So I'm fully aware that I shouldn't ask for a list of favorite podcasts, so I won't, but if anyone recalls a link to a list of favorite podcasts that's been done in the past I'd very much appreciate it.
> Symbolic regression similarly identifies relationships in complicated data sets, but it reports the findings in a format human researchers can understand: a short equation
This is what you call evil and anti-human? A new analysis method?
Then I've spent years trying to create new "evils" in my own field (software development).
That was meant to be a tongue-in-cheek reference to a tweet about dall e 2.
But well, i think I mentioned elsewhere that i'd rather live in a world without personal computers and that I am wary of raising kids in a world with internet, so if you are a software developer we probably have too different sensibilities.
Is this any more evil and anti-human than any other automation? Obviously people care more about stuff that directly impacts them but better to just admit that and not act like this is uniquely scary.
Is finding the equation that describes your data really the hard part of science? I would imagine the hard part is doing the experiment, or finding out what the equation means in a broader sense.
I know I know I'll stop, but it always funny to me that when we develop a magical technology that can save hundreds of thousands of man hours of labor a year and objectively improve the human experience:
The first though is always an anxiety ridden "But my job though :(".
Well, if that impact my job my human experience is not going to be objectively improved, isn't it?
Also, I am yet to be convinced that the technological advancements since the '90s have improved our life. Every time I think that i would want to have a kid, i conclude that a world where internet and smartphones exist is not one where I would want to raise a kid in.
And because of this i don't see science as a way to improve human life, but as a form of art. Yeah, I know it's a provincial concern, who cares, but also we keep improving the human condition in ways that are very easy to measure while at the same time we keep diluting those things that make life nice and meaningful, like art and science
Well in part that's because the parts of science and technology that are making tangible differences aren't visible. The journalism majors that report the news don't really understand them, and they aren't flashy and sexy and/or scary, so they just kind of rumble along in the background.
Teams of chemists and chemical engineers labor to tweak chemical syntheses, to make drugs, prosthetics, materials that work better, last longer, are more economical and less impactful on the environment to manufacture. But unless you're in the biz you won't read about a new and improved polymerization catalyst, or way to manufacture drug ingredient X that eliminates an expensive and toxic solvent in Step #66.
Mechanical engineers tweak car engines and robot manufacturing processes every year, to get more performance, better life, more efficiency in materials, both in manufacturing and use, and you'll never hear about it unless it's your profession and you're reading trade journals or going to conferences. Electrical engineers figure out how to build chips smaller and faster, or using better manufacturing processes. Civil engineers work on improving concrete -- still a work in progress after 2000 years -- or on better ways to build bridges and buildings. People improve the chemistry used to treat water and wastewater. Surgeons figure out how to reduce complications from surgery, radiation oncologists figure out how to target X-rays better so fewer cancer patients heave their guts out. And so on.
You pretty much only read in the broad news about the big, exciting, and/or scary stuff, but there is a giant engine of science, engineering, and technology that chugs along every day making things about infinite percent better than they were even within my lifetime, let alone a century or two.
What you say is clearly true, but at the same time it may also very well be the case that we have reached the point where the negative externalities of the big flashy "innovations" that are forced into our life outweight the positives of the slow technological advancement in other fields
I care little for better plastic when I cannot enjoy the products made with said plastic because internet gave me mental ilnesses. Oh well, at least nowadays i can get new fancy drugs to cope with modern society.
I don't believe that the industrial revolution has been a disaster for the human race. But I do believe that the digital revolution has. I made my limited share of simulations so I appreciate how important computers are in modern reaserch. But i'd rather live in a world where computers never got out of reaserch institutions
What do people think of the claims that omicron improves immunity for vaccinated and boosted people? I'm thinking about exposing myself by going to a probably crowded open air market without a mask.
I'm 69, vaccinated and triple boosted (Pfizer). I have type 2 diabetes, controlled. I'm low-end obese or possibly high-end overweight. (Correction: it was actually two boosters.)
I probably had COVID early on (messed with my sense of taste, but it recovered after about 6 months).
I don't know whether I've already had omicron. I live in southeast Philadelphia, not one of the highest risk zip codes.
I can see wanting to get it just because one gets so sick of feeling *stalked* by the freakin virus. I haven't caught it yet, and every time I have a few tiny cold symptoms I think I have it and feel sort of relieved, like OK, let's get it over with. Katelyn Jetelina's blog is good, and she discusses this sort of thing, i.e. immunity from infection vs. immunity from vaccines, but I can't remember if she addresses your specific question.
Personally, I am afraid of long covid. I think a lot of what's in the press is hysterical and overblown, and that a lot of the people who get diagnosed with long covid are just having a slow recovery from covid, or are suffering from the baseline malaise of everyday life familiar to covid graduates and covid virgins alike: fatigue, vague malaise, difficulty motivating oneself to concentrate . . . However, it does seem like a fraction of those people have something real and unusual going on. I myself suffered from what I am pretty sure was a post-flu syndrome for several years after a mildish case of the flu about 20 years ago. Worst symptom was that I was exhausted and craved a nap every single second every day, no matter how much I slept. In addition I had muscle aches all the time, and several joints so sore I gasped when I even just brushed them against something by accident. Also sore throats that came and went, and temp of 99.5 came that came and went. Internist tested for everything she or I could think of, and all tests gave no clues about what was up. Meanwhile I had a small child and a full-time job. I NEVER want to go through something like that again. It went away after about 3 years, but not before ruining those 3 years of life.
I assume the idea is to get a less dangerous covid variant in order to protect from more dangerous variants? You might not get the variant you're aiming for, even it's the most common one.
It's due to a thing called "affinity maturation" which takes six months and shouldn't be disturbed... Sorry, I can't find an English-language source... does anybody know more about this?
It is unfortunate that this thread is 4 days old. I'd love to hear thoughts on the Oregon 6th district race, especially from Scott himself. Carrick really generated a shitstorm in Oregon for a variety of reasons. "Not political", aka doesn't vote, cryptobro cash, not community oriented, etc. Shouldn't "effective" altruists be smarter than dumping a nobody into a race with crpyto cash?
Yeah this was a huge misstep. This primary flew under the radar so he could win purely on cash advantage. But in a general election funded well on both sides, he’s a really bad candidate. The ads just write themselves. “Doesn’t represent the interest of our district”, “bought by a billionaire from out of state”, “part of a fringe online movement”, “more interested in fanciful AI overloads than the real problems facing our nation right now”.
I feel like this race goes to lean Republican if he wins. And this shouldn’t require saying but I’m not arguing that the above statements are true, but they are highly effective.
Zvi has a post a few months ago where he was doing grant assessments for an EA organization, and a lot of the grants were of the form of
1. increase the status of our group
2. ???
3. good things happen!
and that has lots of problems.
Maybe Scott had really good reasons besides the paragraph that he gave here, but it pattern-matches to that. (And it probably does the opposite of step 1 and lowered the status of EA. It's probably easier to convince junior politicians to do EA than to take an EA person and make them politically savvy.)
> It's probably easier to convince junior politicians to do EA than to take an EA person and make them politically savvy.
Definitely want to signal boost this.
Our system of laws and regulations is unbelievably complex, which leaves many very smart people without a very robust model of how changes actually happen. No slight to them, it's just a crazy system. The scale of it doesn't fit in your head, but it's a system to solve for the preferences of a few hundred million people all at once. None of us can even fully comprehend the complexity of that problem, so naturally the equation we're fitting to that has a gaggle of dimensions.
Imagine all these interest groups scattered around the country that want more funding for law libraries, or harsher penalties for drunk driving, or whatever. They hire an expert in whichever legislature or administrative agency they're trying to persuade. Often, the best way to do this is to draft a few short canned sentences to tuck into a larger bill when it comes to the floor, since any given issue will only be addressed once every few years (maybe once per session on hot issues). You definitely don't have enough money or clout to control the docket. When this does happen, you'll need to scramble to persuade busy legislators to act really quickly, so you'll need to anticipate everything that needs to get done and help them do it.
Lobbying often means patiently sitting on a bench at a station--then suddenly trying to glue a few words onto a passing train.
So for each dollar, you can spend it on:
- getting a politician elected
- influencing existing reps
- convincing more citizens to care
- influencing regulatory agencies
- direct spending on outcomes, bypassing government entirely
I suspect electioneering is vastly oversubscribed. If the candidate is popular enough to win, you're at best getting a time share with a lot of other issues. If your issue was popular enough to be decisive, you'll get the benefit even without campaign contributions. The only thing that makes me question this is EMH, if so many people are buying it, it is probably worth something. But this thing is worth less the more people pay for it, so... I don't know if EMH even works for something with that property.
For niche issues your marginal dollar is probably best spent on awareness. A slogan for some issue that boils down your position to a nonexpert in maybe five words. "Atomkraft nein danke" as a notorious one. What's the "Atomkraft" of EA?
Working at lower levels of government can be a cheap test case. Can you get anything passed as a state law? Can you get somebody elected to a local school board? If not, this hill might be steeper than you expect.
Also, letter writing campaigns. If you want something passed, get a thousand people to write letters saying so. If you have well-founded compelling opinions, maybe just focus on communicating those to decisionmakers until they hear you.
It will take a few tries, other people are constantly trying to message them too.
If they never hear you though, reexamine your priors.
Nate Silver announced last week he was eliminating out-of-state cash from his model because it isn't predictive. I'm guessing that in a general election cash raised is mostly a proxy for the support that candidate has and the cash itself matters little beyond that. So, if the cash is from out of state, it is worthless because those supporters can't vote in the election.
I think the answer is clearly yes. If not, then we should expect the energy crisis to continue indefinitely, because of market expectations for low energy output will keep down investment in all energy extraction methods.
I will happily* wager a 4-figure amount of dollars against that outcome, for exactly the reasons Carl Pham lists in the next comment. Unless you're actually kidding/trolling, which seems possible, in which case never mind.
(* okay not actually _happily_, since I favor expanded investment in nuclear energy generation. But a winning bet as easy as that one would be impossible to resist.)
Of course it won't. Voters are exceedingly reluctant to deregulate even things that are pretty harmless -- like interstate telephony and airline travel -- if they think the field is, or could be, dominated by a few big operators, who may make decisions not in the public's interest. And who thinks public utilities put the public interest first? Essentially nobody. Add to that the potential for danger, real or imagined, and deregulation of nuclear power (which would have to come on top of deregulation of electric utilities) is a nonstarter in every polity of which I can think.
One problem with applying EA to influencing politics is that a lot of political issues are ones where people can and do honestly disagree about what policies are good. In the case of Covid, it sounds as though the EA people see the right approach as greatly increasing federal funding on preparedness. I don't know if that works or not, and would be much more concerned with reductions in regulatory barriers — producing a workable vaccine took less than a week, getting it approved took most of a year.
For a more extreme disagreement, lots of people take it for granted that reducing climate change is an important objective. My view, as some here know, is that it is not clear whether the net effects of climate change are good or bad and there is little reason to think they are very bad. To take an older example of a similar controversy, the EA approach fifty years ago would probably have led to pushing ways of reducing population growth. It now looks as though, at least in the developed world, that would have been the opposite of the right direction.
You can use experiments and statistics to try to estimate the effect of bed nets on mortality, but it's a little harder to do the equivalent in cases like those.
What exactly does "political" mean here? As far as I can tell, it just amounts to "any issue that affects people's lives where there's any significant disagreement about what actions should be taken." The problem is that almost *anything* that has a significant impact on the world will be controversial for one group or another, and thus political. Improvements that are uncontroversial and unambiguously positive but still have a major impact (like mosquito nets) are few and far between, they're the exception rather than the rule. For EA to remain strictly apolitical would massively narrow the scope of what they could focus their efforts on.
As an example, many people think that giving out condoms to poor Africans is a good thing, both because it helps to mitigate overpopulation (which is definitely still a problem in Sub-Saharan Africa, even if it's no longer an issue in developed nations) and because it helps prevent the spread of HIV and other sexually transmitted diseases (which are a huge cause of suffering and death in that region of the world). Yet some people object to the use of condoms on strict deontological religious grounds, and would oppose such a measure on principle even if you could prove that it was a net good from a consequentialist perspective. Should EA reject efforts to distribute condoms in poor parts of Africa because the subject is too political?
So is your critique of EA that being *effective* is not actually desirable, since too much effectiveness could lead us too fast in the wrong direction?
No. My critique of applying the EA approach to politics is that it is hard to know which direction to push in, so you might end up doing net harm instead of net good. The reason people in Africa don't have bed nets is that they can't afford them. The reason people in America don't have gun control/national health care/higher minimum wage is that a lot of people think it is a bad idea — and they might be right.
Sure, but isn't that a fully general argument against any sort of political implication? Presumably you still want people to enter politics for "altruistic" reasons, i.e. to improve the world.
EA basically says: let's think very carefully about what direction we push in, and once we've decided, let's push very hard. If the first part is flawed, then the "effectiveness" of the movement is a bad thing, since it'll push hard in potentially wrong directions. So that would be an argument specifically against pushing hard.
I think I mostly agree with this view, but it's also somewhat weird to argue for. One could *always* be wrong, especially in politics. EAs at least put a lot of energy in trying not to be.
So you're saying a good Bayesian should assign a nonzero but very small number to the probability that another Bayesian assigns a zero probability to something?
Exploring evolutionary psychology using language models. This post looks at the difference between word vector representations of Kin Altruism and Reciprocal Altruism. As one would expect, reciprocal altruism is related to conscientiousness while kin altruism is slightly more related to emotional attachment.
And all this is done without surveying a single psychology undergraduate :p
> Over time, I’ve realized that my default mode is to tune into what’s going on for other people—what they might be feeling, what’s going on with group dynamics, and what I would do in their shoes. Cycling through other people’s perspectives is automatic for me: it runs in a background thread of my mind and I can’t pause it without conscious effort.
I found this statement really interesting, because it is the exact opposite of my own experience. I guess it's another reminder of just how different everyone is.
The FDA has rejected the EUA request for fluvoxamine. I've only read the article[0] and the summary in [1], not all the details, but they sound like legitimate, if perhaps thin, reasons for rejection.
Unfortunately, he's a Democrat. And given how much Democrats are working to make my life worse off, I can't support him. Even if he was "one of the good ones", getting elected would shift power towards the party.
Find me a candidate who won't caucus with the Democrats and I'll reconsider.
I read some of Carrick Flynn's advertising. During primaries the candidates are going for the votes of their party members, so both left and right are more extreme, and they move to the center for the general election. With this in mind, Flynn's advertising is noticeably less extreme when compared to his opponents.
For example, in the material I read, he states that two "500 year" floods ravaged his home town of Vernonia 8 years apart, thereby implying, rather than stating outright, that climate change is to blame. That may sound like a small thing, but in western Oregon that's a relatively conservative position for a Democrat.
One thing I noticed in his TV ads is he wishes his mother had access to abortion. I'm not sure of her circumstances, but does this mean he has siblings he wishes had been aborted? Or possibly himself?
"One thing I noticed in his TV ads is he wishes his mother had access to abortion. I'm not sure of her circumstances, but does this mean he has siblings he wishes had been aborted? Or possibly himself?"
Reading his potted bio on his campaign website, seemingly his family was poor and dysfunctional and he ended up parenting his siblings. So, like a good Democrat, his solution to poverty is kill poor babies before they can be born (okay, that's sarcastic and possibly even mean, but honestly - yes, maybe things would have been better if your parents never had kids, but 'gee I wish Mom had aborted little Pete and little Susie' is not the best way to refer to family, surely?)
"Carrick grew up in Vernonia, a small logging town in the Coastal Range. He had a tough start to life. His family was poor, broken, and after a flood destroyed their house, homeless. While this made him tough, it never made him hard. He took responsibility to shelter his little brother as best he could and up into his teenage years wanted to become a pastor to help others. This combination of strength and kindness remains at the core of his character today."
"yes, maybe things would have been better if your parents never had kids, but 'gee I wish Mom had aborted little Pete and little Susie' is not the best way to refer to family, surely?"
If he'd instead argued "I wish my mother had gotten my tubes tied, we need to make vasectomies and tubal legation surgeries more accessible," or even the socially conservative argument of "I wish my mother had stopped having sex, we need to do a better job teaching abstinence to poor young people," would that have really been better? Regardless of the specific method he'd be endorsing and the associated policy proposals, they would all boil down to him implicitly wishing that his younger siblings hadn't been born.
Literally everyone uses this excuse to be partisan. I am revolted by a great deal of the things that the Democratic party is doing, but if you vote for a Republican your voting for someone who will caucus for a party whose leaders all either wont condemn or actively support what happened on January 6. That's not a small thing.
The point is not that the GOP is worse or better, the point is that no-one is free from the responsibility of actually assessing which candidate they think is better for whatever position and then voting for them. IMO, in general, you should support a competent moderate over an idealogue.
Being revolted by BLM (which I am) does not mean I am not also revolted by January 6 and the attendant conspiracy mongering. If you are choosing one over the other, you are just admitting that you think *your media cult is better than the other one. You are just admitting that you don't really want to be a participant in representative democracy, you want a fun drama with a grand narrative.
And its not a non-sequitur, it clearly follows and fits with what I'm trying to say, your response even makes that clear.
I'm losing my stomach for voting for any Democrat or Republican. We need a new party which denounces the excesses of those old, evil Two parties which continue to hold us hostage.
How about a party which is against the Surveillance State, for instance???
I wouldn't start a new party in any place with First Past The Post voting: no chance of victory. But then again, the US does have especially bad parties, which I guess might help an "alt middle" party in certain specific "purple" locations where both the R and D aren't very well liked. [Edit: It would be extremely tough though, and remember that nowhere in America is there a majority of people who would appreciate ACX, so like, probably none of us should start a party.]
If you are concerned about surveillance, you will find candidates at many levels of government on both sides of the isle who are concerned as well.
Or maybe you're just too good for either party. Maybe you're just too principled, too pure, too *independent.* My heart goes out to you if that's true Jack Wilson who cares so much about privacy, it must be a very hard way to live.
Get someone to run as a single-issue candidate against the Surveillance State, no other platform. Try to get a few percentage points, enough to cover the margin of victory in the given race, to encourage the main parties to take your position.
Would there be a way to make a betting market for financial fraud? Resolutions would be something like "will SEC open an investigation?" (is that public?).
A lot of the data is semi-public if you have know-how and will to sleuth. Would be good to incentivize the sleuthing and public posting of documents + bets.
Yeah, I wonder if limiting the amount any trader can have riding on a bet solves that. If only a fraction of the money can ever be insider money then less risk. As always, need a big market to be functional
This meteorite fragmented. It was seen by dozens of people. Many rocks were recovered, which were geologically distinct from the rocks of the surrounding area, and similar to some other claimed meteorites. The evidence was collected and published by a respectable scientist, Jean-Baptiste Biot, of the French Academy of Sciences.
Meteorites are an a priori dubious hypothesis -- our familiarity with them has hidden that fact, which would've been obvious to earlier scholars. From where would rocks come that fall from the sky? From the Earth itself? That would require events of unimaginable energy, to fling stones so high that they came to the ground not within a thousand miles of any identifiable energetic event (e.g. volcano).
From the heavens, then? But already in 1800 people generally thought the Earth to be quite old, at least millions of years, and how could objects in the heavens be orbiting in stately regular orbits for millions (or longer) of years -- and then those orbits suddenly, after millions of years, intersect the Earth? That's asking for a very peculiar orbit, a great deal of luck, or random novel energetic events in the heavens (that would change well-established orbits) that had not yet been observed. So far as anyone could tell, nothing changed in the sky -- even comets after Halley were thought to be regular in their habits.
I suggest the existence of meteorites actually illustrates a subtle and profound principle of classical mechanics, id est the existence of chaotic dynamics in even relatively few-body systems. That celestial orbits can suddenly hit the Earth after 4.5 billion years of avoiding it is the signpost of dynamic chaos, the fact that even after such a very long time orbits can turn out to be not closed.
The Professor on Gilligan's Island seemed to know about rogue waves when the surfer showed up and then road "reverse tsunami" back home. Big man on little stick episode 1965
It bugged me that he never replied, so I looked it up. Gilligan didn't do anything, the Professor told the surfer who showed up that he'd be crazy to try to surf back to Hawaii, but by the end of the episode he tries it anyway. He makes it back to Hawaii but hits his head on a rock and gets amnesia, so he never tells anybody to go rescue everyone back on the island.
Is there any way to dabble in neurotech by buying a cheap brain computer interface and playing around with the data obtained from it ? That is to say is there a cheap reliable interface can be hooked up to a cheap eeg or something similar ?
Yeah, you can pick up brain interfaces that use skull electrodes mounted to a rubber hat for a few hundred dollars. They normally come with some conducting lube to make a decent electrical connection to your scalp. It helps if you aren't too hairy.
From a cold, amoral economic perspective, is slavery bad? Bad for the economy? I've read some pieces over the years arguing slavery was bad for the US South. Something about free labor being more efficient. But I'm never sure how divorced from politics such arguments are.
Perhaps the concept of AGI gives us a different perspective on the issue. Assume that somehow we know that an AGI has no qualia and can't suffer. Such an AGI would blur the line between human capital and other capital investments. Much like how under a slavery regime such investments would be blurred.
It seems to me that one can't argue that slavery is bad for the economy if total ownership of AGI's is not.
Yes, slavery was unambiguously bad for the US economy. Even ignoring the gargantuan cost, human and economic, of the civil war directly and its long-term impacts, slavery delayed industrialization of the south. Without slaves, wealthy southerners would have been more likely to invest in machinery, railroads etc, things that actually lead to compounding growth. Slavery is an economic dead end.
You need a fairly complex story to explain why the north was so much wealthier than the south even before the civil war, and why they had a more valuable agricultural industry. And why nowhere else on earth that used slaves had anything remotely resembling the economic growth of the US, like Brazil.
Slavery prevents the development of the thing that actually makes a society wealthy, individual capacity. Massive amounts of money and time and talent have to be spent turning one portion of the population against the other, to prevent uprisings, and generally crush spirits on the scale that is necessary to keep a population of people enslaved.
As for comparing it to AGI's I have no idea what the connection you're trying to make is there.
Slavery re-introduces many of the coordination problems that planned economies suffer from. This exists at the level of the manor instead of the nation, which is better. But other problems get worse: finding aptitudes amongst slaves or the best places to invest in non-slave capital become harder.
There is also providing for the needs of the slave. A slave receiving goods from a master is going to get a worse bundle of goods than if they could have spent the money themselves. Plus the monetary amount they'll receive a lot less than what they would have gotten in a market economy. A good definition of "good for the economy" would count this as being bad for the economy. Most arguments for slavery being a net benefit conveniently hide the material needs of the slave and hope you don't notice. Slaves drag the average income level down significantly, so slavery advocates want you to ignore them. They want slaves to count to the population counts for allocating electoral college seat calculations but not for GDP per capita calculations.
>There is also providing for the needs of the slave. A slave receiving goods from a master is going to get a worse bundle of goods than if they could have spent the money themselves. Plus the monetary amount they'll receive a lot less than what they would have gotten in a market economy.
Where's your evidence?
Black slaves were taller than free white men during southern US slavery, but black american men are slightly shorter on average than white american men today (and non-american blacks are much shorter than white men). This indicates that slaves were actually better nourished than free white men. I don't know why people think free white workers in 18th and early 19th southern US experienced anything besides poverty.
>Most arguments for slavery being a net benefit conveniently hide the material needs of the slave and hope you don't notice.
Slaves were unambiguously better off materially than they would have otherwise been (i.e. if they had remained slaves in Africa or had been slaves in Brazil/Arabia instead). And again, I have no idea in the world why you think unskilled free workers in the old american south were well paid, but they weren't. All things considered, they were probably materially no better off on average than slaves were, and may have actually had less access to food.
> Slaves drag the average income level down significantly, so slavery advocates want you to ignore them.
Again, this is all just assumptions. And I mean, who the HELL is a "slavery advocate"? Literally nobody is arguing for slavery, and in fact the overwhelming majority of people I've ever seen arguing that slavery was economically beneficial were left-wingers who are furiously anti-slavery. They desperately want slavery to have been beneficial because it feeds into narratives around reparations and how white people owe black people for their prosperity (despite this being unambiguously false).
Just because they were taller? Geez. First of all, Google's telling me that slaves were *shorter* than whites[1]. Second, there's a lot more to being well-off than having enough nutrients. Ask any lab rat or, you know, slave (because slavery is still a thing).
[1] "adult slaves, Steckel found, were nearly as tall as free whites, and three to five inches taller than the average Africans of the time. [...] Although the adult slaves were clearly well fed, the children were extremely small and malnourished" https://www.newyorker.com/magazine/2004/04/05/the-height-gap
"Black slaves were taller than free white men during southern US slavery, but black american men are slightly shorter on average than white american men today (and non-american blacks are much shorter than white men)"
Do you have a single (reputable) source for any of these rather incredible claims? I would say that people of African descent are in general larger than the average white person, and non-American blacks are absolutely not 'shorter than white men'
Not just the needs of the slave, there's also costs of keeping slaves enslaves and protecting owner from getting attacked by slaves.
I've seen a claim that the profitability of slavery in the south was overestimated because the costs of slave-catching and protection against slave revolts were shuffled off to the governments rather than paid directly by slaveholders.
That doesn't sound like a good argument, because the tax burden would've fallen almost exclusively on slaveholders in the first place -- they were the only people who would be importing finished goods, the way the government got most of its income in those days. So they were still paying -- just indirectly, via taxes.
From a purely economic perspective, yes I think it's bad. To make it interesting let us assume arguendo that we have very enlightened masters, who take good care of their slaves, don't overwork or mistreat them, feed and house them as good or better than they would if they were free laborers and paid for their housing, food, medical care et cetera out of their own wages. We'll also assume the masters take care of healthcare, family support, and old-age support in at least as good a way as imposing pension taxes on the wages of free laborers and then giving them a public pension and/or Medicare would do. And finally we'll have to assume we have happy ideal-communist slaves, who work for the pure joy of doing the job well, and not for mere filthy lucre, and do as good a job as they would if they were being paid a wage and had the prospects of a raise or being fired to motive them.
We still have the problem that the distribution of labor -- who is assigned to what job and what career -- is decided by the masters, not the slaves, and the exact form of healthcare, family care, educational support, and old age support is also decided at that level, and the slaves have no input.
But that is certainly less economically efficient, for the same reason centrally-planned versions of these things always prove less efficient, because it means the people who make the allocation decisions are less than optimally informed about the individual circumstances. In a free labor market, each individual "allocates" himself to a job or career, and he has maximal information about his own skills, wishes, utility functions, et cetera, so given the right price signals as to the value of his labor to others, he can make the optimal choice about where he should work. No central committee or feudal lord could possibly make as good a choice -- because of the lack of detailed individual information. Which means, to achieve the same level of social satisfaction, more economic resources must be spent.
Same thing with appropriate healthcare coverage, appropriate family support, eductional support, appropriate old-age support. The individual is best suited to weigh the pros and cons of various options and choose the best option for himself, and no Federal bureaucracy or oligarchy of slaveowners could choose as well -- so that, again, for equivalent levels of social satisfaction, a larger amount of economic resources must be spent.
Alexis de Tocqueville gives an account of this in his classic book Democracy in America. He describes a journey down the Ohio River, which divided the free state Ohio from the slave state Kentucky: “Upon both banks of the Ohio, the character of the inhabitants is enterprising and energetic; but this vigor is very differently exercised in the two States. The white inhabitant of Ohio, who is obliged to subsist by his own exertions, regards temporal prosperity as the principal aim of his existence; and as the country which he occupies presents inexhaustible resources to his industry and ever-varying lures to his activity, his acquisitive ardor surpasses the ordinary limits of human cupidity: he is tormented by the desire of wealth, and he boldly enters upon every path which fortune opens to him; he becomes a sailor, a pioneer, an artisan, or a laborer with the same indifference, and he supports, with equal constancy, the fatigues and the dangers incidental to these various professions; the resources of his intelligence are astonishing, and his avidity in the pursuit of gain amounts to a species of heroism.
But the Kentuckian scorns not only labor, but all the undertakings which labor promotes; as he lives in an idle independence, his tastes are those of an idle man; money loses a portion of its value in his eyes; he covets wealth much less than pleasure and excitement; and the energy which his neighbor devotes to gain, turns with him to a passionate love of field sports and military exercises; he delights in violent bodily exertion, he is familiar with the use of arms, and is accustomed from a very early age to expose his life in single combat. Thus slavery not only prevents the whites from becoming opulent, but even from desiring to become so.
As the same causes have been continually producing opposite effects for the last two centuries in the British colonies of North America, they have established a very striking difference between the commercial capacity of the inhabitants of the South and those of the North. At the present day it is only the Northern States which are in possession of shipping, manufactures, railroads, and canals.”
I wonder if there's a way to get more data on this hypothesis. I know that white southerners are often described as lazy (Scott mentions this fact in his review of Albion's seed) but one major confounder is the existence of malaria. Malaria-carrying mosquitos thrive above a certain average temperature and die below it, creating 2 lines of latitude in the Western Hemisphere. Between them malaria is common, outside of them it is rare; in North America, this line is essentially the difference between slave and free states because Africans are much more resistant to malaria. And one of the major symptoms of malaria is... fatigue and a lack of desire to work. So I would want to know exactly how close the Ohioans and Kentuckians he is comparing lived to each other, and in particular if they might have had similar rates of malaria.
Well, a major point of his comparison was that that they lived literally on opposite banks of the same river. Which makes it difficult to attribute their differences to anything other than culture.
Right, but I'm not 100% sure how literally to take that. How far inland did he go/look? If he is being really literal then you're right. Although I also looked at a map and for some stretches the river runs North/South, so even if he did look inland a ways the difference couldn't be malaria.
The other thing is, do we have data to back up de Tocqueville's claims? Or is it just his summary/impression?
What is "the economy" and why should we care about it? In my opinion we care about it because represents the ability of people to do and consume things they want. In economist-speak, it allows them to satisfy their preferences. Obviously slaves can't do that, so slavery is by definition bad for the economy.
But also, it does seem like slavery, by providing undercosted unskilled labor, discouraged investment in industry or in improving worker skill. The North had much more industry than the South in the Civil War, and this was likely at least in part a result of slavery itself.
I don't follow your comparison to AI so I can't answer your last question.
"What is "the economy" and why should we care about it? In my opinion we care about it because represents the ability of people to do and consume things they want. In economist-speak, it allows them to satisfy their preferences. Obviously slaves can't do that, so slavery is by definition bad for the economy."
It really should not be that difficult to understand what is meant: Does slavery improve the per capita production of an economy or not (both at the time and over the long run)?
It's also really funny for you to act like you're being anti-slavery by using your definition, because most people who speak endlessly about slavery today are the ones desperately pushing the narrative that slavery is why America is wealthy (because that means that white people "owe" black people today).
Some individual got rich because of slavery; I think "America as a whole got rich because of slavery" is very wrong. There are multiple reasons to believe this--our economic growth post-Civil War has been quite high, much of the accumulated wealth of slavery was destroyed during the CW, the richest regions today are almost all in former free states while the poorest are in slave states (even looking just at whites), industrialization happened much more quickly in the North, etc.
But also, I think it's kind of an irrelevant question to ask. The productive capacity of the economy is not important if a large portion of people are legally barred from getting any benefit from it. It would be like asking if an invasive war is beneficial, while ignoring the dead soldiers, destroyed wealth, etc. and only looking at whether the government of the invading country collected more total taxes or something.
It's not hard to figure out why the south didn't industrialize and the north did. One of the biggest pieces of damage that slavery did to the south was cultural, they couldn't see any way to power other than through keeping people in bondage.
I imagine that part of the reason that slaves make for an unproductive workforce is the fact that they really don't want to be working for you and are going to do the bare minimum to avoid getting beaten. Not to mention the effort you have to invest in overseers, catching runaway slaves, putting down rebellions, etc.
If we make an AI that absolutely loves working in a paperclip factory and thinks it's living its best life, most of those problems wouldn't apply.
Under most circumstances, free labour is a lot more efficient than slave labour. The one possible exception is if you have a huge swath of land and there's not enough people living nearby to work it for you -- under those circumstances it might make sense to buy slaves instead of waiting for the local population to grow.
