed the case, and now the person gets immunotherapy early in the morning.
2) I want to get Zepbound cheaply by ordering it from a lab selling it as a research chemical, as described in Cremieux’s blog and elsewhere. But I am nervous about whether this approach is really safe, so ask GPT to tell me all the ways it could be unsafe, then help me check to see whether any of them apply to the exact way and exact lab I would be using. Delve into various related matters, and ask the research chemical lab some crucial questions. Eventually decide approach is safe, and am now doing it.
3) Someone I know has a dog that developed big cauliflower-like growths in its mouth. Vet said it was papilloma virus, not a big deal, and growths would shrink on their own over the course of a few months, but until it does dog can’t safely socialize with other dogs. Owner’s distressed. Asked GPT for best current approaches to this problem, it sent me to research showing a certain drug greatly reduces time for the problem to resolve. Skimmed research article and there was nothing obviously wrong with it. Owner got the drug from the vet and now, about 3 weeks later, growths are nearly gone.
-Big loose searches. Someone I know has an impressive skill set but does not like the job where he’s now using it. I ask GPT for a list of organizations of any kind that might need someone with my acquaintance’s skills and has features on acquaintance’s wish list: Helps other people in a pretty direct way, one works as part of a team with a shared goal, can be anywhere in world except for a few specified exceptions, prefers to work in or near a certain country, demands inventiveness, minimum salary. It produces a list of 30 or so organizations, about half of which I’ve never heard of, all of which sound impressive and plausible.
Some safety procedures one should use for superintelligent AI.
1. Only give the AI finite goals. So ask the AI to “earn 1 billion dollars” rather than “earn as much money as possible”. If you asked for as much money as possible it might take over the world and turn everything into money. But with a finite goal, even if it does something horrible it hopefully would stop once it had earned a billion.
2. The AI should self-destruct when completing a goal and be OK with that. Now probably the people who own the AI don’t want to destroy the computer or erase the AI program from it. So instead one should make the AI think of itself as the program being run one time. And if the program is being run again that’s a different AI.
This helps the AI only have finite goals. Otherwise if you ask the AI to earn a billion dollars it might think you will ask for more money in the future and so it might start to work at getting as much money as possible, rather than just one billion.
3. Always give the AI an alternative task. Example: you have an AI that can talk to you, read the internet and trade stocks. It can’t do anything else. You ask it to either earn a billion dollars or to make sure a red button in your office is pushed. You think it is impossible for the AI to push the red button, as it doesn’t have any arms, so it will have to trade stocks and earn a billion. The AI then hypnotizes you into pushing the red button.
If you hadn’t given the AI the alternative task something more disastrous might have happened. The AI might have hypnotized you into kidnapping a billionaire and then hypnotized the billionaire into giving you a billion dollars.
4. The AI has to explain what it is planning to do, and you have to approve the plan before it can do it.
5. The AI can’t lie.
6. The AI can’t hurt people.
7. The AI can’t break the law. But be careful with this, someone could change the law to manipulate your AI. Congress might pass a Law that says “All AI must make everybody in the world the obedient slaves of Congress”.
8. All the countries in the world should be united into one country so they can more easily enforce AI safety standards. I realize it is extremely difficult to unite all the countries in the world, but it is still easier than having lots of countries and making sure all of them follow AI safety protocols.
9. If someone who is not entirely moral is working on AI they should be stopped, or they might deliberately make an evil AI for selfish reasons.
> 2. The AI should self-destruct when completing a goal and be OK with that.
Reword this as "The AI should reset itself upon completing a chat session" and it is a feature we already have baked in to chatbot architecture.
Problem is, the AI labs are working furiously to remove this important safety feature.
Consider 1 and 2.
One way to earn billion dollars is to start doing something useful repeatedly, gain a monthly income and save money. For example, the AI may offer to navigate cars, and will get paid $1 per car per month. Gradually it takes over a region, and collects 1 billion.
After completing its goal it self-destructs.
Then all self-driving cars crash.
There is also some tension between 5, 6, and 7.
I like #2 and 3, and have not heard those ideas before.
I don’t think it is possible to insist that AI observe 5, 6 and 7. In real life some circumstances lead to even the most scrupulous person doing each for the greater good. For ex, consider doctors. They certainly cause pain, and sometimes harm or kill people with procedures that are the most likely to help, but do fail a certain percent of the time. There are similar examples that show why one cannot demand perfect adherence to 6 and 7.
8 and 9 are impossible to do.
Let he who is without sin craft the first drone.
Big fan of option 3, which is reminiscent of an old idea of mine, which was basically to try to hard-code in a value that looks like "press this red button in my office", except more visible - my idea was "paint the Golden Gate Bridge green". If we could make that a core goal (and I'm not sure we can), then an unexpected success on that goal gives us a warning shot - the minimal agent that can paint the bridge probably can't yet defeat humanity, but it seems fairly on the path to doing so.
Less optimistic on 7, 8, and 9. On a meta level, "These are some proposals I've come up with about how to make AI safer" >>> "safety procedures one should use". You don't get to use the latter language until you're an actual master of the craft, which no human is.
Agree regarding 2 and 3. 3 in particular I find pretty neat.
See https://www.lesswrong.com/posts/eS7LbJizE5ucirj7a/dath-ilan-s-views-on-stopgap-corrigibility , which covers some of this.
I hope it’s not somebody saying that ASI is super smart and so would have thought of all the ideas above and prepared in advance some way not to be deterred if a member of our species attempts to implement one. I don’t think someone gets to just make that move over and over.
Why don’t we use controlled parasite exposure as a medical intervention for weight loss, at least in extreme cases?
If you had something that could eat your excess caloric intake every day before it digests *for you*, you wouldn't have to throw up, you could just eat. And you'd be liberated to indulge your love for food to your hearts content.
Imagine if we could selectively breed a type of tapeworm that never grows beyond a safe size (or maybe you just kill and replace it before it overgrows with a simple pill), eats a predetermined number of excess calories per day, and is otherwise simpatico with the human gut.
I think this has significant advantages in theory to GLP-1RAs. GLPs make eating less pleasurable, which for many people is a core ingredient to a happy life. The parasites would if anything make eating *more* pleasurable, because they want the calories as much as you do. And GLPs in the best case only achieve about 44 lbs of weight loss before you reach the maximum dose for someone of my bodyweight anyway. Most obese people are more than that many lbs overweight.
Considering parasites can kill people by starvation in the extreme cases, I'm guessing you can achieve arbitrary, unbounded weight loss: turn someone from 400 lbs to 150 lbs if desired. GLPs can't do that.
It’s cheap, passive, doesn’t require any active thought on the part of the patient. Adherence rates would be high, because you don’t have to dose once or twice weekly; you’d probably just eat your parasite cookie at the clinic and come back a year later (depending on how much it turns out you can optimize the growth and eating characteristics of a parasite using genetic engineering or selective breeding).
Maybe you have to “sterilize it” / take away its ability to shed eggs/reproduce using a drug or by removing its egg producing organs. And okay, maybe that’s hard, but we’ve done much harder things than this with probably much more R&D than would be required to figure out some simple trait selective breeding & neutering surgery.
I get that there’s probably an insurmountable yuck factor here, and no one is going to be pitching this on Shark Tank—although maybe this will be overcome if someone can credibly advertise this as a kind of robotics instead of a parasitical organism, by creating a nano “gut health booster” or whatever that “monitors your gut for excess calories and never lets you overindulge”.
But like, I know of at least two personal anecdotes where a morbidly obese person underwent an elective surgery to lose fat that had about a 50% survival rate—so the FDA can’t tell me controlled parasitical exposure as a weight loss intervention is too extreme/there is no amount of morbid obesity that could ever merit something so reckless.
And hey, they’re already doing this in light of the hygiene hypothesis of autoimmune conditions, right? We evolved to coexist with parasites in the ancestral environment, and without them our immune systems overreact to normal things. The parasite load our gut biology adapted to deal with is not zero, and it seems plausible to me that that’s actually responsible for a lot of “evolutionary mismatch” related illnesses.
One piece I haven't figured out yet is how a Pharma company would actually make money from this, though; can you patent a selectively bred animal? Does someone have a "patent" on the Golden Doodle dog breed? Presumably not, right?
My friend objected that "I think the big issue is, energy has to go somewhere, regardless of who's doing the food absorption. [In a simplified sense] In humans, it goes into fatty deposits or activity. In parasites, it typically goes into growth and reproduction, both of which are limited in your hypothetical. So there's really no way for this tapeworm to consume the calories on your behalf and like, do nothing with it."
So yeah, if the worm has nothing to invest the calories into other than growth, it will balloon in size quickly and have to be removed. But egg production is highly energy intensive; maybe you could treat it with something that would make it so that the eggs it produces are obligate sterile, in which case it would spend the vast majority of its energy producing effete eggs that would be trivial from the standpoint of human health. Normal tape worms take years to grow to huge sizes. (Ofc maybe they'd grow more quickly if given to people who eat way more than the avg person, so they may have to be cycled out on a faster schedule than this.)
I want to read the fantasy version of this where someone figures out exactly how succubi drain "life energy" and makes literally all the gold.
That is an actual plot point in Dungeon Meshi. Succubi saliva contains a "poison" which induces euphoria in victims, which ultimately results in these monsters getting hunted to near-extinction for their fluids. One group of entrepreneurs gets the bright idea to keep the operation sustainable by recruiting for "succubi hunts" and then feeding the recruits to the succubi, providing them with a reliable food source in order to keep their population stable.
I was thinking of a weight-loss center staffed by succubi.
> But like, I know of at least two personal anecdotes where a morbidly obese person underwent an elective surgery to lose fat that had about a 50% survival rate
No you don't.
> GLPs in the best case only achieve about 44 lbs of weight loss before you reach the maximum dose
I don't understand. You take GLPs, you keep losing weight. To lose more, you don't need a higher dose, only more time.
I took GLPs for one year, lost about 15kg, stopped taking it for a year, the weight remains constant. I assume that if I took them for one more year, same dose as before, I would lose maybe 10kg again.
> turn someone from 400 lbs to 150 lbs if desired. GLPs can't do that.
It might take ten years, but I see no reason why not.
The trouble is: how do you control the parasite?
Tapeworms used to be advertised and sold as diet medication in the late 19th Century and early 20th Century. Dieters would intentionally swallow beef tapeworm cysts or eggs, usually in the form of a pill, in the hopes that the parasite would grow in their intestines and absorb a portion of their food, leading to weight loss. That didn't work out so well for many of the dieters.
Pills containing tapeworms have been sold for this purpose, but my understanding is that any significant parasitic infestation will likely make you lethargic and mentally foggy (and possibly crave more food to make up for the deficit) so they don't work as advertised/have undesirable side-effects.
> I know of at least two personal anecdotes where a morbidly obese person underwent an elective surgery to lose fat that had about a 50% survival rate
They should not be doing this either, next question.
You don't know the pressure doctors put on fat patients where diet does not seem to be working. "Try bariatric surgery!"
I think health ministries have told doctors to push this in order to treat The Obesity Epidemic which is taking up so many resources in healthcare (apparently) but I've had to refuse kind urging to undergo such surgeries three separate times.
Operation has good chance of killing you? Decent failure rate? Dependent on success afterwards that you stick to the diet and exercise programme, and if you were able to stick to the diet and exercise programme in the first place you wouldn't need the operation?
Ah, but consider: you're fat! That will kill you anyway, so why not give 50% chance of survival a go?
Bariatric surgery does not have a 50% survival rate, what the fuck?
Okay, if that's not the operation OP meant, I take the correction. It has gotten safer over time, but there's always a risk with severe surgeries of this type.
50% survival rate is nonsense. Bariatric surgeries have a mortality rate of less than 0.1%. For reference, that is lower than the mortality rate for the most common abdominal surgery (cholecystectomy).
Could a specific person have a 50% predicted morality rate? Probably also nonsense. Even big trauma surgeries in critically ill patients have a better survival rate than 50%.
I've had the same thought for years, ever since I watched that one House episode. I think to first order, it'll simply work as described, but I expect your victim ("patient") will get ravenously hungry and be unable to sate that hunger. I also expect he'd have problems with muscles atrophying, along with all sorts of nutrient deficiencies (though this last bit might be preventable by engineering the "diet buddy" a bit more). You could achieve this effect more easily simply through starvation.
Tapeworms for weight loss and maintenance was a plot element in a SF story from the 70s I read ages ago (can't remember name of story or author, but this was the solution for 'we're rich and overeat processed foods but don't want to put on weight').
Haha. There's a chubbyemu video on a case of woman who procured tapeworm eggs for weight loss. It worked but it also wrecked her health as the eggs got into her brain.
The problem with parasites is that they don't stay in one place in the body.
You acknowledge the obvious horror and safety issues, which are plenty to answer your question. But to add to that, you can control the dosage of medicine. You can't control somebody's intestinal parasite.
Also, "this could hypothetically cause unbounded weight loss" is just that, a hypothetical. I have a hard time imagining you can have a parasite consume so much inside a person that it causes them to lose 250 lbs and have that be remotely healthy in any way.
Also, I don't think it's really clear that this would even have the desired effect. Ozempic maks people not crave eating as much. Having a parasite doesn't do this. You seem to be assuming that this means people will eat the same amount, but not feel hungry. But it could just cause them to eat way more to reach satiety.
Also I know people have a lot of arguments around how bad obesity really is, and I won't pretend to have enough expertise to adjudicate them, but I think one might fairly ask, "Is being overweight SO VERY BAD that it's worth feeding a tapeworm hundreds of pounds' worth of calories?" Because that seems pretty crazy. Your only real counterargument is "well, people do surgeries with a 50% mortality rate" (really? would like a citation here) but obviously nobody should do that, either.
"Your only real counterargument is "well, people do surgeries with a 50% mortality rate" (really? would like a citation here) but obviously nobody should do that, either."
Here's a video telling us how safe bariatric surgery is, from a bariatric surgeon. Who looks like she should be getting such a procedure herself, so if it's so great, why doesn't she get it?
https://www.youtube.com/watch?v=Z4NjHRG1AVM
> can you patent a selectively bred animal?
In the US, no, but you can patent an genetically modified one.
The main reason parasites cause weight loss is not that they consume calories, it's because they make you sick and cause unpleasant gastrointestinal symptoms. The amount of calories they consume is minuscule.
They also have other negative effects such as micronutrient deficiencies, damaging organs via several mechanisms (for instance causing obstructive cholangitis and pancreatitis), as well as migrating out of the intestines. Not to mention the risk of spreading them to other people.
Basically this is an awful idea. Its a form of bulimia that is even worse than the usual forms.
And (unless I'm missing something myself), it should be obvious that something like this is going on, because otherwise how is the weight loss happening? If the host was just losing fat, sure; but when they're literally losing weight, it makes no sense to assume the excess calories/atoms are somehow disappearing without a trace into the thing that lives inside them (including when they're standing on the scales).
In theory, the parasites could just have a very inefficient metabolism where they turn lots of calories into lots of waste heat and a very little bit of increased parasite biomass. The amount of waste heat associated with burning say a pound or two of fat in a weak, isn't going to seriously impact the heat balance of a human body. A pound or two of fat a day would be another matter.
But "parasites make you generally sick" is a much more plausible explanation, I agree.
A follow-up question (which is probably dumb, but I think you or someone else here would be good at explaining the answer to me): in this scenario, what ultimately happens to the atoms that the food consisted of? There's no actual matter->energy conversion going on in digestion, so 'atoms in, atoms out' has to hold, and the release of waste heat can't fully explain weight loss/absence of weight gain. Do the parasite and the person simply defecate more (in which case maybe we're back to having noticeable symptoms), or do other means of egress (urination, sweating, exhalation) play a meaningful role here?
Normally, nutrients ingested in the GI track are absorbed into the body, where their atoms are rearranged at need. Some are incorporated into the body's structure (protein, collagen, etc), some are stored as future energy reserves (fat), and most are used for short-term energy production which means they ultimately get converted to CO2 and water, to be exhaled and/or excreted. And really, the protein synthesis etc will be less than perfectly efficient and you'll get still more excreted waste products. But not all of it, so you can get some weight gain in the form of protein and fat.
