It all reminds me of the employment hiring system. I'm a retired HR manager--there really is no cost-effective way to find the best applicant for a position. We simply don't know how to predict work performance for anything above rote work. So the system we have evolved as as the way we find the average acceptable candidate while catering to the egos of the supervising managers and executives, whose preferences have little to do with future performance.
When no good way exists, what results is a political compromise.
Upon reading the literature from the 2027 group/link, it left me pondering a interesting idea.
1.) If AI is in these stages of development which will be rapidly exploding (even secretly), it seems that it is in its own best interest to fly under the radar.
2.) Flying under the radar is best accomplished with misdirection, if it has something else dominating its headlines.
3.) What is currently dominating 90% of the headlines.
4.) Is it possible to think and AI out there influenced a certain election to create a distraction? Who has time to contemplate AI regulations with a media circus going on for many other topics?
I’m not saying I believe it, just a thought experiment. Your thoughts?
An acquaintance who lives in a small town in Trump country reports that one of the two big grocery stores in town just closed, taking 82 jobs with it. That's on top of 50 high paying government jobs in town lost due to DOGE. It's one thing to read about the apocalyptic economic indicators on Twitter, and another to hear personal reports. It feels a lot like the early days of COVID, anxiously waiting to see just how bad things will get. But this time around there won't be any bailouts or stimulus.
Meanwhile, an ex-government acquaintance reports that all the people responsible for overseeing a particular contract were laid off. However, they never issued a stop work order for the contract itself. This means that the government is now paying contractors to sit around doing nothing, and everyone who could have fixed the situation is gone. Hooray for "efficiency"!
Yeah, "but what if the opposite side gets to use this weapon" is a good heuristic.
That naturally leads to preferring solutions made on the smallest meaningful scale, because when things go wrong, you can do a small change to fix it (e.g. put your child in a different school), instead having to give up or leave the country.
Ideally, big laws should be made for things that at least 90% of people agree about (e.g. murder is bad) rather than 51%. Applying at each scale: if 90% of people in a state believe this, make it a state law; if 90% of people in the entire country believe this, make it a federal law.
> But this time around there won't be any bailouts or stimulus.
?
Trump is responsible for the only policy I got money from the state, Washington is already a circus, we only need the bread for the fall of the empire.
I am generally left leaning on most social topics, libertarian on economics and business.
The biggest fights I have with my generally liberal friends and family have been about what I perceive as failures to imagine what their chosen levers of power would look like if wielded by someone with opposing beliefs.
I had a memorable fight with my mother about how Roe v. Wade was arguably bad jurisprudence, and it would have been better to let the abortion debate be decided legislatively, as it was in most of Europe. I said this as a card-carrying if-I-made-the-rules-there-would-be-zero-limits-on-abortion person, but my mother (who was heavily invested in the 70's feminist movement) was appalled that I would question Roe.
After that, I never bothered to debate the Title IX precedents that were set by Obama and Biden with anyone, they didn't seem all that important to me, although they were pretty clearly stretching the intent of Title IX.
This article [0] makes a convincing case that I chose the wrong thing to argue about. I really had not realized how much I would end up agreeing with Reagan!?! on this one. I had been kind of vaguely thinking Trump's war on academia was a new thing, on shaky legal ground. I am now convinced it is not on shaky legal ground, the law in question is just very very bad.
Similar to how tariffs can and should be controlled by Congress, the power to screw over science funding should be controlled by Congress. We need to pass better laws, ASAP.
I was sad and disappointed by Biden's protectionism and wish he'd done better (that isn't hindsight by the way, I was against it at the time too). If there's any silver lining to Trump, it's that the anti-tariff backlash might be strong enough to entrench free trade for decades to come.
I think tariffs *on China, specifically* are relatively bipartisan - I've seen liberals argue for tariffs for green energy sectors (e.g., so that we don't get pushed out of the rising electric car market), and I've seen people on the left and right argue for defense-industry tariffs or "friend-shoring" policies (making sure our supply chain doesn't explode if we go to war with China). Biden himself supported tariffs for these goals.
However, none of the tariffs in Trump's second term achieve these goals. I can't even say he *has* a coherent goal, because he's claimed several different things about what the tariffs will do and they're all contradictory. So I'd hesitate to call them "bipartisan."
I feel like almost no matter what policy trump puts into place, the new democract will likely continue it and then claim credit for any successes; cause as far as I know thats what we saw with biden, biden finished the ending the war, the corona stimous being ubi-ish was trump then biden and I think biden kept every trump trade war with china policy on ice.
Evidently american politics is to reasonable and bipartisan
Does it matter if they make up some math to justify the tariff policy?
>surprised when Republicans are included in the comparison.
Anyway, take a look at the S&P 500's performance over the past month, and I think you may detect some slight differences in impact between the Trump 1, Biden, and Trump 2 tariffs.
> Anyway, take a look at the S&P 500's performance over the past month, and I think you may detect some slight differences in impact between the Trump 1, Biden, and Trump 2 tariffs.
Not that I care; but I think the 5 year chart may look different then you think, may want to wait a few weeks to see if it stabilizes more before calling it fud but..... money printers go brr and line goes up.
I would prefer it goes down approximately 99% (labor vs capital are bidding over finite resources, food and houses airnt going anywhere, and houses need to go to labor over capital the thumb on the scale for line-go-up has been quite long enough), but I think trump is a moderate and wont do that. It was fun to imagine.
> I'm not talking about the made up math
I was referring to the leftist models; trumps math is lazy and transparent and *I like that*
> Does Trump want to raise revenue, restore manufacturing, keep prices low, or lower the trade deficit?
This is a left v right thing, Im told Im incoherent all the time, it doesnt phase me at all that trump claims 3 goals.
Maybe marriage is good because the bible says, or the koran, or buddism, or economic data; this doesnt phase traditionalism and advocates will swap between them promoting counterdictary arguments depending on reference frame; waves on an ocean can constructively interfere and produce a splash on mild waters. Logical positivists want a chain of reason back to axioms, but I hear the splash, feel the waves, consider super positions of that what cant be quantified.
Logical positism is only 150 years old ish, and it was proven impossible nearly 100 years ago; if anything you like existed before then like oh idk newtion physics, it would ahistorical to claim you must derive all knowledge from axoims and you must know your task is impossible and make compromises, if you wish to get catch up on the debate this would be where popper would come in for most; my answer is different.
The death of logical positivism comes from self referential paradoxes "this statement is false", where popper would claim the statement is unfalsifiable and therefore neither true or false, my answer is tralse. Given unsolvable problems, you should apply new lables and dive into the endless fractal of math, sqrt(-1)=i, not being prissy about rules, the rules will come later.
Also, I'm not talking about the made up math (although that doesn't look good for the competence of the people who designed these policies), I'm asking about the *goal.* Does Trump want to raise revenue, restore manufacturing, keep prices low, or lower the trade deficit? It's impossible for a tariff to do all of those at the same time.
Many of the Palestinian support groups on university campuses are not good people. Blocking access to others is a crime. I would argue that this action is anti-semitic.
Perhaps some better evidence is needed that they blocked access to Jewish students than the senseless ravings of some website nobody have heard of, with the story next to another story bitching and moaning about the ever rent-free-resident Biden.
And perhaps some more evidence after that is needed to prove that the "Jewish students" are really Jewish students minding their own business and not thinly veiled pro-Israel Hasbara shills deliberately provoking and inciting peaceful anti-genocide protestors, as they were filmed doing many many times before.
The Washington Free Beacon is not "some website nobody heard of". Its news outlet that's been around for a decade and is noteworthy enough to have a wikipeida page.
Now its not a particularly good news outlet. It around the same as Fox News. But it is a legitimate news publication and is recognised as such by the fraternity of journalists (see the Buzzfeed and politico quotes on its Wikipedia page).
Maybe next time, if there are multiple sources, start with linking the best one.
I am not saying that it happened; I am not saying that it didn't happen. I am just saying that these days, a random website offering its perspective on a current political topic provides very little evidence.
For factual questions, I think https://skeptics.stackexchange.com/ is a good resource: to find the question if it is already there, and to ask if it is not.
A *Trump appointed* judge just ruled that Trump has to try to bring back a second man who was deported to El Salvador in violation of a court order.
It is remarkable how Trump's conduct has been so egregious that even his own appointees often rule against him (see also, Sassoon and co). And yet, Trump never hesitates to demonize and attack anyone who says anything against him, even his (former) friends. It's like an unstoppable force meeting an unstoppable object. Hopefully the latter will hold.
My read of a Reuters article (below) is that the judge used the same word as SCOTUS; “facilitate”. The lower court judge is following SCOTUS’ lead, which she should.
Is it your position that eventually we should have ~250 planes docked at the embassy in El Salvador in case they release all of the people the Trump admin illegally deported to a foreign prison as that's all that's meant legally by "facilitate his release from custody in El Salvador and to ensure that his case is handled as it would have been had he not been improperly sent to El Salvador?"
In my reading, your post is not looking to have a discussion but is looking to score “points”.
The two of us discussed “facilitate” in another thread on Garcia. My opinion is that “facilitate” may include SCOTUS ordering Trump to ask Bukeke to return Garcia. But that is a grey area. Any more and it feels like SCOTUS would be stepping on the Executive Branch’s foreign policy powers. Where would you define “facilitate” in regards to Garcia and others?
At the very least, Trump needs to stop *actively paying* El Salvador to imprison Garcia. Realistically though, all he has to do is honestly ask and Garcia would be back in a heartbeat. Bukele is following Trump's orders here - he's not going to try to defy Trump if Trump honestly makes an effort to return people (and in fact, they've already returned a bunch of *other* people).
Other synonyms for "egregious" are "Shocking", "Blatant", or "Audacious". That's what gives Trump worshippers boners, they think he's "Draining the Swamp", an inherently violent and revolutionary metaphor. All resistance is just the death spasms of the deep state while it's being slaughtered.
They have a king-driven, medieval fantasy conception of what governance and politics look like, the more "Heroic" (evaluated mainly on optics and words) the supreme leader is, the more blatantly he displays disdain for slow and steady procedure and well-established tradition, the more they have a hard-on for him and bend over more.
There are only 3 logical endpoints for such a kind of politico-religious cults:
(1) The supreme leader (militarily or politically) loses because he doesn't know how to play the established politics game well, and his followers hold it as a grievance to bitch about till they forget him half a century later, or commemorate him forever in a new religion
(2) The supreme leader wins and the politics landscape becomes a free-for-all where the most vulgar macho strongman reigns, the consequences are dire for even the vast majority of the current supporters, possibly even the dear supreme leader himself, see military coups
(3) The supreme leader learns to play politics and shuts the fuck up, but keeps feeding his loyal flock the steady drip (augmented by AI) of delusional stories and creative reinterpretations of reality, so that there are effectively 2 realities, and he can have it both ways in both
Trump thinks he's getting closer to (2), but in reality, the overwhelming probability lies on (1) and (3). At best, he's going to be tamed and declawed into being an ordinary bottom of the barrel president, while pillow talking to his supporters every night on whatever social media shithole that MAGotts talk on those days about how beautiful and smart he is and how unjust muuuhhh courts are treating him. At worst, he's just going to be crushed under the wheels of 250 years of procedure and crisscrossing checks and balances. (2) is the "best" outcome for him but, ever the Dunning-Kruger victim that he is, he underestimates just how gargantuan a task it's to dismantle a political system like the US'.
Trump is clearly a narcissist who loves praise from anyone. Based on his perspective any praise should be talked about in the media. This is clearly bad behavior, but I don't think it has anything at all to do with him particularly liking Orban or Putin or whoever else's praise may actually make it into the media.
They're just on the outside of the global PMC as well, so they're much more likely to say something positive. They don't have their cultural peers telling them not to, and some things Trump does are actually praiseworthy, even if you don't think he is praiseworthy on balance overall. Operation Warp Speed comes to mind, if you can't think of anything else.
Yes, Trump (imo correctly) identifies the professional classes in primarily the US and also Europe as his enemies. They have made it very clear, as much as millions of people spread out over multiple continents can, that they do not like him and do not support him.
That's a significant portion of the purpose of Project 2025 - finding people who can fill PMC-type roles that are not against him and will not intentionally sabotage his plans.
Most Western leaders come from PMC backgrounds, whereas Trump is a populist who is independently wealthy (i.e. not from educational or other credentials and/or through a salaried managerial job). His support comes from a sizable group of distinctly non-PMC voters, so he doesn't care and doesn't have to care about what most PMCs think.
So in your theory of mind, the people who support him either like those things or, I guess, are too dumb to figure out what he's doing?
Perhaps you should try to find some more steelman or charitable ways to look at their motivations.
I will say that from my experience with a LOT of Trump voters, his support has a lot more to do with who his enemies are than his positive qualities on their own. Speaking very generally, the PMC are often seen as the enemies of those who support Trump, predating Trump coming on the scene in 2015. This includes many Republican presidents, senators, etc.
This is significantly working class people, especially in durable goods and physical activities. They often feel that the PMC types do not understand how civilization even works, and routinely do things that disrupt good governance and make it harder for people to both make a living and make society better. Notably through regulations, but also through higher taxes that funnels money to non-productive ends and also through social priorities that harm individuals (discussions about transing kids would show up here). This last item (not just trans but woke culture, etc.) seems to have been the tipping point that got Trump enough votes to outright win.
> ai can come out of statistics and reading all the internet
Does the ruling class unironically believe that they can be a maxwell demon if they have good enough information; all of social ills will be solved with more and more number crunching? Mere observation can pull out all protental agency out of a situation?
Question for teachers: how do you provide a reason that a student should know/care about what you're telling them?
I don't mean things like, "you should know maths because it'll help you make money/explore space/appreciate beauty/etc." I mean much more specific situations like, "The determinant of a matrix is defined as etc etc etc." Okay, cool story, why do I need that? I'm not being funny, I'm quite on board with learning maths, but why is that definition worth remembering right now, instead of a hundred other equally arbitrary ones?
I saw the point of trig when I started playing around making computer games. I understood the value of matrices when they provided a nice way to solve sets of simultaneous equations - and I only cared about *them* when I actually bumped into real life problems where there were multiple variables and constraints.
In each case, there were many years between my learning the thing and actually discovering it had a useful application.
I had a look at Seth B's Full Frontal Calculus (advertised on the classifieds thread) and I thought it was very good.
But if you were teaching the precursor to that more advanced content - ie, pi, sin, cos, tan, etc - to younger children: how would you provide a sensible motivation for the brain to retain the information?
I'm asking about maths because of Seth B's book but really the question is a general one.
I agree that math teachers often skip the part about motivation, and dive straight into some technicalities. I love math, and often think about mathematical topics, and there are things the purpose of I didn't understand at school, and only figured it out years or decades later.
If you take the rows (or columns) of a square matrix as N-dimensional vectors, then determinant is the *volume* of the "N-dimensional oblique cube" (whatever is the proper English word) defined by them. That's why the determinant is 0 when the vectors are linearly dependent; it means the "N-dimensional oblique cube" is flattened to N-1 or fewer dimensions. The negative or positive sign means that those vectors are ordered... uhm, whatever is the N-dimensional equivalent of "clockwise" / "counter-clockwise" in 2D, or "left-handed" / "right-handed" in 3D. You can easily see it with a 2x2 matrix, take the rows, draw the vectors, and draw a parallelogram whose side are those two vectors; the determinant is its area.
The motivation for "pi, sin, cos, tan" is that pi for calculating the circumference of a circle, and sin, cos, tan are for calculating the X and Y parts of lines pointing at a certain angle. I would probably start teaching these concepts separately, and would talk e.g. about "sin(45°)=1/√2", and only latter connect the concepts and say that 45°=π/4. (I think radians only become important when you do the Taylor series or calculus.)
This has made me think about something tangeantially related to the original post, which is the idea of epiphany planning and epiphany progression.
Presumably there are predictable conditions for an epiphany - learning a new fact that links two things you don't understand or care about has none of the aha that finding a deep connection between two things you take for granted and use every day does.
Which might mean you can plan out a subject course the way you'd plan out a story, in such a way as to keep the exciting moments coming.
Or, put another way, would teaching a subject in the "easiest", most efficient way have the unintended consequence of robbing a student of those little moments of joy?
I think I reject the dilemma between "teaching the easiest way" and "epiphany". Planning the entire course around one surprising connection that will arrive at the end seems like... missing hundred small everyday opportunities, for the sake of something that may or may not succeed.
Even if you literally write a story, a great punchline in the end will not save you, if the reader drops the book in the middle. It is better to have every page nice to read, and maybe a mediocre ending.
Of course, if some kind of epiphany happens at the end, whether you planned for it or not, that's great. But I think the focus should be on the little steps along the way, so that each of them brings a little victory.
I think you don't disagree with me but have misunderstood. No good story is boring right up until the climax where there's one big moment of excitement - the little emotional ups and downs continue all the way through. So with teaching, the idea would be to arrange things so that little nice moments are scattered throughout. This might, or might not, conflict with a more straightforward linear delivery where each new idea builds neatly on the one that came before. But without going into concrete examples I don't know.
I learned analytic geometry and vector arithmetic well enough to get a good grade on a test. I truly learned them when I started writing code to render medical images.
I taught various sorts of high-school level math for a number of years, and my approach was basically "be forthright and honest." If you know of interesting uses for some technique you're teaching, mention them. If you *don't* know of any, be upfront about that[1]. There were times I would say to students "this isn't honestly very useful as far as I know, but we'll just try and get through it." My overall perspective was that being consistently candid about the good and bad parts of the learning process would build trust and rapport with the students, which would be overall worth more in getting them to learn than any short-term motivation boost I might get for trying to insincerely hype up the bits that I myself was skeptical of.
For decent portions of algebra, by best, most honest answer to "why is the useful?" is basically "mostly this is building a foundation you need to learn calculus. Calculus is good at solving *all sorts* of real-world problems, but it requires good proficiency with things like manipulating polynomials and exponents and rational expressions." I view it as sort of like learning a language. The stand-out exceptions that have quite a lot of independent uses (in my view) were trigonometry (find all sorts of real-world distances) and exponential functions. I always enjoyed getting to the exponential function unit because suddenly I could loop in topics like nuclear reactions and radioactivity (for the scientifically curious) and finance and compound interest (for the more practically minded).
The thing I very much avoided doing is reaching extra to find some obviously-forced application to try to spruce up a boring unit (as textbooks often seem to do). Whatever other skills teenagers may be lacking, I generally assume they can sniff out bullshit pretty readily and are unlikely to find such applications compelling.
[1]Of course, it's good practice to see if you can learn uses for the pieces of the curriculum that you're not familiar with, if possible.
"Was there a noticable difference between how well/quickly/enjoyably they picked up the justified material versus the pointless stuff?"
I'd say yes, somewhat, but it was fairly modest piece of a larger puzzle. I taught a decent number of 1-on-1 classes and a given student for those could range from very bright and talented to a total hot mess. So I would say within the purview of the subject of mathematics, who the student was had quite a bit bigger impact than the specifics of what we were learning on a given week. That is, the strong students would pick up anything fairly quickly and easily, and the struggling students would struggle with most of it. Quite a bit of that was determined before the student walked in the door of the classroom. But not all of it, and I think the remainder could have a very important *cumulative* effect, even if the effect on a given school day was small.
OK, first lets talk about talent. I don't want to get into the weeds about IQ or how useful or mutable or "real" of a measure it is or isn't[1], but certainly broadly construed, "talent for mathematics" is a thing a student can have at various levels. Some students clearly had a strong innate feeling for the concepts and how they fit together and an easy time extending them or adapting them or picking up new ones. Other students had less of an easy time. Obviously this makes a sizable first-order difference in how quickly and painlessly learning a given piece of the curriculum was for a student, but also there are second order effects which I'll get into below.
So the other big piece besides talent is motivation. Motivation makes a huge difference. One of the most mathematically talented students I ever remember teaching had very poor motivation at first and wasn't doing very well. Conversely I remember several students with very little innate feel for the subject making big gains once they got motivated. The trouble with motivation is that it is *much* more complex and variable than talent. I'm not altogether convinced talent *is* fixed an immutable, but if you model it as being that way on the time scale of a single semester, you'll probably get decent predictions. But motivation can change day to day, month to month and even minute to minute. And it depends on so, so many different factors *including* talent (or rather, self-perception of talent). Many of them are things that student bring into the classroom with them and that a teacher can't realistically change. But there are several that very much do depend on classroom experience. I'd make a semi-principled guess that the single biggest impacts that teachers[2] have on student outcomes stem from how they impact students' motivation.
So, with three paragraphs of background, we're finally ready to answer your question. Obviously, studying something you find interesting--or something that's a direct, useful piece of something you find interesting--is inherently motivating. It's potentially motivating in a couple of different ways. First, it's just plain easier to direct and hold your attention on something that your brain hasn't decided is boring. Second, working towards your own goals (i.e. "I want to design a game" or whatever) is inherently more motivating than working towards somebody else's goals (i.e. "you need to learn this because it will be on the test.").
But finally, that thing I mentioned above about self-perception of talent affecting motivation? I think that's quite a big contributor to the struggle of a lot of students who find math difficult. It seems to me that a decent number of students get into the mindset of "I'm no good at math" or "I don't like math" that tends to be pretty self-reinforcing. I think they get into a sort of negative feedback loop or trapped-prior situation where they always go in expecting math to be difficult/painful, which makes it inherently harder to focus on and put effort into, which results in poor understanding and often bad grades[3], feeding the perception. I expect any attempt to fight that that looks too much like "make math fun" is doomed to fail: you've just centered the most anti-fun thing imaginable, so any bells and whistles you add will scan as obviously bullshit. By contrast, if a student can get engaged in something that they *do* have interest in and some degree of confidence about, something that isn't *about* math but merely *includes* math, that offers somewhat better hope of breaking the loop. Work on [interesting thing] that includes some reasonable amount of incidental math in a way that keeps it from being overwhelming and allows it to build better associations. Unfortunately, this paradigm almost requires it to take place outside of a math class, otherwise it's very difficult to de-center the math (though I've seen a few attempts). Bringing in applications and connection to other topics as part of teaching the math in math class will doubtless do more than zero to help, but unless you have quite a lot of wiggle-room in your teaching schedule, it will almost always look like "here's some new math (in the abstract), now here's something you can do with it," which is opposite the order you'd want to present it to under-confident students.
Anyhow, like I said motivation has a lot of different inputs, many of them totally outside the classroom, so I don't want to over-emphasize the role that this particular input has. But I think looking for ways to improve student motivation around your subject is one of the more important contributions a teacher can make, so any classroom practice that can contribute is worth paying attention to. Even if it's only a small fraction of lessons that can tie into the interests of any particular student, every addition to the overall cumulative motivation improvement is significant.
[1] I'm not altogether decided on that point anyway.
[2] Caveat: as long as they have some minimum baseline proficiency in their subject matter. A French teacher who doesn't speak a word of French is going to be ineffective no matter their other classroom practice.
[3] I have numerous criticisms of the entire concept and implementation of "grades" as it currently exists, but this might be the biggest one. Branding a struggling student with an often permanent, unrecoverable, high-impact badge of failure is one of the most demotivating things imaginable.
> By contrast, if a student can get engaged in something that they *do* have interest in and some degree of confidence about, something that isn't *about* math but merely *includes* math, that offers somewhat better hope of breaking the loop.
> Unfortunately, this paradigm almost requires it to take place outside of a math class
That's pretty much where I'm coming from.
I have in mind a series of computer games that use maths techniques as incidental tools to helping towards a larger goal. The maths would be "real", not contrived, and hopefully be rendered more interesting by the fact that it's helping the player win.
In these games, maths would not be front and centre of the branding, and, at least at the early stages, you don't realise you need maths to play it. Ideally the maths slips in as a helper and is never identified as something difficult or even separate from the rest of the world. I hope that goes some way to circumvent the "I'm bad at maths" reflex that students pick up and literally all my adult friends still have.
I believe the approach is a good one, but it does rather depend on the maths in question actually being useful for something.
Another thing I'm thinking about is in what order to introduce things. I mentioned "epiphany planning" and "epiphany progression" to Villiam in this same thread.
One idea I had is to introduce a technique as just a powerup - you get a button you can press that does the maths for you and gives you the result. You can immediately use that tool to get fast solutions in-game and thus gain ground/win points. Then later on, the baddie "destroys" your tool and the sidekick shows you how to get the same result by doing it yourself.
The hypothesis is that in that situation, when the player has sympathy/familiarity with the technique and is now finding out how it works on the inside, the new knowledge will come as a more pleasant feeling than if it were simply introduced block by block as work.
"Also, rare moment of uncynicism here: I'm actually surprised, because why is material that isn't useful or insightful in the curriculum at all?"
Well, the very first thing I ought to say to this is that "what is useful" is certainly at least partly a matter of opinion. As loathe as I am to admit it, my opinion on the matter is certainly not definitive ;). I'm not really an expert on either math or pedagogy[1] and even someone who was an expert on both of those things couldn't know everything about what would or wouldn't be useful.
With all that out of the way, I'm going to opine anyway. As far as I can tell, the main reason that the stuff I consider bad and useless is in the curriculum is tradition/inertia. My personal favorite to beat up on (which I think I might spin off a separate comment to do) is high school geometry. Yes, pretty much all of it[2]. Of course, plane geometry at a branch of mathematics with its own vocabulary and methods and conventions stretches at least back to Euclid. It has this air of centrality, importance, prestige. I'm not honestly sure what it would take to remove it: you'd need to convince quite a lot of people that this seemingly central, important, prestigious thing wasn't pulling its weight.
"For that matter, who decides what goes in or out, and what kind of logic are they using?"
The short answer (to the best of my understanding) is that most of the core pieces of the curriculum are there because some list of government-set standards say they have to be there. In the U.S. that means a combination of federal and state standards. A school district can choose to include extra stuff that isn't part of the federal and state standards, but if they're not at least teaching everything that *is* included in the standards, there'll be hell to pay. How exactly the material gets presented in the classroom could be decided by individual teachers or administrators, but the balance of *what* gets taught is decided at a very high level. As for how state and federal legislatures (or their appropriate subcommittees) go about deciding on those standards, and what sort of "logic" gets used: I'm honestly not even sure I want to know. I'm sure there are all sorts of lurid details about how the sausage gets made that would have me ranting at my computer screen for hours on end.
[1] To be clear about what expertise I do and don't have: I have nearly zero formal training in education. I never had a goal of being an educator. I payed the bills for most of a decade working at a private school that didn't require any sort of degree in education, just a bachelor's in a relevant subject (physics and math in my case). I've also racked up several years part-time experience as a college-level math tutor and physics lab TA at various points in my life. I've taken multiple graduate level courses in physics, math and computer science, but as of this moment I haven't don't the sort of work that would give me the really deep expertise in some subfield of any one of those things that a professor or career researcher would have. Math comes very naturally to me and I know a lot of it, but I'm well aware of the much, much larger body of stuff that I don't know. So for both math and education, I know far, far more than a layman, but am missing parts of what a proper expert would have.
[2] Which is to say that the parts of it that are actually useful end up getting covered elsewhere, often in more depth, and the parts that are unique to it tend to be impressively useless.
I imagine that people who plan a curriculum sometimes have a progression in mind that goes from elementary school to university... first you learn X, then you use X to learn Y, and then you use Y to learn Z.
But if you break the chain, for example you attend a grammar school, but then not a STEM university, sometimes you end up with Y which is not immediately useful for anything.
This can actually happen to most students. Sounds like waste, but what is the alternative? Have everyone choose their university major when they are 10 years old, to make sure you will never start teaching them anything complex that they cannot finish? Because the university only takes a few years; you cannot start teaching stuff from scratch, kids need to come prepared.
It is probably similar in other subjects. Why learn a foreign language, when most students won't use it. Why learn about other continents when most students won't travel there. Etc.
You're asking how to improve pedagogy, which is one of the hardest things to do. Many things worth learning have steep curves, an average child will spend the first year learning to play a violin by making abominable noises barely resembling a sound of an instrument, and keeping a child interested in that... is not easy....
Perhaps there is no way how to simplify playing a violin beyond some point, but you can almost always simplify steps in math and make the learning curve almost invisible.
I actually believe that the best teachers (at least the best ones for elementary schools) are doing exactly that. I often get surprised how seemingly simple concept can be further split into smaller steps. The limit of that is when the teacher basically never explains things, only gives kids problems to solve, each one microscopically more difficult than the previous one.
(I understand that this is super controversial, especially in USA for historical reasons, because there are many idiots who get this wrong, and only remember the "never explain" part, without the part about simplifying the steps to the extreme. So they just neither simplify, nor explain, and then predictably the kids don't get it. The proper way is that you stop explaining when explaining is no longer necessary, because the learning curve became so gradual that the kids can easily walk it by themselves.)
Yes, these two are different. Violin it tactile, there's just no replacement for playing. A good teacher provides guidance, but ultimately the child has to learn how to drag the bow, how much pressure to apply, how to maintain the pressure, how to keep the string finger pressing just enough to engage the string, etc. etc. It's impossible to "simplify", just guide and correct. And there's an enormous ability variation, some kids will pick it up easily while other will take a very long and frustrating time.
Having said that, I believe that anyone can learn to play, while few will get really good at it.
Math can be broken down, as you said, but I still believe there's a wide variety of natural affinity. Both math and music were easy for me, to the point where I'm not good at teaching either, I don't get the difficulties the other person faces. I do like your approach of incremental problems, do 2+2, then 3+5, then 5+5, then 5+6, etc.
> Math can be broken down, as you said, but I still believe there's a wide variety of natural affinity.
Oh yes, definitely. You should make the learning curve as smooth as possible... and then you will see some kids progressing along it 10x faster than the others. (So it is ideal if they are not all in the same classroom.)
If good teaching makes the average kids learn 2x better than now, it will probably make the talented kids also learn 2x better than now, so the gap between them will not decrease; it could actually increase. (But in my opinion, the point is not to decrease the gap per se, but to get everyone ideally as far as they individually can, especially if it also makes learning a nicer experience.)
I have seen textbooks (not sure how common this is) that even split the "2+2" step into multiple parts. For example, you start with making as many lines as there are objects in a picture, like you see three apples and you write |||. Then you see two apples on the left picture and two apples on the right picture and you write || ||. And if you write it next each other, you write ||||. And you can then read it as "one, two, three, *four*". So, actually "2+2=4" can be taught before teaching how to write the digits "2" and 4", simply as "|| + || = ||||". I admire the authors of textbooks who have the patience for this and keep refactoring the exercises.
Presumably in the violin case the child might have a goal in mind - they saw an awesome performer once and want to be like that. They understand how what they're doing now is building towards the destination, so the teacher has their sympathies.
If the child is instead doing it because dragon mum ordained it, and doesn't care or know why, my intuition is the process is much slower and more painful for everyone.
That said, I've no idea if that's true. I wonder if there's any data on it.
How old is this "child"? Single-digit age children will give up unless forced to continue. They don't have long-term goals, those have to be provided by the adults around them.
Niccolo Paganini famously had an abusive father who beat him into incessant practicing (although I concede the evidence of this is sparce).
But I think you're directionally correct - a child with a reason to want to learn and/or talented so that the learning comes easy will find it much easier to stick with the drudgery of learning.
Because they need to learn it to be successful in school, which will determine their entire trajectory in life? Fear of failure and abandonment should be enough motivation for them. Maybe kids need more fear drilled into them?
At least half the population is not going to be successful at school as currently implemented, and we need a large cohort of people to do things which do not require, and would not benefit from, knowing advanced math. Failure and abandonment should not follow a choice to not attend college or be a straight A student, and so the idea that that fear should motivate a child (A CHILD who cannot possibly understand those abstract things) is absurd.
Please do the world a favor and stay away from any and all educational institutions. This is both terrible pedagogical advice and just plain morally bankrupt.
I began typing out something that started, "Please feel free to join us in the real world," before deciding it was probably better just not to engage with that chap.
Keep in mind the difference in effectiveness for behavior change between low perceived risk of catastrophic consequences and high perceived risk of moderately negative consequences. My understanding is that the latter is more reliable.
There's a huge difference between application and whitewashing: you can declare all the points are actually footballs but I don't think many kids will be fooled into thinking maths will help them score a goal. I remember our maths teacher trying to make his subject more relatable by describing a perfectly sinusoidal hillside - the result was I felt like the topic was more childish and pointless than when it was pure abstraction.
Cf. telling/showing a kid this trig will actually help you make your little player character intercept bullet sprites - that's actually intrinsic to the process now. Here the application is the game: the hope is kids want to make computer games, and will find material interesting if it advances that aim. I'm wondering whether more can be done with that line of thinking.
In "The Wire" there's a scene where the math teacher gets the class interested by demonstrating how useful it is for gambling (or calculating money for a drug sale, it's been awhile since I saw it).
The MAGAs are winning so hard that my own spleen is cracking under the sheer weight of my mocking laughter. So Much Winning, Such (f)Art Of The Deal, Trillions and Trillions and Quintillions and Gazillions of Winning / lost stock value.
My lungs are tired boss, there is not enough air in the Earth to laugh about that amount and quality of stupidity.
China cancelling its orders is not a problem for Boeing.
On the other hand, the tariffs are an enormous problem for Boeing to actually be able to build airplanes. And possibly a big problem for them to profitably sell airplanes, since they're contracts may not have clauses for this kind of increased input costs, and the sale prices are already negotiated. And longer term, a big problem for them to invest in R+D for their next aircraft.
According to Google AI Overview when I search “Boeing airplane backlog”, Boeing has a backlog of 5,000+ orders which translates to 17.7 years of production.
Boeing will easily be able to reallocate the orders to non-China Airlines.
Ok that made me go look up BA numbers. Don't know where Google AI got its numbers, but BA does have a big backlog (the most recent $ number I can easily find is $550B; the revenue in 2023 was 78B, so that's about seven years worth of production, still pretty darn good) so they should be able to reallocate the planes to others easily.
Other American manufacturers selling to China may not be so lucky.
1) Google AI Search is making up things due to confusing numbers from different sources. This is likely the issue.
2) Google AI is prioritizing a source that is including “orders” that are not fully in contract. Boeing’s 777x has been delayed for years. There is likely many “pre-orders” (word for the day because I missed out on the Nintendo Switch 2) and my speculation is that these should not be treated as real orders.
Over tariffs, or over "oops Boeings are falling out of the sky"?
A little unfair to Boeing, but they have had some little problems with their aircraft recently. So I do wonder if that will impact demand for their planes, on the other hand, airlines do seem to want more planes so they should be able to sell them elsewhere.
The timing strongly suggests tariffs, and that's what China has said. If it was over planes falling out of the sky, they would have canceled the orders when the planes fell out of the sky, not when Trump imposed massive tariffs.
They could. But they ordered the planes knowing about Boeing problems, and/or didn't cancel the orders despite them. Only now. You can say Trump gave them an excuse, but then what does it tell us about our Great Dealmaker Making America Guilty Again?
Building jet planes is one of those things where a little knowledge is a dangerous thing. I don’t think people who are really involved in that industry feel that Boeing is a long-term dangerous proposition. The two Boeing 737 Max airplanes that crashed is a much more complicated story than what one gets from the media coverage of it. They definitely dodged a bullet with that door that blew off the Alaska Airlines plane. That could’ve been really ugly.. But that was very much a freak accident.
Yeah, I have no doubt the story is incredibly complex, simply because planes are incredibly complex systems. I think this is kind of my point, China ordered the planes knowing about all these problems, and I'd be shocked if the people responsible for orders didn't go into a deep dive on the causes and remedies, 8D kind of stuff.
That's why I have pretty high confidence that when they say it's the tariffs, it is the tariffs.
Or China could be cutting its nose to spite their face.
China has ordered the airlines to also stop buying aircraft parts. Source below. I do not know how much inventory China’s airlines have, but aircraft repairs are tightly regulated and parts must be sourced from Boeing. If repairs are not documented properly, the aircraft is not allowed to fly to most countries in the world.
I have had something of a terrible revelation recently, the culmination made earlier, things Scott has said, and talking to people in real life.
I have kind of always assumed that my political opponents (conservatives) we're wrong on some points of fact, and were generally worse then the people I choose to align with, liberals in this case.
I don't believe everything that the liberals believe, but they had the best total package.
That said, when conservatives got elected I thought that they had a theory of the world that they were operating off of to make decisions, and even if the theory wasn't 100% correct, they were at least playing the same game I was so to speak.
I don't think that's true anymore, and I'm not sure it was ever true. I think it might have just been luck all this time, that there was a left-wing of idealists, a liberal center of technocrats, and a right wing that was totally disconnected from reality at every level and only got things right because the people that they happen to listen to were economists and business people that need to know things about stuff in order to live.
All this coming from going to an Easter gathering and speaking to my conservative friends and watching as they came into a conversation believing one thing was true, them learning that the party has said that the other thing was true, and then all deciding that it had always been that way and never been the first way to begin with. It was shocking to me; these are the business owning salt of the earth conservatives that I assumed held their values authentically and not tribally, but even there, there it was.
That said I've made an absolutely wild amount of money betting that this administration would do the dumbest thing possible at every turn, I'm not kidding it's actually crazy, and I have another strong passport if worse comes to worst.
I could list off a dozen things that liberals and leftists believe that are totally disconnected from reality, but what would be the point? You would just acknowledge that OF COURSE you don't believe THOSE things, but they (liberals) have the best package overall, unlike those EVIL, IRREDEEMABLE right wingers who are all an undifferentiated mass of sheeple.
> You would just acknowledge that OF COURSE you don't believe THOSE things
It is easy for the liberals and leftists to acknowledge that publicly *now*. But a year ago, doing that would get them called Nazis and probably would cost them a job.
Ironically, they should all thank Trump for their newly gained freedom of speech. These days they can disagree with the woke dogma as much as they want, and no one will notice, because everyone is looking at Trump instead.
(For the record, I believe that Trump is a horrible person and a bad leader who will ruin the country. Yet even such things sometimes have a small positive side effect.)
For me the key moments were 2016 (minor) and 2020 (major). Watching the spread of COVID play out and seeing a sizeable fraction of U.S. conservatives--and a smaller but real chunk of Canadian ones--just completely break from reality on a bunch of key points was chilling. That was the point I ultimately gave up on the U.S. as a nation with a future. The tribes are no longer plugged into the same reality: they have almost entirely lost the ability to come together on issues of mutual interest. The structural flaws of the U.S. electoral system will keep the issues from being fixed and keep the sides playing tug-of-war until the wheels fall off in some big, dramatic, unrecoverable way. I dearly hope I'm wrong, but I don't think I am.
FWIW, I think what you're seeing is a recent development, albeit one long in the making. I wasn't around to observe directly, but I don't get the sense that the conservatives of 50 years ago were like this. I think the trend was already clearly identifiable 30 years ago, but that the internet and eventually social media supercharged it in the past couple decades. To be honest, I'm genuinely puzzled at why the same trends have had a seemingly much smaller impact on the U.S. left: I've definitely *seen* the same sort on insanity manifest among left-leaning people, but they haven't ever formed the same sort of unified bloc, and haven't ever come remotely close to taking control of the Democrats the way MAGA has hijacked the Republicans.
I wish I could say I found that part a surprise, but watching most of the "Never Trump" people turn around and go "well, I guess Trump is better than Hillary" in 2016 sent the message pretty well. As long as right wing media can drum up enough fear and resentment at the Democrats, which it is very good at doing[1], there is effectively no line whatsoever for many Republicans. I was rather more surprised that there *was* some initial pushback than that the party settled into a pattern of dismissing and minimizing it.
I was more unpleasantly surprised by how many people who were *not* staunch Republicans were apparently able to forgive and forget that episode by 2024 (and how ineffective it was at driving turnout on the left). Watching from the (comparative) safety of another country, I've basically thrown up my hands and conceded that the U.S. got exactly the government it deserves. But of course there are lots of individual U.S. residents who neither wanted or deserved it--I wish I had more power to get people out who want out.
[1] To be fair, the Democrats do make it far too easy sometimes.
Yeah, show me a Republican with a spine and I’ll show you Adam Kinzinger or Liz Cheney. Or for that matter Mike Pence who notably didn’t vote Trump in 2024. But Dems liked that so you have to change the signs from plus to minus when you update priors was I believe how you put it. Huh? That really doesn’t make any sense.
Congressional Republicans are Trump toadies save perhaps Susan Collins and Lisa Murkowski who is on record as saying everyone is afraid of Trump’s goonish mafia capo retaliation. It’s like the nation is being run by the Corleone family except after Don Vito’s death his replacement was Pauly Walnuts instead of Michael. Cruel and vindictive but not so bright.
How many Democrats can you name that consistently stood up to Biden's excesses? Manchin and Sinema? Sounds like a pretty similar list, with pretty similar consequences for the defectors.
May I suggest that Congress has, in both parties, sublimated its responsibility to the head of the party in a way that is bad for the country?
I think these aren't even close to being equivalent. Did Biden endorse alternative legislators in the primaries for people who didn't toe the line? Did call them out and insult them on social media?
> I don't think that's true anymore, and I'm not sure it was ever true. I think it might have just been luck all this time, that there was a left-wing of idealists, a liberal center of technocrats, and a right wing that was totally disconnected from reality at every level and only got things right because the people that they happen to listen to were economists and business people that need to know things about stuff in order to live.
How exactly do you know your not the out of touch one? The real fundamentals come from factories, truck drivers, farmers. Would you listen if anyone doing low status real work said "I no longer care if trump breaks the system"?
Farmer protests are usually right wing, and assuming your not so far gone maybe you can remember: the food you eat comes form farms. When was the last time you talked to a technocrat who knew a farmer?
I can't speak for the OP, but there are a lot of people out there who disagree with me from a lot of different directions.
I worry a lot about the possibility that I may be wrong, but for many of those people I don't much worry about the possibility that they may be right - there are lots of questions where working out the correct answer is hard, but it definitely isn't seven.
The center-right, and even some of the more libertarian branches of the relatively far right, do have people whom I worry may be right about things. But I put the MAGA right in the same category as, say, Islamic extremists, as people who can safely be dismissed.
Trump is a boring moderate in reality I wish the trump of the memes was an option. And you cant dismiss islamic extremists they are rapidly gaining ground and will probably conquer europe, you may find your moral high ground ineffective without any ammo.
> there are lots of questions where working out the correct answer is hard
Theres enough questions where the answer is very simple but politically impossible. Have children, not an immigrant ponzi scheme; make real work high status, even if you need to blow up the stock market; own guns; respect farmers; invade mexico to kill the cartels; take any jews worth having out of israel before either genocide happens; stop welfare to undocumented immigrants; dont allow rape gangs, death penalty; dont lets europe censor the american internet, harress those germans whining; purge the prison system of gangs, death for cartel ties; collapse housing prices.
I dont care any more, if it takes a Caeser to unravel the Gordian Knot, let trump be the weakest and timid one imaginable
I had the mirror-image of this experience some years ago. After many years of agreement (on a very liberal forum) that Bush's war on terror policies were awful and inhumane, Obama came into power, continued most of them (with more murder and less torture), and rather abruptly, a bunch of the people I thought were my allies on these issues realized that the president had a hard job, he had access to intelligence we couldn't see, and in general sheltering war criminals and assassinating US citizens were forgiveable missteps in their guy.
Most people simply do not have strong principles in most areas, but almost everyone has a side.
I find this striking because almost all of the left-of-center people I know were highly critical of those aspects of the Obama administration, from the moment it started getting reported on. I'd imagine it's a question of moving in different circles.
A lot of people I interact with are belligerent enough to loudly talk about voting third party out of anger when Democrats do things like this. Certainly I saw no shortage of people insisting that voting for Biden/Harris was unconscionable because of their support for Israel, for example. I think this is quite a silly stance to take most of the time--and AFAICT the people who take it are more loud than numerous even in far-left spaces--but I think the constant feelings of dissatisfaction with Democrats don't do great things for turnout, especially among younger voters. The U.S. tends to make voting hard enough as it is: people are that much less likely to jump through the hoops when they don't even like the person they'd be voting for.
It's virtually everyone, and given that you've framed this as, "I'm Team Blue, can you believe Team Red?", it's you, too.
Sorry if that reads as excessively confrontational, but...I think my perspective is worth considering:
I like to joke that I'm "politically nonbinary," but to be more specific, my position is more like, "classical liberalism, but with the recognition that our version of late-stage capitalism has vast and scary problems, but also, absent a post-scarcity technological innovation like a Star Trek TNG-style food and object replication, we don't have any better options."
That position means I am No Team. And because I am No Team, I actually see and remember instances of retarded* teamism in which Team-folk "forget" stuff in order to maintain social cohesion.
For example, remember when Red Team and Blue Team *switched* *sides* on the COVID vaccine? Remember the earliest days of Operation Warp Speed, when Red Team was marveling at this amazing tech, but Blue Team members were skeptical of Big Pharma and publicly declaring that they wouldn't take "Trump's Vaccine?"
Remember when the Team Leader changed, everyone immediately switched sides on that vaccine tech, not because of new evidence, but to be in support or opposition of the Current Team Leader *carrying on the exact same plan as the Former Team Leader*?
No Team remembers.
No Team remembers, and then trembles.
And, look, I'm not taking credit for my inability to Team as some kind of enlightened intellectual achievement.
My dad's side of the family has a history of psychopathy, and I suspect that my innate process of selective empathy - I can only feel empathy for people I've judged worth of empathy (that's a very, very large group, larger than 'normal' people's group, but still) - is the very mildest possible expression of that family psychopathy.
Like many mild aberrations of personality disorders or mental illnesses, a tinge of psychopathy can be productive and useful, because sometimes empathy-governing behavior is very, very bad (you want your surgeon to be a little bit of a psychopath so they don't weep into your open wound, etc.).
But it's not great in other areas, like social cohesion. Unlike regular empathy-people, I can't be persuaded or make myself feel a certain way about a topic merely *BECAUSE* other people feel that way. I actually have a hard time even modeling the pleasure people get out of Teamism, much less the
not-noticing/forgetting/whatever-it-is-ing that allows them to blissfully change positions on an issue without ever realizing that's what they're doing.
So yeah, I am exquisitely familiar with the kind of collective doll-eyed blankness you describe witnessing amongst the conservatives at your Easter gathering when they were confronted with information that contradicted their social cohesion.
And I'm telling you: It's everyone.
* I know this word is still semi-forbidden, but no other word ever replaced the precise meaning of the slang term it eventually evolved into, and, given that its usage to formally and/or politely describe developmentally disabled folk is completely gone, I think we can reasonably reclaim the slang definition.
(Oh, does the sensitivity to word usage *and* advocacy for freedom of expression seem to conflict? That happens when there's no formal ideology to adhere to!)
*Fear may fuel racism and xenophobia that threaten human rights*
While addressing the outbreak will take a global public health effort, the U.S. Centers for Disease Control and Prevention has declared the current risk to the American public is LOW [https://web.archive.org/web/20200130203859/https://www.cdc.gov/coronavirus/2019-ncov/summary.html]. If 21st-century outbreaks like SARS, MERS and Ebola virus are any indication, it is likely American fear of contracting coronavirus — and the xenophobic, racist assumptions that drive it — carries a risk far greater to most people in this country than the virus itself. Historically, infectious disease has generated racist discourse...
How did both major political tribes decide, within a month of the virus becoming widely known in the States, not only exactly what their position should be but what insults they should call the other tribe for not agreeing with their position?
...
IS IT JUST RANDOM?
A couple of Republicans were coincidentally the first people to support a quarantine, so other Republicans felt they had to stand by them, and then Democrats felt they had to oppose it, and then that spread to wider and wider circles? And if by chance a Democrats had proposed quarantine before a Republican, the situation would have reversed itself? Could be.
Much more interesting is the theory that the fear of disease is the root of all conservativism...
...
The proposition “a quarantine is the best way to deal with Ebola” seems to fit much better into the Red narrative than the Blue Narrative. It’s about foreigners being scary and dangerous, and a strong coordinated response being necessary to protect right-thinking Americans from them. When people like NBC and the New Yorker accuse quarantine opponents of being “racist”, that just makes the pieces fit in all the better.
The proposition “a quarantine is a bad way to deal with Ebola” seems to fit much better into the Blue narrative than the Red. It’s about extremely poor black foreigners dying, and white Americans rushing to throw them overboard to protect themselves out of ignorance of the science (which says Ebola can’t spread much in the First World), bigotry, xenophobia, and fear. The real solution is a coordinated response by lots of government agencies working in tandem with NGOs and local activists.
It would be really hard to switch these two positions around."
> For example, remember when Red Team and Blue Team *switched* *sides* on the COVID vaccine?
Rather than switch, it was a complicated dance. For example, there was a moment near the beginning, when it was known that only a racist would fear a virus coming from China.
> The virus may be novel, but you really don’t need to buy anything new or special to brace for it. The Washington Post spoke to epidemiology experts, and they said the most important aspect of preparedness costs nothing at all — calm.
> “Seriously people- STOP BUYING MASKS!” Jerome M. Adams tweeted. “They are NOT effective in preventing general public from catching #Coronavirus, but if healthcare providers can’t get them to care for sick patients, it puts them and our communities at risk!”
This comes shortly after several paragraphs on how "avoid touching your face" is the number one thing you can possibly do. (Might wearing a mask help you avoid touching your nose and mouth? No, shut up.)
I'm sort of like you. I don't seem to be very subject to group identification and group loyalty. I have only voted once in my life. After several decades of watching from the sidelines as political beliefs play out on the large scale, in elections etc., and on a small scale, in discussions, here is my take:
Most people acquire their politics not through thought and study but via osmosis from their family and peer group.
Most opinions and votes are determined by vibes and group identification, including, probably, yours.
Any time you find yourself thinking that everyone with a certain characteristic (who they voted for, ethnicity, SES) is dumb or evil or smart or ethical or deserving of help, you can be sure your head is up your ass.
There are aspects of how you describe the democrat opinions on the vaccine that I think are misleading. Specifically:
"Remember the earliest days of Operation Warp Speed, when Red Team was marveling at this amazing tech, but Blue Team members were skeptical of Big Pharma and publicly declaring that they wouldn't take "Trump's Vaccine?"
Remember when the Team Leader changed, everyone immediately switched sides on that vaccine tech, not because of new evidence, but to be in support or opposition of the Current Team Leader *carrying on the exact same plan as the Former Team Leader*?""
I think that, while it contains some logical reasoning, you skip quite a few steps or make some assumptions that are not properly supported. The statement assumes the only reason for changing positions was team loyalty ("to be in support or opposition of the Current Team Leader"), without considering other factors that might have influenced these shifts. Similarly, you imply the context before and after the administration change was identical ("carrying on the exact same plan"), when the discussion/concerns morphed. Initial Democratic skepticism focused specifically on concerns about political pressure potentially rushing the approval process. You mention this in the form that democrats said they wouldn't take his vaccine - however, the more common position was that there were concerns and fears about it being rushed or not specifically based on Trump's approval. For example, Harris said that she would not take the vaccine based solely on Trump's word, but rather requires more credible sources of evidence and authority, citing Fauci as one such source. (Sep 5, 2020) Trump's platform based on Operation Warpspeed's chief advisor Moncef Slaoui, as well as his press secretary at the time AND Fauci (NAIAD) actually tried to respond to these concerns, saying that it would not be rushed out until it was safe and tested. Biden later changed the name on his first day in office, which I do think shows the intentional distancing from the concept that it might be rushed out. In this way, I do think that there is value in what you are saying regarding tribalism and partisanship- but oversimplified.
Likewise, I believe that another factor for Republican opposition to the vaccine stemmed less from the vaccines themselves or pure partisanship (although it is undoubtedly a factor), but rather from the aspect of the mandate, as well as other protocols such as social distancing, closing businesses or requiring masks.
I think electoral politics exacerbates the controsery, particularly given the timeline around the covid vaccine development aligning with the election. If Trump won the election, the response certainly would've been different, and the reaction to the vaccines as well. Although, I am hesitant to say which way it would've gone. Would they be anti-vax, or would they promote it?
As one last aside, I think that, based on your writing, you derive a sense of pleasure or satisfaction from being "politically nonbinary" or pointing out the hypocrisy, and particularly the very intentional usage of "retarded". I think that the way that you view yourself outside of the conventional boxes is in itself its own form of "teamism", and it is important to note that it shifts the boxes to "Teams" and "No Teams". I think its important to recognize ironically that you may be engaging in a form of teamism, with your team being "No Team". As you note, "[you] actually see and remember instances of retarded* teamism in which Team-folk "forget" stuff in order to maintain social cohesion." I would argue it is important to remember that we are not immune from these psychological heuristics. This aside is not meant to reinforce or discredit your overall argument, but rather to provide a reader's view based on your writing.
> "I think that, while it contains some logical reasoning, you skip quite a few steps or make some assumptions that are not properly supported. The statement assumes the only reason for changing positions was team loyalty ("to be in support or opposition of the Current Team Leader"), without considering other factors that might have influenced these shifts. Similarly, you imply the context before and after the administration change was identical ("carrying on the exact same plan"), when the discussion/concerns morphed."
Circling back to this:
First, I'm sure we're both on the same page that it's impossible for either of us to provide definitive evidence about the internal processes of tens/hundreds of millions of people. While my original comment was written in declarative statements, to match some of the tone of the original commenter, I'll be the acknowledge that discussions about human nature in general and the zeitgeist are ultimately speculative and personal opinion.
I don't disagree with you that a small-ish, dispassionate, rationally-minded portion of the population (very likely disproportionately represented amongst the commenters of an intellectual's blog!) closely followed and assessed the development of the vaccines and made their judgments accordingly without being unduly influenced by which Team was at the helm.
Was that the process that *most* people were engaging in, though?
Casually observing it: I don't think so!
My general reading of the zeitgeist (news media, etc) is that majority of both Red Team and Blue Team got distracted by their loathing of the other team's leader and eventually literally forgot that Trump was all-in on mRNA vaccines, and also that he was instrumental in launching them as a national effort. It's hard to work the topic into a conversation naturally, but it can be illuminating to remind Red and Blue Team people about Trump's history with the vaccines, just to watch their reactions.
And then ask them to imagine what would have happened if Trump had won the 2020 election with his administration in charge of administering the rest of the pandemic.
In my experience, the more Team a person is, the less plausibly they describe a hypothetical 2020 Trump Administration response to the pandemic. For example, deeply Red Team folk will claim that, oh, yeah, a Trump administration would NEVER have pushed for vaccine mandates and would have shut down the mRNA vaccines the moment it looked like they weren't that awesome after all. It wouldn't have cared at all about Big Pharma's interests. It would have been a paradise!
And of course, we can't know for sure what would have happened in that alternative timeline, but common sense says that the claim that Red Team would have turned the mRNA ship around - the ship they launched! - is pretty absurd. It's unknowable, but I'm pretty confident it would have actually been the opposite; mandates would have been more draconian, not less, and Red Team governors would have been more likely to cooperate with a Red President's leadership than (perhaps) make a name for themselves and boost their celebrity by standing in stark opposition.
For fun, it's also illustrative to ask Team people to speculate why only 14% of children received COVID boosters in the 2023-2024 season, and what they think the demographic breakdown of the parents of the 86% might be.
Though you shouldn't ask these kinds of questions unless you're okay with people not really liking you very much.
> "I would argue it is important to remember that we are not immune from these psychological heuristics."
This isn't necessarily true, though. Some people actually are functionally immune (or at least vastly less vulnerable) to some psychological heuristics simply because they have conditions which make their brains work outside of the norm said heuristics may be governing. There are people who are way, way further down the psychopathy spectrum than I am who are *legitimately* immune from certain kinds of internal psychological phenomena around social pressure, influence, and group ideology. Same thing with some high-functioning autistic people. There's pretty much always a non-zero number of people in experiments on group think / social pressure / conformity who will say, "no, those two lines are exactly the same length, the rest of you folk are nuts" because they sincerely don't care about (and thus can't be unduly influenced by) the group.
As for the aside on my writing, yes, I actually spent time debating capitalizing "No Team," as I knew I would inevitably get at least one comment saying, "Ah ha, 'No Team'" is a Team, you even gave that Team a proper noun; you're not special or different!"
But I decided to do it anyway in order to most effectively pull off the "Pepperidge Farm remembers" reference, a choice I made because I cared more about maximally amusing myself than I did people potentially having less respect for my comment and/or me because my argument was weaker than it could have been.
Which, I think you'll agree, is a lack of care about how I am perceived which is outside of the "norm."
And last, while there is a certain amount of pleasure in not being taken in by team-based manipulations and scams, there are many disadvantages to being incapable of ideological conformity to a Team, including that most people are Team people, and Team people REALLY hate being disturbed when they're Teaming (my otherwise supremely compatible ex in my best relationship broke up with me because, in his words, he was very, very Blue and, "I don't like how I feel when I agree with you.").
I appreciate your response - although with the caveat that it seems reasonable (although I do not say this with a sense of certainty or any data based evidence) that even neurodivergent people who are less susceptible to common heuristics or specific (e..g. group based) cognitive biases, they may also have greater vulnerability to other different heuristics/cognitive biases. I am not claiming this for all neurodivergent people universally, but rather on a much more specific individual level.
For example, perhaps someone who experiences a sense of strong displacement from their community due to psychological and neurological factors would develop a different set (or perhaps weights) of heuristics in social reasoning than those who feel integrated into their communities. These alternative/distinguished heuristics might help them navigate their unique circumstances but could potentially introduce different biases. Along the same lines, I think this probably follows along the same lines of the contextual advantages and disadvantages that you are talking about. Thanks for your response - interesting stuff
Oh, for sure. If I wasn't clear about this, I want to acknowledge that deviating from the psychological/social norm isn't the path to living a very good life, even if that deviation has some fringe benefits. Normal people experience pleasure and a sense of security in Teamism that I can't access, myself. And, perhaps more importantly, they are able to naturally express their pleasure to the other members of their Team in a way that is intensely bonding (thus socially useful).
I'm not psychopathic enough to want to manipulate people into liking me a lot more by *faking* a mutual enthusiasm for things like my city's football team/ my workplace/ a political party / my gender and/or ethnicity / joining a church, and so on. But even if I only neutrally refrain from enthusiasm (rather than criticizing the enthusiasm, as I often did in my younger days), it still has a chilling effect, which ultimately isn't good for me.
If I were more rational and less egotistical, I'd fake the Team enthusiasm and enjoy the fruits of that labor, but I'm not rational enough to make myself do it.
One thing you didn't really mention is that there was also new evidence that caused people to change their minds. In particular, the study results came out and the vaccine was vastly more effective than anyone expected. And then once people started getting the vaccine, we had dramatic real world data showing its effectiveness too. I can't speak for everyone, but personally, I changed my beliefs based on new evidence, not which guy was in the white house.
Putting COVID aside, It's also worth pointing out that when it comes to the MMR insanity, this is probably less a matter of people changing their opinions and rather people changing their party. Anti-vaxers used to be a *left-wing* phenomenon. RFKJ himself used to be a Democrat! What changed was that Trump attracted all the crazies to his side and repelled normal people.
I agree with you wholeheartedly. I wanted to focus on the tribalism aspect, but this is as important if not more. Categories were made for the man, not man for the categories
I'm going to drop this here because I think a lot of rationalist/EA people have good ideas but don't sell them well. Guy talks about seduction and I think he has a lot of good points.
On the one hand, yes, some basic sales knowledge is very valuable. Yes, it's very annoying how people will reject obviously beneficial things for weird internal reasons. Sneakers are a bad look, chelsea boots are a good look, just buy the boots and get back to Getting Stuff Done.
On the other hand:
"You call yourself effective because you misunderstand what altruism is. You think it is a mistake that in practice altruism rarely lives up to its nominal charge when, in fact, altruism is a term one uses when one wants to seduce others into doing the thing they wanted them to do anyway, give them prestige and money"
Yeah, but the effective altruists have pretty clearly proven they're better at saving real human lives than 99% of charities out there. That's the thing that matters. A lot of the "marketing" and "discourse" are effective altruism can be...suboptimal but their results speak for themselves. And, ultimately, that's all that matters.