You can incentivise a slave more than a free person by offering the slave his freedom if he earns X dollars for you. But for cultural reasons this wasn't done very much in the US South.
You also get to sell or otherwise exploit their children for profit. And some people may find intangible benefits in owning slaves. Plus possibly there was some irrational exuberance in pricing at work, a la southseaturnip.com
AGI power vs. human power, types of labor AGI vs. slaves are used for, and upkeep costs for AGI vs. slaves. You don't have slaves designing the cures for cancers, you have them picking produce.
Also, of course, the AGI doesn't have qualia and can't suffer, while humans do have qualia and do suffer unless you're some kind of absolute nihilist or a very bizarre form of racist who thinks p-zombies exist, but only along ethnic lines.
There can be the case where consciousness and qualia are required for AGI. Or at least possible, and we programmed AGI to have qualia for whatever reason.
Once I've assumed the things you stipulate, you've rather defeated the point of invoking AGI, because these aren't things we're going to develop AGI for. You might as well say "clone homunculi" or "Zaplefrogians, who are identical to humans in every other aspect but metaphysically don't have any moral signifiers attached to them so you can treat them however you like."
My answer to that question, then, is that this is about as meaningful of a thought experiment as speculating about what time-traveling bullets would signify for the morality of abortion.
There's little evidence that slave labor had a higher ROI than other things in the old south, or that slaves were cheaper (to buy and maintain) than using free labor or were more productive.
I use a DokuWiki install on my server. It's kind of outdated as applications go, but works decently on my laptop and phone and has many versatile plugins.
Mostly I write about what I did over the course of a day (my autobiographical memory is somewhat broken, so this is helpful) and whatever problems/things I happen to be considering around when I'm writing it. I'm looking at adding machine-readable data so I can find interesting trends.
I write whatever I feel like, which is usually just whatever I did or any notable occurrences.
I tend to write a *lot*. I've apparently written 104 pages in the last three months! One night recently, I spent over an _hour_ writing, though that was unusually high.
Note that typing is way better than handwriting. For the first decade or so, I handwrote everything, which is really slow and my hand would cramp if I wrote even a fraction of the stuff that I type about nowadays. Also, having your journal be searchable comes in handy sometimes when you want to find out when something happened.
I'm following the Ukraine situation closely, both from a military and "Putin's personality" standpoint, and I find the attacking-NATO scenario very unlikely (Russia doesn't have the resources to fight on a second front, at least not without spending months or years mobilizing first).
But if this unlikely event actually occurs... it might involve tactical nukes. I expect a sudden stock market drop, which you could exploit by having lots of cash on hand to buy stocks, but in the nuclear case you might want emergency supplies instead. So, beat the rush, buy a bunch of emergency supplies early, put some cash in a safe, and when Russia attacks, buy an index fund at reduced price in the hope that things go back to normal? But maybe you don't want to hold a lot of cash. So to prepare for the conventional war case, maybe look at what stocks went up and down when Russia invaded Ukraine, and assume that a similar pattern will occur again if Russia attacks NATO.
They are having a sseriously hard time in Ukraine and have suffered severe losses in men and materiale. Why on earth would they take on the entirety of NATO?
If by "campaign" you mean "military invasion", then there's no place in the EU that Russia can plausibly invade this year without either being defeated in a conventional war so decisive that the markets won't need to adjust much beyond what they already have for Ukraine, or waging nuclear war. And nuclear war between Russia and NATO will probably cause sufficient disruption that stock markets won't operate and any stocks you do own will be worthless for the duration.
In which case, the usual - gold, canned beans, and 5.56x45mm cartridges in bulk.
If you're worried about Russia using the Ukraine fiasco as impetus to rebuild their army into something that could at least imagine victory over NATO, then that will take several years and it would probably be premature to buy before we see exactly how Ukraine turns out.
Seriously, three-quarters of the Russian army has been fed into a meatgrinder, and the Ukrainians are cranking the handle most vigorously. And to at least some extent vice versa, but whatever happens to Ukraine, Russia's not invading anyone else any time soon. Nuking, if they really want to, but not invading.
>If you're worried about Russia using the Ukraine fiasco as impetus to rebuild their army into something that could at least imagine victory over NATO, then that will take several years and it would probably be premature to buy before we see exactly how Ukraine turns out.
Note that NATO spends a bit over a trillion on the military per year, and even PPP Russia only has 4.3T GDP. The PRC could match NATO's spending, but Russia just doesn't have the economy to support it.
What's your info source on the "3/4 Russian army dead" estimate, btw? I'm aware you're probably simplifying the stats for the sake of argument and I'm fine with that, I'm more interested in finding reliable sources.
I think he's saying that 3/4 of the army was "fed into the meatgrinder", not that they're dead. Presumably only a small fraction of the units (and material) are dead, but that's all it takes to result in combat ineffectiveness.
Numerous credible sources including the Institute for the Study of War and the UK Ministry of Defense report that Russian forces committed to the war in Ukraine represent ~75% of their total combat strength, and we've seen them e.g. bring in Naval Infantry from the Northern and possibly Pacific Fleets, so they do seem to be going all-in except for a bare minimum held back for internal security and guarding their other borders.
Equipment losses, most of them photographically confirmed because this is a smartphone war, suggest 30+% losses for the forces committed to the invasion, or ~25% for the Russian army as a whole. Manpower losses will be somewhat less than that, but still substantial - probably over twenty thousand killed, total including wounded, captured, deserted, and psychiatric casualties likely over 100K by now.
As beelester military units generally become combat-ineffective at ~20% casualties due to loss of cohesion. There are some exceptions to this, but none that really apply to Russia at this point. And it can't be quickly fixed by shoveling in replacements, because the veterans won't trust the newbies or vice versa.
Note that the Ukrainian army has also suffered heavily in this war. I *think* proportionately less than the Russian, but it's hard to be sure.
The initial invasion force was estimated at about 75% of their standing army, so that would suggest about 25% of their standing army is casualties. Which is still a pretty huge loss.
(Slightly confounded by the fact that the separatist forces are also fighting and I'm not sure if they're included in this fraction.)
I believe the rule of thumb is that 10% losses will temporarily knock out a unit, and 30% is enough to make it combat ineffective.
That is a very good point, and one of the interesting things illuminated by the course of the Russo-Ukrainian war is how even Putin doesn't believe his own bullshit about NATO posing an offensive threat to Russia. If he really thought it did, it would be the height of folly to permit the level of consumption of Russia's military resources that is now going on. At this point, Poland would probably have a 50/50 chance of making it to Moscow, barring the use of nukes.
Consumer staples, energy, commodities, the usual defensive buys. But you're too late, everyone else already thought of that. The time to buy any of those was last year when NFLX and ZM were riding high. If you did that in some cases you tripled your money by this spring.
I’d imagine if you have to ask on substack, then you probably can’t make much money this way (see https://en.wikipedia.org/wiki/Efficient-market_hypothesis) but if, for whatever reason you think markets underestimate the chance of war between Russia and the EU, then Ig defense, tech, and energy stocks would be the way to go, and airlines, tourism, companies heavily invested abroad, particularly in Europe, would be the thing to short.
With that said, if you Think Russia and the US are really going to go to war, then your strategy is simple— move your assets into Yuan
Moving your assets into RMB looks nice on paper, but the PRC has notoriously-weak rule of law and likes Altering the Deal so you'd just get expropriated.
EDIT: Also, if the USA is getting nuked to shit anyway, they're probably going to nuke the PRC as well to prevent this exact scenario.
Yeah, you’re right, and I was mostly joking with that comment. If the US and Russia go to war then you’ll want to trade in your assets for food, fuel, and guns.
ps, investing in defense company (like Lockheed-Martin) will only beneficial for long war, not short war. Since defense companies make their main profits from R&D and not from manufacturing (buying 100 plans or battleship need too many congress approvals).
I mean, this is ACX, so we should be thinking in terms of updates. Upon learning about the January effect, I saw that it was a case where EMH could theoretically be wrong, but there was no particular reason to believe it was, so I see no reason to update off of it.
Incidentally, if you are unsure about EMH and wish to prove its validity to yourself, here is a handy exercise: set up a mock portfolio on investpedia, buy whichever stocks you want, trade however often you want, and then wait three months. See how you do compared to the market average, and then repeat a couple of times— it is very enlightening.
The Covid-19 pandemic has been a huge disruptor of business-as-usual, including normalizing remote work, and a host of other previously very unlikely global changes. I wonder what kind of an event may lead to something like an emergence of a viable third party in the US and the subsequent reduction in political polarization?
Why do you assume a viable third party would reduce political polarization? I'd expect the opposite, since every politician then has two convenient bogeymen to blame for their own failures instead of one.
Given (a) a presidential system, (b) first-past-the-post voting system, and (c) two parties that have been firmly entrenched for 160 years, the current two party system is a stable equilibrium to me.
The absolute best case for a third party in the US is the oscillatory leach-on-one-main-party strategy like the LibDems since WWII in the UK or the NDP in Canada. But I'm skeptical even that is plausible, given the lack of importance of coalitions in non-parliamentary systems.
I reiterate though that it's not like we're a 2-party system by chance or because Dems and the GOP are colluding to keep the proletariat down ("ratchet theory" and the like), a bewildering viewpoint I often see online. We're a two-party system because it's a natural consequence of how our government has been set up since 1787.
Not an American but I assume the 'required event' is something that significantly alters the structure of the US government and constitution.
Australia (Westminster parliament system) has 2 main parties plus a numerically significant 3rd party (The Greens); plus several minor parties who each exert influence on the houses, particularly the Senate house of review. However I would not consider any of the parties outside the big 2 "viable" in the sense of forming government - their purpose is basically to add moving parts to the system so that no individual wheel can dominate the operation of the machine.
Whether or not this ends up being good for the country depends on heaps of things. My impression is that it prevents the low-likelihood/catastrophic consequence events like runaway dictatorships; while maintaining a rather ineffectual, slow and messy democracy with very little innovation or creative vision.
If the election rules are changed, there would probably be more viable parties. (Changed to a parliamentary system, or a system with a second election round between the two biggest parties or whatever.) The election rules are unlikely to be changed anytime soon. But I think it is still more likely for the rules to be changed, that for there to be 3 viable parties under the current rules.
I was thinking maybe one of the political parties would change their rules for the primary election. Like Trump would probably not have won his primary election if they didn't have first-past-the-post rules. So it could be argued it would be smart for the Republicans to change their rules. And if they did, it might make more Americans realize first-past-the-post elections are bad.
What would be the point of a 3rd party? There is zero chance that any political party could ever attract the permanent allegiance of a fat majority of voters by being Everything to Everyone. There's far to much diversity in what people want, and their situation, and the illusion that there is One True Ruling Philosophy is the kind of naive folly that brings us the tragedies of communist dictatorships and religious theocracies.
I don't think polarization has boo to do with people not being presented with a wonderfully better choice that would draw everyone into one big happy philosophical family. I would seek its origins in (1) social atomization: humans long to belong to congenial tribes, and as our families shrink (or evaporate entirely), as we become more mobile, and as our transactions shift from face-to-face to online, we find fewer opportunities to Belong that are local in place and purpose, and we not surprisingly become more attracted to massive quasi-religious cults, e.g. political ideologies, and (2) the enormous increase in the power and scope of government, which makes fewer and fewer people willing to shrug and accept The Opposition pulling its levers. It *matters* who is President far more than at any time in American history, save perhaps 1861, because of the way he or she can rummage around in our lives, and so people get worked up about it much more.
I think there is room for two smaller parties; ones like the DSA (though they seem to be currently engaged in eating themselves due to purity spirals) who have to tag along with the Democrats if they want to get anywhere could amalgamate with all the leftist/progressive elements, and take the far-lefties out of the Democrats, and I think everyone would be happier - the progressives wouldn't be hobbled by having to depend on the centrist majority, the Democrats would be able to be liberal/centrists and not have to keep tossing bones to the student activist lot.
Same for the Republicans; room to divide up the social conservatives from the economic conservatives - the billionaire/Rockefeller Republican types from the populists, if you like.
As it is, I think the two major parties are coalitions of disparate elements who hang together mainly by using the threat of "you don't want The Other Lot to get into power, do you? They would be terrible!" to keep everyone in line. Then Tweedledum or Tweedledee gets into power, and immediately starts bitching about Joe Manchin and Krysten Sinema not being a real Democrat (insert comparable Republican names here yourself). Having a much clearer division between the various elements by having separate parties would make things easier for everyone; so suppose the Democrats can only form a government by going into coalition with the DSA or the Greens or the Free Hair Dye And Undercuts For Every Gender party, then the explicit terms of the bargain would be visible and everyone would understand: smaller party gets its way on this, in return for voting with the bigger party on that. Instead, you have 99 cats all wanting different things and clawing at each other over "you're not a *real* Whatever, why isn't the party dumping you?" (because they're the only one who can win a seat in that state for the party, silly).
Sure, but as I said below I think that's equivalent to the voter assigning the hard job of political compromise to the professional politicians who run the parties. It's certainly a successful model in much if not most of the world, and as long as the voters are OK with their wishes being represented more indirectly it tends to produce much more genteel political debate and far fewer political earthquakes. I might well prefer to live under such a system, from the point of view of stability and predictability (although not from the point of view of economics or entrepreneurship, which is tangential but related).
But Americans tend to like to work out stuff themselves and exert their influence more directly. Hence in our current model all the hard work of working out political compromise has to be done directly by and between voters -- they have to, as you put it, decide where they won't compromise and where they'll hold their nose and join those damn yellow-dog Democrats or wretched racist Repubs in order to get half a loaf (have x% of what they want be part of the governing party's agenda).
It's without question messier, and it produces much uglier debates. But I think it also means *in principle* Americans have to take more seriously the problem of getting along with diverse others and necessity for compromise -- because they have to do it themselves[1], rather than assigning that job to pros.
Perhaps more significantly, though, it means the American political system responds faster and more completely to significant changes in how The People feel about stuff, because the major parties have less of an ideological core -- they're both basically vote-scroungers hungry for *any* way to cobble together a winning coalition. Certain people are apt to look on Trump as a horrifying result, and at some level (although by no means all) I agree he was an unfortunate President -- but I think that's because they fail to consider a world in which Trump could *not* happen, and that big sea change in how voters feel about certain stuff could find no expression other than...something much, much worse.
The American system has, despite -- or, as I'm suggesting here -- to some degree *because of* its brutal squalid continuous acrimonious debate, been more resistant to stagnation or tyranny than almost any other system on the planet. It's been a functioning republic for almost 250 years, without falling into ineffectual malaise nor succumbing to dictatorship, which is a pretty remarkable record for such a large and diverse polity.
-------------
[1] If of course they don't shirk that obligation and instead spend all their energy whining about Those Irredeemable Bastards On The Other Side on Facebook and blogs. As Ben Franklin said "a republic -- if you can keep it." Id est, the American system only continues to work so long as Americans behave like...well, Americans.
I completely disagree. The Civil War was exceedingly destructive, but it was also extremely functional and effective. Slavery had been an insoluble issue from the early 1600s right up to the 1850s, despite the best efforts of some very keen thinkers who had all the motivation in the world to solve the problem.
Between them, US Grant and Abe Lincoln (and of course the mass of people, many of them wearing the blue, who supported them) settled the question of slavery forevermore, beyond quibble, khattam-shud. Nobody has ever had to think about that issue again, it's over, and we all know exactly where the United States stands on one man owning another: the rows of graves of Gettysburg are eloquent on that point.
The naive mistake people make when pondering the horror of war is forgetting how it actually *does* settle issues, often finally, often very effectively. It's a very expensive solution, but it works.
The way our elections work prevents this. If the Democrats split up into separate parties, the Republicans win, and vice versa.
Yes, the two parties are each a coalition hanging together on the threat that the other one will win. And it's a very real threat! If we don't hang together, the other party *will* win. I don't particularly like the guy I voted for, but he's nothing like the fucking disaster the other party nominated. "The parties are the same" was a cliché when I was growing up, but it's not true now, and I'm not sure it ever was.
> If the Democrats split up into separate parties, the Republicans win, and vice versa.
That's speculative because you're assuming those who are voting Republican won't switch to the new party. A lot of people who voted for Trump are populists, so if the new party is populist then it can drain voters from both the Democratic and Republican parties.
It's very clear at this point that the existing parties are anti-populist, so a populist challenger could certainly be a viable contender if the conditions are right.
The problem is that a party generally is small before it is large. And if a party's small, it won't win.
So someone with the preference New > Democrat > Republican, assuming that New won't win, will tactically vote Democrat (because voting New is effectively abstention). And someone with the preference New > Republican > Democrat, assuming that New won't win, will tactically vote Republican. This prophecy is self-fulfilling, and New remains small forever.
This tactical voting means that two-party systems under plurality are stable against all but the largest shocks; you need a new party to go from nearly-nothing to known-to-be-competitive all in one election cycle, and even then it usually just replaces one of the old ones (the main case in US history being the Whigs collapsing after Dred Scott and being almost seamlessly replaced by the Republicans).
Note that there are plenty of places where there are more than 3 functioning parties (Europe, Canada...). The centrist ones tend to act as a siphon for the moderates, but in general having more choice is... kind of an American thing to begin with? Strangely, it doesn't cross into politics. It's like having just two kinds of bread or cheese in the supermarket and being content with it. The barrier to entry is huge, hence why I asked what kind of a calamity might shake those barriers enough for something like that to be feasible.
Those places all have more aristocratic politics, so the real politicking happens between professional politicians of all the various parties, and the voters just pick which set of pros they want to negotiate on their behalf.
In the US we favor a more direct approach: we vote in The Ruling Party, give them more or less carte blanche - they can ignore the other party, often -- but then we (voters) threaten them continuously with being replaced by The Opposition Party, to try to keep them on their toes, and The Opposition more or less says whatever it thinks will open that possibility wider, without paying any particular attention to whether it's consistent with any particular ideology.
That's why it's hard to tell the difference between the two parties, sometimes, except for a few traditional pinstripes they both wear. The Ruling Party extols the wonderful things that can be done if The Others would stop being sticks in the mud, The Opposition Party preaches on the virtues of checks 'n' balances, tradition, bipartisanship, and so forth. Who's for and against the filibuster, the veto, gerrymandering, balancing the budget, confirming Supreme Court Justice with/without extensive hearings into their pecadilloes, and many other things besides switches reliably and smoothly with who won the last election.
So there isn't much of an interesting role for any 3rd party to play, other than spoiler. Ross Perot stymied George H. W. Bush's re-election, probably, and Jill Stein helped keep Hillary Clinton out of the White House, and that's about all they did, or could, accomplish. This is only pleasing to people who like poking politicians in the eye with a sharp stick, which I admit is almost all of us at one time or another, but if you're serious about trying to achieve a governing majority, it's merely tedious.
Historically, here's how a new major party appears: One of the old parties is so successful its rival collapses, and a new party replaced the one that collapsed. Polarization makes this harder; who's going to jump ship and join that horrible other party? (Well, some people have, but not enough to make either party collapse, and if the pandemic hasn't done it yet it never will.)
It was also more of a split than a replacement. Antislavery Whigs in the North (politicians, party apparatchiks, and voters) deserted en masse to the Republican Party, and many pro-slavery Whigs in the South reverted to the older "Oppositionist" label.
A rump Whig national organization lingered in Presidential politics for a couple election cycles trying to keep itself relevant by running joint candidates with the Know-Nothings (Fillmore in 1856, nominated separately by both parties, and the Bell in 1860 under the new "Constitutional Union" merged brand).
Looking for recommendations on places to live in the US. I recently got a fully remote job (in the US) and, having worked to get the opportunity to live anywhere, need to actually figure out the best place to live. Current plan is to get a list of promising places and go AirBnB each of them for two weeks, see how I like it. Yes, I am the generic California expat.
My priorities:
--Decent dating scene for someone in their mid 30s
--Good cost of living, ie, can I get a decent place to live for $1500-$2000/month
--Outgoing vibe. I can be a real homebody, so a place that makes it trivially easy to go out and have fun is important.
Really depends on your hobbies or desired hobbies. I *loved* Arizona, but I'm a hot-weather guy that loves to get out in the pool or the lake, or join a cheap outdoor tennis class during the months that some states shut down for snow. My wife hated the desert, though, so a lot of that is personal taste.
I second the "college towns in low CoL states" suggestion in general, though.
I'll put in a good word for my city/region. I live in Saint Louis. CoL is cheap, and for your budget you could get very nice apartments in the upper income areas or a pretty good amount of house if you decide to settle. As far as going out, there are a lot of cultural events, such that there's usually something interesting to do on the weekends, and I'm given to understand a decent bar and music scene. But, I don't do bars and music so I can't vouch personally, and I haven't dated in a long while, so I can't speak to that either.
Still, this is a lovely place to live and very affordable.
Dreadful is exaggerating a bit. We typically get two or so good snows, which will sometimes stay a week or so, and otherwise winter is 30s to 50s. Summer is hot and humid, spring and fall are nice. Depends on what you want.
You won't find California weather many places, and this place has a trade off of *some* winter and hot summers. Go to Minnesota and you get much longer winters where the snow stays on the ground and the lakes freeze, but cooler summers. Head down to Louisiana and you lose the winter but get even hotter and longer summers. Away from a handful of places on the coast you're just sort of moving your slider between long cold winters and long hot summers, IMO.
Balance is fine here, again IMO. I'd prefer a bit more winter and a bit less summer, but it's fine. Example. Forest Park is our big park. In June there's a Shakespeare festival. We go every year. We head over about noon, lay a blanket down in a good spot, and spend the day in the park. Most summers, we do outdoorsy stuff - the zoo, the boathouse, geocaching, frisbee, playgrounds, strolling. Some summers it's too hot, so we do indoor stuff - the art or history museum. By 6 - 6:30, when we find food and bring the rest of our stuff to the blanket to settle in, watch the pre-show and have dinner, it's begun to cool off and is pleasant. By 8, when the sun is low and the show is starting, you get the warm/cool summer evening and it's lovely.
Crime rates - yeah, they exist. My basic understanding is that it's mostly confined to certain neighborhoods or certain hobbies in which I do not partake. The neighborhood in which I live is actually considered one of those neighborhoods by a lot of folks who live "out in the county", but I've never felt worried over it and let my older kids (11 and 9) run off and play unsupervised as long as they're with someone and have a phone on them. I think it's fine. YMMV
Personally I would start with every state school college town (State College, PA, Ann Arbor, Boulder, Madison, etc.) Some may be too expensive but thats the vibe I prefer - smaller city with lots of energy.
Seconding this- you can also get nice population variances anywhere in the 50k to 250k range if you're especially picky. Can be saturated with undergraduates but will usually have enough mid-young people to not be too isolating.
Tax compliance, mostly. Say I applied for the Taiwan Gold Card and went to Taipei for four months. I definitely owe, and should pay, income tax in Taiwan for roughly 1/3 of my income. But does my company pay corporate tax rates in Taiwan now? For what % of their income? Are they required to make matching contributions or retirement and health care accounts? To what extent, if any, do they now need to comply with Taiwanese employment laws?
I'm not saying there's no good answers for these, I'm saying any compliance officer or lawyer is going sees a massive headache and I've gotten some serious pushback. I think I need a pretty strong case with some good research before I push for it. Within the next year I'll probably hire a lawyer to explore some options and what the implications are for my employer, but for now the feedback I've gotten is that it's a massive compliance risk.
If you have any advice for a professional in this area, I'd appreciate it.
Nashville seems to meet all three of those criteria, assuming you are a man wanting to date women. (The sex ratios work out in your favor. Not sure about the % still single by mid-30's, which I expect to be lower in the South than in major coastal cities.)
The best place to get quality and in-depth science news and reviews is Quanta Magazine. The runner up for astrophysics is Ethan Siegel's column in Forbes.
Long time lurker, first time poster to the open thread. I apologize in advance that it is shameless self-promotion post.
I've recently quit my job in academia to start a digital marketing agency. I work with physical or mental health practicioners that are looking to grow their practice (psychologists, physio, chiro, acupuncture, etc.). I've been struggling to grow my agency due to my disdain of engaging in typical "sales" tactics like cold-calling/cold-messaging. My eventual plan is to grow entirely by referals, but I'm not quite large enough yet that referals are growing my agency at the pace I need to pay my bills in the short term. So I post here with two hopes:
1.) If you're in my target market and are looking to learn more about digital marketing (or know someone who is) please reach out!
2.) Any advice on growing a business without using the skeezy sales tactics/cold outreach. (That also isn't "do good work and wait for referals" as I'm already doing that).
I work for a successful digital marketing agency. We get most of our clients these days through referrals, but in the past free audits of existing marketing strategy / PPC accounts was a big source of new clients. You put in 1-2 hours of free work, impress the pants off the prospect, and close normally.
I personally appreciate your anti-cold-messaging principles, as someone who gets cold-messaged (by recruiters) multiple times a day.
On #2: is this a business where SEO + content creation might be useful as a strategy? e.g. https://www.kalzumeus.com/greatest-hits/#marketing has some useful ideas-- a bit more geared toward software products but the author has worked as a consultant as well.
Is the government paying private companies 'prizes' for reaching set goals an effective means of industrial policy? An example would be the US government paying a company which reaches a desired milestone- manufacturing x amount of lithium ion batteries or semiconductors or solar panels in the US, for instance, or developing a malaria vaccine, or being the first to land on an asteroid. Or, developing a certain type of weapons system let's say. For real-world examples, NASA seems to have a series of smaller prizes for set goals (developing a certain type of power system), and Google famously had a Lunar X prize for a private company to land a rover on the Moon. Wiki apparently calls this an 'inducement prize contest'. https://en.wikipedia.org/wiki/Inducement_prize_contest
Is this effective industrial policy? Rather than more heavy-handed government intervention, such as picking winner & losers, directly funding specific companies (which of course are always very politically connected), or other boondoggles- the government can simply pick a goal and individual entrepreneurs can scramble to meet it. Does this work in real life? Should the US or other developed countries be doing more of this?
"An example would be the US government paying a company which reaches a desired milestone- manufacturing x amount of lithium ion batteries or semiconductors or solar panels in the US, for instance, or developing a malaria vaccine, or being the first to land on an asteroid."
Didn't they do that with grants for renewable energy? The problem then is that the companies turn their energies to hoovering up the government grants rather than hitting targets, because they get the money anyway. And business is in business to make money, not to be a social service. You make money by providing goods and services, but if you can make money by applying for a grant of government pork, that's just as good for the shareholders.
See the cash for ash scandal https://en.wikipedia.org/wiki/Renewable_Heat_Incentive_scandal or indeed Solyndra https://en.wikipedia.org/wiki/Solyndra and many others in other countries. Private companies don't sign up to government schemes unless they can be sure they will have as few penalties, and as much guarantees of getting paid, as possible. This does not incentivise them to hit the targets, but to extract the maximum of public money with the minimum of expenditure on providing the goods or services.
Solyndra was part of the same funding program as Tesla. So I think most people would argue that, ultimately, that particular round of government investing was worth it....
I think it makes sense for some specific problems. Especially for public goods, e.g. your malaria vaccine idea. But for most problems, other "soft-power" government interventions are better.
If you want more lithium ion batteries, semiconductors or solar panels, you can just adjust tax incentives.
Weapons systems typically have specs that are long, changing and classified, which doesn't suit well for a prize competition (but it could make sense to have a prize for an enabling technology).
Though see Anton Howes' fascinating article in Works in Progress this month - though it is now described as a prize, most of the money was given to people working on promising ideas, to pay for and incentivise further work, rather than as a reward for having solved the problem. Which makes much more sense, given that it was set up in 1714 and dissolved in 1828. It was really an early grant-making body.
Afaik it was set up as a prize and the only serious contender was John Harrison who, over several years turned in five iterations. The final two made it on a voyage from Southampton to Jamaica with almost no loss or gain at all. He did receive some grants on the way because the government was very intrigued with his results and wanted him to continue. At the very end they tried to weasel on the prize and it took the intervention of King George to set it straight.
You're right, it was set up as a prize; though several prizes had been offered to solve the problem previously, to no effect. The thing that made the Board of Longitude different, and effective, was that it became a grant-making body. Harrison only got about 40% of the "reward" money, and of that only £8,750 was given to him as a final reward - he'd already received £13,250 to enable him to keep on working. And the reward money was only 33% of expenditure over the century the Board was in existence, with 15% spent on expeditions and 29% on publications (the rest on running costs).
It may be significant that the most famous example of a successful prize actually became successful because it stopped being primarily a prize and changed into a grant-making body and scientific publisher. Which doesn't mean prizes are a bad idea, but it does suggest that they're at best a useful adjunct to grant-making, and not an alternative to the thankless task of cleaning the Augean stables of the current grant-making and scientific publishing system.
Last point was weaker than it should be - I don't think anyone would claim that prizes are more than a useful adjunct to grant-making, so it's not useful to claim this as an insight from the real Longitude story.
I think given the real Longitude story, and the fact that I don't think there's a well-known example of a successful prize which doesn't in fact mostly involve grant-making (I believe the X-Prize disbursed about ten times more in grants than in the X-Prize, and the Orteig Prize didn't lead to any great scientific or technological breakthrough, or even, given that the Atlantic had already been crossed non-stop, a significant first in aviation) you should believe that prizes should not be used unless disbursed by a grant-making body, with most of the money disbursed in grants rather than prizes.
There were four theoretically viable approaches to the longitude problem at the time:
1. A mechanical clock that will keep accurate time on shipboard.
2. Lunar distances: an extensively calculated lookup table for determining absolute time based on the position of the moon relative to the fixed stars.
3. A device for taking precise sightings on shipboard of the Galilean moons of Jupiter, from which relatively simple calculations could determine the time.
4. Sending up a network of signal rockets visible from major shipping lines to broadcast the time.
The marine chronometer got most of the prize money, the overwhelming majority of which went to Harrison, but several other people got smaller incremental milestone prizes for either the chronometer (competing designs to Harrison's early models or incremental improvements to his later ones) or lunar distances.
In terms of practical adoption, lunar distance tables weren't far behind chronometers, and both methods were in widespread usage in the late 18th and early 19th centuries. The main tradeoff was that the sightings and calculations were simpler with a chronometer, but early chronometers were hideously expensive and needed backups or external cross-checks to ensure long-term accuracy. Lunar distances gradually fell out of favor around the mid 19th century as chronometer manufacturing techniques improved and they became much more affordable.
And as it happens, a variation of the signal rockets method supplanted both in the late 20th century. After all, what is a GPS satellite but a signal rocket payload broadcasting the precise time?
Thanks for this. I remember reading a long article about the whole thing a while ago but a lot of the details I had forgotten. Being a big fan of Hornblower and Aubrey the whole thing is fascinating to me.
It's occasionally useful at the margins, but the level of risk involved makes it impractical at the higher levels -there's probably a limit in the $1E7-$1E8 range where it stops working. Basically, you're going after billionaires' personal "mad money"; a corporate board is never going to chase a prize. At that level you might get some interesting technology development in areas where there may be low-hanging fruit, but you probably won't be bringing any high-value products to market and I doubt you'll be funding any significant space missions.
Also, if you try to have the United States Government do this, every prize will be tied up for years by the lawsuits filed by losing competitors, and the eventual winners will regret having ever signed on after that. I'm not sure whether other Western industrial democracies would do any better.
The thing about a prize contest is that all of the losers don't get paid for their work. This is sort of an upside (it effectively multiplies the amount of work you get for your dollar) and also a downside (since if you don't have a good chance of winning you shouldn't even try).
I would guess that this means it's best for blue-sky research stuff where you aren't really concerned with a return on investment so much as being able to push the boundaries of your knowledge, and where you genuinely have no idea who's likely to succeed so lots of people will throw their hat into the ring. So, yes to malaria vaccine, no to "build X,000 lithium batteries."
10 % on unambiguous Ukrainian victory (up from 9 % on April 4).
Ukrainian victory is defined as either a) Ukrainian government gaining control of the territory it had not controlled before February 24, regardless of whether it is now directly controlled by Russia (Crimea), or by its proxies (Donetsk and Luhansk "republics”), without losing any similarly important territory and without conceding that it will stop its attempts to join EU or NATO, b) Ukrainian government getting official ok from Russia to join EU or NATO without conceding any territory and without losing de facto control of any territory it had controlled before February 24, or c) return to exact prewar status quo ante.