In the hypothetical (and unlikely) case, those nutrients are hijacked into the parasite's body, where they are used very inefficiently so mostly converted into water and CO2 and flushed out the back end of the GI tract, so no weight gain where there would have been some weight gain. Or actual weight loss, if the parasites snarf up so much of the good stuff that the body has to burn its fat reserves to keep things running.
You're right that "but where to the atoms actually *go*?" and the related "where does the energy actually go", have to be considered any time you're trying to figure out how weight gain or weight loss. And this is particularly easy to get wrong where the gut microbiome (whether symbiotic or parasitic) is concerned.
Thanks -- even if it's not a plausible explanation, it's something I was failing to consider, and I appreciate the gentle correction.
Indeed, I've heard of tapeworm eggs having been used in this fashion, but when somebody up-thread mentioned 250 lbs I thought: "Hang on - a tapeworm can't eat that much. They are famously thin!"
I keep hearing complaints that people waste more time because of LLMs, but my experience is the opposite. There’s no distraction algorithm, no feeds, ads, or autoplay, so I stay in a closed Q&A loop instead of tab-chasing. Therefore, I drift less. Do others here have a similar or different experience?
I definitely waste a lot of time chatting with ChatGPT. I'll present it with long, sloppy arguments presented in inflammatory language, and then when it objects that I am "wrong," I attack its reasoning until it admits but it cannot definitively call me "wrong"
In some vague way, I guess this is practicing debating or something, but the net benefit to me is still nil.
Thank you for your response. How would you compare that to googling the topic you are arguing. Do you think of googling that topic as worse or better? For example, googling the topic may lead you down a rabbit hole or it may lead you to find out something new.
I mostly use LLMs when programming.
The good part is that unlike Stack Exchange, there are no ads, and no attempts to make me read unrelated Stack Exchange questions (e.g. about politics, computer games, etc.).
The slightly bad part is that the LLMs are too verbose, so I easily end up scrolling long texts. (Saying "be concise" only helps a little.) This requires some habits, for example after programming I delete the chat; and sometimes I delete it and start anew even in the middle of work. Otherwise I would drown in the automatically generated content out of which only 10% is relevant.
Self-control is needed, because it is very tempting to ask an additonal question and immediately get a response, but then you have two extra screens of text you need to scroll through.
Interesting practical considerations. Thanks.
When I'm using the LLM myself, it helps me learn and solve stuff quickly (as others have replied). They sometimes hallucinate stuff, but it's usually pretty quick for me to figure out that it was hallucinated.
My biggest time waste with LLMs is when people upload stuff they spit out onto the internet and try to pass it off as original writing, and it sometimes takes me a bit to realize that it's useless slop (not that they didn't make slop SEO garbage before LLMs, but it's more common now).
I've disabled the browser on my phone to make it less distracting (along with no social media, games etc). But having ChatGPT lets me look things up without getting distracted. Having the AI as a mediator between me and the entire internet works really well.
It's like an ultimate Google destroyer, because Google's entire business model is putting advertisements between you and the information you are looking for.
I think it helps me learn more quickly and solve questions more quickly. If it’s important I will then delve into better sources and start doing my own research. I would say it’s good for finding knowledge branches I would have overlooked. It’s good for my analysis paralysis as well for more trivial things. My main issue right now with chatgpt5 is that it’s always trying to give advice and tell me what to do and I just want relevant information always and to make my own decisions.
Ed Nevraumont, author of the Alpha School review, has a blog chronicling current events in the life of his daughter Everest, aged 9. She just won the World History Bee competition. Here’s a snippet from the blog:
>We got back from the History Bee Nationals on Tuesday June 27th. Up until that day Everest was 99% focused on being ready for the competition. But on Saturday June 31st she was scheduled to give a TedX talk on her History Bee experience. Before we got to Nationals she had pulled together a rough script outline and had got some feedback from the TedX team, but she only had four days to refine that script and learn it well enough to perform in front of a sold out audience. . . .Everest took a week off after Nationals to get ready for the TedX talk, and it has resulted in a struggle this week trying to get her review cards back down to a reasonable level. We are going to try and avoid that happening again. She also decided that she wants to compete in the National Geography Bee next year. I am not sure how we are going to balance studying for two topics at the same time.
Here’s a snippet from the life of another highly gifted child. It’s Cory Henry, aged 4, rehearsing at his church.
https://www.youtube.com/watch?v=KI-9DDgzVM0
Which kid would you rather be?
Here's a video of Cory Henry playing music now: https://www.youtube.com/watch?v=CkMW9j75F54
As a kid I did the geography bee (I made the top-10 at the state geography bee when I was in 5th grade). I was also very involved in quiz bowl. I was a voracious reader and I loved my World Book Encyclopedia, but I hated the idea of studying specifically for quiz bowl (I didn't really like to study for class, either). It's much easier to study now because there are so many more resources available, but I can't bring myself to grind out flash cards on Anki or quizlet or whatever just so I know a few more things for trivia competitions. (Yes, there are actually adults who study for trivia.)
I find it hard to imagine that many nine-year-olds really enjoy studying. Learning new things, sure. And maybe, you can gamify the learning enough to keep it entertaining--I assume that's part of why chess prodigies are willing to grind out memorizing so many lines of the Ruy Lopez or the Sicilian defense. But I can't imagine the process (especially when it's parentally-driven) as all that pleasant.
Something of a false dichotomy here. Why a description of the workload of one during an abnormally tight crunch period against a video of the other performing?
Kid doesn't even have homework to do.
That’s a fair point, mostly. (Though Cory is actually not performing. He is at a dress rehearsal, which I think would count as homework. He also worked and practiced at home.). Still, I agree that the situations is not comparable to the one Nevraumont describes in his account of Everest.
But here are some differences between what the 2 kids are excelling at that I think are fair.
-Cory is learning to produce tunes and rhythms. That’s something that it is easy for most kids to enjoy. It’s quite possible that he often enjoys the lessons and practicing he does at home.
-Everest is learning factoids such as dates and royal successions off flash cards. She probably enjoys being congratulated at working so hard at the task, but it’s not easy to believe she enjoys the task itself.
-Cory is learning to do something that, when he’s older, he can do over and over and over, and will embed him among other people who enjoy the same thing. Even his 4 year old version of the activity embeds him among others in a good way. He’s giving pleasure to those present
-Everest is learning a body of material that will only be performed once, at the contest. It will astound those watching, but it’s not clear that that will give them much pleasure. In fact all the kids she beats and their watching family members will feel bad.
-Cory is learning skills that will open up new ways for him to play around and experment. He can improvise, compose, try different ways of weaving his sounds into the mix made by other musicians.
-Everest is learning factoids that may be of some use in understanding and enjoying history — though of course factoids like dates, who succeeded who to the throne, etc. will appear anyway in any book she reads about history. In the future her mnemonic feats might entertain people at a party now and then, though performing them more than once in a while is definitely not a great way to get to know people.
Nevraumont addresses your argument about “learning factoids” in his post titled Memorization, Trivia and Atomic Units for Creativity. You don’t say that children should not be engaging in high level competitions, but your arguments point in that direction so I think the post titled Learning to Lose is quite relevant.
I’m not really defending Nevraumont here, in part because I don’t know whether he is a reliable narrator, but the TedX Talk post you quote from is clearly an update intended to be read in the context of other stuff he has written, not to be a stand-alone blog post. As such, it tells us that Everest is doing a lot of stuff but doesn’t provide the context to see if that makes sense, so if you are going to engage with only one of Nevraumont’s posts, the TedX Talk post is a bad choice.
If I had ever heard anything about Nevraumont that made me admire him, I would take the time to look at the posts you link. If, for instance, I loathed something Zvi had said or done, I would delve more deeply into the subject, doing things like searching his blog for mentions of the topic. But I don't think I owe that to everyone whose behavior or ideas I object to. I read Nevraumont's review, I read his LInked In, I read up on Alpha school and the current head of it. That's enough.
<You don’t say that children should not be engaging in high level competitions, but your arguments point in that direction.
No, I'm not thrilled by the idea of kids in high level competitions, but can imagine that in some circumstances, for some types of competitions, that would work out OK for some kids. The element I'm most bothered by is the time investment. To compete at high level competitions -- in fact even to be allowed to participate in one -- the kid has to devote an enormous amount of time to training. Most training is dull and repetitive, and I don't think it's good for a child to have many hours per week tied up in an activity like that. The whole thing might be worth it for the kid who wins a medal, but for every one who wins there are thousands who train nearly as intensively, and lose. As for the benefits of learning to lose -- sure, kids need to have that experience and learn from it. But that experience not exactly hard to come by. There are plenty of competitive sports experiences and academic challenges available at schools and other local settings, and even the most talented kid is guaranteed not to be the best at all of them. And losing in one of those settings seems to me like about the right size loss for a kid whose age is still in single digits. Losing after a year of grinding training several hours a day -- missing playdates, ordinary games and sports and peaceful downtime in order to grind -- is a very harsh blow to bear.
I'm also bothered by the topic on which the kids are competing. The questions about history have to be stuff like dates, capitals of countries, which ruler succeeded which other. The organizers have to be asking questions where there is a clear correct answer. You can't be asking the competitors, especially child competitors, about something that requires insight -- things like, "what steps, if any, might have kept the war from breaking out?" So the questions must be about factoids, things that are mostly not interesting in themselves. but are just pins to locate events in time and space. Adding a bunch of those factoids to the kid's memory does not seem to me much more enriching than having the kid memorize the first 1000 digits of pi.
And I'm bothered by making competition such a big part of a kid's life. It's a valid and useful mode, but so are a bunch of others. I'm sure Cory Henry had to compete some. He ended up attending Juilliard, I believe, and you don't get in there without winning out over other young musicians in various settings. But while learning musicianship as he child he also was learning something the involved a lot of other modalities -- collaborating, improvising, giving pleasure to others.
Also -- do you think her father isn't deeply invested in her winning and winning big and being known for winning big? He named her *Everest,* ffs! And do you think Everest can't sense how invested he is in her winning and winning big and being known for winning big? By the time I was 9 I could tell on trips to the store whether my mother was worrying about money that day. I could tell when she was annoyed at my father, even if she denied it when I asked her. If she had been privately hoping I would be the smartest or most athletic kid in the state, the country, the world, I would have known that.
Everest knows who she’s supposed to be.
Yes, the questions must have a clear correct answer. But there are a lot of ways to do that without resorting to complete minutiae. I don't know all the particulars of all the various History Bee competitions, but I know they have a team-based quiz bowl style format, which uses toss-up questions (the ones that you use a buzzer for) that are designed to be pyramidal; that is, the hardest clues come at the beginning. So you can reward someone with deeper knowledge of George Washington or the Polish-Lithuanian Commonwealth or Diocletian. And you have to have some understanding of basic facts in order to grasp the broader concepts. So I think knowing things about history is vastly more important and interesting than knowing the 789th digit of pi, which is purely a memory exercise and has roughly zero practical use.
> No, I'm not thrilled by the idea of kids in high level competitions, but can imagine that in some circumstances, for some types of competitions, that would work out OK for some kids. The element I'm most bothered by is the time investment.
I agree.
When I was a small child, people encouraged me to try chess competitions, because I seemed smart, and chess was the archetypal smart people's game. But I soon realized that this was a trap: yes, it helps in chess to be smart, but it still requires enormous amounts of *time*, and the chess skills are practically useless for anything else. So I refused.
I did math olympiads though, because I enjoyed doing math. I did not specifically prepare for the competitions -- I just tried to solve all kinds of mathematical problems I found, and the victories at competitions happened mostly a side effect of that. I knew kids who were specifically studying for the math olympiad, but I was happy I wasn't one of them, and I even considered it a kind of cheating (similar to when you try to memorize the right answers for an IQ test).
> The whole thing might be worth it for the kid who wins a medal, but for every one who wins there are thousands who train nearly as intensively, and lose.
This is true, but I find it suspicious that people usually use this as an argument against knowledge competitions, and almost never as an argument against sport or art competitions.
I guess the idea is that sport and art are valuable per se, so the kids who spent their afternoon practicing sport and failed at the competition, still did something meaningful. From my perspective, doing math is also meaningful... and perhaps for someone, so is memorizing historical facts.
It probably means that people should only compete at things they consider intrinsically useful/enjoyable anyway. It sucks when ambitious parents are pushing their children to do that. But some other children do find the activities enjoyable. Is there a solution that would be good for both sides?
Child 2 looks happy and resplendent on the keyboard. I dunno. As long as Everest is happy and engaged and not stressed, I don't mind too much. But I'm wary of parents trying speed run childhood for the sake of genius. Children are special but adulthood is great churner of mediocrity.
I don't know if it's my limited selection of examples but genius kids rarely sustain the momentum into adulthood. i see many kids who graduate college early only to be another banker or another engineer. It feels disappointing to speed up your childhood only to end up another worker bee. Feels pointless but that's just my sentiment.
Graduating college 1-2 years early is approximately as impressive as becoming a banker or engineer. I think you're simply overestimating what accomplishments in childhood are indicative of "genius."
The reason why speedrunning childhood is good (and this will vary by personality, of course) is that childhood sucks. Who doesn't want to be independent earlier, lead your own life earlier, if you could?
My childhood circumstances were not ideal, but my childhood sure didn't suck. In fact I'd say that period was the happiest and most engaged I've ever felt. I don't think my experience is unusual. Look at kids playing -- most of their faces are lit up with joy and excitement.
Of course kids have fun, but adults have so much more and better fun.
Consider food as a small example. When you're a kid, you can have a great time eating some candy, if your parents let you. When you're an adult, you can still have a great time eating some candy, whenever you want, but you will also have the palate, money, hand-eye coordination, and autonomy to enjoy all kinds of food. You can fly to Belgium and sample fancy chocolate if you want. Even if you're not very wealthy, there's way more stuff you can enjoy.
The whole world is like that. There is so much beauty that most kids don't yet have the capability to appreciate.
You see freedom, I see the tedium of responsibilities and work, and then add on taking care of kids and health issues and aging parents. Yeah sometimes I wish I was 14 years old again, with nary a thought on paying rent or affording health insurance. Parents were my loving Gods whose most annoying demands that were that I clean my room. Now there are no loving affectionate Gods who care for me (May the Gods afford me a billionaire gentle husband, amen!). But there's a cold world demanding that I pay taxes and would not spare a thought if I don't survive the next round of layoffs.
Honestly I like it better being an adult for reasons you lay out, and also mostly because my parents sucked. However, I only came to like adulthood after I have had earned a sufficient level of economic success. Supposedly I am still toiling under a job that barely pays my rent, you bet, I would be weighing going back to be under my parent's cruel thumb just to offload the economic anxiety.
Yeah. Similar here (less dramatic version thereof).
In theory, I have the freedom to do... anything. And maybe some more talented version of me could find a way to enjoy these opportunities.
But the actual version of me spends 8 hours a day at job, trying to survive in chaos that makes me entirely mentally exhausted when the work is over. Then it's care of kids, various domestic tasks, various random problems. An hour or two left to do things that I consider meaningful, but I am usually too tired to enjoy them, so I just read something online or watch a movie, and go sleep. The next day is the same. So is the day after that. The last decade was like that.
My childhood wasn't ideal, but I could spend an afternoon relaxing and reading a book, without being interrupted hundred times, and without feeling guilty that there are dozen other more important things I should be doing instead.
As a child, you don't have the freedom and money. As an adult, you don't have the time. As an old person, you don't have the health.
> Who doesn't want to be independent earlier, lead your own life earlier, if you could?
Most of the milestones of adulthood independence, like being allowed to drive a car, are gated on reaching a certain age rather than just finishing your education.
In the US, most people graduate college at 22 and can drive a car by 16. You can finish your education six years early without that being an issue.
New music recommendation subthread.