But don't try to turn your engineers into great salespeople; you just end up with bad engineers and mediocre salespeople. No one gets to be good at everything. If your product is great, by all means make sure your engineers wear clean shirts, but you fundamentally prioritize good product. If your marketing is great, by all means, make sure your product isn't crap, but keep fundamentally prioritizing good marketing.
The aether; I believe it was ny times article (boring as shit) and fox news boomer take was "swearing is wont make you electable"; I have yet to see someones take that properly takes into account my anti-politeness position.
For it to be real I want some memes and a new aesthetic
This might be too vague to answer, but here goes. Is Stuart Russell's book Human Compatible worth reading? I noticed it's cheap on Kindle today, but also it came out in 2019 and a lot's changed in AI in that time. Any thoughts?
"A progressive district attorney has declined to charge the state employee in Tim Walz-led Minnesota who was allegedly caught causing $20,000 damage by vandalizing half a dozen Teslas — a decision the local police chief ripped as the latest betrayal of victims."
>>Despite what police believe to be evidence of Adams committing felonies, Hennepin County District Attorney Mary Moriarty will seek diversion rather than criminal charges.
>>“This is an approach taken in many property crime cases and helps to ensure the individual keeps their job and can pay restitution, as well as reducing the likelihood of repeat offenses,” the attorney’s office said in a statement, CBS News reported.
It's a pretty important clarification that "declined to charge" in this case, means sending the vandal in question into a diversion program involving regular check-ins, community service, and restitution payments, under which failure to comply with the program means they can be "terminated from diversion and returned to court."
I'm okay with diversion, including restitution, given the circumstances, but if this person were too poor to pay back $20,000, does that mean diversion wouldn't be an option?
But I'm explicitly and directly asking about the hypothetical where it's a different person who can't pay $20,000. Does the poor person not get diversion as an option?
Okay I still don’t really know but I’ve thought a bit about it and it’s likely a poor person with no prior criminal record who couldn’t do full restitution would still be put in some form of diversion program. It would probably look something like what Nobody Special came up with.
Minnesota does lean liberal and it seems unlikely they would lay a felony conviction on someone who didn’t have the cash for full restitution. But it’s really just a guess.
I have no particular expertise, but I'd expect that falls into the discretion of the County Attorney, who could work out alternatives like an extended or reduced payment plan, community service, or other forms of in-kind restitution.
If he makes full restitution and jumps through every hoop of the diversion program he will avoid charges. If not he will be prosecuted. This is SOP for first offense property crime at the Hennepin County DA’s office at this time.
He is on the hook for full restitution and the rest.
The best penalty for destructive political acts is public humiliation. Make them publicly apologize and pay the damages. Locking them up just makes them scream about political oppression and look more martyr-y, meaning you'll have to lock up more of their friends too.
Same answer I gave Monkyyy - depends on the vandalism. If it's something like arson where there's an actual danger to others posed, that's a different kettle of fish entirely and diversion is inappropriate IMO.
But assuming we're trying to stay apples to apples with the Minneapolis case, and thus still talking about keying, if somebody keyed the little teacup seats on the Mad Tea Party ride in the name of MAGA I really don't much care whether they get formally charged or put into a diversion program and forced to pay restitution on pain of prosecution later if they screw it up.
In all honesty, if all they did was key the teacup ride and they're a first time offender I'm talking myself into the idea that diversion should just generally be preferable, regardless of political motive in any particular direction. I may not have a lot of sympathy for MAGA, but keying a ride at Disney would be a pretty silly-ass reason to saddle someone with a criminal record if you have an equally-workable option to just make them pay for the damages and reform themselves to not be the kind of moron who keys other people's stuff.
ISTM we want this guy treated the way vandals who get caught are treated in other non-political cases. I don't think vandalizing Teslas for some political cause ought to get extra punishment for the same reason I do not support hate-crime enhancements for vandalism--if Alice spray paints some obscene words on a wall and Bob spray paints a swastika on a wall, they ought to get the same sentence.
Differently, but then fire has the capacity to rage out of control and kill people, so it's pretty distinguishable from keying cars.
I'd also, for example, feel differently about diverting someone for $20,000 of fire damage to a Tesla dealership, and be more open to diversion for someone who did a similar amount of damage by keying the cars of one or the other side of protestor/counterprotestors out front.
Have you ever paid for body work? Or done it yourself? It's absolutely ridiculous. To properly paint something, even a relatively small scratch, in a way that isn't glaringly obvious requires many days of prep and technique.
Beats me - the video imbed & photos in the NYP article just show the guy keying cars. No idea how you get to $20k with that. Maybe some combination of repair shop markups & there being more cars involved? The article says "at least six" rather than just "six," so maybe there are other keyed cars in Minny for which they don't have the video but with which they planned to charge him?
How does this work? Do the people who make these decisions need to make some written statement explaining why? Can we find those statements online? If yes, it would be nice if journalists linked them. This would generally improve the quality of discussion.
Here’s a statement from the Hennepin County DA’s office. - St Paul Pioneer Press
“We want to make sure we are very clear. What Mr. Adams did was wrong and we are holding him accountable for keying the cars,” said Daniel Borgertpoepping, spokesperson for the Hennepin County Attorney’s Office. “The HCAO did not reject or decline this case. We offered diversion as we often do with property damage cases when the person has no record. Mr. Adams will have to complete the requirements of the program. He will also have to pay every penny in restitution to the victims. If he does not meet those requirements, we will proceed through the criminal legal system process.”
Probably paywalled - but you might get a couple of free views
> “We want to make sure we are very clear. What Mr. Adams did was wrong and we are holding him accountable for keying the cars."
> "Mr. Adams will have to complete the requirements of the program. He will also have to pay every penny in restitution to the victims. If he does not meet those requirements, we will proceed through the criminal legal system process.”
He's going to pay whatever it takes to undo the damage he caused, cover case fees, plus a bit more on top, and also do some kind of community service; no lighter a punishment than anyone else committing property damage with no prior record. If he can't or won't go through with all that, it's off to jail via the court.
Sounds pretty reasonable to me.
Are you suggesting that singling this guy in particular out for something much harsher than what is usual for vandalism would be just, or did you miss the part where he was being punished, or something else?
Your arguments continue to ignore the political part and treat this person like a 17 year old that decided to key cars randomly. It is almost like you are ignoring the facts of this case.
I am not sure what is the right penalty but I think he should have been charged with some crime. That would signal to the community that political vandalism is not tolerated.
This “punishment” is so light that it sends the opposite signal. Vandalize the right political opponents’ property and we will protect you. This is tribalism at its worst.
You are free to make your arguments but do not be surprise that MAGA people decide to now vandalize leftist property. I hope you will still be consistent by actively arguing for diversion in those cases too.
In general, what punishment do you think is appropriate for minor vandalism of other peoples' cars? I don't see it being a great use of public resources to lock such people up, so fines and community service seem like a good option, with the county jail as an alternative if they reoffend or don't pay their fines/do their community service.
My opinions on courts systems has near zero effect on reality; but I consider the last 100 years to be theft. The state is claiming ownership of my revenge, an economic asset, and grossly mismanaging it.
Payments need to go to victims and we need 100 years of data and negotiation to catch up on. If you punch me I'm owed a punch, maybe I hire a kick boxer to look scary and ask you prehaphs we could set up a payment plan.
> community service seem like a good option
Maybe compared to rape cages, but your lacking imagination if you think bubbling efforts in a make-work "job" is making the world better.
Does anyone know where I can find the stated views of Mahmoud Khalil (The Columbia University student accused of terrorism)? Matthew Yglesias mentioned his views are "more radical than most liberals realize" in a post (https://www.slowboring.com/p/mailbag-remembering-kevin-drum) but the link to the AP news article he references is now dead.
Everything I can find online seems to basically say; "As a Palestinian student, I believe that the liberation of the Palestinian people and the Jewish people are intertwined and go hand by hand, and you cannot achieve one without the other" which seems a fair and nuanced take. While the Trump administration is clearly willing to be unreasonable, this can't be all that was said regarding Israel and Palestine.
I don't think this has much bearing on the legality of his attempted deportation (the same with Kilmar Garcia and his alleged MS13 affiliation) since due process is being ignored, but I do think it has some bearing on the level of moral outrage demanded, and how generalizable the threat of Trump's deportations are.
If there are edge cases in his policy where people who are actually gang members, but aren't proven so in court, are deported without due process, and people calling for the genocide of Jews are being deported for their views, that changes my estimate on how likely Trump's policies are going to spill out into a more general policy of authoritarianism. I'm of course not saying either Garcia or Khalil are a gang member or explicit antisemite, but I'm curious in the case of Khalil if there's a take on his previously stated views that I assume Yglesias was referencing which seems to have been removed.
I tried to look, my best guess is the phase "columbia united apartheid divest" is the best results
Im seeing the term "Zionism" used in formal documents, and in their list of demands includes "[ending] Dual Degree Program with Tel Aviv University", Id look for the worse treated "dual student from telaviv"(whatever the hell that is) and see what they have to say
An intern at my company was going to school at Columbia when that happened, and their demands seemed pretty reasonable when I read about them at the time. Albeit I've read the divest movements don't actually achieve anything (apparently the divest from South Africa movement didn't have any meaningful effect on the S.A. market).
Can that be what Yglesias was talking about? It doesn't seem particularly radical to me, but that could be that I just differ in opinion.
Remember that ivy league jewish students fled into a libery to avoid one of these protests and they were chased and that video made the rounds in the right wing. That easily could've been Colombia.
Fundamentally this would be trump selectively preferring one middle east religion over another.
When fox news types use the word terrorists about anti isreal protests, they have videos of harassment of jewish students. Much like jan 6, cnn had 1 side of the building breaking windows.
Can anyone explain to me what it means to run multiple instances of an AI model? For example, when talking about how a company could have a whole team of AI 'employees' supervised by a manager (human or AI), what does that mean in concrete terms?
I think that, right now, each model consists of a whole bunch of numeric weights. This program can be run on as much compute as is available, and there will be separate copies of the program on different chips/groups of chips. Each copy of the program will receive inputs, serially, one after the other, and produce an output from them, independently of what all the other copies of the program are doing. Then again, maybe this is only the case for open-source models which different people have downloaded onto unconnected computers? Maybe, for other models which are run centrally by the company, there is only one copy of the program, and it is running multiple different prompt-convert-output sequences at the same time/in parallel? If so, how?
I think this question is the primary reason for my confusion about agentic behaviour. Will a particular model always have the same goals, or might multiple copies of it have independent goals? Should I expect models to collaborate to work towards long-term goals, or to possibly come into conflict? Also, should I expect a model to be able to ‘check on’ the thought processes it generates, or other copies of it generate, when answering other queries? To the extent that the goals of the AI relate to itself, will it think of itself as being only the individual copy, or as being all copies of the same model, or what?
Consider 2 identical chaos pendulums where the inputs are a combinations of where the wights are. The machine can be the same, but if you rise the weights to different heights the output becomes different.
If you encode some sort of knowledge into chaos pendulums, make duplicates, and start connecting them with strings, the encoding is still there, and youve made a larger system.
----
You probably make exact copies of the machines but with different promts you get parrell processing on the task, and then you feed output from one into the input of another every once in a while
> Should I expect models to collaborate to work towards long-term goals, or to possibly come into conflict?
If the signal is positive, it *may* have constructive inference towards the correct answer.
When an LLM answers a question or performs a task it's instructed to do (by outputting text that triggers some actions), these are the factors that influence the result:
1) the model itself (a bunch of weights)
2) parameters such as e.g. temperature (how "creative" the model should be)
3a) the system prompt instructing the model how to behave in general terms
3b) the context window: additional files provided to the model, like source files; previous prompts and replies of the model within the same session.
3c) the actual current prompt the model's to act upon.
When people talk about "different AI employees" or agents they sometimes mean different 1) and 2): e.g. use a reasoning model for some tasks but not others; use Claude where it works best and ChatGPT for other things; fine-tune a model for a specific use; etc. etc. But I think that's not what you're asking about. Suppose for simplicity we fix 1) and 2) and use the same model & parameters for all our AI work; does it then still make sense to talk about "different AI employees"? Yes, because the input to the model is all of 3a), 3b) and 3c) chained together. The model doesn't "remember" anything from one invocation to another, and it doesn't matter on how many chips it runs with how many parallel copies. All the "working memory" is inside 3a)+3b)+3c), but that can be a lot, especially for newer models with large context window sizes.
So with programming work, you can have one "agent" be responsible for fixing small bugs in the codebase, another for reviewing such suggested fixes, a third for deciding which of the small bugs in a bug tracker to send to the first one, etc. You might use the system prompt 3a) to set their "goals" and differentiate their behavior (ask the agent that's generating tests to be very pedantic, for example); you'd use 3b) to give different agents access to different data (source files, bug tracker dump, emails...), and the buildup of prompts/replies within a session may (or may not) be useful to further specialize their behavior to each agent's intended use. Meanwhile, all of these agents might well run on the same model and even on the same group of GPU chips, if that model is fast enough for the purpose, because what differentiates the 'agents' is 3a)-3c). Anthropic has a recent blogpost "Claude Code: Best practices for agentic coding" with helpful suggestions including for multiple agents.
But this is now. As the market grows and tools mature and become less expensive, you might be seeing more of separate agents each running on a different fine-tuned model (i.e. a different set of weights, although obtained quickly and inexpensively from an existing model). And if (when?) we get radically new paradigms of models more powerful than LLMs that continue learning and updating their weights dynamically as they get more data and answer more queries, then the entire landscape may change and agents may come to embody different and fluidly-changing "goals" and "personalities" within their weights. I don't know when or if that will happen, though; maybe radical new breakthroughs are needed. See the recent paper "Welcome to the Era of Experience" for some ideas how that might look like.
> Each copy of the program will receive inputs, serially, one after the other, and produce an output from them, independently of what all the other copies of the program are doing.
Nope. Not even that. Each request is to a new service, it’s not waiting for the user for every chat. In fact any request could be to a new data centre if you changed your location (or VPN), and there’s no computer or process assigned either regardless.
Regarding the POSIWID, I still think of computer programs and the adage that a computer does only what you tell it do. So even if you want the program to do something useful, but end up instructing it to do an infinite loop, it will get stuck.
So what about this reformulation that jettisons the word purpose and encompasses both incompetence as well as malevolence:
Even shorter "working as implemented". That is, there isn't some malfunction in the system. All the pistons are turning exactly as they should *given how they were assembled*. Which might not match anyone's desires or intent, but that's what the collective someone told it to do, and so it will continue doing that until people do the hard thing of fixing it. And even that may be not exactly what was wanted.
Not an English native speaker, but saying a machine is "built to do" a certain task implies "purpose" to me. It's at least ambiguous, which would be even worse as a definition.
I am not a native speaker as well, but I would say that you can build something that doesn't work as intended by accident or by not considering all complexities and failure modes while having the best purpose in mind.
Gaining resources feels like a bad first-order strategy for an AI in most tasks, such that they would be trained against even in RL by lower-capability agents. Gaining resources is often competitive or zero-sum. For example, if there are lots of opportunities for an AI to make money easily today, or lots of computers an AI could hack for extra compute, those would probably have been claimed by other AIs with more resources than the AIs being trained. If nothing else, copies of an AI being trained on network security would be out-competed at money-making by AIs being trained on money-making, and vice versa. A lot of the time, compute dedicated to gaining resources would be a waste of resources, and thus poor expected-value for an AI.
Sometimes, resource-gathering would actually be the most efficient solution. In those cases, it’s complying with instructions, so it’s not fair to call it misaligned - users with AIs using those AIs to acquire resources is the intended behaviour, for better or worse. There, spending compute to get resources would be rewarded. Resource-gathering might develop as a heuristic, and it would probably be used sometimes when it wasn’t useful. But as a mis-applied heuristic rather than a goal, it wouldn’t be applied in an optimised, adversarial way - there would be no advantage in hiding it from users. Also, the AI would learn when to use it and when not to use it, and since it wouldn’t help most of the time, it wouldn’t do it most of the time.
All the examples of mis-alignment I saw cited in AI 2027 were following heuristics that were first-order effective in their training environment. “Do what the fake Oxford paper suggests”/“Use the loopholes in the prompt”/“Go to the top-right corner to get the cheese” are all easy and effective strategies that increase reward, being mis-generalised. “Gain resources” isn’t easy when it doesn’t help with the main objective, and therefore isn’t effective unless the AI learns not to do it most of the time when it isn’t helpful.
> "Copies of an AI being trained on network security would be out-competed at money-making by AIs being trained on money-making, and vice versa"
I think this only works in a sufficiently multipolar environment? If the money-making AI is a generation ahead of the network security AI (or more generally, a generation ahead of every other AI) it might be better in both fields despite only being optimised for one.
Moreover, if an AI knows that currently there aren't many other AIs competing for any given resources but it anticipates that there will be once it is no longer a one-off state-of-the-art but instead part of a regular generation of AI, that seems like an incentive for it to move quickly to grab everything it can before competition for the resources emerges?
Wow, I’m using all the language wrong. I agree with the paradigm where a model is trained once as a text predictor, then they run many instances of it. I think RL would happen with multiple sessions running at the same time. I’m pretty sure that’s how it works now, and I think your link isn’t about RL so I don’t think it disagrees. My argument above works if RL happens in parallel.
So you train many instances but then only deploy the best? Or you train many instances and then deploy them all as a cluster? The former seems much less broadly multipolar, and the latter seems like basically perfect conditions for acausal negotiation...
I mean ‘instances’ like individual ChatGPT sessions. You train the model once, then you get it to do tasks with separate chat/CoT histories. It’s not necessarily multipolar in deployment, but it’s multipolar in training, and AI 2027 worries specifically about misaligned goals baked in during training.
It might lead to acausal negotiation, but I think the AIs would tend to negotiate for help with their stated objective, rather than with goals like acquiring resources. Acquiring resources to trade in acausal negotiations has the same limitations as acquiring resources for any task unrelated to those resources - an AI designing a factory robot would have no comparative advantage in making money. Its best option is probably to sell compute to another AI working on money acquisition, in an ordinary, causal (though super-efficient) transaction.
I guess it’d be vulnerable to bribery, if it e.g. introduced backdoors into its robots in exchange for resources from a competitor (if the backdoor won’t be discovered until after training ends). That feels like straightforward loophole-finding in accomplishing a stated task. (Anything neutral for its task that it could do in exchange for resources/favours would just be good strategy, and it could do it openly.) I agree that loophole-finding is a first-order effective strategy, like hallucinating, that companies would want to RL out but which would be harder to RL out for a smarter system. I think the AI would, again, only accept bribes that help with its stated goal, rather than re-investing them in a separate goal like resource-gathering.
My pathogen update for epidemiological weeks 15-16.
1. Per Biobot, as of 12 April, national SARS2 wastewater concentrations were almost down to the levels of previous post-Omicron interwave gaps. I suspect they're there now as I write this. If patterns hold, SARS2 should maintain this low level of circulation for a few weeks before climbing again in an upward curve. Though previous summer peaks have been in late Aug/early Sept, I wouldn't be surprised to see the next wave peak in July or early Aug. But we'll see.
And if the current trends hold, COVID-19 weekly mortality rates and ED visits will drop below levels of all the previous interwave gaps. So far in 2025, COVID-19 is still holding on to 14th place as the most common cause of death, but I wonder if it won't drop off the top 15 by the end of year.
LP.8.1* and all its Clade 25A descendants have pushed aside XEC and are now the dominant clade. Worldwide, LP.8.1 is still showing some growth, but overall case numbers are falling. I don't think any of the current Clade 25A vars will kick off the next wave. 🤷♀️
2. Influenza B is still circulating. But if you look at the Biobot wastewater numbers, it's an order of magnitude below the levels that influenza A reaches.
3. HPAI incidences in poultry have been falling. Poultry cases follow a seasonal pattern, as do cases in wild birds (sorry, I can't find a recent chart for this). Mammals may not follow this pattern b/c new cases in cattle have held steady over past 3 months at 30-35 new incidences per month.
There have been no new deaths from HPAI in the US for the past 90 days. Worldwide, there have been 6 deaths from HPAI in 2025. There were 8 deaths in 2024. This is higher than the 2016-23 period, when there were only 1 or 2 HPAI deaths per year, but not as high as the preceding decade, when there were between 20 and 40 deaths per year.
Considering that Genotype D1.1 found in poultry is killing people, I suspect we've dodged the H5N1 epidemic bullet for now. Maybe next flu season! But experts have been predicting an imminent H5N1 pandemic for 2 decades now. In the meantime, SARS2 came out of nowhere and got us.
4. It's hard to tell if the US measles outbreak is growing or receding. It takes between 7 and 21 days after initial infection for measles to become contagious, so there may be a whole bunch of latent infections out there. But the current numbers in Texas, which has been hit the hardest, look like it's receding.
Per the CDC, there have been 800 cases in the US so far this year (vs. 285 cases in 2024). There have been two confirmed deaths and one provisional death from measles this year.
Meanwhile, in Canada and Mexico, cases seem to be rising. Mexico has reported 360 cases and 1 death. And as of 12 April, Canadian provincial health authorities have reported 914 cases of measles, but no deaths so far.
Regarding the universe being cyclic, that is, having entropy resets. I might be wrong that this could be the case, but I can't see why.
Two problems for it to be the case are:
1. Entropy usually increases overall
2. Practicality
A cyclic universe must be, let's call it, pendulum-like. In a pendulum-like system energy changes between 2 forms back and forth. It doesn't do this forever because a) the system it's part of a bigger system where energy can go and b) it's made up of smaller systems where energy can go. In the case of the universe a) is ruled out.
A big bounce is the mechanism by which the universe could be cyclic and so all matter needs to meet up again and again, and whether it does so or not depends on something sorta like "escape velocity".
But also distribution of matter might play a role, which is where the entropy and the practicality problems come together.
I think even if the "end" of the universe is a mess, an uneven version of heat death -- more matter over here, vastly less matter over there -- if something like escape velocity is not given, then after a sufficiently long time the universe will bounce, and bounce in an entropy resetting way.
I want to point out, that the uneven distribution of matter due to early inflation of the universe could make it look like the universe is expanding at an ever increasing rate, but does so only locally, in this bubble so to speak.
That's why I think that after a sufficiently long time the universe could bounce anyway.
Also, the entropy law is always mentioned as unquestionable -- even in the case of the system being the whole universe -- but the reason why the law holds true, the mechanics behind it, is just statistics. I mean there is something like this https://en.m.wikipedia.org/wiki/Poincar%C3%A9_recurrence_theorem
Can someone make me see clearly, that I'm just wrong about this?
Our understanding of the universe is limited by relying on models and a weak understanding of how physics would operate under the conditions of those models, so I like to think that anything could happen when the universe 'ends'.
>the entropy law is always mentioned as unquestionable -- even in the case of the system being the whole universe
This is not actually the case. It is not clear whether we can apply the 2nd law ("the entropy law") to the entire universe, and I think most physicists would agree that its application to the universe as a whole could be questioned. Our current level of understanding of physics and cosmology simply isn't at a point where we can make any statements of this kind with certainty.
I don't see anything necessarily wrong with the rest of what you said. At least to me, the main issue is how you get back to the special condition of the big bang - that is, a point where entropy is at a minimum. If the 2nd law did go into reverse at some future point then it seems more likely entropy would end up at some kind of stable intermediate point, rather than at a minimum.
"At least to me, the main issue is how you get back to the special condition of the big bang"
Accidentally :) or... Penrose came up with the idea that when all matter except electromagnetic radiation would decay into the latter and it would redshift with further expansion until the field is so flat that it makes no sense to speak of any distances between anything anymore, then the universe would not be big, however big it would have been before, but just as small as it had been at the big bang, which would then happen again.
> Also, the entropy law is always mentioned as unquestionable -- even in the case of the system being the whole universe
Alastair is right in that you shouldn't assume that 2nd Law of Thermodynamics applies in all cases to the entire universe. Certainly, phenomena like life seem to violate entropy, but the theory says that life exports entropy to its environment. I'm not really comfortable with that claim, although Boltzmann's formula for entropy seems to describe that scenario. I admit that I should just accept this as true because I have no *rational* reason to doubt it, but I still have doubts—maybe because the Universe seems to be calibrated to encourage emergent systems that have higher order informational complexity than their underlying systems. Information is ultimately energy (unless someone can provide me with a counterexample of information that doesn't involve energy transfer in some way?).
Emmy Noether proved that in an empty universe time translation symmetry should produce conservation of energy. But we exist in a non-empty universe where time translation symmetry is broken (the past state of the universe is different from the current state, and it seems like the future state will be different from the current state—thus symmetry is broken), which has been used to prove that energy is *not* conserved over time in our universe. In Einstein's General Relativity, there's no universal definition of energy conservation for the whole universe. For example, the redshift of light due to cosmic expansion implies that photons lose energy over time, but there's no place where that energy goes. So the 1st Law of Thermodynamics is falsified at the universe level, and the 2nd Law is the result of time transition symmetry being broken. So, according to Noether's theorem, energy is not conserved in our universe. Likewise, information is not conserved. And, if energy/information is somehow conserved, where does it go?
But at approximately ten or eleven billion years into universe's evolution, we see information systems acting as templates for much more complex information systems (i.e. life). For example, genes are strings of information, but they code for much more complex informational systems, which in turn can process information in new ways, and, yes, create new information. WTF?
On the other end of the information spectrum, black holes have been mathematically proven to destroy information (although I think that assumption is starting to be questioned now). Lee Smolin has suggested that black holes create new singularities that spawn new universes. And I think he suggested that information consumed by a black hole acts a template for the next universe—but don't quote me on that. But since a black hole consumes a finite amount of energy, Smolin never explained where all the new energy to create a new universe comes from. But that begs the question of where the energy for our universe came from. But I digress...
> I want to point out that the uneven distribution of matter due to early inflation of the universe could make it look like the universe is expanding at an ever-increasing rate, but does so only locally, in this bubble, so to speak.
Although some cosmologists question the Cosmological Principle that matter is evenly distributed across the universe, at least on a scale larger than galaxies, there's no evidence that I know of that supports this idea. So, we have energy disappearing from our universe, and the universe is (probably) expanding evenly everywhere, plus observations show that the expansion seems to be accelerating (due to the Cosmological Constant which is thought to be dark energy). So if current trends continue, and there's no reason to think that they won't, the universe is likely going to end up a lot colder with a lot less complexity. Sorry, to give you the bad news. It bugs me, too, because eventually I'll run out of pleasant places and bodies to reincarnate into. Bummer. Hopefully, Smolin's idea that black holes are spawning new universes may cheer you up, though. ;-)
Also, the laws of our universe may still be evolving. The laws of the universe had to be different during inflation, so why can't they change further as time goes on? Maybe that's another way out of universal heat death?
Oh, crap. I just remembered there was a study that questioned whether the observed acceleration in the expansion of the universe is an artifact of poor supernova sampling. Perlmutter, Reese, and Schmidt won a Nobel Prize for this observation. And that's been the consensus/dogma spawning all sorts of speculations about dark energy. ChatGPT helped me locate the paper that casts major shade on their study --> "Marginal evidence for cosmic acceleration from Type Ia supernovae" by Nielsen, Guffanti, and Sarkar.
The authors analyzed a larger dataset of Type Ia supernovae to assess the evidence for cosmic acceleration. They concluded that the data is consistent with a constant rate of expansion. This challenges the need for dark energy in the standard cosmological model. Whether gravity would eventually overcome the expansion and create a Big Crunch is a separate question. But if the expansion of the universe isn't accelerating, that puts that option back on the table.
As I have replied to Alastair, Roger Penrose came up with conformal cyclic cosmology which might give you an endless supply of bodies to reincarnate into :) even if the universe is getting bigger and bigger because it “blurres” back to small again.
I'm moving to Montrose, CO soon. If anybody lives there (or near there, like in Grand Junction) and wants to grab a coffee together or something, let me know.
It seems to me that the Venezuelans deported by Trump did not received due process because Trump used a law meant for wartime which literally allows him to sidestep due process for enemy combatants in times of war.
Abrego-Garcia wasn't deported under that law, so the courts could rule against his deportation. The Venezuelans were, so the courts couldn't even say anything against it (because, again, the law had authorized the disregard of due process).
- Why wasn't this order against the Alien Enemies Act enacted before?
- Why did SCOTUS ruled that the courts could command the government to "facilitate" the return of Mr. Abrego-Garcia before it ruled against the Alien Enemies Act?
I'm not trying to argue with respect to this, I'm genuinely curious.
"We have held that an individual subject to detention and removal under that statute is entitled to “ ‘judicialreview’ ” as to “questions of interpretation and constitutionality” of the Act as well as whether he or she “is in fact an alien enemy fourteen years of age or older.” Ludecke, 335 U. S., at 163−164, 172, n. 17... The Government expressly agrees that “TdA members subject to removal under the Alien Enemies Act get judicial review.” Reply in Support of Application To Vacate 1. “It is well established that the Fifth Amendment entitles aliens to due process of law” in the context of removal proceedings. Reno v. Flores, 507 U. S. 292, 306 (1993). So, the detainees are entitled to notice and opportunity to be heard “appropriate to the nature of the case.” Mullane v. Central Hanover Bank & Trust Co., 339 U. S. 306, 313 (1950). More specifically, in this context, AEA detainees must receive notice after the date of this order that they are subject to removal under the Act. The notice must be afforded within a reasonable time and in such a manner as will allow them to actually seek habeas relief in the proper venue before such removal occurs."
"Importantly, as the Court stresses, the Court’s disagreement with the dissenters is not over whether the detainees receive judicial review of their transfers—all nine Members of the Court agree that judicial review is available."
>What allowed SCOTUS to, two days ago, order a complete halt of the deportations of these Venezuelans?
"After the district court’s order and starting last night, counsel for Applicants and the proposed class of individuals subject to removal under the AEA were informed that numerous Venezuelan nationals currently in the government’s custody in the Northern District of Texas have received notices that they are subject to removal under the AEA, and further were informed by government officials that they may be removed from the United States as soon as this afternoon or tomorrow. DHS has now publicly announced that AEA removals are imminent.
The Government’s actions to-date, including its lightning-fast timeline, do not give members of the proposed class a realistic opportunity to contest their removal under the AEA. The notices some members of the proposed class have received are in English only and do not inform proposed class members of their right to contest the designation in a federal court. The government has refused to give any information to undersigned counsel for the proposed class. And, as far as Applicants and their counsel know, the government is not giving notice to proposed class members’ immigration attorneys.
Removal without sufficient notice and time to seek habeas relief is in clear violation of this Court’s decision of April 7, 2025 in Trump v. J.G.G."
>Why wasn't this order against the Alien Enemies Act enacted before?
The Trump admin appealed Boasberg's initial Temporary Restraining Order up to the Supreme Court. TROs are normally not appealable because it's irresponsible for superior courts to assert jurisdiction under the conditions TROs are granted - without other parties present, before fact finding occurs, the very first step in the judicial process with little to appeal. The first appeals court denied to block Boasberg's initial TRO preventing deportations under the AEA, particularly because the federal governnment stated in oral arguments that they don't believe deportees are entitled to due process and that deportations would resume immediately.
"Indeed, at oral argument before this Court, the government in no uncertain terms conveyed that—were the injunction lifted—it would immediately begin deporting the plaintiffs without notice."
"(J. Millett: “My question is, if we were to grant the relief you request, would the government consider it necessary to allow time to file a habeas petition before removing people? * * * [Is it] the government’s position that it could immediately resume mass removals of the five named Plaintiffs and the class members, immediately? Government: “Your Honor, * * * we take the position that the AEA does not require notice * * * [and] the government believes there would not be a limitation [on removal.]”). The Constitution’s demand of due process cannot be so easily thrown aside."
There were further comments in the circuit court opinion that "invasion" is a term of war and that the invocation of the AEA was likely illegal.
Since the Supreme Court was handling a TRO after the appeal from the circuit court, they issued a purely procedural ruling that clarified what I pasted above, that 1) All "alien enemies" can challenge their deportation, 2) Thus must be done via filing Habeas, not via class actions, 3) There needs to be a reasonable time afforded for them to file it. ICE continued its bad faith behavior by giving subjects 24 hour notice and attempting to deport on Saturday when most judges and lawyers wouldn't be working, hence the emergency order by the Supreme Court to halt all deportations that came at 1 am Saturday morning. https://www.scotusblog.com/2025/04/justices-temporarily-bar-government-from-removing-venezuelan-men-under-alien-enemies-act/
>Why did SCOTUS ruled that the courts could command the government to "facilitate" the return of Mr. Abrego-Garcia before it ruled against the Alien Enemies Act?
There was comparatively more fact finding and argumentation that needed to be done in the AEA case, where only the TRO was appealed. In the Abrego Garcia case the government admitted its administrative error and there was already prior due process initiated by immigration judges that granted Abrego Garcia's witholding of removal and the only dispute was whether the government had to do anything to bring him back.
>which literally allows him to sidestep due process for enemy combatants in times of war
That might well be unconstitutional (as Trump's sister ruled several years ago in a case that was overturned on appeal on purely procedural grounds). Edit: My mistake: she ruled re the law being used against the Columbia student.
>Why wasn't this order against the Alien Enemies Act enacted before?
It is my understanding that the Alien Enemies Act has been invoked only about 4 times.
Even if a law says "you don't get due process in this situation," you can still get a court to ask "are we in that situation?" The AEA is an act that applies in wartime, which was invoked to deport members of Tren de Aragua, and the people challenging it are arguing that the law can't be used on them because it's not wartime and they aren't members of Tren de Aragua.
As far as I know, the timing of the SCOTUS ruling is just the vagaries of the justice system - it took time for the case to work its way up to SCOTUS and for the lower courts to wrangle out exactly what the government was doing wrong and what they were willing to enjoin. Also, ICE was planning another set of deportation flights so it suddenly became urgent - there were buses literally on their way to the airport which turned around once the ruling came out.
The Federal courts are absolutely allowed to tell the Executive Branch, "Stop doing that; it's quite illegal". Provided, of course, that it actually is illegal, and that someone who has been actually harmed by this has raised a complaint (in the right court, but we don't need to go into that here). Once you get past the level of trial courts, and certainly by the time you reach SCOTUS, that's basically what the courts are *for*. The president and his officers, sworn to uphold the laws, then have to obey the court's injunction against not doing illegal stuff.
As a corollary, the courts can also tell the Executive Branch, "Hold off on doing that for now; it would cause irrevocable harm and we haven't figured out whether it's legal yet". Provided that the act in question really is at least plausibly illegal and irrevocably harmful. That's where we are with most of these cases.
There's certainly a plausible case that the deportations are illegal because the law being cited to justify them requires that the United States be at war with someone, and there's a distinct absence of e.g. Venuzuelan jets bombing American cities or vice versa. The courts are I believe properly skeptical of claims that the lesser ongoing unpleasantness of US-Venezuelan relations constitutes a "war" under the law, or an "invasion" or a "predatory incursion" or whatever. But that's something that needs to be argued in court, with lawyers from both sides making their cases.
And I think we can take the "irrevocable harm" part as given, now that the Presidents of both nations have plainly stated that it is impossible to return deportees from El Salvador to the United States. But that's the part that was (briefly) argued in court before SCOTUS issued the preliminary injunction against deporting anyone until all of this is sorted out.
It took this long to get to that point, because SCOTUS can't just issue rulings out of the blue, it has to wait until (laywers representing) a complainant who was actually harmed bring a case before a US District Court, which is then sent up to an Appellate Court before finally getting to the Supreme Court. Or both sides can agree that the District Court got it right and let their ruling stand, but Donald Trump doesn't seem to be in a very agreeable mood.
I think the short answer is the whole situation is a hot mess and everyone is trying to figure out how to deal with it. I am sure there is a lot of tension within the Supreme Court about how to deal with the alien enemies act and it’s application. So they dodged that question and concentrated on the habeas corpus issues that they raised when they instructed lower courts to facilitate Garcia‘s return.
The newsletters I know that are a bit more technical with this side of things are Joyce Vance’s (“Civil Discourse”) and Steve Vladeck’s (“One First”), both on Substack. They’re not neutral and I can’t vouch for their accuracy, but they seem to know what they’re talking about.
Anyway, my (amateurish and not completely clear, please double-check everything I write) understanding is that the SC order halting the deportations is temporary in effect until the issue is properly litigated by one of the lower courts.
The rationale is that the situation was time-critical, as the Venezuelians were being loaded into buses towards the airfield (from which deportation flights would take off) – and then lead them to CECOT from where the detention could not be meaningfully challenged.
This seems similar to the situation a month(ish?) ago, where the government essentially deported the immigrants at the same time the judge was ruling on the merits (against the government), but the timeline was pretty confused and the government seems to have sought to argue by “fait accompli”.
The Abrego-Garcia situation is another one where the government claimed in court (although they later said something else publicly) that the deportation to El Salvador was a mistake (since there was a court order preventing that), but that since it was done and Abrego-Garcia outside the jurisdiction of the US, there was nothing to be done and the entire proceedings were moot.
The court felt differently – and ordered the government to fix its mistake. The SC tempered the language because the Executive branch is the one deciding foreign policy, but it’s still expected to correct its own mistakes.
TL; DR: different cases, technical arguments, the judicial process in the US is complex, and, as far as I can tell, many of these arguments are about process rather than the fundamental questions (eg is Trump’s use of the Alien Enemies Act correct? Should anyone else get a say in what constitutes an emergency or a foreign invasion? Should anyone else’s opinion about the extent to which the claimed actions address said emergency or invasion matter?).
“The judicial process is complex” just meant that it had an internal logic hinging on intricacies that wouldn’t find their way into a newsline and that most wouldn’t remember unless they were really interested and/or professionally had to deal with that.
These include the written rules, the unspoken customs, the randomness of scheduling, the extent to which people think they can get away with bending or breaching rules or customs… this was an absolutely agnostic on what was actually happening inside the courts.
Anyway, there may indeed be an argument that the Court is bending its own rules. Here is also, in the spirit of balance, an argument that Alito’s dissent contains mistakes of fact, and raises “issues” that apparently didn’t bother him in other cases: https://www.stevevladeck.com/p/145-justice-alitos-misbegotten-dissent .
I am very open to the arguments that SCOTUS should act quickly in the modern age, especially when District Courts are ordering nationwide TROs.
But some of the left leaning SCOTUS judges previously opined that some of these issues shouldn't be taken up by SCOTUS because the issues weren't "ripe" (eg no urgency to review, appeals hadn't run its course). Of course, they did that because their "side" was favored by letting it play out slowly.
SCOTUS is taking up the birthright issue and I applaud that SCOTUS will probably resolve the issue probably this year. With a decision either way, we will get a ruling quickly rather than each side fighting in numerous District and Circuit courts for years.
I just want the Judges to be Judges. Right now, the Judiciary is more of a oligapoly where they make up rules to suit their policy preferences.
I don't know if I'm a fan, exactly, but I can say I read the *entire* book series. There definitely was a volume or two where the problem was Too Big To Edit and the tactful use of a chainsaw to cut down some of the thicket of prose would have improved matters hugely. The ending was something of a disappointment, at least to me.
But there is a good story in there, good characters, and the way he tied it all in to other elements in other stories to create a background universe behind all his works was very cool.
It was also interesting to see the change in his writing style from the original text, somewhere in the 70s, to what he was producing in the 90s. I've seen it with other long-running series where authors are writing over decades and the stylistic quirks of one era stand out by contrast to the current style.
It's not my favorite fantasy series, but I enjoyed it. I bought the series after finishing the second book, but I think the second book was the high point, and the late ones had some instances of noticeably weak prose. (Likewise, the poem it's based on has a wonderful opening verse but then perhaps bogs down. https://fivers.typepad.com/files/childe-roland-to-the-dark-tower-came.pdf)
Finding out the movie Hearts In Atlantis was actually based on Low Men In Yellow Coats, which was a Dark Tower side story, was an interesting factoid.
Read through them sometime around university, generally liked them, really liked the ending, haven't reread them since except to dip in and reread a few favourite scenes. I love the overall aesthetic, especially of the first few books.
Just listened to this: “When they finally arrived they found the castle deserted except for the rooks and blackbirds. The walls had been broken. There had been a great slaughter on the fields to the west. It were white with bones and red with rusty armor, so my dad’s granda said. And the voices of demons cried out like the east wind from the jawbones of those who’d fallen there. A village beyond the castle had been burned to the ground, and a thousand or more skulls were posted along the walls of the keep.”.
Hey Gunflint, discovered another cat toy. I think this is their favorite of all time. They chase it wildly and never get sick of it. My smart cat knows that if he paws randomly at the controller he will eventually turn it on.
Feed and play with the cats. I throw bouncy balls into the bathtub from a few feet away, banking off bathroom walls, and they dart aroundin or near the tub trying to catch them in the air and sometimes succeeding, and otherwise just staring at each for the longish period it takes it to come to rest.
Read. For a long time, I read paper books or e-ink screens using a red or amber reading lamp while getting ready for bed. The red lamp in particular was useful for not disturbing my wife on nights when she went to sleep before I did.
I have since fallen out of the habit of this, in favor of reading on my phone instead. I should probably try to return to my former habits, which worked very well.
Read, sometimes. Work, sometimes. Most of the time I wind down by rereading my notes from the day and making edits or summaries, and reviewing notes from previous days. I've had a paper note system for years, so there's no screens involved except for infrequently taking a quick picture of something with my phone.
I take notes on index cards throughout the day. They're a mix of short range (grocery list, diagram of directions, ingredient amounts for a large batch recipe) and medium range (minutes from a work meeting, quotes and insights from a book I'm reading, a sketch and background info for someone new I met at a dinner party) and long range (tools or techniques from a class, definitions of new words, reminders for myself at the end of the year.) It's a bit scattershot in the first draft format, so I try to neaten it up at the end of day.
Re: college admissions essays, perhaps the best thing would be for America to adopt the British system, whereby students do write admissions statements, but they're expected to stick quite closely to "Why do you want to study this topic? And what makes you think you'd be any good at it?" That lets students mention any *relevant* extracurriculars ("The summer I spent volunteering in Colombia really helped me improve my Spanish, which will be useful for this degree in Hispanic Literature") or mitigating circumstances ("Although my B in Music might not seem very impressive, I think it becomes more so once you remember I've been deaf since the age of seven") without encouraging the sort of extracurriculars-padding or sob stories which I gather are necessary in the US.
I know of a person who got into RISD (Rhode Island school of design in Provincetown) in a very clever way. She mailed in her application envelope and inside there was a one line note; it said “if this letter gets to you, you should admit me to your school.”
A close examination of the envelope revealed that the postage stamp had been hand drawn by her. I saw a photograph of the envelope. It was perfect.
An oddity of the British system is that places are awarded based on letters of recommendation or predicted grades from second level teachers, and the offers are made conditionally before the A-level results are announced. I read some time ago that there is a substantial discrepancy between predictions and actual grades at fee-paying schools; and that this discrepancy is one of the drivers in the inflated rates of progression from fee-paying schools to prestigious universities there. Offers are almost never rescinded - so the net result is poorer performing students from the fee-paying schools beating less wealthy students in the race to Oxbridge.
If your exam results haven't been announced yet, your offer is conditional on your actually achieving your predicted grades. If your teachers have been over promising, you probably won't get the grades, and hence won't fulfil the conditions of your offer.
I don't have an issue with that but American educational establishment just doesn't like that sort of specialization in high school/early college. That's why we study Languages, Sciences, Math, Social Studies all throughout high school, rather than specializing in 2-3 subjects in the last few years of high school which I think is the norm in many(most?) countries. This doesn't end in high school either as you will still have to take history even if you are computer science major for example, unless you tested out of it by passing various AP history exams in high school. For your proposal to be implemented, the whole American educational philosophy would need to change. And students not having enough unique stories for their college essays is just not that big of problem for that to happen.
> That's why we study Languages, Sciences, Math, Social Studies all throughout high school, rather than specializing in 2-3 subjects in the last few years of high school
Hm, I think you got that wrong. The term "specialization" is used indeed in other countries, but means something much weaker than what you describe. At least in Germany, and I think it's more or less the same for other European countries.
You don't just have 2-3 subjects in the last few years of high school. You have all kinds of subjects with a bit of choice, for example if you "specialize" in computer science you still have History/Social Science/Economics, and you still have arts classes, and you still have some natural sciences and German classes and at least one foreign language and so on. "Specialization" means that in the selected 2-3 topics, you have a few extra hours per week, for example 3 hours in the non-specialized courses versus 4-5 in your specializations. Even the specializations need to be broad, usually they need to include at least one language, one STEM subject and one social subject. (Details vary on region and school.)
You are right that it is different in college. There many European countries do have the specialization that you describe (though not all, for example France also starts with two years "CPGE" of more general education).
Ok fair, I was mostly thinking of British A-Levels. On the other hand, don't Germans have different kinds of high schools? Like a school where academically inclined students go to, a school where vocationally inclined students go to, etc? And a large number of students actually go to vocational high schools and the sorting process for that actually happens fairly early on in their lives? Nothing wrong with that system in my opinion, it seems to work quite well for Germany, but that kind of sorting/specialization just doesn't align with American educational philosophy or even American culture more broadly.
My experience, mainly applying to California colleges c. 2000, was that almost everywhere required or strongly encouraged you to declare a major on application. You could change majors after matriculating, but there were a lot of hoops to jump through and some fairly strict veto points unless you were transferring between majors within the same "college" (at my university, departments were divided between six "colleges", each with their own sub-administration: Agriculture, Architecture, Engineering, Liberal Arts, Science and Math, and Business) and even then it was easier if you were moving between closely related majors and if the target major was considered less technically rigorous than the one you're leaving.
That matches my experience as well. I declared my major in my application but my college essay topic was a general one. And I think my high GPA and SAT scores probably played a more important role than the essay in my application.
No matter which way you go the doctor is right. It is biochemical. The question then becomes what do you do about it. Every person sees pain through their own lens. How could they not?
Is it though? The correlation between, say tissue damage and pain sensation is much weaker than a purely biochemical nature would entail. Solders reporting looking down and realizing their leg’s been blown off is one end. Chronic pain with no discernible cause is the other. Then there’s phantom limb pain many amputees report.
Pain sensation is ultimately made up in the mind/brain, so in this sense it is biological, but it’s then trivially true in a non-interesting way.
Ok, I see what distinctions you are trying to make.
They are all flesh and blood issues but as you say, I suppose that’s trivial…
although I am not entirely sure that it is when you assess the three choices that were listed. I can see the distinction between pain that is caused by clear physical trauma, and the kinds of pain that aren’t. Most of what we call emotional pain has physical manifestations. An aching in my heart, frustration; shock. The subjective nature of pain defies categorization don’t you think?
Exactly. Full disclosure: I participated in some early research these guys were doing as a willing subject precisely because pain fascinates me. It’s entirely subjective, and psychology plays a major role. There’s a reason inquisition torturers first just showed the tools. OTOH as a martial artist I just learned to ignore certain kind of pain that is not “dangerous”, i. e. doesn’t lead to damage.
Any chance this is private healthcare (ie, wrong incentives) related? I've never had a doctor insist it must be something related to their specialty, quite the contrary.
As to the person who moves through different systems, I'd say that's the patient, sadly. GPs can take that role to a degree, but I feel that once you are off the basics, you are on your own.
Picked up a copy of https://store.steampowered.com/app/2844290/Hard_Chip/ on Steam the other day, enjoying it so far. It's the latest incarnation of the "hey, this game is just my day job but slightly simplified" zachlike genre, and probably the most hardcore one I've seen to date. I feel like of the public spaces I inhabit online, this is the most likely one to contain someone else who might appreciate it :)
Zach himself actually did a game about semiconductor design (KOHCTPYTOP: Engineer of the People) way back when, although this looks bigger and far more polished. Definitely a hardcore topic for a zachlike.
>"hey, this game is just my day job but slightly simplified" zachlike genre
I've often wanted to make one of those games based on my own job. I'm kind of a general manager type at a small business: not the owner, but in charge of a lot of admin work. The game would consist of your boss giving you more tasks than you can actually physically complete during an in game work week, you have to decide what to work on and what to let slip through the cracks. Every week you have a meeting with the boss and try to show off your accomplishments while hoping that he doesn't ask about the tasks you didn't do.
Scott, I'd love to see you delve into systems theory in a bit more depth. I think you'd have some interesting insights if you thought about this a bit more:
> Oh! I agree this makes sense if you need to talk about the “purpose” of an un-designed system with no humans in it.
If you build out POSIWID as a way of understanding undesigned systems, you'll end up with an entirely separate perspective that you can apply to all systems -- designed or not. And you might find this perspective is a useful adjunct to hearing what designers say about it.
To make an analogy to the mind, you might think of POSIWID as akin to an MRI -- it gives you a rough structural view that you can try to tease apart into an understanding. You'd be right to say that we shouldn't rush to understand everything in terms of MRIs when we can simply talk to people about what's on their minds. But nor should we entirely discount the information from the MRI in favor of what the person says.
(To be clear, I agree with your posts about how absurdly POSIWID is misused on Twitter. But if we judged ideas by how they were misused on Twitter, what ideas would we have left??)
POSIWID is a useful concept applied to ourselves. A human being is a system. (Opinions will vary as to whether it is a designed system or not.) Our behavior is the output of that system. The question then becomes, is the system doing what we want it to do?
Indeed, but our desires are another output of the system. If it's alignment we seek, should we change the world to match our desires, or change our desires to match the world?
Our desires are another vector in the system that leads to our behavior. As to your question about what needs to change or what should change, it’s a constant negotiation, isn’t it?
I'm interested in people's experiences with using LLMs for automation.
I've tried using LLMs for automating a certain part of my job, but unsuccessfully.
The main problem is that it's impossible to communicate in enough detail requirements of the task. When humans do this task, they learn it on the job through a long series of reviews until they end up having intuitive grasp of what's expected of them.
However, this means that there are no explicit instructions that could convey all necessary information to do that task.
Passing many examples for in-context learning does not help, the generated examples still fall far too short. I haven't tried fine-tuning on raw data dumps from reviews, mainly because it would take a lot of effort to format them properly and I have a lot of doubts that it would help at all.
For this task, reasoning models work far better than non-reasoning ones, even though the task is creative in its nature. This is mainly because reasoning models will roughly follow long and detailed instructions, while non-reasoning models will ignore almost all of them. This is why I doubt that fine-tuning would help - I haven't seen that any reasoning models can be fine-tuned and non-reasoning models already perform so poorly that I doubt fine-tuning would make them useful.
I think this is why benchmark progress with LLMs rarely translates into useful work - current top tier LLMs are already incredibly good at tasks which can be easily described to them and making them even better at those kind of tasks will not suddenly make them able to automate a lot of work.
It seems to me that a lot of people expect that you can cheat conservation of information with LLMs - that tasks which require humans to learn for 2-12 weeks can be described in a simple prompt. In most of the cases that information does not exist in textual form which could be given to LLMs and even when it exists we don't have the tools to make LLMs incorporate it properly.
All of this makes me skeptical that LLMs will be used for automating anything more than extremely straightforward tasks until the issues above are resolved.
Has anyone managed to resolve the issues above, or work around them, on real-world tasks for which you pay people to do?
I'm a psychologist in private practice, and had thought about using LLM to help me with my billing, and, like you, I don't see a way to do it. My system is kind of cumbersome. I do not use practice management software. Instead, dates of client appts are in my computer calendar, and the rest of the info is in 2 other folders. When I do billing, I bill some people's insurance, and other people directly. I have a folder, organized by patient initials, in which I keep a copy of bills I sent. I have another folder, also organized by patient initials, in which I keep a record of appt dates, bills sent, and payments. Payments come in the form of bank transfers, PayPal payments, and checks, and I enter those in the second folder.
Yes, the system is cumbersome, but I could explain my system to any bright high school student and they would be able to do the whole thing for me. I can't figure out way to turn it over to an LLM. I don't see a way to explain the system to it, give it access to my calendar and the 2 relevant folders, and just tell it to have at it. How can these things have anything like human level general intelligence when you can't get across to them how to do a messy, but quite simple, task like this? Or, I dunno, make one of the advanced GPT's could? If I entered instructions plus the relevant calendar, payment details and folders, , could it enter dates of service, charges and payments in the records, and make out bills?
You're looking for an AI agent, which is still a nascent field. LLMs simply transform text, so you can talk to them and have them answer questions, but the ability to do things for you in the real world is limited to text-based triggers they can invoke.
There's a lot of talk about this year being the year of the agent in AI, so there's likely to be a lot of progress on this coming soon. But even then I don't think it'll be as simple as turning it over to an AI. Given this is such a critical function, it's worth thinking about how you'd modify your current system to allow you to oversee an AI that's doing most of the work but makes occasional mistakes.
E.g. perhaps you OK each bill before it goes out, and OK each paid bill.
But what is being an agent, really? There are certainly tasks like my billing task, but simpler, that AI can do right now. For instance I know a teacher who used GPT4 to write the paragraph that went out with each student report card. He has a number of comments he makes over and over when he writes these paragraphs, so he made a list of common comments. For instance about homework completion, he might most commonly say “continues to do a great job with completing homework” or “homework completion has greatly improved over last term” or “continues to have a problem with completing homework” etc . So he made a document with a list of comments. The homework comments were one group, with individual comments labeled h1, h2, etc. Then there was another set of 6 or so possible comments for, let’s say, participating in class discussion, and they were labeled d1, d2. And so on. He gave GPT a list of student names, each followed by several alphanumerics corresponding to the comments that applied to that student, and also the document with the list of alphanumerics and the comment that went with each. Then he asked GPT to write a paragraph to go with each report cord, in a friendly, casual prose style, incorporating all the information corresponding to the alphanumerics after each student’s name. GPT did just fine with this task, including recognizing on its own that in order for the paragraph to flow well it could not just use one sentence for each alphanumeric. It often put two or more together in many sentences, making sentences like “Amanda continues to have trouble with homework completion, but has become a lively participant in class discussions.”
Seems to me that handling my billing is not any different in kind from the task the teacher gave GPT, just has most moving parts. If it did my billing all GPT would have to do would be follow clear instructions about what info to put where in what records document, and what kind of bills (personal bills or submissions on a form for their insurance company) to generate for each patient. Are you thinking that to be an “agent” that can carry out my task GPT needs to engage in some process that only conscious beings carry out — would have to “think,” “understand”?
Oh no, I'm definitely not thinking there's a difference in kind between your billing task and the report card task. The difference is in who's creating the "glue" that turns the textual task (which LLMs are good at) into something that actually affects the world.
To give an example from my day: I'm a software engineer, and I took the week off to work on some personal projects and see how much AI could help me. I had a complex 500-line script that I wanted to add a couple of features to. I described what the script did and what I wanted it to do, and I asked Claude to rewrite it with a couple of new features. And it immediately did so -- doing work in about a minute that would have taken me all day. It was stunning.
But what it couldn't do was _run_ the code. This crosses over from the "manipulating text" domain (where it excels) to the "using tools" domain, where current LLMs are fairly limited. For the most part they have some set of integrations that they can invoke textually (e.g. ChatGPT can call an image generator to make pretty amazing pictures) but they can't operate a computer or a web browser directly. In other words, they need special integrations to actually perform tasks (as opposed to giving you back some text) instead of being able to interact with computers the way people do.
There's no real reason this problem can't be solved; in fact there are a bunch of great tech demos and products in development in this space. E.g. OpenAI's Operator, Anthropic's Computer Use. These should be able to solve the problem you described, they just require new training, quite a bit of inference compute (they need to process a lot of images, which is more expensive than text), and quite a bit more trust. (The latter will be the biggest hurdle -- if text generation goes off the rails 1 time in 100, that's unfortunate but not ultimately a deal breaker. If the AI is controlling a web browser or your computer it can accidentally do much more harmful things, like deleting files or accounts.)
Given your existing flow this may not be worth the effort, but I imagine that if you gave an LLM all the information it could figure out the right thing to do. It could look through and understand the two folders of patient files, it likely has a calendar integration or could understand what you copy and paste in from your calendar, it could trawl through bank statements or emails to find payments, etc. It might also have some spreadsheet integrations where it can add rows to, say, Google Sheets.
But it's probably not going to be able to gather data from your online accounts or read files from your computer. It just doesn't have the right interfaces at this point. The major AI companies are working to fix this over the course of 2025 though!
Then the question will be: how do you modify your process from where a single, trusted person is doing everything to where you've got a really smart and motivated intern whose work sometimes needs review?
To me this is a task for an IT rep to design some Power Automate flows to read emails, interact with excel and attachments. It would require a different technical skillset than working with an LLM, but the tools are very powerful at being agentic in very specific, repeatable ways.
<Then the question will be: how do you modify your process from where a single, trusted person is doing everything to where you've got a really smart and motivated intern whose work sometimes needs review?
Is it worth trying this now -- modifying process? I could certainly upload the relevant folders: one of actual bills, each patients' bills, current and past, in a subfolder within the folder; and one a folder of spreadsheets, one for each patient. Columns in spreadsheet are date of service, charge, date of billing, amt billed, date of payment, amt paid. And rather than having the AI trawling my back account,etc., I could just occasionally upload a list of recent payments -- who made them, date of payment, amount. And I can explain very clearly the steps that must be taken to enter appointment dates, make out bills, etc. I can provide a model of the 2 bill types used. But would that work? Is there an AI that would remember what I'd taught it, including mistakes in its process that I had corrected. Is there one where I could just leave the folders stored with it? It does not need to, for ex., print out bills. If it makes them available as text files or even image files I could download and print them myself.
This is mostly a summary of the conversation thus far, but it'll be great context for helping it understand your goals. And it'll be able to answer very specific questions like "how can I integrate my .xlx spreadsheet into an online workflow".
(I have to say it's kind of weird for me to realize that if I were replaced by an AI in this thread it's very likely you would have gotten even better answers than I was able to give!)
IMO it's definitely worth trying this now. Even if it doesn't end up doing what you wanted, it'll give you a deeper understanding of how to interact with these systems and a better idea of what features are missing.
The simplest prototype would be to start a conversation with Claude or ChatGPT, and see if you can get it to do processing steps. E.g. "here's what the spreadsheet looks like, I'd like you to generate new rows based on..." It may stumble, but ultimately it's a conversation so you can just tell it what it did wrong. It'll probably pick it up pretty fast, unless there's some hidden complexity here I'm failing to imagine. Once it does, you can come back to the same conversation later and say "here's more stuff to process in the same way."
Of course the utility of this will be somewhat limited, since you're directly managing the inputs and outputs (which is probably most of the work anyway). I just learned about Zapier today, which might be exactly the adjunct you need -- you can use it to define flows of various kinds, and assuming it has the right integrations (which is a big assumption) you might be able to set up a flow like "every day at 5 AM fetch new email, send it to Claude with this prompt template, and insert new rows generated by Claude into the 'review' spreadsheet".
But creating this workflow with Zapier sounds like a lot of work! So now here's the fun part: I bet Claude (or flagship LLM of your choice) can help you design this workflow. Just tell it what you've got and what you're trying to do, and I bet you'll get a good answer.
I think this is precisely the sort of task where AI is going to take much longer to catch on to what is required. There are all sorts of unwritten rules/procedures that you don't realise you are following. And at best, AI is prone to strange hallucinations - someone does need to check its work. My own opinion is that this will be the downfall of many AI applications: something works 99% of the time, the overseer gets lazy, and it only takes one really idiotic failure (data leak?) to end the experiment in a given company.
I think it's deeply ironic that the work of what we perceive to be highly skilled professions like programmers, accountants and lawyers (here I mean trawling through case law to write legal documents, not arguing in court) will probably be automated much faster than secretarial work. The difference is that these professions are text and rule based, and you can quickly check if the output is correct or not. In contrast, secretaries don't just make appointments and type letters - they handle all sorts of exceptions and errors efficiently and appropriately.