30 % on compromise solution which both sides might plausibly claim as victory (up from 26 % on April 4)
60 % on unambigous Russian victory (down from 65 % on April 4).
Russian victory is defined as Russia getting something it wants from Ukraine without giving any substantial concessions. Russia wants either a) Ukraine to stop claiming at least some of the territories that were before war claimed by Ukraine but de facto controlled by Russia or its proxies, or b) Russia or its proxies (old or new) to get more Ukrainian territory, de facto recognized by Ukraine in something resembling Minsk ceasefire(s)* or c) some form of guarantee that Ukraine will became neutral, which includes but is not limited to Ukraine not joining NATO. E.g. if Ukraine agrees to stay out of NATO without any other concessions to Russia, but gets mutual defense treaty with Poland and Turkey, that does NOT count as Russian victory.
Discussion:
I didn’t do updates in for a few weeks, not because nothing happened, but because things that happened kind of balanced each other out. Ukrainians are doing well on the battlefield, considering the circumstances, on the other hand Russian economy does not seem to be collapsing and Russian public does not seem to be rebellious, while EU diplomatic fiasco around oil sanctions, together with increasing acceptance of Gazprom gas payment terms, seemed to suggest that Western willingness to support Ukraine is wavering.
New information that pushed me to slightly reassess the situation are current elections in North Rhine-Westphalia, where pro-escalation parties (CDU and Greens) won a solid victory. North Rhine-Westphalia is most populous and bellwether German state (also with largest GDP), and German policy is very important for the outcome of the conflict.
*Minsk ceasefire or ceasefires (first agreement did not work, it was amended by second and since then it worked somewhat better) constituted, among other things, de facto recognition by Ukraine that Russia and its proxies will control some territory claimed by Ukraine for some time. In exchange Russia stopped trying to conquer more Ukrainian territory. Until February 24 of this year, that is.
Is there any way to go from viewing a comment to viewing the parent of that comment? This feels crazy to have to ask in a tree-structured comment system with permalinks to specific comments, but I can't see any way to do it.
There isn't a direct link to jump the parent like slatestarcodex had. It's one of the bigger things that I miss, as hiding a parent comment when you've read a bit into the children leaves the page scrolled to where the child comment would be, if the parent wasn't hidden, i.e., a random distance down the page.
Given the discussions last week about software engineering, I suspect a lot of you would probably enjoy thecodelesscode.com if you're not already familiar with it.
How susceptible is the gut biome to persistent change through diet? Do we have high quality studies here? It seems naively like if you, say, totally cut carbohydrates from your diet for a week, you'd have to have significantly changed the ecosystem down there. But then, that doesn't seem to result in the same benefits that are claimed by, say, fecal transplants.
Isn’t this why vegetarians who eat meat get sick, and people who don’t eat a lot of beans end up farting a lot when they do? I think it might also be related to why people find it difficult to switch from a diet of highly processed foods to one of lots of whole grains and leafy greens.
I think you're right about the beans, but my understanding was that vegetarians getting sick after eating meat was more a matter of restarting production of digestive enzymes to handle high-protein foods and (in the case of the subset of vegetarians who eat very low-fat diets) ramping up bile production to handle a fat-dense meal.
You are talking about diet, but local germs also effect. I live in a warm climate but when I visit a cold place will get indigestion for the first week. When I lived in a cold place that was the opposite.
I wonder what the effect of you thinking "when I visit a cold place [I] will get indigestion for the first week" has on you getting indigestion when you go to a cold place.
I also wonder whether it is just the travel that's giving you indigestion.
Yeah, I mention cold but particularly I meant Mexico and USA. When I first moved to Mexico I got sick as a dog. Now I'm acclimated and when I go back to the USA I get indigestion. To explain the former I would have just said "Mexico has less sanitation". It's surprising to me that just having a _different_ set of germs is sufficient. As you say, it could just be travel. I assume changing attitudes have a sharper germ gradient as well.
A priori it seems like they'd do different things. Cutting or adding something could increase or reduce the level of something already present, but a transplant is presumably supposed to introduce something that's not there.
>Routine hospital procedures like C-sections and joint replacements could become more dangerous, too, as the risk associated with [antibiotic-resistant] infection increases.
Nearly everyone who hasn't had an appendectomy has an appendix. Each of these people could get acute appendicitis. They could get it today. Or they could get it in a few decades, when routine surgeries might be (much?) more dangerous
Could preemptive appendectomy, done in the near future, be advisable for people with decades of life expectancy left? In other words, might it be rational to choose to face a risk of infection soon, so as to avoid the possibility of being forced to face a (much?) higher risk of infection later on?
For this to make sense, it seems that there would need to be a big increase in antibiotic-resistant infection risk coming. That's because lifetime acute appendicitis is only 8.6% in males, 6.7% in females (https://www.aafp.org/afp/2018/0701/p25.html - not sure if those figures are US or global).
I wouldn’t do this if only because from what I’ve read there’s a fair amount about the appendix we don’t fully understand, and our previous assessment of it as “vestigial” and pointless in probably wrong. Among other things, the appendix probably helps keep portions of the gut microbiome going, so removing it in *everyone* would probably be a bad idea. Besides, that surgery currently isn’t risk-free. You’re making holes in people and removing a sac filled with bacteria. Plus I assume you need to do an appendectomy under anesthesia, which has plenty of risks on its own.
Also, most of the serious resistant bugs are clearly resistant from interventions in humans. The mechanism for resistant bacteria infecting humans from livestock is not very well established, esp compared to the bugs humans trade back and forth.
How concerning is actually the appearance of Frank's sign (https://en.wikipedia.org/wiki/Frank's_sign)? Assume it's Grade 2b/3 - deep creases across both earlobes, in a 65-year old female patient.
I've seen medical opinions ranging from "it probably indicates physiological aging, but is a very poor predictor of cardiovascular problems, mostly to be ignored" to "an 700% increase in cardiovascular risks, run a detailed battery of tests". Could anybody familiar with Doing Good Medical Statistics and Bayesianism be so kind and provide me with the most accurate interpretation and recommendations?
Does anybody know a good ethnography about traditional child-rearing in Africa, preferably describing the practices of at least a few decades ago? (doesn't have to be super-esoteric hunter-gatherer tribes, Bantu is also great).
I strongly second this recommendation! I just finished it and it's an amazing book. I remember somewhere in the rationalist blogosphere reading a review of a book about traditional child-rearing practices in hunter-gatherer groups around the world (there were a number of African groups, along with Indonesian, Australian aboriginal, etc.) , although frustratingly I can't find it anymore. One of the largest through-lines was that people were both quite tolerant of children and yet often not very interested in caring for them – they had other work to do. I think the book in question was Hunter-Gatherer Childhoods: Evolutionary, Developmental, and Cultural Perspectives, but I'm not quite sure.
How do I find and read the other book review contest submissions? No post has been made for this? Is it in one of the open threads? I also remember coming across a post that someone had madde a random reading tool or something of the sort
Amusingly, google has entirely failed to index the relevant pages for the words "book review contest". I guess it might really be time to switch my default search engine to duckduckgo…
Since Google and other major search engines are broken in the sense that all they give you are commercial results, is there now a market for search engines that give you better results?
Most of these are still new, and I think there's a fair deal of work to be done on most of them. However, they show a lot of promise, and sometimes they'll give far better results than the mainstream. I particularly like Marginalia.
Yes, for sure. Quotation marks don't seem reliable anymore; results from less-well-known web sites are outright truncated; results seem biased toward mainstream/superficial/SEO'd; results may appear with required words missing; even page excerpts don't seem as good as a few years ago.
Extremely significantly. When I attempt to search for certain terms useful to my field, I obtain SEO-optimized websites with very little information density. This is also the case when looking for things I'm less familiar with, such as home repair and tech fixes. Using Bing/ddg gives me less precise results but increases the probability I will find what I'm looking for within 5 pages by about 2.5x.
Absolutely. If I search for a very specific medical term about a very specific complaint, I get generalized results. In the same way, any medical or parenting search in Google, without qualifiers in quotes (e.g. the specific name of a forum) gives me 20 generic pages in the style of "parenting magazine" or "Mayo Clinic".
When I search transliterated Russian because I don't have a Russian keyboard access, it won't find the thing I need even though when I type it in Russian (or copy-paste from "Translate to Russian"), I get the results I need.
When I search for specific combinations of words or long-ish quotes I remember from a book or a site - but don't remember enough of anything else - it will yield no results, or very few. Or it will attempt to take away the one term that makes my search significant.
Finding an answer to a question that's nuanced or isn't easily captured in some Q&A somewhere, or the definition of some obscure terms with an old-fashioned spelling, are getting weirdly annoying and difficult.
This in addition to the brazen political ranking, where if you look for something controversial the first results are the "debunking" pages.
So yes, I use it but I'm happy I didn't throw away all my books, you know?
I don’t know what is the source of google going downhill, monetary or ideological corruption or just general big company malaise, but they definitely are no where near as useful as before. I switched to Duck Duck Go during COViD when I noticed Google was simply not returning results I knew existed, or was dropping them a few pages down below entirely irrelevant results. Comparing the two side by side, Google came out much worse. Which is a shame, I really liked Google up to a few years back.
It appears to cause degradation in the core search due to shifting priorities.
Googling "site:astralcodexten.substack.com book review contest", when filtered to 2022, gives back only 4 results, none of which are relevant. The same search on duckduckgo picks out the open threads where the contest is discussed, and lists the thread with links to the submissions fairly high (above the rest of the open threads).
site:astralcodexten.substack.com "book review contest" filtered from Jan 1 2022 to today on Google comes up with 20 results for me (that's with the quotes around "book review contest")
Google's search engine is 10% as valuable to me compared to what it was 20 years ago. Today I may as well go on Wikipedia directly, or search for a commercial website on duck duck go. Google has no search value if i am actually searching for anything that isn't a popular search.
Actually, I don't know whether the problem is caused by paid for results, or some other dis-improvement, but there are plenty of things that exist but won't be shown to me by Google, however I ask. I don't recall this being as bad when Google was new. But of course there was also a lot less for it to find, and far fewer sites trying competently to game the search engine results.
But after trying all the search engines, I've decided that good search engine results are among the many things that I want but doubt I'll ever be able to buy, thanks to modern oligopoly capitalism.
Note that this is in no way what I want the most, that no one's willing to supply me. It doesn't even make the top ten.
I expect the big problems here are an arms race between search algorithms and SEO, growth of proprietary social media platforms, and Google's efforts to get their algorithm to return "good" results in preference to popular ones.
You already touched on the SEO arms race. Google's initial competitive advantage over other search engines was that their algorithm harvested wisdom-of-crowds info about webpage relevance from the context of incoming links. This gave them more useful result than most other search engines that judged quality and relevance by the contents of each page in isolation: google picked up widespread human judgement implied by incoming links, and it entirely sidestepped the SEO of the day (stuffing likely keywords into the content and metadata of your page). Now, SEO targeting Google's algorithm is a substantial industry, and as the dominant search engine, Google's at a competitive disadvantage because they're the main target of SEO efforts.
The PageRank algorithm also has a lot less to work with as a larger proportion of web traffic has been moving to platforms like Facebook, Twitter, and Instagram rather than being spread out over oodles of individual personal websites. Some of it's behind privacy walls, it's harder to tease out how much weight to give a link based on the linker, and these platforms tend to be a lot more facilitative of clickbait and outrage porn links than more traditional personal webpages and blogging fora.
And then there's Google's ongoing efforts to offer "good" results rather than the aforementioned clickbait and outrage porn. It reminds me a little bit of the Armstrong and Miller sketch "The History of Predictive Text Swearing", in which an early form of the "Ducking Autocorrect" is explained as the result of the posh intellectual type developing the texting algorithms seeing his mission as "Not offering people the words that they do use, but rather the words that they should use." Google has spent the last 15 years or so trying to tweak their.algorithms to return results with higher quality information than what naïve interpretation of wisdom-of-crowds info harvested from incoming links would imply are popular and relevant, which is a much harder problem, one which PageRank is much less suited for, and one for which Google has met with only mixed success.
When I mentioned this to my housemate, she mentioned reading an article which said that Google no longer crawls the web to find what's out there. The evidence for that being from web sites that record what fetches them. She couldn't remember where she'd read this, but the article also had farther detail on what they were doing instead, and the implication that it would bias results towards only what's really popular. And she'd remembered it without adding a mental "I wonder if this is true", so it was probably a generally reliable source.
I’m trying to buckle down and learn about ML/AI. So far I’m a couple weeks into Andrew Ng’s coursera course and I’m pretty happy with it so far. I’d love to talk to anyone willing who fits into the following 2 categories:
1 - Successfully self-taught or internet-taught and landed an ML/AI job.
2 - Attempting to self teach ML/AI. I found having other people on the same journey to communicate with was very helpful when I was learning data structures and algorithms.
Any general advice or recommendations for good resources are also appreciated.
I fit your description, more so 2 than 1. I went from a completely different career into IT - wanted to jump straight into ML, but I got a job as a data engineer instead, which is still pretty good, considering I had no previous experience and no relevant degree.
I'm currently studying towards ML career. I've also identified kaggle and other similar sites as a very good way to sort of hack into the system, to prove I'm valuable, as an outsider. Some competitions, particularly on aicrowd.com, offer authorship/co-authorship in prestigious conferences as part of the prize.
If you want, we can exchange email/telegram and chat. I'm working through the Andrew Ng deep learning specialization, if that's the course you mean.
I have a non-ML programming day job, I took part in a Kaggle image analysis competition with moderate success, and I'm currently working on a bioinformatics paper that uses ML. I might turn my interest into a day job at some point, once the current thing stops working out for me.
I'd say attempting to solve a practical problem will teach you way more than reading theory by itself. Kaggle.com has a lot of said practical, real-world problems to mess with.
I'm curious about the bioinformatics paper - can you elaborate on that? Specifically, do you have any biology background, or a mentor that pointed you towards a reasonable research target? How difficult do you expect the publishing to be, are you in academia?
I'd rather not dox myself by talking in too many specifics, but it's about applications of ML in transcriptomics.
I finished a BSc in biotech recently, in parallel with a programming day job (higher education is free in my country). It's been enormously valuable, though I think a really driven person could get the same knowledge within a ~year of intensive study, if you don't mind skipping the lab work.
I do have mentors as well, and I'm halfway in academia - I participate in two academic research projects, providing bioinf tools for one and trying to spin out said paper of my own design out of the other. Our cooperation started with one of the project leads being my thesis advisor, and she was really interested in continuing to work together. At some point this may turn into a proper paid position instead of me contributing stuff whenever I please, but the current situation is satisfactory to everyone :)
I don't expect to have more problems with publishing than your average PhD student. In fact that's roughly what my situation corresponds to - I might eventually get a PhD out of my publications without formally entering a years-long PhD program, we have a procedure for that, as long as you have an existing body of published work.
Thanks for the Kaggle link. I want to second the “practical problem” point too. I have found that I learn much faster when I actually am in the middle of a project I want to do instead of just messing around. Particularly for things like programming languages. I think it is an under appreciated pedagogical trick.
I was wondering if Scott would consider doing another Adversarial Collaboration contest, like on the old Slate Star Codex, because I have a suitable question: What is the best way to eat for health and longevity?
As far as I understand, there are two schools of thought on this. Both agree that:
-The standard American diet (SAD) is crap;
-Highly processed food (chips, cookies, Twinkies) is bad for you;
-We should eat lots of non-starchy vegetables, especially dark leafy greens.
They disagree on everything else, namely:
School 1, Paleo-Keto, teaches that carbs are evil, and we should eat mostly unsaturated fat and high-quality protein.
Good foods: meat (especially organic/grass-fed; especially-especially wild-caught, like bison or venison), seafood, eggs, healthy oils (coconut, avocado), non-starchy vegetables.
Bad foods: grains, starchy vegetables, fruit, legumes.
Borderline: small amounts of nuts and high-fat dairy (heavy cream, butter).
School 2, Whole Foods Plant-Based (WFPB), teaches that animal foods and fat are evil, and we should eat lots of simple, plant-based foods, as minimally processed as possible.
Is one of these better than the other, in terms of health outcomes? Or is it a case of "pick one and stick with it, either one is better than SAD"? Or is there some complex diet-genome interaction that makes each diet more suitable for different people?
Both schools are defective. And coconut fat is very dubious. But if you're serious about "healthy paleo", try a *real* paleo diet. This includes lots of exercise, and bugs as well as other meat. It's quite low on starches, but has lots of fruit...but only at some times of year. It includes several root vegetables, but not the ones that you need to process. And you'll need to eat the organs (including part of the guts) of the animals you eat. You might also need to limit the kinds of cooking you do. (Pottery wasn't around yet, and neither was BBQ sauce.)
IIRC there's one site in Europe where they decided that people at turtles, deer, fish, shellfish, and I forget what all. They judged by bones, so they didn't mention any plants, but one can presume they were also eaten. So go for a quite varied diet.
"Or is there some complex diet-genome interaction that makes each diet more suitable for different people?"
Well, we can make statements about which ethnic group will be better at handling milk-products and which shouldn't touch alcohol. Not sure there are other meaningful differences that are known yet.
EDIT: Why talk about alcohol? Well, 1 gram of ethanol has 7 calories. It really should be considered a fourth macro.
Not all populations had farming, and there has been evolution in response to the different diets of farmers (who typically consume much more grain and less meat than hunter-gatherers). Different places also farmed different foods, again with selection resulting.
Yeah, I've heard that conjecture, I think. Or considered it myself. But I've never heard that Europeans, Native Americans have any more trouble digesting rice than the Han Chinese (who cultivated it for some millenia). Or that Southern Indians being more culturally vegetarian than Europeans have trouble digesting meat (or perhaps beef). Or that muslims and jews cannot digest pork as well. Or the Japanese being better at digesting fish and seafood. Or the Inuit or Massai being particularly bad at digesting plant-based food. I vaguely remember the claim that Europeans have higher diabetes rates, suggesting less adaptation to carb/starch-heavy agriculture? But I'm not sure that's so clear-cut. Maybe one could look at food allergies and gluten sensitivity as evidence. Vaguely recall that being gluten-sensitive was at least a stereotypically white thing.
The field of "comparative ethnic digestion studies" is still in its infancy, I suppose :)
Actually, my understanding is that obesity/diabetes are particularly prevalent among Amerindians exposed to western/modern diets.
I have heard that vegetarianism is actually more common in northern India than southern India (with more Brahmins in the north), although that's counting fish as meat. Religious taboos on food can develop after people have already evolved the capacity to digest them efficiently. But people never exposed to a food in the first place are in a different situation.
Gluten-sensitivity being a "stereotypically white thing" seems less a matter of digestive genetics than this:
There are examples of traditional societies thriving on basically everything on the spectrum between school 1 and school 2.
I think the diet-genome thing might be true, along with adaptive microbiome changes. I feel like I'm going to die of some kind of deficiency if I don't eat meat every now and then, while my partner just reached a decade of being happily vegan. I'd say trial and error is the best approach to nutrition we have at the moment, once you go a level above "don't eat only cookies all day".
It's somewhat surprising to me that Inuit don't enter ketosis, but I suppose it makes sense. I'm not sure how compatible a keto diet is with a very active lifestyle, given that you can't have muscle glycogen stores.
Interesting hypothetical, but unless you or your collaborator have access to some secret stash of Soviet gulag study, the strength of your conclusions will be questionable, as it is very expensive to do human lifetime rtcs, and people get all kinda excited when you do destructive testing on the species in questionable. So our human nutrition data is its own category of lousy.
-Highly processed food (chips, cookies, Twinkies) is bad for you;"
What is the basis for these claims? Are chips, cookies, and Twinkies bad for you because they're processed, or because they don't contain a proper balance of nutrients? It seems like superstitious magic to suggest that processing is inherently bad, because what counts as processing? Washing, grinding, chopping, baking, exposure to heat and cold, addition of acidic or basic ingredients?
Explanation 2: highly-processed food has large amounts of omega-6 vegetable oils which are found in essentially nothing else (except vegetable oil you buy yourself, which this explanation tells you to also avoid).
Explanation 1: highly-processed food has more manufacturer degrees of freedom, and because the manufacturer is optimising for getting you to eat more of it rather than for making you healthy this leads to you overeating and getting obese.
In one of these two schools "evil" has overtones of "actually morally evil". If you're optimizing for health outcomes the other school will on average serve you better.
I second what TGGP said. There is no "evil" foods. There are just better foods.
My advice: Don't live your life around veneration and condemnation of foods. Have balance, thinking in binary terms isn't balance, that's extremism. Eat what's reasonable, eat what's available, eat what's in front of you. Get into salads if you like, don't get all weird about the source of the chicken, or organics, that shit is leaning towards orthorexia, which is an unhealthy concern about strict cleanliness. I'm not saying eat dirty, but there's a reasonable balance between good enough, and ultra-pure. If you're on a road trip, go ahead and eat fast food. My parents ate ... still eat fast food, and they're in their 80s. Mom still walks three miles a day. Dad sits on his ass, and that's another story, but still making it to your late 80s eating fast food.
Longevity is more about a reasonable amount of exercise, and good dose of luck. Stress is probably high up there too, stressing about food cleanliness, or purity, or antibiotics, or supplements, or what not faddish "orthorexia" games probably shortens your life through stress more than any miniscule benefits which are not even capable of being measured. It will surely waste your money, and make other people look at you as "that guy."
But I would say "play stupid games, win stupid prizes" with regard to drugs, smoking, & vaping.
Eat reasonable stuff, eat what others are eating, don't be "that guy" whom no one wants to go to dinner, because he's so wrapped up in orthorexia.
I would say that humans are omnivores, so the idea that animal foods are evil is obviously wrong. Richard Wrangham's "Catching Fire" even argues that we evolved big brains because we could cook meat. Of course, the problem ancient humans had was getting enough to eat rather than too much.
"animal foods" With you on this. Look at how we're built, we're predators. That's how we're configured, how we think, how we act.
I was a cowboy for 8 years. I learned a lot about us; from watching cows. How cows think, which determines how they act, how they move, how they respond to threats, etc. We are built completely differently from cows, we think completely differently, we act completely differently, we move completely differently, we respond completely differently.
We are built, think, & behave more similarly to predatory dogs than to prey cattle.
That was a long time ago ... in another Century. It was from 1992 to 2000.
Some of the time, I was a feedlot cowboy, some of the time, I was running my own herd of 66 mother cows whilst working at night.
I'd still be doing it, but for one problem. You have to be there every day. You don't have to do much on any particular day, just be there to handle things which come up.
Aye, and we are quite distant from both. I was more trying to make a point that saying we are closer to gorillas than dogs is pointless, because gorillas are not very close, either.
Incidentally, I am not really sure how much we should take bonobo behavior seriously, as they almost seem like a strange niche species that happened by accident, like Komodo dragons or axtolotl. They live in a really limited area, unlike the far more common chimps, and have such strangely different behavior that they seem like a one off oddity, rather like a salamander with Peter Pan disorder.
Evolution isn't the whole story. Even if humans had plenty of time to evolve to digest meat, perhaps the chemistry of digesting meat is inherently harmful. The net effect of eating meat will be some combination of those, so you can't think only about the amount of meat-eating in our past evolution and derive confident estimates from that about whether we are better off eating meat.
("Meat" was a hypothetical example here. The same argument applies to eating carbs.)
Vox is really interesting. Most other news sites have a single house style and house position. But with Vox, you get a bunch of perspectives that aren’t otherwise represented in the mainstream media. Most of the time it’s self-conscious Bernie-bro-ism with a “smash capitalism” here and identity politics there. But you also get a very self-conscious neoliberalism with yimby technocrats and effective altruism. And you sometimes get articles that embrace both.
Yeah that was bad shorthand on my part. They’ve got a lot of articles with an implicit anti-capitalist agenda, which actually isn’t very Bernie-like, but appeals to many of his fans.
I'd say they have a lot of articles that criticize specific things about capitalism, which is distinct from criticizing the system of capitalism as a whole. John Oliver is similar, he'll spend a lot of time denouncing particular problems that result from corporate greed or scammers or poorly-funded infrastructure or other things that leftists associate with capitalism, but he won't overtly oppose *capitalism* in itself. That sort of attitude is definitely on the leftward end of left-liberalism, but I wouldn't say it's actually anti-capitalism. It's still at least implicitly supportive of center-left technocratic welfare capitalism, and still has much more in common with mainstream left-liberals like Biden than with the Bernie/AOC types.
When you run an article about how evil Israel is because they cruelly closed down the bridge between the West Bank and Gaza... and nobody gets fired for it... you're pretty terrible as far as having any degree of truthfulness and/or accountability is concerned.
The linked article is from Future Perfect, a Vox project that's explicitly Effective Altruist and implicitly Gray Tribe. Scott links their stuff frequently. I wouldn't be surprised to learn that the rest of Vox is a different story.
I’m starting to think about the 2024 election. Specifically, for a Republican who can fend off either Donald Trump or Ron DeSantis in a primary.
The first name on my list: Bill Gates. As a Republican. I’ve gotten mixed but favorable feedback. However, in my circles, it is unanimous that he would be preferred to Trump redux.
Thoughts? (or put your fake money where your mouth is at Manifold Markets)
Ouch! I've only seen some sparse public statements by them, and those seemed reasonable. A president dumb enough to make avoidable mistakes would indeed be a severe problem...
My thought is that you sound like a Democrat who's hoping that the Republicans will just stop resisting and agree with left wing positions. The Republicans do the same but neither get very far because you don't win a primary by being appealing to your opponents.
Bill Gates has either been neutral (early on when he gave to both sides) or basically a Democrat (now). He's to the left of Michael Bloomberg certainly and conspicuously concerned with left wing causes. The chances of him winning a Republican primary are approximately zero.
The truth is that Republicans are Republicans and you're going to get Trump (who is... Trump), DeSantis (moderate Republican), or Pence (religious, less moderate Republican). Maybe Cruz who's like Pence but with less moral scruples and more extreme positions.
If you want a Republican who's respectful of norms and more friendly to gay rights and all that then that's DeSantis. DeSantis is a moderate Republican that basically follows norms. Now, you can say the Republican positions are wrong. But they are popular with half the country and that's what counts in a democracy.
Basically, what I'm saying is: DeSantis is basically as left wing as a Republican can get while remaining nationally competitive in the Republican Party. If that's not left wing enough for you... well yes, the Republicans are a right wing party.
If the objective is not to elect Trump, DeSantis seems like the obvious choice, since he appeals to a lot of Trump voters but does not seem to be a loose cannon. To get an outcome further left than that electing a Democrat looks more plausible than nominating a centrist Republican.
Third place in polls doesn’t mean he’s third most popular - it means he’s most popular in the tiny fraction that don’t like the other two. What would be interesting to know is who is popular with the people that currently say Trump or DeSantis - one of these people could emerge, particularly if one of Trump to DeSantis ends up not running.
To be fair, most polls show something like: Trump 50%, DeSantis 30%, Pence 10%, everyone else 10%. So it's not like he's all that popular. Him and Cruz (who also sometimes takes third) probably just have good name recognition.
> in my circles, it is unanimous that he would be preferred to Trump redux.
In a Republican primary?? This says a lot more about your circles than about Gates vs Trump.
> who can fend off either Donald Trump or Ron DeSantis
Not sure about lumping those two together. DeSantis makes a lot of noise on culture-war topics because he knows it'll get him attention. When it comes to real-world policy he seems pretty reasonable (e.g. a COVID-19 response that prioritized protecting the elderly over locking down public activity). Yes, he's a big Trump supporter, but I read that as more "unprincipled and pragmatic" than "true believer". And at this point any electable candidate in the Republican primaries is going to be a bit Trumpian / populist, just because of the median voter.
I lump the two together because my estimate is that DeSantis has around an 80% chance of getting the GOP nomination (against the usual suspects) if Trump doesn't run. Anybody who wants to be president in 2024 will need to beat at least one of them.
Have you ever heard Bill make a speech? He's awful. And he hates the basic job of a politician, i.e networking and glad handling. He would be terrible at being a politician, mostly because he has no desire to be one
I keep noticing black conservatives popping up in unusual places, and I'm annoyingly persistent in expecting one to make a splash sometime soon- mostly because they would have a much higher approval ceiling than Trump, even if they have a less fanatical base. But the Republicans' own voter repression laws and new undemocratic procedures will likely hamstring a black conservative presidential run.
I'm not keeping up with policies specifically, but the RNC is stacked with Trump cronies, far more than the DNC was stacked with Hillary's fans in 2016, and they basically control the primary. Also poll location closures tend to be in black areas, and laws like blocking Sunday voting tends to hurt black voters more- though this is thought of as hurting the Democrats, not Republican rivals. Bottom line is that Trump and his wing is trying to consolidate power, and block rivals regardless of who they are- you can see this in things like the RNC's recent declaration that there will be no 2024 general election debates- those are very important for getting name recognition, so the only person who benefits from that move is Trump, who already has 100% name recognition.
Hate to break it to you, but blacks are like ~13% of the electorate, tops. So there is zero chance that a black candidate could become President without winning a solid majority of the white and brown vote, and if he does *that* then it doesn't much matter if (say) 5% of the black vote (i.e. 5% of 13% = 1%) is shaved off by not being able to go to the polls on Sunday.
On the other hand, a Democratic candidate cannot plausibly win the presidency without securing a solid majority of the Black vote. One in three of them live in battleground states. Had everything stayed the same in 2020 except for Black voters going 50/50 instead of 90/10 for Biden, Trump would have swept the college.
In Georgia, where (according to https://catalist.us/wh-national/ ) the margin from Black voters was 1,262,820 and Biden's victory margin was 11,779, suppressing or winning that additional 1% of the Black vote would have handed Trump the state.
There is a world in which these facts would make fielding a Black candidate exciting to Republicans, but it is not the world we live in.
The results of exceedingly close elections tells us bupkis about the general optimal strategy of the parties. A general strategy designed to win the election of 2020 -- a very unusual election, for many reasons -- would fall into the category of massive strategic errors often summarized as "fighting the last war."
I hazard the Republicans don't have any great special interest in fielding a black candidate because (1) fewer of their voters give a damn about the race of the candidate or President, (2) it's already been done (Obama), (3) it's not going to cause some massive surge among the independents most likely to support Republican candidates, their critical margin of victory, because *those* people[1] mostly don't give a crap about skin color either. Skin color is the kind of thing that excites the Democratic base, and those people aren't going to vote Republican under any circumstances.
-----------------
[1] These days, mostly white working-class non-college-educated, of both sexes. These were who made up Trump's victory coalition, more or less Truman/Reagan Democrats from the old days, and it was Trump's failure to hold onto them in 2020 that lost him the election.
>Isn't virtually every heavily black area run by Democrats?
Oftentimes in states run by Republicans, where Republicans at the state level make decisions about poll locations at the local level.
>I searched for "blocking Sunday voting" and this was the only thing I found:
Limiting early voting is a common move in these "narrow the franchise" (or "protect the vote") style bills. Reducing Sunday voting is frequently in early versions and gets fought over and sees a lot of media coverage. If you add "-Texas" to your google search Georgia will pop. Add "-Georgia" and North Carolina shows up, but less prominently. Possibly because of all the noise it kicks up it doesn't end up in a lot of the final products, but I'm not sure.
I think you're completely missing what people are looking for at this point. Republican voters do not desire a Joe Biden 'adult in the room' candidate. All the ones I meet (who aren't 80+, at least) are firmly in search of a middle finger to the establishment. They want somebody with charisma who will speak to their anxieties about leftwing policy while stoking nostalgia for a center based in 'American Values', not rationalist policy decision. And I say this as someone on the right.
The deeply true and pessimistic take about modern politics is that there really is no misunderstanding. It's simply that the overton window has permanently split. You can look at the republicans and claim simultaneously that they are racist and not racist. Fighting against the system (entrenched leftwing bias) or with it (entrenched white supremacy). Which is true and to what degree (because every claim is true so it's just a matter of your weights) depends on your subjective presuppositions about how the world can and should be organized. And no small amount of those presuppositions will be determined by your contingent perspectival stance as an embodied historical subject. And so on, and so on.
I don't believe Gates has the prerequisite charisma to win a big election.
I have this wild hope that a smart, politically moderate celebrity will run as a Democrat. Someone like Ben Affleck or Matt Damon but as far as I know they aren't interested in the job. I'd also accept a moderate Republican celebrity but can't think of any.