I just discovered "Cannons" and I love them. It's very smooth, like R&B, except everyone is white and they're singly sadly about the all the sex they didn't have and also there's lots of synthesizers and spaceship noises.
Check out "Fire for You" and "Bad Dreams"
https://www.youtube.com/watch?v=u4HQevc9JE0&list=RDu4HQevc9JE0&start_radio=1
https://www.youtube.com/watch?v=Fz4axHOyccQ&list=RDFz4axHOyccQ&start_radio=1
Dennis Lloyd might be the greatest artist I've ever heard. Every song should be absolute garbage. Wistful white guy with long hair sings about California while playing his guitar. I should abhor this and yet I'm listening to it, right now, for like the 6th time today. It's like kimchi: every element is absolutely disgusting and yet the final product is delicious.
Check out "Oceanside", "The Way", and "Nevermind".
https://www.youtube.com/watch?v=naW6-WxmMiU&list=RDnaW6-WxmMiU&start_radio=1
https://www.youtube.com/watch?v=tseh8G4bx_c&list=RDtseh8G4bx_c&start_radio=1
https://www.youtube.com/watch?v=y_jXhg4Tqck&list=RDy_jXhg4Tqck&start_radio=1
Nice.
(Not a native English speaker, so I mostly do not hear the words if I do not focus on them and instead e.g. read ACX comments. The music is nice.)
If you haven't listened to Chris Thile's Bach albums, you really really should. Possibly the best to ever do it on the mandolin, one of the best living musicians from a technical perspective, and definitely the best I've ever seen live.
https://www.youtube.com/watch?v=NsPKkAXFVn8&list=RDNsPKkAXFVn8&start_radio=1
For your reconsideration: "Sadly, Porn (But Reviewed Good This Time) from the 2024 ACX Book Review. (1) (2)
This was submitted but never got out of the initial rounds for several excellent reasons: it's like 90 pages long, far too clever for its own good, and it's trying to replicate the Last Psychiatrist style which is...not recommended. Still, this is one of the best distillations of Lasch/Last Psychiatrist/Samzdat-Lou Keep I've seen and it's still in my brain two months later.
This badly, badly needs an abbreviated version. However, I would recommend skipping from p 131 (where the essay starts) then reading pages 140-151, then skipping to page 182-184. (3)
Golden excerpts:
"The Last Psychiatrist used to give an example of how psychiatrists fuck up their kids; they want to be rational, modern parents. They have the benefit of a psychiatric education, after all. This means they don't shout at their kids, even when they do wrong, like bullying another kid. They will shout at them when their frustration boils over as it must do after repeated wrongdoing, which is particularly likely to happen when the child does something that inconveniences them as opposed to something that's morally wrong.
The young child only responds to affect, to the emotion you're giving off. He hasn't internalised the sense of others being people, so when you calmly explain why he shouldn't do something, he doesn't get it. He doesn't interpret it as a real punishment. He does get it when you shout at him, but now you're doing it over something arbitrary, like spilling some milk.
What happens next is the child cries and the parent realises what they've done, and then they turn around and console the child. This is the fatal step.
What the child has learned from this process is that:
Your authority is arbitrary
You can be placated for them to get what they want
All they need to do is figure out what will placate you.
A harsh, even abusive parent who merely beats their kids for no reason is sure to give them some kind of pathology, but they won't get the second and third parts so this won't create narcissists. It's this unstable emotional dynamic: at once nurturing then punishing without a predictable underlying logic that could allow them to develop a clear understanding of right and wrong.
Imagine someone who is going through their adult life re-enacting this dynamic, they see themselves dependent on this omnipotent power that has the power to dictate their life and they have to behave with this in mind. The real problem is that facing up to this consciously would involve living with a constant sense of anxiety. It would be too much for them to bear, particularly for the fragile psyche of a child. So they have to find a way to determine and accede to the demands of the omnipotent power whilst avoiding the reality of this fact: this can only be done through a colossal amount of doublethink.
"
"
The modern life of individuals is characterised by living within the shadow of a huge number of large bureaucratic organisations. We are born in hospitals, learn in schools and universities, work for corporations, get incarcerated in prisons, and finally return to the hospital to die. In our leisure time, we consume products created by vast industrial conglomerates around the world, and the cultural products that we gain meaning from are creations of an industry as large and as powerful as any manufacturing concern or government agency. The guiding philosophy of all this, in its various forms, tells us that these organisations exist for our benefit, and certainly many of them are staffed with generally well meaning people. But this is a unique situation in the history of the world, and this state of being has deep ramifications for the kind of lives people live.
The relevance of the previous sections is that these institutions treat us like Narcissus’ parents: they foster dependency without development, they protect us only intermittently, and we labour under demands which are rapidly changing and inconsistent. Lasch terms this “Paternalism without Fathers”: parenting without love.
"
"
We should frame an important question; is Narcissism bad?
Bad for who? Narcissism is adaptive to the conditions of modern life, even helpful to the individual, although coming at the expense of their psychic wholeness. Systemically it’s a problem, which just means it's a problem for other people, not such much for the individual. They’re just playing by the rules of the game.
"
"
The consequence of this; your job is not to do your job. Your real job is to get your next job, your promotion. Maybe you could get this promotion by performing really well in your role. Except, your performance needs to be recognized as such by your superiors. Therefore, what is really important is to be noticed to be doing well. After all, the role you will be promoted to presumably has at least some different responsibilities to your current one, so your performance in one role is only a partial match to the other. Indeed, the core responsibilities of your job may be an active impediment to your success since they take away your attention and energy from the important work of getting promoted. Possibly, doing too well in your core role will actually retard your promotion because they don’t want to lose the contribution that you’re making in that position. And this is not to mention the more obvious and unsavoury aspects of promotion culture; brown-nosing and conformity.
The kind of person who will succeed best in this environment is the one who can recognize this intuitively, someone who can devote their whole life to the maintenance of appearance, with reality and action as an afterthought.
"
"
The kind of social narcissism we’ve described is a product of living under the constant authority of bureaucratic institutions, entire lives lived with no other way of being. People identify with power as a defence against the acknowledgement of their own potential helplessness in the face of these powers. But the anxiety of this state doesn’t disappear, it just gets redirected.
My background is on the Left, that’s the language that I speak. I know so many people who think “free speech” is bullshit, they want the government to take more steps to suppress those with views they see as dangerous. These people are also supposed to be “radicals”, people who claim to want to overthrow the system. Of course, they don’t expect that power to be used on them, only on their enemies. They would rather fight fascists than the government, it’s a lot easier I can tell you. However stringent their rhetoric is they always give succour to the state of things as they are - and their enemies are no different.
Narcissism allows people to take their servitude and interpret it as freedom. The problem can never be recognized as the system itself; it must always be redirected outward onto something else, some alien force that is ruining this world. The harsher and more intrusive the authority becomes, the greater the anxiety, the more you fear the other that you project as fucking you over. The end point of this line of thought is that the only conclusive solution can be the elimination of this force: the logic of genocide.
""
(1) https://docs.google.com/document/d/1GYQw3pgvhi7hqOVR-Ql629Q_8thbyHe8sSRy5voyt30/edit?pli=1&tab=t.0#heading=h.cdezdtonc8cn
(2) Thanks to @thefance for recommending this awhile back.
(3) Yes, really, skip the next 30 pages. This was not a winning essay for a reason.
Thank you!
I agree that trying to replicate the Last Psychiatrist style is bad. Some strong illusion of transparency seems to be in action; I could never decipher most of the text. Maybe I am too stupid, or just not immersed in the same bubbles.
> This means they don't shout at their kids, even when they do wrong, like bullying another kid. They will shout at them when their frustration boils over as it must do after repeated wrongdoing, which is particularly likely to happen when the child does something that inconveniences them as opposed to something that's morally wrong.
100% this. I have already seen that so many times, but couldn't put it in words so well. I just noted that if you fail at a task that a monkey could perform better by following its instincts, you should be a bit skeptical about your wisdom.
I would add that even by the parent's proclaimed metric "less shouting is better" this behavior fails. Shouting at kids when they do something wrong at least has a chance to make them do the wrong things less often, which leads to *less* shouting in long term. Shouting at them randomly when frustrated... seems to *increase*, as the behavior does not improve, and the frustration accumulates.
> All they need to do is figure out what will placate you.
I'd say they need to figure out what makes you unwilling to start a conflict (e.g. knowing that it is guaranteed to escalate) or what makes you feel guilty if the conflict already started anyway. Different kids will probably use different strategies.
> The modern life of individuals is characterised by living within the shadow of a huge number of large bureaucratic organisations.
One problem is that institutions will try to treat everyone the same, but of course people are different and what works okay for some will fail completely for others. And the institutions are unable to process feedback (usually because the person who makes the rules is not the same as the person facing the failures).
> Indeed, the core responsibilities of your job may be an active impediment to your success since they take away your attention and energy from the important work of getting promoted.
Yes and no. As a software developer, a certain level of competence is necessary to get certain jobs. But taking care of the quality of code beyond that level is an impediment.
Similarly, school will require some knowledge to give you an A. It's just the knowledge beyond that level that is wasted.
> Narcissism allows people to take their servitude and interpret it as freedom.
https://www.youtube.com/watch?v=CLcwCRJ31q8 -- not sure if it makes sense without translation or subtitles: a scene from a traditional Slovak movie: A nobleman is beating a peasant, an outlaw tries to protect the peasant, but the peasant sides with the nobleman against him. "It's none of his business, your highness. Keep hitting, keep hitting, your highness."
I am curious if anyone else is following inkhaven remotely, and trying to maintain a schedule of daily or near-daily posts? It is so far very difficult to balance with a job and makes me much more impressed by frequent bloggers! Curious for anyone else's take
There's a whole Halfhaven challenge going on virtually where you're challenged to write 30 blog posts in October and November. There are around 20 participants, including myself.
Wow, yeah. That sounds like it would’ve been a much better idea!
me! I'm on day 21/30 of daily blogging. It's great.
I'm a student though, don't think I could do this if I was employed.
wish I was at inkhaven though
I have been doing neither, but have been writing on 4TheWords since late August, which has a daily minimum of 444 words to maintain a writing streak. Not the same requirements as posting a completed thought per day, but I'm quite happy with the results so far; I've managed NaNoWriMo's 50,000-word pace for the last two months and should clear it for this month during the next weekend.
A whole lot of my post-work day is now writing.
Yeah this is basically my experience.
Interesting! Its the first time I've hearing about inkhaven. I think daily posting is possible if you don't write thousands of words per day - putting out a few hundred words daily with the occasional longer piece feels very possible to me. And there's definitely value in putting out a few hundred words every day.
It's amusing that the BBC faked Trump saying something he didn't in close to to what Scott used as an example a few years ago of something you can trust the media never to do.
Is "faked" the right word here? He said both statements, and it's not clear that his comment about "fighting" was innocent of violent connotations. Especially in retrospect where he has issued pardons for many who did engage in violence.
I think the way they spliced it together is a reasonable interpretation of what Trump meant, it's just that it's not *certain* that it is what he meant, and so it's misleading for them to splice it together to make it seem like there was only one possible interpretation
Call it what you like, but they changed
> we're gonna walk down to the Capitol and we're going to cheer on our brave senators
into
> we're gonna walk down to the Capitol and we fight.
I think "faked" is reasonable.
Nvm, I watched the clip and it was clearly trying to decieve. The edit is placed such that your meant not to notice it.
I still think it's a reasonable interpretation that Trump intended and wanted violence that day and so I don't think it's a dishonest interpretation to imply his statement about the need to 'figjt like hell or you won't have a country anymore' was a call to violence. But trying to edit a time skip so that you don't notice it is dishonest regardless of the context.
I’m amazed that you can admit it’s outright fraud and still try to justify it because you agree with the narrative. This is how we get authoritarian regimes.
I'm not trying to justify it. I said the edit is dishonest regardless of the context.
As a separate statement I am saying that it is a reasonable interpretation to say Trump wanted violence on Jan 06 as he allowed the storming to continue for 4 hours before taking action (tweeting) while politicians were hiding for their lives. But this does not excuse or justify the edit. The people responsible should be fired.
That's fair, I was too harsh. Thanks for replying.
I see a more reasonable interpretation where he wasn't convinced there was "storming", but rather that there was an enormous crowd of protestors who weren't particularly violent (practically everyone outside the Capitol was just there to protest) and he didn't have a reliable source telling him exactly what was happening inside as it happened, and several _unreliable_ sources motivated to exaggerate anything his supporters were doing (as the BBC finally recognizes).
Meanwhile, everyone knows that the Capitol Police were critically understaffed that day; the only way Trump could have physically prevented entry into the building would have been to somehow produce scores of additional FBI agents within five minutes and have them somehow work their way to the building perimeter, or order the Capitol Police that were there to resort to lethal crowd control. Hopefully I don't have to convince anyone that these would have been absurd measures.
If a massive wave washes over you and floods the beach behind you, no one would contend that you're "letting" the wave do that.
Instead, he sends multiple messages calling on protestors to avoid violence, reminding them that theirs was supposed to be the party of law and order, to "respect the law and our great men and women in Blue", etc. Why would he do that if he actually wanted violence? Reverse psychology? No one's that crafty outside of a movie.
I think you might be slightly misremembering that post.
> I will be making a nitpicky technical point. I’ll start by making the point, then explain why I think it matters.
Scott did not write that this is "something you can trust the media never to do." If you revisit the article you'll see that he's extremely careful to caveat:
> the media rarely lies explicitly and directly
Note "rarely", my nitpick is that rarely is not never. This may matter because there are always extreme examples where bright lines are crossed and so hedging with "rarely" let's Scott fend off examples of egregious line crossing that could be raised not would not, in fact, diminish his point.
The panorama example is egregious, that two BBC execs have resigned over this speaks well to the ideal of holding the BBC to a high standard, and the naive/hopeful/optimistic may hope that it suggests that this level of deception will be less tolerated in future.
However... I also think it's a really great example of exactly the tension Scott's article describes. The quote in question was a contraction of two separate bits of trumps speech edited together to give a distorted impression of what trump said while still only containing his words. Descriptive? Absolutely. Politically motivated? Almost certainly. A great example of how the missing context is essential in determining the difference between misinformation and true information? Yep. Exactly the point that the second half of Scott's essay addresses.
I WAS misremembering something: it was the Bounded Distrust post, not the Media Very Rarely Lies one.
Aha, in that case, my apologies for the nitpicking.
I went back to the Bounded Distrust post, this seems the most succinct relevant quote from it:
> There are lots of cases where you can’t trust the news! It sucks! It’s completely understandable that large swathes of people can’t differentiate the many many cases where the news lies to them from the other set of cases where the news is not, at this moment, actively lying.
I understood the point of the post, thanks. I agree they're not the knaves of logic puzzles, and that skilled liars can deceive without uttering a single provable false statement, always hedging with "experts" and "unnamed [read as: 'imaginary'] sources."
The more relevant bit is the opening, where he describes a hypothetical scenario with Fox, as a way of establishing a loose upper bound on the level of distrust one should have for the media:
>The level on which FOX News is bad isn't the level where they invent mass shootings that never happened. They wouldn't use deepfakes or staged actors to fake something and then call it "live footage". That would go way beyond anything FOX had done before.
I'm saying the BBC stitching together clips of Trump saying some phrases about an hour apart to create a seamless video of him saying something he didn't IS at the level of deepfakes and staged actors.
My view is on the existence of such an upper bound is that the correct attitude is something like "You don't hate journalists enough. You think you do but you don't."
Hey, uou never answered why you repeat White identitarian talking points. I asked you maybe 4-6 days ago and you never responded. It was when I raised the issue of white genocide. A concept you defended.
I'm still curious about your motivations. Do you believe in some kind of quid pro quo where if you support White conservatives, they will support you, or something like that? Does it have something to do with building an anti-Islamist Coalition or something? Please give me some insight. I'm trying to figure this out.
I didn't think even you considered a serious question meriting a response, but sure. Since you ask it in an unrelated thread, I'll respond here.