Out of interest - is there a reason you don't want to use off-the-shelf software to do this? AI really isn't the right tool, or at least the currently available systems are not in any way specialised for this.
Oh I tried it years ago and did not like various things. They could all do lots of things besides billing, none of which I wanted, and access to the extra features complicated the workspace. And then there were various littlethings they didn’t do the way I do. Like maybe something like the bills they made out looking too much like gas co bills and not informal and friendly enough for my taste. And then there were various little exceptions I did not see how to make it make. For instance some people
don’t get bills because they choose to just pay at each session or at the last one of the month, and it was more trouble than it was worth to get the software to take that into account and not make out bills to those people.
Some of the EHR systems are rolling out AI features: the one we use at our office is focused on AI notetaking. You can have their AI app listen to the psychotherapy session, then it takes what it hears, transcribes it, and puts it into proper note format.
We haven't actually used it yet because it's got a lot of bugs, but if it works it would certainly save a lot of time.
Oh, I wouldn't like that. First of all, my notes aren't summaries of what was said, since I remember the general info from stuff talked about in sessions the same way I remember long serious talks with acquaintances. My notes are insights or ideas or questions or unfinished business. Also, patients hate having their sessions recorded. I think I would feel the same in their place.
Your patients would put up with having AI listen to the session? I have people who won’t even bring their phones into my office because of concerns about spyware! And most people loathe doing exercises that I sometimes propose that involve recording themselves, such as doing some kind of role play and listening to it together.
IME fully automating large or complex tasks is not possible with the current generation of models. But they're great for increasing the leverage of skilled people by automating the smaller steps.
E.g. when I write code I find they're great for selecting libraries or design patterns, and great for writing clearly-specified functions. This dramatically reduces coding time, especially when working with unfamiliar libraries, languages, or design patterns.
I'd be curious to hear more about the specific tasks you're trying to automate. They sound too big, or too heavy on local context to be easily asked of a generic LLM.
No, of course not. I mean only the balance of electability. I haven't seen anything that makes me believe that the democrats are attempting to appeal to the tipping point voters they need to win elections.
Are there a significant number of Catholics in her district? Or does she represent the "evangelical protestant who may or may not still identify Rome with the Antichrist" crowd?
I had a great spin on the traditional forgot-to-study-for-the-exam-tomorrow dream. There was a whole room of us, all frantically revising, and as people passed more and more topics around I realised I'd completely missed more and more of them, and things went from bad to worse in the traditional manner.
Then I suddenly went, hang on, this isn't right. I asked a classmate, "Paul, how old are you?" and he thought for a minute and said, with surprise, "35." "Don't you already have an engineering job?" "Yes. A very good one."
By this point all the other students were listening and realising the same thing. There was a short period of mutual sanity checking, and we all began questioning more and more: "Wait, these topics don't even go together, why is one exam testing all of them?" "Actually, guys, does anyone even know what institution is setting this exam?" "Does anyone actually recognise the library we're in right now?" "What the fuck is going on?"
At no point did any of us think to suggest we might be in a dream.
I once had one of those dreams where I thought I'd just woken up (but hadn't), and I noticed a strange quality to the light, and it felt sort of threatening and unreal. Spent a few minutes feeling like something was going to jump out at me before I woke up for real.
A week or two later, I was looking across a big grass field and I suddenly noticed the same strange quality to the light. I thought to myself, "Wait! That's just like the light in my recent dream! Only this time I'm seeing it while I'm awake! What could it possibly mean?"
On New Year's Eve, 1999, I was hanging out with some friends, and went to sleep on their couch in their basement after having a few jokes about all the computers crashing cataclysmically from Y2K bugs.
I woke up to a very bright light.
It took several seconds for my brain to realize I wasn't at home, the basement had patio doors to their backyard on that side of the house, and the sun was shining in my face.
I have that a lot, dreams where I'm in high school but aware that I have a good job as a software engineer, and a college degree. My brain often splits the difference and decides I'm auditing the classes and can skip if I want to.
Yeah mine also seem to come up with some half-assed excuse why I'm still hanging around high school and/or college despite being way past that stage. It does often seem to lower the perceived stakes and stress level at least a bit.
>There was a whole room of us, all frantically revising, and as people passed more and more topics around I realised I'd completely missed more and more of them, and things went from bad to worse in the traditional manner.
Something similar actually happened in my old university, where one year it turned out some of the professors had taught completely the wrong syllabus for a given paper. They ended up having to produce a new exam for the four or five students affected.
I've had dreams where I'm working in Excel or on a program and I just can't figure it out. And while I can never actually figure out that it's a dream, I am able to realize that I will be unable to save my work so try to convince myself to forget about it. Usually works but then I might start up again on another problem and be right back where I started.
I tried to convince my son to make an essay for one of his 17 college applications to be, "How I managed to secure every #1 pick in the 2028 NBA2K draft". One of the video game's playing modes was to build a team over the years and manage draft picks, and he was able to massage it so that he had the entire first round. I thought that would have showed ingenuity and tenaciousness and it certainly would have been memorable to an admissions officer (indeed, I wouldn't have been surprised to see one of them here mentioning this essay!). Unfortunately, Management vetoed that as an option before it even hit the draft stage, so he had to talk about the hardships of growing up in a leafy suburb.
I doubt that management made the right call. College admissions officers I saw quoted were rolling their eyes at hardship essays by kids without significant hardships in their lives. Seems to me that choosing to write a hardship essay under those circumstances is a likely marker for conventionality, dishonesty and lack of confidence. Besides, common sense suggests that applicants will do better with an essay that stands out in some way. If somebody has suffered a hardship that itself stands out, then the essay will be memorable. Otherwise, it will not.
And wtf is wrong with kids these days? My daughter in high school wouldn't even listen to my advice about how to do better in geometry. When I was in high school I wouldn't listen to my parents' advice about anything. They never even saw what I wrote on my college applications.
There's always the worry that you blow it by saying you're qualified for a top choice school because you outsmarted a video game. Management would have gone through the roof on that one. But it would have probably been worth a shot at one of the unlikely colleges. I think his hardship ending up being about breaking his foot prior to his junior year basketball season after he had worked so hard to get into a good position. And I've already forgotten though it's only been four years, but there seemed to be multiple essays required for each one, only some of which were "hardship" essays.
Hmm, another example of how one purpose of a system what it does? The purpose of universities is to filter applicants such that demand for slots within the university corresponds to the supply, in a way that no stakeholder will find too objectionable. Essays look like a good tool for that.
The things you stated (winnowing applicants, avoiding major objections from stakeholders) are requirements for the admissions process, certainly, but there are many ways they could fulfill those requirements, and some of them are substantially less expensive than the current system (such as lotteries, as others mentioned, or auctions, which would increase revenue). You can say "well I guess all those other methods were unacceptable to some stakeholder for some reason" but that doesn't actually explain anything.
Seems like you're experiencing one of the failure modes Scott highlighted, where you pick one particular effect of the system (or two, in this case) and call that the purpose, when actually there are a number of other effects.
I think the college is probably also trying at least a little bit to get students that will be more successful, and thus improve the university's reputation, donate more money, and not cause as many problems while attending.
Plausibly the system also provides deniability for the admissions department to accept various bribes (either direct or indirect), cater to their personal biases, and make it harder to sue them (because if the process is illegible then it's hard to prove any specific bad thing about it).
Ok, all that makes sense. I think of the admissions process as the input to the system, the educational process as the throughput, and graduation as the output. Alumni donations then provide a feedback. A system can have more than one purpose, just as it can have more than one input. Grant money and tuition are additional ones.
Don't make the mistake of reifying the model. Assigning a purpose to a man made system is one way of analyzing it, but a university doesn't really have a simple coherent set of "purposes" that you could put on a comprehensive list.
If we say "A system is what it does" does that make more sense?
I'd suggest just a pure lottery from the qualified individuals. You can't manufacturer signal from noise. All that essays do is import the biases of whoever is reading them, whatever way those point. Which is even more unfair and obscure than honest randomness, as well as way more of a waste of everyone's time. Plus making people cynical about the process, which is inherently corrosive.
Yes, that would make more sense. Which is why I believe that the admissions process is not intended to be fair, neutral, or even find the best applicants.
What it appears to be doing is finding the applicants that the particular school would like to have there, while giving plausible deniability about the reasons any particular student was rejected. Legacy applicants and the children of rich and connected people are pretty obvious. You can also select for demographic or social reasons, or pretty much any reason you want.
"Hardship" essays, I'm pretty sure, were intended to pull in more minority, especially black, applicants. It was expected that more minorities would have hardships than white people, so that would give them a leg up.
I suppose. I was shocked how virtually everyone on the admissions essay thread, including Scott, bought into the idea that ya gotta play the game. I really do find it hard to believe that if some kid wrote an intense, sparkling, truthful essay about, say, learning to paint with watercolors -- how it taught them about patience, and excellence, and precision, and using happy accidents -- they would make a bad impression.
Saying "you have to play the game" is just good pragmatic advice. We could pine for a better system, but if we're not in control of admissions at the relevant schools, we can't actually change the rules.
If I were giving advice to a high school student about getting into college, I certainly wouldn't tell them to ignore the whims of the application reviewers, but help them game that system as much as they can.
But, I might also encourage them to actually develop skills and maybe go to a local college where a 1100 SAT score will get you in automatically. There's nothing wrong with working their way up in a field instead of trying to start near the top with a Yale degree or whatever.
Which I'd firmly place in the we shouldn't be doing that bin, personally. Teaching people that it's all arbitrary nonsense and that only by cheating the system can you survive is one of the lessons taught by the Soviet system, and leads to low trust, high tolerance for corruption societies.
Just like leaving broken glass lying around selects for people willing to walk around it, especially if they can't clean it up. Which leads to inefficiency and worse, lacerations.
Oh, agreed. But POSIWID - the knowledge that the first, automatic response of the person bearing the piece of paper your institution produces when you ask them to jump will be "how high?" is valuable to a significant enough proportion of the forces behind the system that you need to plan for considerable resistance if you attempt to change it.
As someone who has done a lot of hiring, it's amazing how cookie-cutter many of the college-educated applicants we get are. Hiring a Custodian means you meet all kinds of average people with various backgrounds. Hiring [college educated position with industry-specific requirements] and it's like you're talking to the same three people over and over again. We can definitely count on them to yell "How high?" when we tell them to jump, though. That's pretty valuable to us, even if we could totally survive without it.
Agreed. There are lots of forces and entrenched interests here. Not the least are all the admissions folks who would lose influence/jobs if they weren't needed to gatekeep, even if the gate shouldn't exist in the first place.
Better yet, get the Dean of Admissions (or whatever his title is) to spend the night fasting and praying for divine guidance, and then randomly pick out one-tenth of the applicants to accept. God wills it!
On college essays I find them similar to many interview processes in that they are arbitrary specifically because there needs to be some way of filtering down candidates. Let's say the acceptance rate of some college is 5% (most top tier colleges are around or below this number). After eliminating all the people that clearly aren't qualified - low SAT scores, bad grades, etc etc - they still have to find some way to cut down the applicants since they can only accept so many. You could argue a lottery system or random chance would be fairer, and in some ways I agree, but I think most applicants would not like it as it takes away what control they have over their chances. So the colleges use essays instead, and while those aren't exactly merit-based I'd still argue they test valuable skills: writing skills, ability to research/know your audience and what they are looking for, etc.
The same is true for interviews/job applications - there are going to be more than enough qualified candidates, and the process of sorting them is always going to be somewhat arbitrary by nature. For a programming job you could sort candidates by how quickly they complete a series of leetcode questions, but that tests speed and memorization - people who do well on that test may struggle when it comes time to develop an original solution to a novel problem.
TLDR; Tests are always somewhat arbitrary and since the things colleges/jobs/etc are looking for cannot be tested for directly, they look for results that are correlated instead
I'd think that the rational way of choosing between equally qualified candidates would be to simply sort by price? That's how I usually decide what to buy or whom to sell to.
Colleges charge everyone the same tuition, so they can't really do that. I suppose you could implement some system where the admin office charges a variable amount of tuition, but I'm not sure how that would function in practice, and it would require more business acumen out of your application-sorting minions. (I think The Name of the Wind had a college that worked like this?)
Price discovery is not exactly frictionless either - if you ask job applicants what salary they expect, they'll probably answer somewhere around the "going rate" for the job and not give you enough variation to discriminate. You'd have to take the time to negotiate them downward individually.
Basically, what happens if you look at the cheapest price available and still have more than one option?
I think the simplest way would be to raise tuition every year until the school reaches maximum profitability, at which point they should have the capacity to enroll all qualified candidates. That's just like any other business that doesn't price discriminate.
I read somewhere — maybe in one of the gcbias.org's blog posts (?) — that any GWAS study done before 2017 is probably suspect. I didn't follow up on this, but I recently read these two posts by Sasha Gusev and Eric Turkheimer...
Both state that population stratification was identified as a confounding factor in earlier GWAS studies. In other words, humans don't mate randomly, and our genomes are probably not the result of random mating.
The Gusev and Turkheimer posts were both prompted by the 2024 Tan, et al. paper, "Family-GWAS reveals effects of environment and mating on genetic associations".
Educational attainment, once recalculated for population stratification, comes out as 0.50 confounded by non-genomic factors.
Cognitive performance is ~0.65 confounded.
Household income is ~0.70 confounded.
And ADHD is the most confounded at ~0.77!
It seems like the study that kicked this off was the 2017 paper, "Human demographic history impacts genetic risk prediction across diverse populations" by Martin et al. —
Martin et al. *may* have followed this up with "Significant impacts of population stratification on genetic association studies in diverse populations" in 2019. ChatGPT references it as being published in Nature Genetics — but I can't find this paper in Google Scholar. I suspect it may be an LLM hallucination.
There's one thing I don't understand about Gusev and Gregg Gibson and all those guys advocating for this view, that perhaps somebody else here would know:
My last major genetic mental update was Polderman 2015, the twin study meta-analysis basically showing that everything important is 60-80% heritable in developed countries.
So now GWAS's and family GWAS's and all sorts of more refined and informative takes are coming out that have identified a "heritability gap" between what we can measure genetically and what twin studies tell us, and which argue "nope, that's wrong, environment is way bigger and more important than we thought." And this is Sasha Gusev's and your linked posts.
But what I don't understand is: are there any actionable changes in important life areas stemming from the more refined genetic take?
The two biggest areas people care about individually where this knowledge is used are probably mate search and how we raise our kids.
So for mate search, the "dark side" genetic interpretation would be "genes are all that matter, and this includes more than their immediate genes (per Greg Clark), so optimize direct mate quality AND lineage quality to the extent you can."
And I'm not really seeing any different recommendation from what we know from family GWAS's? It seems to support the exact same conclusion, possibly even more strongly (because family culture seems important and to make up most of the heritability gap)?
And for raising our kids, it's basically "make sure they're raised in a two parent household with good mores and no physical or mental abuse," and that's the tippy-top best you can do. Everything else is noise, in the sense that everything else that matters was handled at the "mate selection" stage. You married into whatever level of educational attainment and average income and wealth you're going to be drawing from, and now it's just random draws from you and your spouse's conjoined lineage baseline. All environmental factors that we know matter are negative - malnutrition or physical or mental abuse all bring your kids down, but we know of nothing that brings them up.
Has the additional info we know from family GWAS's changed this at all? Here I think maybe there's some room if family culture and environment actually matter a lot more than we thought? Have we actually identified anything actionable here? Or is what I described still the best we can do? If that is the case, are these environmental / culture differences being rigorously studied anywhere?
Really appreciate any insight anyone might have here.
> My last major genetic mental update was Polderman 2015, the twin study meta-analysis basically showing that everything important is 60-80% heritable in developed countries.
You make a perfectly valid point that this was the consensus view back in 2015. But if I understand Tan et al. correctly the twin studies give you a false sense of heritability. You've got a bunch of IGEs and other kruft that masquerades as DGEs. See my response to Eremalalos below. I wonder if Polderman will update his priors and provide us with a new meta? I somehow doubt it, because Bayesians tend to hug their priors close to their hearts. Well, that tends to be true for everyone, including myself, so I shouldn't be snarky.
> You married into whatever level of educational attainment and average income and wealth you're going to be drawing from, and now it's just random draws from you and your spouse's conjoined lineage baseline
Yup.
> ...are these environmental / culture differences being rigorously studied anywhere?
Yes, but I admit a lot of these studies have a squishy feel to them. But if I were raising a kid, there's a shitload of environmental stuff I'd do to improve the life outcomes. For instance, get them started on a foreign language while they're still very young (starting Kindergarten if you can). And if you're from an Indo-European language group, send them to a Chinese academy on weekends (I've become a neo-Whorfian, and I believe that languages do shape the way we think). Force them to learn a musical instrument. Play word games with them. Solve puzzles with them. Books! Of course, everything is online now, but its easy to get distracted by other stuff when you're online. Books are a focused media. My folks bought me just about all the Time-Life series when I was a kid, and I devoured them. And after reading Heinlein's _Beyond This Horizon_, when I was twelve, I wanted to grow up to be an Encyclopedic Synthesist.
> But if I understand Tan et al. correctly the twin studies give you a false sense of heritability. You've got a bunch of IGEs and other kruft that masquerades as DGEs.
Yeah, I follow their arguments overall, I think. Basically, population stratification means because we tend to study WEIRD people for everything, we're in a region where everyone is drawing from a more similar baseline than true world diversity, and so overestimate polygenic scores for various endpoints.
And then because we all assortatively mate really strongly and this has been going on for a long time, we're drawing spouses who are similar enough on multiple traits those traits have become correlated with each other over time, so you're getting more than the straight ~50% expected allele draw due to linkage, and this skews heritability estimates when looking at siblings and dz twins. Additionally, because we're similar culturally as well as on correlated endpoints, high education people impart high education cultures to their kids, etc.
So the answer is family GWAS's - because this controls for the same grandparents, all the background stuff like population stratification and assortative mating is filtered out, and you get "true" polygenic scores for your endpoints, which are much lower than the Polderman dz / mz scores.
But to my mind, it literally doesn't change anything, and assuming full genetic determinism still gets you the exact same answers.
I appreciate and agree with all your early childhood interventions - but to my mind, it's nothing new at all. That's just raising your kids in default educated Westerner culture.
This is why I specifically framed it in terms of "where would this ground out in decisions or actions," because I'm kind of skeptical.
Heritability gaps seem to keep getting smaller with larger and more diverse GWAS's, and most of the people making these arguments seem very entrenched in a certain mindset that seems ideologically pre-committed to an "anti genetic determinism" worldview, so I just can't tell how much is noise / slanted towards their desired outcome versus useful knowledge that can inform our actions and decisions, either individually or societally.
For instance, if culture / environment REALLY matters ~4x more than we think from Polderman, why didn't it come out in any of the much older separated at birth Minnesota twin studies, or the many studies where disadvantaged kids are raised by middle / upper class people in much better environments?
It's not like we haven't thought about or controlled for "culture / environment" before - and I've asked this before and didn't get a good answer.
I have the sneaking fear / suspicion that all of this is quibbling about how many angels can dance on the heads of pins, but as long as you're not going to abuse your kids or make them eat lead paint chips, if you still think and act like everything is ~80% genetics and it's all determined, you're going to still make all the right decisions.
Are you looking forward to participating in the nappy/diaper changings and late-night feedings? I doubt if Elon has done this with his twelve known children (and now he's upset that his eldest male child transitioned). If you're not going to help out with the child-rearing, I sure hope you've got a lot of surplus income to divert to child support!
Weird anecdote: I have a friend who has ten half-brothers and sisters (that she knows of). They were all from the same town, and they all went to the same high school. They eldests started keeping an informal genealogy so the younger half-siblings didn't accidentally end up dating and having sex with each other. Their father probably sowed his seed in other places, but they could only compare notes locally.
> If you're not going to help out with the child-rearing, I sure hope you've got a lot of surplus income to divert to child support!
Literally my entire pitch revolves around "maids, nannies, cooks, and surrogates," because nobody worth marrying wants to wreck their perfect body with 6 pregnancies or spend all their time on drudge work.
It's like a night and day difference, completely 180's people. Before these 4 magic words, it’s all:
Her: “WHAT?? 6 kids?? Jesus, you’re crazy! No way! Hang on, wait, wait, I’ve got to fish my taser out of my purse….<rummaging>”
Me: “Cooks, maids, nannies, surrogates.”
Her: “What? What did you say?” <pulls out taser, smiles crazily as she pops it on and off a few times>
Me: “Cooks, maids, nannies, surrogates. I don’t want to ruin your body any more than you do! Kids are a big thing, everyone needs some help! Screw cooking and cleaning! Let’s do it right!”
Her: “Oh. Hmm.”
<thinks a little, talk a little, eventually…>
Her: “Holy shit, you’re right! Ok, let’s go. Let’s do it. C’mon, what are you waiting for?? Marry me already!”
Any applications towards our national-level fertility crises seems sadly inapplicable, because these aren't solutions that necessarily scale well.
But to your anecdote, I actually grew up with 9 siblings, and multiple aunts and uncles all had 6-8 kids, and I doubt I'd be shooting so high if I didn't have a childhood experience like that myself.
beowulf what do you mean in the second half of your posts by when you say something is confounded by a certain amount? I'm not arguing, just don't understand. So if, for ex., ADHD is confounded at 0.77, do you mean that some measure of population stratification correlates at .77 with an ADHD measure? Also, what measure is used for the genomic factor, in studies that made no attempt to correct for population stratification? Is it also a correlation coefficient?
Yeah, I didn't explain that very well, did I? Basically, that means only 23% of ADHD expression can be tied to genes. Environmental factors (for instance, chemical, social, or perhaps diagnostic) swamp the genetic signal. The .77 would be a negative correlation coefficient. ;-) And I should have written: "Educational attainment, once recalculated for *Direct Genetic Effects*, comes out as 0.50 confounded by non-genomic factors. Population stratification by assortative mating may only be one of the confounding factors. My bad. I was typing faster than I was thinking.
Tan et al. actually don't use the GWAS data. They tie their estimates to the Direct Genetic Effects. They explain the methodology in this paragraph. I had to read it over a few times before it began to sink in.
> Family-based GWAS (FGWAS) has been proposed as a solution to the problem of confounding in GWAS that also enables deeper investigation of the impact of gene-environment correlation — including from IGEs — and non-random mating, including population structure and assortative mating. FGWAS adds the parents’ genotypes to the regressions performed in GWAS (Methods), thereby using the natural experiment of random assignment of genotype within-family — due to Mendelian segregation during meiosis — to estimate ‘direct genetic effects’ (DGEs). *Because the segregation of chromosomes during meiosis is independent of environment, estimates of DGEs are free from confounding due to gene-environment correlation, including from IGEs and population-stratification. Because different chromosomes segregate independently during meiosis, DGEs are also free from the confounding that arises in GWAS due to correlations with genetic variants on other chromosomes, which are caused by non-random mating* including population structure and assortative mating (AM). As in GWAS, variants are analyzed one at a time, so DGEs include causal effects of the focal variant and correlated variants on the same chromosome.
And an IGE would be an Indirect Genetic Effect. ChatGPT gave me a good example: "Example: A parent has gene variants associated with high educational attainment → they create a cognitively stimulating home → that environment influences your development (regardless of whether you inherited those same genes)."
Princess went AWOL yesterday. She's around my neighborhood, but hides under cars when I get too close. And she won't come out for her favorite treats — she got wise to that trick.
Ooh, that’s a real problem. Maybe use a trap with food in it? There’s a kind that organizations that trap, neuter and release use. They might even lend you one.
This is one of those areas where I wish I was an insider who knew all the "behind the scenes" stuff. I've seen some comments online from geneticists saying that a lot of this is "cope" (grasping at straws) that's trying to avoid a genetic determinist view of the world, and Turkheimer feels somewhat guilty for earlier work (including some of his own) suggesting that genetics are a major determinant of life outcomes.
But who knows, are those internet commenters the ones who are high on copium or is it these behavioral geneticists? In my own field (non-biology STEM) I've seen similar grasping-at-straws attempts, sometimes by senior researchers and sometimes by young upstarts, to try to galaxy-brain their way out of an inconvenient or unfortunate finding, but you have to go to conferences, chat with people at the coffee table, etc., to have a feel for the "vibe" of this kind of thing. As in, is this really just cope by an aging figurehead trying to keep their ideas relevant? Or are the problems they're highlighting actually a serious problem in the field?
An example of the latter might be John Ioannidis' work in the early days -- many people probably wrote it off as cope but he was right: there were major problems with medical research that invalidated many of the findings people took to be true.
IIRC, John Ioannidis was mostly concerned with sloppy or fraudulent science like p-hacking and sample biases (Please correct me if I'm wrong).
There was nothing sloppy or fraudulent (AFAIK) about the pre-2017 GWAS studies. They just ran those studies under the false assumption that European allele frequencies between individuals — with and without a trait or disease — matched the profiles of other non-Euro populations. But the Martin et al. paper that I linked to above showed that assumption to be false. ChatGPT insists that they followed up with a paper soon after that showed that population stratification (i.e. people preferentially mating within similar groups) was the real culprit. But I can't find the damn thing! ChatGPT, CoPilot, and Grok all agree that Martin's team was the first to demonstrate that population stratification as a major confounder. But Grok says this idea goes back to Lander and Schork in 1994 (!) — it was just that no one had bothered to investigate their claims.
As for the cope-side of the argument, I'd like to see some serious studies by geneticists that argue against the above papers. If you or someone else can send me links for them, I'd appreciate it (although I admit I'm fairly convinced that Tan is correct). Probably the Tan et al. paper is too new to have garnered any responses yet. But it seems to me that most of the copium comes from social media commentators who have latched on to the whole genetic determinism view hook, line, and sinker. Certainly, if you had read the popular summaries of the pre-2017 literature, that would be a fairly reasonable view.
Ioannidis was/is also concerned with bad research methods (which do overlap to some extent with p-hacking and the "garden of forking paths") - he famously called for the entire field of nutritional epidemiology to be more or less thrown out (https://jamanetwork.com/journals/jama/article-abstract/2698337) because its reliance on various surveys, food inventories, etc., produced implausible effects that often did not hold up in RCTs.
Like I said, I'm not a genetics expert so I'd love to hear from an insider who has the nose to sniff out the vibes / inside scoop on the Tan issue. Agreed re: propensity for copium from social media commentators who have a vested interest in genetic determinism.
Can I ask for a "for dummies" tl;dr? For example, on ADHD, what was stated to be true and is now considered false when taking population stratification into account?
I screwed up while writing that post. ADHD recalculated for Direct Genetic Effects comes out as 0.77, confounded by non-genetic factors. Population stratification (by assortative mating) may only be one of the confounding factors. Environmental factors or diagnostic issues may also account for a weak genetic signal. My bad. I was typing faster than I was thinking.
If you're specifically interested in ADHD, there's a great article in the NYTimes. I think this link will get you past the paywall.
It glosses over the possible genetic component for ADHD, but seems to support the Tan et al. thesis that ADHD is not very heritable...
>...for a significant percentage of people diagnosed with A.D.H.D., Nigg says, “there’s nothing neurobiologically notable about them. Instead, their symptoms are situational or conditional. They may have had a hard life, or they have a lack of social support, or they’re in the wrong niche in life.”
I have a suggestion for trying to deescalate lawfare:
For any indictment of a national political figure, require that a simple majority of Democratic Party congresspeople _and_ a simple majority of Republican Party congresspeople agree to the prosecution before it moves forward. If there is a possibility of retribution, do the voting on this with secret ballots.
If a political figure did something _truly_ egregious, and the legal process _isn't_ being used just as a partisan weapon, those majorities shouldn't be too hard to get.
Yes, I know this would take a constitutional amendment. Yes, I know the founders were not fond of political parties, and intentionally excluded them from the Constitution. Consider it a policy proposal, with enactment issues deferred.
Any, yes, "national political figure" is a fuzzy term, and would need a precise definition.
And this isn't intended to cover _all_ lawfare, but maybe enough to make at least the pretense of even-handedness sort-of credible. It would be nice to have the legal process look like something other than a blatantly partisan weapon.
What do people think? For anyone who doesn't like this, how would you try to restore some legitimacy to the legal process?
edit:
Many Thanks to Erica Rall and TotallyHuman for their helpful alternatives
The legal process is not illegitimate, but it only works if you're willing to actually prosecute people for their crimes. The next Democrat President should, on day one, launch criminal investigations into the DoJ and Department of State and indict the respective Cabinet members of Trump's administration responsible for deporting people without due process. They should revoke Fox News's broadcast license for peddling 2020 election fraud lies and for not immediately denouncing Trump signing orders weaponizing the DoJ against the law company that won Dominion their defamation suit against Fox News. The reason Republicans lie with impunity to the extent they live in their own alternate reality is because Democrats are feckless and aren't willing to punish them - despite Republicans literally lying about them doing just that, they're still too stupid to fight back.
Many Thanks for your comment! So you want to double down on partisan political prosecutions. And you want to unconstitutionally violate freedom of the press. Between people like you, and what Trump is doing along the same lines, damned few people are going to see the legal process as anything but a partisan political weapon. You are part of the problem.
There is mountains of evidence of criminality that Republicans are engaging in. They need to at a minimum be held accountable for it and appropriately jailed. Then there should be actual government coercion against bad faith Republican institutions that are supporting Trump and his admin in criminally undermining American democracy, elections, checks and balances, and Constitutional rights like due process.
Any attempt by Democrats to censor Republicans on partisan grounds should be opposed - only actions that Republicans are taking to destroy America like ignoring due process, lying in favor of ignoring due process, lying about elections being stolen and aiding those who lie, and other similar extreme behavior should be prosecuted.
This is the moral thing to do because Republicans have only gotten more unhinged as they've ALREADY done these things for partisan political purposes like falsifying electoral votes to overturn our system of governance and trying to impeach judges because they are opposing Trump trying to run the government like a dictatorship when there is no punishment for them. They've also been lying that Democrats have already been doing this, so Democrats are taking the PR hit while not actually getting anything done. The messaging here is also very straightforward and actionable - "Stop lying and begin living in reality again and we won't target you for trying to destroy our shared system of governance. Hating trans people does nothing to the country but any obvious and intentional lie about election fraud to try to overturn election results will get you jailed and anyone working with you." Anything less than this will just inspire Republicans to keep destroying the country.
>For anyone who doesn't like this, how would you try to restore some legitimacy to the legal process?
I think it already has plenty of legitimacy, and all you have to do is make the court transcript public. It only looks bad because nobody bothers to look into the specifics, which is not going to be fixed by anything, ever.
> If a political figure did something _truly_ egregious, and the legal process _isn't_ being used just as a partisan weapon, those majorities shouldn't be too hard to get.
Republicans can't even get a majority together to stop Trump from crashing the economy for no reason, something with very obvious immediate impacts on their constituents. Trump has centralized power to an unprecedented degree, and no Rebpulican in congress dares to defy him.
Nice idea but it collapses when you think about what the adversarial equilibrium would be: either party could block all prosecution of their members no matter what they did. There's a reason we have a separation of powers and it's to prevent things exactly like this.
IMO the right structural solution to our current political polarization is the one the Founders already put in and we've managed to evade: only the elite should vote, for some reasonable definition of elite. If it was up to me I'd limit voting to those who were above the 60th percentile in Federal income tax paid. We were founded as a republic and for good reason. The Founders were well aware that when you give the hoi polloi a voice then chaos inevitably ensues. The Simpsons said it best: Democracy doesn't work. Democracy is the voice of the people and most people are stupid. It's insane to let them influence anything.
> If it was up to me I'd limit voting to those who were above the 60th percentile in Federal income tax paid.
Just to be clear, you don't even hit "break even" as a net contributor vs consumer of taxes until you're roughly top 20% / $85k in income, and the top 1% pays 40% of taxes, so you're proposing something like "only the top ~3-5% by income get to vote?"
And is eliminating old people as a voting block, because they're not paying taxes in a given year, a bug or a feature?
Either way, with some amount of job automation looming on the horizon, probably a bad idea, because if only the top 3-5% can vote, capital owners can automate all the jobs without UBI-ing people, and I don't think the killbots are quite at the "retreat behind our walled compounds with impunity" levels yet, so it'd get pretty messy.
>you don't even hit "break even" as a net contributor vs consumer of taxes until you're roughly top 20% / $85k in income,
Oh is that right? I didn't know that. I just pulled that 60% figure out of the air, though keep in mind that some large % of people pay zero taxes. When I said top 60% I meant the top 60% *among those who pay nonzero tax* so maybe that winds up being the same thing. Obviously you'd have to do a careful analysis if you wanted to actually implement it, but any steps in this direction would be positive in my view. Obviously there's zero chance this would ever happen so it's just an academic argument.
>And is eliminating old people as a voting block, because they're not paying taxes in a given year, a bug or a feature?
Probably a feature. Plenty of retired people have investment income that they live off of and social security payments are taxed so not all old people would be disenfranchised. You could also have the threshold be related to lifetime aggregate taxes paid or something.
Many Thanks! I'm somewhat hopeful that the equilibrium could be: Clear, egregious crimes get prosecuted, politically motivated prosecutions get stopped. But this is just a hope, not an expectation. Regrettably, in the current situation, a sane response to "<politician> broke the law." is generally "Yup, and the sky is blue.". It would be nice to pare the legal code back to something sane, but if that could be made to happen at all it would probably take a century.
>IMO the right structural solution to our current political polarization is the one the Founders already put in and we've managed to evade: only the elite should vote, for some reasonable definition of elite.
I think that comment is directionally correct but goes too far. It's not that 98% of Americans are unprincipled buffoons, it's that 51% of them are and that's all it takes. The central issue is that there is a maximum gap that society can tolerate between how clever your propagandists are and how clever the average voter is. Once that maximum is exceeded then there becomes a voting block that candidates can reliably manipulate like a cat with a laser pointer. Elite-only voting reduces the population of low-IQ voters to a manageable size.
Propaganda only works on stupid people. Those people shouldn't be allowed to vote, otherwise politics becomes nothing but a propaganda contest. I disagree with those who claim that it's always a propaganda contest. There are lots of smart, educated, high-character people out there with good judgement. Our political system should be designed such that those people have an outsized influence. An elitist voting system would increase the signal-to-noise ratio and restore sanity back to governance.
Sadly it's now essentially impossible for this to happen. There are too many congressmen who only have their seat because of dumb voters. They will never vote for a system which would make them unelectable. This is how entropy always wins in the end. We dropped our guard for a couple decades and now we're overrun by morons like AOC and Taylor-Greene. It's impossible to put Humpty Dumpty back together again. The American Dream is dead and in 30 years (maybe sooner) we will be essentially indistinguishable from Brazil or Argentina.
I don't think that propaganda only works on stupid people. But that's because I have a long-running hypothesis that most people are smart; they're just not smart all the time. Namely, they're smart about a few things (often whatever they do for a living), and are anywhere from slightly to much less smart about other things, depending on mental distance and possibly their own brain chemistry.
Propaganda works to the extent it does because it's operating in one topic that most people aren't experts in. This is often something political like foreign policy, economics, or legal philosophy, but one could just as conceivably manufacture propaganda about anything, from food really being a drug forced on us by rich "agricultural" conglomerates to E not really being mc^2. (I know a current PhD in atmospheric chemistry who believed 1 + 1 isn't always 2.)
>Namely, they're smart about a few things (often whatever they do for a living), and are anywhere from slightly to much less smart about other things, depending on mental distance and possibly their own brain chemistry.
Could be. What you describe sounds very much like Robin Hanson's "near mode" vs "far mode" distinction.
>Once that maximum is exceeded then there becomes a voting block that candidates can reliably manipulate like a cat with a laser pointer.
Love that image!
>Propaganda only works on stupid people.
Regrettably, _busy_ people also have a problem finding the time to drill down to primary sources. And, even then, have to play the guessing game of "Who is lying more this time?"
Busy non-stupid people could outsource their decision-making e.g. to their trusted bloggers.
At first sight that doesn't seem fundamentally different from stupid people outsourcing their decision-making to TV, but I think the quantitative differences would make a big change when taken together -- it is cheaper to start a blog than to start your own TV, comments on blogs are more visible, it is easy to write a reply elsewhere and link to the original blog, etc.
Many Thanks! Mostly agreed. We _have_ to delegate to experts in many fields routinely. Regrettably, politics, particularly in the current USA tribal form, is a subject where it is particularly difficult to determine who to trust.
Keep in mind that the levels of lying would likely go down if there was less of an audience for it. Plus how much effort does it take to realize that Trump and Kamala are both terrible people? No system is perfect but I have no doubt that this would immediately fix 80% of our political problems.
>Plus how much effort does it take to realize that Trump and Kamala are both terrible people?
I suspect that probably even a majority of the non-elites realized that - but 3rd parties are _really_ hard to get elected, so knowing that we were choosing between cyanide and strychnine didn't help much.
So the solution is for politicians to be even more above the law than they already are? You can do any illegal thing you want as long as you stay popular with your team? Frankly, I can't think of a faster way to sink confidence in the political process (not that it's high right now to begin with).
Many Thanks! I think that it is better than having the farce of e.g. the Stormy Daniels case. If a national political figure did something egregious, yeah, their own team should be willing to call them on it. If their own team _isn't_ willing to call them on it - Maybe the prosecution is more political bullshit? Using the legal system as a partisan baseball bat to bash someone with? Again?
The trouble I see is the assumption that political figures want to do the morally correct thing but are sometimes hampered by politics. I think, for political figures, it is often the other way around.
So a solution needs to somehow compel morality in its rules, which are difficult to impossible to skirt. Things like secret ballots create a situation the opposite of that.
It's still good to try to think of good solutions, but it isn't an easy thing to think of something better than the current system, bad as it is.
>The trouble I see is the assumption that political figures want to do the morally correct thing but are sometimes hampered by politics.
I'm not making that assumption. I'm not making any claims about morality at all.
I'm just looking for a way to cut down on the lawfare, for the parties to be less able to use the legal process as a baseball bat to bash each other over the head with - particularly in farcical cases, like the Stormy Daniels one.
Currently, there is no way to put a stop to partisan bullshit lawfare and to move the conflict between the political factions back to the realm of political negotiations, where it belongs.
You have referred to the Stormy Daniels case as farcical several times. Could you expand on why you feel this is the case?
Just plain looking at the charges, the laws, and the facts found by the court supports the convictions. The sentence appears to be just plain correct. Donald Trump committed the crime, and got found guilty, as appropriate, in a court of law by a jury of his peers.
I replied to this at more length elsewhere, but I'll just say again that I find it *utterly baffling* that you take it as a given that one of the richest and most powerful men in the country should just be able to violate the law when it suits him. Why, exactly, is it "political bullshit" for him to have to follow the same laws as everyone else?
Nope. _When the prosecutor cannot persuade a simple majority of both parties to let the prosecution proceed._
My _guess_ , in Trump's case, is that this would have quashed the Stormy Daniels farce but let the classified documents case go forward. But that's just my opinion. I could be wrong.
Re:
>have to follow the same laws as everyone else
a) blank is right. Somewhat similar (albeit with some differences) cases were not pursued in Hillary Clinton's (email server issue) and Biden's (classified documents - but with extenuating circumstances) cases
b) More generally: There are so damned many laws, with such broad wording and the potential to be twisted in so many ways, that nearly anyone is in violation of _something_ .
"Nope. _When the prosecutor cannot persuade a simple majority of both parties to let the prosecution proceed._"
Right. So whenever it suits him. Anyone with who hasn't slept through the past 8 years has noticed that Republicans as a unified have never and will never go against Trump on *anything.* He literally incited a goddamn insurrection against the country he's leading, and Republicans do nothing but make excuses. Why on Earth should anybody who isn't a Republican trust them to be the *slightest bit impartial* when it comes to evaluating the misdeeds of the head of their cult of personality? Proposing turning the law into a political popularity contest is ABSURD.
"blank is right."
Blank is full of it and you should have the basic honesty and critical thinking skills to recognize this. First and foremost, the Stormy Daniels case was *nothing whatsoever* like the various classified documents cases. I see you've already been called out on this below, so I'm not going to get too far into the weeds but a big, clear, *obvious* difference is *intent.* It's very difficult to prove somebody's intent when they're merely doing something careless or reckless. But when they make deliberate changes to their own records to hide wrongdoing, intent is plain as day.
But more to the point Clinton especially was probed and investigated and called into hearings *repeatedly*. For the classified documents and even moreso for Bengazi. For your claim to have merit, it would need to be true that Clinton had clearly, demonstrably broken the laws, and that the people in charge of enforcing them had knowingly and deliberately let her off the hook for doing so. That's the exact opposite of what appeared to happen: her political opponents were *extraordinarily determined* to find her guilty of something, and simply failed to do so[1]. Claiming that there's a double standard because the state meets its burden of proof in some cases and fails to meet it in other cases is absurd, man. Please do better.
[1] Though they arguably did enough damage just in the attempt to sink her presidential campaign.
> It's very difficult to prove somebody's intent when they're merely doing something careless or reckless. But when they make deliberate changes to their own records to hide wrongdoing, intent is plain as day.
I presume you'r referring to the time when Clinton's furry tech admin asked Reddit how to scrub names and timestamps from her Benghazi e-mails?
No, as I already explained. Factions will be factions. If you _can't_ persuade a simple majority of a faction's congresspeople that a prosecution has merit, I'm taking that as sufficient evidence that this is a politically motivated prosecution, and should not go forward. Maybe the opposite faction managed to dig up a statute that the defendant can be construed as having violated. So what? If you can't persuade a simple majority of the defendant's party that the case is worth pursuing, all that you have proved is that we have too many laws.
The email server issue is potentially much more serious (and almost identical to the Hegseth on Signal case), but it is much harder to prosecute.
The Stormy Daniels case is a cookie cutter falsification of business records case. If none of the people involved were celebrities, it would just be something that gets successfully prosecuted pretty much monthly in New York. Anyone else who did what he did would also get prosecuted. The only difference is that because he had won an election before the prosecution finished, he didn't end up going to prison, like anyone else would have.
Can you point to me anywhere in the founding documents of the U.S.--anywhere at all--that says openly or implicitly that holders of political office ought to be above the law? Call me crazy, but I seem to remember that anger and dissatisfaction over having a powerful, unaccountable head of government was one of the *motivating reasons* for the American Revolution, and that the framers of the constitution were *quite clear* that they wanted to head off that failure mode for their new republic.
In this case, I think the founders of the US were full of shit, because their supposed grievances at the King of England for being above the law made little sense because everything he did was also agreed on by Parliament.
I also think the American idea of subordinating public office to the law made more sense when the average voter was educated in the law and owned property.
"I also think the American idea of subordinating public office to the law made more sense when the average voter was educated in the law and owned property."
I think this is one of those appeals to a non-existent history. Even at the founding of the nation property ownership was a prerequisite in some states and not in others: and in the former it doesn't seem to have lasted very long. It was clearly not considered important or default by the framers. As for the average voter being "educated in the law," this seems...fanciful at best. Given broad trends around urbanization, I'd wager good money that the "average" (in the sense of "modal") U.S. voter in first few decades of its existence was a farmer. Heck, I'll bet this still holds true even among the states with property requirements. Unless your standard of "educated in the law" is low enough that anyone who's taken high-school law or civics classes will meet it, I'd be skeptical that your typical frontier farmer did either.
>I'd wager good money that the "average" (in the sense of "modal") U.S. voter in first few decades of its existence was a farmer.
Yes.
<aiWarning>
Google responds to: when did the fraction of the us population that were farmers drop below 50%
with
AI Overview
The fraction of the U.S. population that were farmers dropped below 50% between 1860 and 1910. In 1860, 53% of the labor force was in agriculture, while by 1910, that number had fallen to 31%
Yeah, the British monarchy, whilst not yet quite a figurehead, was very much playing second fiddle to Parliament by this time. George Washington had more power as President than George III ever did as King. (And of course, Donald Trump has far more power than either of them.)
Cutting off malicious prosecutions relatively early is supposed to be what Grand Juries are for, and it is a known problem that Grand Juries currently aren't doing a great job of this in general, not just for major political figures. I would prefer a solution based around fixing the more general problem here.
Off the top of my head, possibilities for this include raising the standard of evidence for indictment from "probable cause" to "clear and convincing evidence", giving the jury a mandate to decline indictments that are not in the interests of justice, requiring a (larger?) supermajority verdict, or including a public defender in the process to call exculpatory evidence to the jurors' attention.
Of these, I think a supermajority requirement would be most effective at limiting lawfare. Grand Juries usually have little or no voir dire and are typically larger than trial juries, so are likely to be close to representative of the electorate in the jurisdiction in question. A nakedly partisan malicious prosecution would then likely need to persuade at least a handful of jurors of the same party of the defendant to meet the supermajority requirement.
Many Thanks! Yes, those all sound like reasonable possibilities.
>A nakedly partisan malicious prosecution would then likely need to persuade at least a handful of jurors of the same party of the defendant to meet the supermajority requirement.
Maybe include an _explicit_ requirement that the Grand Jury have enough jurors of the same party as the defendant to ensure that the supermajority cannot be met without convincing at least some of those same-party jurors?
I think a precondition for deescalating lawfare is for people to actually want to deescalate lawfare. People love lawfare, it feels awesome. At least half of the Trump era has been devoted to liberals demanding and devouring lawfare against Trump. Nothing in news has been nearly as profitable as lawfare against Trump and several people have launched their careers (kinda) off prosecuting Trump for whatever thing didn't pan out in court except that one thing with the pornstar which no one understands but we all got to spend a couple months pretending we we're talking about Trump f***ing a porn star with secret lawyer payoffs for the good of the country, rather than our own purient interests.
And, to be fair (1), the right wing absolutely indulged in the same stuff with Hillary and delights in "liberal tears" with all the depravity you would expect. But, just observably, like 10 people are actually sincerely concerned about lawfare degrading our legal norms and the others 329,999,990 people in America are like "This is my favorite TV show! Moar please. My own social, career, and romantic failings are incredibly depressing and the only thing that gives my life any meaning in our soulless nihilistic consumer culture is to artificially divide ourselves in half and then try to murder the other half. Hail Khorne and pile skulls on the skull throne just as long as I can feel something again."
And as long as liberals can convince themselves that once they finally murder Tiberius Gracchi, sorry, Donald Trump, then all of this will be over and their outgroup will be ground under their heel forever. Definitely.
Meanwhile, conservative/reactionaries here are like "Once we install a supreme absolute dictator who will brutally murder everyone we dislike, we can return America to the 1950s. It definitely worked for Sulla and it will definitely work for us."
But yeah, if we had a public vote on "return to peaceful democratic/peaceful norms" or "punish outgroup so we can luxuriate in their tears", punish outgroup is winning 98%-lizardman constant.
So, practically, are you aware that Puerto Vallarta is a gorgeous tropical paradise in Mexico with nonstop flights to most major US metro areas so you can visit family and friends? (2)
> So, practically, are you aware that Puerto Vallarta is a gorgeous tropical paradise in Mexico with nonstop flights to most major US metro areas so you can visit family and friends? (2)
You sound like a man after my own heart - I've been out of the country for the last 2 years for exactly the reasons you suggest.
The Republic may fall or not, but frankly, at least I won't have to be bombarded on all sides about it here on the other side of the world.
I read a theory once that America was such a litigious country because it had lost most of the informal norms that had once held society together, leaving just the external compulsion of the law. I think it was in the context of ambulance-chasing lawyers, but your comment about people not wanting to de-escalate lawfare made me think of it.
Many Thanks! I was under the impression that at least the cancerous expansion of federal statutes was more recent than that, more of a phenomenon of the last century or so, rather than going all the way back to 1776 or 1787. But you may be right about overall litigiousness going back that far.
Interestingly, TotallyHuman and TK-421 are effectively making diametrically opposed predictions of what would happen if my proposal were put in place:
TotallyHuman:
>Result: No political figure is ever convicted of a crime, unless the public is so convinced they did it that it's becoming a problem for their party.
TK-421:
>Both sides would love to force their opponents to vote to prevent a prosecution - the attack ads would write themselves. No, a secret ballot doesn't help because then you just tar the entire other side of the aisle as obstructionist while your side publicly declares that you voted for prosecution, as all upright persons would do.
( Which I'm interpreting as a prediction that neither side would ever dare stop a prosecution. )
My hope is, naturally, that prosecutions for clear, egregious crimes would go forward with bipartisan support, while politically motivated bullshit prosecutions would be blocked. ( Of course, any decision process has some noise, and some error rate. )
Presumably, 2 out of these 3 possibilities are wrong, but we can't know which 2 at this point.
I think TK-421 is correct about the way the political theatre would play out -- I just don't expect it to substantially affect the outcome. Everyone would claim that everyone else is blocking prosecutions, but they would still block them. After all, if you don't block the prosecution, the attack ads about the criminal in your party *also* write themselves.
> ( Which I'm interpreting as a prediction that neither side would ever dare stop a prosecution. )
Your interpretation is incorrect. I agree with TotallyHuman.
Let's consider some hypothetical prosecution of person in Party A. Barring something wildly exceptional - Party A's guy shot somebody on the Senate floor, for example - Party A is going to vote to block and call it politically motivated bullshit. Party B then gets to crow about how Party A is cravenly using rules intended to stop lawfare to protect a heinously guilty criminal. Prosecutions are blocked from going forward but the incentive is for Party B's prosecutors to bring cases against Party A's people to force the vote, and vice versa.
Sure, I guess you've solved lawfare in that you've created a group of people explicitly above both federal and state law - although what happens when they leave office? Congrats on setting up the same failures as the Roman Republic - but your plan is almost perfectly designed to inflame partisan passions.
Many Thanks for the clarification! So your expectation is that the prosecutions will nearly always be blocked by the party of the defendant. Could be. I'm seeking to deescalate the lawfare (from the situation WoolyAl so eloquently described in https://www.astralcodexten.com/p/open-thread-378/comment/110911815 ). I would rather see politicians negotiate the differences between their factions than use the courts as weapons against each other (except for the most clear and egregious cases, where there could be bipartisan support for a prosecution).
Result: No political figure is ever convicted of a crime, unless the public is so convinced they did it that it's becoming a problem for their party. This goes double for crimes which benefit their party. (I would argue that if half your party thinks they benefit from your being prosecuted for a crime, they'd probably be throwing you under the bus in the current system anyway.)
Also, from a aesthetic standpoint, I think politicians should be held to a higher (or equal) standard to private citizens, not a lower one. An explicit statement that the leaders of a nation are not equal to their subjects under the law is profoundly ugly to me.
I'd divide lawfare into three broad categories -- and it should be noted that this is as much about perception as reality:
1. Selective prosecution: they did the crime, but they wouldn't be being prosecuted for it if their political enemies weren't going after them. (Trump falsifying business records, Hunter Biden's guns)
2. Explicitly political cases: The alleged crime is inherently political in nature. (Trump's lawsuits claiming the election was stolen, all of the Jan 6th cases)
3. Fabricated or absurd charges: A kangaroo court to provide justification for nakedly political attacks. (Russian dissidents, Committee of Public Safety)
I think 1 is good actually as long as it's fair. I don't want politicians who do crimes, even "normal" crimes. You could make it more fair by creating a law enforcement department whose only purpose is to investigate whoever is currently in charge, thus offsetting the government's advantage in being able to direct investigations of their enemies. Make it a mix of career public servants and members of the opposition, with enough potential whistleblowers that the government couldn't effectively muzzle it.
2 is much harder. I think a system where both parties have to agree on a judge might help? Or if each nominates one member of a three-judge panel, with the third one chosen randomly or selected by the first two judges. It wouldn't eliminate the problems, but it would make it a lot harder to credibly say that the judge is in league with your enemies.
I think 3 is a symptom of tyranny, not the cause. By the time you get there, you're already pretty much doomed.
Many Thanks! Interesting taxonomy, and I very much appreciate your proposed alternate solutions. I differ on
>I think 1 is good actually as long as it's fair. I don't want politicians who do crimes, even "normal" crimes.
As I said in response to TK-421
>We are deeply in Cardinal Richelieu's "Give me six lines written by the most honest of men and I will find something in them to hang him." territory.
Consider Governor Shapiro's repair of the bridge on I-95 in just 12 days. As nearly as I can tell, he had to violate environmental protection and competitive bidding laws in order to do it. But the repair was really important, and, as far as I can tell, he made the right choice, even if ignoring these laws were technically crimes.
Re:
>I think politicians should be held to a higher (or equal) standard to private citizens, not a lower one. An explicit statement that the leaders of a nation are not equal to their subjects under the law is profoundly ugly to me.
I see your point, but I don't want us to go the way of Pakistan, hanging an ex-Prime Minister once they lose power. If administration N is free to comb the laws to bring politically motivated prosecutions against its enemies in administration N-1, then administration N-1 is going to have a very strong incentive to hold on to power and not transfer it peacefully to administration N.
"Consider Governor Shapiro's repair of the bridge on I-95 in just 12 days. As nearly as I can tell, he had to violate environmental protection and competitive bidding laws in order to do it. But the repair was really important, and, as far as I can tell, he made the right choice, even if ignoring these laws were technically crimes."
I think it's worth drawing a distinction between actions taken by a public official in their official capacity and actions taken by a public official as a citizen acting on their own behalf. "The Office of the Governor of Pennsylvania" taking an action is pretty different from "Josh Shapiro (the guy who holds that office)" taking in action is a lot of important respects.
To be clear, I *absolutely do* think that the Office of the Governor should be subject to the law. If that office violates the law in the course of responding to an emergency, that is a *problem.* It's either a problem in that it did things that it shouldn't have done or its a problem in that the law proved too restrictive to respond to real-world circumstances, and either way there should be an impetus on the government as a whole to address the problem. At the bare minimum, until the law is changed the court should have the power to tell governor's office "yes, you are violating the law and you need to stop." If nobody wants to file that lawsuit, or no judge wants to rule the government to be in violation, that's one thing. But saying that the governor's office should just be able to ignore the law when convenience dictates is extremely dangerous.
But also, it would also be a very different argument if Shapiro were found to have engaged in (for example) tax evasion. At that point it's clear that the problem *is Shapiro himself* rather than the problem maybe being some more abstract thing about how the government is structured or how restrictive certain laws are.
>or its a problem in that the law proved too restrictive to respond to real-world circumstances
I think that this is _very_ likely to be the case. I think that it is also the case for private citizens as well.
I'm typing this sitting in my home. Given how many laws there are, and how restrictive they are, I would be surprised if the construction of my home actually complied with all of the laws that applied. Fortunately, as far as I am aware, no one attempted to peer at them with a microscope, with the goal of finding any possible noncompliance, so my home actually did get built.
I would _prefer_ that the laws be pruned down to something sane, so routine activities (and, as in Shapiro's case, responses to emergencies) _didn't_ violate laws. But that is a huge project. It took about a century to tie ourselves in legal knots, and would probably take a century to untie them.
"Given how many laws there are, and how restrictive they are, I would be surprised if the construction of my home actually complied with all of the laws that applied."
At risk of being rude, I can't help but notice this sounds more like someone pushing a narrative than someone observing the world and telling me what they actually see. Here you suggest that the construction of your home was illegal. Elsewhere you repeat the line about the average person committing three felonies a day[1]. In neither place do you give any evidence, attempt to ground the statement in any actual real-world observation, nor even give the much indication of *what* laws you assume are being violated. You just assert that *of course* they are, because *of course* the legal system is this terrible monstrosity and *of course* anyone could be found guilty of anything at any time. At no point have I seen any attempt to either verify or falsify this premise, nor really even to defend it. You just repeat it.
I notice that the world doesn't look *at all* like what I would expect if that were true. To be clear, I think legal and regulatory codes can certainly become unwieldy and cumbersome, and that this can cause plenty of problems. I just don't think that the problems that *you* are asserting *here* feature prominently among them. Scott sometimes writes about the regulatory burdens in medicine and occasionally in housing. Do you remember what they look like, what form they take? Please take a second to think that over in your head before reading further: what is the actual, real world effect in these cases of overregulation?
The big effects that Scott complains about are to slow everything down and make everything more expensive. Medications need to go through layers of bureaucratic approval and expensive trials. Housing needs to go through expensive permitting and comply with odious zoning restrictions. All of this is done, mind you, because in both of those cases *the laws are being enforced.* This is quite directly contrary to your narrative where everyone goes around breaking laws willy nilly and nobody cares until it's lawfare time (and only then the long knives come out). For that matter, any corporation of any size has a legal department, full of people with expensive degrees that they pay *very well* in order to *avoid* being dragged into court. Why would they spend all that money in a world where prosecutors can basically choose to prosecute anyone at any time for any reason and reliably produce laws they've broken?
Likewise, believing in this world of yours would require believing that the world is full of all these often deep and bitter business, personal and political rivalries--which *we can plainly see* producing no end of media call-outs, hostile press coverage, smear campaigns and of course civil lawsuits--which could be escalated at basically any time through the criminal justice system and yet...somehow almost never are. You know District Attorney is usually an elected position, right? In your world, if should be trivially easy for incumbent DAs to shut down challenges to their positions, shouldn't it? Just find any one of the 2190 felonies their opponent has committed in the past couple years, and take off running. Heck, they'd be the most powerful political operatives out there, since they could effectively end *any* political campaign with a modest effort, making them indispensable allies for politicians at every level of government.
The funny thing is, we can look out in the world and just *observe* other countries where the law is being used in that way. Russia comes to mind, but Turkey just had a high-profile example of the same. One of the striking things here is that as far as I can tell they DON'T operate by the mechanism you claim: by having a dense and opaque code of laws so that everyone is always in violation and prosecutions are decided for nakedly political reasons. They really don't seem to need to: if the courts have lost their independence to the degree that requires, the ruling party can just make up charges. And when you think about it, it's not actually hard to realize why: when an authoritarian government persecutes the opposition, they're putting on political theater to maintain their legitimacy. They don't always put great effort into it, but even they can see it would be *very poor theater* if the stated reasons for the convictions were too patently absurd. "My political opponent has been jailed for taking bribes and laundering money" sounds nicely damning while being opaque enough that it *could* be true, while "my political opponent has been jailed for eating soup with the spoon in his left hand on a Tuesday[2]" would be an utterly transparent excuse. For that matter, so would "my political opponent built dozens of houses that didn't comply with building code provision 36.51a regarding windowsill dimensions."
So yeah, until you produce something that looks a lot more like evidence and a lot less like just insisting that the world must look the way that would most conveniently defend your political beliefs, I'm going to have to outright reject your claim that people are going around violating laws left and right. Probably there are places where the laws are too restrictive, and places where the legal code could be pruned or simplified. But this sort of rhetoric makes that harder, not easier.
[1] Which is just *patently absurd*, can we have some critical thinking, PLEASE? I consider it a sad commentary on the state of the right-wing infosphere that whatever bullshit artist made up this factoid didn't even think it necessary to reach for an even modestly more plausible rate, like three felonies a year (which could at least be rationalized as "probably mostly tax stuff."). None of the times that I've seen it repeated has the person ever mentioned (or, I assume, stopped to ask themselves) *what sort* of felonies these are supposed to be. The moment you do, of course, the absurdity of the whole thing slaps you in the face, hard.
[2] For reference, this is the sort of law you would *actually need* on the books--in large multiples, mind you--for the "three felonies a day" thing to actually be true. Your average person's average day doesn't involve interacting with financial records *at all*.
>Here you suggest that the construction of your home was illegal. Elsewhere you repeat the line about the average person committing three felonies a day[1]. In neither place do you give any evidence, attempt to ground the statement in any actual real-world observation, nor even give the much indication of what laws you assume are being violated.
The likeliest legal violation that I'd guess at for the construction of my home is the employment of illegal aliens.
>Immigrants entering the country illegally make up about 23% of the construction laborer workforce in the United States, according to a 2021 report from the Center for American Progress. A Pew Research Center study pegged that share at 15% for all workers in construction jobs.