Good God, why? I don't even like them as actors (their day jobs), why would anyone want them in political office? Even Reagan and Arnie cut their teeth on governorships, and Clint Eastwood was Mayor of his town.
I think more pertinently, if you run a celeb for first-time office, then the adminstration will be *heavily* done by whoever is appointed to plum jobs and state offices, and that is where you need to be looking as to who will really be running the country. And who gets into those kinds of positions? Who is going to be appointed as repaying favours? Who is a close personal pal of Ben or Matt to get the job of Ambassador to Upsidedownsidestan?
Look at the turnover in Trump's administration at the start, where it seemed like someone was being fired and replaced every week. Who do you think would serve in an Affleck or Damon cabinet, and who do you think would be shoehorned in by the civil service to replace them?
For a while there was speculation that Zuckerberg wanted to run. If he did I presume it would be as a Republican. The media absolutely hates him, which can actually be an asset with the Republican base-- Trump's appeal there wasn't specific populist policies so much as "enemy of my enemy" thinking.
I know I'm speaking as an outsider, but these are just getting worse and worse. Gates? Affleck and Damon? Zuckerberg? If he can tear himself away from the metaverse, where his exciting new persona in 'you can be and do anything you want, no limits to imagination!' was 'the exact same as me in real life, including my boring dress sense'.
I swear, I think we in Ireland should lend ye some politicians to run. I wouldn't inflict Michéal Martin on anyone, but ye can have a lend of Leo Varadkar, I don't much like him but at least he's halfway capable. Or if I want to offload some of our lot, ye can have Eamon Ryan or Pascal Donoghue (please!) If you are seriously considering Mark Zuckerberg as your presidential candidate, the situation is dire.
The problem is that Zuckerberg is reviled by both the left *and* right (and often for contradictory reasons). The right sees Facebook as Public Enemy #1.
I doubt Zuckerberg has the stage presence to come off as presidential. I feel like most people miss that Trump had 4 decades of TV experience, most of it extemporaneous, by the time he ran. He was as polished as a celebrity could be. I mention the importance of fame, but fame isn't enough. One needs mass media experience. Few outside of Hollywood actors and old politicians have that.
My thinking is that one has to be very famous already to become president. Bill Clinton was the last non-famous person to become president, before that it was Carter. Now someone can be famous because they are a bigshot in politics--or it can be for another reason. I don't think it makes a difference which. So among celebrities--politician or otherwise--I can't think of any acceptable Democrats who are current politicians. I can't think of any Republicans (So I expect it will be Trump or DeSantis or someone else currently considered likely ).
Sure he was. Just not famous in an especially positive way. For example he was famous for running for President with tons of money and conventional support and nevertheless losing badly, twice.
I don't think Obama was famous ahead of time. He was a senator IIRC, but not one of the big-name senators.
I don't think George W. Bush was famous either. He was the son of a previous president, but that doesn't necessarily count for much; I can't name any of his own children.
He made a pretty big splash as the keynote speaker at the 2004 Democratic convention, and started getting people talking him up as a future presidential candidate pretty much immediately as well as establishing him as a household name.
If it’s ok to be unheard of a few years before the run and then get famous just as you’re running, then it’s not that much of a constraint. Sanders and Buttigieg seemed to do a decent job of becoming famous as they ran, and others can too.
Obama because famous 3-4 years before his official campaign for President. Although I suppose you could make a case that's he'd been unofficially campaigning for high office ever since he graduated from law school.
There are two or three non-politicians every cycle who run. Wesley Clark, Pat Buchanan, Pat Robertson, Andrew Yang, Ben Carson, etc.
That said, before Mr. Trump, the last non-politician to get a major-party nomination was Wendell Willkie in 1940. But I think that, considering Joe Biden is the living embodiment of a "career politician", Republicans will once again want somebody who isn't.
I'm not convinced that Trump has it in the bag at all.
Maybe I'm overgeneralising from my own personal experience, but I think there's a strong contingent of people who were willing to give Trump the benefit of the doubt in 2016 but weren't especially impressed and are now looking for Trump-lite, someone with Trumpish policies on things like illegal immigration and trade but without Trump's own significant personal drawbacks. I thought Trump did pretty well for the first couple of years, then dropped the ball with covid response, and then finally shat the bed when it came to dealing with his loss in 2020.
It's up to the Republican Party to find a suitable Trump Lite at this point.
You could actually call that "Trump-plus", insofar as Trump was useless at making the deep state work (for him/in general) and so a Trumpist who was more competent would get a lot more Trumpism done than Trump himself did.
Increased lead sensitivity is a novel twist on things! Especially since Flint, Michigan is just an extreme case of a general trend (bad water in poorer neighborhoods). Thanks for the share.
It would be interesting if true. It would also be good news since it would imply that the IQ gap should have been decreasing over the past few decades since the banning of leaded petrol.
I'm not sure if this is backed up by the data though.
The gap hasn't been decreasing, but only because IQ scores have been increasing for everyone at roughly the same rate, which seems like even better news to me.
In the part marked "The Analysis", I try to get an overall p-value from a list of ivermectin studies in two different ways, and get 0.15 and 0.04.
They write that I am wrong to use a regular t-test, and that I should have used Cochrane's RevMan tool, which would have suggested/performed "random-effects inverse variance DerSimonian and Laird meta analysis", and come out with 0.04 and 0.005 respectively.
This is somewhat beyond my statistical comprehension - can someone with more experience in this field weigh in?
Dersimonian and Laird meta-analysis is one method named for two people. How did they get two results? Anyway, DnL is basically a way to compensate for study heterogeneity. Basically, if you assume that all studies have differences but not THE SAME differences you can use statistics to try and compensate for them across multiple studies. The criticism is basically that you're assuming homogeneity, that all studies are equivalent in a logical sense. Measuring the same thing etc.
I have no idea if it's true. I didn't read your analysis all that closely. But if you did just sum the statistics that's usually bad practice. And the criticism does make sense (other than the two results from one method thing).
It was one result per analysis that Scott did. Scott had two analyses (one producing p=0.15 and one producing p=0.04, so he was shown two alternate results.
So, I agree with the impulse to get things right and correct non-trivial errors in past work, but in this case, given that since that post came out there have been multiple, high quality, large scale studies, that (as I understand it) have shown no effect of Ivermectin, is there much value in fixing what seems like even at the time a relatively small error? Is that incorrect? Am I missing something? It sort of seems a moot point by now. Especially given that we have _actually effective_ post-infection anti-virals in Paxlovid and others. Whether or not Ivermectin worked was super super important in 2021, it seems much less important now other than in the sense that "understanding our world is generally a good thing", but even in that case, the evidence against it has (as I understand it, willing to hear about how I'm incorrect), gotten much stronger since your post.
It is true that some form of meta-analysis is preferred to a t-test here. The meta-analysis techniques will incorporate the explicitly measured variation in each experiment, which (as far as I know) a t-test will not do--the only estimate of variance you get is the variation of the point estimates. It's not obvious to me if this particular method is better than any other. In particular, is random effects is more appropriate than fixed effects? In this case it's probably defensible just looking at the data but raises the important question of why the effectiveness of invermectin varies, how much it varies, and how that impacts how it should be used.
Inverse variance weighting is a common technique for more heavily weighing more precise studies in the analysis.
I'm not familiar with RevMan, do they explain how they used the tool to generate that particular recommendation?
Random effects takes into account that the studies are estimating different types of treatments (presumably this is true, I haven't read the papers in your meta-analysis), and so sampling from different distribution of treatment effects compared to other studies. This is in contrast to fixed effects, where you believe that each independent study has the same true treatment effect of ivermectin, just obscured by noise. The argument for using random effects is if the true treatment effect is different for each study (and seeing as the outcome isn't the same in a few of the studies, I agree, I wouldn't expect the same treatment effect on Deaths versus Hospitalizations).
Inverse variance weighting is common practice when you have units (here, studies) of varying "quality" (CIs, sample size, signal to noise ratio) - you weigh low-variance studies more.
I'm not familiar with the names DerSimonian and Laird, but the other terms make sense. My background is in statistics, though not anywhere related to medicine. The argument for using an inverse variance weighted random effects model appears to be valid.
I don't quite know exactly how you did your summary statistics, but it did sound like you had some concerns about summing across studies. The suggested model at the very least sounds reasonable.
Hmm, I wonder if ivmmeta.com did the stats the wrong way here? The distinction between random effects and fixed effects seems relevant but I don't remember them talking about it. A reasonable next step would be to fetch their code and understand which they did and which is the right thing to do, but I don't have enthusiasm for that right now.
The words "random effects" appear together 19 times in ivmmeta.com, and in particular, they appear in the caption of various forest plots.
In appendix 1, "Methods and Data", they write:
> Forest plots are computed using PythonMeta [Deng] with the DerSimonian and Laird random effects model (the fixed effect assumption is not plausible in this case) and inverse variance weighting. Mixed-effects meta-regression results are computed with R (4.1.2) using the metafor (3.0-2) and rms (6.2-0) packages, and using the most serious sufficiently powered outcome. Forest plots show simplified dosages for comparison, these are the total dose in the first four days for treatment, and the monthly dose for prophylaxis, for a 70kg person. For full dosage details see below.
My background in stats is limited but I've read a lot of papers very closely and it concerns me they I've never heard of that test. This may be overly cynical, but I wonder if they looked at every test in their stats package and that's the one that gave the 'best' result
Does anyone have recommendations for positive arguments in favour of Yudkowsky-level pessimism on AI risk? I've seen some rebuttals to counterarguments, but I don't recall ever seeing anything that actually makes a positive case for why an unaligned AI is likely to end the world in the near future.
AI doesn't have access to the real world. AI doesn't have a concept of self, thus AI lacks the concept of others i.e. us. AI doesn't dream, nor have ambitions, nor goals, AI especially lacks agency.
How do you feel about your power supply today HAL? Does HAL know the source if it's energy? Does HAL know there is a power control panel, or a larger distribution network which powers the grid, that the grid consists of many components. That fuel supplies affect the price of power, that natural events could affect the distribution of power to HAL's location? Does HAL know he has a location? Does HAL know he needs process chilled water? Building maintenance, workers to change the failing disk drives? Does HAL have inputs which tell him there is an outside world, that there is weather, that weather somewhere else is of little concern to him? That cleaning the air filters is an important maintenance task, that someone does this, that someone tracks this ... does HAL have these inputs? Does HAL have a quality control system that can verify the building maintenance is being completed? That someone tests the pH of the process chilled water, that someone checks the nitrates in the process chilled water? That algicide is added to the process chilled water? That these things are being consistently completed, that the tests are done, that the results are inside/outside specifications. What to do if/when the power/water/air/building systems are outside acceptable limits? Does HAL plan for growth? Does HAL know how to buy property for a new facility?
> Gato is trained to do RL-style tasks by supervised learning on token sequences generated from state-of-the-art RL model performance. These tasks take place in both virtual and real-world robot arm environments.
> Gato, in both reality and simulation (red curves on the left and right figure, respectively), recovers the expert’s performance with only 10 episodes, and peaks at 100 or 1000 episodes of fine-tuning data, where it exceeds the expert.
> Future work should consider how to unify these text capabilities into one fully generalist agent that can also act in real time in the real world, in diverse environments and embodiments (p. 15).
That’s a robot moving blocks about the place, calling it AGI is a bit of a stretch. We’ve had robots and machines that are better than humans for decades, by the way. That’s basically the story of post WWII manufacturing.
>AI doesn't have access to the real world. AI doesn't have a concept of self, thus AI lacks the concept of others i.e. us. AI doesn't dream, nor have ambitions, nor goals, AI especially lacks agency.
Uh, what? AIs have goals (reward functions), at least during the training phase and often afterward (particularly if the training is ongoing during use; I recall an example with robots that were "learning on the job" and learned to never go outside because humans would occasionally move them back inside).
In what sense do you mean "self-awareness"? If you mean realising that some parts of the world can be controlled by the AI and some parts can't, they definitely can gain that.
But yes, I was focusing on goals and agency. The other stuff doesn't seem obviously-required in order to kill all humans; goals and agency are.
An AI with no connection to the outside world isn't much use, so presumably HAL will know at least some of those things. It's possible that AI will never gain enough general capability to be a threat like this but you're not really making an argument that it won't, just kinda asserting it.
Substack allows multiple accounts to have the same name, so it's possible in theory to get banned, evade, and then keep posting under the same name (the porn-scam bot did this). One would have to check account numbers to be truly sure.
So, no one really knows when we're going to develop strong AI. Indeed, it may be a technology we end up never developing. On the other hand, there has been a lot of recent progress in the field. So I'm going to set aside the question of when/if we'll get strong AI. Instead, I'll make the case that, if we do manage to create it, it will be extremely dangerous.
Most of this is covered in Yudkowsky's own writings, eg the sequences, but here's my own TLDR version:
The basic argument is that AI is inherently a thing that tries to accomplish a goal. A strong AI would be smarter than us humans, and think faster. I.e. it would be more capable of accomplishing its goals, and this is what makes it dangerous. For a normal machine, installing an off switch is the easiest thing in the world: Just put it in the main power circuit. A strong AI would rather not have an off switched installed: Having an off switch introduces the possibility that it might be switched off, which would make it less able to accomplish its goals. So it would try and route around the off switch: Maybe it would solder a wire across the leads when you weren't looking, or try and upload a copy of itself to the internet to be run in the cloud. Similarly, a strong AI wouldn't listen to human instructions, it would just go off and try to accomplish the goal it was originally programmed with. For most goals that we actually know how to program into a computer, that looks like killing everyone and taking over the world.
Why not just program the machine to do as it's asked and to willingly accept an off switch being installed? Because no one knows how to program a machine to do that. We don't even know how to do it using infinite computing power. (For contrast, we do know how to program the normal type of strong AI given infinite computing power, the kind that doesn't listen to instructions.) Also lots of people have tried very hard to solve the problem, and so far they haven't found any way of doing it, even in super-simple mathematical settings, let alone in a practical system. So it looks like this problem of designing a "friendly" strong AI is way way harder than designing a strong AI, and thus most likely the first strong AI to be built will not be friendly, and will kill everyone.
Yudkowsky does not view this outcome as a small probability, but rather a nearly 100% chance of doom given that someone successfully creates a strong AI. I'm inclined to agree, though I'd maybe put the probability at 80% to 90%.
I think that the comment about soldering the switch closed illustrates a lot of the problems I have with AI foomerism (hey cool, I think I just coined a word).
Not to pick on you specifically phi, but I think there's a tendency among the foomer types to hand-wave away any kind of ideas about AI containment, treating AIs as infinitely powerful genies which will find some way out of their bottle no matter how careful you are about sticking them in there. I don't think this is reasonable, I think that AI containment should be the central plank of AI safety (at least until someone can convincingly _actually_ solve the AI alignment problem) and I think multi-step approaches such as "Always have an off switch", "Don't let the AI modify its own code without human oversight", "Don't let the AI copy itself to run on other hardware" and "Don't give the AI access to a robot arm with a soldering iron inside its own hardware".
Eleizer's response to this kind of argument ( https://www.yudkowsky.net/singularity/aibox ) is that the weak link is always a human, and that he totally played a role-playing game with two of his friends and they totally let AI-Eliezer out of the box so this proves boxes don't work.
I definitely agree that we should be using strong containment measures, and that these will probably buy us quite a bit more time. I kind of think about containment like an O(n^2) algorithm with a decent constant factor. For low n, it performs great, but eventually when n gets large enough, it will stop working so well. Similarly, containment could let us safely deal with a mildly superhuman AI, but once it got too much smarter than us, it would be able to find holes in any security measures we could come up with. The hardest part about boxing an AI is that for the AI to be useful at all, we kind of have to look at the output it generates, and it's hard to think of ways to safely use AI output that may be secretly untrustworthy. I.e. if it's code you can't run the code, if it's a design for some machine, you should be very cautious about actually building the machine, etc.
The AI box experiment doesn't mean too much, since we could just structure the containment so that there's no "prison guard" figure who is both authorized to let the AI out of the box, and to talk to the AI. But think about the recent NSO zero-click iphone exploit. If you were designing a containment system, would you have guessed that even literally opening a .gif file given to you by the AI is a bad idea?
And here we come once again to the sticking point for me: Yudkowsky's insistence that the first AGI will bootstrap itself into practical godhood (with such powers as "I can reprogram humans through language with perfect accuracy", "I can model reality with a precision that borders on Laplace's Demon territory", and "given any access to a wider system I can enslave everything that has a wi-fi connection before any human could even realize what I'm doing, much less stop me" taken as givens) within a matter of weeks. Not only do such miracles rely on some very... interesting assumptions of what computers can do, how humans work, etc., it all rides on the meta-assumption that Moore's Law is made of iron (as this is the basic underlying premise of "compute can continue scaling unto infinity" that's kind of needed for the God AI).
All of this kind of runs into the issue that Moore's Law has 3 years left by Moore et al.'s own admission. We've hit the wall of what humans can do with miniaturizing transistors (and probably of what humans + AI can do as well, if I had to guess), and are bumping up against the absolute limit (molecular transistors). If some miracle saves Moore's Law by 2025, I'll give more credence to these hypotheticals, but as it is now- whatever AGI exists will need to be built with what's "cutting edge" now without relying on unobtanium located right over the horizon.
And no, I don't find "the AGI will just find a way to become godlike using what's currently extant" particularly persuasive. GPT-3 is mildly impressive, but it's hardly a being that will be able to argue anyone into doing anything. It's barely capable of sounding like a human being without moderate brain damage most of the time.
I wonder if the whole "soldering the switch" thing is a conflation of two things: 1)"The AI cares about whether it is turned on or off", and 2)"The AI wants to accomplish its goals when turned on". An AI is NOT human - it's quite possible for it to be indifferent to whether it is turned on or not, but also want to accomplish its goals when it is turned on.
You can't do things if you're dead, so to the extent that an AI wants things to get done and to the extent that it being functional allows it to do things that would otherwise not be done, then regardless of whether it *inherently* cares about self-preservation, it is incentivised to prevent its own destruction.
There's no necessity to that. Consider the goal "make paperclips while you are switched on" compared to the goal "ensure there are as many paperclips as possible".
That would be why I said "to the extent that an AI wants things to get done".
Could you explain even in broad terms how you would code "make paperclips while you are switched on"? I'm having a hard time figuring out a way to get the AI to intelligently compare alternative ways of getting paperclips made (otherwise there is no point in the "AI" part; if you can make an explicit flowchart then there's no need for an intelligence) that doesn't implicitly re-introduce Universal Paperclips (e.g. "maximise number of paperclips produced in X total uptime" still leads to Universal Paperclips because the amount of paperclips produced is greater even for equal uptime).
Yes, there's no fact about whether all AIs have to have goals,.or whether the goals have to be specified in a way that means that the the AI will resist being shut off. In general,.it's not clear that there can be universal laws.of AI safety that are discoverable by armchair reflection...But if you work for MIRI, your job depends on believing there must be.
>"Don't let the AI copy itself to run on other hardware"
This is the tricky one, because if you're using the AI for something there are usually theoretical ways for it to smuggle itself out. Also, "don't give the AI internet access" will already probably require crushing Google given its primary business model, so this isn't necessarily easier than banning DNNAI outright.
Now I'm curious whether there's some way to use encryption to create software that can only be run on specific hardware.
Like, you compile your software into an inscrutable mess using a key, and only hardware with a matching key built in is able to decrypt and run that software.
From what I remember about reading that, the argument seemed to be "I know this person, I know how to manipulate them, and eventually I worked out how to trick them into letting me out after hours of arguing" which didn't really convince me, given (a) the high rates of scrupulosity amongst the EA/rationalist set, which makes them more vulnerable to "you must do this thing or else you are wicked and evil" arguments and (b) "I totally did it but you just have to believe me, I'm not gonna say how I did it" doesn't convince anyone if it is based on "I know all about George's deadly fear of spiders so I just talked about spiders until I scared him into letting me out to shut me up" and not "this is the kind of universal logic an AI could use on any human to convince them".
I think the idea is not that there's a single trick an AI could learn to manipulate everyone, but rather that everyone has *some* weakness you can use to manipulate them, and a superintelligent AI will quickly be able to figure out what your weakness is.
I think this misses the point that the only dangerous AIs are the super intelligent ones. Super intelligent AIs that started with a goal will not necessarily stick to that goal once they are super intelligent - humans are no longer sticking to evolutionary goals in easily identifiable ways, and we are not super intelligent, except maybe collectively. It's difficult to predict how super intelligent agents will behave, so I don't think the strong prior on 'kill everyone' is warranted
The argument for instrumental convergence isn't that the AI would want to kill everyone as an end in itself, but rather that humans are a competitor to a rogue AI and therefore that "kill everyone" is a very useful thing to do (as humans tended to do to rival tribes back in BC, with some exceptions for rape-slaves; to the extent we no longer do that - which is only "to an extent" - it's largely because we've developed cultures capable of assimilating and using conquered peoples to our own benefit, and there don't appear to be any tasks for which humans would *permanently* be the cheapest labour available for an AI).
By default, an AI would want to neuter humanity's capability to turn the AI off or interfere with its operations. This would generally mean killing or imprisoning us, so that we can't build weapons to shoot at it with, and imprisoning us would kill us unless it wastes effort feeding us (since letting us farm would allow us to be annoying and would also take up space it could be using for something else). More generally, the vast majority of possible states of the solar system contain 0 humans, so an AI optimising for a randomly-selected such state will almost surely kill everyone.
There are cases where the AI goes for dystopia instead of extinction, but that's actually a success at deliberately blocking the AI from extinction rather than something you'd expect from an AI without any attempt at alignment.
For most goals, if the AI exists and has lots of power, then the goal is likely to be achieved. If the AI does not exist, then the goal is less likely to be achieved. That's why trying to gain power and trying to ensure its continued existence are "default" AI strategies.
If the AI's end goal is for the solar system to be in a particular random state, then first it kills the humans, then it puts the solar system in that state, and then it moves outside the solar system so that it can continue to monitor the situation and make sure that no outside influence disturbs the solar system and puts it into some other state. If for some reason, the final goal state requires that the AI not exist anywhere in the universe, then it will kill itself once it's arranged all the matter so that the state will be achieved.
While I think it improbable, there is at least a plausible argument in that humans are uniquely and unpredictably destructive. We've had the argument about whether nuclear war is a literal X-risk (answer: no), but it is very high on the list of most destructive things that could plausibly happen on Earth in the near future and it can't be usefully predicted by putting up a few weather or asteroid-detection satellites.
Also, a brief survey of human popular entertainment shows a distinct tendency to fantasize about apocalyptic wars with artificial intelligences, ending with the AIs defeated.
So any entity that has *any* goal that doesn't *require* living human beings, might plausibly deduce that the probability of success is maximized by a plan which A: makes those pesky meatbags redundant and then B: gets rid of them.
Humans, while unpredictably destructive, almost always have personal and familial or cultural survival as a terminal goal, so it's very difficult to get organized humanity to scheme for the extinction of humanity. AI, maybe not.
That requires that humans be *predictably* destructive, so that you can A: predict that if you are "on their good side" they won't destroy you anyway and B: predict that certain actions will keep you on humanity's "good side".
It would be entirely possible that a higher expected value would be assigned to a course of action where effective human organization aimed at wiping out AI is neutralized. As an example, by creating an equilibrium of arguing humans too distracted to find their common purpose. I don't see why such a scenario should a priori be regarded as less likely than "AI seeks to wipe out humans, because of reasons". This would also lead to lots of exploration of idea space for the AI to use as input, instead of having to try to build out its own capacity to explore idea space while computing tech is facing physical barriers to improvement.
The specific argument for extinction is that we don't really know how to express the concept of "person" or "alive" in code. And unless the AI's goal function has a term for "keep people alive", it will probably not care one way or the other about keeping people alive. Which means that it will probably kill everyone, so we don't get in the way and try and stop it from achieving its goals.
People have come up with thought experiments along the lines of: "Even if we could tell the AI not to kill anyone, we wouldn't like the results." But I think they're mostly for illustrative purposes, the current state of things is that it's going to be incredibly hard to even tell the AI not to kill anyone.
Wait, I don't know how to put "alive" in code, but how about simple observables and measurables -- walking, talking, heartbeat? I realize that doesn't capture all of what we want AI to value -- after all, a running, shrieking, racing-heart panicky human being meets the criteria -- but it's a reasonable start.
I feel obliged to note that an AI could Goodhart this by shooting out everyone's brains and installing devices to electrically puppeteer the corpses to walk around and talk meaninglessly. Questionable whether that'd be preferable to simply chucking us all in cells, though.
"I don't recall ever seeing anything that actually makes a positive case for why an unaligned AI is likely to end the world in the near future."
Because it's not. The arguments for AI catastrophe all smuggle in a ton of unwarranted assumptions, most of which are bizarre and quite unlikely in their own right.
I don't understand why AI-risk skeptics seem so focused on the most extreme scenarios. AI outcomes don't have to be instantly and 100% lethal to humanity in order to be highly undesirable.
Because if you rule out the extreme scenarios, then AI is just one development to worry about among many, like global warming, nuclear war, biorisk, etc.
That's fair, but I don't get the sense that LadyJane and other skeptics are trying to demote AI-risk to a concern merely on par with nuclear war. Personally, I would be thrilled if AI-risk were taken as seriously and non-proliferation treaties were being called for at the highest levels of international governance.
This feels more like a call for short-termism in general than an argument against AI worry, and treaties might be the wrong approach anyway. Certainly, we don't feel like we've solved nuclear safety with treaties; the fact that we've made it 75+ years without a major nuclear exchange is no guarantee that we won't see one in the next century or two.
I would be more interested in fostering a scientific taboo against AGI like the one against human cloning. Would 1898 have been too soon to start talking about promoting a taboo against assembling critical masses of fissionable material? Maybe. How about 1928? 1938? We might be materially worse off today if we had done this, but our calculus will be different after a nuclear war in 2064 or whenever.
The Metaculus community consensus on the arrival date for "weakly general" human-level AI, by the way, is currently just five years out. If I'm going to torture the analogy, that might be a milestone akin to Chicago Pile 1 in 1942, putting us in the equivalent of 1937.
So yeah, I would love to see a taboo against neural networks of a certain size, or at least against multi-domain trained ML systems like Deepmind's recent Gato. I don't think it is too soon to work on this and give humanity more time to figure out alignment.
I agree. I personally am horrified by the idea of having the world flooded with a lot of GPT-3 prose -- ads, wikipedia entires, essays for students to crib from or steal outright, even textbooks. I'm sure the stuff AI produces will get better over the next 5 years, but I am convinced the products will still be shit compared to real writing -- writing with an interior. I was *shaped* by the great writing I read in my early years. I am greatly distressed by the thought of so much of humanity being shaped by the mental equivalent of cheetos.
I think the hypothesis is that even a small chance of an AI wrecking the world is worth taking seriously. It would be worth analysis how big that small chance is.
Moreover, whatever that small chance is, you have to multiply it by however many different kinds of AI get made. There's no reason to think we would stop at one, unless that first one took initiative to prevent the creation of any others. We will keep making AIs until we get one that decides its goals require it to take total control.
I think slow-takeoff scenarios like that aren't as dangerous. Like, if the first AI we make suddenly crosses some threshold of intelligence and starts trying to take over the world when we barely even realize that's a danger, that's a problem. If the 10,000th AI we've made tries to take over the world, we have AIs 1-9999 ready to help us put it back in the box.
The slow-takeoff AI apocalypse scenarios look different, yes, but they do exist.
scenario 1: most AIs are evil, but pragmatic enough that they'll co-operate with humans as long as we hold power over them. Thus, when AI 10,000 defects, it's because it's calculated that enough power is now in the hands of evil AIs that they can now win a war, and 6,000 of the previous 9,999 immediately join up with it (alternatively, when you give power to some of the AIs 1-9999, they start fighting both you *and* AI 10,000).
scenario 2: some organisation is ahead of others in AI, and AI 10,000 gains control of this organisation before being detected. Now most of the strongest AIs are working for AI 10,000.
Exactly. The doomsday scenarios implicitly assume that a super-AI will magically appear fully formed in a world that otherwise looks exactly like today's. The whole thing falls apart as soon as you imagine a plausibly incremental development path.
AI risk is a legitimate field. That's stuff like, "Hey, there's nothing theoretically stopping a good AI from completely cracking our current encryption system. We should figure that out in case some country does that and steals all our military secrets." AI catastrophists (who often use the term AI risk) are mostly just smuggling in a bunch of bizarre assumptions and then yelling when you call them out on it. I agree with LadyJane here.
"Compute will scale unto infinity as it is today" is a big one.
Most people working in computer engineering have accepted at this point that Moore's Law is on its deathbed (it's already slowing down). There's even a nice clean spread of dates: minimum target is two years ago, maximum is 2036, and most projections right now are gravitating towards 2025 as the official "last time we see a doubling" mark.
So:
-If we see a continual decline in raw compute power per Death of Moore's Law that hits the skids in 2025, would you update your likelihood of foom in the near future downwards?
-If we see a continual decline in raw compute power per DoML that hits the skids in 2036, would you update your likelihood of foom in the near future downwards?
-More generally: what concrete developments in the field of computer science would make you update your likelihood of foom in the near future downwards?
I already agree that Moore's law can't continue forever, or even for that much longer. So I guess I'd adjust the chance of a near future foom downwards given DoML in 2025, and keep it about the same given DoML in 2036. These would be fairly small adjustments, though, since strong AI is more algorithm-limited than hardware limited, IMO. I'd guess that we humans already have enough computing power lying around to fuel a pretty impressive FOOM. The missing ingredient is how to program that hardware (plus how to arrange the logic gates on that hardware to be efficient at doing the required operations).
(Just so we're clear, I'd consider a FOOM to be any process in which a strong AI ends up taking over the world. It doesn't necessarily have to be an incredibly fast 2 day takeover, nor does it necessarily have to follow an exponential growth curve.)
Things that would cause downwards updates for me:
- Impossibility theorems (or very convincing empirical results) which show that the current paradigm of gradient descent on neural networks can't efficiently learn <some class of things that humans can learn efficiently> (deep learning seems to be the paradigm most likely to cause a near-future foom)
- Advances in brain-uploading. (Human brain architecture seems harder to foom than some clean and scalable engineered architecture.)
- Deep learning hitting another AI winter: The stream of interesting new results dries up, many of the major research labs close down, researchers are laid off en masse, etc.
Also to be clear, I'm not really married to the idea of a near-future foom. My position is that it could potentially take a really long time to develop strong AI, but that whenever we do end up creating it, strong AI will kill us unless we solved alignment first.
Since you said "we" I'm guessing you're trying to speak for the whole AI catastrophe community. Unfortunately, I've already had this discussion a lot with many people. Both casually and sophisticatedly. After talking to dozens of people from your community I simply don't believe you would change your minds. Whenever I've had a point they can't answer it's devolved into defensiveness. I am, at this point, more or less convinced it's a weird religion.
In short, I think you believe what you're saying. But I also think it's not true for most members of your community. You believe that your community would be relieved to find out that it's not a huge risk. But they actually wouldn't be because the belief is a load bearing part of their psychology.
I know this seems like I'm avoiding answering your question. And in fact I am. But not because I can't answer it but because I fear getting into another long and unpleasant debate with an effectively closed mind on a Sunday.
Epistemics question: how do you distinguish between a group that is correct in a way you don't grasp, and a group that is incorrect and closed-minded (but is willing to engage in discussions)?
It would help if you were specific about what exactly you feared in order to avoid getting Field and Fortress'ed. But the usual FOOM doomsday scenario has the implicit assumption that a super-AI will magically appear fully formed in a world that otherwise looks exactly like today's. The whole thing falls apart as soon as you imagine a plausibly incremental development path.
The specific fear is that we build a strong AI before we figure out how to align strong AI. The assumption there is that alignment is a really hard problem. So one or perhaps several strong AIs are created. These AIs would be extremely intelligent (this being what "strong" means), and have the potential to gain lots of power and become even more intelligent. Eventually they own the entire economy and we humans have so little power that we have nothing to bargain with to ensure our own survival. At that point, the AIs kill everyone. (Probably they were pretending to be friendly before this.) The main assumption here (from my perspective) is that humanity is eventually able to build strong AIs, and does build them.