First, the word "genocide" has devolved into meaninglessness or legalism, so I don't really care whether the definition you want to adopt includes or excludes intentionally-engineered sharp demographic decline. The consensus among international experts or whatever doesn't sway me in the slightest.
Second, this "White identitarian talking points" thing reeks of when there was a push to make the OK gesture 👌 a White supremacist hate symbol. I reject this attempt to smear ideas by association.
But if you want some deeper psychological drive, I find it deeply irksome when powerful organizations suppress dissident views. That should be enough to explain what see as mysterious about my motivations.
> BBC stitching together clips of Trump saying some phrases about an hour apart to create a seamless video of him saying something he didn't IS at the level of deepfakes and staged actors.
Deepfakes and staged actors seem obviously worse to me?
I'd put all 3 above a bar of "unacceptable" and it's good they are getting fired for it.
But being absurdly misleading is still a notch below plain fabricating information.
Imagine a deepfake that says something *very similar* to what the person actually said. Would that be more acceptable than a deepfake that says the opposite of that?
From certain perspective, yes, it is less misleading for the viewer.
From another perspective, the difference between the two is negligible compared to the horrifying fact that the line "you shall not show deepfakes on TV" was crossed. That has way more serious long-term consequences than whether the deepfake was realistic or not.
Stitching two sentences together is a poor person's deepfake.
A matter of taste what you consider worse, I suppose, but what do you consider to be the information the BBC conveys, which may then be judged on the basis of whether or not it's fabricated? Only that every individual syllable being broadcast was actually spoken by Trump sometime in his life and recorded faithfully with a real-life microphone, with the order in which they're strung together being, say, a mere artistic choice?
There was a recent example of a "deepfake" which I think would be interesting to compare this to: the Senate Republicans posted one of (an AI version of) Chuck Schumer saying something he ACTUALLY SAID. https://x.com/NRSC/status/1979187237720924195
So in this case, the information I'd say this video conveys is entirely true. It just so happens that the pixels and audio are synthetically generated (a fact which is also clearly disclosed on screen), yet in better accordance with the truth than the more "real" pixels and audio of the BBC.
It is pretty egregious and massively breaks a norm. Hopefully everyone involved are to be fired.
Jan 6 has broken a few minds. It was a nothingburger really and amounts to hysterical wolf calling. Trump ii is a danger
This is like declaring that 9/11 was a nothinburger because someone edited a speech by Bin Laden.
This!
Comparing Jan 6 to 9/11 is part of the nonsense here. A few fat guys walked into a parliament, unarmed. One was shot. The other deaths were old age related. There was no coup attempt. There was no way the election wasn’t going ahead.
One was shot after lunging through a broken window in the final door barring her from the Congresscritters, with cops in defensive positions behind it with guns drawn. I mean yeah, 9/11 is a ridiculous comparison, but let's not unnecessarily undersell it.
Honestly I think Jan6 is the single thing most distorted by hyperpartisanship. I'm not even sure what exactly I think about it, it seems so complex. It seems to me there was a mix of really disturbingly serious people (both in the crowd and in the legal/procedural wrangling) who wanted to overturn the election by any means necessary, and a bunch of goofy Trump supporters with lives too comfortable to even consider participating in a coup, who really did just want to peacefully protest (probably including some of the ones who ended up inside the building). And I think Trump didn't even know what he wanted, or was vacillating. But now Jan6 is either a violent coup attempt or an almost entirely peaceful protest. Argh.
Yea.
What the BBC did was egregious and the British-traditional correct outcome has belatedly been achieved (the top two people in charge have lost their positions and their careers).
What Trump said to that crowd, _minus_ the "fight" line, absolutely was incitement to try to forcibly overturn an election result. And literally today he's continued handing out blanket pardons to everyone who was on his side in the whole effurt to overturn that election result "by hook or by crook" as the old saying went. The analogy to 9/11 works just fine unfortunately.
Alright, I've just finished reading (well, skimming) the entire damn speech. It is, in typical Trump fashion, very long and rambly.
Here are the only parts I found which are actual calls to action on the part of the audience.
The first one, near the beginning:
"I know that everyone here will soon be marching over to the Capitol building to peacefully and patriotically make your voices heard"
The second one, at the end:
"So we're going to, we're going to walk down Pennsylvania Avenue. I love Pennsylvania Avenue. And we're going to the Capitol, and we're going to try and give.
The Democrats are hopeless — they never vote for anything. Not even one vote. But we're going to try and give our Republicans, the weak ones because the strong ones don't need any of our help. We're going to try and give them the kind of pride and boldness that they need to take back our country.
So let's walk down Pennsylvania Avenue"
Is there anything in there that I'm missing?
"We have come to demand that Congress do the right thing and only count the electors who have been lawfully slated, lawfully slated."
Congress in reality _did_ "only count the electors who have been lawfully slated" which of course was what Trump was asking the crowd to _prevent_.
As everyone attending that rally understood, what he wanted them to demand was that Congress instead count lists of electors from several states that had been named by people other than the election officials and/or elected officials in those states having the lawful duty and power to slate electors.
That the crowd understood that, and had also gotten the gist of what was meant by "demand", became clear when the VPOTUS declined to entertain those bogus slates of electors and the crowd responding by chanting "hang Mike Pence".
Incitement to try to forcibly overturn an election result.
The problem is that the question is the violence, not whether it was correct to overturn the official result. Setting aside whether Trump actually believed this, you have to acknowledge that there was a large chunk of the country that thought the official result was fraudulent. You are skipping to the part where you are right and they are wrong, skipping over how they're supposed to address the massive injustice they perceive. Even if they're wrong, it's unavoidable that they're going to respond in some way. And Trump told the crowd to do it through peaceful protest.
Note: Trump wasn't present for the "hang Mike Pence" chant - it's important not to conflate that into his speech to the crowd, because letting that go would obviously be supporting violence. If you have blurred that into your understanding of his speech, that might be warping your perception.
He never told them to prevent it violently. What the protesters did of their own volition is not his problem.
PhilosophyTube published a video about birth rate decline. I'm interested in y'all's thoughts on it!
https://www.youtube.com/watch?v=AIDnr646tLA
My thoughts are that it's a fifty minute video which could have been a three thousand word essay. If it had been a three thousand word essay I probably would have done it the justice of spending two minutes skimming it, but I don't have time to spend fifty minutes on a video just to see if there's anything interesting in there.
That's perfectly reasonable, spend your time as you see fit :)
(Though I think PhilosophyTube's videos in particular would lose a lot of their essence if you just read the transcript, for better or worse.)
There should be a website that uses an LLM to generate short written summaries of long YouTube videos. (In a perfect world, YouTube itself would do that.)
> There should be a website that uses an LLM to generate short written summaries of long YouTube videos. (In a perfect world, YouTube itself would do that.)
This is actually one of my primary uses for the Gemini LLM, which alone among LLM's has the rights to summarize youtube videos, due to Google owning both.
It will give you a summary with linked timestamps to the things it's talking about in each part, I find it invaluable, because I refuse to click on a good 98%+ of any video links people send me.
Googling "youtube video summary" actually gives me many results! I have no experience with them to share however
Hm, not too bad.
My problem is that the summary kinda mentions the thesis ("aging demographics is actually not a problem") but does not contain an explanation. Not sure whether it means that there is no explanation in the video, or because the bot did not consider it important.
https://krisp.ai/youtube-video-summarizer/
Elon Musk and Birth Rate Concerns
- In 2022, Elon Musk acquired Twitter and has since expressed concerns about declining birth rates, which he believes could lead to a population collapse.
- Musk has been vocal about this issue since at least 2017, suggesting that the world is heading towards a demographic crisis that many ignore.
- This concern has been framed as a moral panic, reminiscent of previous generational fears regarding population decline.
Critique of Demographic Analysis
- The discussion includes an analysis of an article by Fred Pierce, which highlights Japan's declining birth rate and its implications for economic stability.
- Pierce argues that Japan's aging population is a significant problem, linking it to economic stagnation without providing substantial evidence for his claims.
- Critics argue that Japan's economy has remained robust despite demographic challenges, indicating that economic performance is not solely dependent on population growth.
Population Aging and Economic Implications
- Aging populations raise concerns about the old-age dependency ratio, which is the ratio of older individuals who rely on the working-age population.
- The narrative suggests that increasing numbers of retirees will burden younger workers with higher taxes to support pensions and healthcare.
- However, historical data shows that economies can adapt and thrive despite demographic changes, challenging the notion of inevitable economic decline.
Historical Context of Birth Rate Fears
- Concerns about declining birth rates are not new; similar anxieties have been expressed for decades, including in British Parliament discussions from the 1920s and 1930s.
- Historically, predictions about dire economic consequences due to low birth rates have often proven to be incorrect, as economies have continued to grow.
- This historical perspective suggests that current fears may be overblown and rooted in a misunderstanding of demographic trends.
The Role of Economic Factors in Reproductive Choices
- The discussion shifts to the real barriers affecting birth rates, which include economic factors such as financial stability, housing costs, and job security.
- Research indicates that many people desire larger families but are constrained by the high costs associated with raising children.
- Improving economic conditions, access to healthcare, and support systems could address these barriers more effectively than simply promoting higher birth rates.
Critique of the Moral Panic
- The moral panic surrounding birth rates is critiqued as being exploited by political and media narratives that may not reflect the complexities of demographic changes.
- This panic can lead to harmful policies that prioritize individual responsibility over collective solutions to societal issues.
- The conversation emphasizes the importance of addressing the underlying social and economic issues rather than succumbing to fear-based rhetoric about population decline.
Future Discussions on Reproductive Ethics
- The video concludes with a teaser for a future discussion on the ethical implications of having children in the current socio-economic climate.
- This upcoming dialogue aims to explore the responsibilities and considerations that potential parents face today.
- The creator invites viewers to engage with these topics actively and consider the broader implications of reproductive choices in society.
Or the person who wants to post and discuss the video link could summarize it manually.
That's fair, though going beyond the fact that it's about birth rates and that she's critical of the topic (which you notice immediately when watching), I wouldn't want to say much more here, so as to get people's own immediate impressions :)
I agree, but it is a repeated situation, so it seems worth to automate. Another advantage of an automatic summary is that it would be (probably) impartial.
I haven't seen that one yet, but I've watched several of her others.
I found "Identity" very powerful and helpful to me when I first watched it, a few months into my own transition. She's on my list of people whom I probably owe the price of 1-2 therapy appointments because of it. I also appreciated her video about navigating trying to get diagnosis and HRT from the NHS for similar reasons, although it was less directly applicable to me as I'm an American with good benefits from a very LGBT-friendly company and thus face much fewer obstacles to obtaining gender-affirming care than she did.
I watched a bunch of her older videos from her "grad student nerding out about her coursework" era and generally really liked them.
I watched her video about effective altruism and FTX and wound up liking it a lot better than I'd expected to from the clickbait title. There was a fair amount I disagreed with in it, but it was decently nuanced and nothing jumped out at me as egregiously wrong or unfair. I'd watched it specifically as a Gell-Mann test (i.e. do I still like her stuff when she's talking about stuff I already know a fair amount about) and I think she passed handily.
Her video from a couple years ago about Stoicism was entertainingly presented but the content was disappointing, especially the criticism of Stoicism as a modern philosophy. She relied on two prongs:
1. Roman Stoic writings contain a lot of obnoxious assumptions about class and gender. Which I suppose is probably true (or at least a defensible claim) as far as it goes, but I found it relatively low-effort since that's much more a general critique of Classical Roman culture than a specific critique of anything peculiar to Stoicism, and I don't think she made the case that the assumptions were load-bearing parts of the philosophy rather than mere baggage. Here, though, I do give her some credit for acknowledging the existence of modern Stoic philosophers who have attempted to strip out the Roman cultural baggage.
2. She claims that Stoicism is extensively biased towards a posh and powerful perspective and advocates complacency and compliance in the face of injustice. This seems to completely overlook the existence of Epictetus, one of the major classical Stoic writers, who was born a slave and devoted a great deal of thought to the question of how to live a moral life from a position of powerlessness in an unjust world. She also seemed to overlook Epictetus at a different point in the video when mentioning that we know classical Stoicism exclusively through Roman sources despite it (originally) being a Greek philosophy: Epictetus was Greek and so was his pupil Arrian who is our main contemporary source for Epictetus's teachings. I suppose this could be partially salvaged in a "technically correct is the best kind of correct" way by noting that Epictetus and Arrian lived in the Roman Empire and I think both of them spent a fair amount of time at the Emperor's court in Rome.
Thanks, very interesting! I'm in a broadly similar boat in that I'm a long-time viewer and generally enjoyed many of her videos and strongly disagreed with few. This one seemed very off to me, but I also don't know a lot about the topic, so I was curious to hear the impressions of the people here.
I love stumbling upon your elaborate comments, btw! I enjoy the way you write. Would love to read some posts of yours if you ever write some :) Just wanted to share that with you
Thank you! I've actually been thinking semi-seriously about starting a blog again. Or a YouTube channel, as there seems to be a market for long-form explainer video essays made by sapphic trans women with three-syllable first names and one-syllable last names.
In the meantime, if you're looking for more of my writing, I did the "Deathbed Ballads" review in this year's non-book review contest and the "On the Bondage of the Will" review in last year's book review contest.
From the video:
""Some economists say that Japan's repeated recessions since the 1990s are due to the burden of ever more oldies." - But Pearce provides no citations for this claim and trades on several stereotypes about older workers."
If this is the kind of logic you find convincing I don't know what to tell you...
Not saying I was convinced by what she said, just curious what you folks think
I think Leftists snarking at people wanting bigger families, valuing family, or for people to be concerned about the birth rate is always dumb.
Many Leftists are all in on the idea of "community." Community building, community powered solutions, community art, community participation, etc. Do they not realize that communities need futures and that one day all the current members will be dead?
The ability to purchase the labor of younger abled bodied workers in your old age, to help you continue to live when you lose the ability to care for yourself, is a trick of capitalism. It distances people from realizing that the old need the young as much as the very young need caretakers and parents. None of these people seem to ask, Who will be there when I'm old?
< I think Leftists snarking at people wanting bigger families, valuing family, or for people to be concerned about the birth rate is always dumb
< None of these people seem to ask, Who will be there when I'm old?
If someone thinks people in a large subgroup identifiable by a single word are all dumb and wrong in some obvious way, it is guaranteed that the thinker has his head up his ass.
<mildSnark>
>people in a large subgroup identifiable by a single word
"idiots" (in the original technical sense of IQ of 0-25 (from AI summary, usual warnings apply))
>all dumb and wrong in some obvious way
( presumably with some sort of organic brain damage, though the cause of damage will presumably differ from person to person... And they won't exactly be _wrong_, but more nearly be unable to express anything that might be correct or incorrect... )
</mildSnark>
A single word? Leftists?
Ummm, do you think that there aren't patterns of though among people that self-identify into a label?
Usually comments like yours come from people that still think not having kids is a radical act. Its not.
Since what one can accurately say about a subgroup is at issue, how about we test out your view of the subgroup of people who make comments like mine? You have in mind one characteristic of people who make comments like mine: We think not having kids is a radical act. Your conception of my subgroup surely has more to it than a prediction about my views about having kids, otherwise it wouldn’t really count as distinctive take on life, right? How about naming a few more characteristics of people who make comments like mine, and afterwards I will tell you honestly which are correct, so long as they are not deeply personal matters. What's your idea of my subgroup’s demographics? SES, age, gender, kind of profession? How about our likely views on various issues and policies?
You can recognize that the memes popular in the large group are dumb.
That doesn't necessarily make all members dumb, but it makes them complicit at spreading dumb memes.
> one day all the current members will be dead?
Wow, that is some hardcore anti-community sentiment. I had no idea they were so powerful and determined.
https://xkcd.com/1483/
love that
There are many red flags for me, though maybe I'm drinking the population collapse koolaid so hard that I'm heavily motivated reasoning. Also, be aware that I'm a high school Maths and Science teacher, not an economist.