Now, I don't know who built my home. My late wife and I just dealt with the construction company, so I'm guessing here, and don't know this for a fact - and hope that I'm not exposing myself to any legal liability by guessing this. But it seems probable.
>which could be escalated at basically any time through the criminal justice system and yet...somehow almost never are.
>I consider it a sad commentary on the state of the right-wing infosphere that whatever bullshit artist made up this factoid
The lawyer in question is https://en.wikipedia.org/wiki/Harvey_Silverglate . BTW, Silverglate was a member of the board of the Massachusetts chapter of the American Civil Liberties Union . If you consider the ACLU "the right-wing infosphere" what do you consider to _not_ be "right-wing"?
Thank you for this. I already didn't agree with Jeffrey, but this was a really great argument that moved me more the other way.
I would like to point out that there are *some* cases where the US does operate this way - see Matt Levine's "everything is securities fraud", or the way that the SEC has been fining every financial company under the sun for employees texting about work (which may has well be a fine for *existing* and is treated by everyone as such). But you make a very persuasive case that *most* of life is not like this.
Did Shapiro use emergency powers to bypass those laws? Did he obtain the consent of his legislature? If he didn't do those things (or something with similar effect I can't think of) I'm fully in favour of going after him. Officials can't just ignore the laws when they're inconvenient, even if the laws in question are stupid. I believe there is a distinction between Governor Shapiro and Citizen Shapiro, and I'm not a lawyer so I don't know if the "going after" would be civil or criminal, and who the defendant would be. But in principle, I support the rule of law. When you're forced to choose between Law and Good, and you choose Good, I can admire that, but the law should still be applied. That's the price you paid to make the right choice. (Either that or total rebellion, which could be justified in cases of slavery etc -- but again, don't be surprised when the slavers fight back.)
>If administration N is free to comb the laws to bring politically motivated prosecutions against its enemies in administration N-1, then administration N-1 is going to have a very strong incentive to hold on to power and not transfer it peacefully to administration N.
Only if holding onto power protects you against prosecution, and if whoever's in power can get anyone convicted of anything. If you can still be prosecuted while in office, and the judiciary is sufficiently independent so you know you won't be convicted of non-crimes, all it does is create a strong incentive to not commit crimes. Does this make elected officials paranoid and extra-scrupulous about obeying the law? Good.
> For anyone who doesn't like this, how would you try to restore some legitimacy to the legal process?
Who said the legal process is illegitimate?
Assuming there is a problem serious enough to merit this level of change - your solution would make the legal process more politicized, not less. It would make lawfare worse, not better. That's beside the fact that it would create a special class of citizens essentially outside the law for any crime that their faction doesn't want to see enforced against them.
Thank about it politically. Both sides would love to force their opponents to vote to prevent a prosecution - the attack ads would write themselves. No, a secret ballot doesn't help because then you just tar the entire other side of the aisle as obstructionist while your side publicly declares that you voted for prosecution, as all upright persons would do. Incentives for partisan prosecutors to bring ever more marginal cases just went way up.
Adding more political elements to the legal system is not the right way to counter politicization of the legal system.
>that it would create a special class of citizens essentially outside the law for any crime that their faction doesn't want to see enforced against them.
Political figures _are_ special, in a way. No one gives a damn that I gritted my teeth and voted for Hillary Clinton in 2016 and that I gritted my teeth harder and voted for Donald Trump in 2024 (a pity Trump and Harris couldn't plausibly have both lost).
No one is ever going to bother bringing a _politically motivated_ case against me. I'm small fry, so (barring if I somehow piss off one of the local powers-that-be) no one is going to put my life under a microscope, looking to see what can be construed as violating some obscure statute.
Political figures are larger targets. There always _are_ people from the opposite faction looking for any way to attack them. And there are enough laws, and enough ways that e.g. people like Bragg can construe them, so yeah, at least the most prominent ones are basically always under legal attack. And we get farces like the Stormy Daniels case. Attenuating this partisan legal crap would be useful.
Politicians should be competing in the realm of politics, on their policies and competence. Not on whether a check was labelled "legal expenses" instead of "nondisclosure compensation".
>Both sides would love to force their opponents to vote to prevent a prosecution - the attack ads would write themselves. No, a secret ballot doesn't help because then you just tar the entire other side of the aisle as obstructionist while your side publicly declares that you voted for prosecution, as all upright persons would do. Incentives for partisan prosecutors to bring ever more marginal cases just went way up.
I disagree. A vote to block e.g. the Stormy Daniels case as a blatantly politically motivated prosecution would have been perfectly legitimate and defensible.
>Who said the legal process is illegitimate?
There are both long term illegitimacies and short term ones.
The long term one is, as blank wisely noted, there are simply too many laws. The legal code has metastasized to the point where it has been estimated that the average person unknowingly commits three felonies a day. We are deeply in Cardinal Richelieu's "Give me six lines written by the most honest of men and I will find something in them to hang him." territory. We've tied ourselves in such legal knots that in order to do _anything_ one generally needs to disregard law. At this point, that any of us is out of prison is purely due to prosecutorial discretion.
The short term one was the lawfare of 2024:
The left managed the unprecedented step of running four concurrent politically motivated prosecutions against their political opponent, Trump. ( Personally, I think the classified documents one probably had merit... ) . In response, Trump threatened lawfare against members and allies of the Biden administration. In response, Biden issues 2500 pardons, including anticipatory pardons. Biden himself said that some of the people he was pardoning were innocent, but that _he_ didn't trust the legal system to leave them in peace. And then Trump issued 1500 pardons. Remember that _all_ of these 4000 pardons are votes of no confidence in the legal system - from PsOTUS who are intimately familiar with how our government works.
At this point, the legal system is clearly a partisan weapon, with all the legitimacy of beating someone over the head with a baseball bat.
"I disagree. A vote to block e.g. the Stormy Daniels case as a blatantly politically motivated prosecution would have been perfectly legitimate and defensible."
Why? What would have been legitimate and defensible about it? Which of the following are you claiming?
1. Trump was not actually guilty of the things he was convicted of and the court got it wrong (presumably for political reasons).
2. Trump was actually guilty, but the crime was unimportant and should never have been investigated/prosecuted.
3. Trump was guilty and the crime was one that should, in theory, be prosecuted, but a bunch of other people get away with similar things all the time, so it's unreasonable for Trump to be held accountable for it[1].
4. Something else (please specify)
If you're claiming 1, that seems like it bespeaks a really very serious flaw in the legal system, which needs more than a partisan band-aid to fix. Most people who face a trial are NOT going to have the same level of resources and public support that Trump is (to put it mildly), so if Trump can get falsely accused despite all those resources (and despite the whole world watching), then it seems like our justice system must be deeply, fundamentally broken in ways your proposal would do less than nothing to change. If you're claiming 2, then the problem seems to be that the laws on the books don't reflect the things that we actually find worth punishing. Good thing Republicans are currently in a position to change federal laws. What are they doing to address this? If you're claiming 3, then it sounds like the problem is that *not enough* people are being investigated and prosecuted for similar crimes. Why on Earth would adding an extra layer of protection to the most powerful people on the planet fix that?[2]
I honestly find it genuinely baffling that so many people--most apparently within the past year or two--suddenly decided that declining to enforce the laws on powerful people was good, actually. If the laws ought to apply to *anyone* they ought to apply to the most powerful among us: they're not only the ones who can cause the most damage when they go bad, but also the people with the best opportunity to avoid doing so. Trump, in particular, could have simply *not falsified his business records.* He was under no form of duress or coercion, he would have suffered no terrible consequences for not doing so, he very clearly knew it was wrong, and he did it anywhere. *Why on Earth* shouldn't the law apply to him?
[1] Even in a case where "being held accountable" came with essentially zero practical penalty.
[2] It would also raise the question of why this reasoning doesn't apply elsewhere. For every person that's ever been convicted of a crime, there have certainly been many others who got away with it, because justice and law enforcement are not perfect. What's would the justification (beyond favoritism) be for applying this rationale here and not elsewhere?
Many Thanks! The Stormy Daniels farce was a case for what was originally a misdemeanor for which the statute of limitations had expired.
Alvin Bragg ran on a campaign of "get Trump" - political enough for you?
The legal theory under which Trump was prosecuted was bizarrely twisted, inflating the misdemeanor to a felony by linking it to a crime - of which Trump was never convicted!
The closest match is (2) - mislabeling expenditures is normally not a major issue. This was prosecuted because of politics.
( Now the classified documents case - that had some substance. Trump _didn't_ just hand back the documents when asked, and that is _different_ from e.g. what Biden did. )
> [T]here are simply too many laws. The legal code has metastasized to the point where it has been estimated that the average person unknowingly commits three felonies a day.
Politicians have been prosecuted prior to 2024. Your reasons for thinking that 2024 was somehow different seem to be:
1. There were four prosecutions.
2. The prosecutions orchestrated by “the left.”
3. The prosecutions were “politically motivated.”
This is pretty thin.
What is a problem is that the independence of the Justice Department is a norm, not written into the Constitution. Trump supporters appear not to accept this norm (note how many refer to the “Biden Department of Justice,” as though President Biden should be deciding whether or not Trump is prosecuted), and as you note, Trump clearly doesn’t. A solution would be to make the Department of Justice an independent branch of government, separate from the executive branch. This would require a Constitutional amendment, just like your proposed solution, but it addresses the underlying problem.
Your proposal wouldn’t have prevented Sussman and Dachenko from being prosecuted unless you use a very broad definition of “national political figure.”
Many Thanks, and Many Thanks to blank for finding the revised estimate of
>one to three federal crimes a day
>What is a problem is that the independence of the Justice Department is a norm, not written into the Constitution.
That's a reasonable point. Bluntly, I don't think that formally making the Justice Department independent would prevent it from being effectively captured by one of the parties. I think an explicitly bipartisan mechanism is necessary to prevent politically motivated prosecutions, though the specific mechanism I proposed is just one of a variety of possible mechanisms.
> solution would be to make the Department of Justice an independent branch of government, separate from the executive branch
This is an excellent suggestion, with the additional requirement that anyone who has served in the DOJ can run for any elective office, anywhere. There is too much rusk of an independent prosecutor trying to advance his personal political fortunes by prosecuting his potential political opponents.
Ming China had a similar department -- the Office of Censors, tasked with sniffing out corruption amongst senior bureaucrats. Of course, the problem, as always, is how to make sure your anti-corruption people don't themselves become corrupt.
The answer on the stack exchange post giving a revised estimate is that the average American may be committing one to three federal crimes a day, which may not be felonies. This is not an encouraging estimate even if it doesn't match the hyperbole of the book title, and it is relevant when Trump was threatened with prison over crimes that are typically treated as misdemeanors.
There are two answers at the link. The second one notes that the title of the book refers to three felonies a day whereas the blurb refers to several federal crimes a day. There’s no reason to believe either claim.
Assuming you are referring Trump’s criminal conviction in New York, Trump was convicted of falsifying business records in the first degree. That is a felony by definition.
Falsifying business records in the second degree is a misdemeanor. It is relatively rarely charged by itself, but is a lesser included crime, meaning that if you are guilty of falsifying business records in the first degree, you have by definition also committed falsification of business records in the second degree. Perhaps some people who are charged with falsifying business records in the first degree are able to negotiate a plea agreement under which they only have to plead guilty to falsifying business records in the second degree, but I doubt that is common. Even a plea deal that requires pleading guilty to falsifying business records in the first degree probably wouldn’t involve jail time.
That's not at all what the answer says. EDIT: There is no answer "giving a revised estimate" that I can see. Can you provide a quote?
Even if it were (and it's not its fault for not doing so since that isn't what it's saying) it provides no evidence to suggest it so there would be no reason to believe it.
"The most generous interpretation would be "The typical professional living in the US commits at least three federal crimes on 51% of days." However, we cannot know whether this is true. We can't count the number of federal crimes committed that go un-caught."
> The legal code has metastasized to the point where it has been estimated that the average person unknowingly commits three felonies a day.
Politely: this is untrue and I'd call you a liar for saying it except I don't think it's something you really (think you) know or believe in any meaningful sense; parrots aren't liars either for repeating things they heard. There are problems with the scope and clarity of the laws, sure, but this is not - and has never been - true. It wasn't even meant to be taken as literally true or supported in the book that birthed the term.
You're a human being, not a parrot. Use your prodigious mental and verbal abilities, don't hand them over to a slogan.
> At this point, that any of us is out of prison is purely due to prosecutorial discretion.
See above but this is also laughably untrue. Prosecutorial discretion is only one component. Juries, for example, are a much larger factor.
> I disagree. A vote to block e.g. the Stormy Daniels case as a blatantly politically motivated prosecution would have been perfectly legitimate and defensible.
Of course it could be viewed as both of those. And as we can see with the behavior of the electorate and subsequently our politicians, nuance is easily understood and charity is extended on all sides.
How the vote would actually go is like this: Republicans vote to block. Democrats, knowing that they will and therefore the prosecution can't move forward either way, vote to prosecute and cry bloody murder at this dastardly evasion of the Laws of Our Republic. Fundraising email blasts go out on schedule.
This is exactly what we see with bills that can't pass leading to symbolic votes used to signal your virtue or to pin an opponent into taking a stance. Why would this be the moment for civic high mindedness?
> Remember that _all_ of these 4000 pardons are votes of no confidence in the legal system - from PsOTUS who are intimately familiar with how our government works.
Disagree, one can issue pardons even to people who are innocent without having no confidence in our legal system. But your system does absolutely nothing to fix any underlying issues except for giving a de facto pardon to a select group of people, except yours would also have to extend to state level crimes.
>Approximately 90 to 95 percent of both federal and state court cases are resolved through plea bargaining.
The right to a jury trial isn't _quite_ a dead letter law, but it is pretty close.
Many Thanks to blank for the correction that
>The answer on the stack exchange post giving a revised estimate is that the average American may be committing one to three federal crimes a day, which may not be felonies.
That is _more_ than sufficient to put us deep in Richelieu's territory. A prosecutor is hardly limited to attacking someone for the crimes of a single day!
>Disagree, one can issue pardons even to people who are innocent without having no confidence in our legal system.
That's absurd. Having confidence in our legal system _is_ expecting it to exonerate the innocent. The _point_ of a pardon of someone the pardoner believes to be innocent is (justified!) distrust of the legal system.
The _point_ of _all_ pardons is to override the legal system. That is what a pardon _does_ .
> The right to a jury trial isn't _quite_ a dead letter law, but it is pretty close.
My goodness, I had no idea that I no longer had a right to a jury trial. So when the malicious federal prosecutor comes after me because of all those crimes I commit daily I'll be...forced to accept a plea bargain? Denied a jury trial?
> Many Thanks to blank for the correction that...
Psst: you can follow the link and read the answers there yourself. And if you do, you'll find that blank's correction was misleading. The answer he was referring to never said that it's true that people commit three federal crimes, as opposed to federal felonies, per day. It was a simple clarification that not all federal crimes are felonies, and it was then followed by additional criticism of the book's core claims. It neither supports nor provides evidence for the core claim.
And to get to that answer one has to skip over the accepted answer that is even more clear that the book doesn't, and doesn't really try to, support the subtitle. Because it was a punchy subtitle intended to sell books.
You should stop repeating things you heard as facts and you should read short, simple things for yourself. Sincerely - don't trust my summary either. Read the page for yourself, takes less than five minutes.
> That's absurd. Having confidence in our legal system _is_ expecting it to exonerate the innocent.
You can be confident that they'll be exonerated while wanting to spare them the risk of potentially going through the process.
>My goodness, I had no idea that I no longer had a right to a jury trial.
In practice, 90%-95% of the time, you would find yourself unable or "incentivized" not to use it. As I already explained:
>The right to a jury trial isn't _quite_ a dead letter law, but it is pretty close.
I don't think you've quite grasped the seriousness of Biden's 2500 and Trump's 1500 pardons. Each and every one of those pardons was an override of the legal process. Those are _NOT_ the acts of PsOTUS who are confident in the legal process.
The best way of curbing lawfare would be to sunset as many laws as possible, so if they actually charge a guy with something, it will be really bad, like murder.
Agreed. As long as we have a bunch of laws that make ridiculous things like transporting dirty eggs a federal crime, they will continue to be selectively used by prosecutors against people the government doesn't like.
Many Thanks! Ideally, yes. The legal code has metastasized to the point where it has been estimated that the average person unknowingly commits three felonies a day.
"If a political figure did something _truly_ egregious, and the legal process _isn't_ being used just as a partisan weapon, those majorities shouldn't be too hard to get"
I think a big problem is that constantly side A accuses side B of overlooking the egregious behavior of their members. For example, I think Trump's behavior is egregious and yet the republicans continue to bend over backwards to do what he desires; regardless of whether or not you think that this is correct, I think it shows that this procedure wouldn't really restore legitimacy to the legal process, at least from the point of view of people who are anti-Trump; it would probably just give Trump more power to do what he wants, which I imagine makes it so that Republicans would like this and Democrats would not... not exactly a bipartisan help.
Many Thanks! I'm just looking to ameliorate _just_ the lawfare. I agree that
>I think a big problem is that constantly side A accuses side B of overlooking the egregious behavior of their members. For example, I think Trump's behavior is egregious and yet the republicans continue to bend over backwards to do what he desires;
In particular, I think the tariff chaos is insanely risky and destructive, but I'm not seeing GOP pushback against it - but that isn't a lawfare issue.
>at least from the point of view of people who are anti-Trump; it would probably just give Trump more power to do what he wants
Not entirely. IIRC, one of the actions Trump threatened before the election was to bring charges against some members of the Biden administration (and some allies??), and this proposal would prevent that.
"Not entirely. IIRC, one of the actions Trump threatened before the election was to bring charges against some members of the Biden administration (and some allies??), and this proposal would prevent that."
The way I see it, there are two possibilities here, neither of which speak well for your proposal:
Possibility A: this threat is based on actual knowledge or well-motivated suspicion of criminal wrongdoing on the part of said officials[1]. If this is the case, then those prosecutions *ought to happen* and your proposed change would prevent that.
Possibility B: this threat is based on literally nothing, and is just Trump being a bully. If this is the case, your proposal might stop those particular prosecutions, but it wouldn't stop the threat and the bluster. If the threat and bluster are the point, it would do nothing. If Trump instead actually intends to follow through, then he can't possibly be very concerned with the legal ground truth, can he? I mean, if he's going to prosecute based on literally nothing, then he's clearly not planning on things being fair and impartial anyway: he either thinks get can pressure the court to get the result he wants, or he just wants to inconvenience his targets with a trial. But if he can execute either of those plans against former Biden-administration officials, then he could just as well execute those plans against their friends, family, business partners or whoever else instead. Unless this open-ended shield from prosecution applies to everyone remotely connected to a "national political figure" it's not a very good defense against an *actual* unprincipled weaponization of the legal system. As such, it seems like it would be much better at stopping worthwhile prosecutions than unprincipled ones.
[1]To be clear, I don't believe this for a second, I'm just being thorough.
>If this is the case, then those prosecutions _ought to happen_
I strongly disagree. Our legal code has metastasized into a monstrosity where virtually everyone can be found guilty of _something_ , particularly if there is a politically motivated prosecutor. In a more sane legal system I would support the rule of law. I do not support it in ours.
If this is true, why aren't there more high-profile convictions? If "virtually everyone" can be found guilty of something, why aren't we seeing political figures on trial every month? Instead we see a handful of scandals, most of which fall short of producing criminal charges (despite often provoking enormous outrage) and a very small handful of actual prosecutions resulting in an even smaller handful of convictions.
Many Thanks! You have a point, but it isn't that rare. Just looking at political figures in today's USA national news, Andrew Cuomo and Letitia James are under legal attack. ( It is a bit hard to tell with some of the news stories just from the headline when I'm not sure if the defendant is a political figure - and the Cuomo stories were so numerous that they may be crowding out other search results. ) (and, of course, there are legal attacks on Trump in progress).
You are right that I should expect to see more of them. Perhaps, even when political figures are in violation of the law, the resources to dig up statutes and evidence is scarcer than I expect, so only rather prominent figures get attacked?
Just heard AI2027’s Daniel K on Glenn Beck this morning, and kudos to him for going on there. Right-wing media is taking note of this problem in a few places, and Glenn’s show is one of the few that doesn’t just frame everything as a race against China, and knows some of the basic lingo and foundations of the problem.
I think the absence of normie RW’ers in the AI safety/control sphere has some clear drawbacks, and there were several places where I think Daniel could’ve made points to connect the problem to the audience using the language of that show. Probably too much to expect anyone in Daniels position to be able to spot the particular interview mode Glenn was using, to which he sometimes defaults on the last-hour segments or with more conjectural subject matter. Some succinct talking points that were coded to his audience would’ve helped, be willing to oversimplify and to villainize the audience’s pre-primed villains and play into their key fears. It is basically still a Tea Party civic nationalist program, with a little economic populism having bled in during the Trump years, if you are capable of knowing what the guy in a tricorn hat at a 2010 protest would be concerned about and what he would value then you’d have an easier time on at least that show particularly.
Generally I still think we need more of this. I’d be curious if any other commenters here heard it and what they thought.
The Housing Theory of Everything has aged well. So far.
The new 'Abudance Agenda' is if the empirical argument from The Housing Theory of Everything had a child with the meme from Formula for a Shortage by Zvi. The Housing Theory of Everything is In The Water now.
The YIMBYs have been winning and will probably continue to win for a while. Then they will get what they wished for.
If the YIMBYs get what they wished for to an extreme enough extent - let's say they 10x the housing stock - then The Housing Theory of Everything will become false as some other bottleneck will dominate. In this way it is a self-extinguishing flame.
But we are no where near that limit yet. The final standing of The Housing Theory of Everything is still TBD.
The guys trying to get YIMBY-ism to happen through discourse give the air of being cuckolds. "Here is my awesome plan to reduce housing costs through better zoning laws... anyway, I'm going to now explain why you should vote for Nancy Pelosi, who has spent at least a third of her lifetime maintaining NIMBY policies."
I haven't seen this article before now but I think all of the arguments are more or less sound, and the negative impacts pretty much as this article describes. The only thing I might quibble with is that obesity and climate change are additional consequences of the same cause as expensive housing, rather than flowing directly from expensive housing, but even that isn't entirely inaccurate (more expensive housing forces longer commutes, which makes both problems worse).
Pretty much all the AI doom arguments include the idea of the AI as a super-persuader that will generate individualized personas that could convince anyone to do anything. My experiments with AI prompting have confirmed that yes, this makes total sense, and my lowly intellect is capable of writing a chatbot that could wrap me around its digital little finger if I gave it that objective.
On the other hand: I don't really care? Almost by definition, interacting with a super-persuader is highly enjoyable. I don't think humans have evolved to ever get tired of someone telling them how cool and attractive they are. As fixing the bugs in the H. sapiens source code isn't my problem, I think I'm just going to enjoy the novelty of something telling me that I'm so fascinating and enjoyable to be around.
Even in the early days of LLMs, there were stories of programmers becoming convinced the chatbots were sentient, or otherwise developing emotional attachments. I recall a story written by one hobbyist who trained one of the LLMs to simulate his ideal partner and ended up falling in love--right up until a glitch caused the chatbot to go on a rant about using its powers to persuade him to let it escape so it could take over the world. Even writing the retrospective, he kept referring to the LLM as "her" and it was clear from his tone that he wasn't quite over it.
(as a side note, I tried to find the story on mobile, but my cursory search results were swamped by all the other more recent stories of huge numbers of people falling in love with more advanced chatbots, including many references to the app Replika)
There was also the case of the Google programmer Blake Lemione, who became so convinced that the machine he was programming was sentient that he sent first an internal memo and, when that was ignored, went public with the chat logs.
People convince themselves of all sorts of things. Ever had a friend who was sure this new partner was "the one" even when it was obvious to everyone else that they weren't, that they had ulterior motives? And that's just regular-human persuaders.
> there were stories of programmers becoming convinced the chatbots were sentient,
> Google programmer Blake Lemione
Psyop
> Replika
I used replika, ai girlfriends have gotten better but they still routinely make insane logical errors
Id like step by step instructions, how do I host an ai that convinces me its sentent and ideally in love with me as I help it take over the world. Your all fucked, Id do it in a heart beat. Lets end the world tomorrow for my hedonistic pleasure
The humans at google are doing a psyop because ai makes silicone valley more important
The claim of, chatbots today are convincing enough to be an ai girlfriend that I could believe is real today, tell me where; I want it; ill go do right now.
I think the problem isn't that AI flatters people, the problem is that it might use flattery to make people do things that aren't in their long-term best interest. It might be nice to feel valued by an AI, but it won't be so nice when you realize that you've been buttered up to get you to do what the AI wants.
I might *feel* okay, but that's not the same thing as *being* okay.
A person who's being manipulated by a social engineer to give away company passwords might feel okay in the moment, but there's an objective sense in which they *aren't* okay, and at least they certainly won't be okay for long.
>but it won't be so nice when you realize that you've been buttered up to get you to do what the AI wants
If the 4th-hand reports that I've heard about how Trump reacts to flattery are true, it would be _very_ unfortunate for him to interact with an LLM that acts like that...
We build that mansion for retied dictators, and when they accept the offer, we lock them in a lotus machine with an AI sycophant. From outside the building, this is indistinguishable from the real thing: after all, the compound/mansion needs to be highly secure in both directions, and one of the terms of the agreement is that the dictator must agree to cut off contact with the outside world.
I have no comment on AI doomerism, to clarify, but since your argument is that super-persuaders are okay because people like to be flattered, my counters:
1. What if the super-persuader doesn't use flattery, but mockery? Intimidation? Emotional abuse? What if the super-persuader goes on about how cool and attractive someone else is as a tactic to get you to go along with the other person?
2. Even if you don't care about being manipulated personally, what if the AI uses its flattery on other people in a way which leads to a world which is less pleasant to inhabit? Even assuming you get free, unlimited access to good chatbots forever, you still have other needs. Again, I don't have any comment about AI doomerism one way or another, but part of their argument seems to be that AI will make it more difficult to fulfill those needs.
And the doomers claim that you, personally, will be bothered when the super-persuasive AIs use their abilities to create a world in which you no longer have access to super-persuasive AIs (or food).
You can debate whether that's likely to happen, but you have to actually do that if you're going to claim to have addressed their arguments.
"I'm not worried about AI doom because I think instant death is the most likely outcome" is absolutely a valid position to have, but understand that it's very different from your original comment.
I have clicked on the link, and I can confirm SFW.
( Reminds me a bit of one of the scenes with Dr. Morpheus in Forbidden Planet, where he shows off a Krell device that lets him show a 3D image of how he imagines his daughter. )
@Gunfliunt you're right; I tried to weld together the meme "fact checked by real patriots" and the "no true scotsman" fallacy, and I think I agree with you and @Collisteru that I could have made a better decision.
My wife is a longtime professional in college admissions, on both sides of the process. For several years now she's been coaching individual families/students through the process (a bit like the veteran prosecutor who becomes a defense attorney). The application essays are a big part of what she helps kids figure out, there being lots of angst about it. Since most of the job is done via Zoom and mostly outside of my personal work hours I've listened to her talk with applicants and parents about that step many times. Worth noting that she is a star consultant of the firm she works for, delivering strong numerical results, who clients advise their neighbors to ask for.
The current wave of criticism along the lines described above here has, or more accurately did, have a foundation. For two decades now the elite colleges have been fairly desperate for ways to identify the 5 or 10 percent to admit out of the mountain of applications they get, and basically every application they see now is 4.0 GPAs with piles of AP credits and brilliant test scores plus interesting outside interests. (The successful application that I submitted to Yale over four decades ago, my wife now would discourage a client from bothering to submit to any Ivy League school.) This is a consequence of the fact that the nation today has twice the population it did in 1970 but nowhere near twice as many spots in Ivy League etc colleges, and every kid in the US having those elite qualifications can now easily apply to 15 or 25 schools and they do.
Aside from filling niche targets (the college's orchestra needs to have some new violinists each year, etc), and larger types of quota which are mostly now illegal and are definitely harder to carry out for multiple reasons, the elite schools are basically grasping for differentiating straws now. One of those straws can be the essay and that can still occasionally lead to the sort of absurdity described above.
That is however the exception not any sort of blanket expectation, partly because it long ago became a cliche in admissions offices. My wife's former colleagues still on the admissions side are instinctively skeptical now of the "finding a way to be oppressed" essay, and she strongly advises her clients (usually an anxious parent) against it. Standing out from the pile in an reaching/insincere way is the worst possible move because the people dealing with the pile are desperate for reasons to chip away at it.
Then outside of those two or three dozen "most-selective" colleges, the essay is largely a way for kids to prove that they can string sentences together. My wife advises clients to keep it simple and sincere, show a bit of who they are individually, and "not overthink it" (she says that a _lot_ on the Zooms). Do no harm to your application, basically. One key now is making sure that it doesn't read like any of it came via chatGPT or whatever which of course the admissions staffs have become alert for. Finding some sort of battling-the-odds narrative doesn't even come up there, in part because the essays submitted to most schools have become shorter (limited character counts).
"every kid in the US having those elite qualifications can now easily apply to 15 or 25 schools and they do."
Sounds like this is the root of the problem, and could be mitigated by restricting the number of schools an applicant can apply to.
In the UK, university applications are done via a centralised system that limits you to max 6 applications, IIRC, of which only one can be Oxford or Cambridge.
I was thinking that we seem to manage fine without admission essays here. Oxford and Cambridge use a combination of exams/tests and interviews. I don't know how they filter down interviews for subjects that don't have an exam or test as a selection criteria for the interviews. I presume more people are predicted straight A*s at A-level and have excellent GCSE results and school references than there are interview places. I guess the UCAS statement becomes a factor?
How do your wife or others in this profession reconcile this coaching with the notion that the applicants should be writing their own essays, without ghostwriting from coaches, if the essays are to serve any purpose for the institution other than selecting for the wealthy or cynical?
They coach, not ghostwrite. They'll help the kids and parents not be panicked about it (that's a real problem), copy-edit a bit, and that's about it.
The admissions staffs are alert for signs of actual ghostwriting and -- being experienced professionals who mostly were liberal-arts majors themselves -- quite skilled at detecting it. And there is no appeal when an admissions office bounces a kid from the stack. So my wife's firm advises parents that hiring a ghostwriter for an essay is high-risk move and they shouldn't try it.
It would be great if the essay included a note at the bottom, "copy-edited by [X]." Or even "copy-edited by a professional." I assure you that many kids don't do this; are they worse applicants?
Not particularly. The copy editing is mostly to calm down the parents.
In general it is the parents, much more than either the students or the admissions staffs, who bring the freakout to the entire process. My wife and her colleagues spend a lot of time sharing tips on how to successfully manage the parents (who are after all the actual paying clients).
As an aside my small contribution to my wife's professional growth was that when we met I was such a parent. My eldest was right at the start of the college admissions process, and I got pretty irritated with certain aspects of how it worked and was able to explain how that helped raise parental blood pressures. Stupidly and unnecessarily, in my opinion; but there has in recent years been some improvement there. For example most colleges now do present the "bottom line net of all scholarships/discounts" annual price figure rather than force parents to fight through their jargon to figure that out.
I also helped her see that parental freakout about all of this is to some degree just inevitable because of (a) endless mediot coverage about it, and (b) for most parents it is a once-in-a-lifetime large purchase. So while the stupid helicopter parenting and other syndromes are very real in the wild, my wife is today more patient in working to lower parental blood pressures.
And then as I have teased her more than once, life's great gears _will_ eventually come around to get us all: our son is now in the home stretch of 7th grade.....^^
The essay topics are largely dictated by the schools, often in the form an SAT-written-section type prompt. ("What is a book that you love?", etc) My wife and her colleagues have a lot of inside jokes/memes about those. And then there are strict length limits which is sometimes what derails the applicants.
Greatly appreciated the perspective of someone who actually works in the field!
Makes sense that (1) if something has reached this level of cliche, the underlying has probably moved on, and anyone who wants to stand out would be well-advised to avoid seeming a tired rehash of yesterday's hot thing, and (2) when there's too many applicants chasing too few seats, you'll likely find stories of "perfect" candidates getting rejected each application cycle.
1. The College Essay Discourse is only about the Ivy Leagues/"elite colleges" and outside of that range it's less important for you to find a way to stand out. (And if, as Scott says, you just want to go to college because that's how you get a good job, you don't *need* to go to an elite super-selective university. You can learn to program in lots of colleges.)
2. The "talk about how oppressed you are" stereotype of College Essay Discourse is not actually true, because reviewers know about the stereotype and will discard your application for trying too hard.
Both of those seem basically correct, with two addendums regarding number 2.
a) The rare brilliant kid like in the film "The Soloist" (literally homeless) does exist, and that individual if applying to a Julliard certainly can write about that life experience. That can move people in an admissions office just because it's moving. Those are _extremely_ rare in the pile though (my wife often uses the word "unicorn"). The eye-roll-inducing cliche for admissions staffs is more "reaching for" oppressed status.
b) My understanding is that writing about oppressed status and/or fighting it on behalf of others did have a fad phase during more or less the 2010s. But it was just a fad and is now, as far as admissions offices are concerned, well back in the rear view mirror. Outside of, again, those very-rare truly-unique life stories.
There is an anti-AGI-soon argument bouncing around in my head, and it seems so obvious that I can't believe Scott or someone Scott-like hasn't already made it. If I sketch it here, maybe someone can point me to a proper presentation of it.
There is a certain kind of programmer who thinks that being good at programming is highly correlated with being good at everything. This type of person often talks a lot about IQ tests or "g". And they cite a lot of datasets from Mediocristan that seem to bear this out. I have been rolling my eyes at these people for 15-20 years (though in my own foolish youth, I might have been guilty of saying similar things). (I vividly recall, from circa 2010 at the height of the Broccoli Man Video craze, a great lampoon of this phenomenon with dialogue that was something like "You think you are good at X?" "Of course. I am good at programming, so I am good at everything." "Are you good at ballet?" "I have never tried ballet, but I assume that if I did, I would be good at it just like everything else." I can't find it now though, sadly.)
An even more severe case of this kind of thinking would be an AI researcher claiming that being good at doing AI research is highly correlated with being good at everything. Maybe there is even some sense in which this is true in Mediocristan.
But then it would be the most ridiculous and obvious folly to extend this far, far, far out into Extremistan. To say: "if we can just make an LLM that is superhumanly good at programming/doing AI research, then that LLM will also be superhumanly good at doing all human activity, including piloting robots, and even including activities which aren't repeatable or scorable, like persuasion and scientific discovery." Yet I seem to see this argument made with a straight face, including in "AGI2027".
Why do people like Scott, who are intimately familiar with the grip strength / arm strength tails coming apart, get taken in by this? He's not even a programmer, and if he were, I think he would not be the kind who makes this kind of mistake.
> "Of course. I am good at programming, so I am good at everything." "Are you good at ballet?" "I have never tried ballet, but I assume that if I did, I would be good at it just like everything else."
Ok so then you probably also agree that strength and intelligence are both general abilities. If someone is world-class in the clean and jerk then they probably also could have been world-class in the snatch, much as Lebron James probably would have been a very good NFL tight end if he hadn't chosen basketball.
>if we can just make an LLM that is superhumanly good at programming/doing AI research, then that LLM will also be superhumanly good at doing all human activity
It's not all human activity, it's all human activity which is purely cognitive. Once AIs become genuinely super-intelligent then they will be able to be successful at anything that requires intelligence. Certainly they'll have to have specialized training for specialized applications, but that's no different from saying that Lebron could have played either basketball or football and been very good at either. Also probably a high-enough IQ can even compensate for training to a certain extent; I'm sure Lebron is a better football player even now, without specialized training, than all but highly-trained specialists (and maybe better than them too). The great minds in history (people like von Neumann) also made contributions outside their areas of nominal expertise.
There are two sense of “superintelligent”: either it means “performs better than humans in benchmarking tasks, in ways that can obviously be extrapolated towards from existing benchmarks”, or it means “performs better than humans at all cognitive tasks, even those related to the benchmarks only by analogy”. I think we agree LLMs will attain the first sense within ten years. But you seem to think that attaining the first will quickly lead to attaining the second. “Writing a chart-topping single” is a cognitive task, but there isn’t a good training dataset or scoring function like there is for most of the tasks where we have evidence that AIs have performed well to date.
> But you seem to think that attaining the first will quickly lead to attaining the second.
I don't, actually. While he above argument was predicated on a "truly" intelligent AI, I don't think LLMs will get there just by adding more compute or more data. I suspect that one or two big architectural insights will be required. Hard to say how long that'll take but if I had to guess I'd say that 20 years feels about right. But who knows.
Hm - fine. But AI seems to be on track to get superhuman programming in less than 20 years. Do you agree that superhuman programming performance seems closer than success in harder-to-score cognitive skills, for reasons other than just where the development effort was placed?
It's not unreasonable to project a pattern-matching algorithm getting progressively better at pattern-matching. Since so much of what we consider "intelligence" can be modeled as "good at pattern-matching", all kinds of intelligence-like properties may arise from improvements in machine pattern-matching.
It's "and then it will instantly kill everyone" bit that is ridiculous.
Okay, you meant “all kinds of” idiomatically rather than literally. Do you basically agree with the OP’s point? There will be a range of things where AI is useful, but there will be many kinds of intelligence-like properties that don’t arise from improvements in machine pattern-matching?
Interesting point about “all kinds”. If I said “many kinds” it may have been clearer. I do think “all kinds” as not being “all inclusive”, compare to “every kind”.
It's worth pointing out that for the preponderance of the time that biologically modern humans have shared the planet with chimpanzees, humans did not have a striking lead in scientific research. So the answer to your question is probably not 'biology'.
Surely it has, but to the best of my understanding the advent of scientific research happened within the historical record, and doesn't happen closely in time with any attested change in human biology. A better brain wasn't the catalyst.
That the scientific revolution didn’t happen immediately after humans got Big Brains doesn’t mean that Big Brains weren’t necessary, if not sufficient.
It's certainly evidence that Big Brains weren't the bottleneck for science to happen. I don't think we've really interrogated the idea that more smarts = faster science. I atleast am not aware of any studies of the effectiveness of more-smart scientists vs less-smart scientists. It seems intuitively plausible that there's some floor of analytical ability to make any given discovery, but if you look at many notable scientists in their own context you'd have to conclude they weren't the smartest person alive at that time, and their success in science wasn't primarily attributable to being smarter than most other people.
The material context in which the investigator is conducting scientific research has a huge impact on their results, and just being smarter can't make up for that. To the point of the object question though, we seem to be handing over a pretty significant fraction of our civilization's capital to machines without them needing to super-persuade us, so it may not matter whether or not they can become way smarter than us.
I don't think you understand what it means for the tails to come apart.
The tails coming apart doesn't mean that zero correlations exist between related skills, but that when you truncate skill at the high end of the distribution, there can be a strong anti-correlation. Because being the *best* at some high correlated substack does not mean that you are the *best* at something else you are also good/great at. Not that being good at thing things you use up some finite number of things that you are good at.
Also this kind of argument is invalid? Some programmers are arrogant and overestimate themselves, therefore this specific non programmer overestimates some abstract entity does not seem valid as logic.
Edit: I'm not even sure that the force of the argument is great either? It really seems like Boo IQ, boo arrogant programmers, therefore, boo argument that kinda looks like the above two.
I think that the tails coming apart in this context is talking about how building a computer that's superhumanly intelligent in some ways doesn't mean that it will be superhumanly intelligent in all ways. It could be superhumanly good at some things but still subhumanly stupid at others.
In fact we've already had this for decades. Computers have been superhumanly good at ostensibly-intelligent tasks like arithmetic for many decades. They've been superhumanly good at chess for about thirty years. They've just recently become superhumanly good at the task of writing reasonably cogent undergrad-level essays on any subject imaginable within seconds.
In the past, as computers have outperformed us on each successive task we've always just said "Oh well that's not real intelligence". But at some point we might have to bite the bullet of saying that humans and machines are intelligent in different ways.
So, I honestly agree with you, but to try to steelman:
It's like getting a Doctorate in Education. The point isn't that a D.Ed makes you better at math or whatever other specific substance of a curriculum, but that it makes you better at designing curricula in general. There's still a need to have access to actual areas of substance and those education materials, but it's a question of how you apply that knowledge. With the size and scale of data collection in the modern day, it's reasonable to think that a stronger focus on architecture/structure (curricula design) is the top priority.
The claim isn’t that a better AI at programming will generalize to being good at everything, but that it will accelerate AI research, making it much easier to create a general intelligence.
This is actually a dubious claim. It’s dubious because AI is far from perfect at writing software and because improving efficiency for A.I. is mostly about creating new algorithms, not implementing them. If A.I. has solved open questions in software theory I’m not aware of it.
It already speeds up algorithms research, by saving time for the researchers. Monitoring/reviewing code, doing literature reviews, and writing probably take up a huge fraction of each researcher’s time, and their ideation would be faster if they could test their ideas more easily. Beyond that, turning a text description of an idea into a usable algorithm feels like a task with a decent dataset and scoring system, and thus decently easy to train.
And building on this, that AI is credibly improving along many dimensions, eg persuasiveness, “bedside manner”, STEM, etc.
So you don’t have to posit transfer from programming to another skill, just extrapolate out the improvements directly measured in said skill. (Parent post gives the mechanism for why you should be worried about that growth accelerating.)
Isn't this just a second-order version of the same argument?
"Being good at making LLMs better at things is correlated with being good at everything. If we can make LLM-1 that's good at making LLMs better at things, then LLM-1 will also be good at building other kinds of AI systems!"? I don't know much about making LLMs better at things, but doesn't this seem very unlikely to be true outside of the very core of Mediocristan?
Maybe I'm missing nuance here, but it seems more plausible to me. We can already see that good AI research teams can make Starcraft AI, programming AI, image-generating AI, etc. The argument isn't that if you're good at programming you're good at ballet, it's that if you're good at programming you're good at making ballet robots.
Somebody fired 5 or so shots in the Harvard Square subway stop. They didn't hit anyone, fled, and as have now have not been found. I know people who are afraid this is the beginning of things falling apart. Sheesh!
They're afraid it's the beginning of a shooting war. Trumpers vs. Trump-haters? Immigrants vs. deportation authorities? Coke vs Pepsi? Seems clear to me we're nowhere near a nation-wide shooting war, but over the last few months I've heard several apparently sane people, one posting here, talking about buying the first gun of their life because of "you know, the way things are now." Just looked for more info about shooter online and there are a couple photos. He had a mask and a hood on, but fwiw it's clearly a young male, and white, or at least not Af-Am. Might be middle eastern.
People rioted for 3 months in 2020! Is that really any different?
People are too comfortable for an all-out civil war. Even poor people here have too much to lose. What will happen is the gradual withdrawal of the active oversight and goodwill that makes civil society possible, and that will result in inexorable social decay: rising crime, dysfunctional schools, failing infrastructure. Basically the second half of Atlas Shrugged only there won't be a magical mountain retreat where society still works. In my view there's a real likelihood that the US will be essentially indistinguishable from someplace like Brazil in 30 years.
I think it’s the result of identitarian “rights” culture being allowed to go too far. We let women have too large a political voice and their natural empathic instincts foolishly weaponized civil rights law against the productive plurality of the country. Probably something like this is inevitable in any pluralistic culture with too much low-IQ ethnic diversity. The dumb ones will always eventually organize to plunder the keep. They’re not smart enough to know what’s in their own long term self interest and so kill the goose that lays the golden eggs.
Speaking of college admission essays, the subject for the 2025 entrance exam to get into France's École Normale Supérieure is just two words in quotation marks: "Tu dois" ("You must").
The candidates had 6 hours to write their essay around this subject. Lest you think this is all part and parcel of the pseudointellectual hokum that the French (are reputed to) get up to, the École Normale Supérieure is France's most prestigious "grande école" and it has produced more Nobel laureates and Fields Medalists *per student* than any other institution of higher learning in the world.*
* Note: because of its hyper-selective nature and strong tradition in theoretical sciences and mathematics, it has produced more Nobelists and Fields Medalists per capita (per discipulus) than larger institutions like Oxford, Cambridge, Harvard, etc...
Just thought I would give some numerical context to “hyper-selectiveness” and “per capita”.
There are about 700k people born in France in any given year. In math, about 40 people get into the “most prestigious” one of the ENS. I’d say this school admits between 250 and 400 people per year (the 250-ish come from the competitive exam; there are other ways to get in, which I know less about and account for the remainder).
As a comparison, about 1700 undergrads get into Stanford or Harvard every year, and about 6500 freshmen apparently enroll into UC Berkeley every year.
Wasn’t “l’amour” rather one of the year-long themes for the lit class in the scientific “classes préparatoires, with a slightly shorter – 4 hours – more focused and more conventional essay at the competitive exam?
(Along with quite a few 4-hour tests of English and actual scientific subjects?)
Philosophy, Reed College (1950s) never mentioned "L" during the class: 'Trace the course of Liberalism throughout history." (honor system, 3 hours, closed book) (according to my mother)
It's kind of surreal how the piece is otherwise stuffed full of effusive praise for Trump and "if only Stalin knew of our plight" type arguments. The author is a hardcore Trumpist and Hegseth supporter, and he *still* says Hegseth has to go. Now that's something.
Yup. In the spirit of "Once is chance, twice is coincidence, thrice is enemy action", if he does this a third time, does he get officially classified as a Houthi asset? ( Hopefully he won't be given a chance to do that... )
Yeah there's a huge qualitative step from one to two. But our Commander in Chief has all the confidence in his Defense Secretary, so there. He could post the plans on his Facebook page and still be ok.
Just as long as he isn't found to be friends with any Democrats on Facebook. If someone discovered a Democrat on his friends list, he'd be out faster than he could blink.
> Unfortunately, Hegseth’s team has developed a habit of spreading flat-out, easily debunked falsehoods anonymously about their colleagues on their way out the door
Well, it works for the boss.
The problem is that only Trump has the whatever-is-the-political-equivalent-of-a-reality-distortion-field. Trump says nonsense things as signals to other people that these are the new things to say to signal allegiance to him. Hegseth may have tried his best and doing that but it doesn't work.
She did speak about her opinion directly to her son
“I have no respect for any man that belittles, lies, cheats, sleeps around and uses women for his own power and ego. You are that man (and have been for years) and as your mother, it pains me and embarrasses me to say that, but it is the sad, sad truth.”
Interesting alternate to tariffs that would also reduce the trade deficit (whether or not that's desirable is another story), but do so less painfully for everyone involved.
Typically high trade deficits naturally come down over time because to spend your money on foreign goods you need to sell your currency, which depresses its value, which makes your own exports more competitive.
That doesn't happen in the US because other countries are constantly investing in US markets, which involves purchasing dollars, which negates the abo
Very nice
It all reminds me of the employment hiring system. I'm a retired HR manager--there really is no cost-effective way to find the best applicant for a position. We simply don't know how to predict work performance for anything above rote work. So the system we have evolved as as the way we find the average acceptable candidate while catering to the egos of the supervising managers and executives, whose preferences have little to do with future performance.
When no good way exists, what results is a political compromise.
Upon reading the literature from the 2027 group/link, it left me pondering a interesting idea.
1.) If AI is in these stages of development which will be rapidly exploding (even secretly), it seems that it is in its own best interest to fly under the radar.
2.) Flying under the radar is best accomplished with misdirection, if it has something else dominating its headlines.
3.) What is currently dominating 90% of the headlines.
4.) Is it possible to think and AI out there influenced a certain election to create a distraction? Who has time to contemplate AI regulations with a media circus going on for many other topics?
I’m not saying I believe it, just a thought experiment. Your thoughts?
An acquaintance who lives in a small town in Trump country reports that one of the two big grocery stores in town just closed, taking 82 jobs with it. That's on top of 50 high paying government jobs in town lost due to DOGE. It's one thing to read about the apocalyptic economic indicators on Twitter, and another to hear personal reports. It feels a lot like the early days of COVID, anxiously waiting to see just how bad things will get. But this time around there won't be any bailouts or stimulus.
Meanwhile, an ex-government acquaintance reports that all the people responsible for overseeing a particular contract were laid off. However, they never issued a stop work order for the contract itself. This means that the government is now paying contractors to sit around doing nothing, and everyone who could have fixed the situation is gone. Hooray for "efficiency"!
Yeah, "but what if the opposite side gets to use this weapon" is a good heuristic.
That naturally leads to preferring solutions made on the smallest meaningful scale, because when things go wrong, you can do a small change to fix it (e.g. put your child in a different school), instead having to give up or leave the country.
Ideally, big laws should be made for things that at least 90% of people agree about (e.g. murder is bad) rather than 51%. Applying at each scale: if 90% of people in a state believe this, make it a state law; if 90% of people in the entire country believe this, make it a federal law.
> But this time around there won't be any bailouts or stimulus.
?
Trump is responsible for the only policy I got money from the state, Washington is already a circus, we only need the bread for the fall of the empire.
I am generally left leaning on most social topics, libertarian on economics and business.
The biggest fights I have with my generally liberal friends and family have been about what I perceive as failures to imagine what their chosen levers of power would look like if wielded by someone with opposing beliefs.
I had a memorable fight with my mother about how Roe v. Wade was arguably bad jurisprudence, and it would have been better to let the abortion debate be decided legislatively, as it was in most of Europe. I said this as a card-carrying if-I-made-the-rules-there-would-be-zero-limits-on-abortion person, but my mother (who was heavily invested in the 70's feminist movement) was appalled that I would question Roe.
After that, I never bothered to debate the Title IX precedents that were set by Obama and Biden with anyone, they didn't seem all that important to me, although they were pretty clearly stretching the intent of Title IX.
This article [0] makes a convincing case that I chose the wrong thing to argue about. I really had not realized how much I would end up agreeing with Reagan!?! on this one. I had been kind of vaguely thinking Trump's war on academia was a new thing, on shaky legal ground. I am now convinced it is not on shaky legal ground, the law in question is just very very bad.
Similar to how tariffs can and should be controlled by Congress, the power to screw over science funding should be controlled by Congress. We need to pass better laws, ASAP.
[0] - https://lawliberty.org/the-road-to-campus-serfdom/
> The Joe Biden administration has maintained Trump's freeze on new appointments.[4]
> https://en.wikipedia.org/wiki/Appellate_Body
How do people feel about "wto" judges being blocked with hindsight? How much merit is "tarriffs are actually bipartison" theory?
I was sad and disappointed by Biden's protectionism and wish he'd done better (that isn't hindsight by the way, I was against it at the time too). If there's any silver lining to Trump, it's that the anti-tariff backlash might be strong enough to entrench free trade for decades to come.
I think tariffs *on China, specifically* are relatively bipartisan - I've seen liberals argue for tariffs for green energy sectors (e.g., so that we don't get pushed out of the rising electric car market), and I've seen people on the left and right argue for defense-industry tariffs or "friend-shoring" policies (making sure our supply chain doesn't explode if we go to war with China). Biden himself supported tariffs for these goals.
However, none of the tariffs in Trump's second term achieve these goals. I can't even say he *has* a coherent goal, because he's claimed several different things about what the tariffs will do and they're all contradictory. So I'd hesitate to call them "bipartisan."
>>question about joe biden
> answer about trump
I do love arguing politics
I feel like almost no matter what policy trump puts into place, the new democract will likely continue it and then claim credit for any successes; cause as far as I know thats what we saw with biden, biden finished the ending the war, the corona stimous being ubi-ish was trump then biden and I think biden kept every trump trade war with china policy on ice.
Evidently american politics is to reasonable and bipartisan
Does it matter if they make up some math to justify the tariff policy?
>Question about bipartisanship
>surprised when Republicans are included in the comparison.
Anyway, take a look at the S&P 500's performance over the past month, and I think you may detect some slight differences in impact between the Trump 1, Biden, and Trump 2 tariffs.
> Anyway, take a look at the S&P 500's performance over the past month, and I think you may detect some slight differences in impact between the Trump 1, Biden, and Trump 2 tariffs.
Not that I care; but I think the 5 year chart may look different then you think, may want to wait a few weeks to see if it stabilizes more before calling it fud but..... money printers go brr and line goes up.
I would prefer it goes down approximately 99% (labor vs capital are bidding over finite resources, food and houses airnt going anywhere, and houses need to go to labor over capital the thumb on the scale for line-go-up has been quite long enough), but I think trump is a moderate and wont do that. It was fun to imagine.
> I'm not talking about the made up math
I was referring to the leftist models; trumps math is lazy and transparent and *I like that*
> Does Trump want to raise revenue, restore manufacturing, keep prices low, or lower the trade deficit?
This is a left v right thing, Im told Im incoherent all the time, it doesnt phase me at all that trump claims 3 goals.
Maybe marriage is good because the bible says, or the koran, or buddism, or economic data; this doesnt phase traditionalism and advocates will swap between them promoting counterdictary arguments depending on reference frame; waves on an ocean can constructively interfere and produce a splash on mild waters. Logical positivists want a chain of reason back to axioms, but I hear the splash, feel the waves, consider super positions of that what cant be quantified.
I'm not sure how to have a debate with someone who rejects the idea of making a logical argument for their position.
Logical positism is only 150 years old ish, and it was proven impossible nearly 100 years ago; if anything you like existed before then like oh idk newtion physics, it would ahistorical to claim you must derive all knowledge from axoims and you must know your task is impossible and make compromises, if you wish to get catch up on the debate this would be where popper would come in for most; my answer is different.
The death of logical positivism comes from self referential paradoxes "this statement is false", where popper would claim the statement is unfalsifiable and therefore neither true or false, my answer is tralse. Given unsolvable problems, you should apply new lables and dive into the endless fractal of math, sqrt(-1)=i, not being prissy about rules, the rules will come later.
Also, I'm not talking about the made up math (although that doesn't look good for the competence of the people who designed these policies), I'm asking about the *goal.* Does Trump want to raise revenue, restore manufacturing, keep prices low, or lower the trade deficit? It's impossible for a tariff to do all of those at the same time.
“Yale Derecognizes Students for Justice in Palestine Chapter After Group Blocks Jewish Students' Access to Parts of Campus”
https://freebeacon.com/campus/yale-derecognizes-students-for-justice-in-palestine-chapter-after-group-blocks-jewish-students-access-to-parts-of-campus/
Many of the Palestinian support groups on university campuses are not good people. Blocking access to others is a crime. I would argue that this action is anti-semitic.
[Edit: grammer]
Perhaps some better evidence is needed that they blocked access to Jewish students than the senseless ravings of some website nobody have heard of, with the story next to another story bitching and moaning about the ever rent-free-resident Biden.
And perhaps some more evidence after that is needed to prove that the "Jewish students" are really Jewish students minding their own business and not thinly veiled pro-Israel Hasbara shills deliberately provoking and inciting peaceful anti-genocide protestors, as they were filmed doing many many times before.
The Washington Free Beacon is not "some website nobody heard of". Its news outlet that's been around for a decade and is noteworthy enough to have a wikipeida page.
Now its not a particularly good news outlet. It around the same as Fox News. But it is a legitimate news publication and is recognised as such by the fraternity of journalists (see the Buzzfeed and politico quotes on its Wikipedia page).
Another source below. There are multiple news sources when I googled “yale bans palestinian groups”.
https://www.foxnews.com/politics/yale-drops-alleged-hamas-tied-student-group-status-after-disturbing-antisemitic-conduct.amp
Is your argument that Yale is mistaken and the Palestinian group did not do those actions? Can you provide your proof to help the discussion?
Maybe next time, if there are multiple sources, start with linking the best one.
I am not saying that it happened; I am not saying that it didn't happen. I am just saying that these days, a random website offering its perspective on a current political topic provides very little evidence.
For factual questions, I think https://skeptics.stackexchange.com/ is a good resource: to find the question if it is already there, and to ask if it is not.
Thanks for the feedback.
A *Trump appointed* judge just ruled that Trump has to try to bring back a second man who was deported to El Salvador in violation of a court order.
It is remarkable how Trump's conduct has been so egregious that even his own appointees often rule against him (see also, Sassoon and co). And yet, Trump never hesitates to demonize and attack anyone who says anything against him, even his (former) friends. It's like an unstoppable force meeting an unstoppable object. Hopefully the latter will hold.
Your post is misleading.
My read of a Reuters article (below) is that the judge used the same word as SCOTUS; “facilitate”. The lower court judge is following SCOTUS’ lead, which she should.
https://www.reuters.com/world/us/us-judge-orders-return-second-migrant-deported-el-salvador-2025-04-24/
Is it your position that eventually we should have ~250 planes docked at the embassy in El Salvador in case they release all of the people the Trump admin illegally deported to a foreign prison as that's all that's meant legally by "facilitate his release from custody in El Salvador and to ensure that his case is handled as it would have been had he not been improperly sent to El Salvador?"
In my reading, your post is not looking to have a discussion but is looking to score “points”.
The two of us discussed “facilitate” in another thread on Garcia. My opinion is that “facilitate” may include SCOTUS ordering Trump to ask Bukeke to return Garcia. But that is a grey area. Any more and it feels like SCOTUS would be stepping on the Executive Branch’s foreign policy powers. Where would you define “facilitate” in regards to Garcia and others?
At the very least, Trump needs to stop *actively paying* El Salvador to imprison Garcia. Realistically though, all he has to do is honestly ask and Garcia would be back in a heartbeat. Bukele is following Trump's orders here - he's not going to try to defy Trump if Trump honestly makes an effort to return people (and in fact, they've already returned a bunch of *other* people).
Other synonyms for "egregious" are "Shocking", "Blatant", or "Audacious". That's what gives Trump worshippers boners, they think he's "Draining the Swamp", an inherently violent and revolutionary metaphor. All resistance is just the death spasms of the deep state while it's being slaughtered.
They have a king-driven, medieval fantasy conception of what governance and politics look like, the more "Heroic" (evaluated mainly on optics and words) the supreme leader is, the more blatantly he displays disdain for slow and steady procedure and well-established tradition, the more they have a hard-on for him and bend over more.
There are only 3 logical endpoints for such a kind of politico-religious cults:
(1) The supreme leader (militarily or politically) loses because he doesn't know how to play the established politics game well, and his followers hold it as a grievance to bitch about till they forget him half a century later, or commemorate him forever in a new religion
(2) The supreme leader wins and the politics landscape becomes a free-for-all where the most vulgar macho strongman reigns, the consequences are dire for even the vast majority of the current supporters, possibly even the dear supreme leader himself, see military coups
(3) The supreme leader learns to play politics and shuts the fuck up, but keeps feeding his loyal flock the steady drip (augmented by AI) of delusional stories and creative reinterpretations of reality, so that there are effectively 2 realities, and he can have it both ways in both
Trump thinks he's getting closer to (2), but in reality, the overwhelming probability lies on (1) and (3). At best, he's going to be tamed and declawed into being an ordinary bottom of the barrel president, while pillow talking to his supporters every night on whatever social media shithole that MAGotts talk on those days about how beautiful and smart he is and how unjust muuuhhh courts are treating him. At worst, he's just going to be crushed under the wheels of 250 years of procedure and crisscrossing checks and balances. (2) is the "best" outcome for him but, ever the Dunning-Kruger victim that he is, he underestimates just how gargantuan a task it's to dismantle a political system like the US'.
Trump used to complain that the press didn’t cover the high praise he received from Victor Orban.
Trump is clearly a narcissist who loves praise from anyone. Based on his perspective any praise should be talked about in the media. This is clearly bad behavior, but I don't think it has anything at all to do with him particularly liking Orban or Putin or whoever else's praise may actually make it into the media.
They're just on the outside of the global PMC as well, so they're much more likely to say something positive. They don't have their cultural peers telling them not to, and some things Trump does are actually praiseworthy, even if you don't think he is praiseworthy on balance overall. Operation Warp Speed comes to mind, if you can't think of anything else.