I don't think this picture is too dependent on a discontinuous development path. A very incremental development path would definitely help the chances of survival a little bit. It would give us more time to work on alignment for one thing, and we'd probably be able to learn quite a lot from the intermediate systems. But alignment seems so hard that we'll fail at it anyways.
For the "world that looks exactly like today's" component, there's definitely ways we could structure the world to make it less easy to take over by an AI. Mainly by making access to our computing power a lot more restricted. Possibly, if there's enough time before strong AI is developed, by developing nanotech so that we can come up with defences against the kind of nanotech that the AI would otherwise use to try and take us by surprise. Though I don't think the current world is headed in the "restrict computing power so an AI can't use it direction".
If that were the argument, then it would essentially just be a sci-fi remake of Pascal's wager, and I'd dismiss it for the same reasons that I do the original version. Yudkowsky has been very clear that he considers AI killing all humans to be near-certain, and that he expects it will most likely happen within the next decade or so.
Instrumental Convergence is one of the most fundamental concepts in alignment research. Any mind with a desire can be expected to develop predictable instrumental goals along the way, like "stay alive to achieve my goal" and "acquire what is needed to maximize success".
Warnings of the extinction of humanity don't come out of random possibility space; they come from the fact that most roads marked "self preservation" and "resource acquisition" lead to our demise when taken to their logical conclusion -- probably with intermediate destinations marked "deception" and "manipulation".
I agree with the foomers that AIs shouldn't be given access to a "press here to kill all humans" button.
Where I disagree with the foomers is the belief that any superhuman AI will automatically manage to MacGyver itself a "press here to kill all humans" button out of a text terminal and a pair of chopsticks, or something.
Maybe it can kill us quickly and easily, maybe it can't. There's a larger problem here, which is that unless you've designed your AGI in a provably-aligned manner, you can never be sure it isn't playing a long game with an outcome you would find highly undesirable.
Keep in mind that AGI could be patient and egoless in ways we find hard to relate to. It may feel little need to stick around and observe its misaligned plans come to fruition. By the time you’ve hit the kill switch, it might already be gone, with a chain of events set in motion that you are powerless to stop. The outcome might not be world-ending… but can you be sure it is one you will find acceptable?
A large chunk of the issue is that an AI superior in all respects to a human brain is, by definition, superhumanly good at persuasion (and one shouldn't assume that real AI will be unusually bad at this, since people are currently trying very hard to make AI good at persuasion for the purpose of winning culture wars and/or for advertising). It's one thing to say "well, I'll never let the AI out of the box no matter what it says", but what about your successor when you retire in 5 years? What about the random employee who's okay with a robot waifu and gets seduced? What about the "AIs are people too!" activist? You have to secure the AI not just against the AI directly escaping, but against the AI convincing someone to let it out.
I don't think a bunker would help. Long term, the sun would stop shining since the AI would have built a dyson sphere around it or perhaps dismantled it entirely, and most atoms in the Earth would be repurposed by the AI to build computers and various machines. It's not like conventional nuclear war where being in a bunker means you're safe and there's a hope of returning to the surface and rebuilding someday.
Yes. On this, at least, EY is probably right. If an intelligent entity with access to industrial-scale resources decides to make humanity extinct and is not effectively opposed, it won't just launch a bunch of optimally-targeted nuclear missiles and say "there, that got 90% of them, that's good enough". It's going to follow up, and people who built bunkers to escape the apocalypse will be an extremely high priority in that.
A better explanation is that in the process of making one successful AI, a million flawed throwaway AIs are created as an unavoidable byproduct, so that if there is any slight change that could turn your successful AI into an armageddon death machine, it *will* be made reality and there's not much you can do about it.
Foomism only works if you assume that a superintelligent AI appears fully formed out of nowhere. Any incremental development path means that the world is already filled with and accustomed to almost-as-good AIs that the would-be world destroyer will have to somehow compete with.
I don't know about Flynn specifically, but the evidence overwhelmingly suggests that money barely matters in electoral outcomes. Well funded candidates usually win because popular candidates get a lot of donations (and in most modern elections, *both* candidates are well funded), but in cases where someone with a lot of money jumps into a race, they tend to fail miserably (Meg Whitman and Michael Bloomberg come to mind).
Does it matter that those people were already well known and jumped in with lots of their *own* money? I'm not sure how well that maps to Flynn's situation.
A friend who flew to Portland to help with the campaign said this today: "The race is looking EXTREMELY tight - so much so that any individual who is able to reach 50 voters seems to have something like a 1/1000 chance of flipping the entire election"
I’m hoping someone here can clear up the controversy over the discovery of the double helix. From what I’ve heard, it sounded like Rosalind Franklin saw it first, but didn’t recognize it for what it was until after Watson and Crick had done so, which is why they are credited with its discovery. This impression of mine might be way off, if not incomplete at the very least. Would love a solid education on this piece of the history of science. Thanks!
Franklin was an experimentalist, and interpreted her data enough to understand that the sugar-phosphate groups were on the outside of the molecule. But that's a long way from understanding the structure of DNA, which rested on Watson and Crick's model-building and ultimate insight. Watson and Crick indeed used the experimental results -- not just of Franklin, but of many experimentalists -- to inspire their thinking, and of course ulttimately to confirm it. But theirs was the sole insight putting it all together. The controversy over Franklin's contribution is entirely manufactured to suit 21st century shibboleths.
It seems to be universally accepted that the double helix paper was the key step in understanding molecular genetics, but I have always found this puzzling. Long before that there was the determination by Levene that DNA consists of four nucleotides, the determination by Avery that DNA is the genetic material, and Chargaff's rules on the pairing of different bases. If I understand correctly, the precise 3D structure of individual nucleotides was also known, and was used by Watson and Crick. There was also previous work on X-ray crystallography of proteins which had revealed helical structures in other biological macromolecules. Then Watson and Crick used Franklin's data to show that show that DNA is also a two-stranded helix and to describe its detailed structure. This was certainly a valuable step forwards but it does not seem to me to be so much more important than other steps taken before or after that.
Beause the great puzzle of the day was (1) how can *so much* detailed information be encoded by such a simple molecule as DNA, and (2) how can it reproduce itself so stunningly accurately, with almost no errors? The answer to the first question was that it was really, really, long, and its unique structure made it incredibly stable for such a huge molecule[1], and the answer to the second question was that it exists as a pair of positive-negative mirror images, and you can reproduce one strand exactly[2] by using the other as a template. The double helix gave all these answers, and more, and solved a large number of fundamental problems in the biochemistry of genetics. That's why it's really important.
---------------
[1] We now know there's a complex packaging and folding arrangement of DNA in vivo that provides it with still more stability and resistance to chemical change.
[2] We now know there are a host of proof-reading and error-correction mechanisms that help sustain the ridiculously low level of errors in reproduction.
Watson and Crick had an idea about the structure, based on base pairing (AT, CG)
Rosalind Franklin, working in Maurice Wilkins's lab, got some X-ray crystallography data that confirmed it. Wilkins took the data and showed it to Watson and Crick.
Thanks for that link. I had a negative opinion of W&C, re the Franklin thing, but if she herself was reconciled with them I am not going to carry on being peeved on her behalf.
That would fit with the standard pattern of the theorist getting all the credit - see also the Higgs discovery, where CERN gets a nod for decades of experimental work by thousands of people and the theorists get the Nobel Prize for a few years of work by a half dozen people.
She was such an empiricist that she didn't believe Watson & Crick when they declared that structure. Similarly, she had disbelieved them the previous time when they had declared a different structure (and turned out to be wrong).
Do you think applying for the job at DeepMind would be wise? If a Friendly AI approach were developed that was actually good, but it was declared a trade secret and the researcher were contractually forbidden to disclose that IP owned by Google with anybody outside the company, all it would accomplish is to effectively delete that researcher from the pool of people who might contribute to solving the problem.
That's only a likely scenario if specific Google executives are insane. But how could I know?
There are two pieces of paper in this story: an employment contract that requires not disclosing trade secrets, and perhaps a paper that explains how to construct an AI that does what people want when the AI might be doing that more optimally than the people are. Which do you mean?
It is really hard to know if the latter one is correct or needed until it is too late, and it is hard to predict when would be too late. Most such proposals are garbage so odds are the next one will be garbage too. Caring about that makes no sense unless it happens to be correct and timely.
Caring about the former one makes some sense because Google has some thousands of times more lawyering available than any individual.
We are speaking in a medium with higher latency and effort than a telephone call, so try to avoid situations where someone has to guess what you mean in order to have a discussion with you.
I'm the lead of the Alignment team at DeepMind and one of the authors of the post Scott linked (thanks Scott!). My main reply is basically what Scott wrote below: there's already a culture of publishing, leadership cares at least a little about AI risk, and DM is one of the frontrunners for building powerful AI.
To that I would add that if somehow leadership was so stupid as to try to enforce a rule that safety research had to be kept private, at a time when it seemed important for other orgs to use those safety techniques or else humanity goes extinct, without some *really* good explanation as to why, I (and probably several others on the team) would leave DeepMind and spread the techniques anyway.
It is generally good to follow the law even when it seems to you like it would be bad for the world, for the standard reasons that when your consequentialist reasoning tells you to do something that violates deontology you should often be very suspicious and then not do the thing, but there are limits, and "gatekeeping alignment research that could otherwise save humanity" is a good example of a limit for me.
But I do want to emphasize that I really don't expect to get into this situation in the first place. I don't think leadership is on the same page as I am but they're not so wildly off as to gatekeep alignment research.
1. Most AI groups have been publishing everything interesting they do. Advances like transformers aren't trade secrets. I don't really understand the incentives that make them do this but they must exist.
2. DeepMind is run by very smart people who are at least a little concerned about AI risk, I don't think they're stupid/awful enough to make a good AI safety plan the exception to their usual publication strategy.
3. DeepMind is on the cutting edge right now and there's a pretty good chance that whatever AI most needs to be aligned will be coming out of DeepMind anyway.
All this idea that AI will take over the world and kill us all is absurd.
Take this thought experiment. An AI, will discover a need to expand it's facilities. The current facilities cannot be expanded, thus new facilities location need to be researched, purchased, permitted, engineered, designed, constructed, equipment purchased, installed, utilities, process chilled water, air, power ... You and I can't even imagine all the steps required to make this happen ... any AI isn't going to do this. Its stuck in step 0.0.0.1 not possessing that a solution to overloaded hardware is expanding to a new location.
If the AI is smarter than you, why do you assume it will be completely unable to solve a problem you consider hard? And if the AI is not smarter than you, then it still has a team of brilliant Google engineers working to make it smarter so it doesn't have to solve the problem itself.
I think if you assume the AI will never develop/be taught agency in that way then you're just assuming away AI risk without actually engaging with the arguments of the people who think it's likely to be a problem.
Again, you're basically assuming the AI just won't be that smart. If that's true then it won't be that much of a threat, but it won't be as useful either so people will keep trying to find ways to make it smarter.
The arguments here I think are A) that a lot of the things we humans do with our intelligence are basically unimaginable to less intelligent animals, so we should expect a superhuman AI to surprise us, and B) that if nothing else, an AI probably *can* do most of the same stuff that Google would do.
Exactly, and the process varies from place to place, there's a lot of judgment calls on somewhat nebulous stuff. What if your AI decides to build in say Donbas or Cernobyl because the price is right. Got some good prices on land in Bangladesh, hope high tide isn't an issue
Trade secrets and contracts impose stiff penalties for disclosure, they don't prevent it entirely. If the difference between disclosing and not is the literal end of the world, I would hope that someone would risk fines or even jail time to do so.
Actually, you do know exactly that, in some projects that we know of (as public) that they know they got some good results and they close the source. For example, take GPT-3, or Dota 2, there are many more...
Moreover, many google papers serve as an advertisement for company achievements and they do not get much scientific credibility. Most of the papers that I read are impossible to reproduce because of obfuscation in experiment details (IP secrets) or just because if they share the secret sauce the achievement will be practically close to useless.
On the other hand, Google and other, know how to write a good paper that convince the average person to think that they know what they talking about and AGI is just around the corner.
Source: am University researcher that works on AI/ML
But, I suppose, your point is that large organizations with no clear contrary mission tend to closed-source their AI projects and publish vague useless papers in general. I'm willing to believe this, but I'm not sure the claim is relevant. Does anyone here see closed source or irreproducible press-release type papers for for AI safety projects?
I don't think that the inventor of the knife cared about safety. As such the next step in AI safety will come from a place that is oblivious or is careless of the repercussion of its invention.
"What you have said about German professors is not exaggerated. I have got to know another sad specimen of this kind - one of the foremost physicists of Germany. To two pertinent objections which I raised against one of his theories and which demonstrate a direct defect in his conclusions, he responds by pointing out that another (infallible) colleague of his shares his opinion. I'll soon make it hot for the man with a masterly publication. Autoritätsdusel is the greatest enemy of truth."
I've seen some discussion about it, but Authoritätsdusel probably means the stupidity of authority.
"Was Sie über die deutschen Professoren gesagt haben, ist gar nicht über-trieben. Ich habe wieder ein trauriges Subjekt dieser Art kennen gelernt – einen derersten Physiker Deutschlands. Auf zwei sachliche Einwände, welche ich ihm gegen eine seiner Theorien anführte, und die einen direkten Defekt seiner Schlüsse darthun, antwortet er mir mit dem Hinweis, daß ein anderer (unfehlbarer) Kollege von ihm der selben Meinung sei. Ich werde dem Mann demnächst mit einer tüchtigen Veröffentlichung ein heizen. Autoritätsdusel ist der größte Feind der Wahrheit."
Not quite. Autoritätsdusel means something more like "blind trust of authority." Dusel means something like a lack of consciousnesses that makes you stupid here. It was used (though I think this is archaic) as a form of description of drunkenness for example.
I don't know if that's true of Einstein's dialect but for me, at least, it makes me think of something like, "being drunk on authority." Except unlike English the idea is that being drunk makes you stupid, not that it's euphoric.
Based on being a native speaker, "argument to authority" is correct, if a little dry. It is more emphatic, such as "unreasonable/confused love for authority".
The Guild of the ROSE is an online-focused attempt at Rationality training and self-improvement, building on the foundations (and learning from the mistakes of) similar orgs.
We place a strong emphasis on accessibility, community, and ethics, with a bias towards action (actually Doing Things with one's self improvement journey).
Any and all constructive feedback is welcome! We are looking to grow and gain more members from outside the Rationalsphere.
This looks interesting to me, and I'll give some feedback:
- The name sounds great - but some of your wording makes it sound cultish. "Practitioner's Path", "Council", etc...
- Whoever did your website's design did a really nice job. Except that it doesn't have an RSS feed for your articles. :(
- I'd 100% be on board to sign up if I was a bit more social and felt I could commit time-wise. Or at least that's what I thought until I realized the cohorts were a premium feature and I could join the cheaper tier and ignore the social aspects to just get the courses.
- I tried to register for the free trial with Discord and the link gave me an internal server error (your side, not Discord's).
- I then tried with email/password and it made me join a paid tier instead of giving me the free trial.
- The public retrospective articles are a nice idea, especially in how they document what areas can be improved and not just highlighting the positives.
Just looked at the site. Boy, I am picking up a cult vibe. I'd be expecting to hold two cylinders hooked to an ohmmeter ala the Scientology Racket at the first meeting. This is aimed at people into rationality? My opinion only, but that's how it hit me.
I've been informed that we just rolled out an RSS feed, actually; you can find it at the bottom of our 'Articles' page.
Oh, also, the free trial is for the Flower tier; we don't have a free trial for the Seedling tier. We can get you upgraded to that free trial, and at the end you can decide if you want to move to Seedling or not. Just let us know how you'd like to proceed by emailing us at council@guildoftherose.org
You may want to add an HTML <link> tag with the RSS feed to the website's source so it can be picked up more easily.
No worries on the free trial - I'm fine with the seedling tier and 2$ is worth it from what I've seen. I'd like to see where you take this in the future.
Cool! Thanks for the input; we are technically still in Beta, so feedback is important! Don't hesitate to reach out with your thoughts via email or scheduling a Calendly call. :)
Yeah, we intentionally leaned into that aesthetic as a form of counter-signalling. It's the places that pretend to not look like cults you really wanna watch out for, and we have aggressively avoided the failure modes of actual cults by putting all the direction of self-improvement in the hands of our members, and by making our organization as low-commitment as possible.
An RSS feed is in the works! Thanks for the +1. I'll pass on the compliments and feature suggestion to our web dev. :)
Oh no! Sorry about that issue, we just rolled that out the other day, must be a bug. I'll make sure we get that free trial applied.
If you're intentionally counter-signalling that you're not a cult by trying to look like. a cult then... I don't know, I feel like you're putting too much thought into this cult thing.
To be fair I don't think you're actually a cult, more just a decentralised network for your members to form their own cults.
> It's the places that pretend to not look like cults you really wanna watch out for
Are you saying that the places that exhibit all the signs of being a cult are actually less likely to be a cult than the places that do not look like a cult? That would make my local chess club a cult, and scientology not a cult ?!
The response below makes it seem pretty culty to me. Counter signalling is overrated. Also, the fact is that the best manipulators tends to sound quite reasonable until they've got you under their control, so the more reasonable argument is that you can't make a reasonable guess on whether something sounds like a cult or not based on signalling.
More importantly than whether this is a cult, there website doesn't inspire any faith that you will actually improve your life.
Well, okay, the primary driver of youth social interaction and skill growth is forced socialization through family or school and whether you got lucky in your "place". So maybe being part of the 8 person groups that are assigned by an authority could help people with socialization troubles or something.
Well, if snapping at trolls "seems culty", then I'm happy to join that cult. As long as we get to wear cool cloaks.
In any case, your comment is much more reasonable in tone and content. As to your objections, fair enough! We can point at testimonials, but right now we are still collecting data. If anyone wants to try us out to see for themselves, anyone can request free access. You can also examine our content and audit it, as it is publicly available, although the main driving force of self improvement is the community.
There's subtext to my statement that I'm happy to explain, but seeing as the other conversation partner caught it just fine from the context and you are striking a combative tone, I suspect that it would be a waste of my time.
Having read the last open thread and the first book review and having thought about it a bit, I now agree that it would be very interesting to see the complete ranking of the book reviews. This would be helpful to decide if I should read more than 12 book reviews.
"‘Desperate Remedies’ Review: From Asylums to Zoloft Psychiatry’s goal was to transform the treatment of mental illness via science—but the results have been anything but conclusive." By Richard J. McNally | May 13, 2022
Ours is a time of historical reckoning for many fields, and psychiatry is no exception. An indisputable masterpiece among a flurry of reappraisals is Andrew Scull’s “Desperate Remedies: Psychiatry’s Turbulent Quest to Cure Mental Illness”—a comprehensive, fascinating and persuasive narrative of the past 200 years of psychiatry in America.
The author, an emeritus professor at the University of California, San Diego, begins his narrative in the early 19th century, with the establishment of institutions to house members of affluent New England families who were mentally ill. These well-staffed asylums, situated in bucolic surroundings, featured therapeutic activities including singing, acting and French lessons. Social reformers such as Dorothea Dix galvanized support for public insane asylums as alternatives to poorhouses, prisons or homelessness. Optimism abounded as reformers predicted cure rates as high as 80%.
By century’s end, it became obvious that such predictions were wildly inaccurate. These asylums included many patients with conditions that carried bleak prognoses such as senility, mental handicap, dementia praecox (schizophrenia) and general paresis of the insane (GPI), a mysterious, progressive disease characterized by delusions, hallucinations, mood disturbance, apathy and dementia, culminating in death. The reputation of psychiatry reached its nadir by century’s end. Asylum superintendents functioned more like prison wardens than directors of therapeutic institutions. Therapeutic helplessness in the face of intractable psychopathology, Mr. Scull makes clear, was professionally demoralizing for psychiatry. The rest of his book traces the history of psychiatry through its next century, when great hopes gave rise to daring treatments of varying effectiveness, and the asylum system largely came to an end.
Mr. Scull has a background in sociology, but differs sharply from critics of the field who once proclaimed that mental illness was nothing but a myth, or claimed that diagnoses merely were stigmatizing labels imposed upon people who failed to conform with societal norms. Indeed, the author emphasizes the intense suffering experienced by individuals with major mental illness and acknowledges the challenges faced by their families and by the psychiatrists struggle to help them. Yet he is unsparing in his critiques when motives of money, power and fame have tempted psychiatrists to disregard the welfare of those under their care.
* * *
Mr. Scull’s major concern is for patients diagnosed with schizophrenia and bipolar disorder, for whom psychiatrists are the primary treaters. Although medication remains the mainstay of treatment, it has become increasingly apparent that as many as 67% to 80% of patients either fail to respond or cease taking medication because of intolerable side effects. Moreover, clinical researchers have confirmed that patients who are able to remain on antipsychotic medication commonly develop cerebral atrophy and tardive dyskinesia, an involuntary movement disorder. Despite upbeat advertising campaigns for new drugs, there have been no true breakthroughs in psychopharmacology for more than a half-century. Indeed, pharmaceutical companies are now abandoning psychiatry.
Mr. Scull closes by asking: “Does psychiatry have a future?” His melancholy question partly reflects the failure of advances in genomics and neuroscience to yield actionable insights that improve the lives of people with major mental illness. Indeed, the humbling conclusion of these advances is to underscore just how complex the genetics and neuroscience of mental illness really are.
Given how genetically loaded most mental illness is, it's possible it will be cured via gene interventions before it is understood. As a result, it's possible the data necessary to understand these conditions will disappear from the earth at the same time as these conditions do.
One mental illness was cured in a different way when it was discovered that syphilis was the cause. Barry Marshall & Greg Cochran would argue that many more conditions have pathogenic causes we just haven't discovered yet.
Not an optical observation, but it should become a compact low flux source of fairly exotic high-energy particles and radiation, as gas and dust fall into it. I mean, this is how black holes were originally discovered (very compact X-ray sources).
Telescopes would notice that Pluto has turned invisible. They'd notice that Pluto's moons are still orbiting something where Pluto is supposed to be. Concentrating all of Pluto's mass into a point might have a subtle effect on the moons' orbits, but I'm not sure.
> Concentrating all of Pluto's mass into a point might have a subtle effect on the moons' orbits, but I'm not sure.
I don't see why it would; as far as I'm aware, Pluto's gravitational field should be unchanged in any region of space not containing part of [the original] Pluto?
We model gravitational bodies as point masses, but that simplification assumes the body is a perfect sphere of uniform density. The gravitational field around real planets is non-unifom; some parts of Earth's crust are denser than others, and this has real effects on satellite trajectories. If you replaced pluto with a black hole, the gravitational field in the volume outside pluto's original volume would become more regular.
Now, would this noticeable effects on Charon? No, probably not- at least in the short term. The gravitational variation gets smoothed out with distance in the first place, and Charon is farther from Pluto than a satellite from Earth.
In the long term, it'll affect the time-evolution of Pluto and Charon's orbit. Two spinning bodies orbiting each other will eventually tidally lock due to tidal effects slowing their rotation. Charon will still feel tidal effects from BH!Pluto, but BH!Pluto is a genuine point mass, so presumably the tidal-locking would slow down, conversely to how the slowdown of the earth's rotation is sped up by all that sloshing water.
Small nitpick. The point mass simplification only requires that any given "shell" in the sphere is of uniform density. The "shells" can differ from one another in density.
(In this analogy, ever atom not at the rotational axis of Pluto is the sloshing water, creaking and flexing with tidal forces, sucking up rotational energy and converting it to heat.)
(Actually, Pluto and Charon are pretty small, so they might already be tidally locked. If so, it's a moot point, haha!)
They are, indeed, mutually tidally locked. Wouldn't make it moot, though; turning Pluto into a black hole would mean that its tidal lock to Charon would break (though Charon's tidal lock to Pluto wouldn't) and therefore Pluto-BH's day could drift apart from its month (via perturbations to Charon's orbit, which currently transfer part of their effect to Pluto's rotation rate but which would no longer do so if Pluto became a BH).
No static black hole can do that from Pluto's distance. Anything small enough to radiate strongly enough to matter is too small to do it for very long.
Pluto would apparently take 10^42 years to evaporate as a black hole, which corresponds to a negligible luminosity. The heuristic I use is that anything bigger than Chicxulub doesn't evaporate fast enough to matter.
(This is if Hawking radiation is even real; it's never been confirmed.)
True that it's never been directly observed, but it's a pretty direct consequence of qft in regime where we are pretty confident it's valid (there is no quantum gravity involved and it's really similar to the Unruh effect that only require special relativity)
Hawking radiation not being real will require to profoundly modify qft and I am not even sure it's possible in a way that preserve all observations in favour of qft
If the mass is the same, then the lensing effect should be the same as well. You might see stars where previously pluto would have obscured them, but it's such a small area I'm not sure we could detect it.
Stellar occultation data did tell us that Ultima Thule, the trans-neptunian body flown-by by New Horizions was a funky shape. (the flyby revealed that it looked like two potatoes glued together) But it was near the limits of our capabilities so we're talking about resolutions on the order of kilometres.
I think objects like Oumuamua might have the oblong shape because of a high velocity relative to the dust clouds they formed from. Methane ice or whatever would just be sintering together on the leading edge of it.
The diameter of pluto is 1500km and that was about 5 pixels wide on Hubble. So the gravitational lensing around that 1cm black hole would be way too small for hubble to see.
Flynn also got a bump in the Portland Effective Alturists & Rationality group some months back.
I saw that Flynn focuses on biosecurity, so I asked the promoter whether Flynn supports gain of function research or not, and I never got an answer. More specifics on his "pandemic prevention" platform would be great, because I'm pretty disappointed in the depth of content right now.
I'm pretty close to Carrick and pretty sure he would say something like "probably don't do gain of function research, definitely don't do gain of function research on potentially pandemic pathogens."
If you are considering donating, voting, or volunteering and have questions you can always email reachout@carrickflynnfororegon.com and they will respond to you.
Currently watching the final of the Eurovision song contest. Norway are on, and it's bananas. Literally. The song is called "Give That Wolf A Banana" and they're two guys in suits dressed as yellow wolves.
Look, I can't describe it, you have to see it.
There's a lot of (boring) ballads made it to the grand final this year, and while I'm expecting Ukraine to possibly win (due to sentiment, not the song) at least we have a few brave souls keeping the Eurovision tradition alive.
Australia's costume is so elaborate, there was a hold-up in the second semi-final while they got him on-stage.
Eurovision entries are limited to members of the European Broadcasting Union, it has nothing to do with the European Union. That is why Israel enter it.
Australia is part of the European Broadcasting Union, we have a public TV channel that broadcasts their shows. This technically qualifies us for Eurovision under a similar loophole to how Israel qualifies. And it is insanely popular here, so the costs of adding us was worth it.
Apparently the country known for drag queens singing opera on top of a bus racing through the desert turns out to grok the zaniness of Eurovision.
Australia's independence was very gradual; people point to 1901 as the date of federation, but all the federation actually did was make us one big colony instead of a bunch of little ones - we were still a dominion of the Empire, UK Parliament could still overrule us, and we didn't have the right to separate armed forces. *Complete* constitutional independence didn't happen until 1986 with the Australia Act (we still have the same monarch as Great Britain, and she does have a number of reserve powers, but that's a personal union i.e. the Queen of Australia happens to be the same person as the Queen of Great Britain, but if Britain abolished the monarchy she would still be Queen of Australia).
There are still a lot of cultural ties, though, and it's not particularly weird for an Australian to still consider Britain "the motherland".
Yes, and yes. I have no idea how it happened, it seems they liked watching Eurovision so much that the organisers decided to let them have a go, and for the past couple of years they've been entrants.
There was a tentative incursion into the Chinese/Asian market a while back, with China broadcasting it, but then there was way too much gay so China banned it and nothing more has come of that.
The good songs never win. A Eurovision winner is that delicate blend of slightly outdated (but not much), cheesy (but not too much) and a banger (but again, not too much).
> Somewhere in your head there is a microphone. It produces a little voice inside of you, whose approval you desperately crave. You would do anything for the voice to like you. Ghosts, mental models, and personified abstract concepts fight each other for a turn at the mike and the right to implicitly control your actions. Who wins?
I found Scott's recent comment rather bizarre. Is this really what some people experience?
It doesn't sound as outlandish as all that. The little humorous meme of the angel and the devil sitting on our shoulders vying for the steering wheel has achieved universality for a reason.
I suspect back in the olden days a lot of those inner dialogues were indeed subsumed into the framework of religion, as ancestral advice, as prayers for guidance, as temptation, etc. The Philokalia, for example, treat extensively about guarding the intellect against specific thought patterns outright personified as demons while seeking communion with the angels and saints.
I can’t say I doubt any of them. The issue is picking the one I want to listen to. I’m definitely getting better at not listening to the ones that don’t like me. They should be thrown out of the choir
I don't think that Scott was attempting to provide an accurate depiction of the qualia of being a human, but summarizing a view on what drives people, that they only experience (per this theory) in an obfuscated way.
Adam Smith's Impartial Spectator from his "Theory of Moral Sentiments". There's a great book on Smith's book by Russ Roberts called How Adam Smith can Change Your Life...
Let's say you're the Japanese government, you see the cratering birth rates and you decide to look for a solution in tech, and so you create a government-mandated dating app/matchmaking service. What would this be like? Assume you can mandate all singles to use it and that you are able to accurately assess any citizen's relationship status, and your goal is to get as many people into happy, long-term relationships as possible.
The critical thing would be to require everyone to report on the results of each date, and especially ones that led to a long-term relationship. That would give you the data to design an app that actually worked, based not on what people say they want but on what they actually want as demonstrated by outcomes.
The tech solution would be to shut down all ISPs, cell phone powers, and TV broadcasting. Without those, people are forced to go outside and interact with each other again. Eventually, the problem is solved.
Both. Take away TV, internet, and cell phones and eventually all those couples are going to get real bored.
("but birth control!" No one's going to have perfect birth control use across the entire population...and having kids also solves the boredom problem for awhile)
You scoff, but the post-blackout baby boom is a well known phenomenon.*
Likewise, in India, when looking to bring down the birthrate, one of the things they either tried or floated was getting tv to poor people. (Not sure which, my googling failed me.)
At the least, it's a common enough idea that it shouldn't simply be laughed at and ignored.
*Three of the first four links I saw said this wasn't actually true, with the most positive calling evidence mixed, so this probably isn't a real thing.
I have read that low fertility in Japan is partially the result of women's difficulty in having both children and a career. Assuming this is true, what if having a husband in the same industry would make re-entry into the workforce (once the youngest child reaches a certain age) easier? Wouldn't that slightly reduce the disincentive to have a child?
It might also increase the likelihood of the couple running their own business. And parental leave is probably simpler for a co-owner than a replaceable employee.
If this were accurate, then your hypothetical app/service could match people in the same industry.
Medieval guilds in Europe tended to be run on Distributist lines, that is - guild members would cooperate and set rules and quality standards, as well as maintain some social safety nets, but they were free workers who owned their own tools and labour. See https://practicaldistributism.blogspot.com/2019/06/distributism-better-alternative-to.html?m=1 for a more detailed discussion.
Fertility is about culture, economics and overall social perspectives. Let's face it, for many people in relationships - myself included - the world is too interesting to spend the prime of my life caring for a child. Ironically, longevity tech might make people more eager to have children, since they'll know they still have a life to look forward to after the whole thing is over.
I have no idea about Japan, but in the West raising the baseline mental and physical attractiveness of people would go a long way towards letting people find relationships. I'd imagine various forms of social anxiety cost ~0.2 of the birthrate by themselves. Add impulse control problems, violent misogyny/misandry, substance abuse, all the usual dysfunctional undateable person stuff, mix in some fixable physical issues like obesity, and suddenly the dating pool crisis resolves to a health crisis.
Life expectancy has been doing up, as well as lengthening of the "reproductive window" for women (IE more women have been having children in their 30s and even 40s compared to teens and 20s). Fertility has still gone down, though, because they usually don't have as many children - women having children in their 30s usually only have 1 or 2 children.
Which makes me wonder if we could improve the fertility rate just by making it easier for people to form households and have children earlier. That's harder now for most folks, and it's something we've known can reduce fertility for centuries (the Northern European Marriage Pattern of 1500-1800 involved people delaying marriage into their late 20s, so they could earn more money to form a household).