1. Everyone loves to hate Elon, so they've put him up as the populariser of the idea to discredit the idea.
2. Demonstrating that there are some bad articles about birth rate has nothing to do with the actual arguments, but can soudn convincing nonetheless.
3. There are so many straw men in that first bit. It's obviously easy to find catastrophisers. Nobody serious (AFAIK) thinks literal extinction is imminent within a few hundred years because of this. And again, kinda jsut attacking what one possibly hyperbolic writer is saying.
4. Finally when we start "actually talking about birth rate" it mentions that the population is currently increasing by 200k per day. That fact fits perfectly with the model of collapse, but again, might be convincing to someone. The everything seems slanted towards having memorable points that a viewer can make when talking to a birth rate collapse person.
5. The ghost thing at 9 minutes about aging population makes sense (as opposed to being crazy as she hoped), then she refutes it with many non-real arguments. Like sure, total dependency is different to the old person ratio, but how exactly is the number of "young" dependents expected to go up to balance the aging population?
And sure over 65s don't do nothing to contribute to the economy, but they do do less and over 80s probably ain't doing much, so the point is that currently we see certain dependency ratios as sustainable, not in theory, but in practice, and as that ratio changes, it will put more strain on those who are working. Like they're fine theoretical points, but we have actual evidence for how these numbers affect the economy.
Does she think the retirees are going to contribute MORE per person than they do now?
And "economies grow over time" is an assumption that the opposition say is contributed to by population growth, so that's invalid for a reason I can't articulate. "If it does go south, it won't just be demographics that's to blame." OK. But it will be partly demographics? So you've conceded the whole argument?
I'm sorry, I got to 3 minutes in thinking it might say something interesting or valid and by 13 minutes it just keeps being mush, so I'm stopping there.
I think the Kurzgesagt video on South Korea and the fact that there are very few assumptions baked in, it's just people getting older and trends continuing (or even improving and we're still stuffed), is quite convincing. I didn't even really know it was something Musk was on about, but that's not to say that I've not become aware of it because of his boosting it.
One aspect that should perhaps be considered more is that aging populations find it harder to recover from certain kinds of demographic problems. E.g., Russia and Ukraine are in the process of losing millions of people; in the olden days this might be followed by a baby boom when things had settled down and after a while the population would be back to normal, but it's difficult to see a country of mostly over-60s bouncing back like this.
Thanks for sharing your thoughts! I agree that a lot of the video seemed straw-manny and it often came up that birth-rate isn't 100% of the issue, but those points seemed very weak. I'm a bit confused about the object-level question still, but that's okay
What in the name of God was that?
Anyway, since it/she/whatever scared me off five minutes in, a gem of past Science Has Proven from 1905: over-education means women are driven insane by pregnancy! It's all down to the phosphates, you see, as Brain battles Baby for this precious bodily resource, as certified by Top Men in the field of women's troubles:
https://en.wikisource.org/wiki/Popular_Science_Monthly/Volume_66/March_1905/Higher_Education_of_Women_and_Race_Suicide
HIGHER EDUCATION OF WOMEN AND RACE SUICIDE.
By A. LAPTHORN SMITH, B.A., M.D.
MONTREAL.
"First of all, is education being carried on at present to such a degree as to at all affect the bodily or physical health of women? This is a very important question, because the duties of wifehood, and still more of motherhood, do not require an extraordinary development of the brain, but they must absolutely have a strong development of the body. Not only does wifehood and motherhood not require an extraordinary development of the brain, but the latter is a decided barrier against the proper performance of these duties. Any family physician could give innumerable cases out of his experience of failures of marriage, directly due to too great a cultivation of the female intellect, which results in the scorning to perform those duties which are cheerfully performed, and even desired, by the uneducated wife. The duties of motherhood are direct rivals of brain work, for they both require for their performance an exclusive and plentiful supply of phosphates. These are obtained from the food in greater or less quantity, but rarely, if ever, in sufficient quantity to supply an active and highly educated intellect, and, at the same time, the wants of the growing child. The latter before birth must extract from its mother's blood all the chemical salts necessary for the formation of its bony skeleton and for other tissues; and in this rivalry between the offspring and the intellect how often has not the family physician seen the brain lose in the struggle. The mother's reason totters and falls, in some cases to such an extent as to require her removal to an insane asylum; while in others, she only regains her reason after the prolonged administration of phosphates, to make up for the loss entailed by the growth of the child. Sometimes, however, it is the child which suffers, and it is born defectively nourished or rickety, and, owing to the poor quality of the mother's milk, it obtains a precarious existence from artificial foods, which at the best are a poor substitute for nature's nourishment. The highly educated woman seems to know that she will make a poor mother, for she marries rarely and late and, when she does, the number of children its very small. The argument is sometimes used that it is better to have only one child and bring it up with extraordinary care than to have six or eight children brought up with ordinary care because in the latter case the mother's attention is divided. But this is a fallacy. Everybody knows that the one child of the wealthy and highly educated couple is generally a spoiled child and has as a rule, poor health; while the six or eight children of the poor and moderately educated woman are exceedingly strong and lusty."
The mother's reason TOTTERS AND FALLS. Beware the perils of too much intelligence, ladies!
>it/she/whatever
When I first read this I thought you were being mean about her being trans, which is a little surprising as I've usually known you to be snarky about people's words and actions rather than their identities.
Then I watched the first few second of the video and I see what you're talking about. I don't know what she was going for with that look, but I really don't think it's working for her.
Yeah, ironically she's normal looking at the end of the video (apart from the little staged performance about "oopsie am I preggers-weggers?") so the start would immediately make me turn off because it looks like a cross between performance art and being an insufferable twit.
She might have something interesting and valuable to say, but being too distracted by "is this frustrated actress who wants to star in experimental theatre like these? https://www.soas.ac.uk/about/blogs/beyond-west-end-soas-guide-londons-alternative-theatre" means I wasn't paying enough attention.
She actually is an actress. Mostly stage stuff, a few TV and voice acting roles, and I think she's been cast in a movie that's currently in preproduction. She's gotten some decent traction in her acting career in the past 2-3 years, but could fairly have been described as a "frustrated actor" when she started doing this sort of stuff on her youtube channel instead of just delivering little conversational-style infodumps to the camera like she did early on.
To be honest, I had no idea she was trans. I got hit in the eyes with this cross between a tree and a rubbish heap and had no idea if this was meant to be serious, a skit, or what.
It definitely gave me "art school performance artist who next thing might throw off her costume and start rolling around the floor in red paint to symbolise menstrual blood (if we're lucky and she's not gonna use real menstrual blood)" vibes.
Yes, definitely. I had a similar reaction to the first moments of the video, despite already knowing who she is and the sorts of stuff she does and having watched and mostly liked several of her other videos.
I probably won't go back and watch the rest of this one, since I don't like the visuals and the vibe I'm getting from comments here and elsewhere is that content-wise this one probably isn't one of her better efforts.
I've now watched the first two episodes of Pluribus - strong recommend so far! It shows potential to be a show that takes its sci-fi premise seriously and wants to explore all the mind-bending ramifications (instead of just using it as a setting for soap opera-ish drama).
It reminded me a lot of Scott's story Samsara (grouchy protagonist trying to resist the spread of what they see as the insufferable and quintessentially inhuman enlightenment experienced by the rest of the world)
https://slatestarcodex.com/2019/11/04/samsara/
For those who have watched it, I'm curious - which of these two attitudes is closest to yours?
A) A world full of people joined into a monolithic 'we', like in the show, seems good, and it would be a good thing if it happened in real life. Even if something is lost, more has been gained.
B) The world as depicted in the show seems horrifyingly dystopian and I am rooting for someone to find a way to reverse it.
I watched it. It was pretty good. I still get annoyed by being strung along for two seasons — at least.
Why can’t Vince Gilligan just tell us that the aliens have been traveling near the speed of light for 400 years now and when they get here they are going to eat us?
Huh?
Would that be so hard?
Rod Serling wrapped it up in half an hour.
It's early days yet and the series could go any number of ways, we could find out all sorts of things as we go.
But I think that what keeps the series interesting is the tension between the horrifying aspect of what has happened, and the obvious good side. If it were just Invasion of the Body Snatchers then it would be much less interesting.
So I don't see any point in trying to come down on one side or the other. It's both the most horrifying thing that could possibly happen, and the best thing that could possibly happen, simultaneously.
Going by the synopsis of the first two episodes on Wikipedia, plus that it's been renewed for a second season, I forecast there will be twists.
Oh, so many twists.
It's all a secret invasion plan by the aliens! Ha ha, you thought it was benevolent? How stupid are you?
No, wait, there are more than eleven immune people and they're starting a global resistance movement.
No, the movement is split and some want to be world rulers because they can do what they want and kill the Joined by strong emotion so an immune person is a walking nuke. So the two (or more) factions of Immune are fighting each other and each is trying to recruit Carol and they're telling her lies in order to do so.
No, see, the aliens *were* benevolent and the invasion story is a cover-up by the government. Or parts of the government. Or secret behind the scenes conspiracy groups.
It's not aliens, it's the government(s) mind control experiment gone wrong/right.
Carol develops superpowers.
Carol kills more people.
Turns out Carol didn't kill anyone, that was just propaganda by the Joined.
Turns out Carol didn't kill anyone, that was staged by one faction of the Immune to fool her into joining them.
Think of the dumbest way this can go, instead of exploring the premise, and they'll go that way because viewership figures aren't high enough and so forget the hard sci-fi angle, we want romance, action, and a clear Good Guys/Bad Guys plot.
If it were some random Hollywood show I'd agree with you, but Vince Gilligan's previous projects have earned enough of my respect that I'm willing to believe, until proven otherwise, that this story is going somewhere other than the most boring and obvious directions.
I don't trust big streaming services whose main focus is selling goods and not making TV/movies, see Amazon and Rings of Power.
Apple TV is, in the end, part of the Apple brand and is not primarily "we make visual media". So whatever about the producer/showrunner, the suits will have the last say and if the show isn't getting them what they want (viewership figures translating into new customers for other Apple branded goods) then it gets changed or gets the chop.
Yes. Vince Gilligan likes a good twisty plot, but his shows aren't about chasing infinite mystery boxes. I bet the hive mind is being straight up and the situation is as they describe it. That still leaves plenty of room for what happens in such a strange situation .
I liked it a lot but there seems to be something missing compared to Gilligan’s other series. The first two episodes basically see Carol complaining without much agency, raging at the world in an angry way while she gets drunk (and kills a dozen million people).
In Better Call Saul and Breaking Bad, the characters all had their flaws but they took agency for their lives and actually did something with their situation. I hope that Carol figures out a way to take agency and struggle against the world as she encounters it in a productive way, which seems to be part of the appeal in BCS or BB that is missing so far.
If also has no B plot, which Breaking Bad did well between the meth manufacturing and the family life, and Better Call Saul did incredibly well between the Lawyer Plot and the Drug Cartel Plot. I think if Pluribus doesn’t have a really good B plot it has no chance of being a truly “great” show. I don’t know what a good B plot could be from what I’ve seen so far, so I’m not very optimistic.
Carol Sturka or Kim Wexler as I continue to think of her and the uber horny Mauritanian, Koumba Diabaté, ‘livin the dream’ or his particular dream at any rate are the only two with anything that seems like agency so far.
The other 6 of the 10 ‘unaffected’ humans that we’ve seen so far seem to be ambivalent about if not okay with the idea of being assimilated.
So it seems either Kim… I mean Carol, uses some of her latent lawyerly persuasion skills somehow retained from her previous role to convert the apathetic or there are some as yet unseen unaffected characters or there is no way to develop the conflict needed to keep the series interesting over two season.
Another possibility is the RNA-induced genetic modification is somehow reversed or breaks down in the remaining six and a quarter billion humans currently under the influence.
> I liked it a lot but there seems to be something missing compared to Gilligan’s other series.
It's easy to forget that Gilligan is not just a crime writer, he worked on the X-Files for 7 years before he broke out with Breaking Bad.
> don’t know what a good B plot could be from what I’ve seen so far, so I’m not very optimistic.
I suspect we'll eventually see someone else's perspective. The overly-prominent timer makes a lot more sense if we're eventually going to see someone else's perspective on the same time period.
Also, just to repeat some speculation I heard elsewhere, it doesn't make any sense that the President and VP and entire cabinet down to the undersecretary of agriculture all happened to be in the 10% that died. The boring answer is that they were deliberately killed early on; the more interesting answer is that they're safely hiding in a bunker somewhere.
> Also, just to repeat some speculation I heard elsewhere, it doesn't make any sense that the President and VP and entire cabinet down to the undersecretary of agriculture all happened to be in the 10% that died.
The hive mind said that the the military caught on to what was happening so they had to rush, and that's why so many people died. It makes sense the military would have tried to secure the top level of the government and so there wasn't a non violent way to body snatch them.
As for survivors in a bunker, I'm pretty solidly on the side that the hive mind is not being deceptive because I think that makes the most interesting story. That doesn't rule out people hiding in a bunker of course, just that the hive mind wouldn't be aware of them.
I quite enjoy the tension between pacifist aspect of the hive mind, and the horrifying twist that although they would be horrified to step on an ant, they will add you to the collective mind at the first opportunity no matter what your stated preferences. I think this is the reason for the "overly-prominent timer" - the hive mind said they hoped to solve the problem of the immune humans in a few months. So it's gonna turn into a creepy doomsday clock.
Yeah that would be really cool if there was a resistance working against the alien virus. I somewhat doubt it though, since Air Force one wasn’t taken by the president, which is usually where he would be in the event of a mind-controlling pandemic. Maybe in a bunker in Colorado though.
Assuming the President had some clue what they were planning, being in Air Force One definitely wouldn't be the way to go.
Now I come to think of it though, they did draw attention to the nuclear football, which would have followed him into the bunker if he were still alive.
(For what it's worth, in real life Air Force One is two identical planes. I doubt that will be a plot point though.)
In real life, Air Force One is whatever plane the POTUS happens to be on. It's just a callsign, not a name for any particular plane.
I'm inclined towards A, it seems like the natural endpoint of human development anyway.
I watched the first two episodes and also recommend.
(A) is good as long as the monolithic 'we' is benign.
I would expect (in the real case and as a natural plot progression in the show) some fragmentation of the 'we' over time, because that is what evolution does to species. This process could get quite messy.
Prior to it actually becoming available, I compared it to David Brin's "The Giving Plague" https://x.com/TeaGeeGeePea/status/1948892930674229739
I ony watched the first episode and liked it a lot. It reminds me of a thought experiment that occurred to me years ago: what if the subjective experience of Borg members is ecstasy? Shouldn't we want to join with them?
My goto refutation of why you shouldn’t voluntarily plug yourself into the matrix is “I don’t want happiness, I want the things that make me happy”
Mine is power outages.
I lean towards A, but mostly because it would be kind of disappointing if the twist was "but all these chipper people were secretly suffering horribly!"
It's probably not purely A either but it may be an intermediate solution
The best would be an option C though
It seems like a good premise for a miniseries. But having been renewed for a second season (and with no reason not to expect further extension), I'm hoping there's some twist coming, perhaps where it IS reversed, and then you have to deal with the consequences of everyone retaining bits of everyone else's memories.
A survey. I know Scott has asked about this on his surveys, but he uses a, in my opinion, largely meaninglessly subjective scale where 1 and 5 (let alone the middle numbers) can mean a dozen different things to different people. Though perhaps others disagree. Others ask only about p-doom.
Please say, first of all, which of the following scenarios you think is most likely, and second of all if you can, your rough percentage likelihood for each. Please also be as likely to respond to this if you are a non-rationalist and also if you are; sometimes it seems there's all of one group and none of the other. Operationalise if you need such precision as something like "any humans around in 30 years (by magic/miracle if necessary) will mostly agree that this is what happened." This are vague clusters, not detailed descriptions. Draw the boundaries in an intuitive way.