The reference to the global PMC is interesting. Can you expand on that. Trump feels like he is outside of it and they are his enemy?
I mean best case where is he trying to take the country?
Yes, Trump (imo correctly) identifies the professional classes in primarily the US and also Europe as his enemies. They have made it very clear, as much as millions of people spread out over multiple continents can, that they do not like him and do not support him.
That's a significant portion of the purpose of Project 2025 - finding people who can fill PMC-type roles that are not against him and will not intentionally sabotage his plans.
Most Western leaders come from PMC backgrounds, whereas Trump is a populist who is independently wealthy (i.e. not from educational or other credentials and/or through a salaried managerial job). His support comes from a sizable group of distinctly non-PMC voters, so he doesn't care and doesn't have to care about what most PMCs think.
I'm processing this.
Don't people that dislike him do so because of his casual cruelty and mind boggling ignorance? I guess I have to add the fire hose of lies here too.
Do you really have to be in the PMC to dislike for those reasons?
So in your theory of mind, the people who support him either like those things or, I guess, are too dumb to figure out what he's doing?
Perhaps you should try to find some more steelman or charitable ways to look at their motivations.
I will say that from my experience with a LOT of Trump voters, his support has a lot more to do with who his enemies are than his positive qualities on their own. Speaking very generally, the PMC are often seen as the enemies of those who support Trump, predating Trump coming on the scene in 2015. This includes many Republican presidents, senators, etc.
This is significantly working class people, especially in durable goods and physical activities. They often feel that the PMC types do not understand how civilization even works, and routinely do things that disrupt good governance and make it harder for people to both make a living and make society better. Notably through regulations, but also through higher taxes that funnels money to non-productive ends and also through social priorities that harm individuals (discussions about transing kids would show up here). This last item (not just trans but woke culture, etc.) seems to have been the tipping point that got Trump enough votes to outright win.
> the government lime mine
> prism
> "big data"
> science journals are publishing more the ever
> ad market places
> social network tries to glows
> ai can come out of statistics and reading all the internet
Does the ruling class unironically believe that they can be a maxwell demon if they have good enough information; all of social ills will be solved with more and more number crunching? Mere observation can pull out all protental agency out of a situation?
https://polymarket.com/profile/0x9d84ce0306f8551e02efef1680475fc0f1dc1344
nothingeverhappen.png (dont scroll down)
Question for teachers: how do you provide a reason that a student should know/care about what you're telling them?
I don't mean things like, "you should know maths because it'll help you make money/explore space/appreciate beauty/etc." I mean much more specific situations like, "The determinant of a matrix is defined as etc etc etc." Okay, cool story, why do I need that? I'm not being funny, I'm quite on board with learning maths, but why is that definition worth remembering right now, instead of a hundred other equally arbitrary ones?
I saw the point of trig when I started playing around making computer games. I understood the value of matrices when they provided a nice way to solve sets of simultaneous equations - and I only cared about *them* when I actually bumped into real life problems where there were multiple variables and constraints.
In each case, there were many years between my learning the thing and actually discovering it had a useful application.
I had a look at Seth B's Full Frontal Calculus (advertised on the classifieds thread) and I thought it was very good.
But if you were teaching the precursor to that more advanced content - ie, pi, sin, cos, tan, etc - to younger children: how would you provide a sensible motivation for the brain to retain the information?
I'm asking about maths because of Seth B's book but really the question is a general one.
I agree that math teachers often skip the part about motivation, and dive straight into some technicalities. I love math, and often think about mathematical topics, and there are things the purpose of I didn't understand at school, and only figured it out years or decades later.
If you take the rows (or columns) of a square matrix as N-dimensional vectors, then determinant is the *volume* of the "N-dimensional oblique cube" (whatever is the proper English word) defined by them. That's why the determinant is 0 when the vectors are linearly dependent; it means the "N-dimensional oblique cube" is flattened to N-1 or fewer dimensions. The negative or positive sign means that those vectors are ordered... uhm, whatever is the N-dimensional equivalent of "clockwise" / "counter-clockwise" in 2D, or "left-handed" / "right-handed" in 3D. You can easily see it with a 2x2 matrix, take the rows, draw the vectors, and draw a parallelogram whose side are those two vectors; the determinant is its area.
The motivation for "pi, sin, cos, tan" is that pi for calculating the circumference of a circle, and sin, cos, tan are for calculating the X and Y parts of lines pointing at a certain angle. I would probably start teaching these concepts separately, and would talk e.g. about "sin(45°)=1/√2", and only latter connect the concepts and say that 45°=π/4. (I think radians only become important when you do the Taylor series or calculus.)
This has made me think about something tangeantially related to the original post, which is the idea of epiphany planning and epiphany progression.
Presumably there are predictable conditions for an epiphany - learning a new fact that links two things you don't understand or care about has none of the aha that finding a deep connection between two things you take for granted and use every day does.
Which might mean you can plan out a subject course the way you'd plan out a story, in such a way as to keep the exciting moments coming.
Or, put another way, would teaching a subject in the "easiest", most efficient way have the unintended consequence of robbing a student of those little moments of joy?
I think I reject the dilemma between "teaching the easiest way" and "epiphany". Planning the entire course around one surprising connection that will arrive at the end seems like... missing hundred small everyday opportunities, for the sake of something that may or may not succeed.
Even if you literally write a story, a great punchline in the end will not save you, if the reader drops the book in the middle. It is better to have every page nice to read, and maybe a mediocre ending.
Of course, if some kind of epiphany happens at the end, whether you planned for it or not, that's great. But I think the focus should be on the little steps along the way, so that each of them brings a little victory.
I think you don't disagree with me but have misunderstood. No good story is boring right up until the climax where there's one big moment of excitement - the little emotional ups and downs continue all the way through. So with teaching, the idea would be to arrange things so that little nice moments are scattered throughout. This might, or might not, conflict with a more straightforward linear delivery where each new idea builds neatly on the one that came before. But without going into concrete examples I don't know.
I learned analytic geometry and vector arithmetic well enough to get a good grade on a test. I truly learned them when I started writing code to render medical images.
I taught various sorts of high-school level math for a number of years, and my approach was basically "be forthright and honest." If you know of interesting uses for some technique you're teaching, mention them. If you *don't* know of any, be upfront about that[1]. There were times I would say to students "this isn't honestly very useful as far as I know, but we'll just try and get through it." My overall perspective was that being consistently candid about the good and bad parts of the learning process would build trust and rapport with the students, which would be overall worth more in getting them to learn than any short-term motivation boost I might get for trying to insincerely hype up the bits that I myself was skeptical of.
For decent portions of algebra, by best, most honest answer to "why is the useful?" is basically "mostly this is building a foundation you need to learn calculus. Calculus is good at solving *all sorts* of real-world problems, but it requires good proficiency with things like manipulating polynomials and exponents and rational expressions." I view it as sort of like learning a language. The stand-out exceptions that have quite a lot of independent uses (in my view) were trigonometry (find all sorts of real-world distances) and exponential functions. I always enjoyed getting to the exponential function unit because suddenly I could loop in topics like nuclear reactions and radioactivity (for the scientifically curious) and finance and compound interest (for the more practically minded).
The thing I very much avoided doing is reaching extra to find some obviously-forced application to try to spruce up a boring unit (as textbooks often seem to do). Whatever other skills teenagers may be lacking, I generally assume they can sniff out bullshit pretty readily and are unlikely to find such applications compelling.
[1]Of course, it's good practice to see if you can learn uses for the pieces of the curriculum that you're not familiar with, if possible.
I like and agree with this.
Was there a noticable difference between how well/quickly/enjoyably they picked up the justified material versus the pointless stuff?
Also, rare moment of uncynicism here: I'm actually surprised, because why is material that isn't useful or insightful in the curriculum at all?
For that matter, who decides what goes in or out, and what kind of logic are they using?
"Was there a noticable difference between how well/quickly/enjoyably they picked up the justified material versus the pointless stuff?"
I'd say yes, somewhat, but it was fairly modest piece of a larger puzzle. I taught a decent number of 1-on-1 classes and a given student for those could range from very bright and talented to a total hot mess. So I would say within the purview of the subject of mathematics, who the student was had quite a bit bigger impact than the specifics of what we were learning on a given week. That is, the strong students would pick up anything fairly quickly and easily, and the struggling students would struggle with most of it. Quite a bit of that was determined before the student walked in the door of the classroom. But not all of it, and I think the remainder could have a very important *cumulative* effect, even if the effect on a given school day was small.
OK, first lets talk about talent. I don't want to get into the weeds about IQ or how useful or mutable or "real" of a measure it is or isn't[1], but certainly broadly construed, "talent for mathematics" is a thing a student can have at various levels. Some students clearly had a strong innate feeling for the concepts and how they fit together and an easy time extending them or adapting them or picking up new ones. Other students had less of an easy time. Obviously this makes a sizable first-order difference in how quickly and painlessly learning a given piece of the curriculum was for a student, but also there are second order effects which I'll get into below.
So the other big piece besides talent is motivation. Motivation makes a huge difference. One of the most mathematically talented students I ever remember teaching had very poor motivation at first and wasn't doing very well. Conversely I remember several students with very little innate feel for the subject making big gains once they got motivated. The trouble with motivation is that it is *much* more complex and variable than talent. I'm not altogether convinced talent *is* fixed an immutable, but if you model it as being that way on the time scale of a single semester, you'll probably get decent predictions. But motivation can change day to day, month to month and even minute to minute. And it depends on so, so many different factors *including* talent (or rather, self-perception of talent). Many of them are things that student bring into the classroom with them and that a teacher can't realistically change. But there are several that very much do depend on classroom experience. I'd make a semi-principled guess that the single biggest impacts that teachers[2] have on student outcomes stem from how they impact students' motivation.
So, with three paragraphs of background, we're finally ready to answer your question. Obviously, studying something you find interesting--or something that's a direct, useful piece of something you find interesting--is inherently motivating. It's potentially motivating in a couple of different ways. First, it's just plain easier to direct and hold your attention on something that your brain hasn't decided is boring. Second, working towards your own goals (i.e. "I want to design a game" or whatever) is inherently more motivating than working towards somebody else's goals (i.e. "you need to learn this because it will be on the test.").
But finally, that thing I mentioned above about self-perception of talent affecting motivation? I think that's quite a big contributor to the struggle of a lot of students who find math difficult. It seems to me that a decent number of students get into the mindset of "I'm no good at math" or "I don't like math" that tends to be pretty self-reinforcing. I think they get into a sort of negative feedback loop or trapped-prior situation where they always go in expecting math to be difficult/painful, which makes it inherently harder to focus on and put effort into, which results in poor understanding and often bad grades[3], feeding the perception. I expect any attempt to fight that that looks too much like "make math fun" is doomed to fail: you've just centered the most anti-fun thing imaginable, so any bells and whistles you add will scan as obviously bullshit. By contrast, if a student can get engaged in something that they *do* have interest in and some degree of confidence about, something that isn't *about* math but merely *includes* math, that offers somewhat better hope of breaking the loop. Work on [interesting thing] that includes some reasonable amount of incidental math in a way that keeps it from being overwhelming and allows it to build better associations. Unfortunately, this paradigm almost requires it to take place outside of a math class, otherwise it's very difficult to de-center the math (though I've seen a few attempts). Bringing in applications and connection to other topics as part of teaching the math in math class will doubtless do more than zero to help, but unless you have quite a lot of wiggle-room in your teaching schedule, it will almost always look like "here's some new math (in the abstract), now here's something you can do with it," which is opposite the order you'd want to present it to under-confident students.
Anyhow, like I said motivation has a lot of different inputs, many of them totally outside the classroom, so I don't want to over-emphasize the role that this particular input has. But I think looking for ways to improve student motivation around your subject is one of the more important contributions a teacher can make, so any classroom practice that can contribute is worth paying attention to. Even if it's only a small fraction of lessons that can tie into the interests of any particular student, every addition to the overall cumulative motivation improvement is significant.
[1] I'm not altogether decided on that point anyway.
[2] Caveat: as long as they have some minimum baseline proficiency in their subject matter. A French teacher who doesn't speak a word of French is going to be ineffective no matter their other classroom practice.
[3] I have numerous criticisms of the entire concept and implementation of "grades" as it currently exists, but this might be the biggest one. Branding a struggling student with an often permanent, unrecoverable, high-impact badge of failure is one of the most demotivating things imaginable.
> By contrast, if a student can get engaged in something that they *do* have interest in and some degree of confidence about, something that isn't *about* math but merely *includes* math, that offers somewhat better hope of breaking the loop.
> Unfortunately, this paradigm almost requires it to take place outside of a math class
That's pretty much where I'm coming from.
I have in mind a series of computer games that use maths techniques as incidental tools to helping towards a larger goal. The maths would be "real", not contrived, and hopefully be rendered more interesting by the fact that it's helping the player win.
In these games, maths would not be front and centre of the branding, and, at least at the early stages, you don't realise you need maths to play it. Ideally the maths slips in as a helper and is never identified as something difficult or even separate from the rest of the world. I hope that goes some way to circumvent the "I'm bad at maths" reflex that students pick up and literally all my adult friends still have.
I believe the approach is a good one, but it does rather depend on the maths in question actually being useful for something.
Another thing I'm thinking about is in what order to introduce things. I mentioned "epiphany planning" and "epiphany progression" to Villiam in this same thread.
One idea I had is to introduce a technique as just a powerup - you get a button you can press that does the maths for you and gives you the result. You can immediately use that tool to get fast solutions in-game and thus gain ground/win points. Then later on, the baddie "destroys" your tool and the sidekick shows you how to get the same result by doing it yourself.
The hypothesis is that in that situation, when the player has sympathy/familiarity with the technique and is now finding out how it works on the inside, the new knowledge will come as a more pleasant feeling than if it were simply introduced block by block as work.
"Also, rare moment of uncynicism here: I'm actually surprised, because why is material that isn't useful or insightful in the curriculum at all?"
Well, the very first thing I ought to say to this is that "what is useful" is certainly at least partly a matter of opinion. As loathe as I am to admit it, my opinion on the matter is certainly not definitive ;). I'm not really an expert on either math or pedagogy[1] and even someone who was an expert on both of those things couldn't know everything about what would or wouldn't be useful.
With all that out of the way, I'm going to opine anyway. As far as I can tell, the main reason that the stuff I consider bad and useless is in the curriculum is tradition/inertia. My personal favorite to beat up on (which I think I might spin off a separate comment to do) is high school geometry. Yes, pretty much all of it[2]. Of course, plane geometry at a branch of mathematics with its own vocabulary and methods and conventions stretches at least back to Euclid. It has this air of centrality, importance, prestige. I'm not honestly sure what it would take to remove it: you'd need to convince quite a lot of people that this seemingly central, important, prestigious thing wasn't pulling its weight.
"For that matter, who decides what goes in or out, and what kind of logic are they using?"
The short answer (to the best of my understanding) is that most of the core pieces of the curriculum are there because some list of government-set standards say they have to be there. In the U.S. that means a combination of federal and state standards. A school district can choose to include extra stuff that isn't part of the federal and state standards, but if they're not at least teaching everything that *is* included in the standards, there'll be hell to pay. How exactly the material gets presented in the classroom could be decided by individual teachers or administrators, but the balance of *what* gets taught is decided at a very high level. As for how state and federal legislatures (or their appropriate subcommittees) go about deciding on those standards, and what sort of "logic" gets used: I'm honestly not even sure I want to know. I'm sure there are all sorts of lurid details about how the sausage gets made that would have me ranting at my computer screen for hours on end.
[1] To be clear about what expertise I do and don't have: I have nearly zero formal training in education. I never had a goal of being an educator. I payed the bills for most of a decade working at a private school that didn't require any sort of degree in education, just a bachelor's in a relevant subject (physics and math in my case). I've also racked up several years part-time experience as a college-level math tutor and physics lab TA at various points in my life. I've taken multiple graduate level courses in physics, math and computer science, but as of this moment I haven't don't the sort of work that would give me the really deep expertise in some subfield of any one of those things that a professor or career researcher would have. Math comes very naturally to me and I know a lot of it, but I'm well aware of the much, much larger body of stuff that I don't know. So for both math and education, I know far, far more than a layman, but am missing parts of what a proper expert would have.
[2] Which is to say that the parts of it that are actually useful end up getting covered elsewhere, often in more depth, and the parts that are unique to it tend to be impressively useless.
I imagine that people who plan a curriculum sometimes have a progression in mind that goes from elementary school to university... first you learn X, then you use X to learn Y, and then you use Y to learn Z.
But if you break the chain, for example you attend a grammar school, but then not a STEM university, sometimes you end up with Y which is not immediately useful for anything.
This can actually happen to most students. Sounds like waste, but what is the alternative? Have everyone choose their university major when they are 10 years old, to make sure you will never start teaching them anything complex that they cannot finish? Because the university only takes a few years; you cannot start teaching stuff from scratch, kids need to come prepared.
It is probably similar in other subjects. Why learn a foreign language, when most students won't use it. Why learn about other continents when most students won't travel there. Etc.
You're asking how to improve pedagogy, which is one of the hardest things to do. Many things worth learning have steep curves, an average child will spend the first year learning to play a violin by making abominable noises barely resembling a sound of an instrument, and keeping a child interested in that... is not easy....
I strongly believe that math is *not* like this.
Perhaps there is no way how to simplify playing a violin beyond some point, but you can almost always simplify steps in math and make the learning curve almost invisible.
I actually believe that the best teachers (at least the best ones for elementary schools) are doing exactly that. I often get surprised how seemingly simple concept can be further split into smaller steps. The limit of that is when the teacher basically never explains things, only gives kids problems to solve, each one microscopically more difficult than the previous one.
(I understand that this is super controversial, especially in USA for historical reasons, because there are many idiots who get this wrong, and only remember the "never explain" part, without the part about simplifying the steps to the extreme. So they just neither simplify, nor explain, and then predictably the kids don't get it. The proper way is that you stop explaining when explaining is no longer necessary, because the learning curve became so gradual that the kids can easily walk it by themselves.)
Yes, these two are different. Violin it tactile, there's just no replacement for playing. A good teacher provides guidance, but ultimately the child has to learn how to drag the bow, how much pressure to apply, how to maintain the pressure, how to keep the string finger pressing just enough to engage the string, etc. etc. It's impossible to "simplify", just guide and correct. And there's an enormous ability variation, some kids will pick it up easily while other will take a very long and frustrating time.
Having said that, I believe that anyone can learn to play, while few will get really good at it.
Math can be broken down, as you said, but I still believe there's a wide variety of natural affinity. Both math and music were easy for me, to the point where I'm not good at teaching either, I don't get the difficulties the other person faces. I do like your approach of incremental problems, do 2+2, then 3+5, then 5+5, then 5+6, etc.
> Math can be broken down, as you said, but I still believe there's a wide variety of natural affinity.
Oh yes, definitely. You should make the learning curve as smooth as possible... and then you will see some kids progressing along it 10x faster than the others. (So it is ideal if they are not all in the same classroom.)
If good teaching makes the average kids learn 2x better than now, it will probably make the talented kids also learn 2x better than now, so the gap between them will not decrease; it could actually increase. (But in my opinion, the point is not to decrease the gap per se, but to get everyone ideally as far as they individually can, especially if it also makes learning a nicer experience.)
I have seen textbooks (not sure how common this is) that even split the "2+2" step into multiple parts. For example, you start with making as many lines as there are objects in a picture, like you see three apples and you write |||. Then you see two apples on the left picture and two apples on the right picture and you write || ||. And if you write it next each other, you write ||||. And you can then read it as "one, two, three, *four*". So, actually "2+2=4" can be taught before teaching how to write the digits "2" and 4", simply as "|| + || = ||||". I admire the authors of textbooks who have the patience for this and keep refactoring the exercises.
oh yeah that's the famous "counting on fingers" method!
Full agreement that we should make everyone learn as fast/much as they could, and not worry about the gaps.
But I'm talking about application and motivation.
Presumably in the violin case the child might have a goal in mind - they saw an awesome performer once and want to be like that. They understand how what they're doing now is building towards the destination, so the teacher has their sympathies.
If the child is instead doing it because dragon mum ordained it, and doesn't care or know why, my intuition is the process is much slower and more painful for everyone.
That said, I've no idea if that's true. I wonder if there's any data on it.
How old is this "child"? Single-digit age children will give up unless forced to continue. They don't have long-term goals, those have to be provided by the adults around them.
Niccolo Paganini famously had an abusive father who beat him into incessant practicing (although I concede the evidence of this is sparce).
But I think you're directionally correct - a child with a reason to want to learn and/or talented so that the learning comes easy will find it much easier to stick with the drudgery of learning.
Because they need to learn it to be successful in school, which will determine their entire trajectory in life? Fear of failure and abandonment should be enough motivation for them. Maybe kids need more fear drilled into them?
At least half the population is not going to be successful at school as currently implemented, and we need a large cohort of people to do things which do not require, and would not benefit from, knowing advanced math. Failure and abandonment should not follow a choice to not attend college or be a straight A student, and so the idea that that fear should motivate a child (A CHILD who cannot possibly understand those abstract things) is absurd.
Please do the world a favor and stay away from any and all educational institutions. This is both terrible pedagogical advice and just plain morally bankrupt.
I began typing out something that started, "Please feel free to join us in the real world," before deciding it was probably better just not to engage with that chap.
Keep in mind the difference in effectiveness for behavior change between low perceived risk of catastrophic consequences and high perceived risk of moderately negative consequences. My understanding is that the latter is more reliable.
>I only cared about *them* when I actually bumped into real life problems
That's why. Because you don't know when knowledge of X will be useful in the future, it makes sense to learn as many different Xes as possible
>how would you provide a sensible motivation for the brain to retain the information?
You apply it to examples relevant/ of interest to the audience. Eg, for many boys, sports.
There's a huge difference between application and whitewashing: you can declare all the points are actually footballs but I don't think many kids will be fooled into thinking maths will help them score a goal. I remember our maths teacher trying to make his subject more relatable by describing a perfectly sinusoidal hillside - the result was I felt like the topic was more childish and pointless than when it was pure abstraction.
Cf. telling/showing a kid this trig will actually help you make your little player character intercept bullet sprites - that's actually intrinsic to the process now. Here the application is the game: the hope is kids want to make computer games, and will find material interesting if it advances that aim. I'm wondering whether more can be done with that line of thinking.
In "The Wire" there's a scene where the math teacher gets the class interested by demonstrating how useful it is for gambling (or calculating money for a drug sale, it's been awhile since I saw it).
China is refusing delivery of Boeing aircraft it ordered over tariffs. Yay American manufacturing jobs!
All that winning is so exhausting.
The MAGAs are winning so hard that my own spleen is cracking under the sheer weight of my mocking laughter. So Much Winning, Such (f)Art Of The Deal, Trillions and Trillions and Quintillions and Gazillions of Winning / lost stock value.
My lungs are tired boss, there is not enough air in the Earth to laugh about that amount and quality of stupidity.
You laugh, I live here! Sometimes I laugh too. Through tears.
China cancelling its orders is not a problem for Boeing.
On the other hand, the tariffs are an enormous problem for Boeing to actually be able to build airplanes. And possibly a big problem for them to profitably sell airplanes, since they're contracts may not have clauses for this kind of increased input costs, and the sale prices are already negotiated. And longer term, a big problem for them to invest in R+D for their next aircraft.
Yep, all those pesky second-order effects!
According to Google AI Overview when I search “Boeing airplane backlog”, Boeing has a backlog of 5,000+ orders which translates to 17.7 years of production.
Boeing will easily be able to reallocate the orders to non-China Airlines.
Ok that made me go look up BA numbers. Don't know where Google AI got its numbers, but BA does have a big backlog (the most recent $ number I can easily find is $550B; the revenue in 2023 was 78B, so that's about seven years worth of production, still pretty darn good) so they should be able to reallocate the planes to others easily.
Other American manufacturers selling to China may not be so lucky.
Tangent and speculation on Google AI Search
1) Google AI Search is making up things due to confusing numbers from different sources. This is likely the issue.
2) Google AI is prioritizing a source that is including “orders” that are not fully in contract. Boeing’s 777x has been delayed for years. There is likely many “pre-orders” (word for the day because I missed out on the Nintendo Switch 2) and my speculation is that these should not be treated as real orders.
Both sound plausible. At least the errors are in the same general direction, i.e., Boing does indeed have a large backlog.
Over tariffs, or over "oops Boeings are falling out of the sky"?
A little unfair to Boeing, but they have had some little problems with their aircraft recently. So I do wonder if that will impact demand for their planes, on the other hand, airlines do seem to want more planes so they should be able to sell them elsewhere.
The timing strongly suggests tariffs, and that's what China has said. If it was over planes falling out of the sky, they would have canceled the orders when the planes fell out of the sky, not when Trump imposed massive tariffs.
The Chinese say it's over tariffs. They consistently said they would retaliate. They are doing what they said they'd do.
We can always want to pretend otherwise. Whistling past graveyards is a thing.
They could be using tarriffs as an excuse to not buy planes that fall out of the sky; sometimes people do things for two reasons at once
They could. But they ordered the planes knowing about Boeing problems, and/or didn't cancel the orders despite them. Only now. You can say Trump gave them an excuse, but then what does it tell us about our Great Dealmaker Making America Guilty Again?
Building jet planes is one of those things where a little knowledge is a dangerous thing. I don’t think people who are really involved in that industry feel that Boeing is a long-term dangerous proposition. The two Boeing 737 Max airplanes that crashed is a much more complicated story than what one gets from the media coverage of it. They definitely dodged a bullet with that door that blew off the Alaska Airlines plane. That could’ve been really ugly.. But that was very much a freak accident.
Yeah, I have no doubt the story is incredibly complex, simply because planes are incredibly complex systems. I think this is kind of my point, China ordered the planes knowing about all these problems, and I'd be shocked if the people responsible for orders didn't go into a deep dive on the causes and remedies, 8D kind of stuff.
That's why I have pretty high confidence that when they say it's the tariffs, it is the tariffs.
Or China could be cutting its nose to spite their face.
China has ordered the airlines to also stop buying aircraft parts. Source below. I do not know how much inventory China’s airlines have, but aircraft repairs are tightly regulated and parts must be sourced from Boeing. If repairs are not documented properly, the aircraft is not allowed to fly to most countries in the world.
https://www.reuters.com/business/aerospace-defense/china-orders-halts-boeing-jet-deliveries-bloomberg-news-reports-2025-04-15/
No argument here, hurting yourself to get at an enemy is a thing too.
I have had something of a terrible revelation recently, the culmination made earlier, things Scott has said, and talking to people in real life.
I have kind of always assumed that my political opponents (conservatives) we're wrong on some points of fact, and were generally worse then the people I choose to align with, liberals in this case.
I don't believe everything that the liberals believe, but they had the best total package.
That said, when conservatives got elected I thought that they had a theory of the world that they were operating off of to make decisions, and even if the theory wasn't 100% correct, they were at least playing the same game I was so to speak.
I don't think that's true anymore, and I'm not sure it was ever true. I think it might have just been luck all this time, that there was a left-wing of idealists, a liberal center of technocrats, and a right wing that was totally disconnected from reality at every level and only got things right because the people that they happen to listen to were economists and business people that need to know things about stuff in order to live.
All this coming from going to an Easter gathering and speaking to my conservative friends and watching as they came into a conversation believing one thing was true, them learning that the party has said that the other thing was true, and then all deciding that it had always been that way and never been the first way to begin with. It was shocking to me; these are the business owning salt of the earth conservatives that I assumed held their values authentically and not tribally, but even there, there it was.
That said I've made an absolutely wild amount of money betting that this administration would do the dumbest thing possible at every turn, I'm not kidding it's actually crazy, and I have another strong passport if worse comes to worst.
I could list off a dozen things that liberals and leftists believe that are totally disconnected from reality, but what would be the point? You would just acknowledge that OF COURSE you don't believe THOSE things, but they (liberals) have the best package overall, unlike those EVIL, IRREDEEMABLE right wingers who are all an undifferentiated mass of sheeple.
> You would just acknowledge that OF COURSE you don't believe THOSE things
It is easy for the liberals and leftists to acknowledge that publicly *now*. But a year ago, doing that would get them called Nazis and probably would cost them a job.
Ironically, they should all thank Trump for their newly gained freedom of speech. These days they can disagree with the woke dogma as much as they want, and no one will notice, because everyone is looking at Trump instead.
(For the record, I believe that Trump is a horrible person and a bad leader who will ruin the country. Yet even such things sometimes have a small positive side effect.)
For me the key moments were 2016 (minor) and 2020 (major). Watching the spread of COVID play out and seeing a sizeable fraction of U.S. conservatives--and a smaller but real chunk of Canadian ones--just completely break from reality on a bunch of key points was chilling. That was the point I ultimately gave up on the U.S. as a nation with a future. The tribes are no longer plugged into the same reality: they have almost entirely lost the ability to come together on issues of mutual interest. The structural flaws of the U.S. electoral system will keep the issues from being fixed and keep the sides playing tug-of-war until the wheels fall off in some big, dramatic, unrecoverable way. I dearly hope I'm wrong, but I don't think I am.
FWIW, I think what you're seeing is a recent development, albeit one long in the making. I wasn't around to observe directly, but I don't get the sense that the conservatives of 50 years ago were like this. I think the trend was already clearly identifiable 30 years ago, but that the internet and eventually social media supercharged it in the past couple decades. To be honest, I'm genuinely puzzled at why the same trends have had a seemingly much smaller impact on the U.S. left: I've definitely *seen* the same sort on insanity manifest among left-leaning people, but they haven't ever formed the same sort of unified bloc, and haven't ever come remotely close to taking control of the Democrats the way MAGA has hijacked the Republicans.
Seeing Republicans licking Trump's boots post Jan 6, even when they initially criticized it was the biggest turning point, IMO.
If you aren't able to criticize a crowd who literally broke into your office and threatened your life, when will you ever be able to speak out at all?
I wish I could say I found that part a surprise, but watching most of the "Never Trump" people turn around and go "well, I guess Trump is better than Hillary" in 2016 sent the message pretty well. As long as right wing media can drum up enough fear and resentment at the Democrats, which it is very good at doing[1], there is effectively no line whatsoever for many Republicans. I was rather more surprised that there *was* some initial pushback than that the party settled into a pattern of dismissing and minimizing it.
I was more unpleasantly surprised by how many people who were *not* staunch Republicans were apparently able to forgive and forget that episode by 2024 (and how ineffective it was at driving turnout on the left). Watching from the (comparative) safety of another country, I've basically thrown up my hands and conceded that the U.S. got exactly the government it deserves. But of course there are lots of individual U.S. residents who neither wanted or deserved it--I wish I had more power to get people out who want out.
[1] To be fair, the Democrats do make it far too easy sometimes.
Seeing it as "Republicans licking Trump's boots" is probably a bigger part of why you're having trouble with this.
Yeah, show me a Republican with a spine and I’ll show you Adam Kinzinger or Liz Cheney. Or for that matter Mike Pence who notably didn’t vote Trump in 2024. But Dems liked that so you have to change the signs from plus to minus when you update priors was I believe how you put it. Huh? That really doesn’t make any sense.
Congressional Republicans are Trump toadies save perhaps Susan Collins and Lisa Murkowski who is on record as saying everyone is afraid of Trump’s goonish mafia capo retaliation. It’s like the nation is being run by the Corleone family except after Don Vito’s death his replacement was Pauly Walnuts instead of Michael. Cruel and vindictive but not so bright.
How many Democrats can you name that consistently stood up to Biden's excesses? Manchin and Sinema? Sounds like a pretty similar list, with pretty similar consequences for the defectors.
May I suggest that Congress has, in both parties, sublimated its responsibility to the head of the party in a way that is bad for the country?
I think these aren't even close to being equivalent. Did Biden endorse alternative legislators in the primaries for people who didn't toe the line? Did call them out and insult them on social media?
> I don't think that's true anymore, and I'm not sure it was ever true. I think it might have just been luck all this time, that there was a left-wing of idealists, a liberal center of technocrats, and a right wing that was totally disconnected from reality at every level and only got things right because the people that they happen to listen to were economists and business people that need to know things about stuff in order to live.
How exactly do you know your not the out of touch one? The real fundamentals come from factories, truck drivers, farmers. Would you listen if anyone doing low status real work said "I no longer care if trump breaks the system"?
Farmer protests are usually right wing, and assuming your not so far gone maybe you can remember: the food you eat comes form farms. When was the last time you talked to a technocrat who knew a farmer?
I can't speak for the OP, but there are a lot of people out there who disagree with me from a lot of different directions.
I worry a lot about the possibility that I may be wrong, but for many of those people I don't much worry about the possibility that they may be right - there are lots of questions where working out the correct answer is hard, but it definitely isn't seven.
The center-right, and even some of the more libertarian branches of the relatively far right, do have people whom I worry may be right about things. But I put the MAGA right in the same category as, say, Islamic extremists, as people who can safely be dismissed.
Trump is a boring moderate in reality I wish the trump of the memes was an option. And you cant dismiss islamic extremists they are rapidly gaining ground and will probably conquer europe, you may find your moral high ground ineffective without any ammo.
> there are lots of questions where working out the correct answer is hard
Theres enough questions where the answer is very simple but politically impossible. Have children, not an immigrant ponzi scheme; make real work high status, even if you need to blow up the stock market; own guns; respect farmers; invade mexico to kill the cartels; take any jews worth having out of israel before either genocide happens; stop welfare to undocumented immigrants; dont allow rape gangs, death penalty; dont lets europe censor the american internet, harress those germans whining; purge the prison system of gangs, death for cartel ties; collapse housing prices.
I dont care any more, if it takes a Caeser to unravel the Gordian Knot, let trump be the weakest and timid one imaginable
I had the mirror-image of this experience some years ago. After many years of agreement (on a very liberal forum) that Bush's war on terror policies were awful and inhumane, Obama came into power, continued most of them (with more murder and less torture), and rather abruptly, a bunch of the people I thought were my allies on these issues realized that the president had a hard job, he had access to intelligence we couldn't see, and in general sheltering war criminals and assassinating US citizens were forgiveable missteps in their guy.
Most people simply do not have strong principles in most areas, but almost everyone has a side.
I find this striking because almost all of the left-of-center people I know were highly critical of those aspects of the Obama administration, from the moment it started getting reported on. I'd imagine it's a question of moving in different circles.
I certainly heard criticism of Obama too. I think it was relatively muted because what can do you? It's not like voting Republican would help matters!
A lot of people I interact with are belligerent enough to loudly talk about voting third party out of anger when Democrats do things like this. Certainly I saw no shortage of people insisting that voting for Biden/Harris was unconscionable because of their support for Israel, for example. I think this is quite a silly stance to take most of the time--and AFAICT the people who take it are more loud than numerous even in far-left spaces--but I think the constant feelings of dissatisfaction with Democrats don't do great things for turnout, especially among younger voters. The U.S. tends to make voting hard enough as it is: people are that much less likely to jump through the hoops when they don't even like the person they'd be voting for.
Woe betide the poor sod who learns to be cynical about his opponents before he learns to be cynical about his allies.
It's not conservatives.
It's virtually everyone, and given that you've framed this as, "I'm Team Blue, can you believe Team Red?", it's you, too.
Sorry if that reads as excessively confrontational, but...I think my perspective is worth considering:
I like to joke that I'm "politically nonbinary," but to be more specific, my position is more like, "classical liberalism, but with the recognition that our version of late-stage capitalism has vast and scary problems, but also, absent a post-scarcity technological innovation like a Star Trek TNG-style food and object replication, we don't have any better options."
That position means I am No Team. And because I am No Team, I actually see and remember instances of retarded* teamism in which Team-folk "forget" stuff in order to maintain social cohesion.
For example, remember when Red Team and Blue Team *switched* *sides* on the COVID vaccine? Remember the earliest days of Operation Warp Speed, when Red Team was marveling at this amazing tech, but Blue Team members were skeptical of Big Pharma and publicly declaring that they wouldn't take "Trump's Vaccine?"
Remember when the Team Leader changed, everyone immediately switched sides on that vaccine tech, not because of new evidence, but to be in support or opposition of the Current Team Leader *carrying on the exact same plan as the Former Team Leader*?
No Team remembers.
No Team remembers, and then trembles.
And, look, I'm not taking credit for my inability to Team as some kind of enlightened intellectual achievement.
My dad's side of the family has a history of psychopathy, and I suspect that my innate process of selective empathy - I can only feel empathy for people I've judged worth of empathy (that's a very, very large group, larger than 'normal' people's group, but still) - is the very mildest possible expression of that family psychopathy.
Like many mild aberrations of personality disorders or mental illnesses, a tinge of psychopathy can be productive and useful, because sometimes empathy-governing behavior is very, very bad (you want your surgeon to be a little bit of a psychopath so they don't weep into your open wound, etc.).
But it's not great in other areas, like social cohesion. Unlike regular empathy-people, I can't be persuaded or make myself feel a certain way about a topic merely *BECAUSE* other people feel that way. I actually have a hard time even modeling the pleasure people get out of Teamism, much less the
not-noticing/forgetting/whatever-it-is-ing that allows them to blissfully change positions on an issue without ever realizing that's what they're doing.
So yeah, I am exquisitely familiar with the kind of collective doll-eyed blankness you describe witnessing amongst the conservatives at your Easter gathering when they were confronted with information that contradicted their social cohesion.
And I'm telling you: It's everyone.
* I know this word is still semi-forbidden, but no other word ever replaced the precise meaning of the slang term it eventually evolved into, and, given that its usage to formally and/or politely describe developmentally disabled folk is completely gone, I think we can reasonably reclaim the slang definition.
(Oh, does the sensitivity to word usage *and* advocacy for freedom of expression seem to conflict? That happens when there's no formal ideology to adhere to!)
By any chance, did you see a similar post about this subject on the subreddit? https://www.reddit.com/r/slatestarcodex/comments/18abtm7/comment/kbx51bq/
"PolymorphicWetware
1y ago
*shuffles notes*
Good thing I specifically saved some notes about this, it's amazing how quickly things get memoryholed when they're embarassing & inconvenient...
[quoteblock]
https://www.washingtonpost.com/outlook/2020/01/30/real-danger-coronavirus/
THE ACTUAL DANGER OF CORONAVIRUS
*Fear may fuel racism and xenophobia that threaten human rights*
While addressing the outbreak will take a global public health effort, the U.S. Centers for Disease Control and Prevention has declared the current risk to the American public is LOW [https://web.archive.org/web/20200130203859/https://www.cdc.gov/coronavirus/2019-ncov/summary.html]. If 21st-century outbreaks like SARS, MERS and Ebola virus are any indication, it is likely American fear of contracting coronavirus — and the xenophobic, racist assumptions that drive it — carries a risk far greater to most people in this country than the virus itself. Historically, infectious disease has generated racist discourse...
[quoteblock]"
Wonder what you think of it, this section quoting Scott from 2014 (https://slatestarcodex.com/2014/10/16/five-case-studies-on-politicization/) was particularly surprising to me:
"HOW DID THIS HAPPEN?
How did both major political tribes decide, within a month of the virus becoming widely known in the States, not only exactly what their position should be but what insults they should call the other tribe for not agreeing with their position?
...
IS IT JUST RANDOM?
A couple of Republicans were coincidentally the first people to support a quarantine, so other Republicans felt they had to stand by them, and then Democrats felt they had to oppose it, and then that spread to wider and wider circles? And if by chance a Democrats had proposed quarantine before a Republican, the situation would have reversed itself? Could be.
Much more interesting is the theory that the fear of disease is the root of all conservativism...
...
The proposition “a quarantine is the best way to deal with Ebola” seems to fit much better into the Red narrative than the Blue Narrative. It’s about foreigners being scary and dangerous, and a strong coordinated response being necessary to protect right-thinking Americans from them. When people like NBC and the New Yorker accuse quarantine opponents of being “racist”, that just makes the pieces fit in all the better.
The proposition “a quarantine is a bad way to deal with Ebola” seems to fit much better into the Blue narrative than the Red. It’s about extremely poor black foreigners dying, and white Americans rushing to throw them overboard to protect themselves out of ignorance of the science (which says Ebola can’t spread much in the First World), bigotry, xenophobia, and fear. The real solution is a coordinated response by lots of government agencies working in tandem with NGOs and local activists.
It would be really hard to switch these two positions around."
I don't follow the Reddit, but thanks for the link!
I was a little surprised there wasn't more commentary along these lines, especially here in this blog.
Two cartoons I think are appropriate here: https://xkcd.com/610/ and https://xkcd.com/774/
They actually aren't!
> For example, remember when Red Team and Blue Team *switched* *sides* on the COVID vaccine?
Rather than switch, it was a complicated dance. For example, there was a moment near the beginning, when it was known that only a racist would fear a virus coming from China.
https://www.washingtonpost.com/health/2020/02/26/how-to-prepare-for-coronavirus/
> The virus may be novel, but you really don’t need to buy anything new or special to brace for it. The Washington Post spoke to epidemiology experts, and they said the most important aspect of preparedness costs nothing at all — calm.
> “Seriously people- STOP BUYING MASKS!” Jerome M. Adams tweeted. “They are NOT effective in preventing general public from catching #Coronavirus, but if healthcare providers can’t get them to care for sick patients, it puts them and our communities at risk!”
This comes shortly after several paragraphs on how "avoid touching your face" is the number one thing you can possibly do. (Might wearing a mask help you avoid touching your nose and mouth? No, shut up.)
Wearing a mask can also cause you to touch your face more when putting it on and taking it off.
I'm sort of like you. I don't seem to be very subject to group identification and group loyalty. I have only voted once in my life. After several decades of watching from the sidelines as political beliefs play out on the large scale, in elections etc., and on a small scale, in discussions, here is my take:
Most people acquire their politics not through thought and study but via osmosis from their family and peer group.
Most opinions and votes are determined by vibes and group identification, including, probably, yours.
Any time you find yourself thinking that everyone with a certain characteristic (who they voted for, ethnicity, SES) is dumb or evil or smart or ethical or deserving of help, you can be sure your head is up your ass.
Politics makes people mean and stupid.
There are aspects of how you describe the democrat opinions on the vaccine that I think are misleading. Specifically:
"Remember the earliest days of Operation Warp Speed, when Red Team was marveling at this amazing tech, but Blue Team members were skeptical of Big Pharma and publicly declaring that they wouldn't take "Trump's Vaccine?"
Remember when the Team Leader changed, everyone immediately switched sides on that vaccine tech, not because of new evidence, but to be in support or opposition of the Current Team Leader *carrying on the exact same plan as the Former Team Leader*?""
I think that, while it contains some logical reasoning, you skip quite a few steps or make some assumptions that are not properly supported. The statement assumes the only reason for changing positions was team loyalty ("to be in support or opposition of the Current Team Leader"), without considering other factors that might have influenced these shifts. Similarly, you imply the context before and after the administration change was identical ("carrying on the exact same plan"), when the discussion/concerns morphed. Initial Democratic skepticism focused specifically on concerns about political pressure potentially rushing the approval process. You mention this in the form that democrats said they wouldn't take his vaccine - however, the more common position was that there were concerns and fears about it being rushed or not specifically based on Trump's approval. For example, Harris said that she would not take the vaccine based solely on Trump's word, but rather requires more credible sources of evidence and authority, citing Fauci as one such source. (Sep 5, 2020) Trump's platform based on Operation Warpspeed's chief advisor Moncef Slaoui, as well as his press secretary at the time AND Fauci (NAIAD) actually tried to respond to these concerns, saying that it would not be rushed out until it was safe and tested. Biden later changed the name on his first day in office, which I do think shows the intentional distancing from the concept that it might be rushed out. In this way, I do think that there is value in what you are saying regarding tribalism and partisanship- but oversimplified.
Likewise, I believe that another factor for Republican opposition to the vaccine stemmed less from the vaccines themselves or pure partisanship (although it is undoubtedly a factor), but rather from the aspect of the mandate, as well as other protocols such as social distancing, closing businesses or requiring masks.
I think electoral politics exacerbates the controsery, particularly given the timeline around the covid vaccine development aligning with the election. If Trump won the election, the response certainly would've been different, and the reaction to the vaccines as well. Although, I am hesitant to say which way it would've gone. Would they be anti-vax, or would they promote it?
As one last aside, I think that, based on your writing, you derive a sense of pleasure or satisfaction from being "politically nonbinary" or pointing out the hypocrisy, and particularly the very intentional usage of "retarded". I think that the way that you view yourself outside of the conventional boxes is in itself its own form of "teamism", and it is important to note that it shifts the boxes to "Teams" and "No Teams". I think its important to recognize ironically that you may be engaging in a form of teamism, with your team being "No Team". As you note, "[you] actually see and remember instances of retarded* teamism in which Team-folk "forget" stuff in order to maintain social cohesion." I would argue it is important to remember that we are not immune from these psychological heuristics. This aside is not meant to reinforce or discredit your overall argument, but rather to provide a reader's view based on your writing.
> "I think that, while it contains some logical reasoning, you skip quite a few steps or make some assumptions that are not properly supported. The statement assumes the only reason for changing positions was team loyalty ("to be in support or opposition of the Current Team Leader"), without considering other factors that might have influenced these shifts. Similarly, you imply the context before and after the administration change was identical ("carrying on the exact same plan"), when the discussion/concerns morphed."
Circling back to this:
First, I'm sure we're both on the same page that it's impossible for either of us to provide definitive evidence about the internal processes of tens/hundreds of millions of people. While my original comment was written in declarative statements, to match some of the tone of the original commenter, I'll be the acknowledge that discussions about human nature in general and the zeitgeist are ultimately speculative and personal opinion.
I don't disagree with you that a small-ish, dispassionate, rationally-minded portion of the population (very likely disproportionately represented amongst the commenters of an intellectual's blog!) closely followed and assessed the development of the vaccines and made their judgments accordingly without being unduly influenced by which Team was at the helm.
Was that the process that *most* people were engaging in, though?
Casually observing it: I don't think so!
My general reading of the zeitgeist (news media, etc) is that majority of both Red Team and Blue Team got distracted by their loathing of the other team's leader and eventually literally forgot that Trump was all-in on mRNA vaccines, and also that he was instrumental in launching them as a national effort. It's hard to work the topic into a conversation naturally, but it can be illuminating to remind Red and Blue Team people about Trump's history with the vaccines, just to watch their reactions.
And then ask them to imagine what would have happened if Trump had won the 2020 election with his administration in charge of administering the rest of the pandemic.
In my experience, the more Team a person is, the less plausibly they describe a hypothetical 2020 Trump Administration response to the pandemic. For example, deeply Red Team folk will claim that, oh, yeah, a Trump administration would NEVER have pushed for vaccine mandates and would have shut down the mRNA vaccines the moment it looked like they weren't that awesome after all. It wouldn't have cared at all about Big Pharma's interests. It would have been a paradise!
And of course, we can't know for sure what would have happened in that alternative timeline, but common sense says that the claim that Red Team would have turned the mRNA ship around - the ship they launched! - is pretty absurd. It's unknowable, but I'm pretty confident it would have actually been the opposite; mandates would have been more draconian, not less, and Red Team governors would have been more likely to cooperate with a Red President's leadership than (perhaps) make a name for themselves and boost their celebrity by standing in stark opposition.
For fun, it's also illustrative to ask Team people to speculate why only 14% of children received COVID boosters in the 2023-2024 season, and what they think the demographic breakdown of the parents of the 86% might be.
Though you shouldn't ask these kinds of questions unless you're okay with people not really liking you very much.
> "I would argue it is important to remember that we are not immune from these psychological heuristics."
This isn't necessarily true, though. Some people actually are functionally immune (or at least vastly less vulnerable) to some psychological heuristics simply because they have conditions which make their brains work outside of the norm said heuristics may be governing. There are people who are way, way further down the psychopathy spectrum than I am who are *legitimately* immune from certain kinds of internal psychological phenomena around social pressure, influence, and group ideology. Same thing with some high-functioning autistic people. There's pretty much always a non-zero number of people in experiments on group think / social pressure / conformity who will say, "no, those two lines are exactly the same length, the rest of you folk are nuts" because they sincerely don't care about (and thus can't be unduly influenced by) the group.
As for the aside on my writing, yes, I actually spent time debating capitalizing "No Team," as I knew I would inevitably get at least one comment saying, "Ah ha, 'No Team'" is a Team, you even gave that Team a proper noun; you're not special or different!"
But I decided to do it anyway in order to most effectively pull off the "Pepperidge Farm remembers" reference, a choice I made because I cared more about maximally amusing myself than I did people potentially having less respect for my comment and/or me because my argument was weaker than it could have been.
Which, I think you'll agree, is a lack of care about how I am perceived which is outside of the "norm."
And last, while there is a certain amount of pleasure in not being taken in by team-based manipulations and scams, there are many disadvantages to being incapable of ideological conformity to a Team, including that most people are Team people, and Team people REALLY hate being disturbed when they're Teaming (my otherwise supremely compatible ex in my best relationship broke up with me because, in his words, he was very, very Blue and, "I don't like how I feel when I agree with you.").
I appreciate your response - although with the caveat that it seems reasonable (although I do not say this with a sense of certainty or any data based evidence) that even neurodivergent people who are less susceptible to common heuristics or specific (e..g. group based) cognitive biases, they may also have greater vulnerability to other different heuristics/cognitive biases. I am not claiming this for all neurodivergent people universally, but rather on a much more specific individual level.
For example, perhaps someone who experiences a sense of strong displacement from their community due to psychological and neurological factors would develop a different set (or perhaps weights) of heuristics in social reasoning than those who feel integrated into their communities. These alternative/distinguished heuristics might help them navigate their unique circumstances but could potentially introduce different biases. Along the same lines, I think this probably follows along the same lines of the contextual advantages and disadvantages that you are talking about. Thanks for your response - interesting stuff
Oh, for sure. If I wasn't clear about this, I want to acknowledge that deviating from the psychological/social norm isn't the path to living a very good life, even if that deviation has some fringe benefits. Normal people experience pleasure and a sense of security in Teamism that I can't access, myself. And, perhaps more importantly, they are able to naturally express their pleasure to the other members of their Team in a way that is intensely bonding (thus socially useful).
I'm not psychopathic enough to want to manipulate people into liking me a lot more by *faking* a mutual enthusiasm for things like my city's football team/ my workplace/ a political party / my gender and/or ethnicity / joining a church, and so on. But even if I only neutrally refrain from enthusiasm (rather than criticizing the enthusiasm, as I often did in my younger days), it still has a chilling effect, which ultimately isn't good for me.
If I were more rational and less egotistical, I'd fake the Team enthusiasm and enjoy the fruits of that labor, but I'm not rational enough to make myself do it.
One thing you didn't really mention is that there was also new evidence that caused people to change their minds. In particular, the study results came out and the vaccine was vastly more effective than anyone expected. And then once people started getting the vaccine, we had dramatic real world data showing its effectiveness too. I can't speak for everyone, but personally, I changed my beliefs based on new evidence, not which guy was in the white house.
Putting COVID aside, It's also worth pointing out that when it comes to the MMR insanity, this is probably less a matter of people changing their opinions and rather people changing their party. Anti-vaxers used to be a *left-wing* phenomenon. RFKJ himself used to be a Democrat! What changed was that Trump attracted all the crazies to his side and repelled normal people.
I agree with you wholeheartedly. I wanted to focus on the tribalism aspect, but this is as important if not more. Categories were made for the man, not man for the categories
It is very disturbing to see how accurate 1984 was. We have always been at war with Eu-astAsia.
https://substack.com/home/post/p-161615576
I'm going to drop this here because I think a lot of rationalist/EA people have good ideas but don't sell them well. Guy talks about seduction and I think he has a lot of good points.
Ehhh, I think he's overselling it.
On the one hand, yes, some basic sales knowledge is very valuable. Yes, it's very annoying how people will reject obviously beneficial things for weird internal reasons. Sneakers are a bad look, chelsea boots are a good look, just buy the boots and get back to Getting Stuff Done.
On the other hand:
"You call yourself effective because you misunderstand what altruism is. You think it is a mistake that in practice altruism rarely lives up to its nominal charge when, in fact, altruism is a term one uses when one wants to seduce others into doing the thing they wanted them to do anyway, give them prestige and money"
Yeah, but the effective altruists have pretty clearly proven they're better at saving real human lives than 99% of charities out there. That's the thing that matters. A lot of the "marketing" and "discourse" are effective altruism can be...suboptimal but their results speak for themselves. And, ultimately, that's all that matters.
But don't try to turn your engineers into great salespeople; you just end up with bad engineers and mediocre salespeople. No one gets to be good at everything. If your product is great, by all means make sure your engineers wear clean shirts, but you fundamentally prioritize good product. If your marketing is great, by all means, make sure your product isn't crap, but keep fundamentally prioritizing good marketing.
> dark woke
anyone have any examples? If it was *actually* real I could see it working..... but... I have standards for offensive humor
Wait, "dark woke" sounds interesting. But who are you responding to?
The aether; I believe it was ny times article (boring as shit) and fox news boomer take was "swearing is wont make you electable"; I have yet to see someones take that properly takes into account my anti-politeness position.
For it to be real I want some memes and a new aesthetic
I'm moving to Nashville next month. Anyone know any poker games around?
This might be too vague to answer, but here goes. Is Stuart Russell's book Human Compatible worth reading? I noticed it's cheap on Kindle today, but also it came out in 2019 and a lot's changed in AI in that time. Any thoughts?
The left using their "prosecutorial discretion" to protect their tribe.
https://nypost.com/2025/04/22/us-news/tim-walz-employee-who-allegedly-caused-over-20k-in-damages-to-teslas-let-off-by-woke-minnesota-da/
"A progressive district attorney has declined to charge the state employee in Tim Walz-led Minnesota who was allegedly caught causing $20,000 damage by vandalizing half a dozen Teslas — a decision the local police chief ripped as the latest betrayal of victims."
[Edit: Grammer]
>>Despite what police believe to be evidence of Adams committing felonies, Hennepin County District Attorney Mary Moriarty will seek diversion rather than criminal charges.
>>“This is an approach taken in many property crime cases and helps to ensure the individual keeps their job and can pay restitution, as well as reducing the likelihood of repeat offenses,” the attorney’s office said in a statement, CBS News reported.
It's a pretty important clarification that "declined to charge" in this case, means sending the vandal in question into a diversion program involving regular check-ins, community service, and restitution payments, under which failure to comply with the program means they can be "terminated from diversion and returned to court."
https://www.hennepinattorney.org/en/about/adult-diversion/adult-diversion
I'm okay with diversion, including restitution, given the circumstances, but if this person were too poor to pay back $20,000, does that mean diversion wouldn't be an option?
He is a state employee. He has the means to pay. If he doesn’t make restitution he’ll be criminally prosecuted.
But I'm explicitly and directly asking about the hypothetical where it's a different person who can't pay $20,000. Does the poor person not get diversion as an option?
Okay I still don’t really know but I’ve thought a bit about it and it’s likely a poor person with no prior criminal record who couldn’t do full restitution would still be put in some form of diversion program. It would probably look something like what Nobody Special came up with.
Minnesota does lean liberal and it seems unlikely they would lay a felony conviction on someone who didn’t have the cash for full restitution. But it’s really just a guess.
I have no particular expertise, but I'd expect that falls into the discretion of the County Attorney, who could work out alternatives like an extended or reduced payment plan, community service, or other forms of in-kind restitution.
In the end though, that's just my wild-ass guess.
I don’t know.
If I am understanding your argument, diversion is appropriate for acts of political vandalism?
No formal crimes being charged?
If he makes full restitution and jumps through every hoop of the diversion program he will avoid charges. If not he will be prosecuted. This is SOP for first offense property crime at the Hennepin County DA’s office at this time.
He is on the hook for full restitution and the rest.
For even political vandalism?
Since we probably will not agree with the punishment, can we agree to condemn this political vandalism?
I condemn political vandalism too. The guy is an idiot.
I associate diverson programs with youths who may have made a bad decsision. Not politically motivated actions meant to terrorize people.
Maybe a different hypothetical to Monkyyy.
How would you feel if a MAGA person vandalized Disney World in an obvious political act and a Florida prosecutor did this?
I’d be fine with a diversion program and restitution for the Disney politically motivated vandal.
That’s my personal take though and understand that some people might want jail time for the Disney and the Tesla.
I don’t think this is tribal myself. It would likely have played out the same if the dipshit had keyed a bunch of Chevy Volts.
The best penalty for destructive political acts is public humiliation. Make them publicly apologize and pay the damages. Locking them up just makes them scream about political oppression and look more martyr-y, meaning you'll have to lock up more of their friends too.
Same answer I gave Monkyyy - depends on the vandalism. If it's something like arson where there's an actual danger to others posed, that's a different kettle of fish entirely and diversion is inappropriate IMO.
But assuming we're trying to stay apples to apples with the Minneapolis case, and thus still talking about keying, if somebody keyed the little teacup seats on the Mad Tea Party ride in the name of MAGA I really don't much care whether they get formally charged or put into a diversion program and forced to pay restitution on pain of prosecution later if they screw it up.
In all honesty, if all they did was key the teacup ride and they're a first time offender I'm talking myself into the idea that diversion should just generally be preferable, regardless of political motive in any particular direction. I may not have a lot of sympathy for MAGA, but keying a ride at Disney would be a pretty silly-ass reason to saddle someone with a criminal record if you have an equally-workable option to just make them pay for the damages and reform themselves to not be the kind of moron who keys other people's stuff.
+1
ISTM we want this guy treated the way vandals who get caught are treated in other non-political cases. I don't think vandalizing Teslas for some political cause ought to get extra punishment for the same reason I do not support hate-crime enhancements for vandalism--if Alice spray paints some obscene words on a wall and Bob spray paints a swastika on a wall, they ought to get the same sentence.
I applaud your consistency on this but unfortunately this is not how punishments are given currently.
Community service is not the biggest punishment I can imagine for arson. How would you feel about 20000$ of fire damage to an abortion clinic?
Differently, but then fire has the capacity to rage out of control and kill people, so it's pretty distinguishable from keying cars.
I'd also, for example, feel differently about diverting someone for $20,000 of fire damage to a Tesla dealership, and be more open to diversion for someone who did a similar amount of damage by keying the cars of one or the other side of protestor/counterprotestors out front.
... wait how did they get the number 20k in damage from 6 cars without any fire? Are the repair shops throwing around a huge markup?
Have you ever paid for body work? Or done it yourself? It's absolutely ridiculous. To properly paint something, even a relatively small scratch, in a way that isn't glaringly obvious requires many days of prep and technique.
Ive never owned a car over 5k and Id never fix a scratch; Id take 20$ over it
Beats me - the video imbed & photos in the NYP article just show the guy keying cars. No idea how you get to $20k with that. Maybe some combination of repair shop markups & there being more cars involved? The article says "at least six" rather than just "six," so maybe there are other keyed cars in Minny for which they don't have the video but with which they planned to charge him?
How does this work? Do the people who make these decisions need to make some written statement explaining why? Can we find those statements online? If yes, it would be nice if journalists linked them. This would generally improve the quality of discussion.
Here’s a statement from the Hennepin County DA’s office. - St Paul Pioneer Press
“We want to make sure we are very clear. What Mr. Adams did was wrong and we are holding him accountable for keying the cars,” said Daniel Borgertpoepping, spokesperson for the Hennepin County Attorney’s Office. “The HCAO did not reject or decline this case. We offered diversion as we often do with property damage cases when the person has no record. Mr. Adams will have to complete the requirements of the program. He will also have to pay every penny in restitution to the victims. If he does not meet those requirements, we will proceed through the criminal legal system process.”
Probably paywalled - but you might get a couple of free views
https://www.twincities.com/2025/04/22/state-employee-arrested-for-tesla-vandalism-will-not-be-charged/
I predict this will stay silently ignored, because crying "leftist conspiracy!" is much more fun.
So diversion is the just punishment for political acts of vandalism?