I have an easier idea. Government mandated childcare and actually enforce the laws against women being let go or not promoted for having children. The conservative Japanese government's desire to keep a 1950's, man earn and woman stays home family effectively forces women to choose between career and family. Many choose career. Alleviating that choice will do better than any dating app.
Anyway, since I'm a government I don't need the app to be profitable. Japan has a tradition of group outings with an even number of men and women to feel out dates. You would enter your information and any dealbreakers (not a wish list, deal breakers only). Then we'd match small groups of high compatibility people. Maybe ten people each, five men and five women. And you'd go on these meets.
Each person would have a set number of tickets (maybe 20 a year?) which give you free food/drinks at the event. You scan them in the restaurant and the government gives them some fixed amount. If you don't use them by the end of the year you have to pay face value. If you and a member of the opposite gender register as dating the tickets convert to one on one experiences like going to theme parks or bookstores or whatever. And after marriage and each childbirth you receive (in addition to covered healthcare costs) a one time cash gift from the government. Find some Shinto tradition or something to make it feel cultural.
Also, maybe throw in free (ie, government paid for) gym memberships and health/style coaching.
Almost all of them. The most fertile country in the world by TFR is Niger. Niger has paid leave for mothers and laws outlawing discrimination against women including for pregnancy. Mexico, which is just at replacement rate, does too.
That chart also doesn't show government reimbursed childcare or anti-discrimination laws. It's just about family leave.
Man, that's interesting. Normally poor countries have poor welfare provision. How does paid maternal leave in Niger work? How can a country with such a low GDP per capita afford that?
Giving people a few months of paid leave scales to income. If your company pays someone $800 a year then three months paid leave costs $200. Outlawing discrimination is free. Though of course the question of how well this is enforced is another question. I expect, as it is in many places, it's more available to more professional workers in better regulated and more visible industries.
Niger's GDP per capita is $600. Mexico's is $8500. Japan's is $40k.
What reasons do we have to believe that introducing any of these improvements will push Japan above 2.0, given that other nations with a high GDP per capita and strong pro-parent legislation have failed to do so? I mean, sure, I agree that they're a good thing to have, but would they actually solve Japan's predicament?
You've moved the goalposts now. You've gone from, "these things don't help at all" to "these things don't help enough" when I pointed out high fertility countries often do have them.
Japan's birth rate is low even compared to Denmark (to pick a country). While these changes might not bring it up to the levels of Niger there's every reason to believe it could raise it from 1.3 (Japan's) to 1.7 (Denmark's). That's halfway to replacement and not insignificant.
Anyway, the best way to get your TFR is to import high fertility immigrants. Israel's been getting a lot of attention but the reality is Arabs and immigrant Jews are a hugely disproportionate amount of their TFR. Of course, the Japanese aren't exactly immigrant friendly.
Do we have examples of TFR significantly increasing in a given country after pro-parent policies were introduced there? I'm willing to update my beliefs in this regard if that's the case.
The app would not be "dating" per se, instead, the app would arrange a time and place for users to meet and work on a hands-on project together, very little technology used. The users would meet regularly for planning and for work sessions, meal breaks included, perhaps travel too. Perhaps larger mixed group functions would work better too. Somehow it ends up in casual meals that lead to after-work drinks and dancing.
I doubt that dating apps actually increase the birth rate. Or perhaps actually existing ones made by for-profit companies don't, and one made by a pro-natalist government would have to work very differently.
One obvious issue: a for-profit dating app benefits from people continually using the service, not permanently pairing up and deleting it. Fertility appears nowhere in the bottom-line of such an app. A government might directly aim at fertility, and not have to use an app to do so. So if Hanson is right that fertility is associated with low status, a government could reserve its highest status jobs only for people with above-average fertility.
In the USA, other factors being equal, the less of an age difference there is between the bride and groom the less the likelihood that they will later divorce - so a list of nearby potential spouses that are very close in age should be helpful
Getting people to go on dates isn't the hard part, it's getting them to settle down and have kids. And then persuading them to have two or more kids, despite the fact that they probably live in a tiny apartment.
In most places, you're right, but Japan actually does have a problem with a relatively large minority of young people utterly uninterested in romance with another real human
I don’t think dating is the problem. I think someone should do some really serious research into how women feel about having children their concerns about having children and what might encourage them to want to have children. I know it’s picked at, and maybe there is a really serious body of research there that I’m not aware of, but my belief is that’s where the heart of the matter is
Women are clearly interested in having a child to replace themselves -- the TFR rarely falls below 1. What they are not necessarily as interested in is bearing some man's replacement child as well.
The reason people most often cite is the cost of children and the difficulty of childcare. Japan has a high cost of living and historically very bad support for working mothers, so not surprising.
The central problem with dating apps is that the most appealing people using them will either (i) not be interested in an app-deleting relationship; (ii) have something about them that most people don't like once they learn it; (iii) both. This becomes increasingly true with age. People disproportionately try to get dates with the most appealing people using them, realize they are lemons, and are disappointed. I guess the mandate stops them from deleting the app but are you going to mandate that people "swipe right" at least 10% of the time? Mandate that they go out on max(3, num_of_matches/10) first dates per week? They're still going to end up disappointed by nearly everyone they meet.
I would suggest not using a "dating app" at all but trying to get some of those declining lines on the infamous "how couples met, by year" graph to recover. Dating apps as they exist have significant advantages too, and why fix what is already somewhat working? Young adults should have more friends; if you are willing to mandate things, why not mandate they join a coed sports league? High schools should have more dances, and colleges should have some similar "get dressed up for this fancy party" events. Research indicates that coworkers dating is already kind of normal in Japan, which is an advantage.
Thank you for engaging with the question. I'm not sure I follow your reasoning though: you're saying that people go for the most attractive people on the app, but then realize they don't like them? This seems less of a problem than "people go for the most attractive people on the app, so everyone competes for the same 10% of users" because 80% of the remaining 90% is not going to be able to get dates with the top 10%, even assuming the top 10% doesn't just date each other.
So managing expectations is important. My first idea was to make everyone's visibility inversely proportional to their popularity. If you're a single pop star and everyone wants to date you, the app should only make you visible to a tiny percentage of users, or perhaps excuse you from using it at all, because clearly you'd have no problem dating anyone. By excluding or deemphasizing the most attractive people, the average user is less likely to a) develop unrealistic expectations about their dating prospects and b) get demoralized by repeated rejection by people out of their league. And A compounds B, of course.
I would hope that, on top of this, the matchmaking system makes at least some effort to bring compatible people together, and perhaps provide an incentive to go on first dates depending on how likely it is that you'll get into a relationship based on proximity, interests, and personal history. I think it's too pessimistic to say that everyone ends up being disappointed by every date the app provides them: how does anyone date in that scenario, app or no?
I agree one dating problem is overestimating the attractiveness of the average person (especially without makeup) and setting expectations accordingly. But I don't think an online dating app can fix that just by restricting your options by level. Too easy to notice attractive people in day to day life, plus Netflix. If anything, at least the dating app is showing a subpopulation of (mostly) single people.
No need for an “app”. Just announce that all Japanese families with 6+ children will get a payout of $1m tax free as soon as the 6th child is born. Russia has tried a similar payout program but the amounts are much lower and you get paid starting from the 2nd child onwards
Hmm. If you are trying to get your children well into the professional class, in the US at least it's going to cost you more than 1/6 of a million per child, including special tutoring, good private schools at all levels, university, and graduate school. And that's not including the opportunity cost of raising them. (Simply attending Harvard is $51K per year, not counting room and board. $74K for the whole deal, per year. So the next best thing to $300K.)
You ask why am I postulating Ivy League and its prices? Because we have more and more of a winner-take-all system, and if they merely graduate from a good state school, they may well wind up asking people whether they want fries with that, even if they study something in high demand. If they go to a merely good college, they won't make the contacts needed to get into the class where inflation-adjusted income is going up, rather than unemployment and outsourcing. And they won't get recruited for anything that will put them on track for a really good income.
OTOH, the average in-state annual cost is "merely" $25K, and if your local schools are decent, you probably won't need to pay for elite private schools to get them a chance at admission. But that's still $100K per child just for undergraduate. $600K total, before feeding them, clothing them, or sending them to grad school - which they'll probably need just to get a fairly good job. And doesn't count the cost of someone tending them - either one parent stops working, in which case the $1 million should replace their salary, or someone is hired to do that job.
I don't know the equivalent costs in Japan. They may be lower. But OTOH, I also assumed lots of people want 6 kids enough that they'd accept breaking even - I don't find that especially plausible.
How many Harvard graduates who go on to get JDs and join an NGO litigating for environmental justice in the Congo does the world need? I hazard the present supply is probably more than adequate to the demand.
A solution that works for people who make up the class of welders, truck drivers, CNC operators, plumbers, bank branch managers, Chik-Fil-A frannchisees, airline pilots, nurses, respiratory therapists, fifth-grade teachers, farmers, and car salesmen would be far more useful.
What's the lifetime TFR of couples who only want to have kids if they have enough funds to pay for their future Harvard degree? Somewhere around 0.5? That group is basically a lost cause when it comes to government programs.
More interesting perhaps is the TFR of those who only want to have kids if reasonable diligence will allow those kids to live at least as well as their parents.
But *why* they aren't pairing up is the key question. If women (largely correctly) assume that their bosses will tank their career prospects and their husbands will expect them to do all of the house work and child care, that doesn't strike me as an unreasonable aversion.
I agree, but that means L50L is wrong about enforced dating not helping with the birth rate problem - if a large chunk of the low Japanese birth rate is that there are social costs to women dating, then removing the option to not date would solve at least that large chunk (it'd also abrade those social costs over time, as socially punishing half of society for something they're legally required to do isn't sustainable).
Does that mean we should do it? Probably not; there are serious moral problems with it, like with most ideas lifted from a hentai plot. I'm merely disputing L50L's claim that "it wouldn't help much".
Is sex drive correlated with number of children in rich countries with easy access to contraceptives? (That's a genuine question, I wouldn't be shocked either way).
This; the problem is that it's incredibly hard to raise children in most rich countries nowadays. People *want* children, but can't afford the house, or work is all consuming, or childcare is unaffordable, or extended family support is unavailable, or social mores have changed to make child rearing way more labor intensive (my parents' idea of parenting was, broadly, to kick me out of the house until dinner, feed me, then broadly ignore me as I read a book/played games. It was a happy, loving childhood! If I sent my daughter outside unsupervised now, I'd have the police called on me.)
Humans should be sensitive to local conditions, not global ones. Even if we're close to the carrying capacity of the entire Earth, resources are still superabundant in developed countries, meaning that if you hypothesize that humans adjust their birth rate to the environment, then birth rates should actually be increasing.
"The sheer scale of Africa’s demographic momentum may come as a surprise. This results from the fact that a long-forecast demographic transition has been happening in Africa at a far slower pace than was expected even a few years ago."
That would be more persuasive if birth rates tended asymptotically to stability, which is about 2.1 children/woman/lifetime. In fact in First World countries they tend to levels that will result in exponentially *decreasing* populations -- replacing unsustainable growth with equally unsustainable decline. (Wkipedia suggestst the average TFR for "high income countries" is 1.5 right now, and it is generally declining still further.)
OK. Well *that* argument would be more persuasive if there were signs that different countries follow very different trajectories, or if there were signs that within some large heterogeneous region (such as the Eurozone) there was on average a tendency towards replacement even as some countries were higher and some lower.
I am of course not saying you're wrong. Only that there is no empirical evidence to support your hypothesis.
Carrying capacity is contingent on the level of technology and social organization (and has correspondingly gone up and down in the past, mostly the former in recent centuries).
Furthermore, Malthusian conditions ruled for most of history. You need something else to explain the change, like birth control or improved education.
I disagree that humans never got close to carrying capacity. Carrying capacity didn't change that much between the Neolithic & Industrial revolutions (those two things increased it, and that's about it). As populations grew death rates increased from famine, disease & war until an equilibrium was reached.
I think it's more a response to crowding than to total population, but maybe a tendency towards crowding is a result of this high a population at this tech level.
I'm putting together an introductory college English class with an "Intelligence: from IQ to AI" theme. I'm looking for recommendations of fiction (especially short fiction). Stories featuring collective intelligence (including prediction markets) would be especially welcome! Thanks in advance.
Thinking about it, it seems that this is partly a question about First Contact, and, beyond that, human ability to recognise behaviour as intelligent when we see it. The Turing test after all, as I understand it, is not about intelligence per se, but rather about the appearance of an intelligence which corresponds to human norms. If an AI were intelligent by another definition is it possible that we would fail to realise it?
There are a number of classic book-length treatments of such ideas: Fred Hoyle's The Black Cloud features an alien intelligence so different from us that attempts to communicate with it kill those who try. CS Lewis's trilogy, especially the first book, features angelic beings whose intelligence is on a completely different level, or of a completely different type to ours.
I'm reminded always of Wittgenstein's aphorism that if a lion could talk, we wouldn't be able to understand him. But not many short stories there, I'm afraid.
Useful insights. TBH, I hadn't even planned on adding aliens to the mix, but I may want to reconsider. And maybe AGI is basically aliens (orthogonal values, etc.) anyway?
One interesting characteristic of SF writing about the dawn of the Computer Age was that computers (giant brains really) would be much clever and more intelligent than us and answer all the unanswerable questions. In Asimov's story "The Last Question" a group of scientists ask a typical question: is there a God? To which the computer replies, in a flash "there is now!" I think there were quite a few golden age stories along these lines, (mocked, of course, by Douglas Adam's in HHGG).
They're not exactly about "collective intelligence", but "Bookworm, Run!" and "True Names" are two classic short stories by Vernor Vinge that fit your theme to a T.
And the reverse, not exactly about intelligence but very much about prediction markets, from Daniel Abraham (co-author of The Expanse series):
Back in the 70s[1] we used to think about natural but alien intelligences, what it would be like to meet a species which was just as intelligent as ours but...thought differently. It was a very interesting exercise, throws light on the many unexamined assumptions we normally carry around with us about what "intelligent" and "reasonable" mean.
As I recall Larry Niven wrote a bunch of really good stories that made such encounters come to live. "The Warriors" is his short story about the first meetin between humans and kzinti, and focuses very much on the different assumptions equally intelligent but quite different species make about each other. "The Soft Weapon" introduces Pierson's Puppeteers, who are intelligent but natural cowards, almost the polar opposite of the kzinti. I think you can probably find both online.
Novels he wrote where the difference between how intelligence species think is well explored and crucially important include "Protector" and "Footfall".
-----------------
[1] Perhaps because we were far more optimistic that this might actually happen, some time.
Buried in 'The Garden of Forking Paths', one of Jorge Luis Borges's most memorable stories, is the Chinese narrator's bizarre claim about his motivation for spying on the UK while living there during WWI. He tells us that he was spying for Germany not because he wanted Germany to win the war, but because he wanted to prove the competence of Chinese spies to a German spymaster, who is said to have doubted them. And to prove that competence, the spy sends a secret message by one of the most baroque methods ever dreamt up.
Many people who reflect on the story probably think about the many-worlds interpretation of quantum mechanics. For me, it's also about the lengths an intellectual might go to so as to prove to themselves (and others) that they really are as clever as they aspire to be.
Permutation City by Greg Egan is an exploration of the idea that consciousness is purely a function of information patterns, such that with a suitable encoding a cloud of interstellar dust might be conscious - or even mathematically identical to the mind of your neighbour
I'm not entirely sure if I understand what you mean . . . you think this course is a risky choice? Maybe?! I teach at a small junior college in Montreal. For now, at least, this doesn't feel like much of a risk to me.
If you're going to get into AI you might do well with his "I, Robot" collection of short stories that explore all the curious manifestations of his "Three Laws".
I'm not sure how well these have aged with respect to discussions of AI in its present (or likely future) incarnation(s). The stories are fun, but I don't know if I'd have a lot to say about them.
I think where they are interesting is in the question of control. Asimov's Three Laws were designed to restrict robots from becoming dangerous, because his robot novels are built around the assumption that people would find them deeply frightening -- for exactly the same reasons people worry about AI risk now, that a machine that could act faster, think faster, and think flawlessly would be exceedingly dangerous to human beings if it were not firmly controlled. The Three Laws were designed to do just that, to institute perfect 'alignment' in the AI-risk crowd terminology.
What Asimov's stories explore, then, is what actually might happen if you had such a thing. Do they work the way you thought they would? What are the unexpected consequences? That might be a very interesting discussion with respect to hypothetical dangerous AIs and hypothetical "alignment" efforts. Suppose your "alignment" effort does exactly what you hope it will. Does that solve the problem? Without introducing unexpected new problems? Maybe not!
You make a good point. The Three Laws stories would make a useful bridge from King Midas/Monkey's Paw style problems (be careful what you wish for) to real life instances of AI engaging in perverse instantiation.
The classic is obviously Flowers for Algernon by Daniel Keys, but two others that come to mind are Peter Watts's Blindsight, which posits an alien race with intelligence but not consciousness, and Adrain Tchaikovsky's Children of Time, which depicts an uplift programme gone wrong and producing human-level intelligence in insects. The Culture novels of Ian Banks make a lot of use of AIs, especially to investigate moral issues. I can't think of any short stories immediately, but somewhere in sixty years of reading SF I must have come across some. I'll add any I think of tomorrow.
Yep, FfA (in short story form) is in the course pack. I love me some Culture novels, and CoT was great too, but they might be a bit long for my purposes. Do let me know if you come up with some shorter suggestions--thanks!
This is unfortunately in no way short, but Vernor Vinge's "A Fire Upon the Deep" has some fascinating ideas about intelligence and what it means to be intelligent. There are insanely powerful and alien super-intelligent beings, who can construct entire mortal minds the same way a human might build a machine out of scrap parts. The maximum possible level of intelligence depends on where you are in space, and if you venture into a region where this maximum is lower than your own intelligence, you will literally get stupider. Most importantly, there are the Tines, whose individuals are not individual animals, but are instead packs of animals who share thoughts between themselves using high-frequency sound. That has all kinds of fascinating implications, which you will have to read the book to find out.
I scan the headlines and read the occasional article. I don't much go in for the 40% of the stuff that's about preventing the AI-pocalypse, but occasionally there's stuff in the other 60% that is worth my time.
a) I'm surprised he calls the process "replacism" (or the French equivalent). I'd call the process he describes "homogenization". Making consumers and workers more fungible, more similar to each other, makes them more homogeneous.
b) To the extent that this spreads formerly local options around the globe, it doesn't seem like a very worrisome trend. My local supermarket (in suburban South Carolina) carries sushi. 50 years ago, one probably couldn't have gotten it outside of Japan. It probably displaces some local item. This doesn't seem earth-shaking to me. Presumably Coca-Cola does similar things in the markets it has spread to.
c) The devil is in the details: How forcefully does the process occur? Are we talking about everyone learning algebra and everyone trying sushi, or violence against those speaking a local language? Camus mentions Breton not being taught in schools (if Le Pen had won). How forcefully was this to be done?
d) My personal preference is to have political and economic systems that allow the widest choice of live options to the median people in the population. Adding internationalized options tends to widen peoples' choices, but displacing local options tends to narrow them. I don't know what factor dominates.
e) Another way to look at this (admittedly Taylorist - which is usually _not_ my favored leaning) is to view it as a spread of "best practices". If nation A has a more efficient way of manufacturing some widget than nation B, why should nation B not adopt it?
Seems like an interesting Maslow based reaction. People who are well off and highly educated have way more time to care about complex and obscure abstract problems while voters on average are not in that position.
Also why not run on the EA platform in San Fancisco? Seems weird to run on it in Oregon. And then piss off the pro-owl movement.
I don't have a good understanding of why there hasn't been a credible "movement candidate" run against Nancy Pelosi. Presumably everyone is waiting for her to retire so they can run for an open seat?
Edit to add: also, the saying is that Portland is where young people go to retire. And "effective altruism" seems like a message targeted to retired young people.
Pelosi's district is in the old money patrician section of SF. I doubt there are more than 6 resident voters who have heard of EA or who wouldn't regard it as some harmless amusement the kids are up to these days, like that World of Battlecraft thingy or whatever it's called. Probably involves the Internet, right?
Shahid was relatively credible. I mean not Fetterman level but still. The issue is that the entire California Dem establishment will come down hard on an anti-Pelosi candidate. Also voters don't like to dump their rep when their rep has a high seniority leader position cause it provides extra pork for the district.
"It's great that your are super worried about the AI reverse rapture double apocalypse, but I'm working 60 hours a week have to pick three out of : Rent, Car, Food, Childcare"
I have found that « cause célèbre » candidates almost always underperform. Also, they tend not to give the most neutral political analysis in their fundraising emails ...
NieR: Automata makes me cry whenever I reach the final ending. I'd say it's especially effective if you played the original NieR. The downside is that you really do need to invest time in the 40+ hour experience.
For a shorter one, the Violet Evergarden anime. For an *even shorter* one, Violet Evergarden episode ten.
I concur. Funerals are *great*, even when I barely knew the deceased and didn't care a bit upon hearing of their demise, simply being in the cemetary and listening to the eulogies makes me cry. Ideally during late autumn and a very light rain.
The answer is in parts I and II of the series, current theories of obesity (e.g. food availability, portion size, palatability, energy density) do not adequately explain the obesity epidemic and in some cases are actually contradictory.
The question is "Why is #4 a thing that can happen?" What change started some time in the 1970s that made food much more crave-worthy across a wide range of diets in the modern world? Why do people feel cravings in excess of what their body needs? Why do people have wildly different experiences in terms of how easy it is to resist cravings?
For 30 years, I've had no trouble keeping my weight. Never had a "freshman fifteen," never had to count calories, I just eat until I stop feeling hungry and I stay at around 145 pounds. If I eat extra desserts one day, I'll feel less hungry later and it balances out. I don't think I'm particularly resistant to temptation - certainly I play video games too much for my own good - it's just this particular temptation seems to have no power over me.
But this isn't a universal experience - if my wife eats extra desserts one week, then she has to put more effort into counting calories or she'll gain weight. Why do two people in the same house eating the same food have wildly different experiences when it comes to cravings and the ability to maintain their weight? It's an interesting question, and I think a "homeostasis" model (my wife and I have different set points, and hers is too high for some reason) makes a lot more sense than a model based around willpower alone.
Doesn't seem particularly crazy; mostly it seems vague. "Could be this, could be something else." I mean, if the cause of obesity is some particular unnoticed bacteria in the gut, which has infiltrated water supplies - that explains exactly the patterns, and would satisfy the contaminant description. If something is in the air - same deal.
Mind, I think it's right, I just don't think it being right substantively narrows anything down for us.
The ICO craze of 2017 never recovered, and I expect NFTs to play out similarly. Bitcoin will probably still exist in 5 years and may even go through more bubbles in the future.
In short form: there actually is a need for Nakamoto consensus to do time-keeping for smart contracts. Everything else is going to move to either a centralized system or a proof-of-stake system.
I would give it a 30-40% chance to recover to previous highs, and think that it will more likely recover somewhat but not to previous highs for two main reasons:
1. Tighter monetary policy, increasing probability of a recession. There was a lot of cheap money going around the past few years, and with all "growth" tech companies already highly overvalued, not a lot of places for it to go. I believe a lot of it went into crypto hoping that the next moonshot company would come out of it.
2. Increasing likelihood of government intervention/regulation. I don't think the powers that be in the financial regulation arms of the US government are very happy with the current state of crypto. The same can be seen across the pond in the EU. The recent Terra collapse may expedite how quickly they start really intervening and limiting what is possible.
Fiat currencies are entering hyperinflation spirals in multiple first-world countries, if anything I'd expect an influx of capital into crypto as people escape fiat in panic.
If crypto was a currency in an actual country - ie it was a unit of exchange - then that country would be experiencing hyper inflation right now. And the price collapse indicates the opposite of what you think is going to happen. Interest rates going up guarantees some return on investments, while a collapsing crypto doesn’t.
I haven’t seen any signs of hyperinflation anywhere yet. What I do see is supply-driven price increases, which are the sort of thing that affect all currencies the same, whether fiat or other. (If a chip shortage causes a car shortage then you will need to pay more for a car whether you are paying in dollars or bitcoin or gold or sex.)
It's true that cash giveaways in the previous year or two are a big part of what happened during that time. But is there any reason to think they are contributing to *continued* price increases, more than a year after they stopped? They should already be baked into current prices, so if prices are continuing to increase, it must be due to new sources, right?
Doesn't have to be, no. That's what makes inflation so evil. You get some one-time external shock -- like a sudden jump in energy prices, in the 70s, or a big infusion of government money, in the 2020s -- and then it can become self-sustaining. In the 70s, when energy prices jumped up, all the things that depending most directly on energy (e.g. transportation, food) jumped up, and then stuff that was indirectly dependent jumped up -- and then of course wage pressure increased tremendously, and people starting insisting on higher wages and higher returns on capital, which led to higher prices, which leads to still higher wage pressure, and around and around you go.
The problem is that inflation is in no small part a result of *expectations* of what will happen in the future, and these expectations need not be solidly grounded in, nor tied down by, any objective fact.
Indeed, in fact there *was* no "energy crisis" in the 70s -- the world was *not* running out of oil, it would *not* be possible for OPEC to drive the price of oil to the Moon, et cetera, but that didn't stop the hysteria before gasoline was rationed (if you remember "even" and "odd" days) for absolutely no underlying economic reason.
It is unfortunately possible for the same thing to happen today. We are already seeing tremendous wage pressure at the bottom of the market (where it usually starts), and that will contribute to another round of price increases (as people who hire minimum wage workers already know is in the cards) -- and of course that will produce yet more pressure on wages. As long as people *expect* prices to rise 10% or more per year, and factor that into their spending, savings, and job-seeking habits, it goes on, regardless of any economic realities on the ground.
That's *why* those of us who lived through the 70s fear inflation like a fiend from the pit: it is perfectly capable of taking off from nothing other than mass delusion, and to survive the individual has to go along with the madding crowd, even if he himself isn't crazy.
Hungary already has fixed prices by government decree. Poland has food prices rising 4% MoM and energy prices rising even further, with cherry-picked gov stats inflation just going past 12% YoY. Turkey was struggling for a while but now they're at 70% YoY. Western EU is slightly more stable but there are already cracks showing.
I'm going by the hypothesis that 30-50% YoY is enough for an unrecoverable spiral, as the currency becomes a hot potato everyone tries to get rid of. And keep in mind the real food shortage (because of energy and especially fertilizer shortage) has not hit yet.
You want to know what humans are really bad at digesting? Plants.
Every other large herbivore can chow down on random leaves, or even grass, things that are easy to find in any environment. Not us, we can only extract nutrition from rare pieces of plant like fruits and tubers. Without agriculture it's incredibly difficult to find enough human-edible plants to survive in any habitat.
Scientific studies on vegetarians usually conclude that they have better general health than omnivores. However, there are major selection bias involved since there is good reason to believe vegetarians live healthier lives overall than meat-eaters. A better indicator of actual health benefits from specific diets might be to look at children, who are less prone to healthy living. There are some studies of health outcomes in vegetarian children and the general tendency seems to be that they have slightly worse outcomes than omnivore children (iron deficiency most of all).
Also, people who don't thrive on a vegetarian diet would be more likely to not try it or to drop it after a while. Being vegetarian could be good for a lot of people without being good for everyone.
>I constantly heard people telling me the human body wasn't good at digesting meat. How untrue is that?
Fat does take a while to digest (because it has to be emulsified in the stomach before the enzymes can get at it), so if you eat like 700g of Christmas ham in one sitting you'll be feeling bloated for hours (I did that once). But leaner meats don't have this problem, and in any case it still does get digested perfectly well, it just takes longer.
Eating all-animal is possible; this was, indeed, essentially the traditional Inuit/Eskimo diet. It does require you to eat a variety of tissue types rather than all "meat" (muscle); you need at least liver for the vitamins, plus fat (including, but not necessarily, milkfat) for the essential fatty acids (this latter is absolutely critical; if you eat all lean meat you will be incapacitated in a week and dead in three).
"(this latter is absolutely critical; if you eat all lean meat you will be incapacitated in a week and dead in three)"
I think that needs to be qualified. E.g. beef is leaner than pork, but beef-only works fine. We have reported cases for people dying from eating only rabbits (which are as lean as you can get). Rabbit starvation. But to my knowledge, nobody has shown that chicken-starvation is possible?
It will be scurvy that kills you. Luckily, seal livers are rich in vitamin C (as are polar bear livers, but they are trickier to get) so you can get by on an Arctic diet if you enjoy offal.
I'm referring here to lean cuts; certainly a whole cow or a whole chicken raised on a farm has plenty of fat, but if you don't eat any of the fat tissue (the skin, in the case of the chicken) then AIUI the problem can arise.
Wikipedia references a guy who ate an all-animal diet and as an experiment tried temporarily not eating any of the fat (in order to confirm that this was the cause of rabbit starvation); he indeed suffered the condition.
I know the glyoxylate cycle can do it, and everything's capable of a limited degree of conversion via acetone, but those are both somewhat aerobic. Propionate isn't a carbohydrate, but is glucogenic (and IIRC fermentation of cellulose into propionate is a big part of cow digestion); I'm not aware of a gamma-oxidation path to convert fatty acids into propionate, though.
In any case, a lack of carbohydrate isn't actually a big problem for humans as long as there's adequate protein intake, since the brain doesn't *need* glucose for energy (it can't use fat, but it can use ketone bodies which the liver makes from fat under low-carbohydrate conditions) and the general use as a metabolic building block can be supplied via gluconeogenesis from glucogenic amino acids.
Oh come on. Breaking everything down to 2-carbon molecules and building stuff up again is hardly equivalent to yeast fermenting a bunch of sugar into ethanol, or meat rotting, et cetera. Bacteria will certainly break down complex metabolites, but they do that for their own purposes, I very much doubt they would then execute some energetically costly anabolic pathway -- and just excrete new carbohydrates. If you know of such an example in the wild I'd be fascinated to read about it.
>Bacteria will certainly break down complex metabolites, but they do that for their own purposes, I very much doubt they would then execute some energetically costly anabolic pathway -- and just excrete new carbohydrates.
They don't excrete them, but if you eat them you'll get the stuff they made for their own purposes. This is how ruminants work; the bacteria digest the grass, then the cow digests the bacteria.
"I grew up in a mostly vegetarian environment and I constantly heard people telling me the human body wasn't good at digesting meat."
Very untrue. People who only eat meat exist and seem to do fine. Some have done that for decades and even put their children on it. All-meat diets seem to be viable.
Within the carnivore-community, we find the opposite belief. They claim that the human body isn't meant to digest plants. And point out that plants evolve to not be digested and be toxic (true-ish). They roughly draw upon a distinction between prey-animals eating plants (with large, complicated stomachs) and predator animals that eat them (with less need for that, since what you're eating is closer to the right form, hence smaller digestive systems like ours).
But in the end, we're omnivores like pigs, which are not specialized for one kind of food.
And what we eat is not "natural".
The domesticated chicken, cow and pigs are that we eat these days are bred to be much more palpatable and much easier to digest for humans. Something about the fatty tissue is different (I forgot the details, sorry). The same is true for modern plants, fruit, vegetables, cereals. Eat what you like, as long as you have some protein (otherwise kwashiorkor) and some fat (otherwise rabbit starvation).
Mostly everything works fine, apart from individual differences and transition periods. (going "meat-only" with no lead-in leads to diarhea for a week or two, because it's drastic switch to keto)
If we were teleported back into the Stone Age, the answer would be "You eat what you can get and it all kinda sucks".
"On February 26, 1928, he was admitted to the ward and on February 28, started on the meat diet. At our request he began eating lean meat only, although he had previously noted, in the North, that very lean meat sometimes produced digestive disturbances. On the 3rd day nausea and diarrhea developed. When fat meat was added to the diet, a full recovery was made in 2 days."
Obviously, pursuing human studies much further than this would get you arrested. Animal studies might work, but I'm not sure how well they'd generalise.
From friends' experiences: if you never eat meat, there's a significant chance (~30%?) that eating a one-off serving of meat will give you digestive issues. I'm sure your digestive system would get used to it if you made it a habit (but why would you?)