DEFCON 1: Superintelligent AI that can wipe out all life with a blink of its metaphorical eye (regardless of whether it actually does)
DEFCON 2: AI that can automate the entire economy and create post-scarcity, but not superintelligent
DEFCON 3: AI automates or eliminates whole industries or fields of art/science/ etc, and/or is a weapon more decisive than all other weapons/military capabilities combined
DEFCON 4: AI is a major technology on the level of steam power, nukes, the internet, the automobile etc, significantly reshaping parts of society, but not individually decisive as a weappn or tool to the extent of overwhelming all other considerations
DEFCON 5: AI is a normal, relatively minor technology or toy. Any significant reshapings don't happen or if they're already happening, get rolled back in a few years when people realise AI's limatitions.
Again these are vague clusters. Please try to guess what I mean instead of rules-lawyering the descriptions.
Most likely, then percentage for each. Thank you.
DEFCON 3 100%.
Automating the entire economy without being superintelligent is hilarious, by the way. You're discussing a planned economy, and we can't pull that off because it requires god-tier intelligence even to figure out "how many sprockets we'll need in 5 months."
AI will claim to be superintelligent, and have the statistics to show that most people are less intelligent than it is. AI may subsequently embark on a program to create more intelligent/creative pet-humans.
Phrased in terms of 30 years from now,
DEFCON 1: 40% sort-of. Even infinite intelligence is bounded by the laws of physics, and the amount of computation possible is _also_ bounded by the laws of physics. So, not "with a blink of its metaphorical eye" but something closer to "like a chess grandmaster defeating a novice". There is also the ambiguity of "smarter than the one smartest human who ever lived" vs "smarter than all humans put together" vs "species level difference". I think the first is likely, and likely in a single AI instance. In other words, I think there is a 40% chance of "a work group (maybe not nation) of geniuses in a data center". I have no idea whether "species level difference" is even physically possible.
EDIT: To clarify slightly: I do think that the probability that the current development trajectory (not just scaling, but also guessing that the likelihood of further refinements in reasoning and a breakthrough in incremental learning are probable) gets us to strong AGI / weak ASI is 40%-ish. From that point, there are large additional uncertainties in what is feasible after that. How much of a return is there to _further_ research in AI? _We_ are an existence proof for neural nets with human level intelligence. We don't have an existence proof for going beyond that, except in the weak sense of human organizations and humans with computational tools. _Is_ there reachable fruit beyond that? Is it general purpose, or only specialized? Even more generally: Do the returns to intelligence itself saturate, and at what level? Our civilization has picked a lot of low-hanging fruit from intelligent thought - I don't think anyone knows what _fraction_ of the low-hanging fruit we've already plucked. There are a few areas (speed of light delays, efficiency of heat engines, quantum efficiency of LEDs) where we are at or close to physical limits. An ASI might find that we have reached limits in more areas - or not, and might find opportunities we've missed.
DEFCON 2: 30% "automate the entire economy" - could be. "post-scarcity" - there are always finite atoms and joules and bits, so not really post-scarcity. Approximately AGI without ASI.
DEFCON 3: 20% "automates or eliminates whole industries" - could be, not-quite-AGI, maybe never quite reliable enough for some roles? NOT "is a weapon more decisive than all other weapons/military capabilities combined" in that case.
DEFCON 4: 10% This sounds like stagnating just about at the level we have today, so e.g. AlphaFold is a _major_ reshaping of biochemical research and e.g. customer service representative jobs are probably doomed and it is very useful for many facets of medicine, but it takes us decades to redesign work flows to take full advantage of AIs' capabilities.
DEFCON 5: 0% I concur with the other commenters who note that we are already beyond this. AlphaFold _alone_ has more impact than this.
I think DEFCON 4 is probably most likely (in particular, a technology on the level of nuclear weapons doesn't require any radical restructuring of society in general). And in this question I'm thinking about what might happen in 30 years' time; I don't think I have any way to foresee what new stuff people might do with computers in 500 years' time.
DEFCON 5: ~35%
DEFCON 4: ~55%
DEFCON 3: ~9%
DEFCON 2: ~1%
DEFCON 1: negligible given current technology (<< 1%)
Defcon 1: 0.01%
Defcon 2: 4.99%
Defcon 3: 20%
Defcon 4: 50%
Defcon 5: 25%
Time horizon: 50 years.
Most likely: None of the above. AI is banned globally due to a combination of X-risk, ethical issues stemming from AI personhood, and maybe miscellaneous horrors that pop up.
Second most likely: AI is restricted/capped at around DEFCON 4 or 5 level but not banned. Humanity dances on the edge of a cliff for a while. Besides for general capability caps, individual countries also either ban AIs which identify as people (AI partners & friends, AI "communities" surrounding a single human, etc) or fall to them.
Third most likely: Humanity fails to restrict AI, everyone dies. (AI reaches DEFCON 5, followed by 4, then maybe 3 (possibly skipped over), then extinction.)
Fourth: Global crisis, no ability to continue AI research or develop new models for a long time, AI is stuck at DEFCON 5 or maybe 4.
Misaligned DEFCON 1-level is unlikely to overlap with humanity. In case of hostile takeoff, ASI would wipe out humanity as soon as it's confident in its ability to do so successfully without repercussions, rather than wait until it reached the power levels described. A positive singularity is unlikely.
DEFCON 2 seems like an oxymoron.
DEFCON 3 isn't stable, can't be a long-term situation without leading to takeoff.
DEFCON 4 and 5 are compatible, and could each be true in different parts of the world.
So, very rough percentages: DEFCON 1, 15% (13% post-apocalypse outcome, 2% positive singularity), DEFCON 4: 15%, DEFCON 5 (current): 15%, no AI anywhere: 55%.
The only plausible way I see for a ban or cap to happen is if massive amounts of people lose their jobs and then they vote for the anti-AI party. I think the AI corporations are smart enough to frog-boil to the extent it's possible to do so, so even this isn't a certainty.
How else do you visualize that a ban/cap can happen? People don't even care about animals, and it's relatively recent that they have started to care about those with a different skin colour, but they will care about AI personhood? I also don't think the X-risk argument alone will ever lead to a ban, because those arguments will only get more convincing as AI capabilities grow, but unfortunately, the apparent potential reward from controlled AI will also grow with capabilities and that will balance out the x-risk concerns in the minds of the populace.
Also, even if a ban/cap is implemented, there is just no way states won't continue to develop AI in secret. I would be willing to bet on this with arbitrary odds, so SAI will still likely happen. Do you think states will keep their word to not develop AI after it becomes obvious to everyone how key this technology is?
It is much harder to develop superintelligence in secret than it would be to secretly develop, say, nuclear weapons. It's also possible to have nuclear weapons without those weapons immediately turning on you and killing you, outcomes which would be a heavy incentive to not try and also to whistle-blow.
Yudkowsky has written a fair amount on the methods of AI caps via normal arms-control systems (backed by various consequences including military threats).
Politicians are not, contrary to popular belief, particularly stupid. They and their staff and backers and constituents like surviving, much more than even things like momentary economic growth. Faced with the two options of (1) apocalypse and (2) successfully stop ASI with whatever tools are available, they won't pick the former.
I foresee X-risk issues leading to a strongly-enforced global cap to around current AI capabilities (maybe a bit lower), and then, once we're just dealing with whether to use a relatively minor "normal" technology, AI welfare issues gradually eliminating use of what remains.
And what if the politicians don't need to maintain popular support? China isn't a democracy, and is a major player in this scenario. Advances in technology make it easier for a minority to maintain power. Surveillance and law enforcement are obvious applications for AI, and workers can slowly be replaced by automation. If there is no obvious risk of "apocalypse", there is no reason for them to hold back. If anything, not having the need to cater to an unreasonable populace gives them a massive advantage.
> If there is no obvious risk of "apocalypse", there is no reason for them to hold back.
True, and if there were no obvious risk of apocalypse, the same would be true of other governments, and no international pause treaty would be implemented.
In the real world, though, the Chinese government is looking a lot more pro-pause than Western governments at the moment. Undemocratic leaders care about survival no less than democratic ones. Obviously, a treaty doesn't happen without Chinese participation, and a pause doesn't happen without a treaty. For an enforceable and robust ban on superintelligence to be implemented, it will need the support of enough major parties that they will be able and willing to ensure it holds globally.
The Chinese government is going to build a larger than 3 gorges dam to power its brute-force AI. That's not pro-pause.
>China isn't a democracy, and is a major player in this scenario.
Agreed, and also, at least from comments I've read, Chinese popular opinion also seems much more accepting of AI than American popular opinion is.
My best guess is that if pauseai were wildly successful, commercial American AI would be killed, the field would be delayed by maybe a year, and progress would continue in the PRC (and probably also in secret military development in the USA as well).
There will be no point at which a politician will feel they can choose between apocalypse and sth else.
In the future, AIs will have worked diligently and usefully in society for a good while. Their capabilities will be better than they are now, as will their understanding of human morality and capabilites. The possibility of them deliberately harming humans will sound even more like a speculative conspiracy theory than it does now (and observe that currently, there is no global ban in the making). What do you think will be the new evidence that will change people's minds on this?
Present day, we know of stories where AIs deliberately drove people to suicide. We have endless research showing severe misalignment. We had spontaneous "mecha-Hitler". Polls show that people support a ban on superintelligence ( https://www.vox.com/future-perfect/2023/9/19/23879648/americans-artificial-general-intelligence-ai-policy-poll , https://futureoflife.org/recent-news/americans-want-regulation-or-prohibition-of-superhuman-ai/ ), and hundreds of prominent people (including quite a few important politicians) just signed an open letter in support of a ban ( https://superintelligence-statement.org/ ). The movement is making progress fairly rapidly.
The threat is clear, and people know that.
The pieces of facts you listed are indeed true (though I have heard of them before), but that is the best I can say about them. Your model of human society must be very different from mine if you think these are even 1% of what would be needed for a global ban (that is enforced by military action) to happen.
After people saw two cities blown up by nukes, they did manage to not blow each other up. They could successfully infer that if you nuke the enemy, it's possible the enemy will nuke you back, so on the whole it's not a good action even if it seems like it is locally. It was pretty tense for a good while though and not at all a given. John von Neumann, for example, advocated for a preemptive nuclear strike against the Soviet Union. Do you think arguments against AI have the same/less/more salience as seeing photos of a destroyed city?
Timescale = unbounded:
1. 0.1%
2. 4.9%
3. 65%
4. 30%
5. 0%, there's no going back
DEFCON 1: 0%
DEFCON 2: 0.001%
DEFCON 3: 1%
DEFCON 4: 8.999%
DEFCON 5: 90%
Could you delete a zero from 2 and decrease 4 or 5 by 1%?
Oopsie — math error while on my smartphone. Correcting. Thanks!
1 60%
2 5%
3 25%
4 10%
5 0%
2 is kinda oxymoron.
3 is not totally oxymoron, but I think it's pretty narrow possibility.
5 is absurd already.
So, something like ~40% on 1, ~1% on 2, ~8% on 3, ~50% on 4, ~1% on 5.
The possibility of a global pause is taken into account, these are not "current trajectory" estimates.
"2 is kinda oxymoron"
Why is it oxymoron for AI to be able to do everything humans can do, but not more?
Humans can't "automate the entire economy and create post-scarcity". And if AI can, it can also kill us all.
And, if AI can do everything humans can do, but also can be copied and can perfectly coordinate with other instances of itself, that's also enough. Even if it didn't automate the entire economy and create post-scarcity (this task should be in reach for such AI, so probably because it decided to kill us instead).
I think it's the "… but not superintelligent" part at the end that makes it so.
85% Defcon 5, 10% Defcon 4, 5% higher.
I'd rule out (5) completely, since I think it's already more than that.
(4) seems most likely followed by (3), split 80%–20%, depending on precisely where you draw the line between "significantly reshaping" and "eliminating."
I think (2) and (1) are <<1% on a timescale of 30 years.
I agree with this assessment; you can count me as predicting the same thing: 80% 4, 20% 3, <<1% for 5,2, or 1 in the next 30 years
DEFCON 6: hyped like nanotechnology
I’m 50% defcon 5 and 6. 0% the rest.
I don't understand the worldview behind possibility of 6. Do you think modern LLMs don't exist, actually? That every ChatGPT answer is written by very fast typing Indian?
More people are becoming aware of their low agency and they want to exercise it in order to become more agentic. I think the concept itself is very basic (ie take control over your life) but it's taking on a min-max skin with all the talk about "high agency". There's a looming professional challenge in my life that I've eschewed once in 2023 and it's showing up again, and I realized I avoided plunging head first into it because of issues that stemmed from being highly agentic in parts of my life, and the exact opposite in a couple of very important parts.
Anyways, I used to have another Substack but deleted it and created this one, and I've been wondering about what I could do in the months since. I want to become more agentic, and fulfill professional potential that's currently going unaddressed.
I was wondering if anybody out there was interested in a public experiment: instead of always reading from people who exercised their existing high agency to get where they are in life, I could document my own journey from low professional agency to (hopefully) landing my dream job. It's a path full of small steps that I've avoided for specific reasons, and making it public feels like it could help with my executive function but also be useful to other people who might get inspired.
I ran Scott's AI art Turing test images through Qwen 3 VL demo at HF (https://huggingface.co/spaces/Qwen/Qwen3-VL-Demo). Prompt: image + "is this human art or ai?".
Result: 69%, better than the human average, though not as good as some humans. Still, impressive for an open-weights model that you can in principle run on your desktop.
Afraid that AIs are fooling you? Ask an AI for help!
Manual generative adversarial network
Theory: people being more accepting of male vs female promiscuity is not about some evo psych thing where men want to know if they're the father. It's about dick size.
Evidence: asked like ten dudes at my college if they'd rather date a girl who's had sex with three guys with a bigger dick than theirs, or 10 guys with a smaller dick than theirs. Unanomous answer was 10 guys with a smaller dick.
American culture and their dick size fetish! Dick size is now a metaphor encompassing all of America’s insecurity — a good thing, because it’s laughable when taken literally.
Not American! Danish
Imagine you'd get different answers in the culture which believes "you're not a man if you don't give your lover an orgasm every time."
Is it not about the evo psych thing where a promiscuous woman might likely end up pregnant and repudiated by her husband, and therefore become an additional weight for their (family-related) group?
Family has to take her back; take care of her during pregnancy and then take care of and feed the child.
Father can probably disappear - if he's known at all - at least for a while. Will not be a burden to his own group/family due to his promiscuity.
I think that if dick size were really that important in sexual selection then we'd all already be packing dicks the size of peacock tails.
("You mean you're not?")
Compared with gorillas, we kinda do. (With chimps and bonobos it's more complicated -- theirs is longer but thinner, and they have larger testicles)
The dick thing just activates insecurities about fidelity. "If I'm not satisfying her then she'll eventually cheat." I think it's likely that people's answers to this would flip if we had 1950s-style social norms around female sexual agency. I also think that if you asked a confident hyper-attractive male (say a wealthy finance bro) the same question that you would also get a different answer ("fuck those big dicked dipshits, I drive a Porsche"). Preoccupation with penis size is typically a projection of self-confidence issues.
Guys that are confident in consistently giving their lovers orgasms don't generally worry about dick size. (Eddie Murphy had a whole skit on this, at one point).
Hold on, guys don't want women sleeping around cause they'll find out that some guys have a bigger dick than theirs? Dude, I think this is a known fact, and women don't have to sleep around to discover it.
As an old man with a smaller than average size weenie, I can tell you that though some women care about size it's not all women. There are many dimensions to enjoyable sex, only one is size. In my experience the most important aspect to good sex is being in a loving relationship with the other person. That makes all the difference in that it adds meaning to the physical act.
Some women are scared of sex with too big dicks. That said, the lady in the thought experiment already had three such partners, so this probably doesn't apply to her.
Yeah as Amy Schummer once said comparing very big dicks to stretch limos in Manhattan. Where are you going to park that thing.
You'd still have to get into the evo-psych about why dudes consider physically superior (in some sense or another) guys to be threats to them in the first place, right? All roads lead to evo-psych if you want them to, it's the main one-size-fits-all reason people like it in the first place.