> “We want to make sure we are very clear. What Mr. Adams did was wrong and we are holding him accountable for keying the cars."
> "Mr. Adams will have to complete the requirements of the program. He will also have to pay every penny in restitution to the victims. If he does not meet those requirements, we will proceed through the criminal legal system process.”
He's going to pay whatever it takes to undo the damage he caused, cover case fees, plus a bit more on top, and also do some kind of community service; no lighter a punishment than anyone else committing property damage with no prior record. If he can't or won't go through with all that, it's off to jail via the court.
Sounds pretty reasonable to me.
Are you suggesting that singling this guy in particular out for something much harsher than what is usual for vandalism would be just, or did you miss the part where he was being punished, or something else?
Your arguments continue to ignore the political part and treat this person like a 17 year old that decided to key cars randomly. It is almost like you are ignoring the facts of this case.
I am not sure what is the right penalty but I think he should have been charged with some crime. That would signal to the community that political vandalism is not tolerated.
This “punishment” is so light that it sends the opposite signal. Vandalize the right political opponents’ property and we will protect you. This is tribalism at its worst.
You are free to make your arguments but do not be surprise that MAGA people decide to now vandalize leftist property. I hope you will still be consistent by actively arguing for diversion in those cases too.
> Do the people who make these decisions need to make some written statement explaining why?
Im aware of no requirements, I tried to find a rant in the local news but found nothing; this has a genetic statement for a different case
https://youtu.be/PXj9Dy0FC34?si=13xFV9AV0eiB6tQ7&t=225
> it would be nice if journalists linked them.
Why provide names and statements when you can say "tim walz['s employee] IS EVIL"
"tim walz['s employee] IS EVIL"
That's an interesting thing to write. Didn't this guy do something really, really wrong? And is corroding the civil discourse in the US?
"Evil" is too strong of a word for my taste, but this is not a good guy.
> Didn't this guy do something really, really wrong?
Idk theres 2 different stories; if its 20k in damages sure, if 6 keyed cars is being charged 20k someones stealing money.
6 keyed cars getting community service is soft on crime vs broken window policing; not news.
In general, what punishment do you think is appropriate for minor vandalism of other peoples' cars? I don't see it being a great use of public resources to lock such people up, so fines and community service seem like a good option, with the county jail as an alternative if they reoffend or don't pay their fines/do their community service.
He reoffended 5 times, deliberately, with forethought, with adult capacity, in order to destroy property for political reasons.
“minor” vandalism? These words lead me to conclude that you are ignoring and/or downplaying the political part of this vandalism. Can you clarify?
My opinions on courts systems has near zero effect on reality; but I consider the last 100 years to be theft. The state is claiming ownership of my revenge, an economic asset, and grossly mismanaging it.
Payments need to go to victims and we need 100 years of data and negotiation to catch up on. If you punch me I'm owed a punch, maybe I hire a kick boxer to look scary and ask you prehaphs we could set up a payment plan.
> community service seem like a good option
Maybe compared to rape cages, but your lacking imagination if you think bubbling efforts in a make-work "job" is making the world better.
> Ive never owned a car over 5k and Id never fix a scratch; Id take 20$ over it
I'm not sure you're the best judge of the standard costs of car repairs.
Does anyone know where I can find the stated views of Mahmoud Khalil (The Columbia University student accused of terrorism)? Matthew Yglesias mentioned his views are "more radical than most liberals realize" in a post (https://www.slowboring.com/p/mailbag-remembering-kevin-drum) but the link to the AP news article he references is now dead.
Everything I can find online seems to basically say; "As a Palestinian student, I believe that the liberation of the Palestinian people and the Jewish people are intertwined and go hand by hand, and you cannot achieve one without the other" which seems a fair and nuanced take. While the Trump administration is clearly willing to be unreasonable, this can't be all that was said regarding Israel and Palestine.
I don't think this has much bearing on the legality of his attempted deportation (the same with Kilmar Garcia and his alleged MS13 affiliation) since due process is being ignored, but I do think it has some bearing on the level of moral outrage demanded, and how generalizable the threat of Trump's deportations are.
If there are edge cases in his policy where people who are actually gang members, but aren't proven so in court, are deported without due process, and people calling for the genocide of Jews are being deported for their views, that changes my estimate on how likely Trump's policies are going to spill out into a more general policy of authoritarianism. I'm of course not saying either Garcia or Khalil are a gang member or explicit antisemite, but I'm curious in the case of Khalil if there's a take on his previously stated views that I assume Yglesias was referencing which seems to have been removed.
I tried to look, my best guess is the phase "columbia united apartheid divest" is the best results
Im seeing the term "Zionism" used in formal documents, and in their list of demands includes "[ending] Dual Degree Program with Tel Aviv University", Id look for the worse treated "dual student from telaviv"(whatever the hell that is) and see what they have to say
> "dual student from telaviv"(whatever the hell that is)
That sounds like it would be a student from Tel Aviv who is involved in a study abroad program.
Im questioning whats the politics, I can put the words together.
Maybe well enough for your own blog, but I regularly struggle to parse what you're trying to say on here. I wish you would hire a line editor.
An intern at my company was going to school at Columbia when that happened, and their demands seemed pretty reasonable when I read about them at the time. Albeit I've read the divest movements don't actually achieve anything (apparently the divest from South Africa movement didn't have any meaningful effect on the S.A. market).
Can that be what Yglesias was talking about? It doesn't seem particularly radical to me, but that could be that I just differ in opinion.
Remember that ivy league jewish students fled into a libery to avoid one of these protests and they were chased and that video made the rounds in the right wing. That easily could've been Colombia.
Fundamentally this would be trump selectively preferring one middle east religion over another.
When fox news types use the word terrorists about anti isreal protests, they have videos of harassment of jewish students. Much like jan 6, cnn had 1 side of the building breaking windows.
Can anyone explain to me what it means to run multiple instances of an AI model? For example, when talking about how a company could have a whole team of AI 'employees' supervised by a manager (human or AI), what does that mean in concrete terms?
I think that, right now, each model consists of a whole bunch of numeric weights. This program can be run on as much compute as is available, and there will be separate copies of the program on different chips/groups of chips. Each copy of the program will receive inputs, serially, one after the other, and produce an output from them, independently of what all the other copies of the program are doing. Then again, maybe this is only the case for open-source models which different people have downloaded onto unconnected computers? Maybe, for other models which are run centrally by the company, there is only one copy of the program, and it is running multiple different prompt-convert-output sequences at the same time/in parallel? If so, how?
I think this question is the primary reason for my confusion about agentic behaviour. Will a particular model always have the same goals, or might multiple copies of it have independent goals? Should I expect models to collaborate to work towards long-term goals, or to possibly come into conflict? Also, should I expect a model to be able to ‘check on’ the thought processes it generates, or other copies of it generate, when answering other queries? To the extent that the goals of the AI relate to itself, will it think of itself as being only the individual copy, or as being all copies of the same model, or what?
> Will a particular model always have the same goals, or might multiple copies of it have independent goals?
#2, 100%. It's trivially easy to set up two prompts for the same model instance and have the two personas "battle" each other.
Consider 2 identical chaos pendulums where the inputs are a combinations of where the wights are. The machine can be the same, but if you rise the weights to different heights the output becomes different.
If you encode some sort of knowledge into chaos pendulums, make duplicates, and start connecting them with strings, the encoding is still there, and youve made a larger system.
----
You probably make exact copies of the machines but with different promts you get parrell processing on the task, and then you feed output from one into the input of another every once in a while
> Should I expect models to collaborate to work towards long-term goals, or to possibly come into conflict?
If the signal is positive, it *may* have constructive inference towards the correct answer.
When an LLM answers a question or performs a task it's instructed to do (by outputting text that triggers some actions), these are the factors that influence the result:
1) the model itself (a bunch of weights)
2) parameters such as e.g. temperature (how "creative" the model should be)
3a) the system prompt instructing the model how to behave in general terms
3b) the context window: additional files provided to the model, like source files; previous prompts and replies of the model within the same session.
3c) the actual current prompt the model's to act upon.
When people talk about "different AI employees" or agents they sometimes mean different 1) and 2): e.g. use a reasoning model for some tasks but not others; use Claude where it works best and ChatGPT for other things; fine-tune a model for a specific use; etc. etc. But I think that's not what you're asking about. Suppose for simplicity we fix 1) and 2) and use the same model & parameters for all our AI work; does it then still make sense to talk about "different AI employees"? Yes, because the input to the model is all of 3a), 3b) and 3c) chained together. The model doesn't "remember" anything from one invocation to another, and it doesn't matter on how many chips it runs with how many parallel copies. All the "working memory" is inside 3a)+3b)+3c), but that can be a lot, especially for newer models with large context window sizes.
So with programming work, you can have one "agent" be responsible for fixing small bugs in the codebase, another for reviewing such suggested fixes, a third for deciding which of the small bugs in a bug tracker to send to the first one, etc. You might use the system prompt 3a) to set their "goals" and differentiate their behavior (ask the agent that's generating tests to be very pedantic, for example); you'd use 3b) to give different agents access to different data (source files, bug tracker dump, emails...), and the buildup of prompts/replies within a session may (or may not) be useful to further specialize their behavior to each agent's intended use. Meanwhile, all of these agents might well run on the same model and even on the same group of GPU chips, if that model is fast enough for the purpose, because what differentiates the 'agents' is 3a)-3c). Anthropic has a recent blogpost "Claude Code: Best practices for agentic coding" with helpful suggestions including for multiple agents.
But this is now. As the market grows and tools mature and become less expensive, you might be seeing more of separate agents each running on a different fine-tuned model (i.e. a different set of weights, although obtained quickly and inexpensively from an existing model). And if (when?) we get radically new paradigms of models more powerful than LLMs that continue learning and updating their weights dynamically as they get more data and answer more queries, then the entire landscape may change and agents may come to embody different and fluidly-changing "goals" and "personalities" within their weights. I don't know when or if that will happen, though; maybe radical new breakthroughs are needed. See the recent paper "Welcome to the Era of Experience" for some ideas how that might look like.
> Each copy of the program will receive inputs, serially, one after the other, and produce an output from them, independently of what all the other copies of the program are doing.
Nope. Not even that. Each request is to a new service, it’s not waiting for the user for every chat. In fact any request could be to a new data centre if you changed your location (or VPN), and there’s no computer or process assigned either regardless.
Regarding the POSIWID, I still think of computer programs and the adage that a computer does only what you tell it do. So even if you want the program to do something useful, but end up instructing it to do an infinite loop, it will get stuck.
So what about this reformulation that jettisons the word purpose and encompasses both incompetence as well as malevolence:
The system does what it is built to do.
or
What the system does is what it was built to do.
Even shorter "working as implemented". That is, there isn't some malfunction in the system. All the pistons are turning exactly as they should *given how they were assembled*. Which might not match anyone's desires or intent, but that's what the collective someone told it to do, and so it will continue doing that until people do the hard thing of fixing it. And even that may be not exactly what was wanted.
Not an English native speaker, but saying a machine is "built to do" a certain task implies "purpose" to me. It's at least ambiguous, which would be even worse as a definition.
I am not a native speaker as well, but I would say that you can build something that doesn't work as intended by accident or by not considering all complexities and failure modes while having the best purpose in mind.
Gaining resources feels like a bad first-order strategy for an AI in most tasks, such that they would be trained against even in RL by lower-capability agents. Gaining resources is often competitive or zero-sum. For example, if there are lots of opportunities for an AI to make money easily today, or lots of computers an AI could hack for extra compute, those would probably have been claimed by other AIs with more resources than the AIs being trained. If nothing else, copies of an AI being trained on network security would be out-competed at money-making by AIs being trained on money-making, and vice versa. A lot of the time, compute dedicated to gaining resources would be a waste of resources, and thus poor expected-value for an AI.
Sometimes, resource-gathering would actually be the most efficient solution. In those cases, it’s complying with instructions, so it’s not fair to call it misaligned - users with AIs using those AIs to acquire resources is the intended behaviour, for better or worse. There, spending compute to get resources would be rewarded. Resource-gathering might develop as a heuristic, and it would probably be used sometimes when it wasn’t useful. But as a mis-applied heuristic rather than a goal, it wouldn’t be applied in an optimised, adversarial way - there would be no advantage in hiding it from users. Also, the AI would learn when to use it and when not to use it, and since it wouldn’t help most of the time, it wouldn’t do it most of the time.
All the examples of mis-alignment I saw cited in AI 2027 were following heuristics that were first-order effective in their training environment. “Do what the fake Oxford paper suggests”/“Use the loopholes in the prompt”/“Go to the top-right corner to get the cheese” are all easy and effective strategies that increase reward, being mis-generalised. “Gain resources” isn’t easy when it doesn’t help with the main objective, and therefore isn’t effective unless the AI learns not to do it most of the time when it isn’t helpful.
> "Copies of an AI being trained on network security would be out-competed at money-making by AIs being trained on money-making, and vice versa"
I think this only works in a sufficiently multipolar environment? If the money-making AI is a generation ahead of the network security AI (or more generally, a generation ahead of every other AI) it might be better in both fields despite only being optimised for one.
Moreover, if an AI knows that currently there aren't many other AIs competing for any given resources but it anticipates that there will be once it is no longer a one-off state-of-the-art but instead part of a regular generation of AI, that seems like an incentive for it to move quickly to grab everything it can before competition for the resources emerges?
I’m assuming it’s trained alongside many instances of itself, in parallel.
That's not how it works. They train one instance and then deploy many copies. It's literally called "train once, deploy many".
https://epoch.ai/blog/train-once-deploy-many-ai-and-increasing-returns
Wow, I’m using all the language wrong. I agree with the paradigm where a model is trained once as a text predictor, then they run many instances of it. I think RL would happen with multiple sessions running at the same time. I’m pretty sure that’s how it works now, and I think your link isn’t about RL so I don’t think it disagrees. My argument above works if RL happens in parallel.
So you train many instances but then only deploy the best? Or you train many instances and then deploy them all as a cluster? The former seems much less broadly multipolar, and the latter seems like basically perfect conditions for acausal negotiation...
I mean ‘instances’ like individual ChatGPT sessions. You train the model once, then you get it to do tasks with separate chat/CoT histories. It’s not necessarily multipolar in deployment, but it’s multipolar in training, and AI 2027 worries specifically about misaligned goals baked in during training.
It might lead to acausal negotiation, but I think the AIs would tend to negotiate for help with their stated objective, rather than with goals like acquiring resources. Acquiring resources to trade in acausal negotiations has the same limitations as acquiring resources for any task unrelated to those resources - an AI designing a factory robot would have no comparative advantage in making money. Its best option is probably to sell compute to another AI working on money acquisition, in an ordinary, causal (though super-efficient) transaction.
I guess it’d be vulnerable to bribery, if it e.g. introduced backdoors into its robots in exchange for resources from a competitor (if the backdoor won’t be discovered until after training ends). That feels like straightforward loophole-finding in accomplishing a stated task. (Anything neutral for its task that it could do in exchange for resources/favours would just be good strategy, and it could do it openly.) I agree that loophole-finding is a first-order effective strategy, like hallucinating, that companies would want to RL out but which would be harder to RL out for a smarter system. I think the AI would, again, only accept bribes that help with its stated goal, rather than re-investing them in a separate goal like resource-gathering.
My pathogen update for epidemiological weeks 15-16.
1. Per Biobot, as of 12 April, national SARS2 wastewater concentrations were almost down to the levels of previous post-Omicron interwave gaps. I suspect they're there now as I write this. If patterns hold, SARS2 should maintain this low level of circulation for a few weeks before climbing again in an upward curve. Though previous summer peaks have been in late Aug/early Sept, I wouldn't be surprised to see the next wave peak in July or early Aug. But we'll see.
And if the current trends hold, COVID-19 weekly mortality rates and ED visits will drop below levels of all the previous interwave gaps. So far in 2025, COVID-19 is still holding on to 14th place as the most common cause of death, but I wonder if it won't drop off the top 15 by the end of year.
LP.8.1* and all its Clade 25A descendants have pushed aside XEC and are now the dominant clade. Worldwide, LP.8.1 is still showing some growth, but overall case numbers are falling. I don't think any of the current Clade 25A vars will kick off the next wave. 🤷♀️
2. Influenza B is still circulating. But if you look at the Biobot wastewater numbers, it's an order of magnitude below the levels that influenza A reaches.
3. HPAI incidences in poultry have been falling. Poultry cases follow a seasonal pattern, as do cases in wild birds (sorry, I can't find a recent chart for this). Mammals may not follow this pattern b/c new cases in cattle have held steady over past 3 months at 30-35 new incidences per month.
There have been no new deaths from HPAI in the US for the past 90 days. Worldwide, there have been 6 deaths from HPAI in 2025. There were 8 deaths in 2024. This is higher than the 2016-23 period, when there were only 1 or 2 HPAI deaths per year, but not as high as the preceding decade, when there were between 20 and 40 deaths per year.
Considering that Genotype D1.1 found in poultry is killing people, I suspect we've dodged the H5N1 epidemic bullet for now. Maybe next flu season! But experts have been predicting an imminent H5N1 pandemic for 2 decades now. In the meantime, SARS2 came out of nowhere and got us.
4. It's hard to tell if the US measles outbreak is growing or receding. It takes between 7 and 21 days after initial infection for measles to become contagious, so there may be a whole bunch of latent infections out there. But the current numbers in Texas, which has been hit the hardest, look like it's receding.
Per the CDC, there have been 800 cases in the US so far this year (vs. 285 cases in 2024). There have been two confirmed deaths and one provisional death from measles this year.
Meanwhile, in Canada and Mexico, cases seem to be rising. Mexico has reported 360 cases and 1 death. And as of 12 April, Canadian provincial health authorities have reported 914 cases of measles, but no deaths so far.
https://tinyurl.com/3exwp3v9
https://tinyurl.com/5n83rbb9
Slides here...
https://t.co/EofSMV6kTh
Regarding the universe being cyclic, that is, having entropy resets. I might be wrong that this could be the case, but I can't see why.
Two problems for it to be the case are:
1. Entropy usually increases overall
2. Practicality
A cyclic universe must be, let's call it, pendulum-like. In a pendulum-like system energy changes between 2 forms back and forth. It doesn't do this forever because a) the system it's part of a bigger system where energy can go and b) it's made up of smaller systems where energy can go. In the case of the universe a) is ruled out.
A big bounce is the mechanism by which the universe could be cyclic and so all matter needs to meet up again and again, and whether it does so or not depends on something sorta like "escape velocity".
But also distribution of matter might play a role, which is where the entropy and the practicality problems come together.
I think even if the "end" of the universe is a mess, an uneven version of heat death -- more matter over here, vastly less matter over there -- if something like escape velocity is not given, then after a sufficiently long time the universe will bounce, and bounce in an entropy resetting way.
I want to point out, that the uneven distribution of matter due to early inflation of the universe could make it look like the universe is expanding at an ever increasing rate, but does so only locally, in this bubble so to speak.
That's why I think that after a sufficiently long time the universe could bounce anyway.
Also, the entropy law is always mentioned as unquestionable -- even in the case of the system being the whole universe -- but the reason why the law holds true, the mechanics behind it, is just statistics. I mean there is something like this https://en.m.wikipedia.org/wiki/Poincar%C3%A9_recurrence_theorem
Can someone make me see clearly, that I'm just wrong about this?
Our understanding of the universe is limited by relying on models and a weak understanding of how physics would operate under the conditions of those models, so I like to think that anything could happen when the universe 'ends'.
>the entropy law is always mentioned as unquestionable -- even in the case of the system being the whole universe
This is not actually the case. It is not clear whether we can apply the 2nd law ("the entropy law") to the entire universe, and I think most physicists would agree that its application to the universe as a whole could be questioned. Our current level of understanding of physics and cosmology simply isn't at a point where we can make any statements of this kind with certainty.
I don't see anything necessarily wrong with the rest of what you said. At least to me, the main issue is how you get back to the special condition of the big bang - that is, a point where entropy is at a minimum. If the 2nd law did go into reverse at some future point then it seems more likely entropy would end up at some kind of stable intermediate point, rather than at a minimum.
Thanks.
"At least to me, the main issue is how you get back to the special condition of the big bang"
Accidentally :) or... Penrose came up with the idea that when all matter except electromagnetic radiation would decay into the latter and it would redshift with further expansion until the field is so flat that it makes no sense to speak of any distances between anything anymore, then the universe would not be big, however big it would have been before, but just as small as it had been at the big bang, which would then happen again.
He called this conformal cyclic cosmology.
> Also, the entropy law is always mentioned as unquestionable -- even in the case of the system being the whole universe
Alastair is right in that you shouldn't assume that 2nd Law of Thermodynamics applies in all cases to the entire universe. Certainly, phenomena like life seem to violate entropy, but the theory says that life exports entropy to its environment. I'm not really comfortable with that claim, although Boltzmann's formula for entropy seems to describe that scenario. I admit that I should just accept this as true because I have no *rational* reason to doubt it, but I still have doubts—maybe because the Universe seems to be calibrated to encourage emergent systems that have higher order informational complexity than their underlying systems. Information is ultimately energy (unless someone can provide me with a counterexample of information that doesn't involve energy transfer in some way?).
Emmy Noether proved that in an empty universe time translation symmetry should produce conservation of energy. But we exist in a non-empty universe where time translation symmetry is broken (the past state of the universe is different from the current state, and it seems like the future state will be different from the current state—thus symmetry is broken), which has been used to prove that energy is *not* conserved over time in our universe. In Einstein's General Relativity, there's no universal definition of energy conservation for the whole universe. For example, the redshift of light due to cosmic expansion implies that photons lose energy over time, but there's no place where that energy goes. So the 1st Law of Thermodynamics is falsified at the universe level, and the 2nd Law is the result of time transition symmetry being broken. So, according to Noether's theorem, energy is not conserved in our universe. Likewise, information is not conserved. And, if energy/information is somehow conserved, where does it go?
But at approximately ten or eleven billion years into universe's evolution, we see information systems acting as templates for much more complex information systems (i.e. life). For example, genes are strings of information, but they code for much more complex informational systems, which in turn can process information in new ways, and, yes, create new information. WTF?
On the other end of the information spectrum, black holes have been mathematically proven to destroy information (although I think that assumption is starting to be questioned now). Lee Smolin has suggested that black holes create new singularities that spawn new universes. And I think he suggested that information consumed by a black hole acts a template for the next universe—but don't quote me on that. But since a black hole consumes a finite amount of energy, Smolin never explained where all the new energy to create a new universe comes from. But that begs the question of where the energy for our universe came from. But I digress...
> I want to point out that the uneven distribution of matter due to early inflation of the universe could make it look like the universe is expanding at an ever-increasing rate, but does so only locally, in this bubble, so to speak.
Although some cosmologists question the Cosmological Principle that matter is evenly distributed across the universe, at least on a scale larger than galaxies, there's no evidence that I know of that supports this idea. So, we have energy disappearing from our universe, and the universe is (probably) expanding evenly everywhere, plus observations show that the expansion seems to be accelerating (due to the Cosmological Constant which is thought to be dark energy). So if current trends continue, and there's no reason to think that they won't, the universe is likely going to end up a lot colder with a lot less complexity. Sorry, to give you the bad news. It bugs me, too, because eventually I'll run out of pleasant places and bodies to reincarnate into. Bummer. Hopefully, Smolin's idea that black holes are spawning new universes may cheer you up, though. ;-)
Also, the laws of our universe may still be evolving. The laws of the universe had to be different during inflation, so why can't they change further as time goes on? Maybe that's another way out of universal heat death?
Oh, crap. I just remembered there was a study that questioned whether the observed acceleration in the expansion of the universe is an artifact of poor supernova sampling. Perlmutter, Reese, and Schmidt won a Nobel Prize for this observation. And that's been the consensus/dogma spawning all sorts of speculations about dark energy. ChatGPT helped me locate the paper that casts major shade on their study --> "Marginal evidence for cosmic acceleration from Type Ia supernovae" by Nielsen, Guffanti, and Sarkar.
https://www.nature.com/articles/srep35596
The authors analyzed a larger dataset of Type Ia supernovae to assess the evidence for cosmic acceleration. They concluded that the data is consistent with a constant rate of expansion. This challenges the need for dark energy in the standard cosmological model. Whether gravity would eventually overcome the expansion and create a Big Crunch is a separate question. But if the expansion of the universe isn't accelerating, that puts that option back on the table.
As I have replied to Alastair, Roger Penrose came up with conformal cyclic cosmology which might give you an endless supply of bodies to reincarnate into :) even if the universe is getting bigger and bigger because it “blurres” back to small again.
I'm moving to Montrose, CO soon. If anybody lives there (or near there, like in Grand Junction) and wants to grab a coffee together or something, let me know.
Re: Deportations.
It seems to me that the Venezuelans deported by Trump did not received due process because Trump used a law meant for wartime which literally allows him to sidestep due process for enemy combatants in times of war.
Abrego-Garcia wasn't deported under that law, so the courts could rule against his deportation. The Venezuelans were, so the courts couldn't even say anything against it (because, again, the law had authorized the disregard of due process).
In light of this, I'm curious:
- What allowed SCOTUS to, two days ago, order a complete halt of the deportations of these Venezuelans? (see https://www.bbc.com/news/articles/czd3rdjn81lo)
- Why wasn't this order against the Alien Enemies Act enacted before?
- Why did SCOTUS ruled that the courts could command the government to "facilitate" the return of Mr. Abrego-Garcia before it ruled against the Alien Enemies Act?
I'm not trying to argue with respect to this, I'm genuinely curious.
>allows him to sidestep due process for enemy combatants in times of war
>(because, again, the law had authorized the disregard of due process).
https://www.supremecourt.gov/opinions/24pdf/24a931_2c83.pdf
"We have held that an individual subject to detention and removal under that statute is entitled to “ ‘judicialreview’ ” as to “questions of interpretation and constitutionality” of the Act as well as whether he or she “is in fact an alien enemy fourteen years of age or older.” Ludecke, 335 U. S., at 163−164, 172, n. 17... The Government expressly agrees that “TdA members subject to removal under the Alien Enemies Act get judicial review.” Reply in Support of Application To Vacate 1. “It is well established that the Fifth Amendment entitles aliens to due process of law” in the context of removal proceedings. Reno v. Flores, 507 U. S. 292, 306 (1993). So, the detainees are entitled to notice and opportunity to be heard “appropriate to the nature of the case.” Mullane v. Central Hanover Bank & Trust Co., 339 U. S. 306, 313 (1950). More specifically, in this context, AEA detainees must receive notice after the date of this order that they are subject to removal under the Act. The notice must be afforded within a reasonable time and in such a manner as will allow them to actually seek habeas relief in the proper venue before such removal occurs."
"Importantly, as the Court stresses, the Court’s disagreement with the dissenters is not over whether the detainees receive judicial review of their transfers—all nine Members of the Court agree that judicial review is available."
>What allowed SCOTUS to, two days ago, order a complete halt of the deportations of these Venezuelans?
https://assets.aclu.org/live/uploads/2025/04/SCOTUS-application.pdf
"After the district court’s order and starting last night, counsel for Applicants and the proposed class of individuals subject to removal under the AEA were informed that numerous Venezuelan nationals currently in the government’s custody in the Northern District of Texas have received notices that they are subject to removal under the AEA, and further were informed by government officials that they may be removed from the United States as soon as this afternoon or tomorrow. DHS has now publicly announced that AEA removals are imminent.
The Government’s actions to-date, including its lightning-fast timeline, do not give members of the proposed class a realistic opportunity to contest their removal under the AEA. The notices some members of the proposed class have received are in English only and do not inform proposed class members of their right to contest the designation in a federal court. The government has refused to give any information to undersigned counsel for the proposed class. And, as far as Applicants and their counsel know, the government is not giving notice to proposed class members’ immigration attorneys.
Removal without sufficient notice and time to seek habeas relief is in clear violation of this Court’s decision of April 7, 2025 in Trump v. J.G.G."
>Why wasn't this order against the Alien Enemies Act enacted before?
The Trump admin appealed Boasberg's initial Temporary Restraining Order up to the Supreme Court. TROs are normally not appealable because it's irresponsible for superior courts to assert jurisdiction under the conditions TROs are granted - without other parties present, before fact finding occurs, the very first step in the judicial process with little to appeal. The first appeals court denied to block Boasberg's initial TRO preventing deportations under the AEA, particularly because the federal governnment stated in oral arguments that they don't believe deportees are entitled to due process and that deportations would resume immediately.
https://media.cadc.uscourts.gov/orders/docs/2025/03/25-5067.FINAL.2.pdf
"Indeed, at oral argument before this Court, the government in no uncertain terms conveyed that—were the injunction lifted—it would immediately begin deporting the plaintiffs without notice."
"(J. Millett: “My question is, if we were to grant the relief you request, would the government consider it necessary to allow time to file a habeas petition before removing people? * * * [Is it] the government’s position that it could immediately resume mass removals of the five named Plaintiffs and the class members, immediately? Government: “Your Honor, * * * we take the position that the AEA does not require notice * * * [and] the government believes there would not be a limitation [on removal.]”). The Constitution’s demand of due process cannot be so easily thrown aside."
There were further comments in the circuit court opinion that "invasion" is a term of war and that the invocation of the AEA was likely illegal.
Since the Supreme Court was handling a TRO after the appeal from the circuit court, they issued a purely procedural ruling that clarified what I pasted above, that 1) All "alien enemies" can challenge their deportation, 2) Thus must be done via filing Habeas, not via class actions, 3) There needs to be a reasonable time afforded for them to file it. ICE continued its bad faith behavior by giving subjects 24 hour notice and attempting to deport on Saturday when most judges and lawyers wouldn't be working, hence the emergency order by the Supreme Court to halt all deportations that came at 1 am Saturday morning. https://www.scotusblog.com/2025/04/justices-temporarily-bar-government-from-removing-venezuelan-men-under-alien-enemies-act/
>Why did SCOTUS ruled that the courts could command the government to "facilitate" the return of Mr. Abrego-Garcia before it ruled against the Alien Enemies Act?
There was comparatively more fact finding and argumentation that needed to be done in the AEA case, where only the TRO was appealed. In the Abrego Garcia case the government admitted its administrative error and there was already prior due process initiated by immigration judges that granted Abrego Garcia's witholding of removal and the only dispute was whether the government had to do anything to bring him back.
>which literally allows him to sidestep due process for enemy combatants in times of war
That might well be unconstitutional (as Trump's sister ruled several years ago in a case that was overturned on appeal on purely procedural grounds). Edit: My mistake: she ruled re the law being used against the Columbia student.
>Why wasn't this order against the Alien Enemies Act enacted before?
It is my understanding that the Alien Enemies Act has been invoked only about 4 times.
Even if a law says "you don't get due process in this situation," you can still get a court to ask "are we in that situation?" The AEA is an act that applies in wartime, which was invoked to deport members of Tren de Aragua, and the people challenging it are arguing that the law can't be used on them because it's not wartime and they aren't members of Tren de Aragua.
As far as I know, the timing of the SCOTUS ruling is just the vagaries of the justice system - it took time for the case to work its way up to SCOTUS and for the lower courts to wrangle out exactly what the government was doing wrong and what they were willing to enjoin. Also, ICE was planning another set of deportation flights so it suddenly became urgent - there were buses literally on their way to the airport which turned around once the ruling came out.
The Federal courts are absolutely allowed to tell the Executive Branch, "Stop doing that; it's quite illegal". Provided, of course, that it actually is illegal, and that someone who has been actually harmed by this has raised a complaint (in the right court, but we don't need to go into that here). Once you get past the level of trial courts, and certainly by the time you reach SCOTUS, that's basically what the courts are *for*. The president and his officers, sworn to uphold the laws, then have to obey the court's injunction against not doing illegal stuff.
As a corollary, the courts can also tell the Executive Branch, "Hold off on doing that for now; it would cause irrevocable harm and we haven't figured out whether it's legal yet". Provided that the act in question really is at least plausibly illegal and irrevocably harmful. That's where we are with most of these cases.
There's certainly a plausible case that the deportations are illegal because the law being cited to justify them requires that the United States be at war with someone, and there's a distinct absence of e.g. Venuzuelan jets bombing American cities or vice versa. The courts are I believe properly skeptical of claims that the lesser ongoing unpleasantness of US-Venezuelan relations constitutes a "war" under the law, or an "invasion" or a "predatory incursion" or whatever. But that's something that needs to be argued in court, with lawyers from both sides making their cases.
And I think we can take the "irrevocable harm" part as given, now that the Presidents of both nations have plainly stated that it is impossible to return deportees from El Salvador to the United States. But that's the part that was (briefly) argued in court before SCOTUS issued the preliminary injunction against deporting anyone until all of this is sorted out.
It took this long to get to that point, because SCOTUS can't just issue rulings out of the blue, it has to wait until (laywers representing) a complainant who was actually harmed bring a case before a US District Court, which is then sent up to an Appellate Court before finally getting to the Supreme Court. Or both sides can agree that the District Court got it right and let their ruling stand, but Donald Trump doesn't seem to be in a very agreeable mood.
>"…the law had authorized the disregard of due process"
Quibble: it's just that less process than default is actually due under the conditions covered by the AEA, not that due process was disregarded.
I think the short answer is the whole situation is a hot mess and everyone is trying to figure out how to deal with it. I am sure there is a lot of tension within the Supreme Court about how to deal with the alien enemies act and it’s application. So they dodged that question and concentrated on the habeas corpus issues that they raised when they instructed lower courts to facilitate Garcia‘s return.
The newsletters I know that are a bit more technical with this side of things are Joyce Vance’s (“Civil Discourse”) and Steve Vladeck’s (“One First”), both on Substack. They’re not neutral and I can’t vouch for their accuracy, but they seem to know what they’re talking about.
Anyway, my (amateurish and not completely clear, please double-check everything I write) understanding is that the SC order halting the deportations is temporary in effect until the issue is properly litigated by one of the lower courts.
The rationale is that the situation was time-critical, as the Venezuelians were being loaded into buses towards the airfield (from which deportation flights would take off) – and then lead them to CECOT from where the detention could not be meaningfully challenged.
This seems similar to the situation a month(ish?) ago, where the government essentially deported the immigrants at the same time the judge was ruling on the merits (against the government), but the timeline was pretty confused and the government seems to have sought to argue by “fait accompli”.
The Abrego-Garcia situation is another one where the government claimed in court (although they later said something else publicly) that the deportation to El Salvador was a mistake (since there was a court order preventing that), but that since it was done and Abrego-Garcia outside the jurisdiction of the US, there was nothing to be done and the entire proceedings were moot.
The court felt differently – and ordered the government to fix its mistake. The SC tempered the language because the Executive branch is the one deciding foreign policy, but it’s still expected to correct its own mistakes.
TL; DR: different cases, technical arguments, the judicial process in the US is complex, and, as far as I can tell, many of these arguments are about process rather than the fundamental questions (eg is Trump’s use of the Alien Enemies Act correct? Should anyone else get a say in what constitutes an emergency or a foreign invasion? Should anyone else’s opinion about the extent to which the claimed actions address said emergency or invasion matter?).
The judicial process is complex.
But doesn't the Judiciary need to follow rules?
https://reason.com/volokh/2025/04/20/justice-alito-dissents-both-the-executive-and-the-judiciary-have-an-obligation-to-follow-the-law/
There is an argument that the Judiciary are at least bending rules, if not out right not following their own rules.
“The judicial process is complex” just meant that it had an internal logic hinging on intricacies that wouldn’t find their way into a newsline and that most wouldn’t remember unless they were really interested and/or professionally had to deal with that.
These include the written rules, the unspoken customs, the randomness of scheduling, the extent to which people think they can get away with bending or breaching rules or customs… this was an absolutely agnostic on what was actually happening inside the courts.
Anyway, there may indeed be an argument that the Court is bending its own rules. Here is also, in the spirit of balance, an argument that Alito’s dissent contains mistakes of fact, and raises “issues” that apparently didn’t bother him in other cases: https://www.stevevladeck.com/p/145-justice-alitos-misbegotten-dissent .
I am very open to the arguments that SCOTUS should act quickly in the modern age, especially when District Courts are ordering nationwide TROs.
But some of the left leaning SCOTUS judges previously opined that some of these issues shouldn't be taken up by SCOTUS because the issues weren't "ripe" (eg no urgency to review, appeals hadn't run its course). Of course, they did that because their "side" was favored by letting it play out slowly.
SCOTUS is taking up the birthright issue and I applaud that SCOTUS will probably resolve the issue probably this year. With a decision either way, we will get a ruling quickly rather than each side fighting in numerous District and Circuit courts for years.
I just want the Judges to be Judges. Right now, the Judiciary is more of a oligapoly where they make up rules to suit their policy preferences.
>SCOTUS is taking up the birthright issue
It actually isn't. It is only taking up the universal jurisdiction issue.
https://www.scotusblog.com/cases/case-files/trump-v-casa-inc/
Edit: Should read universal injunction
You are right. I read the sources carelessly.
The Alien Enemies Act is not solely applicable to wartime. "... war, invasion, or predatory incursion."
... by any foreign nation or government"
Gangs often collude with foreign governments in some capacity. So did pirates when the law was written.
Are you claiming that Venezuela is coordinating with MS-13?
I think it's very likely there is some connection between them.
I'm listening to Stephen King's Dark Tower series as an audiobook -- it's got a great reader -- and am loving the series. Anyone else here a fan?
Loved the first three books, couldn't get through the 4th no matter how many times I tried so stopped there.
I don't know if I'm a fan, exactly, but I can say I read the *entire* book series. There definitely was a volume or two where the problem was Too Big To Edit and the tactful use of a chainsaw to cut down some of the thicket of prose would have improved matters hugely. The ending was something of a disappointment, at least to me.
But there is a good story in there, good characters, and the way he tied it all in to other elements in other stories to create a background universe behind all his works was very cool.
It was also interesting to see the change in his writing style from the original text, somewhere in the 70s, to what he was producing in the 90s. I've seen it with other long-running series where authors are writing over decades and the stylistic quirks of one era stand out by contrast to the current style.
It's not my favorite fantasy series, but I enjoyed it. I bought the series after finishing the second book, but I think the second book was the high point, and the late ones had some instances of noticeably weak prose. (Likewise, the poem it's based on has a wonderful opening verse but then perhaps bogs down. https://fivers.typepad.com/files/childe-roland-to-the-dark-tower-came.pdf)
Finding out the movie Hearts In Atlantis was actually based on Low Men In Yellow Coats, which was a Dark Tower side story, was an interesting factoid.
Read through them sometime around university, generally liked them, really liked the ending, haven't reread them since except to dip in and reread a few favourite scenes. I love the overall aesthetic, especially of the first few books.
Just listened to this: “When they finally arrived they found the castle deserted except for the rooks and blackbirds. The walls had been broken. There had been a great slaughter on the fields to the west. It were white with bones and red with rusty armor, so my dad’s granda said. And the voices of demons cried out like the east wind from the jawbones of those who’d fallen there. A village beyond the castle had been burned to the ground, and a thousand or more skulls were posted along the walls of the keep.”.
I was a big fan, right up until the ending. I thought it was sort of a cop-out. Maybe you'll like it. I don't know. But I won't spoil it for you.
Ever since I watched the death of Gus Fring scene, which wasn't even the end of Breaking Bad, just a season finale, all other endings seem a bit lame.
Fring got dinged by Hector and Walt
Hey Gunflint, discovered another cat toy. I think this is their favorite of all time. They chase it wildly and never get sick of it. My smart cat knows that if he paws randomly at the controller he will eventually turn it on.
Here’s Popo playing with it.
https://tinyurl.com/yej3ymke
https://www.amazon.com/AsFrost-Upgraded-Interactive-Treadmill-Stimulation/dp/B0DHZZSPMB?ref_=ast_sto_dp
That's awesome, I'm getting one!
Cool
Thanks!
What do you do to relax before going to bed? Read?
I'm considering taking up whittling as a manual hobby before bed, to get off screens for at least an hour before bed.
If you like crafts, kumihimo braiding on a marudai is very meditative. https://m.youtube.com/watch?v=4sqB9WgGYbI
Marudai part is about 50 secs in. The man’s doing a very complex pattern but you can make striking things with just 6 strands.
I typically throw on one of a couple dozen Youtube playlists I've seen a thousand times, and use it as white noise to fall asleep.
Foreign language videos work as well; just, something my brain stops trying to pay attention to.
Feed and play with the cats. I throw bouncy balls into the bathtub from a few feet away, banking off bathroom walls, and they dart aroundin or near the tub trying to catch them in the air and sometimes succeeding, and otherwise just staring at each for the longish period it takes it to come to rest.
Read. For a long time, I read paper books or e-ink screens using a red or amber reading lamp while getting ready for bed. The red lamp in particular was useful for not disturbing my wife on nights when she went to sleep before I did.
I have since fallen out of the habit of this, in favor of reading on my phone instead. I should probably try to return to my former habits, which worked very well.
Read, mostly. I use lower color temp option for the kindle app to make it easier on the eyes.
Read, sometimes. Work, sometimes. Most of the time I wind down by rereading my notes from the day and making edits or summaries, and reviewing notes from previous days. I've had a paper note system for years, so there's no screens involved except for infrequently taking a quick picture of something with my phone.
What are these notes you talk about?
I take notes on index cards throughout the day. They're a mix of short range (grocery list, diagram of directions, ingredient amounts for a large batch recipe) and medium range (minutes from a work meeting, quotes and insights from a book I'm reading, a sketch and background info for someone new I met at a dinner party) and long range (tools or techniques from a class, definitions of new words, reminders for myself at the end of the year.) It's a bit scattershot in the first draft format, so I try to neaten it up at the end of day.
Re: college admissions essays, perhaps the best thing would be for America to adopt the British system, whereby students do write admissions statements, but they're expected to stick quite closely to "Why do you want to study this topic? And what makes you think you'd be any good at it?" That lets students mention any *relevant* extracurriculars ("The summer I spent volunteering in Colombia really helped me improve my Spanish, which will be useful for this degree in Hispanic Literature") or mitigating circumstances ("Although my B in Music might not seem very impressive, I think it becomes more so once you remember I've been deaf since the age of seven") without encouraging the sort of extracurriculars-padding or sob stories which I gather are necessary in the US.
I know of a person who got into RISD (Rhode Island school of design in Provincetown) in a very clever way. She mailed in her application envelope and inside there was a one line note; it said “if this letter gets to you, you should admit me to your school.”
A close examination of the envelope revealed that the postage stamp had been hand drawn by her. I saw a photograph of the envelope. It was perfect.
She did get in.
An oddity of the British system is that places are awarded based on letters of recommendation or predicted grades from second level teachers, and the offers are made conditionally before the A-level results are announced. I read some time ago that there is a substantial discrepancy between predictions and actual grades at fee-paying schools; and that this discrepancy is one of the drivers in the inflated rates of progression from fee-paying schools to prestigious universities there. Offers are almost never rescinded - so the net result is poorer performing students from the fee-paying schools beating less wealthy students in the race to Oxbridge.
If your exam results haven't been announced yet, your offer is conditional on your actually achieving your predicted grades. If your teachers have been over promising, you probably won't get the grades, and hence won't fulfil the conditions of your offer.
As I understand it, in the UK, the offer typically stands even if the student doesn't achieve the predicted grade:
https://www.edsk.org/publications/university-admissions/
I've never understood their system.
That certainly wasn't the case when I was applying; maybe it's an issue that started cropping up recently.
I don't have an issue with that but American educational establishment just doesn't like that sort of specialization in high school/early college. That's why we study Languages, Sciences, Math, Social Studies all throughout high school, rather than specializing in 2-3 subjects in the last few years of high school which I think is the norm in many(most?) countries. This doesn't end in high school either as you will still have to take history even if you are computer science major for example, unless you tested out of it by passing various AP history exams in high school. For your proposal to be implemented, the whole American educational philosophy would need to change. And students not having enough unique stories for their college essays is just not that big of problem for that to happen.
> That's why we study Languages, Sciences, Math, Social Studies all throughout high school, rather than specializing in 2-3 subjects in the last few years of high school
Hm, I think you got that wrong. The term "specialization" is used indeed in other countries, but means something much weaker than what you describe. At least in Germany, and I think it's more or less the same for other European countries.
You don't just have 2-3 subjects in the last few years of high school. You have all kinds of subjects with a bit of choice, for example if you "specialize" in computer science you still have History/Social Science/Economics, and you still have arts classes, and you still have some natural sciences and German classes and at least one foreign language and so on. "Specialization" means that in the selected 2-3 topics, you have a few extra hours per week, for example 3 hours in the non-specialized courses versus 4-5 in your specializations. Even the specializations need to be broad, usually they need to include at least one language, one STEM subject and one social subject. (Details vary on region and school.)
You are right that it is different in college. There many European countries do have the specialization that you describe (though not all, for example France also starts with two years "CPGE" of more general education).
Ok fair, I was mostly thinking of British A-Levels. On the other hand, don't Germans have different kinds of high schools? Like a school where academically inclined students go to, a school where vocationally inclined students go to, etc? And a large number of students actually go to vocational high schools and the sorting process for that actually happens fairly early on in their lives? Nothing wrong with that system in my opinion, it seems to work quite well for Germany, but that kind of sorting/specialization just doesn't align with American educational philosophy or even American culture more broadly.
My experience, mainly applying to California colleges c. 2000, was that almost everywhere required or strongly encouraged you to declare a major on application. You could change majors after matriculating, but there were a lot of hoops to jump through and some fairly strict veto points unless you were transferring between majors within the same "college" (at my university, departments were divided between six "colleges", each with their own sub-administration: Agriculture, Architecture, Engineering, Liberal Arts, Science and Math, and Business) and even then it was easier if you were moving between closely related majors and if the target major was considered less technically rigorous than the one you're leaving.
That matches my experience as well. I declared my major in my application but my college essay topic was a general one. And I think my high GPA and SAT scores probably played a more important role than the essay in my application.
Can your therapist go meta?
Excerpt: https://debugyourpain.substack.com/p/can-your-therapist-go-meta
> Something I've noticed is that every specialist sees pain through their own lens:
- Physiotherapist: "It's muscular imbalance"
- Doctor: "It's biochemical"
- IFS therapist: "It's your exiled child part"
Where are the people who can move between the different systems?
No matter which way you go the doctor is right. It is biochemical. The question then becomes what do you do about it. Every person sees pain through their own lens. How could they not?
“ It is biochemical”
Is it though? The correlation between, say tissue damage and pain sensation is much weaker than a purely biochemical nature would entail. Solders reporting looking down and realizing their leg’s been blown off is one end. Chronic pain with no discernible cause is the other. Then there’s phantom limb pain many amputees report.
Pain sensation is ultimately made up in the mind/brain, so in this sense it is biological, but it’s then trivially true in a non-interesting way.
Ok, I see what distinctions you are trying to make.
They are all flesh and blood issues but as you say, I suppose that’s trivial…
although I am not entirely sure that it is when you assess the three choices that were listed. I can see the distinction between pain that is caused by clear physical trauma, and the kinds of pain that aren’t. Most of what we call emotional pain has physical manifestations. An aching in my heart, frustration; shock. The subjective nature of pain defies categorization don’t you think?
Exactly. Full disclosure: I participated in some early research these guys were doing as a willing subject precisely because pain fascinates me. It’s entirely subjective, and psychology plays a major role. There’s a reason inquisition torturers first just showed the tools. OTOH as a martial artist I just learned to ignore certain kind of pain that is not “dangerous”, i. e. doesn’t lead to damage.
Any chance this is private healthcare (ie, wrong incentives) related? I've never had a doctor insist it must be something related to their specialty, quite the contrary.
As to the person who moves through different systems, I'd say that's the patient, sadly. GPs can take that role to a degree, but I feel that once you are off the basics, you are on your own.
Picked up a copy of https://store.steampowered.com/app/2844290/Hard_Chip/ on Steam the other day, enjoying it so far. It's the latest incarnation of the "hey, this game is just my day job but slightly simplified" zachlike genre, and probably the most hardcore one I've seen to date. I feel like of the public spaces I inhabit online, this is the most likely one to contain someone else who might appreciate it :)
It looks like a non-trivial version of Turing Complete. I'll definitely try it the next time I'll have time to play anything. Thanks for the rec!
Zach himself actually did a game about semiconductor design (KOHCTPYTOP: Engineer of the People) way back when, although this looks bigger and far more polished. Definitely a hardcore topic for a zachlike.
>"hey, this game is just my day job but slightly simplified" zachlike genre
I've often wanted to make one of those games based on my own job. I'm kind of a general manager type at a small business: not the owner, but in charge of a lot of admin work. The game would consist of your boss giving you more tasks than you can actually physically complete during an in game work week, you have to decide what to work on and what to let slip through the cracks. Every week you have a meeting with the boss and try to show off your accomplishments while hoping that he doesn't ask about the tasks you didn't do.
...oh, kind of like "papers, please" but banal rather than pointedly political?
Yeah, I just want to capture the stressful feeling of having too many plates to spin at once and having to hide the broken plates.
Scott, I'd love to see you delve into systems theory in a bit more depth. I think you'd have some interesting insights if you thought about this a bit more:
> Oh! I agree this makes sense if you need to talk about the “purpose” of an un-designed system with no humans in it.
If you build out POSIWID as a way of understanding undesigned systems, you'll end up with an entirely separate perspective that you can apply to all systems -- designed or not. And you might find this perspective is a useful adjunct to hearing what designers say about it.
To make an analogy to the mind, you might think of POSIWID as akin to an MRI -- it gives you a rough structural view that you can try to tease apart into an understanding. You'd be right to say that we shouldn't rush to understand everything in terms of MRIs when we can simply talk to people about what's on their minds. But nor should we entirely discount the information from the MRI in favor of what the person says.
(To be clear, I agree with your posts about how absurdly POSIWID is misused on Twitter. But if we judged ideas by how they were misused on Twitter, what ideas would we have left??)
POSIWID is a useful concept applied to ourselves. A human being is a system. (Opinions will vary as to whether it is a designed system or not.) Our behavior is the output of that system. The question then becomes, is the system doing what we want it to do?
Indeed, but our desires are another output of the system. If it's alignment we seek, should we change the world to match our desires, or change our desires to match the world?
Our desires are another vector in the system that leads to our behavior. As to your question about what needs to change or what should change, it’s a constant negotiation, isn’t it?
I'm interested in people's experiences with using LLMs for automation.
I've tried using LLMs for automating a certain part of my job, but unsuccessfully.
The main problem is that it's impossible to communicate in enough detail requirements of the task. When humans do this task, they learn it on the job through a long series of reviews until they end up having intuitive grasp of what's expected of them.
However, this means that there are no explicit instructions that could convey all necessary information to do that task.
Passing many examples for in-context learning does not help, the generated examples still fall far too short. I haven't tried fine-tuning on raw data dumps from reviews, mainly because it would take a lot of effort to format them properly and I have a lot of doubts that it would help at all.
For this task, reasoning models work far better than non-reasoning ones, even though the task is creative in its nature. This is mainly because reasoning models will roughly follow long and detailed instructions, while non-reasoning models will ignore almost all of them. This is why I doubt that fine-tuning would help - I haven't seen that any reasoning models can be fine-tuned and non-reasoning models already perform so poorly that I doubt fine-tuning would make them useful.
I think this is why benchmark progress with LLMs rarely translates into useful work - current top tier LLMs are already incredibly good at tasks which can be easily described to them and making them even better at those kind of tasks will not suddenly make them able to automate a lot of work.
It seems to me that a lot of people expect that you can cheat conservation of information with LLMs - that tasks which require humans to learn for 2-12 weeks can be described in a simple prompt. In most of the cases that information does not exist in textual form which could be given to LLMs and even when it exists we don't have the tools to make LLMs incorporate it properly.
All of this makes me skeptical that LLMs will be used for automating anything more than extremely straightforward tasks until the issues above are resolved.
Has anyone managed to resolve the issues above, or work around them, on real-world tasks for which you pay people to do?
I'm a psychologist in private practice, and had thought about using LLM to help me with my billing, and, like you, I don't see a way to do it. My system is kind of cumbersome. I do not use practice management software. Instead, dates of client appts are in my computer calendar, and the rest of the info is in 2 other folders. When I do billing, I bill some people's insurance, and other people directly. I have a folder, organized by patient initials, in which I keep a copy of bills I sent. I have another folder, also organized by patient initials, in which I keep a record of appt dates, bills sent, and payments. Payments come in the form of bank transfers, PayPal payments, and checks, and I enter those in the second folder.
Yes, the system is cumbersome, but I could explain my system to any bright high school student and they would be able to do the whole thing for me. I can't figure out way to turn it over to an LLM. I don't see a way to explain the system to it, give it access to my calendar and the 2 relevant folders, and just tell it to have at it. How can these things have anything like human level general intelligence when you can't get across to them how to do a messy, but quite simple, task like this? Or, I dunno, make one of the advanced GPT's could? If I entered instructions plus the relevant calendar, payment details and folders, , could it enter dates of service, charges and payments in the records, and make out bills?
You're looking for an AI agent, which is still a nascent field. LLMs simply transform text, so you can talk to them and have them answer questions, but the ability to do things for you in the real world is limited to text-based triggers they can invoke.
There's a lot of talk about this year being the year of the agent in AI, so there's likely to be a lot of progress on this coming soon. But even then I don't think it'll be as simple as turning it over to an AI. Given this is such a critical function, it's worth thinking about how you'd modify your current system to allow you to oversee an AI that's doing most of the work but makes occasional mistakes.
E.g. perhaps you OK each bill before it goes out, and OK each paid bill.
But what is being an agent, really? There are certainly tasks like my billing task, but simpler, that AI can do right now. For instance I know a teacher who used GPT4 to write the paragraph that went out with each student report card. He has a number of comments he makes over and over when he writes these paragraphs, so he made a list of common comments. For instance about homework completion, he might most commonly say “continues to do a great job with completing homework” or “homework completion has greatly improved over last term” or “continues to have a problem with completing homework” etc . So he made a document with a list of comments. The homework comments were one group, with individual comments labeled h1, h2, etc. Then there was another set of 6 or so possible comments for, let’s say, participating in class discussion, and they were labeled d1, d2. And so on. He gave GPT a list of student names, each followed by several alphanumerics corresponding to the comments that applied to that student, and also the document with the list of alphanumerics and the comment that went with each. Then he asked GPT to write a paragraph to go with each report cord, in a friendly, casual prose style, incorporating all the information corresponding to the alphanumerics after each student’s name. GPT did just fine with this task, including recognizing on its own that in order for the paragraph to flow well it could not just use one sentence for each alphanumeric. It often put two or more together in many sentences, making sentences like “Amanda continues to have trouble with homework completion, but has become a lively participant in class discussions.”
Seems to me that handling my billing is not any different in kind from the task the teacher gave GPT, just has most moving parts. If it did my billing all GPT would have to do would be follow clear instructions about what info to put where in what records document, and what kind of bills (personal bills or submissions on a form for their insurance company) to generate for each patient. Are you thinking that to be an “agent” that can carry out my task GPT needs to engage in some process that only conscious beings carry out — would have to “think,” “understand”?
Oh no, I'm definitely not thinking there's a difference in kind between your billing task and the report card task. The difference is in who's creating the "glue" that turns the textual task (which LLMs are good at) into something that actually affects the world.
To give an example from my day: I'm a software engineer, and I took the week off to work on some personal projects and see how much AI could help me. I had a complex 500-line script that I wanted to add a couple of features to. I described what the script did and what I wanted it to do, and I asked Claude to rewrite it with a couple of new features. And it immediately did so -- doing work in about a minute that would have taken me all day. It was stunning.
But what it couldn't do was _run_ the code. This crosses over from the "manipulating text" domain (where it excels) to the "using tools" domain, where current LLMs are fairly limited. For the most part they have some set of integrations that they can invoke textually (e.g. ChatGPT can call an image generator to make pretty amazing pictures) but they can't operate a computer or a web browser directly. In other words, they need special integrations to actually perform tasks (as opposed to giving you back some text) instead of being able to interact with computers the way people do.
There's no real reason this problem can't be solved; in fact there are a bunch of great tech demos and products in development in this space. E.g. OpenAI's Operator, Anthropic's Computer Use. These should be able to solve the problem you described, they just require new training, quite a bit of inference compute (they need to process a lot of images, which is more expensive than text), and quite a bit more trust. (The latter will be the biggest hurdle -- if text generation goes off the rails 1 time in 100, that's unfortunate but not ultimately a deal breaker. If the AI is controlling a web browser or your computer it can accidentally do much more harmful things, like deleting files or accounts.)
Given your existing flow this may not be worth the effort, but I imagine that if you gave an LLM all the information it could figure out the right thing to do. It could look through and understand the two folders of patient files, it likely has a calendar integration or could understand what you copy and paste in from your calendar, it could trawl through bank statements or emails to find payments, etc. It might also have some spreadsheet integrations where it can add rows to, say, Google Sheets.
But it's probably not going to be able to gather data from your online accounts or read files from your computer. It just doesn't have the right interfaces at this point. The major AI companies are working to fix this over the course of 2025 though!
Then the question will be: how do you modify your process from where a single, trusted person is doing everything to where you've got a really smart and motivated intern whose work sometimes needs review?
To me this is a task for an IT rep to design some Power Automate flows to read emails, interact with excel and attachments. It would require a different technical skillset than working with an LLM, but the tools are very powerful at being agentic in very specific, repeatable ways.
<Then the question will be: how do you modify your process from where a single, trusted person is doing everything to where you've got a really smart and motivated intern whose work sometimes needs review?
Is it worth trying this now -- modifying process? I could certainly upload the relevant folders: one of actual bills, each patients' bills, current and past, in a subfolder within the folder; and one a folder of spreadsheets, one for each patient. Columns in spreadsheet are date of service, charge, date of billing, amt billed, date of payment, amt paid. And rather than having the AI trawling my back account,etc., I could just occasionally upload a list of recent payments -- who made them, date of payment, amount. And I can explain very clearly the steps that must be taken to enter appointment dates, make out bills, etc. I can provide a model of the 2 bill types used. But would that work? Is there an AI that would remember what I'd taught it, including mistakes in its process that I had corrected. Is there one where I could just leave the folders stored with it? It does not need to, for ex., print out bills. If it makes them available as text files or even image files I could download and print them myself.
Ok, I had to try running our conversation through Claude to see what it would say:
https://claude.ai/share/7d24f156-c0a8-4f14-82e5-3496cbd24e63
This is mostly a summary of the conversation thus far, but it'll be great context for helping it understand your goals. And it'll be able to answer very specific questions like "how can I integrate my .xlx spreadsheet into an online workflow".
(I have to say it's kind of weird for me to realize that if I were replaced by an AI in this thread it's very likely you would have gotten even better answers than I was able to give!)
IMO it's definitely worth trying this now. Even if it doesn't end up doing what you wanted, it'll give you a deeper understanding of how to interact with these systems and a better idea of what features are missing.
The simplest prototype would be to start a conversation with Claude or ChatGPT, and see if you can get it to do processing steps. E.g. "here's what the spreadsheet looks like, I'd like you to generate new rows based on..." It may stumble, but ultimately it's a conversation so you can just tell it what it did wrong. It'll probably pick it up pretty fast, unless there's some hidden complexity here I'm failing to imagine. Once it does, you can come back to the same conversation later and say "here's more stuff to process in the same way."
Of course the utility of this will be somewhat limited, since you're directly managing the inputs and outputs (which is probably most of the work anyway). I just learned about Zapier today, which might be exactly the adjunct you need -- you can use it to define flows of various kinds, and assuming it has the right integrations (which is a big assumption) you might be able to set up a flow like "every day at 5 AM fetch new email, send it to Claude with this prompt template, and insert new rows generated by Claude into the 'review' spreadsheet".
But creating this workflow with Zapier sounds like a lot of work! So now here's the fun part: I bet Claude (or flagship LLM of your choice) can help you design this workflow. Just tell it what you've got and what you're trying to do, and I bet you'll get a good answer.