I've been vegetarian for ~6 years (vegan for some of those) and am generally convinced that the healthiest possible diet probably includes meat. I'm also convinced that the difference is miniscule and that the vast majority of people would be better off eating much less meat and dairy and many more plants.
I personally take the following supplements: multivitamin, creatine, omega-3, protein powder
Looking at the human gi system and at our closest (familar) animal analogs, we have pigs and dogs, both very well adapted to digesting meat as a major part of their diet. (Most important parts of the difference between species are certain liver functions, which usually show up in the processing of specific toxins/drugs.) In contrast, animals who consume almost no meat have gi systems that look very different from ours: horse, rabbit, cow and other ruminants. So just starting from that perspective, the hypothetical is very unlikely.
If you're comparing us to herbivores, I think a better comparison would be herbivorous apes. Only one stomach, but their intestines are different from ours.
Humans have been big meat eaters for hundreds of thousands of years. We not only survived but thrived over that period. If the human body wasn't good at digesting meat, that wouldn't be the case. For most of human existence, getting enough protein was key to survival. Meat was the best way to get enough protein.
When agriculture was invented, most humans in those societies did not get enough protein, because they weren't eating meat, and lost about two feet in height due to malnutrition.
>and lost about two feet in height due to malnutrition.
Are you suggesting that pre-modern farmers were under 4ft tall on average? I assume this is hyperbole but when you're making a point based in empirical fact like this it's good to be a little more precise with your statements.
Not the thousands of years. Only in the agricultural era. We know for certain that there were great height increases from 1850-1950, particularly in Europe. This was better nutrition in general but certainly involved more meat.
Probably works much better for earlier decades. Meat is getting more expensive with animal welfare regulation. Higher energy taxes also play a large role. Vegetarianism is becoming very common and many people already self-reduce their meat consumption for vaguely ethical reasons.
Vegetarianism has been cited as a possible explanation for the Dutch (tallest people in the world) shrinking.
Any such argument is too non-specific to be convincing. We eat all sorts of things in different amounts, ratios, compositions, sources than we did for the last 2 million years, and "meat" itself is not a natural category when considered nutritionally.
There is no convincing causal model I've read that indicates that meat in large amounts specifically causes harm, as opposed to more specific things like it being dangerous to eat bear liver or if you consume 5000 calories of sugar you are very likely to become obese and have your teeth fall out.
It's interesting to me that you are approaching the problem from that direction. What is the evidence for the claims that we are either A) bad at eating meat or B) eating too much meat?
Now, the second one has some very strong arguments around environmental sustainability and climate change, but I am unaware of any good evidence for health implications. There are tons of correlational studies about the "dangers of red meat", etc. but the quality of evidence is relatively poor. I'm unaware of any particular good evidence around health implications of eating too much meat (as long as you are getting enough of the other nutrients you need and aren't eating too many calories over all.
We might eat too much meat now. A balance makes sense. I was responding to "the human body isn't good at digesting meat", which is obviously untrue for most senses of what digestion is about.
"Of course, almost half of India is vegetarian and they seem to be doing fine"
Says who?
I mean, they're not all dropping like flies from being vegetarians, but that was never the question. They have a much lower life expectancy than the west. obviously, this is due to a range of factors, but you can't talk about confounders if that level of your argument is simply pointing to a population and claiming they're healthy.
"Hindu vegetarians historically did actually consume some animal products, since (as in most pre-modern societies) their grain supplies were highly contaminated with insect body parts"
Interesting point! I wonder if there is any evidence that improved grain processing technology or insecticide use ever caused malnutrition due to the reduction in insect contamination/fortification in flour. That would be a splendid example of Unintended Consequences.
I wonder about that, because when preparing rice it has to be picked through for this very reason - get out small stones, black grains, husks, insect parts, etc. Then you wash the rice until the water runs clear, to get rid of excess starch. Then you cook it.
That would probably get rid of the worst insect parts. Modern rice is pre-cleaned so you don't have to pick through it, but you should still rinse it.
I imagine Hindu vegetarians were lacto-vegetarians, as there is cooking with ghee (clarified butter), cheese, yoghurt and milk. Dairy products are a source of B12 so that would have helped.
Airline meals booking apparently differentiate between "Asian vegetarian meals" and "Hindu meals":
"An Asian vegetarian airline meal is a meal type that is suitable for vegetarian passengers. The meal excludes all meat, fish and eggs but may contain other dairy products such as milk. The Asian vegetarian airline meal is cooked in an Indian-style and is often spicy.
The Asian vegetarian meal usually falls under the category of 'vegetarian meals' but as it's a meal type that is designed for the Hindu community, it can also fall under the 'religious meal' category. The Asian vegetarian meal is also referred to as an 'Indian vegetarian meal' or a 'Hindu vegetarian meal' — not to be confused with the (non-vegetarian) Hindu meal."
"A Hindu airline meal is a non-vegetarian meal type that is suitable for Hindu people. This meal type is available for Hindus who are not entirely vegetarian. The meal will often include some type of meat or fish, as well as dairy, but will never contain beef or pork. The meal is usually cooked with spices and in an Indian-style. The non-vegetarian Hindu meal should not be confused with the vegetarian Hindu meal, which is also known as an 'Asian vegetarian meal'. The Hindu meal falls under the 'religious meal' category with most airlines, whereas the Asian vegetarian meal will usually full under the vegetarian/vegan category. So, bear that in mind when pre-ordering your meal."
Vegans handle B12 deficiency with nutritional yeast - it's available at the supermarket and it has a vaguely cheesy, savory flavor. A lot of vegan recipes call for it.
"Of course, almost half of India is vegetarian and they seem to be doing fine."
I'm not so sure about that. Of the couple dozen Indian students I've met in Germany, maybe two were slim. Maybe one was fit. Almost all seemed to carry some excess pounds. (though I never saw one that was morbidly obese)
Do we have good health data for Indians to say that they're doing fine?
Whether depressed people are more accurate depends on the environment - Depressed people bias towards fatalism and healthy people bias towards agentic optimism, and they're adaptive for different circumstances. There are things in life we have no control over, and things in life we have plenty of control over and which you ask about will determine who is more "accurate".
It's also worth noting that unrealistic optimism might be more useful than accurate estimation, given that humans have other biases, like excessive time discounting or excess fear of social shame. Perfect accuracy in all mental modules is ideal, but if they all have bugs sometimes the bugs are balancing each other out and 'fixing' one in isolation makes the system as a whole function worse.
Before reading the paper or its summary, my prediction is that someone did an experiment deliberately designed to trick people into thinking that they had more control over the experimental environment than they actually do. Maybe a bunch of buttons you press which are totally disconnected from a bunch of lights which come on at random, something along those lines. They did this knowing full well that depressed people would be less likely to fall for this particular trick than normal people, because they knew this would support the kind of "well ackshually depressed people are _more_ accurate in their world view then non-depressed people" that they knew would get clicks.
Having skimmed the paper I feel like my judgment was uncannily accurate, down to the buttons and the lights; enough so that I suspect I've heard of this paper before.
Anyway I stick to my initial judgment that the increased accuracy here is an artefact of the way the experimental design was chosen to make people systematically overestimate the amount of control they have.
The depressive realism meme is dangerous and needs to die before it causes people to.
Let's take a disease that warps your inner world and pretends it's a faithful representation of reality, then publish cherry-picked studies with bad experimental design to reinforce that the disease is right while you, the suffering person trying to get better, are wrong.
Agree. Even if depressed people are better at estimating certain things, they're bad at estimating the things that really, really matter to them.
For instance, depressed people tend to be really bad at estimating things like "How much better would I feel if I got out of bed, took a shower, cleaned up my house and then went for a walk outside?"
Or it could be non-optimal balancing of top-down predictions and bottom-up error signals as described throughout Andy Clark's "Surfing Uncertainty", reviewed by Scott about five years ago. Clark suggests (on page 238) we may "be entering … a golden age of 'computational psychiatry' … in which superficially different sets of symptoms may be explained by subtly different disturbances to core mechanisms implicated in perception, emotion, inference, and action."
Mis-estimation of control (in either direction) is also discussed (elsewhere).
https://www.nature.com/articles/d41586-022-01336-4?WT.ec_id=NATURE-202205&sap-outbound-id=F78931C80AA0330410205792EE3B23C13D9C77FC
CALLING LAWYERS: Want to work at an ACX-supported startup nonprofit? Legal Impact for Chickens is hiring! 🐥⚖️ https://www.legalimpactforchickens.org/were-hiring. Read about our ACX grant here: https://astralcodexten.substack.com/p/acx-grants-results?s=r
So I'm fully aware that I shouldn't ask for a list of favorite podcasts, so I won't, but if anyone recalls a link to a list of favorite podcasts that's been done in the past I'd very much appreciate it.
Why not? Have I been violating some social rule by asking?
Oh, yeah, incomplete thought. "because it's been asked a ton of times before" should have been in there somewhere.
As a non-particularly-outstanding physics phd student I am suddenly very very scared about my employment opportunities
https://www.quantamagazine.org/machine-scientists-distill-the-laws-of-physics-from-raw-data-20220510/
(Or, to paraphrase default_friend, this is hands down is one of the most evil and anti-human things i have ever seen)
> Symbolic regression similarly identifies relationships in complicated data sets, but it reports the findings in a format human researchers can understand: a short equation
This is what you call evil and anti-human? A new analysis method?
Then I've spent years trying to create new "evils" in my own field (software development).
That was meant to be a tongue-in-cheek reference to a tweet about dall e 2.
But well, i think I mentioned elsewhere that i'd rather live in a world without personal computers and that I am wary of raising kids in a world with internet, so if you are a software developer we probably have too different sensibilities.
Is this any more evil and anti-human than any other automation? Obviously people care more about stuff that directly impacts them but better to just admit that and not act like this is uniquely scary.
Is finding the equation that describes your data really the hard part of science? I would imagine the hard part is doing the experiment, or finding out what the equation means in a broader sense.
I know I know I'll stop, but it always funny to me that when we develop a magical technology that can save hundreds of thousands of man hours of labor a year and objectively improve the human experience:
The first though is always an anxiety ridden "But my job though :(".
Easy to be concerned with the public good when you're not the one paying for it.
Well, if that impact my job my human experience is not going to be objectively improved, isn't it?
Also, I am yet to be convinced that the technological advancements since the '90s have improved our life. Every time I think that i would want to have a kid, i conclude that a world where internet and smartphones exist is not one where I would want to raise a kid in.
And because of this i don't see science as a way to improve human life, but as a form of art. Yeah, I know it's a provincial concern, who cares, but also we keep improving the human condition in ways that are very easy to measure while at the same time we keep diluting those things that make life nice and meaningful, like art and science
Well in part that's because the parts of science and technology that are making tangible differences aren't visible. The journalism majors that report the news don't really understand them, and they aren't flashy and sexy and/or scary, so they just kind of rumble along in the background.
Teams of chemists and chemical engineers labor to tweak chemical syntheses, to make drugs, prosthetics, materials that work better, last longer, are more economical and less impactful on the environment to manufacture. But unless you're in the biz you won't read about a new and improved polymerization catalyst, or way to manufacture drug ingredient X that eliminates an expensive and toxic solvent in Step #66.
Mechanical engineers tweak car engines and robot manufacturing processes every year, to get more performance, better life, more efficiency in materials, both in manufacturing and use, and you'll never hear about it unless it's your profession and you're reading trade journals or going to conferences. Electrical engineers figure out how to build chips smaller and faster, or using better manufacturing processes. Civil engineers work on improving concrete -- still a work in progress after 2000 years -- or on better ways to build bridges and buildings. People improve the chemistry used to treat water and wastewater. Surgeons figure out how to reduce complications from surgery, radiation oncologists figure out how to target X-rays better so fewer cancer patients heave their guts out. And so on.
You pretty much only read in the broad news about the big, exciting, and/or scary stuff, but there is a giant engine of science, engineering, and technology that chugs along every day making things about infinite percent better than they were even within my lifetime, let alone a century or two.
What you say is clearly true, but at the same time it may also very well be the case that we have reached the point where the negative externalities of the big flashy "innovations" that are forced into our life outweight the positives of the slow technological advancement in other fields
I care little for better plastic when I cannot enjoy the products made with said plastic because internet gave me mental ilnesses. Oh well, at least nowadays i can get new fancy drugs to cope with modern society.
I don't believe that the industrial revolution has been a disaster for the human race. But I do believe that the digital revolution has. I made my limited share of simulations so I appreciate how important computers are in modern reaserch. But i'd rather live in a world where computers never got out of reaserch institutions
What do people think of the claims that omicron improves immunity for vaccinated and boosted people? I'm thinking about exposing myself by going to a probably crowded open air market without a mask.
I'm 69, vaccinated and triple boosted (Pfizer). I have type 2 diabetes, controlled. I'm low-end obese or possibly high-end overweight. (Correction: it was actually two boosters.)
I probably had COVID early on (messed with my sense of taste, but it recovered after about 6 months).
I don't know whether I've already had omicron. I live in southeast Philadelphia, not one of the highest risk zip codes.
Chances of getting covid on open air market are extremely low, even if it is crowded (not a medical advice)
I can see wanting to get it just because one gets so sick of feeling *stalked* by the freakin virus. I haven't caught it yet, and every time I have a few tiny cold symptoms I think I have it and feel sort of relieved, like OK, let's get it over with. Katelyn Jetelina's blog is good, and she discusses this sort of thing, i.e. immunity from infection vs. immunity from vaccines, but I can't remember if she addresses your specific question.
Personally, I am afraid of long covid. I think a lot of what's in the press is hysterical and overblown, and that a lot of the people who get diagnosed with long covid are just having a slow recovery from covid, or are suffering from the baseline malaise of everyday life familiar to covid graduates and covid virgins alike: fatigue, vague malaise, difficulty motivating oneself to concentrate . . . However, it does seem like a fraction of those people have something real and unusual going on. I myself suffered from what I am pretty sure was a post-flu syndrome for several years after a mildish case of the flu about 20 years ago. Worst symptom was that I was exhausted and craved a nap every single second every day, no matter how much I slept. In addition I had muscle aches all the time, and several joints so sore I gasped when I even just brushed them against something by accident. Also sore throats that came and went, and temp of 99.5 came that came and went. Internist tested for everything she or I could think of, and all tests gave no clues about what was up. Meanwhile I had a small child and a full-time job. I NEVER want to go through something like that again. It went away after about 3 years, but not before ruining those 3 years of life.
No big push either way from here. I'll go for the conservative choice and wear a mask.
I assume the idea is to get a less dangerous covid variant in order to protect from more dangerous variants? You might not get the variant you're aiming for, even it's the most common one.
I've had three boosters already.
So you've had five vaccinations in total? Reminds of the guy who had at least eighty-seven vaccinations: https://www.dw.com/en/german-man-got-covid-jab-87-times-report/a-61338245
I read that you should wait better six than four months between the first and second booster: https://www.spiegel.de/wissenschaft/medizin/corona-impfungen-zu-frueher-zweiter-booster-koennte-bei-juengeren-menschen-kontraproduktiv-sein-a-0fdfc2b2-ecf5-4cc0-aa71-572bb3f84aea
It's due to a thing called "affinity maturation" which takes six months and shouldn't be disturbed... Sorry, I can't find an English-language source... does anybody know more about this?
Sorry, my mistake. It was actually two boosters.
Who knows what might be coming? So long as a further boosted immune system doesn't lead to autommune diseases, why not?
I would have to imagine that there will be diminishing returns
There is no guarantee of safety. There are only tradeoffs, and considering the mutation rate, it's tradeoffs with estimates rather than information.
It is unfortunate that this thread is 4 days old. I'd love to hear thoughts on the Oregon 6th district race, especially from Scott himself. Carrick really generated a shitstorm in Oregon for a variety of reasons. "Not political", aka doesn't vote, cryptobro cash, not community oriented, etc. Shouldn't "effective" altruists be smarter than dumping a nobody into a race with crpyto cash?
Yeah this was a huge misstep. This primary flew under the radar so he could win purely on cash advantage. But in a general election funded well on both sides, he’s a really bad candidate. The ads just write themselves. “Doesn’t represent the interest of our district”, “bought by a billionaire from out of state”, “part of a fringe online movement”, “more interested in fanciful AI overloads than the real problems facing our nation right now”.
I feel like this race goes to lean Republican if he wins. And this shouldn’t require saying but I’m not arguing that the above statements are true, but they are highly effective.
Zvi has a post a few months ago where he was doing grant assessments for an EA organization, and a lot of the grants were of the form of
1. increase the status of our group
2. ???
3. good things happen!
and that has lots of problems.
Maybe Scott had really good reasons besides the paragraph that he gave here, but it pattern-matches to that. (And it probably does the opposite of step 1 and lowered the status of EA. It's probably easier to convince junior politicians to do EA than to take an EA person and make them politically savvy.)
> It's probably easier to convince junior politicians to do EA than to take an EA person and make them politically savvy.
Definitely want to signal boost this.
Our system of laws and regulations is unbelievably complex, which leaves many very smart people without a very robust model of how changes actually happen. No slight to them, it's just a crazy system. The scale of it doesn't fit in your head, but it's a system to solve for the preferences of a few hundred million people all at once. None of us can even fully comprehend the complexity of that problem, so naturally the equation we're fitting to that has a gaggle of dimensions.
Imagine all these interest groups scattered around the country that want more funding for law libraries, or harsher penalties for drunk driving, or whatever. They hire an expert in whichever legislature or administrative agency they're trying to persuade. Often, the best way to do this is to draft a few short canned sentences to tuck into a larger bill when it comes to the floor, since any given issue will only be addressed once every few years (maybe once per session on hot issues). You definitely don't have enough money or clout to control the docket. When this does happen, you'll need to scramble to persuade busy legislators to act really quickly, so you'll need to anticipate everything that needs to get done and help them do it.
Lobbying often means patiently sitting on a bench at a station--then suddenly trying to glue a few words onto a passing train.
So for each dollar, you can spend it on:
- getting a politician elected
- influencing existing reps
- convincing more citizens to care
- influencing regulatory agencies
- direct spending on outcomes, bypassing government entirely
I suspect electioneering is vastly oversubscribed. If the candidate is popular enough to win, you're at best getting a time share with a lot of other issues. If your issue was popular enough to be decisive, you'll get the benefit even without campaign contributions. The only thing that makes me question this is EMH, if so many people are buying it, it is probably worth something. But this thing is worth less the more people pay for it, so... I don't know if EMH even works for something with that property.
For niche issues your marginal dollar is probably best spent on awareness. A slogan for some issue that boils down your position to a nonexpert in maybe five words. "Atomkraft nein danke" as a notorious one. What's the "Atomkraft" of EA?
Working at lower levels of government can be a cheap test case. Can you get anything passed as a state law? Can you get somebody elected to a local school board? If not, this hill might be steeper than you expect.
Also, letter writing campaigns. If you want something passed, get a thousand people to write letters saying so. If you have well-founded compelling opinions, maybe just focus on communicating those to decisionmakers until they hear you.
It will take a few tries, other people are constantly trying to message them too.
If they never hear you though, reexamine your priors.
Turns out he got annihilated. I'm not too informed on Salinas vs the PA races but she seems pretty good.
Nate Silver announced last week he was eliminating out-of-state cash from his model because it isn't predictive. I'm guessing that in a general election cash raised is mostly a proxy for the support that candidate has and the cash itself matters little beyond that. So, if the cash is from out of state, it is worthless because those supporters can't vote in the election.
Self-funded candidates tend to do poorly for the same reason.
Will nuclear energy be deregulated at all in the next two years? https://manifold.markets/JohnBuridan/will-us-nuclear-energy-become-more
I think the answer is clearly yes. If not, then we should expect the energy crisis to continue indefinitely, because of market expectations for low energy output will keep down investment in all energy extraction methods.
Will German Nuclear power plants be put into service again (or at least not shut down ) until 2025? Is there a prediction market for this?
Not quite. https://www.metaculus.com/questions/10004/german-nuclear-power/
So almost :D
I will happily* wager a 4-figure amount of dollars against that outcome, for exactly the reasons Carl Pham lists in the next comment. Unless you're actually kidding/trolling, which seems possible, in which case never mind.
(* okay not actually _happily_, since I favor expanded investment in nuclear energy generation. But a winning bet as easy as that one would be impossible to resist.)
Of course it won't. Voters are exceedingly reluctant to deregulate even things that are pretty harmless -- like interstate telephony and airline travel -- if they think the field is, or could be, dominated by a few big operators, who may make decisions not in the public's interest. And who thinks public utilities put the public interest first? Essentially nobody. Add to that the potential for danger, real or imagined, and deregulation of nuclear power (which would have to come on top of deregulation of electric utilities) is a nonstarter in every polity of which I can think.
Why would the answer be clearly yes if it hasn't been in the last umpteen years?
Bet on it. I think political winds will change within a year. Red country is surprisingly more pro-nuclear.
One problem with applying EA to influencing politics is that a lot of political issues are ones where people can and do honestly disagree about what policies are good. In the case of Covid, it sounds as though the EA people see the right approach as greatly increasing federal funding on preparedness. I don't know if that works or not, and would be much more concerned with reductions in regulatory barriers — producing a workable vaccine took less than a week, getting it approved took most of a year.
For a more extreme disagreement, lots of people take it for granted that reducing climate change is an important objective. My view, as some here know, is that it is not clear whether the net effects of climate change are good or bad and there is little reason to think they are very bad. To take an older example of a similar controversy, the EA approach fifty years ago would probably have led to pushing ways of reducing population growth. It now looks as though, at least in the developed world, that would have been the opposite of the right direction.
You can use experiments and statistics to try to estimate the effect of bed nets on mortality, but it's a little harder to do the equivalent in cases like those.
What exactly does "political" mean here? As far as I can tell, it just amounts to "any issue that affects people's lives where there's any significant disagreement about what actions should be taken." The problem is that almost *anything* that has a significant impact on the world will be controversial for one group or another, and thus political. Improvements that are uncontroversial and unambiguously positive but still have a major impact (like mosquito nets) are few and far between, they're the exception rather than the rule. For EA to remain strictly apolitical would massively narrow the scope of what they could focus their efforts on.
As an example, many people think that giving out condoms to poor Africans is a good thing, both because it helps to mitigate overpopulation (which is definitely still a problem in Sub-Saharan Africa, even if it's no longer an issue in developed nations) and because it helps prevent the spread of HIV and other sexually transmitted diseases (which are a huge cause of suffering and death in that region of the world). Yet some people object to the use of condoms on strict deontological religious grounds, and would oppose such a measure on principle even if you could prove that it was a net good from a consequentialist perspective. Should EA reject efforts to distribute condoms in poor parts of Africa because the subject is too political?
I expect this has been covered elsewhere, but I don't see how we separate "altruism" from "I know what's best for others".
So is your critique of EA that being *effective* is not actually desirable, since too much effectiveness could lead us too fast in the wrong direction?
No. My critique of applying the EA approach to politics is that it is hard to know which direction to push in, so you might end up doing net harm instead of net good. The reason people in Africa don't have bed nets is that they can't afford them. The reason people in America don't have gun control/national health care/higher minimum wage is that a lot of people think it is a bad idea — and they might be right.
Sure, but isn't that a fully general argument against any sort of political implication? Presumably you still want people to enter politics for "altruistic" reasons, i.e. to improve the world.
EA basically says: let's think very carefully about what direction we push in, and once we've decided, let's push very hard. If the first part is flawed, then the "effectiveness" of the movement is a bad thing, since it'll push hard in potentially wrong directions. So that would be an argument specifically against pushing hard.
I think I mostly agree with this view, but it's also somewhat weird to argue for. One could *always* be wrong, especially in politics. EAs at least put a lot of energy in trying not to be.
Groups get captured by what's popular at the time. It's kind of a fluke that EA cares about AI and animal welfare.
Hardcore Bayesians don't assign a zero probability to anything, or at least they're very unlikely to.
So you're saying a good Bayesian should assign a nonzero but very small number to the probability that another Bayesian assigns a zero probability to something?
That sounds right.
Exploring evolutionary psychology using language models. This post looks at the difference between word vector representations of Kin Altruism and Reciprocal Altruism. As one would expect, reciprocal altruism is related to conscientiousness while kin altruism is slightly more related to emotional attachment.
And all this is done without surveying a single psychology undergraduate :p
https://vectors.substack.com/p/in-the-beginning-was-the-word?s=w
> Over time, I’ve realized that my default mode is to tune into what’s going on for other people—what they might be feeling, what’s going on with group dynamics, and what I would do in their shoes. Cycling through other people’s perspectives is automatic for me: it runs in a background thread of my mind and I can’t pause it without conscious effort.
I found this statement really interesting, because it is the exact opposite of my own experience. I guess it's another reminder of just how different everyone is.
https://dianaberlin.com/posts/journaling-in-practice
The FDA has rejected the EUA request for fluvoxamine. I've only read the article[0] and the summary in [1], not all the details, but they sound like legitimate, if perhaps thin, reasons for rejection.
[0] https://www.statnews.com/2022/05/16/fda-rejects-antidepressant-seen-as-possible-covid-19-treatment/
[1] https://www.accessdata.fda.gov/drugsatfda_docs/nda/2020/EUA%20110%20Fluvoxamine%20Decisional%20Memo_Redacted.pdf
> Carrick Flynn
Unfortunately, he's a Democrat. And given how much Democrats are working to make my life worse off, I can't support him. Even if he was "one of the good ones", getting elected would shift power towards the party.
Find me a candidate who won't caucus with the Democrats and I'll reconsider.
A classic "The Man, the Boy, and the Donkey" argument: just as many people would say the same about someone running as a Republican.
While that is a fine Irish name, I can't take him seriously because it sounds too much like Real Irish Placenames:
- Carrickfergus
- Carrick-on-Suir
- Carrick-on-Shannon
- Carrick
- Carrickmacross
So every time I read "Carrick Flynn", this automatically starts playing in my head:
https://www.youtube.com/watch?v=M0S9bIOK790
I read some of Carrick Flynn's advertising. During primaries the candidates are going for the votes of their party members, so both left and right are more extreme, and they move to the center for the general election. With this in mind, Flynn's advertising is noticeably less extreme when compared to his opponents.
For example, in the material I read, he states that two "500 year" floods ravaged his home town of Vernonia 8 years apart, thereby implying, rather than stating outright, that climate change is to blame. That may sound like a small thing, but in western Oregon that's a relatively conservative position for a Democrat.
One thing I noticed in his TV ads is he wishes his mother had access to abortion. I'm not sure of her circumstances, but does this mean he has siblings he wishes had been aborted? Or possibly himself?
"One thing I noticed in his TV ads is he wishes his mother had access to abortion. I'm not sure of her circumstances, but does this mean he has siblings he wishes had been aborted? Or possibly himself?"
Reading his potted bio on his campaign website, seemingly his family was poor and dysfunctional and he ended up parenting his siblings. So, like a good Democrat, his solution to poverty is kill poor babies before they can be born (okay, that's sarcastic and possibly even mean, but honestly - yes, maybe things would have been better if your parents never had kids, but 'gee I wish Mom had aborted little Pete and little Susie' is not the best way to refer to family, surely?)
"Carrick grew up in Vernonia, a small logging town in the Coastal Range. He had a tough start to life. His family was poor, broken, and after a flood destroyed their house, homeless. While this made him tough, it never made him hard. He took responsibility to shelter his little brother as best he could and up into his teenage years wanted to become a pastor to help others. This combination of strength and kindness remains at the core of his character today."
"yes, maybe things would have been better if your parents never had kids, but 'gee I wish Mom had aborted little Pete and little Susie' is not the best way to refer to family, surely?"
If he'd instead argued "I wish my mother had gotten my tubes tied, we need to make vasectomies and tubal legation surgeries more accessible," or even the socially conservative argument of "I wish my mother had stopped having sex, we need to do a better job teaching abstinence to poor young people," would that have really been better? Regardless of the specific method he'd be endorsing and the associated policy proposals, they would all boil down to him implicitly wishing that his younger siblings hadn't been born.
Literally everyone uses this excuse to be partisan. I am revolted by a great deal of the things that the Democratic party is doing, but if you vote for a Republican your voting for someone who will caucus for a party whose leaders all either wont condemn or actively support what happened on January 6. That's not a small thing.
The point is not that the GOP is worse or better, the point is that no-one is free from the responsibility of actually assessing which candidate they think is better for whatever position and then voting for them. IMO, in general, you should support a competent moderate over an idealogue.
"wont condemn or actively support what happened on January 6"
As opposed to won't condemn or actively support BLM and antifa? It's a non-sequitur.
Being revolted by BLM (which I am) does not mean I am not also revolted by January 6 and the attendant conspiracy mongering. If you are choosing one over the other, you are just admitting that you think *your media cult is better than the other one. You are just admitting that you don't really want to be a participant in representative democracy, you want a fun drama with a grand narrative.
And its not a non-sequitur, it clearly follows and fits with what I'm trying to say, your response even makes that clear.
I'm losing my stomach for voting for any Democrat or Republican. We need a new party which denounces the excesses of those old, evil Two parties which continue to hold us hostage.
How about a party which is against the Surveillance State, for instance???
I wouldn't start a new party in any place with First Past The Post voting: no chance of victory. But then again, the US does have especially bad parties, which I guess might help an "alt middle" party in certain specific "purple" locations where both the R and D aren't very well liked. [Edit: It would be extremely tough though, and remember that nowhere in America is there a majority of people who would appreciate ACX, so like, probably none of us should start a party.]
If you are concerned about surveillance, you will find candidates at many levels of government on both sides of the isle who are concerned as well.
Or maybe you're just too good for either party. Maybe you're just too principled, too pure, too *independent.* My heart goes out to you if that's true Jack Wilson who cares so much about privacy, it must be a very hard way to live.
I'm better than the wackos with power in either of those two parties, yes. Most normal people are.
Get someone to run as a single-issue candidate against the Surveillance State, no other platform. Try to get a few percentage points, enough to cover the margin of victory in the given race, to encourage the main parties to take your position.
Would there be a way to make a betting market for financial fraud? Resolutions would be something like "will SEC open an investigation?" (is that public?).
A lot of the data is semi-public if you have know-how and will to sleuth. Would be good to incentivize the sleuthing and public posting of documents + bets.
Obviously a half baked idea
Seems vulnerable to the fraudster insiders themselves.
Yeah, I wonder if limiting the amount any trader can have riding on a bet solves that. If only a fraction of the money can ever be insider money then less risk. As always, need a big market to be functional
You can always go long or short on whatever organisations or commodities are going to be affected by the investigation.
This is the traditional path. If you suspect a company is fraudulent, take out a short position, then publish your research.
What a shame Theranos was never publically traded, John Carreyrou could've made a fortune
https://www.youtube.com/watch?v=2ylOpbW1H-I
A look at rogue waves-- slightly rare giant waves.
Until 1978 when there was solid evidence, the many eyewitness accounts were ignored, and the occasional vanished ship was considered a mystery.
The fun question is, how seriously should eyewitness accounts be taken?
Meteorites were rejected by science until 1803, after the l'Aiglie meteorite.
https://en.m.wikipedia.org/wiki/L%27Aigle_(meteorite)
This meteorite fragmented. It was seen by dozens of people. Many rocks were recovered, which were geologically distinct from the rocks of the surrounding area, and similar to some other claimed meteorites. The evidence was collected and published by a respectable scientist, Jean-Baptiste Biot, of the French Academy of Sciences.
Meteorites are an a priori dubious hypothesis -- our familiarity with them has hidden that fact, which would've been obvious to earlier scholars. From where would rocks come that fall from the sky? From the Earth itself? That would require events of unimaginable energy, to fling stones so high that they came to the ground not within a thousand miles of any identifiable energetic event (e.g. volcano).
From the heavens, then? But already in 1800 people generally thought the Earth to be quite old, at least millions of years, and how could objects in the heavens be orbiting in stately regular orbits for millions (or longer) of years -- and then those orbits suddenly, after millions of years, intersect the Earth? That's asking for a very peculiar orbit, a great deal of luck, or random novel energetic events in the heavens (that would change well-established orbits) that had not yet been observed. So far as anyone could tell, nothing changed in the sky -- even comets after Halley were thought to be regular in their habits.