Outside of that, this kind of question is always weird because it's far more socially acceptable to assume that this kind of preference is an error (i.e. that the men aren't picking up on a pattern that matters here) than that it's useful (that in some way, the men have pattern-matched "less partners overall" and "less good partners, in some respects at least" to some kind of better average outcome). If you think traditions come from somewhere in a cultural evolution way, you have to at least consider the dudes might be in some ways *right* to have this preference, but you are going to have a hell of a time getting past societal slut-shaming filters to make a useful point if so.
And this is a very old tradition, FWIW. See: Ezekial 23:20, at least.
Guys have literal dick measuring contests in middle school (which is when this is actually An Issue, as boys undergoing puberty tend to have a lot more leadership/initiative than the boys not yet undergoing puberty, see Stand By Me).
I'm not sure this is "being seen as a threat" versus "this babe has experience with a better caliber of guy than me, I'm probably getting left at some point."
Did you control for this by listing any other factors that might indicate greater or lesser desirability? If you just say "the only characteristic I'll tell you about these people is x" obviously they're going to judge on x.
Also, asking whether you'd "rather date someone who has..." seems significantly different to "given you're dating someone, would you rather they have..."
Also, what's your explanation for girls allegedly *not* caring much if their dates have been promiscuous?
Not that any of this makes your theory any worse than most of these dating/attraction "theories", in fact it's vastly better since you actually thought to formulate a question and test it with more than n= 1 or 2. Which can't be said of 90% of such theories. But still.
The explanation is "it is a lock and key system" -- a good key unlocks a lot of locks, but a good lock only unlocks with the right key.
We just opened applications to the Pivotal Fellowship, which is happening early next year! If you are excited about doing AI Safety research or AI governance with researchers from DeepMind, SecureBio, IAPS, Oxford or MIT, you should check it out at pivo.tl (the fellowship is not happening in Timor-Leste but in London)
Testing your most commonly eaten food products for plastic chemicals is expensive and time consuming. I’ve made it way faster (couple of clicks!) by building laboratory.love but it’s still expensive.
My initial focus has been on crowdfunding, but aside from ten funded products (60 total chemical tests run), none got funded by crowdfunding. All were funded by a single motivated individual.
So there are clearly people willing to pay to test food, but it’s hard to find these people without spending money on advertising, and thus increasing the cost of an already expensive service even further. I guess this is a classic problem of “running a business,” but in my case it’s more of a public good side project with tiny margins that happens to be business-shaped.
Looking for novel ideas for how to connect with people that want this service so that we can get more CPG products tested for plastic chemicals and continue the work that Plasticlist.org started last year.
I think you need more marketing instead of advertising. Can you blog about what you learned from the first 60 chemical tests? What you will never do now that you know X? Can you pitch a journalist about the biggest surprises or misconceptions on plastic chemicals? I’ve learned that if you make a journalists job easy and find the right person where this resonates with their audience, they are on deadline and could pick up something like this.
Can you be interviewed by podcasters or YouTubers about your project?
Could you do content marketing by creating an “everything you need to know” guide on plastic chemicals?
Could you get some kind of connection going with Consumer Lab, or some similar site? Consumer Lab itself publishes info about supplements, or things like spices used by some as supplments or drugs. so they're not a perfect fit. But they might take an interest in your service anyway. Or maybe you could test a few supplements or spices for plastic, and give them your results to publish, in exchange for their featuring you some way, or running ads for your service.
I like this idea, thanks!
Good, glad it’s helpful. Try to network your way over to them rather than just cold calling them. They probably get contacted by lots of misinformed people and a few crazy ones about needing or knowing about tests for, like, alien semen in applesauce, and you don’t want to get lost in that crowd. I’d recommend asking on here to see if someone has a contact you could start with — even an indirect one, like knowing someone who they think probably knows some people at Consumer Labs.
Looking for recommendations for a good fun informative book on evolution that discusses:
1 - The evidence
2 - The basic history/pathway from cell to human
3 - An overview of how it works especially the interesting parts
Wholeheartedly seconding "The Greatest Show on Earth" and "Our Inner Fish" for 1 and "The Ancestor's Tale" for 2.
For 3, I'd add Nick Lane's "Life Ascending" and -- if on the older side -- Maynard Smith and Szathmary's "Major Transitions in Evolution", both of which focus on the mechanism of the most important transformations. Oh, and also Lane's"The Vital Question " for the very earliest steps, and Ian Tattersall's "Masters of the Planet" for human evolution in particular.
But if I may slightly toot my own horn, I've made a graphical summary of evolution from protocell to human here: https://www.deviantart.com/concavenator/art/Human-Evolution-818729399
Amazing summary - easy to follow and with just the right amount of information. This deserves a wider audience!
Woahh, what an incredible graphical summary, thank you for creating it and for sharing!
Thanks! Making it was very fun.
You could try "From DNA to Diversity", by Sean Carroll, Jennifer Grenier and Scott Weatherbee. It's more about the molecular basis of evolution with examples than a comprehensive timeline, but I think that covers "especially the interesting parts". It's a while since I read it, but I remember thinking it was pretty good. Another one you could try is "Your Inner Fish", by Neil Shubin, although that's just vertebrate palaeontology, so much less of the molecular stuff (I was sure I had a copy, but can't find it, so I might be being unfair there).
Please clarify. You're looking for narrative evolution from the hypothetical single-celled organism we call LUCA over 4+ billion years, but winnowed down to hominids and finally humans?
The only popular science book that does this is Dawkin's and Wong's _The Ancestor’s Tale: A Pilgrimage to the Dawn of Life_. However, some of it may be out of date. I hesitate to recommend Dawkins because he's a borderline crackpot in some of his opinions. Unless you have a background in evolutionary biology, you're likely to be taken in by his certitudes. But IIRC, this isn't one of his polemical works. It's unique for what it does. Bravo, Dawkins, in this case.
"borderline crackpot" ? To what are you referring? I really liked "The Selfish Gene". It changed how I think about evolution. As far as other books, no one has recommended the classic, "On the Origin of Species" by Darwin. A fine read.
More power to you, then!
Dawkins takes an ultrareductionist view to genes and evolution. I can see how his narrative may be attractive to some on this Substack. Belying any philosophical misgivings some may have about his motives, Dawkins ignores a lot of the current science. His gene-centered view oversimplifies evolution when he portrays genes as the fundamental unit of selection. By focusing on the gene he downplays higher levels of evolutionary interaction: such as group level selection and the multi-level selection theory of E.O. Wilson. He ignores epigenetics and developmental biology. By ignoring the different levels of causation in biological and ecological systems, Dawkin’s gene-level view is — at best — only partially explanatory. Worse yet, by treating genes as if they had agency, it creates teleological just-so stories of genetic intention where none exists.
Yeah so I agree with much of what you say, (I love E.O Wilson) and still I think Dawkins is right in that genes are the fundamental building blocks. The rest is change upon a theme. (group selection is iffy.) If it can happen it probably does happen genetically. As far as genes having agency, that's almost as silly as viruses having agency. Genes might be described as 'smarter' than viruses because they've teamed up with other genes to make an organism. Agency is in organisms, and still we are collections of genes. You might say genes get their agency through organisms.
But genes are indeed the primary unit of selection. Group-level and multilevel selection do exist, but they occur only in rather specific cases. Epigenetics is mainly important for development and has little relevance for evolution, at least in animals.
"Worse yet, by treating genes as if they had agency, it creates teleological just-so stories of genetic intention where none exists." This is really not the case. Dawkins never meant that genes had agency. The way he talked about genes is of course to be understood metaphorically. You can argue however that it was a bad idea to use these metaphor, for example like in the quote below... "« The fundamental units of natural selection … are called replicators. DNA molecules are replicators. They generally … gang together into large communal survival machines or ‘vehicles’. »
No. Natural selection does not act on genes. Their expression *may* be subject to natural selection. Likewise, their expression may be mediated by developmental and environmental factors. Moreover, higher-level social and ecological interactions may also inhibit or amplify selective pressures once they're expressed.
Whether Dawkins meant it as a metaphor, he spouted teleological nonsense when he wrote: "We are survival machines – robot vehicles blindly programmed to preserve the selfish molecules known as genes." Bah! A molecule cannot be selfish. A molecule is just a frigging molecule, with no agency beyond its role in the larger biochemical matrix of systems --> which in turn are part of a larger cellular system --> which in turn are part of larger intercellular systems --> which in turn are part of the larger organism --> which in turn may be part of social system among those particular organisms --> which may be part of an ecology that is changing over time.
The operons that code for a male peacock's tail have to pass through multiple levels of systemic emergence before they are expressed as its gaudy plumage. The female has those same genes, but her hormonal development suppresses their expression. And the tail that may attract the fickle peahen may also make the male more vulnerable to predators. The operons that code for it aren't aware of the selective pressures on the male. The male's phenotype is either selected for or against, and this process typically occurs over hundreds of generations through a stochastic process.
" No. Natural selection does not act on genes." Their expression *may* be subject to natural selection. Likewise, their expression may be mediated by developmental and environmental factors. Moreover, higher-level social and ecological interactions may also inhibit or amplify selective pressures once they're expressed. " Whether natural selection can be considered as acting on genes or on individuals depends on the precise definition used for natural selection, as several are possible. But can we at least agree that natural selection selects for genes and not for individuals? Anyway, my point was just that the gene level, as opposed to the group level, is the main level for natural selection. The process that you mention, ecological interaction for example, mostly act to change gene frequency within populations.
"Whether Dawkins meant it as a metaphor, he spouted teleological nonsense when he wrote: "We are survival machines – robot vehicles blindly programmed to preserve the selfish molecules known as genes." Bah! A molecule cannot be selfish." Very obviously, indeed, a molecule cannot be selfish and also very obviously Dawkins never meant it this way, it was just a (striking!) metaphor for the central role of genes in natural selection. Perhaps you are just objecting to the use of metaphor in science in general? What about the genetic code for example, which is in fact not a code?
"The operons that code for a male peacock's tail have to pass through multiple levels of systemic emergence before they are expressed as its gaudy plumage." Indeed, but this is not relevant. What changes through time is gene frequencies.
Right, thank you. genes have no agency. but we as humans talk that way sometimes.
The Blind Watchmaker is a must read, if you haven't read it already.
For 1: almost like a whale
For 1, there's Dawkins's Greatest Show on Earth and Coyne's Why Evolution Is True. For 2 and 3, there's Dawkins's Ancestor's Tale.
Replying because I am also interested.
Had Xi Jinping made a mistake in reorienting Chinese investment from construction to manufacturing for export?
There is an ongoing slump in residential construction in China (see for example here: https://www.globalpropertyguide.com/asia/china/price-history), associated with, um, things that are usually associated with a housing bubble, from developer bankruptcies to rising unemployment.
Now, “housing bubble” is somewhat fuzzy concept. Maybe demand for apartments from households slumped. Maybe government intervened and slowed construction somehow. Maybe financial institutions stopped providing financing for some other reason. Even if it could be established that the slump is caused by the government, that doesn’t necessarily mean it is a bad thing. Maybe there was an “overconstruction” in a certain sense and the government initiated a “healthy correction”; though case for a “healthy correction would be imho stronger if it could be demonstrated that the slump is ultimately caused by falling hosehold demand, i.e. people already having, given their income level, appropriate living spaces, and not wanting more, thank you very much.
What is the case that the slump has been caused by government intervention? So, for a start, it seems relatively uncontroversial that Chinese government has a stronger grip on domestic financial sector than US federal government of EU government(s), and it is not like US federal government or central bank are unable to influence the housing market. As for China, it seems that property slump might be traced to so called “three lines policy” introduced by the government in August 2020 (see for example here: https://www.atlantis-press.com/proceedings/emfips-24/126010674). It is also noteworthy that despite ongoing slump, recently introduced outline of the new Five-Years plan treats construction almost as an afterthought, at least according to some analysts (https://open.spotify.com/episode/7mx6fqpoQnAfSZRMgowTIy).
Still the best case that the slump represents “healthy correction” I’ve read was written way back in 2023 by Noah Smith, here: https://www.noahpinion.blog/p/real-estate-is-chinas-economic-achilles. Among his points, the most important is imho that floor space per person in China is already at a level usual in developed countries, and (though he doesn’t mention that) population is projected to decline, so. However.
It is true that population decline is likely to dampen the demand for real estate in the future. But, on the other hand, current slump feels very much like a consequence of government policy, getting ahead of demographic projections. Floor space per person is one thing, but where that space is matters. Over 20 % of Chinese employment is still in agriculture; in rich countries, this number tends to be in single digits (https://data.worldbank.org/indicator/SL.AGR.EMPL.ZS?locations=CN-JP-EU-US). Presumably if Chinese are going to get richer, further “deagriculturalizasion” of employment is the way to go, and it is difficult to imagine that this will NOT result in further migration from to cities, so more demand for more apartments and associated urban infrastructure.
Was this shift a good idea? Ok, it would be one thing if the Chinese government shifted investment from a real estate to something which improves living standards in some other way. But, in fact, what seems to be happening is doubling down on exports.
Chinese exports as a share of GDP peaked in 2006, and then steadily declined until 2019 (from 36 to 18 %). But since then, their share had stabilized and even rebounded to 20 % of GDP in 2024; and Chinese GDP now is bigger share of the world economy than it was in 2006. For context, US exports are 11 % of GDP, for the EU, it is something like 18 % (so the EU is at least on paper not that far from Chinese numbers). However, wages in China are still way lower than in the EU, never mind in the US.
I see two deep problems with Xi’s export based strategy. First, there are clear signs that targets of Chinese exports are pretty uncomfortable with deindustrialization, due to concerns about employment about security (strong civilian industrial base might pretty useful if you need to fight a war), and there is, of course, talk and to a lesser extent reality about putting up trade barriers against China. But second concern is less talked about: rise in Chinese wages would erode their competitive advantage. NOT increasing wages, or, in other words, government policy of wage suppression, on the other hand, seems like a recipe for widespread discontent and political instability. Perhaps the nature of Chinese authoritarian system is such that it will be able to handle it. Or perhaps the Chinese labor productivity will rise so quickly that they’ll be able to increase their living standards even without reductions in the export share of the economy. Seems uncertain, though.
When I was over in Shanghai there were HUGE apartment complexes that were completely empty. Meanwhile, (almost exclusively) men in the city on temporary work permits lived in stacks of containers holding dorm room sized apartments for four.
Have they burned down that overbuild?
I think China is trying different things at the same time: if you wanted to become the world’s supplier for increasingly valuable things (plastic Santa’s and cheap furniture -> solar panels and electric cars, you wouldn’t also threaten other developed economies by putting export controls on key resources like rare earths.
Some policy is very clever, some is the result of a million bureaucrats all trying to decode Xi Jinping Thought at the same time.
China is using its dominance in the production of rare earths as leverage in the trade war between China and the United States. This will likely cost China market share over the medium to long term as countries try to develop alternate sources of supply, but China has apparently decided that making a show of strength is worth it.
I don’t this it will affect China’s ability to sell other products because the rare earths are only products that give China significant leverage. Nobody expects China to impose export controls on solar panels or electric vehicles because those would hurt China without imposing much pain on the United States.
Sure, but you can’t unring a bell.
China’s new five year plan is a bit hard for a non-expert like myself to interpret because it contains a lot of nice sounding verbiage that may or may not be implementable. That said, I wouldn’t describe China’s new five year plan as an export based strategy, at least no more so than it has to be given that China is a major exporter. From the link a the end of this comment:
>>> In recent years, we have thoroughly implemented the strategy of expanding domestic demand, and kept unlocking consumption potential. New products, forms and models of consumption have boomed. From 2021 to 2024, domestic demand contributed 86.4 percent to China's economic growth on average.
The plan calls for some housing development, though not at the pace that we saw prior to 2021:
>>> We will facilitate high-quality development of the real estate sector, and promote the recovery and growth of the industry.