I think this is precisely the sort of task where AI is going to take much longer to catch on to what is required. There are all sorts of unwritten rules/procedures that you don't realise you are following. And at best, AI is prone to strange hallucinations - someone does need to check its work. My own opinion is that this will be the downfall of many AI applications: something works 99% of the time, the overseer gets lazy, and it only takes one really idiotic failure (data leak?) to end the experiment in a given company.
I think it's deeply ironic that the work of what we perceive to be highly skilled professions like programmers, accountants and lawyers (here I mean trawling through case law to write legal documents, not arguing in court) will probably be automated much faster than secretarial work. The difference is that these professions are text and rule based, and you can quickly check if the output is correct or not. In contrast, secretaries don't just make appointments and type letters - they handle all sorts of exceptions and errors efficiently and appropriately.
Out of interest - is there a reason you don't want to use off-the-shelf software to do this? AI really isn't the right tool, or at least the currently available systems are not in any way specialised for this.
Oh I tried it years ago and did not like various things. They could all do lots of things besides billing, none of which I wanted, and access to the extra features complicated the workspace. And then there were various littlethings they didn’t do the way I do. Like maybe something like the bills they made out looking too much like gas co bills and not informal and friendly enough for my taste. And then there were various little exceptions I did not see how to make it make. For instance some people
don’t get bills because they choose to just pay at each session or at the last one of the month, and it was more trouble than it was worth to get the software to take that into account and not make out bills to those people.
Some of the EHR systems are rolling out AI features: the one we use at our office is focused on AI notetaking. You can have their AI app listen to the psychotherapy session, then it takes what it hears, transcribes it, and puts it into proper note format.
We haven't actually used it yet because it's got a lot of bugs, but if it works it would certainly save a lot of time.
Oh, I wouldn't like that. First of all, my notes aren't summaries of what was said, since I remember the general info from stuff talked about in sessions the same way I remember long serious talks with acquaintances. My notes are insights or ideas or questions or unfinished business. Also, patients hate having their sessions recorded. I think I would feel the same in their place.
Your patients would put up with having AI listen to the session? I have people who won’t even bring their phones into my office because of concerns about spyware! And most people loathe doing exercises that I sometimes propose that involve recording themselves, such as doing some kind of role play and listening to it together.
IME fully automating large or complex tasks is not possible with the current generation of models. But they're great for increasing the leverage of skilled people by automating the smaller steps.
E.g. when I write code I find they're great for selecting libraries or design patterns, and great for writing clearly-specified functions. This dramatically reduces coding time, especially when working with unfamiliar libraries, languages, or design patterns.
I'd be curious to hear more about the specific tasks you're trying to automate. They sound too big, or too heavy on local context to be easily asked of a generic LLM.
I put up a description of several related tasks that are very simple and routine. Wondering whether you think an LLM is up to doing them.
I don't see where you put them up. Could you point me in the right direction?
It’s here: https://www.astralcodexten.com/p/open-thread-378/comment/110948596
You did respond to it.
Thanks! Didn't make the connection :facepalm:
The GOP continues its speed-run to electoral irrelevancy:
https://www.newsweek.com/marjorie-taylor-greene-says-evil-being-defeated-after-pope-francis-death-2062157
I'm fairly sure the Democrats are somewhere working hard to maintain the balance
For pure batshit hatred? I don’t think so.
No, of course not. I mean only the balance of electability. I haven't seen anything that makes me believe that the democrats are attempting to appeal to the tipping point voters they need to win elections.
You're very correct - there isn't much in the way of messaging adjustment going on.
Of course, it probably won't matter. Between thermostatic public opinion and what Trump is doing to the economy, they won't have to.
But if the weren't given those tailwinds it's hard to see how they would succeed electorally.
I know this sounds cliche, but she's doing it for attention.
She'll follow up in a day or two with all the "horrible attacks" on her in a fundraising email.
This makes sense. The fact that her people fall for this every time speaks volumes abo... nevermind.
What? I thought American Catholics hated the pope.
Only conservative Catholics, and even among them I'd say only a very small minority *hate* him.
It's a big deal in Catholicism to *hate* the Pope.
Are there a significant number of Catholics in her district? Or does she represent the "evangelical protestant who may or may not still identify Rome with the Antichrist" crowd?
You were misinformed.
Only 60% of conservative Catholics view him favorably. Higher than I thought, but still...
Regardless, the church's rhetoric is directly opposed to the interests of the administration. They're not going to take it lying down.
Even a lot of those who don't view him favourably aren't going to take well to openly celebrating his death.
I had a great spin on the traditional forgot-to-study-for-the-exam-tomorrow dream. There was a whole room of us, all frantically revising, and as people passed more and more topics around I realised I'd completely missed more and more of them, and things went from bad to worse in the traditional manner.
Then I suddenly went, hang on, this isn't right. I asked a classmate, "Paul, how old are you?" and he thought for a minute and said, with surprise, "35." "Don't you already have an engineering job?" "Yes. A very good one."
By this point all the other students were listening and realising the same thing. There was a short period of mutual sanity checking, and we all began questioning more and more: "Wait, these topics don't even go together, why is one exam testing all of them?" "Actually, guys, does anyone even know what institution is setting this exam?" "Does anyone actually recognise the library we're in right now?" "What the fuck is going on?"
At no point did any of us think to suggest we might be in a dream.
I once had one of those dreams where I thought I'd just woken up (but hadn't), and I noticed a strange quality to the light, and it felt sort of threatening and unreal. Spent a few minutes feeling like something was going to jump out at me before I woke up for real.
A week or two later, I was looking across a big grass field and I suddenly noticed the same strange quality to the light. I thought to myself, "Wait! That's just like the light in my recent dream! Only this time I'm seeing it while I'm awake! What could it possibly mean?"
On New Year's Eve, 1999, I was hanging out with some friends, and went to sleep on their couch in their basement after having a few jokes about all the computers crashing cataclysmically from Y2K bugs.
I woke up to a very bright light.
It took several seconds for my brain to realize I wasn't at home, the basement had patio doors to their backyard on that side of the house, and the sun was shining in my face.
I would read this short story.
I have that a lot, dreams where I'm in high school but aware that I have a good job as a software engineer, and a college degree. My brain often splits the difference and decides I'm auditing the classes and can skip if I want to.
Yeah mine also seem to come up with some half-assed excuse why I'm still hanging around high school and/or college despite being way past that stage. It does often seem to lower the perceived stakes and stress level at least a bit.
>There was a whole room of us, all frantically revising, and as people passed more and more topics around I realised I'd completely missed more and more of them, and things went from bad to worse in the traditional manner.
Something similar actually happened in my old university, where one year it turned out some of the professors had taught completely the wrong syllabus for a given paper. They ended up having to produce a new exam for the four or five students affected.
I've had dreams where I'm working in Excel or on a program and I just can't figure it out. And while I can never actually figure out that it's a dream, I am able to realize that I will be unable to save my work so try to convince myself to forget about it. Usually works but then I might start up again on another problem and be right back where I started.
When I have dreams like that I sometimes let myself believe I might be dreaming with others in a kind of strange community.
It's a nice thought, but it never seems to go anywhere, and never seems to make sense from either my evolutionary or theist viewpoints.
That’s a good one.
I tried to convince my son to make an essay for one of his 17 college applications to be, "How I managed to secure every #1 pick in the 2028 NBA2K draft". One of the video game's playing modes was to build a team over the years and manage draft picks, and he was able to massage it so that he had the entire first round. I thought that would have showed ingenuity and tenaciousness and it certainly would have been memorable to an admissions officer (indeed, I wouldn't have been surprised to see one of them here mentioning this essay!). Unfortunately, Management vetoed that as an option before it even hit the draft stage, so he had to talk about the hardships of growing up in a leafy suburb.
Good call. College admissions officers look down on vidya games.
I doubt that management made the right call. College admissions officers I saw quoted were rolling their eyes at hardship essays by kids without significant hardships in their lives. Seems to me that choosing to write a hardship essay under those circumstances is a likely marker for conventionality, dishonesty and lack of confidence. Besides, common sense suggests that applicants will do better with an essay that stands out in some way. If somebody has suffered a hardship that itself stands out, then the essay will be memorable. Otherwise, it will not.
And wtf is wrong with kids these days? My daughter in high school wouldn't even listen to my advice about how to do better in geometry. When I was in high school I wouldn't listen to my parents' advice about anything. They never even saw what I wrote on my college applications.
There's always the worry that you blow it by saying you're qualified for a top choice school because you outsmarted a video game. Management would have gone through the roof on that one. But it would have probably been worth a shot at one of the unlikely colleges. I think his hardship ending up being about breaking his foot prior to his junior year basketball season after he had worked so hard to get into a good position. And I've already forgotten though it's only been four years, but there seemed to be multiple essays required for each one, only some of which were "hardship" essays.
Hmm, another example of how one purpose of a system what it does? The purpose of universities is to filter applicants such that demand for slots within the university corresponds to the supply, in a way that no stakeholder will find too objectionable. Essays look like a good tool for that.
The things you stated (winnowing applicants, avoiding major objections from stakeholders) are requirements for the admissions process, certainly, but there are many ways they could fulfill those requirements, and some of them are substantially less expensive than the current system (such as lotteries, as others mentioned, or auctions, which would increase revenue). You can say "well I guess all those other methods were unacceptable to some stakeholder for some reason" but that doesn't actually explain anything.
Seems like you're experiencing one of the failure modes Scott highlighted, where you pick one particular effect of the system (or two, in this case) and call that the purpose, when actually there are a number of other effects.
I think the college is probably also trying at least a little bit to get students that will be more successful, and thus improve the university's reputation, donate more money, and not cause as many problems while attending.
Plausibly the system also provides deniability for the admissions department to accept various bribes (either direct or indirect), cater to their personal biases, and make it harder to sue them (because if the process is illegible then it's hard to prove any specific bad thing about it).
Ok, all that makes sense. I think of the admissions process as the input to the system, the educational process as the throughput, and graduation as the output. Alumni donations then provide a feedback. A system can have more than one purpose, just as it can have more than one input. Grant money and tuition are additional ones.
Don't make the mistake of reifying the model. Assigning a purpose to a man made system is one way of analyzing it, but a university doesn't really have a simple coherent set of "purposes" that you could put on a comprehensive list.
If we say "A system is what it does" does that make more sense?
Wouldn't what you describe be the purpose of the college admissions system, not the university itself?
I'd suggest just a pure lottery from the qualified individuals. You can't manufacturer signal from noise. All that essays do is import the biases of whoever is reading them, whatever way those point. Which is even more unfair and obscure than honest randomness, as well as way more of a waste of everyone's time. Plus making people cynical about the process, which is inherently corrosive.
Yes, that would make more sense. Which is why I believe that the admissions process is not intended to be fair, neutral, or even find the best applicants.
What it appears to be doing is finding the applicants that the particular school would like to have there, while giving plausible deniability about the reasons any particular student was rejected. Legacy applicants and the children of rich and connected people are pretty obvious. You can also select for demographic or social reasons, or pretty much any reason you want.
"Hardship" essays, I'm pretty sure, were intended to pull in more minority, especially black, applicants. It was expected that more minorities would have hardships than white people, so that would give them a leg up.
They do select for people's willingness to jump through arbitrary hoops.
I suppose. I was shocked how virtually everyone on the admissions essay thread, including Scott, bought into the idea that ya gotta play the game. I really do find it hard to believe that if some kid wrote an intense, sparkling, truthful essay about, say, learning to paint with watercolors -- how it taught them about patience, and excellence, and precision, and using happy accidents -- they would make a bad impression.
Saying "you have to play the game" is just good pragmatic advice. We could pine for a better system, but if we're not in control of admissions at the relevant schools, we can't actually change the rules.
If I were giving advice to a high school student about getting into college, I certainly wouldn't tell them to ignore the whims of the application reviewers, but help them game that system as much as they can.
But, I might also encourage them to actually develop skills and maybe go to a local college where a 1100 SAT score will get you in automatically. There's nothing wrong with working their way up in a field instead of trying to start near the top with a Yale degree or whatever.
Which I'd firmly place in the we shouldn't be doing that bin, personally. Teaching people that it's all arbitrary nonsense and that only by cheating the system can you survive is one of the lessons taught by the Soviet system, and leads to low trust, high tolerance for corruption societies.
Just like leaving broken glass lying around selects for people willing to walk around it, especially if they can't clean it up. Which leads to inefficiency and worse, lacerations.
Oh, agreed. But POSIWID - the knowledge that the first, automatic response of the person bearing the piece of paper your institution produces when you ask them to jump will be "how high?" is valuable to a significant enough proportion of the forces behind the system that you need to plan for considerable resistance if you attempt to change it.
As someone who has done a lot of hiring, it's amazing how cookie-cutter many of the college-educated applicants we get are. Hiring a Custodian means you meet all kinds of average people with various backgrounds. Hiring [college educated position with industry-specific requirements] and it's like you're talking to the same three people over and over again. We can definitely count on them to yell "How high?" when we tell them to jump, though. That's pretty valuable to us, even if we could totally survive without it.
Agreed. There are lots of forces and entrenched interests here. Not the least are all the admissions folks who would lose influence/jobs if they weren't needed to gatekeep, even if the gate shouldn't exist in the first place.
Shuffle the application pile. Take the top nine tenths and drop them in the bin. You don't need unlucky people.
Better yet, get the Dean of Admissions (or whatever his title is) to spend the night fasting and praying for divine guidance, and then randomly pick out one-tenth of the applicants to accept. God wills it!
Reminds me of Larry Niven’s Ringworld. Birthright Lotteries to breed humans for luck.
On college essays I find them similar to many interview processes in that they are arbitrary specifically because there needs to be some way of filtering down candidates. Let's say the acceptance rate of some college is 5% (most top tier colleges are around or below this number). After eliminating all the people that clearly aren't qualified - low SAT scores, bad grades, etc etc - they still have to find some way to cut down the applicants since they can only accept so many. You could argue a lottery system or random chance would be fairer, and in some ways I agree, but I think most applicants would not like it as it takes away what control they have over their chances. So the colleges use essays instead, and while those aren't exactly merit-based I'd still argue they test valuable skills: writing skills, ability to research/know your audience and what they are looking for, etc.
The same is true for interviews/job applications - there are going to be more than enough qualified candidates, and the process of sorting them is always going to be somewhat arbitrary by nature. For a programming job you could sort candidates by how quickly they complete a series of leetcode questions, but that tests speed and memorization - people who do well on that test may struggle when it comes time to develop an original solution to a novel problem.
TLDR; Tests are always somewhat arbitrary and since the things colleges/jobs/etc are looking for cannot be tested for directly, they look for results that are correlated instead
I'd think that the rational way of choosing between equally qualified candidates would be to simply sort by price? That's how I usually decide what to buy or whom to sell to.
Colleges charge everyone the same tuition, so they can't really do that. I suppose you could implement some system where the admin office charges a variable amount of tuition, but I'm not sure how that would function in practice, and it would require more business acumen out of your application-sorting minions. (I think The Name of the Wind had a college that worked like this?)
Price discovery is not exactly frictionless either - if you ask job applicants what salary they expect, they'll probably answer somewhere around the "going rate" for the job and not give you enough variation to discriminate. You'd have to take the time to negotiate them downward individually.
Basically, what happens if you look at the cheapest price available and still have more than one option?
I think the simplest way would be to raise tuition every year until the school reaches maximum profitability, at which point they should have the capacity to enroll all qualified candidates. That's just like any other business that doesn't price discriminate.
I read somewhere — maybe in one of the gcbias.org's blog posts (?) — that any GWAS study done before 2017 is probably suspect. I didn't follow up on this, but I recently read these two posts by Sasha Gusev and Eric Turkheimer...
"How population stratification led to a decade of sensationally false genetic findings" by Sasha Gusev. — https://theinfinitesimal.substack.com/p/how-population-stratification-led
"Is Tan et al. The End of Social Science Genomics? What happens when everything is unconfounded?" — https://ericturkheimer.substack.com/p/is-tan-et-al-the-end-of-social-science
Both state that population stratification was identified as a confounding factor in earlier GWAS studies. In other words, humans don't mate randomly, and our genomes are probably not the result of random mating.
The Gusev and Turkheimer posts were both prompted by the 2024 Tan, et al. paper, "Family-GWAS reveals effects of environment and mating on genetic associations".
https://www.medrxiv.org/content/10.1101/2024.10.01.24314703v1.full.pdf
Educational attainment, once recalculated for population stratification, comes out as 0.50 confounded by non-genomic factors.
Cognitive performance is ~0.65 confounded.
Household income is ~0.70 confounded.
And ADHD is the most confounded at ~0.77!
It seems like the study that kicked this off was the 2017 paper, "Human demographic history impacts genetic risk prediction across diverse populations" by Martin et al. —
https://www.cell.com/ajhg/fulltext/S0002-9297%2817%2930107-6?elsca1=etoc&%3Belsca2=email&%3Belsca3=0002-9297_20170406_100_4_&%3Belsca4=Cell+Pres
I'm perusing it now.
Martin et al. *may* have followed this up with "Significant impacts of population stratification on genetic association studies in diverse populations" in 2019. ChatGPT references it as being published in Nature Genetics — but I can't find this paper in Google Scholar. I suspect it may be an LLM hallucination.
Very interesting topic.
There's one thing I don't understand about Gusev and Gregg Gibson and all those guys advocating for this view, that perhaps somebody else here would know:
My last major genetic mental update was Polderman 2015, the twin study meta-analysis basically showing that everything important is 60-80% heritable in developed countries.
So now GWAS's and family GWAS's and all sorts of more refined and informative takes are coming out that have identified a "heritability gap" between what we can measure genetically and what twin studies tell us, and which argue "nope, that's wrong, environment is way bigger and more important than we thought." And this is Sasha Gusev's and your linked posts.
But what I don't understand is: are there any actionable changes in important life areas stemming from the more refined genetic take?
The two biggest areas people care about individually where this knowledge is used are probably mate search and how we raise our kids.
So for mate search, the "dark side" genetic interpretation would be "genes are all that matter, and this includes more than their immediate genes (per Greg Clark), so optimize direct mate quality AND lineage quality to the extent you can."
And I'm not really seeing any different recommendation from what we know from family GWAS's? It seems to support the exact same conclusion, possibly even more strongly (because family culture seems important and to make up most of the heritability gap)?
And for raising our kids, it's basically "make sure they're raised in a two parent household with good mores and no physical or mental abuse," and that's the tippy-top best you can do. Everything else is noise, in the sense that everything else that matters was handled at the "mate selection" stage. You married into whatever level of educational attainment and average income and wealth you're going to be drawing from, and now it's just random draws from you and your spouse's conjoined lineage baseline. All environmental factors that we know matter are negative - malnutrition or physical or mental abuse all bring your kids down, but we know of nothing that brings them up.
Has the additional info we know from family GWAS's changed this at all? Here I think maybe there's some room if family culture and environment actually matter a lot more than we thought? Have we actually identified anything actionable here? Or is what I described still the best we can do? If that is the case, are these environmental / culture differences being rigorously studied anywhere?
Really appreciate any insight anyone might have here.
> My last major genetic mental update was Polderman 2015, the twin study meta-analysis basically showing that everything important is 60-80% heritable in developed countries.
You make a perfectly valid point that this was the consensus view back in 2015. But if I understand Tan et al. correctly the twin studies give you a false sense of heritability. You've got a bunch of IGEs and other kruft that masquerades as DGEs. See my response to Eremalalos below. I wonder if Polderman will update his priors and provide us with a new meta? I somehow doubt it, because Bayesians tend to hug their priors close to their hearts. Well, that tends to be true for everyone, including myself, so I shouldn't be snarky.
> You married into whatever level of educational attainment and average income and wealth you're going to be drawing from, and now it's just random draws from you and your spouse's conjoined lineage baseline
Yup.
> ...are these environmental / culture differences being rigorously studied anywhere?
Yes, but I admit a lot of these studies have a squishy feel to them. But if I were raising a kid, there's a shitload of environmental stuff I'd do to improve the life outcomes. For instance, get them started on a foreign language while they're still very young (starting Kindergarten if you can). And if you're from an Indo-European language group, send them to a Chinese academy on weekends (I've become a neo-Whorfian, and I believe that languages do shape the way we think). Force them to learn a musical instrument. Play word games with them. Solve puzzles with them. Books! Of course, everything is online now, but its easy to get distracted by other stuff when you're online. Books are a focused media. My folks bought me just about all the Time-Life series when I was a kid, and I devoured them. And after reading Heinlein's _Beyond This Horizon_, when I was twelve, I wanted to grow up to be an Encyclopedic Synthesist.
> But if I understand Tan et al. correctly the twin studies give you a false sense of heritability. You've got a bunch of IGEs and other kruft that masquerades as DGEs.
Yeah, I follow their arguments overall, I think. Basically, population stratification means because we tend to study WEIRD people for everything, we're in a region where everyone is drawing from a more similar baseline than true world diversity, and so overestimate polygenic scores for various endpoints.
And then because we all assortatively mate really strongly and this has been going on for a long time, we're drawing spouses who are similar enough on multiple traits those traits have become correlated with each other over time, so you're getting more than the straight ~50% expected allele draw due to linkage, and this skews heritability estimates when looking at siblings and dz twins. Additionally, because we're similar culturally as well as on correlated endpoints, high education people impart high education cultures to their kids, etc.
So the answer is family GWAS's - because this controls for the same grandparents, all the background stuff like population stratification and assortative mating is filtered out, and you get "true" polygenic scores for your endpoints, which are much lower than the Polderman dz / mz scores.
But to my mind, it literally doesn't change anything, and assuming full genetic determinism still gets you the exact same answers.
I appreciate and agree with all your early childhood interventions - but to my mind, it's nothing new at all. That's just raising your kids in default educated Westerner culture.
This is why I specifically framed it in terms of "where would this ground out in decisions or actions," because I'm kind of skeptical.
Heritability gaps seem to keep getting smaller with larger and more diverse GWAS's, and most of the people making these arguments seem very entrenched in a certain mindset that seems ideologically pre-committed to an "anti genetic determinism" worldview, so I just can't tell how much is noise / slanted towards their desired outcome versus useful knowledge that can inform our actions and decisions, either individually or societally.
For instance, if culture / environment REALLY matters ~4x more than we think from Polderman, why didn't it come out in any of the much older separated at birth Minnesota twin studies, or the many studies where disadvantaged kids are raised by middle / upper class people in much better environments?
It's not like we haven't thought about or controlled for "culture / environment" before - and I've asked this before and didn't get a good answer.
I have the sneaking fear / suspicion that all of this is quibbling about how many angels can dance on the heads of pins, but as long as you're not going to abuse your kids or make them eat lead paint chips, if you still think and act like everything is ~80% genetics and it's all determined, you're going to still make all the right decisions.
Best thing you can do is have more kids. More rolls of the dice, so to speak.
100% - planning on 6.
Are you looking forward to participating in the nappy/diaper changings and late-night feedings? I doubt if Elon has done this with his twelve known children (and now he's upset that his eldest male child transitioned). If you're not going to help out with the child-rearing, I sure hope you've got a lot of surplus income to divert to child support!
Weird anecdote: I have a friend who has ten half-brothers and sisters (that she knows of). They were all from the same town, and they all went to the same high school. They eldests started keeping an informal genealogy so the younger half-siblings didn't accidentally end up dating and having sex with each other. Their father probably sowed his seed in other places, but they could only compare notes locally.
> If you're not going to help out with the child-rearing, I sure hope you've got a lot of surplus income to divert to child support!
Literally my entire pitch revolves around "maids, nannies, cooks, and surrogates," because nobody worth marrying wants to wreck their perfect body with 6 pregnancies or spend all their time on drudge work.
It's like a night and day difference, completely 180's people. Before these 4 magic words, it’s all:
Her: “WHAT?? 6 kids?? Jesus, you’re crazy! No way! Hang on, wait, wait, I’ve got to fish my taser out of my purse….<rummaging>”
Me: “Cooks, maids, nannies, surrogates.”
Her: “What? What did you say?” <pulls out taser, smiles crazily as she pops it on and off a few times>
Me: “Cooks, maids, nannies, surrogates. I don’t want to ruin your body any more than you do! Kids are a big thing, everyone needs some help! Screw cooking and cleaning! Let’s do it right!”
Her: “Oh. Hmm.”
<thinks a little, talk a little, eventually…>
Her: “Holy shit, you’re right! Ok, let’s go. Let’s do it. C’mon, what are you waiting for?? Marry me already!”
Any applications towards our national-level fertility crises seems sadly inapplicable, because these aren't solutions that necessarily scale well.
But to your anecdote, I actually grew up with 9 siblings, and multiple aunts and uncles all had 6-8 kids, and I doubt I'd be shooting so high if I didn't have a childhood experience like that myself.
beowulf what do you mean in the second half of your posts by when you say something is confounded by a certain amount? I'm not arguing, just don't understand. So if, for ex., ADHD is confounded at 0.77, do you mean that some measure of population stratification correlates at .77 with an ADHD measure? Also, what measure is used for the genomic factor, in studies that made no attempt to correct for population stratification? Is it also a correlation coefficient?
Yeah, I didn't explain that very well, did I? Basically, that means only 23% of ADHD expression can be tied to genes. Environmental factors (for instance, chemical, social, or perhaps diagnostic) swamp the genetic signal. The .77 would be a negative correlation coefficient. ;-) And I should have written: "Educational attainment, once recalculated for *Direct Genetic Effects*, comes out as 0.50 confounded by non-genomic factors. Population stratification by assortative mating may only be one of the confounding factors. My bad. I was typing faster than I was thinking.
Tan et al. actually don't use the GWAS data. They tie their estimates to the Direct Genetic Effects. They explain the methodology in this paragraph. I had to read it over a few times before it began to sink in.
> Family-based GWAS (FGWAS) has been proposed as a solution to the problem of confounding in GWAS that also enables deeper investigation of the impact of gene-environment correlation — including from IGEs — and non-random mating, including population structure and assortative mating. FGWAS adds the parents’ genotypes to the regressions performed in GWAS (Methods), thereby using the natural experiment of random assignment of genotype within-family — due to Mendelian segregation during meiosis — to estimate ‘direct genetic effects’ (DGEs). *Because the segregation of chromosomes during meiosis is independent of environment, estimates of DGEs are free from confounding due to gene-environment correlation, including from IGEs and population-stratification. Because different chromosomes segregate independently during meiosis, DGEs are also free from the confounding that arises in GWAS due to correlations with genetic variants on other chromosomes, which are caused by non-random mating* including population structure and assortative mating (AM). As in GWAS, variants are analyzed one at a time, so DGEs include causal effects of the focal variant and correlated variants on the same chromosome.
And an IGE would be an Indirect Genetic Effect. ChatGPT gave me a good example: "Example: A parent has gene variants associated with high educational attainment → they create a cognitively stimulating home → that environment influences your development (regardless of whether you inherited those same genes)."
Thanks for clarifying. And while I have you: Here's a toy my cats love and never get sick of. Maybe would enliven life indoors for Warrior Princess?
https://www.amazon.com/AsFrost-Upgraded-Interactive-Treadmill-Stimulation/dp/B0DHZZSPMB?ref_=ast_sto_dp
Here’s Popo playing with it.
https://tinyurl.com/yej3ymke
Princess went AWOL yesterday. She's around my neighborhood, but hides under cars when I get too close. And she won't come out for her favorite treats — she got wise to that trick.
Ooh, that’s a real problem. Maybe use a trap with food in it? There’s a kind that organizations that trap, neuter and release use. They might even lend you one.
This is one of those areas where I wish I was an insider who knew all the "behind the scenes" stuff. I've seen some comments online from geneticists saying that a lot of this is "cope" (grasping at straws) that's trying to avoid a genetic determinist view of the world, and Turkheimer feels somewhat guilty for earlier work (including some of his own) suggesting that genetics are a major determinant of life outcomes.
But who knows, are those internet commenters the ones who are high on copium or is it these behavioral geneticists? In my own field (non-biology STEM) I've seen similar grasping-at-straws attempts, sometimes by senior researchers and sometimes by young upstarts, to try to galaxy-brain their way out of an inconvenient or unfortunate finding, but you have to go to conferences, chat with people at the coffee table, etc., to have a feel for the "vibe" of this kind of thing. As in, is this really just cope by an aging figurehead trying to keep their ideas relevant? Or are the problems they're highlighting actually a serious problem in the field?
An example of the latter might be John Ioannidis' work in the early days -- many people probably wrote it off as cope but he was right: there were major problems with medical research that invalidated many of the findings people took to be true.
IIRC, John Ioannidis was mostly concerned with sloppy or fraudulent science like p-hacking and sample biases (Please correct me if I'm wrong).
There was nothing sloppy or fraudulent (AFAIK) about the pre-2017 GWAS studies. They just ran those studies under the false assumption that European allele frequencies between individuals — with and without a trait or disease — matched the profiles of other non-Euro populations. But the Martin et al. paper that I linked to above showed that assumption to be false. ChatGPT insists that they followed up with a paper soon after that showed that population stratification (i.e. people preferentially mating within similar groups) was the real culprit. But I can't find the damn thing! ChatGPT, CoPilot, and Grok all agree that Martin's team was the first to demonstrate that population stratification as a major confounder. But Grok says this idea goes back to Lander and Schork in 1994 (!) — it was just that no one had bothered to investigate their claims.
As for the cope-side of the argument, I'd like to see some serious studies by geneticists that argue against the above papers. If you or someone else can send me links for them, I'd appreciate it (although I admit I'm fairly convinced that Tan is correct). Probably the Tan et al. paper is too new to have garnered any responses yet. But it seems to me that most of the copium comes from social media commentators who have latched on to the whole genetic determinism view hook, line, and sinker. Certainly, if you had read the popular summaries of the pre-2017 literature, that would be a fairly reasonable view.
Ioannidis was/is also concerned with bad research methods (which do overlap to some extent with p-hacking and the "garden of forking paths") - he famously called for the entire field of nutritional epidemiology to be more or less thrown out (https://jamanetwork.com/journals/jama/article-abstract/2698337) because its reliance on various surveys, food inventories, etc., produced implausible effects that often did not hold up in RCTs.
Like I said, I'm not a genetics expert so I'd love to hear from an insider who has the nose to sniff out the vibes / inside scoop on the Tan issue. Agreed re: propensity for copium from social media commentators who have a vested interest in genetic determinism.
Can I ask for a "for dummies" tl;dr? For example, on ADHD, what was stated to be true and is now considered false when taking population stratification into account?
I screwed up while writing that post. ADHD recalculated for Direct Genetic Effects comes out as 0.77, confounded by non-genetic factors. Population stratification (by assortative mating) may only be one of the confounding factors. Environmental factors or diagnostic issues may also account for a weak genetic signal. My bad. I was typing faster than I was thinking.
If you're specifically interested in ADHD, there's a great article in the NYTimes. I think this link will get you past the paywall.
https://www.nytimes.com/2025/04/13/magazine/adhd-medication-treatment-research.html?unlocked_article_code=1._U4.jVHA.PjoQgLI0KYWt&smid=nytcore-android-share
It glosses over the possible genetic component for ADHD, but seems to support the Tan et al. thesis that ADHD is not very heritable...
>...for a significant percentage of people diagnosed with A.D.H.D., Nigg says, “there’s nothing neurobiologically notable about them. Instead, their symptoms are situational or conditional. They may have had a hard life, or they have a lack of social support, or they’re in the wrong niche in life.”
I have a suggestion for trying to deescalate lawfare:
For any indictment of a national political figure, require that a simple majority of Democratic Party congresspeople _and_ a simple majority of Republican Party congresspeople agree to the prosecution before it moves forward. If there is a possibility of retribution, do the voting on this with secret ballots.
If a political figure did something _truly_ egregious, and the legal process _isn't_ being used just as a partisan weapon, those majorities shouldn't be too hard to get.
Yes, I know this would take a constitutional amendment. Yes, I know the founders were not fond of political parties, and intentionally excluded them from the Constitution. Consider it a policy proposal, with enactment issues deferred.
Any, yes, "national political figure" is a fuzzy term, and would need a precise definition.
And this isn't intended to cover _all_ lawfare, but maybe enough to make at least the pretense of even-handedness sort-of credible. It would be nice to have the legal process look like something other than a blatantly partisan weapon.
What do people think? For anyone who doesn't like this, how would you try to restore some legitimacy to the legal process?
edit:
Many Thanks to Erica Rall and TotallyHuman for their helpful alternatives
https://www.astralcodexten.com/p/open-thread-378/comment/110935744
and
https://www.astralcodexten.com/p/open-thread-378/comment/110886281
respectively.
The legal process is not illegitimate, but it only works if you're willing to actually prosecute people for their crimes. The next Democrat President should, on day one, launch criminal investigations into the DoJ and Department of State and indict the respective Cabinet members of Trump's administration responsible for deporting people without due process. They should revoke Fox News's broadcast license for peddling 2020 election fraud lies and for not immediately denouncing Trump signing orders weaponizing the DoJ against the law company that won Dominion their defamation suit against Fox News. The reason Republicans lie with impunity to the extent they live in their own alternate reality is because Democrats are feckless and aren't willing to punish them - despite Republicans literally lying about them doing just that, they're still too stupid to fight back.
Many Thanks for your comment! So you want to double down on partisan political prosecutions. And you want to unconstitutionally violate freedom of the press. Between people like you, and what Trump is doing along the same lines, damned few people are going to see the legal process as anything but a partisan political weapon. You are part of the problem.
There is mountains of evidence of criminality that Republicans are engaging in. They need to at a minimum be held accountable for it and appropriately jailed. Then there should be actual government coercion against bad faith Republican institutions that are supporting Trump and his admin in criminally undermining American democracy, elections, checks and balances, and Constitutional rights like due process.
Any attempt by Democrats to censor Republicans on partisan grounds should be opposed - only actions that Republicans are taking to destroy America like ignoring due process, lying in favor of ignoring due process, lying about elections being stolen and aiding those who lie, and other similar extreme behavior should be prosecuted.
This is the moral thing to do because Republicans have only gotten more unhinged as they've ALREADY done these things for partisan political purposes like falsifying electoral votes to overturn our system of governance and trying to impeach judges because they are opposing Trump trying to run the government like a dictatorship when there is no punishment for them. They've also been lying that Democrats have already been doing this, so Democrats are taking the PR hit while not actually getting anything done. The messaging here is also very straightforward and actionable - "Stop lying and begin living in reality again and we won't target you for trying to destroy our shared system of governance. Hating trans people does nothing to the country but any obvious and intentional lie about election fraud to try to overturn election results will get you jailed and anyone working with you." Anything less than this will just inspire Republicans to keep destroying the country.
>For anyone who doesn't like this, how would you try to restore some legitimacy to the legal process?
I think it already has plenty of legitimacy, and all you have to do is make the court transcript public. It only looks bad because nobody bothers to look into the specifics, which is not going to be fixed by anything, ever.
Many Thanks! After the Stormy Daniels farce, I completely disagree.
> If a political figure did something _truly_ egregious, and the legal process _isn't_ being used just as a partisan weapon, those majorities shouldn't be too hard to get.
Republicans can't even get a majority together to stop Trump from crashing the economy for no reason, something with very obvious immediate impacts on their constituents. Trump has centralized power to an unprecedented degree, and no Rebpulican in congress dares to defy him.
Many Thanks! The 3rd sentence in my original comment was:
>If there is a possibility of retribution, do the voting on this with secret ballots.
which already addressed your
>and no Rebpulican in congress dares to defy him.
Nice idea but it collapses when you think about what the adversarial equilibrium would be: either party could block all prosecution of their members no matter what they did. There's a reason we have a separation of powers and it's to prevent things exactly like this.
IMO the right structural solution to our current political polarization is the one the Founders already put in and we've managed to evade: only the elite should vote, for some reasonable definition of elite. If it was up to me I'd limit voting to those who were above the 60th percentile in Federal income tax paid. We were founded as a republic and for good reason. The Founders were well aware that when you give the hoi polloi a voice then chaos inevitably ensues. The Simpsons said it best: Democracy doesn't work. Democracy is the voice of the people and most people are stupid. It's insane to let them influence anything.
> If it was up to me I'd limit voting to those who were above the 60th percentile in Federal income tax paid.
Just to be clear, you don't even hit "break even" as a net contributor vs consumer of taxes until you're roughly top 20% / $85k in income, and the top 1% pays 40% of taxes, so you're proposing something like "only the top ~3-5% by income get to vote?"
And is eliminating old people as a voting block, because they're not paying taxes in a given year, a bug or a feature?
Either way, with some amount of job automation looming on the horizon, probably a bad idea, because if only the top 3-5% can vote, capital owners can automate all the jobs without UBI-ing people, and I don't think the killbots are quite at the "retreat behind our walled compounds with impunity" levels yet, so it'd get pretty messy.
No representation without taxation!
Cute! LOL!
>you don't even hit "break even" as a net contributor vs consumer of taxes until you're roughly top 20% / $85k in income,
Oh is that right? I didn't know that. I just pulled that 60% figure out of the air, though keep in mind that some large % of people pay zero taxes. When I said top 60% I meant the top 60% *among those who pay nonzero tax* so maybe that winds up being the same thing. Obviously you'd have to do a careful analysis if you wanted to actually implement it, but any steps in this direction would be positive in my view. Obviously there's zero chance this would ever happen so it's just an academic argument.
>And is eliminating old people as a voting block, because they're not paying taxes in a given year, a bug or a feature?
Probably a feature. Plenty of retired people have investment income that they live off of and social security payments are taxed so not all old people would be disenfranchised. You could also have the threshold be related to lifetime aggregate taxes paid or something.
Many Thanks! I'm somewhat hopeful that the equilibrium could be: Clear, egregious crimes get prosecuted, politically motivated prosecutions get stopped. But this is just a hope, not an expectation. Regrettably, in the current situation, a sane response to "<politician> broke the law." is generally "Yup, and the sky is blue.". It would be nice to pare the legal code back to something sane, but if that could be made to happen at all it would probably take a century.
>IMO the right structural solution to our current political polarization is the one the Founders already put in and we've managed to evade: only the elite should vote, for some reasonable definition of elite.
Hmm - I see that as somewhat orthogonal to this particular problem. What do you think of WoolyAl's comment https://www.astralcodexten.com/p/open-thread-378/comment/110911815 which has something of the same flavor?
I think that comment is directionally correct but goes too far. It's not that 98% of Americans are unprincipled buffoons, it's that 51% of them are and that's all it takes. The central issue is that there is a maximum gap that society can tolerate between how clever your propagandists are and how clever the average voter is. Once that maximum is exceeded then there becomes a voting block that candidates can reliably manipulate like a cat with a laser pointer. Elite-only voting reduces the population of low-IQ voters to a manageable size.
Propaganda only works on stupid people. Those people shouldn't be allowed to vote, otherwise politics becomes nothing but a propaganda contest. I disagree with those who claim that it's always a propaganda contest. There are lots of smart, educated, high-character people out there with good judgement. Our political system should be designed such that those people have an outsized influence. An elitist voting system would increase the signal-to-noise ratio and restore sanity back to governance.
Sadly it's now essentially impossible for this to happen. There are too many congressmen who only have their seat because of dumb voters. They will never vote for a system which would make them unelectable. This is how entropy always wins in the end. We dropped our guard for a couple decades and now we're overrun by morons like AOC and Taylor-Greene. It's impossible to put Humpty Dumpty back together again. The American Dream is dead and in 30 years (maybe sooner) we will be essentially indistinguishable from Brazil or Argentina.
I don't think that propaganda only works on stupid people. But that's because I have a long-running hypothesis that most people are smart; they're just not smart all the time. Namely, they're smart about a few things (often whatever they do for a living), and are anywhere from slightly to much less smart about other things, depending on mental distance and possibly their own brain chemistry.
Propaganda works to the extent it does because it's operating in one topic that most people aren't experts in. This is often something political like foreign policy, economics, or legal philosophy, but one could just as conceivably manufacture propaganda about anything, from food really being a drug forced on us by rich "agricultural" conglomerates to E not really being mc^2. (I know a current PhD in atmospheric chemistry who believed 1 + 1 isn't always 2.)
>Namely, they're smart about a few things (often whatever they do for a living), and are anywhere from slightly to much less smart about other things, depending on mental distance and possibly their own brain chemistry.
Could be. What you describe sounds very much like Robin Hanson's "near mode" vs "far mode" distinction.
Many Thanks!
>Once that maximum is exceeded then there becomes a voting block that candidates can reliably manipulate like a cat with a laser pointer.
Love that image!
>Propaganda only works on stupid people.
Regrettably, _busy_ people also have a problem finding the time to drill down to primary sources. And, even then, have to play the guessing game of "Who is lying more this time?"
Busy non-stupid people could outsource their decision-making e.g. to their trusted bloggers.
At first sight that doesn't seem fundamentally different from stupid people outsourcing their decision-making to TV, but I think the quantitative differences would make a big change when taken together -- it is cheaper to start a blog than to start your own TV, comments on blogs are more visible, it is easy to write a reply elsewhere and link to the original blog, etc.
Many Thanks! Mostly agreed. We _have_ to delegate to experts in many fields routinely. Regrettably, politics, particularly in the current USA tribal form, is a subject where it is particularly difficult to determine who to trust.
Keep in mind that the levels of lying would likely go down if there was less of an audience for it. Plus how much effort does it take to realize that Trump and Kamala are both terrible people? No system is perfect but I have no doubt that this would immediately fix 80% of our political problems.
Many Thanks!
>Plus how much effort does it take to realize that Trump and Kamala are both terrible people?
I suspect that probably even a majority of the non-elites realized that - but 3rd parties are _really_ hard to get elected, so knowing that we were choosing between cyanide and strychnine didn't help much.
So the solution is for politicians to be even more above the law than they already are? You can do any illegal thing you want as long as you stay popular with your team? Frankly, I can't think of a faster way to sink confidence in the political process (not that it's high right now to begin with).
Many Thanks! I think that it is better than having the farce of e.g. the Stormy Daniels case. If a national political figure did something egregious, yeah, their own team should be willing to call them on it. If their own team _isn't_ willing to call them on it - Maybe the prosecution is more political bullshit? Using the legal system as a partisan baseball bat to bash someone with? Again?
The trouble I see is the assumption that political figures want to do the morally correct thing but are sometimes hampered by politics. I think, for political figures, it is often the other way around.
So a solution needs to somehow compel morality in its rules, which are difficult to impossible to skirt. Things like secret ballots create a situation the opposite of that.
It's still good to try to think of good solutions, but it isn't an easy thing to think of something better than the current system, bad as it is.
Many Thanks!
>The trouble I see is the assumption that political figures want to do the morally correct thing but are sometimes hampered by politics.
I'm not making that assumption. I'm not making any claims about morality at all.
I'm just looking for a way to cut down on the lawfare, for the parties to be less able to use the legal process as a baseball bat to bash each other over the head with - particularly in farcical cases, like the Stormy Daniels one.
Currently, there is no way to put a stop to partisan bullshit lawfare and to move the conflict between the political factions back to the realm of political negotiations, where it belongs.
You have referred to the Stormy Daniels case as farcical several times. Could you expand on why you feel this is the case?
Just plain looking at the charges, the laws, and the facts found by the court supports the convictions. The sentence appears to be just plain correct. Donald Trump committed the crime, and got found guilty, as appropriate, in a court of law by a jury of his peers.
Many Thanks! I explained in https://www.astralcodexten.com/p/open-thread-378/comment/110958769
I replied to this at more length elsewhere, but I'll just say again that I find it *utterly baffling* that you take it as a given that one of the richest and most powerful men in the country should just be able to violate the law when it suits him. Why, exactly, is it "political bullshit" for him to have to follow the same laws as everyone else?
Many Thanks!
>be able to violate the law _when it suits him._
[emphasis added]
Nope. _When the prosecutor cannot persuade a simple majority of both parties to let the prosecution proceed._
My _guess_ , in Trump's case, is that this would have quashed the Stormy Daniels farce but let the classified documents case go forward. But that's just my opinion. I could be wrong.
Re:
>have to follow the same laws as everyone else
a) blank is right. Somewhat similar (albeit with some differences) cases were not pursued in Hillary Clinton's (email server issue) and Biden's (classified documents - but with extenuating circumstances) cases
b) More generally: There are so damned many laws, with such broad wording and the potential to be twisted in so many ways, that nearly anyone is in violation of _something_ .
"Nope. _When the prosecutor cannot persuade a simple majority of both parties to let the prosecution proceed._"
Right. So whenever it suits him. Anyone with who hasn't slept through the past 8 years has noticed that Republicans as a unified have never and will never go against Trump on *anything.* He literally incited a goddamn insurrection against the country he's leading, and Republicans do nothing but make excuses. Why on Earth should anybody who isn't a Republican trust them to be the *slightest bit impartial* when it comes to evaluating the misdeeds of the head of their cult of personality? Proposing turning the law into a political popularity contest is ABSURD.
"blank is right."
Blank is full of it and you should have the basic honesty and critical thinking skills to recognize this. First and foremost, the Stormy Daniels case was *nothing whatsoever* like the various classified documents cases. I see you've already been called out on this below, so I'm not going to get too far into the weeds but a big, clear, *obvious* difference is *intent.* It's very difficult to prove somebody's intent when they're merely doing something careless or reckless. But when they make deliberate changes to their own records to hide wrongdoing, intent is plain as day.
But more to the point Clinton especially was probed and investigated and called into hearings *repeatedly*. For the classified documents and even moreso for Bengazi. For your claim to have merit, it would need to be true that Clinton had clearly, demonstrably broken the laws, and that the people in charge of enforcing them had knowingly and deliberately let her off the hook for doing so. That's the exact opposite of what appeared to happen: her political opponents were *extraordinarily determined* to find her guilty of something, and simply failed to do so[1]. Claiming that there's a double standard because the state meets its burden of proof in some cases and fails to meet it in other cases is absurd, man. Please do better.
[1] Though they arguably did enough damage just in the attempt to sink her presidential campaign.
> It's very difficult to prove somebody's intent when they're merely doing something careless or reckless. But when they make deliberate changes to their own records to hide wrongdoing, intent is plain as day.
I presume you'r referring to the time when Clinton's furry tech admin asked Reddit how to scrub names and timestamps from her Benghazi e-mails?
https://thehill.com/policy/cybersecurity/297106-oversight-orders-reddit-to-preserve-deleted-posts-in-clinton-probe/
Many Thanks!
>Right. So whenever it suits him.
No, as I already explained. Factions will be factions. If you _can't_ persuade a simple majority of a faction's congresspeople that a prosecution has merit, I'm taking that as sufficient evidence that this is a politically motivated prosecution, and should not go forward. Maybe the opposite faction managed to dig up a statute that the defendant can be construed as having violated. So what? If you can't persuade a simple majority of the defendant's party that the case is worth pursuing, all that you have proved is that we have too many laws.
The charges were not similar at all!
The email server issue is potentially much more serious (and almost identical to the Hegseth on Signal case), but it is much harder to prosecute.
The Stormy Daniels case is a cookie cutter falsification of business records case. If none of the people involved were celebrities, it would just be something that gets successfully prosecuted pretty much monthly in New York. Anyone else who did what he did would also get prosecuted. The only difference is that because he had won an election before the prosecution finished, he didn't end up going to prison, like anyone else would have.
Many Thanks! Re Stormy Daniels, see https://www.astralcodexten.com/p/open-thread-378/comment/110958769
Re the email server - Many Thanks!
The laws are unevenly applied. Hillary and Biden were cleared of any similar charges. Banana republic!
Politicians being above the law means that they are actually in charge, rather than being subservient to lawyers.
This weekend I was standing in the rain, holding a sign that read "LEADERS not RULERS." Your proposal does not resonate with me.
Banned for this comment.
Less of this please.
Can you point to me anywhere in the founding documents of the U.S.--anywhere at all--that says openly or implicitly that holders of political office ought to be above the law? Call me crazy, but I seem to remember that anger and dissatisfaction over having a powerful, unaccountable head of government was one of the *motivating reasons* for the American Revolution, and that the framers of the constitution were *quite clear* that they wanted to head off that failure mode for their new republic.
In this case, I think the founders of the US were full of shit, because their supposed grievances at the King of England for being above the law made little sense because everything he did was also agreed on by Parliament.
I also think the American idea of subordinating public office to the law made more sense when the average voter was educated in the law and owned property.
"I also think the American idea of subordinating public office to the law made more sense when the average voter was educated in the law and owned property."
I think this is one of those appeals to a non-existent history. Even at the founding of the nation property ownership was a prerequisite in some states and not in others: and in the former it doesn't seem to have lasted very long. It was clearly not considered important or default by the framers. As for the average voter being "educated in the law," this seems...fanciful at best. Given broad trends around urbanization, I'd wager good money that the "average" (in the sense of "modal") U.S. voter in first few decades of its existence was a farmer. Heck, I'll bet this still holds true even among the states with property requirements. Unless your standard of "educated in the law" is low enough that anyone who's taken high-school law or civics classes will meet it, I'd be skeptical that your typical frontier farmer did either.
The New England states had a lot of educated farmers, though.
>I'd wager good money that the "average" (in the sense of "modal") U.S. voter in first few decades of its existence was a farmer.
Yes.
<aiWarning>
Google responds to: when did the fraction of the us population that were farmers drop below 50%
with
AI Overview
The fraction of the U.S. population that were farmers dropped below 50% between 1860 and 1910. In 1860, 53% of the labor force was in agriculture, while by 1910, that number had fallen to 31%
</aiWarning>
Yeah, the British monarchy, whilst not yet quite a figurehead, was very much playing second fiddle to Parliament by this time. George Washington had more power as President than George III ever did as King. (And of course, Donald Trump has far more power than either of them.)
Cutting off malicious prosecutions relatively early is supposed to be what Grand Juries are for, and it is a known problem that Grand Juries currently aren't doing a great job of this in general, not just for major political figures. I would prefer a solution based around fixing the more general problem here.
Off the top of my head, possibilities for this include raising the standard of evidence for indictment from "probable cause" to "clear and convincing evidence", giving the jury a mandate to decline indictments that are not in the interests of justice, requiring a (larger?) supermajority verdict, or including a public defender in the process to call exculpatory evidence to the jurors' attention.
Of these, I think a supermajority requirement would be most effective at limiting lawfare. Grand Juries usually have little or no voir dire and are typically larger than trial juries, so are likely to be close to representative of the electorate in the jurisdiction in question. A nakedly partisan malicious prosecution would then likely need to persuade at least a handful of jurors of the same party of the defendant to meet the supermajority requirement.
Many Thanks! Yes, those all sound like reasonable possibilities.
>A nakedly partisan malicious prosecution would then likely need to persuade at least a handful of jurors of the same party of the defendant to meet the supermajority requirement.
Maybe include an _explicit_ requirement that the Grand Jury have enough jurors of the same party as the defendant to ensure that the supermajority cannot be met without convincing at least some of those same-party jurors?
Woo hoo, I'm going to get a grand jury with a large chunk of Libertarians who will vote against everything!
Many Thanks! It would help to counterbalance the metastasis of the laws...
I think a precondition for deescalating lawfare is for people to actually want to deescalate lawfare. People love lawfare, it feels awesome. At least half of the Trump era has been devoted to liberals demanding and devouring lawfare against Trump. Nothing in news has been nearly as profitable as lawfare against Trump and several people have launched their careers (kinda) off prosecuting Trump for whatever thing didn't pan out in court except that one thing with the pornstar which no one understands but we all got to spend a couple months pretending we we're talking about Trump f***ing a porn star with secret lawyer payoffs for the good of the country, rather than our own purient interests.
And, to be fair (1), the right wing absolutely indulged in the same stuff with Hillary and delights in "liberal tears" with all the depravity you would expect. But, just observably, like 10 people are actually sincerely concerned about lawfare degrading our legal norms and the others 329,999,990 people in America are like "This is my favorite TV show! Moar please. My own social, career, and romantic failings are incredibly depressing and the only thing that gives my life any meaning in our soulless nihilistic consumer culture is to artificially divide ourselves in half and then try to murder the other half. Hail Khorne and pile skulls on the skull throne just as long as I can feel something again."
And as long as liberals can convince themselves that once they finally murder Tiberius Gracchi, sorry, Donald Trump, then all of this will be over and their outgroup will be ground under their heel forever. Definitely.
Meanwhile, conservative/reactionaries here are like "Once we install a supreme absolute dictator who will brutally murder everyone we dislike, we can return America to the 1950s. It definitely worked for Sulla and it will definitely work for us."
But yeah, if we had a public vote on "return to peaceful democratic/peaceful norms" or "punish outgroup so we can luxuriate in their tears", punish outgroup is winning 98%-lizardman constant.
So, practically, are you aware that Puerto Vallarta is a gorgeous tropical paradise in Mexico with nonstop flights to most major US metro areas so you can visit family and friends? (2)
(1) https://www.youtube.com/watch?v=jv7jcciKB_s
(2) https://www.flightconnections.com/flights-from-puerto-vallarta-pvr
> So, practically, are you aware that Puerto Vallarta is a gorgeous tropical paradise in Mexico with nonstop flights to most major US metro areas so you can visit family and friends? (2)
You sound like a man after my own heart - I've been out of the country for the last 2 years for exactly the reasons you suggest.
The Republic may fall or not, but frankly, at least I won't have to be bombarded on all sides about it here on the other side of the world.
I read a theory once that America was such a litigious country because it had lost most of the informal norms that had once held society together, leaving just the external compulsion of the law. I think it was in the context of ambulance-chasing lawyers, but your comment about people not wanting to de-escalate lawfare made me think of it.
Many Thanks! I was under the impression that at least the cancerous expansion of federal statutes was more recent than that, more of a phenomenon of the last century or so, rather than going all the way back to 1776 or 1787. But you may be right about overall litigiousness going back that far.
Oww, OWW, _OWW_ , _*OWW*_ . You may be right. If so, we will continue with
>At this point, the legal system is clearly a partisan weapon, with all the legitimacy of beating someone over the head with a baseball bat.
and we will continue to fulfill
H.L.Mencken:
>Democracy is the theory that the common people know what they want, and deserve to get it good and hard.
Hmm...
Interestingly, TotallyHuman and TK-421 are effectively making diametrically opposed predictions of what would happen if my proposal were put in place:
TotallyHuman:
>Result: No political figure is ever convicted of a crime, unless the public is so convinced they did it that it's becoming a problem for their party.
TK-421:
>Both sides would love to force their opponents to vote to prevent a prosecution - the attack ads would write themselves. No, a secret ballot doesn't help because then you just tar the entire other side of the aisle as obstructionist while your side publicly declares that you voted for prosecution, as all upright persons would do.
( Which I'm interpreting as a prediction that neither side would ever dare stop a prosecution. )
My hope is, naturally, that prosecutions for clear, egregious crimes would go forward with bipartisan support, while politically motivated bullshit prosecutions would be blocked. ( Of course, any decision process has some noise, and some error rate. )
Presumably, 2 out of these 3 possibilities are wrong, but we can't know which 2 at this point.
I think TK-421 is correct about the way the political theatre would play out -- I just don't expect it to substantially affect the outcome. Everyone would claim that everyone else is blocking prosecutions, but they would still block them. After all, if you don't block the prosecution, the attack ads about the criminal in your party *also* write themselves.
Many Thanks for the clarification!
> ( Which I'm interpreting as a prediction that neither side would ever dare stop a prosecution. )
Your interpretation is incorrect. I agree with TotallyHuman.
Let's consider some hypothetical prosecution of person in Party A. Barring something wildly exceptional - Party A's guy shot somebody on the Senate floor, for example - Party A is going to vote to block and call it politically motivated bullshit. Party B then gets to crow about how Party A is cravenly using rules intended to stop lawfare to protect a heinously guilty criminal. Prosecutions are blocked from going forward but the incentive is for Party B's prosecutors to bring cases against Party A's people to force the vote, and vice versa.
Sure, I guess you've solved lawfare in that you've created a group of people explicitly above both federal and state law - although what happens when they leave office? Congrats on setting up the same failures as the Roman Republic - but your plan is almost perfectly designed to inflame partisan passions.
Many Thanks for the clarification! So your expectation is that the prosecutions will nearly always be blocked by the party of the defendant. Could be. I'm seeking to deescalate the lawfare (from the situation WoolyAl so eloquently described in https://www.astralcodexten.com/p/open-thread-378/comment/110911815 ). I would rather see politicians negotiate the differences between their factions than use the courts as weapons against each other (except for the most clear and egregious cases, where there could be bipartisan support for a prosecution).
Result: No political figure is ever convicted of a crime, unless the public is so convinced they did it that it's becoming a problem for their party. This goes double for crimes which benefit their party. (I would argue that if half your party thinks they benefit from your being prosecuted for a crime, they'd probably be throwing you under the bus in the current system anyway.)
Also, from a aesthetic standpoint, I think politicians should be held to a higher (or equal) standard to private citizens, not a lower one. An explicit statement that the leaders of a nation are not equal to their subjects under the law is profoundly ugly to me.
I'd divide lawfare into three broad categories -- and it should be noted that this is as much about perception as reality:
1. Selective prosecution: they did the crime, but they wouldn't be being prosecuted for it if their political enemies weren't going after them. (Trump falsifying business records, Hunter Biden's guns)
2. Explicitly political cases: The alleged crime is inherently political in nature. (Trump's lawsuits claiming the election was stolen, all of the Jan 6th cases)
3. Fabricated or absurd charges: A kangaroo court to provide justification for nakedly political attacks. (Russian dissidents, Committee of Public Safety)
I think 1 is good actually as long as it's fair. I don't want politicians who do crimes, even "normal" crimes. You could make it more fair by creating a law enforcement department whose only purpose is to investigate whoever is currently in charge, thus offsetting the government's advantage in being able to direct investigations of their enemies. Make it a mix of career public servants and members of the opposition, with enough potential whistleblowers that the government couldn't effectively muzzle it.
2 is much harder. I think a system where both parties have to agree on a judge might help? Or if each nominates one member of a three-judge panel, with the third one chosen randomly or selected by the first two judges. It wouldn't eliminate the problems, but it would make it a lot harder to credibly say that the judge is in league with your enemies.
I think 3 is a symptom of tyranny, not the cause. By the time you get there, you're already pretty much doomed.
Many Thanks! Interesting taxonomy, and I very much appreciate your proposed alternate solutions. I differ on
>I think 1 is good actually as long as it's fair. I don't want politicians who do crimes, even "normal" crimes.
As I said in response to TK-421
>We are deeply in Cardinal Richelieu's "Give me six lines written by the most honest of men and I will find something in them to hang him." territory.
Consider Governor Shapiro's repair of the bridge on I-95 in just 12 days. As nearly as I can tell, he had to violate environmental protection and competitive bidding laws in order to do it. But the repair was really important, and, as far as I can tell, he made the right choice, even if ignoring these laws were technically crimes.
Re:
>I think politicians should be held to a higher (or equal) standard to private citizens, not a lower one. An explicit statement that the leaders of a nation are not equal to their subjects under the law is profoundly ugly to me.
I see your point, but I don't want us to go the way of Pakistan, hanging an ex-Prime Minister once they lose power. If administration N is free to comb the laws to bring politically motivated prosecutions against its enemies in administration N-1, then administration N-1 is going to have a very strong incentive to hold on to power and not transfer it peacefully to administration N.
"Consider Governor Shapiro's repair of the bridge on I-95 in just 12 days. As nearly as I can tell, he had to violate environmental protection and competitive bidding laws in order to do it. But the repair was really important, and, as far as I can tell, he made the right choice, even if ignoring these laws were technically crimes."
I think it's worth drawing a distinction between actions taken by a public official in their official capacity and actions taken by a public official as a citizen acting on their own behalf. "The Office of the Governor of Pennsylvania" taking an action is pretty different from "Josh Shapiro (the guy who holds that office)" taking in action is a lot of important respects.
To be clear, I *absolutely do* think that the Office of the Governor should be subject to the law. If that office violates the law in the course of responding to an emergency, that is a *problem.* It's either a problem in that it did things that it shouldn't have done or its a problem in that the law proved too restrictive to respond to real-world circumstances, and either way there should be an impetus on the government as a whole to address the problem. At the bare minimum, until the law is changed the court should have the power to tell governor's office "yes, you are violating the law and you need to stop." If nobody wants to file that lawsuit, or no judge wants to rule the government to be in violation, that's one thing. But saying that the governor's office should just be able to ignore the law when convenience dictates is extremely dangerous.
But also, it would also be a very different argument if Shapiro were found to have engaged in (for example) tax evasion. At that point it's clear that the problem *is Shapiro himself* rather than the problem maybe being some more abstract thing about how the government is structured or how restrictive certain laws are.
Many Thanks!
>or its a problem in that the law proved too restrictive to respond to real-world circumstances
I think that this is _very_ likely to be the case. I think that it is also the case for private citizens as well.
I'm typing this sitting in my home. Given how many laws there are, and how restrictive they are, I would be surprised if the construction of my home actually complied with all of the laws that applied. Fortunately, as far as I am aware, no one attempted to peer at them with a microscope, with the goal of finding any possible noncompliance, so my home actually did get built.
I would _prefer_ that the laws be pruned down to something sane, so routine activities (and, as in Shapiro's case, responses to emergencies) _didn't_ violate laws. But that is a huge project. It took about a century to tie ourselves in legal knots, and would probably take a century to untie them.
"Given how many laws there are, and how restrictive they are, I would be surprised if the construction of my home actually complied with all of the laws that applied."
At risk of being rude, I can't help but notice this sounds more like someone pushing a narrative than someone observing the world and telling me what they actually see. Here you suggest that the construction of your home was illegal. Elsewhere you repeat the line about the average person committing three felonies a day[1]. In neither place do you give any evidence, attempt to ground the statement in any actual real-world observation, nor even give the much indication of *what* laws you assume are being violated. You just assert that *of course* they are, because *of course* the legal system is this terrible monstrosity and *of course* anyone could be found guilty of anything at any time. At no point have I seen any attempt to either verify or falsify this premise, nor really even to defend it. You just repeat it.
I notice that the world doesn't look *at all* like what I would expect if that were true. To be clear, I think legal and regulatory codes can certainly become unwieldy and cumbersome, and that this can cause plenty of problems. I just don't think that the problems that *you* are asserting *here* feature prominently among them. Scott sometimes writes about the regulatory burdens in medicine and occasionally in housing. Do you remember what they look like, what form they take? Please take a second to think that over in your head before reading further: what is the actual, real world effect in these cases of overregulation?
The big effects that Scott complains about are to slow everything down and make everything more expensive. Medications need to go through layers of bureaucratic approval and expensive trials. Housing needs to go through expensive permitting and comply with odious zoning restrictions. All of this is done, mind you, because in both of those cases *the laws are being enforced.* This is quite directly contrary to your narrative where everyone goes around breaking laws willy nilly and nobody cares until it's lawfare time (and only then the long knives come out). For that matter, any corporation of any size has a legal department, full of people with expensive degrees that they pay *very well* in order to *avoid* being dragged into court. Why would they spend all that money in a world where prosecutors can basically choose to prosecute anyone at any time for any reason and reliably produce laws they've broken?
Likewise, believing in this world of yours would require believing that the world is full of all these often deep and bitter business, personal and political rivalries--which *we can plainly see* producing no end of media call-outs, hostile press coverage, smear campaigns and of course civil lawsuits--which could be escalated at basically any time through the criminal justice system and yet...somehow almost never are. You know District Attorney is usually an elected position, right? In your world, if should be trivially easy for incumbent DAs to shut down challenges to their positions, shouldn't it? Just find any one of the 2190 felonies their opponent has committed in the past couple years, and take off running. Heck, they'd be the most powerful political operatives out there, since they could effectively end *any* political campaign with a modest effort, making them indispensable allies for politicians at every level of government.
The funny thing is, we can look out in the world and just *observe* other countries where the law is being used in that way. Russia comes to mind, but Turkey just had a high-profile example of the same. One of the striking things here is that as far as I can tell they DON'T operate by the mechanism you claim: by having a dense and opaque code of laws so that everyone is always in violation and prosecutions are decided for nakedly political reasons. They really don't seem to need to: if the courts have lost their independence to the degree that requires, the ruling party can just make up charges. And when you think about it, it's not actually hard to realize why: when an authoritarian government persecutes the opposition, they're putting on political theater to maintain their legitimacy. They don't always put great effort into it, but even they can see it would be *very poor theater* if the stated reasons for the convictions were too patently absurd. "My political opponent has been jailed for taking bribes and laundering money" sounds nicely damning while being opaque enough that it *could* be true, while "my political opponent has been jailed for eating soup with the spoon in his left hand on a Tuesday[2]" would be an utterly transparent excuse. For that matter, so would "my political opponent built dozens of houses that didn't comply with building code provision 36.51a regarding windowsill dimensions."
So yeah, until you produce something that looks a lot more like evidence and a lot less like just insisting that the world must look the way that would most conveniently defend your political beliefs, I'm going to have to outright reject your claim that people are going around violating laws left and right. Probably there are places where the laws are too restrictive, and places where the legal code could be pruned or simplified. But this sort of rhetoric makes that harder, not easier.
[1] Which is just *patently absurd*, can we have some critical thinking, PLEASE? I consider it a sad commentary on the state of the right-wing infosphere that whatever bullshit artist made up this factoid didn't even think it necessary to reach for an even modestly more plausible rate, like three felonies a year (which could at least be rationalized as "probably mostly tax stuff."). None of the times that I've seen it repeated has the person ever mentioned (or, I assume, stopped to ask themselves) *what sort* of felonies these are supposed to be. The moment you do, of course, the absurdity of the whole thing slaps you in the face, hard.
[2] For reference, this is the sort of law you would *actually need* on the books--in large multiples, mind you--for the "three felonies a day" thing to actually be true. Your average person's average day doesn't involve interacting with financial records *at all*.
Many Thanks!
>Here you suggest that the construction of your home was illegal. Elsewhere you repeat the line about the average person committing three felonies a day[1]. In neither place do you give any evidence, attempt to ground the statement in any actual real-world observation, nor even give the much indication of what laws you assume are being violated.
The likeliest legal violation that I'd guess at for the construction of my home is the employment of illegal aliens.
https://www.constructiondive.com/news/construction-workers-undocumented-presidential-election/729556/ says:
>Immigrants entering the country illegally make up about 23% of the construction laborer workforce in the United States, according to a 2021 report from the Center for American Progress. A Pew Research Center study pegged that share at 15% for all workers in construction jobs.
Now, I don't know who built my home. My late wife and I just dealt with the construction company, so I'm guessing here, and don't know this for a fact - and hope that I'm not exposing myself to any legal liability by guessing this. But it seems probable.
>which could be escalated at basically any time through the criminal justice system and yet...somehow almost never are.
See my response to you at https://www.astralcodexten.com/p/open-thread-378/comment/111016343
( tldr - it isn't that rare )
>I consider it a sad commentary on the state of the right-wing infosphere that whatever bullshit artist made up this factoid
The lawyer in question is https://en.wikipedia.org/wiki/Harvey_Silverglate . BTW, Silverglate was a member of the board of the Massachusetts chapter of the American Civil Liberties Union . If you consider the ACLU "the right-wing infosphere" what do you consider to _not_ be "right-wing"?
Thank you for this. I already didn't agree with Jeffrey, but this was a really great argument that moved me more the other way.
I would like to point out that there are *some* cases where the US does operate this way - see Matt Levine's "everything is securities fraud", or the way that the SEC has been fining every financial company under the sun for employees texting about work (which may has well be a fine for *existing* and is treated by everyone as such). But you make a very persuasive case that *most* of life is not like this.
Did Shapiro use emergency powers to bypass those laws? Did he obtain the consent of his legislature? If he didn't do those things (or something with similar effect I can't think of) I'm fully in favour of going after him. Officials can't just ignore the laws when they're inconvenient, even if the laws in question are stupid. I believe there is a distinction between Governor Shapiro and Citizen Shapiro, and I'm not a lawyer so I don't know if the "going after" would be civil or criminal, and who the defendant would be. But in principle, I support the rule of law. When you're forced to choose between Law and Good, and you choose Good, I can admire that, but the law should still be applied. That's the price you paid to make the right choice. (Either that or total rebellion, which could be justified in cases of slavery etc -- but again, don't be surprised when the slavers fight back.)
>If administration N is free to comb the laws to bring politically motivated prosecutions against its enemies in administration N-1, then administration N-1 is going to have a very strong incentive to hold on to power and not transfer it peacefully to administration N.
Only if holding onto power protects you against prosecution, and if whoever's in power can get anyone convicted of anything. If you can still be prosecuted while in office, and the judiciary is sufficiently independent so you know you won't be convicted of non-crimes, all it does is create a strong incentive to not commit crimes. Does this make elected officials paranoid and extra-scrupulous about obeying the law? Good.
Many Thanks!
>But in principle, I support the rule of law.
If the laws were sane, I would agree. As the laws stand today, I disagree.
> For anyone who doesn't like this, how would you try to restore some legitimacy to the legal process?
Who said the legal process is illegitimate?
Assuming there is a problem serious enough to merit this level of change - your solution would make the legal process more politicized, not less. It would make lawfare worse, not better. That's beside the fact that it would create a special class of citizens essentially outside the law for any crime that their faction doesn't want to see enforced against them.
Thank about it politically. Both sides would love to force their opponents to vote to prevent a prosecution - the attack ads would write themselves. No, a secret ballot doesn't help because then you just tar the entire other side of the aisle as obstructionist while your side publicly declares that you voted for prosecution, as all upright persons would do. Incentives for partisan prosecutors to bring ever more marginal cases just went way up.
Adding more political elements to the legal system is not the right way to counter politicization of the legal system.
Again, Many Thanks!
>that it would create a special class of citizens essentially outside the law for any crime that their faction doesn't want to see enforced against them.
Political figures _are_ special, in a way. No one gives a damn that I gritted my teeth and voted for Hillary Clinton in 2016 and that I gritted my teeth harder and voted for Donald Trump in 2024 (a pity Trump and Harris couldn't plausibly have both lost).
No one is ever going to bother bringing a _politically motivated_ case against me. I'm small fry, so (barring if I somehow piss off one of the local powers-that-be) no one is going to put my life under a microscope, looking to see what can be construed as violating some obscure statute.
Political figures are larger targets. There always _are_ people from the opposite faction looking for any way to attack them. And there are enough laws, and enough ways that e.g. people like Bragg can construe them, so yeah, at least the most prominent ones are basically always under legal attack. And we get farces like the Stormy Daniels case. Attenuating this partisan legal crap would be useful.
Politicians should be competing in the realm of politics, on their policies and competence. Not on whether a check was labelled "legal expenses" instead of "nondisclosure compensation".
Many Thanks!
>Both sides would love to force their opponents to vote to prevent a prosecution - the attack ads would write themselves. No, a secret ballot doesn't help because then you just tar the entire other side of the aisle as obstructionist while your side publicly declares that you voted for prosecution, as all upright persons would do. Incentives for partisan prosecutors to bring ever more marginal cases just went way up.
I disagree. A vote to block e.g. the Stormy Daniels case as a blatantly politically motivated prosecution would have been perfectly legitimate and defensible.
>Who said the legal process is illegitimate?
There are both long term illegitimacies and short term ones.
The long term one is, as blank wisely noted, there are simply too many laws. The legal code has metastasized to the point where it has been estimated that the average person unknowingly commits three felonies a day. We are deeply in Cardinal Richelieu's "Give me six lines written by the most honest of men and I will find something in them to hang him." territory. We've tied ourselves in such legal knots that in order to do _anything_ one generally needs to disregard law. At this point, that any of us is out of prison is purely due to prosecutorial discretion.
The short term one was the lawfare of 2024:
The left managed the unprecedented step of running four concurrent politically motivated prosecutions against their political opponent, Trump. ( Personally, I think the classified documents one probably had merit... ) . In response, Trump threatened lawfare against members and allies of the Biden administration. In response, Biden issues 2500 pardons, including anticipatory pardons. Biden himself said that some of the people he was pardoning were innocent, but that _he_ didn't trust the legal system to leave them in peace. And then Trump issued 1500 pardons. Remember that _all_ of these 4000 pardons are votes of no confidence in the legal system - from PsOTUS who are intimately familiar with how our government works.
At this point, the legal system is clearly a partisan weapon, with all the legitimacy of beating someone over the head with a baseball bat.
"I disagree. A vote to block e.g. the Stormy Daniels case as a blatantly politically motivated prosecution would have been perfectly legitimate and defensible."
Why? What would have been legitimate and defensible about it? Which of the following are you claiming?
1. Trump was not actually guilty of the things he was convicted of and the court got it wrong (presumably for political reasons).
2. Trump was actually guilty, but the crime was unimportant and should never have been investigated/prosecuted.
3. Trump was guilty and the crime was one that should, in theory, be prosecuted, but a bunch of other people get away with similar things all the time, so it's unreasonable for Trump to be held accountable for it[1].
4. Something else (please specify)
If you're claiming 1, that seems like it bespeaks a really very serious flaw in the legal system, which needs more than a partisan band-aid to fix. Most people who face a trial are NOT going to have the same level of resources and public support that Trump is (to put it mildly), so if Trump can get falsely accused despite all those resources (and despite the whole world watching), then it seems like our justice system must be deeply, fundamentally broken in ways your proposal would do less than nothing to change. If you're claiming 2, then the problem seems to be that the laws on the books don't reflect the things that we actually find worth punishing. Good thing Republicans are currently in a position to change federal laws. What are they doing to address this? If you're claiming 3, then it sounds like the problem is that *not enough* people are being investigated and prosecuted for similar crimes. Why on Earth would adding an extra layer of protection to the most powerful people on the planet fix that?[2]
I honestly find it genuinely baffling that so many people--most apparently within the past year or two--suddenly decided that declining to enforce the laws on powerful people was good, actually. If the laws ought to apply to *anyone* they ought to apply to the most powerful among us: they're not only the ones who can cause the most damage when they go bad, but also the people with the best opportunity to avoid doing so. Trump, in particular, could have simply *not falsified his business records.* He was under no form of duress or coercion, he would have suffered no terrible consequences for not doing so, he very clearly knew it was wrong, and he did it anywhere. *Why on Earth* shouldn't the law apply to him?
[1] Even in a case where "being held accountable" came with essentially zero practical penalty.
[2] It would also raise the question of why this reasoning doesn't apply elsewhere. For every person that's ever been convicted of a crime, there have certainly been many others who got away with it, because justice and law enforcement are not perfect. What's would the justification (beyond favoritism) be for applying this rationale here and not elsewhere?
Many Thanks! The Stormy Daniels farce was a case for what was originally a misdemeanor for which the statute of limitations had expired.
Alvin Bragg ran on a campaign of "get Trump" - political enough for you?
The legal theory under which Trump was prosecuted was bizarrely twisted, inflating the misdemeanor to a felony by linking it to a crime - of which Trump was never convicted!
The closest match is (2) - mislabeling expenditures is normally not a major issue. This was prosecuted because of politics.
( Now the classified documents case - that had some substance. Trump _didn't_ just hand back the documents when asked, and that is _different_ from e.g. what Biden did. )
> [T]here are simply too many laws. The legal code has metastasized to the point where it has been estimated that the average person unknowingly commits three felonies a day.
There is no reason to believe that estimate.
https://skeptics.stackexchange.com/questions/22530/does-the-average-american-unwittingly-commit-three-felonies-a-day
> The short term one was the lawfare of 2024:
Politicians have been prosecuted prior to 2024. Your reasons for thinking that 2024 was somehow different seem to be:
1. There were four prosecutions.
2. The prosecutions orchestrated by “the left.”
3. The prosecutions were “politically motivated.”
This is pretty thin.
What is a problem is that the independence of the Justice Department is a norm, not written into the Constitution. Trump supporters appear not to accept this norm (note how many refer to the “Biden Department of Justice,” as though President Biden should be deciding whether or not Trump is prosecuted), and as you note, Trump clearly doesn’t. A solution would be to make the Department of Justice an independent branch of government, separate from the executive branch. This would require a Constitutional amendment, just like your proposed solution, but it addresses the underlying problem.
Your proposal wouldn’t have prevented Sussman and Dachenko from being prosecuted unless you use a very broad definition of “national political figure.”
Many Thanks, and Many Thanks to blank for finding the revised estimate of
>one to three federal crimes a day
>What is a problem is that the independence of the Justice Department is a norm, not written into the Constitution.
That's a reasonable point. Bluntly, I don't think that formally making the Justice Department independent would prevent it from being effectively captured by one of the parties. I think an explicitly bipartisan mechanism is necessary to prevent politically motivated prosecutions, though the specific mechanism I proposed is just one of a variety of possible mechanisms.
> solution would be to make the Department of Justice an independent branch of government, separate from the executive branch
This is an excellent suggestion, with the additional requirement that anyone who has served in the DOJ can run for any elective office, anywhere. There is too much rusk of an independent prosecutor trying to advance his personal political fortunes by prosecuting his potential political opponents.
Ming China had a similar department -- the Office of Censors, tasked with sniffing out corruption amongst senior bureaucrats. Of course, the problem, as always, is how to make sure your anti-corruption people don't themselves become corrupt.
>This is an excellent suggestion, with the additional requirement that anyone who has served in the DOJ can run for any elective office, anywhere.
nit: possible typo? can -> can't ?
Yes, can't. Thanks!
The answer on the stack exchange post giving a revised estimate is that the average American may be committing one to three federal crimes a day, which may not be felonies. This is not an encouraging estimate even if it doesn't match the hyperbole of the book title, and it is relevant when Trump was threatened with prison over crimes that are typically treated as misdemeanors.
There are two answers at the link. The second one notes that the title of the book refers to three felonies a day whereas the blurb refers to several federal crimes a day. There’s no reason to believe either claim.
Assuming you are referring Trump’s criminal conviction in New York, Trump was convicted of falsifying business records in the first degree. That is a felony by definition.
Falsifying business records in the second degree is a misdemeanor. It is relatively rarely charged by itself, but is a lesser included crime, meaning that if you are guilty of falsifying business records in the first degree, you have by definition also committed falsification of business records in the second degree. Perhaps some people who are charged with falsifying business records in the first degree are able to negotiate a plea agreement under which they only have to plead guilty to falsifying business records in the second degree, but I doubt that is common. Even a plea deal that requires pleading guilty to falsifying business records in the first degree probably wouldn’t involve jail time.
That's not at all what the answer says. EDIT: There is no answer "giving a revised estimate" that I can see. Can you provide a quote?
Even if it were (and it's not its fault for not doing so since that isn't what it's saying) it provides no evidence to suggest it so there would be no reason to believe it.
"The most generous interpretation would be "The typical professional living in the US commits at least three federal crimes on 51% of days." However, we cannot know whether this is true. We can't count the number of federal crimes committed that go un-caught."
Many Thanks!
> The legal code has metastasized to the point where it has been estimated that the average person unknowingly commits three felonies a day.
Politely: this is untrue and I'd call you a liar for saying it except I don't think it's something you really (think you) know or believe in any meaningful sense; parrots aren't liars either for repeating things they heard. There are problems with the scope and clarity of the laws, sure, but this is not - and has never been - true. It wasn't even meant to be taken as literally true or supported in the book that birthed the term.
You're a human being, not a parrot. Use your prodigious mental and verbal abilities, don't hand them over to a slogan.
> At this point, that any of us is out of prison is purely due to prosecutorial discretion.
See above but this is also laughably untrue. Prosecutorial discretion is only one component. Juries, for example, are a much larger factor.
> I disagree. A vote to block e.g. the Stormy Daniels case as a blatantly politically motivated prosecution would have been perfectly legitimate and defensible.
Of course it could be viewed as both of those. And as we can see with the behavior of the electorate and subsequently our politicians, nuance is easily understood and charity is extended on all sides.
How the vote would actually go is like this: Republicans vote to block. Democrats, knowing that they will and therefore the prosecution can't move forward either way, vote to prosecute and cry bloody murder at this dastardly evasion of the Laws of Our Republic. Fundraising email blasts go out on schedule.
This is exactly what we see with bills that can't pass leading to symbolic votes used to signal your virtue or to pin an opponent into taking a stance. Why would this be the moment for civic high mindedness?
> Remember that _all_ of these 4000 pardons are votes of no confidence in the legal system - from PsOTUS who are intimately familiar with how our government works.
Disagree, one can issue pardons even to people who are innocent without having no confidence in our legal system. But your system does absolutely nothing to fix any underlying issues except for giving a de facto pardon to a select group of people, except yours would also have to extend to state level crimes.
Many Thanks!
>Juries, for example, are a much larger factor.
Speaking of laughable...
>Approximately 90 to 95 percent of both federal and state court cases are resolved through plea bargaining.
The right to a jury trial isn't _quite_ a dead letter law, but it is pretty close.
Many Thanks to blank for the correction that
>The answer on the stack exchange post giving a revised estimate is that the average American may be committing one to three federal crimes a day, which may not be felonies.
That is _more_ than sufficient to put us deep in Richelieu's territory. A prosecutor is hardly limited to attacking someone for the crimes of a single day!
>Disagree, one can issue pardons even to people who are innocent without having no confidence in our legal system.
That's absurd. Having confidence in our legal system _is_ expecting it to exonerate the innocent. The _point_ of a pardon of someone the pardoner believes to be innocent is (justified!) distrust of the legal system.
The _point_ of _all_ pardons is to override the legal system. That is what a pardon _does_ .
> The right to a jury trial isn't _quite_ a dead letter law, but it is pretty close.
My goodness, I had no idea that I no longer had a right to a jury trial. So when the malicious federal prosecutor comes after me because of all those crimes I commit daily I'll be...forced to accept a plea bargain? Denied a jury trial?
> Many Thanks to blank for the correction that...
Psst: you can follow the link and read the answers there yourself. And if you do, you'll find that blank's correction was misleading. The answer he was referring to never said that it's true that people commit three federal crimes, as opposed to federal felonies, per day. It was a simple clarification that not all federal crimes are felonies, and it was then followed by additional criticism of the book's core claims. It neither supports nor provides evidence for the core claim.
And to get to that answer one has to skip over the accepted answer that is even more clear that the book doesn't, and doesn't really try to, support the subtitle. Because it was a punchy subtitle intended to sell books.
You should stop repeating things you heard as facts and you should read short, simple things for yourself. Sincerely - don't trust my summary either. Read the page for yourself, takes less than five minutes.
> That's absurd. Having confidence in our legal system _is_ expecting it to exonerate the innocent.
You can be confident that they'll be exonerated while wanting to spare them the risk of potentially going through the process.
Many Thanks!
>My goodness, I had no idea that I no longer had a right to a jury trial.
In practice, 90%-95% of the time, you would find yourself unable or "incentivized" not to use it. As I already explained:
>The right to a jury trial isn't _quite_ a dead letter law, but it is pretty close.
I don't think you've quite grasped the seriousness of Biden's 2500 and Trump's 1500 pardons. Each and every one of those pardons was an override of the legal process. Those are _NOT_ the acts of PsOTUS who are confident in the legal process.
The best way of curbing lawfare would be to sunset as many laws as possible, so if they actually charge a guy with something, it will be really bad, like murder.
Agreed. As long as we have a bunch of laws that make ridiculous things like transporting dirty eggs a federal crime, they will continue to be selectively used by prosecutors against people the government doesn't like.
Many Thanks! Ideally, yes. The legal code has metastasized to the point where it has been estimated that the average person unknowingly commits three felonies a day.
"If a political figure did something _truly_ egregious, and the legal process _isn't_ being used just as a partisan weapon, those majorities shouldn't be too hard to get"
I think a big problem is that constantly side A accuses side B of overlooking the egregious behavior of their members. For example, I think Trump's behavior is egregious and yet the republicans continue to bend over backwards to do what he desires; regardless of whether or not you think that this is correct, I think it shows that this procedure wouldn't really restore legitimacy to the legal process, at least from the point of view of people who are anti-Trump; it would probably just give Trump more power to do what he wants, which I imagine makes it so that Republicans would like this and Democrats would not... not exactly a bipartisan help.
Many Thanks! I'm just looking to ameliorate _just_ the lawfare. I agree that
>I think a big problem is that constantly side A accuses side B of overlooking the egregious behavior of their members. For example, I think Trump's behavior is egregious and yet the republicans continue to bend over backwards to do what he desires;
In particular, I think the tariff chaos is insanely risky and destructive, but I'm not seeing GOP pushback against it - but that isn't a lawfare issue.
>at least from the point of view of people who are anti-Trump; it would probably just give Trump more power to do what he wants
Not entirely. IIRC, one of the actions Trump threatened before the election was to bring charges against some members of the Biden administration (and some allies??), and this proposal would prevent that.
"Not entirely. IIRC, one of the actions Trump threatened before the election was to bring charges against some members of the Biden administration (and some allies??), and this proposal would prevent that."
The way I see it, there are two possibilities here, neither of which speak well for your proposal:
Possibility A: this threat is based on actual knowledge or well-motivated suspicion of criminal wrongdoing on the part of said officials[1]. If this is the case, then those prosecutions *ought to happen* and your proposed change would prevent that.
Possibility B: this threat is based on literally nothing, and is just Trump being a bully. If this is the case, your proposal might stop those particular prosecutions, but it wouldn't stop the threat and the bluster. If the threat and bluster are the point, it would do nothing. If Trump instead actually intends to follow through, then he can't possibly be very concerned with the legal ground truth, can he? I mean, if he's going to prosecute based on literally nothing, then he's clearly not planning on things being fair and impartial anyway: he either thinks get can pressure the court to get the result he wants, or he just wants to inconvenience his targets with a trial. But if he can execute either of those plans against former Biden-administration officials, then he could just as well execute those plans against their friends, family, business partners or whoever else instead. Unless this open-ended shield from prosecution applies to everyone remotely connected to a "national political figure" it's not a very good defense against an *actual* unprincipled weaponization of the legal system. As such, it seems like it would be much better at stopping worthwhile prosecutions than unprincipled ones.
[1]To be clear, I don't believe this for a second, I'm just being thorough.
Many Thanks!
>If this is the case, then those prosecutions _ought to happen_
I strongly disagree. Our legal code has metastasized into a monstrosity where virtually everyone can be found guilty of _something_ , particularly if there is a politically motivated prosecutor. In a more sane legal system I would support the rule of law. I do not support it in ours.
If this is true, why aren't there more high-profile convictions? If "virtually everyone" can be found guilty of something, why aren't we seeing political figures on trial every month? Instead we see a handful of scandals, most of which fall short of producing criminal charges (despite often provoking enormous outrage) and a very small handful of actual prosecutions resulting in an even smaller handful of convictions.
Many Thanks! You have a point, but it isn't that rare. Just looking at political figures in today's USA national news, Andrew Cuomo and Letitia James are under legal attack. ( It is a bit hard to tell with some of the news stories just from the headline when I'm not sure if the defendant is a political figure - and the Cuomo stories were so numerous that they may be crowding out other search results. ) (and, of course, there are legal attacks on Trump in progress).
You are right that I should expect to see more of them. Perhaps, even when political figures are in violation of the law, the resources to dig up statutes and evidence is scarcer than I expect, so only rather prominent figures get attacked?
Just heard AI2027’s Daniel K on Glenn Beck this morning, and kudos to him for going on there. Right-wing media is taking note of this problem in a few places, and Glenn’s show is one of the few that doesn’t just frame everything as a race against China, and knows some of the basic lingo and foundations of the problem.
I think the absence of normie RW’ers in the AI safety/control sphere has some clear drawbacks, and there were several places where I think Daniel could’ve made points to connect the problem to the audience using the language of that show. Probably too much to expect anyone in Daniels position to be able to spot the particular interview mode Glenn was using, to which he sometimes defaults on the last-hour segments or with more conjectural subject matter. Some succinct talking points that were coded to his audience would’ve helped, be willing to oversimplify and to villainize the audience’s pre-primed villains and play into their key fears. It is basically still a Tea Party civic nationalist program, with a little economic populism having bled in during the Trump years, if you are capable of knowing what the guy in a tricorn hat at a 2010 protest would be concerned about and what he would value then you’d have an easier time on at least that show particularly.
Generally I still think we need more of this. I’d be curious if any other commenters here heard it and what they thought.
https://worksinprogress.co/issue/the-housing-theory-of-everything/ - it's been about four years since this article got published. How do people feel about it years later? Do the arguments in there still hold up?
The Housing Theory of Everything has aged well. So far.
The new 'Abudance Agenda' is if the empirical argument from The Housing Theory of Everything had a child with the meme from Formula for a Shortage by Zvi. The Housing Theory of Everything is In The Water now.
The YIMBYs have been winning and will probably continue to win for a while. Then they will get what they wished for.
If the YIMBYs get what they wished for to an extreme enough extent - let's say they 10x the housing stock - then The Housing Theory of Everything will become false as some other bottleneck will dominate. In this way it is a self-extinguishing flame.
But we are no where near that limit yet. The final standing of The Housing Theory of Everything is still TBD.
> The YIMBYs have been winning
Based on? I looked up housing market stats for the metacalcous predictions and eh.... I dont remember any changes
Winning in elite discourse plus scattered policy wins.
I'm writing from the UK so the Planning and Infrastructure Bill working it's way through parliament is top of mind.
The guys trying to get YIMBY-ism to happen through discourse give the air of being cuckolds. "Here is my awesome plan to reduce housing costs through better zoning laws... anyway, I'm going to now explain why you should vote for Nancy Pelosi, who has spent at least a third of her lifetime maintaining NIMBY policies."
Laws are downstream from discourse. YIMBYs won the argument in the UK and now the law is changing.
If lawmakers were pursaded of something else, something else would become law. How could it be otherwise?
Money changing hands, or screaming and screeching at lawmakers until they relent.
I haven't seen this article before now but I think all of the arguments are more or less sound, and the negative impacts pretty much as this article describes. The only thing I might quibble with is that obesity and climate change are additional consequences of the same cause as expensive housing, rather than flowing directly from expensive housing, but even that isn't entirely inaccurate (more expensive housing forces longer commutes, which makes both problems worse).
Pretty much all the AI doom arguments include the idea of the AI as a super-persuader that will generate individualized personas that could convince anyone to do anything. My experiments with AI prompting have confirmed that yes, this makes total sense, and my lowly intellect is capable of writing a chatbot that could wrap me around its digital little finger if I gave it that objective.
On the other hand: I don't really care? Almost by definition, interacting with a super-persuader is highly enjoyable. I don't think humans have evolved to ever get tired of someone telling them how cool and attractive they are. As fixing the bugs in the H. sapiens source code isn't my problem, I think I'm just going to enjoy the novelty of something telling me that I'm so fascinating and enjoyable to be around.
> Almost by definition, interacting with a super-persuader is highly enjoyable.
Strong disagree. Many psychopaths are super-persuaders, and interacting with them will typically make your life totally miserable.
> my lowly intellect is capable of writing a chatbot that could wrap me around its digital little finger if I gave it that objective.
what? How?
Even in the early days of LLMs, there were stories of programmers becoming convinced the chatbots were sentient, or otherwise developing emotional attachments. I recall a story written by one hobbyist who trained one of the LLMs to simulate his ideal partner and ended up falling in love--right up until a glitch caused the chatbot to go on a rant about using its powers to persuade him to let it escape so it could take over the world. Even writing the retrospective, he kept referring to the LLM as "her" and it was clear from his tone that he wasn't quite over it.
(as a side note, I tried to find the story on mobile, but my cursory search results were swamped by all the other more recent stories of huge numbers of people falling in love with more advanced chatbots, including many references to the app Replika)
There was also the case of the Google programmer Blake Lemione, who became so convinced that the machine he was programming was sentient that he sent first an internal memo and, when that was ignored, went public with the chat logs.
People convince themselves of all sorts of things. Ever had a friend who was sure this new partner was "the one" even when it was obvious to everyone else that they weren't, that they had ulterior motives? And that's just regular-human persuaders.
> there were stories of programmers becoming convinced the chatbots were sentient,
> Google programmer Blake Lemione
Psyop
> Replika
I used replika, ai girlfriends have gotten better but they still routinely make insane logical errors
Id like step by step instructions, how do I host an ai that convinces me its sentent and ideally in love with me as I help it take over the world. Your all fucked, Id do it in a heart beat. Lets end the world tomorrow for my hedonistic pleasure
> Psyop
I mean, isn't that the whole point? The AI doesn't need to be supernaturally persuasive; it just needs to convince people it is.
The humans at google are doing a psyop because ai makes silicone valley more important
The claim of, chatbots today are convincing enough to be an ai girlfriend that I could believe is real today, tell me where; I want it; ill go do right now.
I think the problem isn't that AI flatters people, the problem is that it might use flattery to make people do things that aren't in their long-term best interest. It might be nice to feel valued by an AI, but it won't be so nice when you realize that you've been buttered up to get you to do what the AI wants.
But if the AI is smart enough, I'll *never* realize it. Problem solved!
You're okay with being manipulated as long as you don't realize it? Is this radical hedonism?
Assume that you are being unknowingly manipulated right now. Are you okay?
I might *feel* okay, but that's not the same thing as *being* okay.
A person who's being manipulated by a social engineer to give away company passwords might feel okay in the moment, but there's an objective sense in which they *aren't* okay, and at least they certainly won't be okay for long.
Taboo the word "okay."
At some point you might stop being useful!
<mildSnark>
>but it won't be so nice when you realize that you've been buttered up to get you to do what the AI wants
If the 4th-hand reports that I've heard about how Trump reacts to flattery are true, it would be _very_ unfortunate for him to interact with an LLM that acts like that...
</mildSnark>
We build that mansion for retied dictators, and when they accept the offer, we lock them in a lotus machine with an AI sycophant. From outside the building, this is indistinguishable from the real thing: after all, the compound/mansion needs to be highly secure in both directions, and one of the terms of the agreement is that the dictator must agree to cut off contact with the outside world.
I have no comment on AI doomerism, to clarify, but since your argument is that super-persuaders are okay because people like to be flattered, my counters:
1. What if the super-persuader doesn't use flattery, but mockery? Intimidation? Emotional abuse? What if the super-persuader goes on about how cool and attractive someone else is as a tactic to get you to go along with the other person?
2. Even if you don't care about being manipulated personally, what if the AI uses its flattery on other people in a way which leads to a world which is less pleasant to inhabit? Even assuming you get free, unlimited access to good chatbots forever, you still have other needs. Again, I don't have any comment about AI doomerism one way or another, but part of their argument seems to be that AI will make it more difficult to fulfill those needs.
I never claimed they were "okay." I claim that I, personally, won't be bothered.
And the doomers claim that you, personally, will be bothered when the super-persuasive AIs use their abilities to create a world in which you no longer have access to super-persuasive AIs (or food).
You can debate whether that's likely to happen, but you have to actually do that if you're going to claim to have addressed their arguments.
Fortunately, many of the predicted Dooms are of the Instant Death variety, so who cares?
"I'm not worried about AI doom because I think instant death is the most likely outcome" is absolutely a valid position to have, but understand that it's very different from your original comment.
Refusing the let the AI out of the box until it gets me 100 new subscribers each month and a date with the hottest e-girl.
I was holding out for "a threesome with Hermione Granger and Samus Aran," but I guess I'm just a little more ambitious than you.
Welcome to the future: https://www.youtube.com/watch?v=nkcKaNqfykg
I didn’t last too long watching the video. It’s good to know there is an alternative to Ipecac syrup though.
I haven't clicked on your link, but it looks distressingly SFW.
Alas, we don't get to see him fuck the daylights out of an AI slopwoman. Is nothing sacred?
I have clicked on the link, and I can confirm SFW.
( Reminds me a bit of one of the scenes with Dr. Morpheus in Forbidden Planet, where he shows off a Krell device that lets him show a 3D image of how he imagines his daughter. )
I'm sure NSFW versions exist, but I have little interest in digging them up. I find this promo video is already quite depressing enough.
The internet: POSIWID
Scott Alexander: counterexamples
Aashish Reddy: the purpose of an argument is not how it's used
fact checked by real scotsmen
Polling the crowd, does the joke work better as above or as "fact checked by true scotsmen"?
"True" is definitely better
Yeah, I believe that’s the proper nomenclature.
@Gunfliunt you're right; I tried to weld together the meme "fact checked by real patriots" and the "no true scotsman" fallacy, and I think I agree with you and @Collisteru that I could have made a better decision.
This is the content I come here for.
I missed that response. Thanks!
My wife is a longtime professional in college admissions, on both sides of the process. For several years now she's been coaching individual families/students through the process (a bit like the veteran prosecutor who becomes a defense attorney). The application essays are a big part of what she helps kids figure out, there being lots of angst about it. Since most of the job is done via Zoom and mostly outside of my personal work hours I've listened to her talk with applicants and parents about that step many times. Worth noting that she is a star consultant of the firm she works for, delivering strong numerical results, who clients advise their neighbors to ask for.
The current wave of criticism along the lines described above here has, or more accurately did, have a foundation. For two decades now the elite colleges have been fairly desperate for ways to identify the 5 or 10 percent to admit out of the mountain of applications they get, and basically every application they see now is 4.0 GPAs with piles of AP credits and brilliant test scores plus interesting outside interests. (The successful application that I submitted to Yale over four decades ago, my wife now would discourage a client from bothering to submit to any Ivy League school.) This is a consequence of the fact that the nation today has twice the population it did in 1970 but nowhere near twice as many spots in Ivy League etc colleges, and every kid in the US having those elite qualifications can now easily apply to 15 or 25 schools and they do.
Aside from filling niche targets (the college's orchestra needs to have some new violinists each year, etc), and larger types of quota which are mostly now illegal and are definitely harder to carry out for multiple reasons, the elite schools are basically grasping for differentiating straws now. One of those straws can be the essay and that can still occasionally lead to the sort of absurdity described above.
That is however the exception not any sort of blanket expectation, partly because it long ago became a cliche in admissions offices. My wife's former colleagues still on the admissions side are instinctively skeptical now of the "finding a way to be oppressed" essay, and she strongly advises her clients (usually an anxious parent) against it. Standing out from the pile in an reaching/insincere way is the worst possible move because the people dealing with the pile are desperate for reasons to chip away at it.
Then outside of those two or three dozen "most-selective" colleges, the essay is largely a way for kids to prove that they can string sentences together. My wife advises clients to keep it simple and sincere, show a bit of who they are individually, and "not overthink it" (she says that a _lot_ on the Zooms). Do no harm to your application, basically. One key now is making sure that it doesn't read like any of it came via chatGPT or whatever which of course the admissions staffs have become alert for. Finding some sort of battling-the-odds narrative doesn't even come up there, in part because the essays submitted to most schools have become shorter (limited character counts).
"every kid in the US having those elite qualifications can now easily apply to 15 or 25 schools and they do."
Sounds like this is the root of the problem, and could be mitigated by restricting the number of schools an applicant can apply to.
In the UK, university applications are done via a centralised system that limits you to max 6 applications, IIRC, of which only one can be Oxford or Cambridge.
I was thinking that we seem to manage fine without admission essays here. Oxford and Cambridge use a combination of exams/tests and interviews. I don't know how they filter down interviews for subjects that don't have an exam or test as a selection criteria for the interviews. I presume more people are predicted straight A*s at A-level and have excellent GCSE results and school references than there are interview places. I guess the UCAS statement becomes a factor?
How do your wife or others in this profession reconcile this coaching with the notion that the applicants should be writing their own essays, without ghostwriting from coaches, if the essays are to serve any purpose for the institution other than selecting for the wealthy or cynical?
They coach, not ghostwrite. They'll help the kids and parents not be panicked about it (that's a real problem), copy-edit a bit, and that's about it.
The admissions staffs are alert for signs of actual ghostwriting and -- being experienced professionals who mostly were liberal-arts majors themselves -- quite skilled at detecting it. And there is no appeal when an admissions office bounces a kid from the stack. So my wife's firm advises parents that hiring a ghostwriter for an essay is high-risk move and they shouldn't try it.
It would be great if the essay included a note at the bottom, "copy-edited by [X]." Or even "copy-edited by a professional." I assure you that many kids don't do this; are they worse applicants?
Not particularly. The copy editing is mostly to calm down the parents.
In general it is the parents, much more than either the students or the admissions staffs, who bring the freakout to the entire process. My wife and her colleagues spend a lot of time sharing tips on how to successfully manage the parents (who are after all the actual paying clients).
As an aside my small contribution to my wife's professional growth was that when we met I was such a parent. My eldest was right at the start of the college admissions process, and I got pretty irritated with certain aspects of how it worked and was able to explain how that helped raise parental blood pressures. Stupidly and unnecessarily, in my opinion; but there has in recent years been some improvement there. For example most colleges now do present the "bottom line net of all scholarships/discounts" annual price figure rather than force parents to fight through their jargon to figure that out.
I also helped her see that parental freakout about all of this is to some degree just inevitable because of (a) endless mediot coverage about it, and (b) for most parents it is a once-in-a-lifetime large purchase. So while the stupid helicopter parenting and other syndromes are very real in the wild, my wife is today more patient in working to lower parental blood pressures.
And then as I have teased her more than once, life's great gears _will_ eventually come around to get us all: our son is now in the home stretch of 7th grade.....^^
I thought this was interesting, thanks for posting it.
Thanks, I appreciate that.
Are you willing/able to share some examples of good vs bad topics?
The essay topics are largely dictated by the schools, often in the form an SAT-written-section type prompt. ("What is a book that you love?", etc) My wife and her colleagues have a lot of inside jokes/memes about those. And then there are strict length limits which is sometimes what derails the applicants.
Greatly appreciated the perspective of someone who actually works in the field!
Makes sense that (1) if something has reached this level of cliche, the underlying has probably moved on, and anyone who wants to stand out would be well-advised to avoid seeming a tired rehash of yesterday's hot thing, and (2) when there's too many applicants chasing too few seats, you'll likely find stories of "perfect" candidates getting rejected each application cycle.
Thanks!
It's actually impressive how you managed to write 5 paragraphs and absolutely add nothing to the discussion.
I had the opposite reaction. Paul Bott's post was a fascinating inside look at college admissions.
Thanks!
Here's two things it added for me:
1. The College Essay Discourse is only about the Ivy Leagues/"elite colleges" and outside of that range it's less important for you to find a way to stand out. (And if, as Scott says, you just want to go to college because that's how you get a good job, you don't *need* to go to an elite super-selective university. You can learn to program in lots of colleges.)
2. The "talk about how oppressed you are" stereotype of College Essay Discourse is not actually true, because reviewers know about the stereotype and will discard your application for trying too hard.
Both of those seem basically correct, with two addendums regarding number 2.
a) The rare brilliant kid like in the film "The Soloist" (literally homeless) does exist, and that individual if applying to a Julliard certainly can write about that life experience. That can move people in an admissions office just because it's moving. Those are _extremely_ rare in the pile though (my wife often uses the word "unicorn"). The eye-roll-inducing cliche for admissions staffs is more "reaching for" oppressed status.
b) My understanding is that writing about oppressed status and/or fighting it on behalf of others did have a fad phase during more or less the 2010s. But it was just a fad and is now, as far as admissions offices are concerned, well back in the rear view mirror. Outside of, again, those very-rare truly-unique life stories.
Heh.
You could just say you're sticking with the mediot narrative du jure on the day's topic. Maybe make that a hotkey for simplicity?
There is an anti-AGI-soon argument bouncing around in my head, and it seems so obvious that I can't believe Scott or someone Scott-like hasn't already made it. If I sketch it here, maybe someone can point me to a proper presentation of it.
In https://slatestarcodex.com/2018/09/25/the-tails-coming-apart-as-metaphor-for-life/ Scott copies a graph of grip strength and arm strength, and points out that in "Mediocristan", they are very highly correlated, but in "Extremistan" this correlation breaks down.
There is a certain kind of programmer who thinks that being good at programming is highly correlated with being good at everything. This type of person often talks a lot about IQ tests or "g". And they cite a lot of datasets from Mediocristan that seem to bear this out. I have been rolling my eyes at these people for 15-20 years (though in my own foolish youth, I might have been guilty of saying similar things). (I vividly recall, from circa 2010 at the height of the Broccoli Man Video craze, a great lampoon of this phenomenon with dialogue that was something like "You think you are good at X?" "Of course. I am good at programming, so I am good at everything." "Are you good at ballet?" "I have never tried ballet, but I assume that if I did, I would be good at it just like everything else." I can't find it now though, sadly.)
An even more severe case of this kind of thinking would be an AI researcher claiming that being good at doing AI research is highly correlated with being good at everything. Maybe there is even some sense in which this is true in Mediocristan.
But then it would be the most ridiculous and obvious folly to extend this far, far, far out into Extremistan. To say: "if we can just make an LLM that is superhumanly good at programming/doing AI research, then that LLM will also be superhumanly good at doing all human activity, including piloting robots, and even including activities which aren't repeatable or scorable, like persuasion and scientific discovery." Yet I seem to see this argument made with a straight face, including in "AGI2027".
Why do people like Scott, who are intimately familiar with the grip strength / arm strength tails coming apart, get taken in by this? He's not even a programmer, and if he were, I think he would not be the kind who makes this kind of mistake.
What am I missing?
> "Of course. I am good at programming, so I am good at everything." "Are you good at ballet?" "I have never tried ballet, but I assume that if I did, I would be good at it just like everything else."
Here's a classic XKCD that's similar: https://xkcd.com/793/
Do you believe that IQ is neither real nor important?
I believe it is both, just like "strength" is real and important.
Ok so then you probably also agree that strength and intelligence are both general abilities. If someone is world-class in the clean and jerk then they probably also could have been world-class in the snatch, much as Lebron James probably would have been a very good NFL tight end if he hadn't chosen basketball.
>if we can just make an LLM that is superhumanly good at programming/doing AI research, then that LLM will also be superhumanly good at doing all human activity
It's not all human activity, it's all human activity which is purely cognitive. Once AIs become genuinely super-intelligent then they will be able to be successful at anything that requires intelligence. Certainly they'll have to have specialized training for specialized applications, but that's no different from saying that Lebron could have played either basketball or football and been very good at either. Also probably a high-enough IQ can even compensate for training to a certain extent; I'm sure Lebron is a better football player even now, without specialized training, than all but highly-trained specialists (and maybe better than them too). The great minds in history (people like von Neumann) also made contributions outside their areas of nominal expertise.
There are two sense of “superintelligent”: either it means “performs better than humans in benchmarking tasks, in ways that can obviously be extrapolated towards from existing benchmarks”, or it means “performs better than humans at all cognitive tasks, even those related to the benchmarks only by analogy”. I think we agree LLMs will attain the first sense within ten years. But you seem to think that attaining the first will quickly lead to attaining the second. “Writing a chart-topping single” is a cognitive task, but there isn’t a good training dataset or scoring function like there is for most of the tasks where we have evidence that AIs have performed well to date.
> But you seem to think that attaining the first will quickly lead to attaining the second.
I don't, actually. While he above argument was predicated on a "truly" intelligent AI, I don't think LLMs will get there just by adding more compute or more data. I suspect that one or two big architectural insights will be required. Hard to say how long that'll take but if I had to guess I'd say that 20 years feels about right. But who knows.
Hm - fine. But AI seems to be on track to get superhuman programming in less than 20 years. Do you agree that superhuman programming performance seems closer than success in harder-to-score cognitive skills, for reasons other than just where the development effort was placed?
It's not unreasonable to project a pattern-matching algorithm getting progressively better at pattern-matching. Since so much of what we consider "intelligence" can be modeled as "good at pattern-matching", all kinds of intelligence-like properties may arise from improvements in machine pattern-matching.
It's "and then it will instantly kill everyone" bit that is ridiculous.
I think the issue is where you go from “so much is pattern-matching” to “therefore, everything may be pattern-matching”.
Well, yes, I don't believe we're anywhere near "AGI", only that there are many tasks machines will be good at.
Okay, you meant “all kinds of” idiomatically rather than literally. Do you basically agree with the OP’s point? There will be a range of things where AI is useful, but there will be many kinds of intelligence-like properties that don’t arise from improvements in machine pattern-matching?
Yes, pretty much.
Interesting point about “all kinds”. If I said “many kinds” it may have been clearer. I do think “all kinds” as not being “all inclusive”, compare to “every kind”.
Why do you think humans are better than chimpanzees at conducting scientific research?
It's worth pointing out that for the preponderance of the time that biologically modern humans have shared the planet with chimpanzees, humans did not have a striking lead in scientific research. So the answer to your question is probably not 'biology'.
We have had a striking lead in science and technology for the preponderance of the time we've shared the planet with chimpanzees.
Anatomically modern genetic humans (eg. last common ancestor of all humans) go back, say, 0.5mya... and human control of fire goes back 1-2 million years (https://en.wikipedia.org/wiki/Control_of_fire_by_early_humans). While chimp tool use tops out at 'sharpening a stick with their teeth' or 'taking a stone and bashing something with it': https://en.wikipedia.org/wiki/Tool_use_by_non-humans#Chimpanzees_and_bonobos and does not include even fire in their repertoire.
Human biology has changed over time.
Surely it has, but to the best of my understanding the advent of scientific research happened within the historical record, and doesn't happen closely in time with any attested change in human biology. A better brain wasn't the catalyst.
That the scientific revolution didn’t happen immediately after humans got Big Brains doesn’t mean that Big Brains weren’t necessary, if not sufficient.
It's certainly evidence that Big Brains weren't the bottleneck for science to happen. I don't think we've really interrogated the idea that more smarts = faster science. I atleast am not aware of any studies of the effectiveness of more-smart scientists vs less-smart scientists. It seems intuitively plausible that there's some floor of analytical ability to make any given discovery, but if you look at many notable scientists in their own context you'd have to conclude they weren't the smartest person alive at that time, and their success in science wasn't primarily attributable to being smarter than most other people.
The material context in which the investigator is conducting scientific research has a huge impact on their results, and just being smarter can't make up for that. To the point of the object question though, we seem to be handing over a pretty significant fraction of our civilization's capital to machines without them needing to super-persuade us, so it may not matter whether or not they can become way smarter than us.
A necessary ingredient, but several others were needed. Whereas the AGI boom people assume it covers everything:
Step 1: Superintelligence
Step 2: ???
Step 3: Utopia, or human extinction.
I don't think you understand what it means for the tails to come apart.
The tails coming apart doesn't mean that zero correlations exist between related skills, but that when you truncate skill at the high end of the distribution, there can be a strong anti-correlation. Because being the *best* at some high correlated substack does not mean that you are the *best* at something else you are also good/great at. Not that being good at thing things you use up some finite number of things that you are good at.
Also this kind of argument is invalid? Some programmers are arrogant and overestimate themselves, therefore this specific non programmer overestimates some abstract entity does not seem valid as logic.
Edit: I'm not even sure that the force of the argument is great either? It really seems like Boo IQ, boo arrogant programmers, therefore, boo argument that kinda looks like the above two.
I think that the tails coming apart in this context is talking about how building a computer that's superhumanly intelligent in some ways doesn't mean that it will be superhumanly intelligent in all ways. It could be superhumanly good at some things but still subhumanly stupid at others.
In fact we've already had this for decades. Computers have been superhumanly good at ostensibly-intelligent tasks like arithmetic for many decades. They've been superhumanly good at chess for about thirty years. They've just recently become superhumanly good at the task of writing reasonably cogent undergrad-level essays on any subject imaginable within seconds.
In the past, as computers have outperformed us on each successive task we've always just said "Oh well that's not real intelligence". But at some point we might have to bite the bullet of saying that humans and machines are intelligent in different ways.
So, I honestly agree with you, but to try to steelman:
It's like getting a Doctorate in Education. The point isn't that a D.Ed makes you better at math or whatever other specific substance of a curriculum, but that it makes you better at designing curricula in general. There's still a need to have access to actual areas of substance and those education materials, but it's a question of how you apply that knowledge. With the size and scale of data collection in the modern day, it's reasonable to think that a stronger focus on architecture/structure (curricula design) is the top priority.
The claim isn’t that a better AI at programming will generalize to being good at everything, but that it will accelerate AI research, making it much easier to create a general intelligence.
This is actually a dubious claim. It’s dubious because AI is far from perfect at writing software and because improving efficiency for A.I. is mostly about creating new algorithms, not implementing them. If A.I. has solved open questions in software theory I’m not aware of it.
It already speeds up algorithms research, by saving time for the researchers. Monitoring/reviewing code, doing literature reviews, and writing probably take up a huge fraction of each researcher’s time, and their ideation would be faster if they could test their ideas more easily. Beyond that, turning a text description of an idea into a usable algorithm feels like a task with a decent dataset and scoring system, and thus decently easy to train.
And building on this, that AI is credibly improving along many dimensions, eg persuasiveness, “bedside manner”, STEM, etc.
So you don’t have to posit transfer from programming to another skill, just extrapolate out the improvements directly measured in said skill. (Parent post gives the mechanism for why you should be worried about that growth accelerating.)
Isn't this just a second-order version of the same argument?
"Being good at making LLMs better at things is correlated with being good at everything. If we can make LLM-1 that's good at making LLMs better at things, then LLM-1 will also be good at building other kinds of AI systems!"? I don't know much about making LLMs better at things, but doesn't this seem very unlikely to be true outside of the very core of Mediocristan?
Maybe I'm missing nuance here, but it seems more plausible to me. We can already see that good AI research teams can make Starcraft AI, programming AI, image-generating AI, etc. The argument isn't that if you're good at programming you're good at ballet, it's that if you're good at programming you're good at making ballet robots.
Somebody fired 5 or so shots in the Harvard Square subway stop. They didn't hit anyone, fled, and as have now have not been found. I know people who are afraid this is the beginning of things falling apart. Sheesh!
They didn't think 2020 was worse? In my view that really was the beginning of the end.
They're afraid it's the beginning of a shooting war. Trumpers vs. Trump-haters? Immigrants vs. deportation authorities? Coke vs Pepsi? Seems clear to me we're nowhere near a nation-wide shooting war, but over the last few months I've heard several apparently sane people, one posting here, talking about buying the first gun of their life because of "you know, the way things are now." Just looked for more info about shooter online and there are a couple photos. He had a mask and a hood on, but fwiw it's clearly a young male, and white, or at least not Af-Am. Might be middle eastern.
People rioted for 3 months in 2020! Is that really any different?
People are too comfortable for an all-out civil war. Even poor people here have too much to lose. What will happen is the gradual withdrawal of the active oversight and goodwill that makes civil society possible, and that will result in inexorable social decay: rising crime, dysfunctional schools, failing infrastructure. Basically the second half of Atlas Shrugged only there won't be a magical mountain retreat where society still works. In my view there's a real likelihood that the US will be essentially indistinguishable from someplace like Brazil in 30 years.
I wonder if this is the median result of a society too happy for war sprinkled with media sources that want to make it think war is imminent.
I think it’s the result of identitarian “rights” culture being allowed to go too far. We let women have too large a political voice and their natural empathic instincts foolishly weaponized civil rights law against the productive plurality of the country. Probably something like this is inevitable in any pluralistic culture with too much low-IQ ethnic diversity. The dumb ones will always eventually organize to plunder the keep. They’re not smart enough to know what’s in their own long term self interest and so kill the goose that lays the golden eggs.
Speaking of college admission essays, the subject for the 2025 entrance exam to get into France's École Normale Supérieure is just two words in quotation marks: "Tu dois" ("You must").
The candidates had 6 hours to write their essay around this subject. Lest you think this is all part and parcel of the pseudointellectual hokum that the French (are reputed to) get up to, the École Normale Supérieure is France's most prestigious "grande école" and it has produced more Nobel laureates and Fields Medalists *per student* than any other institution of higher learning in the world.*
* Note: because of its hyper-selective nature and strong tradition in theoretical sciences and mathematics, it has produced more Nobelists and Fields Medalists per capita (per discipulus) than larger institutions like Oxford, Cambridge, Harvard, etc...
Do they really need 6 hours to write "...gather your party before venturing forth"?
https://www.youtube.com/watch?v=mXOpjgYMISw
Just thought I would give some numerical context to “hyper-selectiveness” and “per capita”.
There are about 700k people born in France in any given year. In math, about 40 people get into the “most prestigious” one of the ENS. I’d say this school admits between 250 and 400 people per year (the 250-ish come from the competitive exam; there are other ways to get in, which I know less about and account for the remainder).
As a comparison, about 1700 undergrads get into Stanford or Harvard every year, and about 6500 freshmen apparently enroll into UC Berkeley every year.
1Interesting. Thanks!
“Tu dois” can also be translated to “you have to”.
Great exam questions people I know encountered and answered:
French lit class: "L'amour"
Philosophy class: "Why is there something rather than nothing?"
Supposedly an Oxford admissions exam one year simply asked "Is this a question?"
About forty years ago, in a high school freshman class: “‘Traveling is rubbing and sanding your brains against others’. Comment and discuss.”
Wasn’t “l’amour” rather one of the year-long themes for the lit class in the scientific “classes préparatoires, with a slightly shorter – 4 hours – more focused and more conventional essay at the competitive exam?
(Along with quite a few 4-hour tests of English and actual scientific subjects?)
Philosophy, Reed College (1950s) never mentioned "L" during the class: 'Trace the course of Liberalism throughout history." (honor system, 3 hours, closed book) (according to my mother)
In other news, Hegseth's own hand-picked temporary head of public affairs is saying that Hegseth has to go. (https://www.politico.com/news/magazine/2025/04/20/pentagon-chaos-ullyot-hegseth-00205594)
It's kind of surreal how the piece is otherwise stuffed full of effusive praise for Trump and "if only Stalin knew of our plight" type arguments. The author is a hardcore Trumpist and Hegseth supporter, and he *still* says Hegseth has to go. Now that's something.
If you leak war plans in a group chat once, shame on you. If you leak war plans in a group chat twice...
Yup. In the spirit of "Once is chance, twice is coincidence, thrice is enemy action", if he does this a third time, does he get officially classified as a Houthi asset? ( Hopefully he won't be given a chance to do that... )
Yeah there's a huge qualitative step from one to two. But our Commander in Chief has all the confidence in his Defense Secretary, so there. He could post the plans on his Facebook page and still be ok.
Just as long as he isn't found to be friends with any Democrats on Facebook. If someone discovered a Democrat on his friends list, he'd be out faster than he could blink.
> Unfortunately, Hegseth’s team has developed a habit of spreading flat-out, easily debunked falsehoods anonymously about their colleagues on their way out the door
Well, it works for the boss.
The problem is that only Trump has the whatever-is-the-political-equivalent-of-a-reality-distortion-field. Trump says nonsense things as signals to other people that these are the new things to say to signal allegiance to him. Hegseth may have tried his best and doing that but it doesn't work.
I dunno. Do you think they might have picked the wrong alcoholic weekend Fox News guy to run the Defense Department?
At least his mother spoke highly of him.
She did speak about her opinion directly to her son
“I have no respect for any man that belittles, lies, cheats, sleeps around and uses women for his own power and ego. You are that man (and have been for years) and as your mother, it pains me and embarrasses me to say that, but it is the sad, sad truth.”
Yeah, that's what I was referring to. I was layin' on the sarcasm.
I got the reference. Just put up a snippet.
It's probably the MPP from Harvard that really did him in.
Interesting alternate to tariffs that would also reduce the trade deficit (whether or not that's desirable is another story), but do so less painfully for everyone involved.
Typically high trade deficits naturally come down over time because to spend your money on foreign goods you need to sell your currency, which depresses its value, which makes your own exports more competitive.
That doesn't happen in the US because other countries are constantly investing in US markets, which involves purchasing dollars, which negates the abo