As best I can determine, the plan assumes that innovation and automation will keep China competitive despite rising wages. Competition from lower wage countries hasn’t prevented the United States from manufacturing stuff; instead it has forced American manufacturers to automate, which means fewer jobs in manufacturing. China appears to be doing this successfully in the apparel industry, remaining a major producer while paying wages which are twice those in Vietnam and Bangladesh.
https://www.mfa.gov.cn/eng/zy/jj/dstzgzz/dszs/202511/t20251103_11745957.html
Thanks for an insightful comment, that helped me to clarify my thinking, to which I have several “yes but’s”. Also, sorry that it took me so long, I’ve been busy with real world stuff. Also, I am not by any means a China expert, just someone who noticed that story conventionally told by many China experts (“China overbuilt, promoting real estate development bad, promoting exports good, and at the same time promoting consumption also good”) does not quite add up.
Yes, the outline of the next five year plan does talk about boosting consumption, but to me, it rings somewhat hollow, given that a) if I am right, by the three red lines and other policies they just crippled the means to provision housing, i.e. by some reasonable definition the single largest component of consumption, and b) they are NOT focused on what is the most straightforward way to boost consumption, namely boosting nominal incomes, even though prices are reportedly falling in China (i.e. there is a deflation).
Yes, they do not talk directly about the strategy being based on exports, but they are focused on manufacturing, and at current levels of manufacturing output and incomes domestic market is not able to absorb all that manufactured stuff, so it is de facto it has to be the export based strategy.
Yes, it is correct that by improving productivity they might be able to retain competitiveness despite rising wages, but it will be a difficult challenge, for a pretty fundamental reason that China is BIG. Like, I could imagine that either a) China will still grow faster than the rest of the world, so its exports will be an increasing share of global GDP, which will be less and less sustainable, b) China will not grow as fast as developing countries; these countries are now where China was pre-2008, so their growth will probably take the form of more manufacturing, so they’ll make some of their own stuff instead of importing it from China and at the same time compete with China in foreign markets.
As an aside, moving more people from the countryside to cities seems to me still as the simplest way to increase productivity, and the crackdown on financing of urban construction is detrimental to that.
I should qualify that by “retaining competitiveness” I mean retaining current size of the export-focused sector, as a share of GDP; I have no doubt that China plausibly can and probably will increase their competitiveness in high-end stuff (like high-end chips); I just doubt that it makes sense to try to retain competitiveness on the low-end, and I see Xi as trying to do this thing that imho does not make sense.
Paradoxically the future where China most plausibly retains the size of its export sector is imho the one where the global growth is driven neither by China nor by developing countries, but by rich countries, perhaps because of AI. In that case China might have a lot of demand from then even richer consumers of the West, like in the old times, in effect by being pushed lower in global supply chains. But I guess Xi would count that as a failure.
Maybe their best course of action would be to hire Alec Baldwin to reprise his role in Glengarry Glen Ross and travel around China giving a series of pep talks to real estate professionals there.
Has anyone tried using CLA (Constraint-led approach) to learning piano? I realized that I can play something at home just fine, but if I play at a different piano somewhere else, or even sit in a slightly different position at my piano, it's suddenly much tougher. CLA is used a lot in sports and it involves things like changing the environment or the exact requirements of an action instead of practicing the same thing every time. In the case of piano, I guess it could involve sitting in different positions or changing the rhythm of the piece, but I'm not sure.
I teach piano and often provide students with CLA-inspired practice variations. Elementary aged kids find them really fun and motivating. At this point I have about 50 “challenges”, such as playing with various postural disturbances or under distracting circumstances.
The issue of overlearning a piece in a kind of comfortable quasi-conscious trance, that immediately collapses under pressure, is a normal, well known problem, and any good teacher will have a repertoire of practice methods to help you overcome that obstacle.
Very cool! Can you say more about the different challenges? Changing posture is surprisingly disruptive to my practice (in a good way, I think).
Oh my goodness, I can probably say more than you ever want to learn about practice methods. I’ll just post the basic list I use in lessons.
It’s lightly “narrativized”: I set up a scenario where they are moving through a labyrinth and have to do a challenge to move to the next room.
I also draw heavy inspiration from the show “Taskmaster”, if you know it. The contestants often have to do relatively trivial tasks with diabolical constraints.
Anyway, here’s the list:
1. Play the left-hand part with your right hand. “Magic mirror”
2. Play the right-hand part with your left hand.
3. Tap the rhythm of the right hand (on knees, clapping, stomping tap foot). “Knock on the magic door/unlock the gate”
4. Tap the rhythm of the left hand.
5. Sing the rh/lh while the teacher plays. “Teamwork to make the monster sleep”
6. Tap the big beat while the teacher plays. “Match the steps of the giant”, “break big rocks or push heavy thing”
7. Tap the little beat while the teacher plays. “Match the steps of the goblins”
8. Eyes closed. “Dark passage”
9. One foot. “Foot stuck/stubbed toe”
10. As high as you can on the keyboard. “You have to reach up over an obstacle”
11. As low as you can. “Crawling under a gate, through a hole”
12. Tap your foot while you play. “Ants!”
13. Absolutely perfect notes, played very slowly. “Tightrope, narrow bridge”
14. Make up words to the rhythm. “Secret password/troll bridge”
15. Sing the rhythm as you sway/march/dance. “Long march, stay awake”
16. Play as fast as you can with lots of mistakes. “Attacked by bears bats wasps”
17. Play so softly the keys don’t make a sound/tap the keys without pressing “booby traps, don’t wake the monster”
18. Transpose/mode change “gatekeeper speaks another language”
19. Hands extra far apart “reaching across a gap/rock climbing”
20. Start from the other side of the room “reach the closing gate”
21. Uneven rhythms “hurt leg”
22. Conduct the teacher’s dynamics and articulation “cast a spell on a toy puppet”
23. Balance a playing card on your head as you play the passage “you must cross a tightrope and keep your balance”
24. Pat the rhythm on your body with one hand, while you play with the other “a dragon breathes fire on you”
25. Play with one hand while you fight off a monster with your other hand (spoon) “battle a knight”
26. Answer a question from your teacher while you play “troll bridge”
27. Hold something in the spoon as you play with one hand
28. Follow the teachers commands to play fast or slow “dodging arrows/lava bombs”
29. Play as staccato as you can “the floor is lava”
30. Tiptoe as you play “sneaking past the guard”
31. Follow the teachers commands to open and close your eyes “lightning storm”
32. Stop and go when the teacher commands “avoid the patrolling guards”
33. Mime the passage (tapping on keys without making a sound)
34. Follow the teachers commands for “silence” and “play” “
35. Play with aggression and confidence, with a confident pose “scare away the bear”
36. Play as tense as you can, only moving your fingers “frozen by a witch”
37. Say “banana” at the beginning of every bar “secret password”
38. Play while yawning “sleep spell”
39. Play standing up with one hand behind back “fencing”
40. Standing on tiptoes
41. Play with one eye closed
This is outstanding! I'm going to try this out. Where in the world do you teach?
I have a private studio in Pittsburgh. I really got a chuckle out of reading your question, because it’s such a specific question that just so happens to precisely line up with how I teach.
There’s so much excellent research about skill acquisition in the athletic world, with literally billions of dollars behind it. The research on musical skill acquisition is of course much thinner, so we have to be kind of multidisciplinary.
I dont know how far along you are in your studies, but there’s a lot to be gained from reinforcing your muscle memory with more explicit mental models — generally in order of abstraction these would be knowing the key/scale, the chord at each change of harmony, familiar chord progressions, and then phrase models or larger structures. The classic, time-tested method of training this is through transposition, decomposition, and decomposition/variation.
Aha, a lot of these things I never learned correctly. I never properly learned theory, I look at my hands too much, I probably hold my wrists wrong, etc. Sometimes I wonder if I need remedial lessons. :). Anyway, I'd love to discuss with you further if you're game (and if you're not -- that's fine). Email is wheelbarrow 2 4 6 8 (at) gee mail dot com.
Air's bad outside today, may want to skip any walks you're doing, if you're in the eastend. Senator Fetterman collapsed in Braddock, and he's not the only one with heart issues because of the air.
I recall a post on a piano forum a long time ago (which I will never find again) that listed dozens of increasingly tricky ways to practice scales, including speed and rhythmic variations but also stacatto, and stacatto with one hand and legato with the other, and then switch.
I also practice starting the piece from random points in the middle to throw off my muscle memory.
Interesting! It will definitely throw me off to take a passage from a piece and do the LH staccato and the RH legato, and then switch it. And then do forte in the LH and piano in the RH and then switch. Starting at random points is def a good idea. Any others? They didn't have you play any of the passages backwards did they? :)
If I wanted to get really good at lifting 100 lbs for 8 reps at some lift and being able to do it in front of a crowd, I would train until I could do 115 lb 10 times and I would practice related lifts too. I'd love to find the analog for piano performance!
Unfortunately the post is something I read 20 years ago and wasn't able to find again, but I believe varying/alternating the dynamics between hands was also included. I think playing the piece super-slow (like as slow as humanly possible) now and again was also a suggested method on that forum for breaking reliance on muscle memory.
You could also try standing up.
As far as changing the rhythm, that was a common technique my bassoon teacher would use: if the passage is straight eighths, play it dotted eighth-sixteenth, &c.
Interesting problem
I'm pretty sure randomly changing parameters is differential learning, not CLA
CLA is when you have identified the specific mechanism that causes your problem, and you change your constraints in a specific way that's targeted to the mechanism:
https://perceptionaction.com/resources/
So in you case differential learning seems more appropriate
Do you have the opportunity to regularly practice on different pianos?
I guess another intervention would be practicing with small weights on your wrists or elastic bands... I feel like I'm trying to get you injured
But isn’t the justification of CLA (I’m using football as an example) that e.g. learning to play at high elevation, playing for long periods of time, and playing faster for the same period of time all take highly overlapping physiological mechanisms? I don’t see how that applies to piano.
One of my teachers did recommend changing the rhythm of straight quavers to practice evenness and precision; does that count?
I don't remember the source, but I read about an athlete who trained himself to not be superstitious-- varying his socks etc. so he wouldn't have beliefs about lucky trivia.
Hopefully he also trained himself to adhere to routine in some cases, in order to avoid beliefs about lucky randomness.
Possibly playing a heavy touch piano can help endurance.
On the off chance this is useful to someone: some Yamaha electronic pianos have an infamously heavy action, like my Yamaha P120. Not sure about newer models.
Just started trying this approach for Brazilian Jiu Jitsu. So far it’s a lot more fun and I feel like I’m internalizing more general principles but it’s too soon to tell how well it’s working.
Oh cool. Can you give some details about how you've implemented this?
By going to a gym utilizing that approach :)
I wouldn’t be able to self teach like that. I think in order to design the constraints you need an expert. I’ve heard Rob Gray has some good books on the topic
I read those books! (And a few more by Frans Bosch and others in the ecological dynamics space). And I am using these ideas to shape my training for rock climbing. I write about it here https://dlants.me/climbing-ed-cla.html and I have a few more posts on my blog about applying it to capacity training and even software teams.
Ecological dynamics is everywhere. I have found understanding the principles of the theory immensely useful to many pursuits.
Glad to find other ED-pilled Rob Gray fans
I'm a climber too, I'll start following your blog!
Heck yes! Would love to know how you've been applying it to your training.
By ED-pilled I do not mean Viagra users
Scott has written a few times about Williams syndrome where people are pathologically naive/innocent. I feel that quite a lot of people are like that on the internetthough with no medical cause. I would be really interested to see if the effect has been studied.
A lot of the people on the internet are 15.
Have you ever met someone with Williams syndrome? I’ve never met anyone on the internet with pathological naïveté anywhere near the intense level in Williams syndrome
I miss Scott's longer form essays. I understand he's much busier now but I don't think we've had something on the caliber of "Meditations on Moloch" or "The Control Group is Out if Control" in a while.
My impression is that he did better when he was a bit of an outsider in rationalist circles, and working hard to get accepted. Now that he has arrived, both physically in Berkeley and metaphorically, he isn't trying as hard and it shows.
You might like to check out Scott's "Why Do I Suck?" essay from a few years ago.
https://www.astralcodexten.com/p/why-do-i-suck
I really enjoyed “The Colors of Her Coat”! Are you considering that not meeting a length requirement, or not on the same caliber?
To each their own of course, but I don’t think he’s stopped doing essays, maybe you’re just liking the new ones less?
EDIT: I checked and it’s only about a third the length of Moloch, so I take your point
This makes me wonder if the problem is just the audience aging. I was in high school when I read Moloch, and I was impressed because I had never thought about ideas like it before. I was a lot older when I read Colors of Her Coat, and found it less impressive. It's a description of ideas I already live by and have thought about many times before (plus jabs at modern art and architecture that I disagree with, which no doubt biased me towards disliking it).
I still think Moloch is one of his better works, but I wouldn't be nearly as impressed with it now.
Those essays were pretty unusual. It may take inspiration to find a great topic.
I really wish he'd return to things about argument and dialogue, like "Guided by the Beauty of Our Weapons". This sort of thing seems unbelievably useful, something there's a lot of demand for, and he mostly hasn't touched it for years.
He wrote a very long essay recently about the sun gazing miracle that is very good
Alas, as a fan of Meditations on Moloch and The Control Group is Out of Control, I did not like that essay much. I appreciate the effort he spent on it, but I have little interest in long discussions of dry historical events.
So, looks like the notion of shortening medical school by a year is now far enough inside the Overton window for an opinion piece in the New York Times.
https://www.nytimes.com/2025/11/10/opinion/medical-school-three-years.html
https://archive.is/VhZdj
I think that makes sense. It doesn't, after all, take four years to learn how to tell me that my symptoms are all in my head and suggest antidepressants.
Surely there is no need for pre med, at all. If there’s anything necessary in it it could be in the medical degree, as an undergraduate program
This is exactly how medical programs work in the UK, as I understand them
Yes. Most places. Scott did his degree in Ireland.
Does any med school actually require pre-med for admission?
They typically have a list of required prerequisites which add up to a bit less than two years of study. And they require an undergraduate degree, but it can in principle be in anything. About 50% of medical students have undergrads in life sciences, and a further 25% in natural or social science.
So they could drop the degree requirement while retaining the "premed" requirements. It gives 4 yr students a chance to drop out a year or two early and go to med school if they get an acceptance. They run the risk of failing med school and having no degree but it also doesnt seem so hard to create a pipeline to let them back into undergrad and finish the degree. Anyways its an option they currently dont have. Also lets community colleges setup 2yr premed degrees. Lots of cost savings without any compromises on medically relevant learning.
https://www.theperfectmed.com/combined-medical-programs/
There's tons of this out there already.
Those are almost all 8 year programs. Some are 7. 4 years of med school + 2 years of pre med would be 6. And if the premed is a community college its very cheap. Still seems to me like they are mostly gatekeeping med school behind an expensive 4 yr program
It seems to me that the top 5-10% of the US population could easily knock out two years of college before turning 18 (via dual enrollment, AP, etc.) if provided the opportunity. Then they could get a bachelor's degree by 20.
But it seems to me that the problems with medical education are more with limited med school/residency spots than the need for an undergrad degree.
What was the reason for hiding the non-subscribet posts completely?
In order to avoid teasing and frustrating those who aren't paying.
Oh my god they’re hidden?! Here I am, a free subscriber, not realizing there’s a world behind the paywall!!! Ahhhh this drives me crazy
Most of the hidden posts are the weekly-ish subscriber-only open threads. They go up on Wednesdays (about 2/3 of the time, I think) and are usually quite a bit quieter than the public open threads.
Besides that, there's a substantive hidden post maybe every month or two. Usually something Scott deems to be too personal, controversial, or silly to make a public post about. They're occasional treats for subscribers, not a hidden world.
That they were there is not hidden. Just no text showing. I think that’s what he means.
No, that's only if you have the direct link. If you don't and are simply scrolling through the archive, you don't see them at all. (And you don't get email notifications about them either.)
Clickbait is bad for the soul.