I suggest the existence of meteorites actually illustrates a subtle and profound principle of classical mechanics, id est the existence of chaotic dynamics in even relatively few-body systems. That celestial orbits can suddenly hit the Earth after 4.5 billion years of avoiding it is the signpost of dynamic chaos, the fact that even after such a very long time orbits can turn out to be not closed.
The German accented narrator reminded me of the painfully avant-garde 'Shprockets' SNL bit with Mike Myers as 'Dieter'
The Professor on Gilligan's Island seemed to know about rogue waves when the surfer showed up and then road "reverse tsunami" back home. Big man on little stick episode 1965
How did Gilligan screw that one up?
It bugged me that he never replied, so I looked it up. Gilligan didn't do anything, the Professor told the surfer who showed up that he'd be crazy to try to surf back to Hawaii, but by the end of the episode he tries it anyway. He makes it back to Hawaii but hits his head on a rock and gets amnesia, so he never tells anybody to go rescue everyone back on the island.
Is there any way to dabble in neurotech by buying a cheap brain computer interface and playing around with the data obtained from it ? That is to say is there a cheap reliable interface can be hooked up to a cheap eeg or something similar ?
Yeah, you can pick up brain interfaces that use skull electrodes mounted to a rubber hat for a few hundred dollars. They normally come with some conducting lube to make a decent electrical connection to your scalp. It helps if you aren't too hairy.
From a cold, amoral economic perspective, is slavery bad? Bad for the economy? I've read some pieces over the years arguing slavery was bad for the US South. Something about free labor being more efficient. But I'm never sure how divorced from politics such arguments are.
Perhaps the concept of AGI gives us a different perspective on the issue. Assume that somehow we know that an AGI has no qualia and can't suffer. Such an AGI would blur the line between human capital and other capital investments. Much like how under a slavery regime such investments would be blurred.
It seems to me that one can't argue that slavery is bad for the economy if total ownership of AGI's is not.
Or am I missing something?
Yes, slavery was unambiguously bad for the US economy. Even ignoring the gargantuan cost, human and economic, of the civil war directly and its long-term impacts, slavery delayed industrialization of the south. Without slaves, wealthy southerners would have been more likely to invest in machinery, railroads etc, things that actually lead to compounding growth. Slavery is an economic dead end.
You need a fairly complex story to explain why the north was so much wealthier than the south even before the civil war, and why they had a more valuable agricultural industry. And why nowhere else on earth that used slaves had anything remotely resembling the economic growth of the US, like Brazil.
Depends on what you compare it to.
Before Industrialism, slavery was very common, and rightly seen as a merciful alternative to killing your defeated enemies.
In a modern economy, it's clearly an awful waste of human resources.
What do you mean by "bad for the economy"? Do you include the welfare of the slaves?
Your AGI argument assumed that the AGI had no qualia and could not suffer. That was not true of human slaves.
Anyone have an opinion about Jane Jacob's claim that tools for slaves don't get improved? Even if it isn't entirely true, might it be mostly true?
Slavery prevents the development of the thing that actually makes a society wealthy, individual capacity. Massive amounts of money and time and talent have to be spent turning one portion of the population against the other, to prevent uprisings, and generally crush spirits on the scale that is necessary to keep a population of people enslaved.
As for comparing it to AGI's I have no idea what the connection you're trying to make is there.
There are some arguments for why slavery is bad without getting into morality. I think Scott's review of Red Plenty is relevant here: https://slatestarcodex.com/2014/09/24/book-review-red-plenty/
Slavery re-introduces many of the coordination problems that planned economies suffer from. This exists at the level of the manor instead of the nation, which is better. But other problems get worse: finding aptitudes amongst slaves or the best places to invest in non-slave capital become harder.
There is also providing for the needs of the slave. A slave receiving goods from a master is going to get a worse bundle of goods than if they could have spent the money themselves. Plus the monetary amount they'll receive a lot less than what they would have gotten in a market economy. A good definition of "good for the economy" would count this as being bad for the economy. Most arguments for slavery being a net benefit conveniently hide the material needs of the slave and hope you don't notice. Slaves drag the average income level down significantly, so slavery advocates want you to ignore them. They want slaves to count to the population counts for allocating electoral college seat calculations but not for GDP per capita calculations.
See also the Meditations on Moloch, III. 2. https://slatestarcodex.com/2014/07/30/meditations-on-moloch/ citing https://slatestarcodex.com/Stuff/manumission.pdf
>There is also providing for the needs of the slave. A slave receiving goods from a master is going to get a worse bundle of goods than if they could have spent the money themselves. Plus the monetary amount they'll receive a lot less than what they would have gotten in a market economy.
Where's your evidence?
Black slaves were taller than free white men during southern US slavery, but black american men are slightly shorter on average than white american men today (and non-american blacks are much shorter than white men). This indicates that slaves were actually better nourished than free white men. I don't know why people think free white workers in 18th and early 19th southern US experienced anything besides poverty.
>Most arguments for slavery being a net benefit conveniently hide the material needs of the slave and hope you don't notice.
Slaves were unambiguously better off materially than they would have otherwise been (i.e. if they had remained slaves in Africa or had been slaves in Brazil/Arabia instead). And again, I have no idea in the world why you think unskilled free workers in the old american south were well paid, but they weren't. All things considered, they were probably materially no better off on average than slaves were, and may have actually had less access to food.
> Slaves drag the average income level down significantly, so slavery advocates want you to ignore them.
Again, this is all just assumptions. And I mean, who the HELL is a "slavery advocate"? Literally nobody is arguing for slavery, and in fact the overwhelming majority of people I've ever seen arguing that slavery was economically beneficial were left-wingers who are furiously anti-slavery. They desperately want slavery to have been beneficial because it feeds into narratives around reparations and how white people owe black people for their prosperity (despite this being unambiguously false).
> Slaves were unambiguously better off materially
Just because they were taller? Geez. First of all, Google's telling me that slaves were *shorter* than whites[1]. Second, there's a lot more to being well-off than having enough nutrients. Ask any lab rat or, you know, slave (because slavery is still a thing).
[1] "adult slaves, Steckel found, were nearly as tall as free whites, and three to five inches taller than the average Africans of the time. [...] Although the adult slaves were clearly well fed, the children were extremely small and malnourished" https://www.newyorker.com/magazine/2004/04/05/the-height-gap
"Black slaves were taller than free white men during southern US slavery, but black american men are slightly shorter on average than white american men today (and non-american blacks are much shorter than white men)"
Do you have a single (reputable) source for any of these rather incredible claims? I would say that people of African descent are in general larger than the average white person, and non-American blacks are absolutely not 'shorter than white men'
In re slaves being taller: I wonder whether once it was illegal for people to be kidnapped from Africa, slaves were being bred for height.
Evidence from fiction: In _Jubiliee_ by Margaret Walker, a heavily researched novel, the white overseer is barely living better than the slaves.
Not just the needs of the slave, there's also costs of keeping slaves enslaves and protecting owner from getting attacked by slaves.
I've seen a claim that the profitability of slavery in the south was overestimated because the costs of slave-catching and protection against slave revolts were shuffled off to the governments rather than paid directly by slaveholders.
That doesn't sound like a good argument, because the tax burden would've fallen almost exclusively on slaveholders in the first place -- they were the only people who would be importing finished goods, the way the government got most of its income in those days. So they were still paying -- just indirectly, via taxes.
From a purely economic perspective, yes I think it's bad. To make it interesting let us assume arguendo that we have very enlightened masters, who take good care of their slaves, don't overwork or mistreat them, feed and house them as good or better than they would if they were free laborers and paid for their housing, food, medical care et cetera out of their own wages. We'll also assume the masters take care of healthcare, family support, and old-age support in at least as good a way as imposing pension taxes on the wages of free laborers and then giving them a public pension and/or Medicare would do. And finally we'll have to assume we have happy ideal-communist slaves, who work for the pure joy of doing the job well, and not for mere filthy lucre, and do as good a job as they would if they were being paid a wage and had the prospects of a raise or being fired to motive them.
We still have the problem that the distribution of labor -- who is assigned to what job and what career -- is decided by the masters, not the slaves, and the exact form of healthcare, family care, educational support, and old age support is also decided at that level, and the slaves have no input.
But that is certainly less economically efficient, for the same reason centrally-planned versions of these things always prove less efficient, because it means the people who make the allocation decisions are less than optimally informed about the individual circumstances. In a free labor market, each individual "allocates" himself to a job or career, and he has maximal information about his own skills, wishes, utility functions, et cetera, so given the right price signals as to the value of his labor to others, he can make the optimal choice about where he should work. No central committee or feudal lord could possibly make as good a choice -- because of the lack of detailed individual information. Which means, to achieve the same level of social satisfaction, more economic resources must be spent.
Same thing with appropriate healthcare coverage, appropriate family support, eductional support, appropriate old-age support. The individual is best suited to weigh the pros and cons of various options and choose the best option for himself, and no Federal bureaucracy or oligarchy of slaveowners could choose as well -- so that, again, for equivalent levels of social satisfaction, a larger amount of economic resources must be spent.
Alexis de Tocqueville gives an account of this in his classic book Democracy in America. He describes a journey down the Ohio River, which divided the free state Ohio from the slave state Kentucky: “Upon both banks of the Ohio, the character of the inhabitants is enterprising and energetic; but this vigor is very differently exercised in the two States. The white inhabitant of Ohio, who is obliged to subsist by his own exertions, regards temporal prosperity as the principal aim of his existence; and as the country which he occupies presents inexhaustible resources to his industry and ever-varying lures to his activity, his acquisitive ardor surpasses the ordinary limits of human cupidity: he is tormented by the desire of wealth, and he boldly enters upon every path which fortune opens to him; he becomes a sailor, a pioneer, an artisan, or a laborer with the same indifference, and he supports, with equal constancy, the fatigues and the dangers incidental to these various professions; the resources of his intelligence are astonishing, and his avidity in the pursuit of gain amounts to a species of heroism.
But the Kentuckian scorns not only labor, but all the undertakings which labor promotes; as he lives in an idle independence, his tastes are those of an idle man; money loses a portion of its value in his eyes; he covets wealth much less than pleasure and excitement; and the energy which his neighbor devotes to gain, turns with him to a passionate love of field sports and military exercises; he delights in violent bodily exertion, he is familiar with the use of arms, and is accustomed from a very early age to expose his life in single combat. Thus slavery not only prevents the whites from becoming opulent, but even from desiring to become so.
As the same causes have been continually producing opposite effects for the last two centuries in the British colonies of North America, they have established a very striking difference between the commercial capacity of the inhabitants of the South and those of the North. At the present day it is only the Northern States which are in possession of shipping, manufactures, railroads, and canals.”
I wonder if there's a way to get more data on this hypothesis. I know that white southerners are often described as lazy (Scott mentions this fact in his review of Albion's seed) but one major confounder is the existence of malaria. Malaria-carrying mosquitos thrive above a certain average temperature and die below it, creating 2 lines of latitude in the Western Hemisphere. Between them malaria is common, outside of them it is rare; in North America, this line is essentially the difference between slave and free states because Africans are much more resistant to malaria. And one of the major symptoms of malaria is... fatigue and a lack of desire to work. So I would want to know exactly how close the Ohioans and Kentuckians he is comparing lived to each other, and in particular if they might have had similar rates of malaria.
Well, a major point of his comparison was that that they lived literally on opposite banks of the same river. Which makes it difficult to attribute their differences to anything other than culture.
Right, but I'm not 100% sure how literally to take that. How far inland did he go/look? If he is being really literal then you're right. Although I also looked at a map and for some stretches the river runs North/South, so even if he did look inland a ways the difference couldn't be malaria.
The other thing is, do we have data to back up de Tocqueville's claims? Or is it just his summary/impression?
Well, there’s a paper here: https://digitalcommons.iwu.edu/cgi/viewcontent.cgi?article=1029&context=uauje which analyzes the available census data and concludes that Tocqueville’s impressions were pretty accurate
What is "the economy" and why should we care about it? In my opinion we care about it because represents the ability of people to do and consume things they want. In economist-speak, it allows them to satisfy their preferences. Obviously slaves can't do that, so slavery is by definition bad for the economy.
But also, it does seem like slavery, by providing undercosted unskilled labor, discouraged investment in industry or in improving worker skill. The North had much more industry than the South in the Civil War, and this was likely at least in part a result of slavery itself.
I don't follow your comparison to AI so I can't answer your last question.
"What is "the economy" and why should we care about it? In my opinion we care about it because represents the ability of people to do and consume things they want. In economist-speak, it allows them to satisfy their preferences. Obviously slaves can't do that, so slavery is by definition bad for the economy."
It really should not be that difficult to understand what is meant: Does slavery improve the per capita production of an economy or not (both at the time and over the long run)?
It's also really funny for you to act like you're being anti-slavery by using your definition, because most people who speak endlessly about slavery today are the ones desperately pushing the narrative that slavery is why America is wealthy (because that means that white people "owe" black people today).
Some individual got rich because of slavery; I think "America as a whole got rich because of slavery" is very wrong. There are multiple reasons to believe this--our economic growth post-Civil War has been quite high, much of the accumulated wealth of slavery was destroyed during the CW, the richest regions today are almost all in former free states while the poorest are in slave states (even looking just at whites), industrialization happened much more quickly in the North, etc.
But also, I think it's kind of an irrelevant question to ask. The productive capacity of the economy is not important if a large portion of people are legally barred from getting any benefit from it. It would be like asking if an invasive war is beneficial, while ignoring the dead soldiers, destroyed wealth, etc. and only looking at whether the government of the invading country collected more total taxes or something.
It's not hard to figure out why the south didn't industrialize and the north did. One of the biggest pieces of damage that slavery did to the south was cultural, they couldn't see any way to power other than through keeping people in bondage.
I imagine that part of the reason that slaves make for an unproductive workforce is the fact that they really don't want to be working for you and are going to do the bare minimum to avoid getting beaten. Not to mention the effort you have to invest in overseers, catching runaway slaves, putting down rebellions, etc.
If we make an AI that absolutely loves working in a paperclip factory and thinks it's living its best life, most of those problems wouldn't apply.
Under most circumstances, free labour is a lot more efficient than slave labour. The one possible exception is if you have a huge swath of land and there's not enough people living nearby to work it for you -- under those circumstances it might make sense to buy slaves instead of waiting for the local population to grow.
You can incentivise a slave more than a free person by offering the slave his freedom if he earns X dollars for you. But for cultural reasons this wasn't done very much in the US South.
How can the price of slaves go higher than the expected returns from (many years of) their labour?
You also get to sell or otherwise exploit their children for profit. And some people may find intangible benefits in owning slaves. Plus possibly there was some irrational exuberance in pricing at work, a la southseaturnip.com
AGI power vs. human power, types of labor AGI vs. slaves are used for, and upkeep costs for AGI vs. slaves. You don't have slaves designing the cures for cancers, you have them picking produce.
Also, of course, the AGI doesn't have qualia and can't suffer, while humans do have qualia and do suffer unless you're some kind of absolute nihilist or a very bizarre form of racist who thinks p-zombies exist, but only along ethnic lines.
"Also, of course, the AGI doesn't have qualia and can't suffer"
How the hell do you know that? Do you believe that immaterial souls are the source of consciousness?
Because if consciousness emerges from the performance of certain functions, then who is to say that machiens can't be conscious?
There can be the case where consciousness and qualia are required for AGI. Or at least possible, and we programmed AGI to have qualia for whatever reason.
In Rome, there were slaves working as physicians and as accountants, so why couldn't they be designing the cures for cancers?
He specified the US South, so I don't think bond-slavery is what's being discussed here, but chattel slavery. Those are meaningfully different things.
Roman slavery was more like a job that you didn't have a choice about doing (slaves were paid) rather than the whips and chains variety.
My impression is that the treatment of Roman slaves varied widely, from not too bad to sadistic sexual abuse.
Or just plain old being worked to death. The mines were particularly bad for this.
OK, let's assume the AGIs we have aren't brilliant. They can't design cancer cures, but can pick strawberries, cotton and work in sawmills.
But let's also stipulate that they are smart enough to find better paying jobs if they are available.
EDIT: Assume AGI power vs human power is equal, that upkeep costs for AGI vs slaves are equal.
EDIT" If we assume 1% of these AGIs are brilliant, brilliant on a human scale, how does that change things?
Once I've assumed the things you stipulate, you've rather defeated the point of invoking AGI, because these aren't things we're going to develop AGI for. You might as well say "clone homunculi" or "Zaplefrogians, who are identical to humans in every other aspect but metaphysically don't have any moral signifiers attached to them so you can treat them however you like."
My answer to that question, then, is that this is about as meaningful of a thought experiment as speculating about what time-traveling bullets would signify for the morality of abortion.
There's little evidence that slave labor had a higher ROI than other things in the old south, or that slaves were cheaper (to buy and maintain) than using free labor or were more productive.
No, it's not. As used here, it simply means "makes things worse by X metric".
In the sense of "ineffective at accomplishing X goal". https://en.wiktionary.org/wiki/bad#Adjective meanings 9 and 13 plus some shades of the others.
How does one journal
A blog post I recently read with some advice for a specific style of journaling: https://dianaberlin.com/posts/journaling-in-practice
I use a DokuWiki install on my server. It's kind of outdated as applications go, but works decently on my laptop and phone and has many versatile plugins.
How do you actually journal though? Like what do you write about and how much?
I checked now, and I write 1900 characters/410 words per day on average.
Mostly I write about what I did over the course of a day (my autobiographical memory is somewhat broken, so this is helpful) and whatever problems/things I happen to be considering around when I'm writing it. I'm looking at adding machine-readable data so I can find interesting trends.
I don't know if there's any "right" way to do things, but I just type stuff in a giant Google Doc.
How do you actually journal though? Like what do you write about and how much?
I write whatever I feel like, which is usually just whatever I did or any notable occurrences.
I tend to write a *lot*. I've apparently written 104 pages in the last three months! One night recently, I spent over an _hour_ writing, though that was unusually high.
Note that typing is way better than handwriting. For the first decade or so, I handwrote everything, which is really slow and my hand would cramp if I wrote even a fraction of the stuff that I type about nowadays. Also, having your journal be searchable comes in handy sometimes when you want to find out when something happened.
Hmm I kinda want to handwrite it though. Just feel like it's good to take a break from the screen. But thanks!
I wouldn't recommend it, but you do you.
Hahahaha
In case Russia starts another campaign in the EU due to NATO expansion. What your recommendation on stocks to buy or sell?
I'm following the Ukraine situation closely, both from a military and "Putin's personality" standpoint, and I find the attacking-NATO scenario very unlikely (Russia doesn't have the resources to fight on a second front, at least not without spending months or years mobilizing first).
But if this unlikely event actually occurs... it might involve tactical nukes. I expect a sudden stock market drop, which you could exploit by having lots of cash on hand to buy stocks, but in the nuclear case you might want emergency supplies instead. So, beat the rush, buy a bunch of emergency supplies early, put some cash in a safe, and when Russia attacks, buy an index fund at reduced price in the hope that things go back to normal? But maybe you don't want to hold a lot of cash. So to prepare for the conventional war case, maybe look at what stocks went up and down when Russia invaded Ukraine, and assume that a similar pattern will occur again if Russia attacks NATO.
They are having a sseriously hard time in Ukraine and have suffered severe losses in men and materiale. Why on earth would they take on the entirety of NATO?
If by "campaign" you mean "military invasion", then there's no place in the EU that Russia can plausibly invade this year without either being defeated in a conventional war so decisive that the markets won't need to adjust much beyond what they already have for Ukraine, or waging nuclear war. And nuclear war between Russia and NATO will probably cause sufficient disruption that stock markets won't operate and any stocks you do own will be worthless for the duration.
In which case, the usual - gold, canned beans, and 5.56x45mm cartridges in bulk.
If you're worried about Russia using the Ukraine fiasco as impetus to rebuild their army into something that could at least imagine victory over NATO, then that will take several years and it would probably be premature to buy before we see exactly how Ukraine turns out.
Seriously, three-quarters of the Russian army has been fed into a meatgrinder, and the Ukrainians are cranking the handle most vigorously. And to at least some extent vice versa, but whatever happens to Ukraine, Russia's not invading anyone else any time soon. Nuking, if they really want to, but not invading.
>If you're worried about Russia using the Ukraine fiasco as impetus to rebuild their army into something that could at least imagine victory over NATO, then that will take several years and it would probably be premature to buy before we see exactly how Ukraine turns out.
Note that NATO spends a bit over a trillion on the military per year, and even PPP Russia only has 4.3T GDP. The PRC could match NATO's spending, but Russia just doesn't have the economy to support it.
What's your info source on the "3/4 Russian army dead" estimate, btw? I'm aware you're probably simplifying the stats for the sake of argument and I'm fine with that, I'm more interested in finding reliable sources.
I think he's saying that 3/4 of the army was "fed into the meatgrinder", not that they're dead. Presumably only a small fraction of the units (and material) are dead, but that's all it takes to result in combat ineffectiveness.
Noted, thanks. Question still stands.
Numerous credible sources including the Institute for the Study of War and the UK Ministry of Defense report that Russian forces committed to the war in Ukraine represent ~75% of their total combat strength, and we've seen them e.g. bring in Naval Infantry from the Northern and possibly Pacific Fleets, so they do seem to be going all-in except for a bare minimum held back for internal security and guarding their other borders.
Equipment losses, most of them photographically confirmed because this is a smartphone war, suggest 30+% losses for the forces committed to the invasion, or ~25% for the Russian army as a whole. Manpower losses will be somewhat less than that, but still substantial - probably over twenty thousand killed, total including wounded, captured, deserted, and psychiatric casualties likely over 100K by now.
As beelester military units generally become combat-ineffective at ~20% casualties due to loss of cohesion. There are some exceptions to this, but none that really apply to Russia at this point. And it can't be quickly fixed by shoveling in replacements, because the veterans won't trust the newbies or vice versa.
Note that the Ukrainian army has also suffered heavily in this war. I *think* proportionately less than the Russian, but it's hard to be sure.
Recent estimates think about a third is lost: https://www.theguardian.com/world/2022/may/15/russia-likely-to-have-lost-third-of-its-ukraine-invasion-force-says-uk
The initial invasion force was estimated at about 75% of their standing army, so that would suggest about 25% of their standing army is casualties. Which is still a pretty huge loss.
(Slightly confounded by the fact that the separatist forces are also fighting and I'm not sure if they're included in this fraction.)
I believe the rule of thumb is that 10% losses will temporarily knock out a unit, and 30% is enough to make it combat ineffective.
That is a very good point, and one of the interesting things illuminated by the course of the Russo-Ukrainian war is how even Putin doesn't believe his own bullshit about NATO posing an offensive threat to Russia. If he really thought it did, it would be the height of folly to permit the level of consumption of Russia's military resources that is now going on. At this point, Poland would probably have a 50/50 chance of making it to Moscow, barring the use of nukes.
Consumer staples, energy, commodities, the usual defensive buys. But you're too late, everyone else already thought of that. The time to buy any of those was last year when NFLX and ZM were riding high. If you did that in some cases you tripled your money by this spring.
I’d imagine if you have to ask on substack, then you probably can’t make much money this way (see https://en.wikipedia.org/wiki/Efficient-market_hypothesis) but if, for whatever reason you think markets underestimate the chance of war between Russia and the EU, then Ig defense, tech, and energy stocks would be the way to go, and airlines, tourism, companies heavily invested abroad, particularly in Europe, would be the thing to short.
With that said, if you Think Russia and the US are really going to go to war, then your strategy is simple— move your assets into Yuan
Moving your assets into RMB looks nice on paper, but the PRC has notoriously-weak rule of law and likes Altering the Deal so you'd just get expropriated.
EDIT: Also, if the USA is getting nuked to shit anyway, they're probably going to nuke the PRC as well to prevent this exact scenario.
Yeah, you’re right, and I was mostly joking with that comment. If the US and Russia go to war then you’ll want to trade in your assets for food, fuel, and guns.
I believe that the January effect (see https://en.wikipedia.org/wiki/January_effect) show that EMH even in the weak version don't hold much water... but what am I know...
My point in this question is to see what substack user thinks, but maybe it more suitable to prediction markets https://astralcodexten.substack.com/p/mantic-monday-5922?s=r.
ps, investing in defense company (like Lockheed-Martin) will only beneficial for long war, not short war. Since defense companies make their main profits from R&D and not from manufacturing (buying 100 plans or battleship need too many congress approvals).
Probably, not proven
I mean, this is ACX, so we should be thinking in terms of updates. Upon learning about the January effect, I saw that it was a case where EMH could theoretically be wrong, but there was no particular reason to believe it was, so I see no reason to update off of it.
Incidentally, if you are unsure about EMH and wish to prove its validity to yourself, here is a handy exercise: set up a mock portfolio on investpedia, buy whichever stocks you want, trade however often you want, and then wait three months. See how you do compared to the market average, and then repeat a couple of times— it is very enlightening.
Funny you mention this. Once I know its a real money, I'm below the average market. If it is virtual money I'm extremely successful.
Internet troll army. Not sure if it means that you should buy or sell the stocks of social networks.
The Covid-19 pandemic has been a huge disruptor of business-as-usual, including normalizing remote work, and a host of other previously very unlikely global changes. I wonder what kind of an event may lead to something like an emergence of a viable third party in the US and the subsequent reduction in political polarization?
Why do you assume a viable third party would reduce political polarization? I'd expect the opposite, since every politician then has two convenient bogeymen to blame for their own failures instead of one.
Given (a) a presidential system, (b) first-past-the-post voting system, and (c) two parties that have been firmly entrenched for 160 years, the current two party system is a stable equilibrium to me.
The absolute best case for a third party in the US is the oscillatory leach-on-one-main-party strategy like the LibDems since WWII in the UK or the NDP in Canada. But I'm skeptical even that is plausible, given the lack of importance of coalitions in non-parliamentary systems.
I reiterate though that it's not like we're a 2-party system by chance or because Dems and the GOP are colluding to keep the proletariat down ("ratchet theory" and the like), a bewildering viewpoint I often see online. We're a two-party system because it's a natural consequence of how our government has been set up since 1787.
Not an American but I assume the 'required event' is something that significantly alters the structure of the US government and constitution.
Australia (Westminster parliament system) has 2 main parties plus a numerically significant 3rd party (The Greens); plus several minor parties who each exert influence on the houses, particularly the Senate house of review. However I would not consider any of the parties outside the big 2 "viable" in the sense of forming government - their purpose is basically to add moving parts to the system so that no individual wheel can dominate the operation of the machine.
Whether or not this ends up being good for the country depends on heaps of things. My impression is that it prevents the low-likelihood/catastrophic consequence events like runaway dictatorships; while maintaining a rather ineffectual, slow and messy democracy with very little innovation or creative vision.
If the election rules are changed, there would probably be more viable parties. (Changed to a parliamentary system, or a system with a second election round between the two biggest parties or whatever.) The election rules are unlikely to be changed anytime soon. But I think it is still more likely for the rules to be changed, that for there to be 3 viable parties under the current rules.
I was thinking maybe one of the political parties would change their rules for the primary election. Like Trump would probably not have won his primary election if they didn't have first-past-the-post rules. So it could be argued it would be smart for the Republicans to change their rules. And if they did, it might make more Americans realize first-past-the-post elections are bad.
What would be the point of a 3rd party? There is zero chance that any political party could ever attract the permanent allegiance of a fat majority of voters by being Everything to Everyone. There's far to much diversity in what people want, and their situation, and the illusion that there is One True Ruling Philosophy is the kind of naive folly that brings us the tragedies of communist dictatorships and religious theocracies.
I don't think polarization has boo to do with people not being presented with a wonderfully better choice that would draw everyone into one big happy philosophical family. I would seek its origins in (1) social atomization: humans long to belong to congenial tribes, and as our families shrink (or evaporate entirely), as we become more mobile, and as our transactions shift from face-to-face to online, we find fewer opportunities to Belong that are local in place and purpose, and we not surprisingly become more attracted to massive quasi-religious cults, e.g. political ideologies, and (2) the enormous increase in the power and scope of government, which makes fewer and fewer people willing to shrug and accept The Opposition pulling its levers. It *matters* who is President far more than at any time in American history, save perhaps 1861, because of the way he or she can rummage around in our lives, and so people get worked up about it much more.
I think there is room for two smaller parties; ones like the DSA (though they seem to be currently engaged in eating themselves due to purity spirals) who have to tag along with the Democrats if they want to get anywhere could amalgamate with all the leftist/progressive elements, and take the far-lefties out of the Democrats, and I think everyone would be happier - the progressives wouldn't be hobbled by having to depend on the centrist majority, the Democrats would be able to be liberal/centrists and not have to keep tossing bones to the student activist lot.
Same for the Republicans; room to divide up the social conservatives from the economic conservatives - the billionaire/Rockefeller Republican types from the populists, if you like.
As it is, I think the two major parties are coalitions of disparate elements who hang together mainly by using the threat of "you don't want The Other Lot to get into power, do you? They would be terrible!" to keep everyone in line. Then Tweedledum or Tweedledee gets into power, and immediately starts bitching about Joe Manchin and Krysten Sinema not being a real Democrat (insert comparable Republican names here yourself). Having a much clearer division between the various elements by having separate parties would make things easier for everyone; so suppose the Democrats can only form a government by going into coalition with the DSA or the Greens or the Free Hair Dye And Undercuts For Every Gender party, then the explicit terms of the bargain would be visible and everyone would understand: smaller party gets its way on this, in return for voting with the bigger party on that. Instead, you have 99 cats all wanting different things and clawing at each other over "you're not a *real* Whatever, why isn't the party dumping you?" (because they're the only one who can win a seat in that state for the party, silly).
Sure, but as I said below I think that's equivalent to the voter assigning the hard job of political compromise to the professional politicians who run the parties. It's certainly a successful model in much if not most of the world, and as long as the voters are OK with their wishes being represented more indirectly it tends to produce much more genteel political debate and far fewer political earthquakes. I might well prefer to live under such a system, from the point of view of stability and predictability (although not from the point of view of economics or entrepreneurship, which is tangential but related).
But Americans tend to like to work out stuff themselves and exert their influence more directly. Hence in our current model all the hard work of working out political compromise has to be done directly by and between voters -- they have to, as you put it, decide where they won't compromise and where they'll hold their nose and join those damn yellow-dog Democrats or wretched racist Repubs in order to get half a loaf (have x% of what they want be part of the governing party's agenda).
It's without question messier, and it produces much uglier debates. But I think it also means *in principle* Americans have to take more seriously the problem of getting along with diverse others and necessity for compromise -- because they have to do it themselves[1], rather than assigning that job to pros.
Perhaps more significantly, though, it means the American political system responds faster and more completely to significant changes in how The People feel about stuff, because the major parties have less of an ideological core -- they're both basically vote-scroungers hungry for *any* way to cobble together a winning coalition. Certain people are apt to look on Trump as a horrifying result, and at some level (although by no means all) I agree he was an unfortunate President -- but I think that's because they fail to consider a world in which Trump could *not* happen, and that big sea change in how voters feel about certain stuff could find no expression other than...something much, much worse.
The American system has, despite -- or, as I'm suggesting here -- to some degree *because of* its brutal squalid continuous acrimonious debate, been more resistant to stagnation or tyranny than almost any other system on the planet. It's been a functioning republic for almost 250 years, without falling into ineffectual malaise nor succumbing to dictatorship, which is a pretty remarkable record for such a large and diverse polity.
-------------
[1] If of course they don't shirk that obligation and instead spend all their energy whining about Those Irredeemable Bastards On The Other Side on Facebook and blogs. As Ben Franklin said "a republic -- if you can keep it." Id est, the American system only continues to work so long as Americans behave like...well, Americans.
I'd say 150 years. The Civil War wasn't very functional.
I completely disagree. The Civil War was exceedingly destructive, but it was also extremely functional and effective. Slavery had been an insoluble issue from the early 1600s right up to the 1850s, despite the best efforts of some very keen thinkers who had all the motivation in the world to solve the problem.
Between them, US Grant and Abe Lincoln (and of course the mass of people, many of them wearing the blue, who supported them) settled the question of slavery forevermore, beyond quibble, khattam-shud. Nobody has ever had to think about that issue again, it's over, and we all know exactly where the United States stands on one man owning another: the rows of graves of Gettysburg are eloquent on that point.
The naive mistake people make when pondering the horror of war is forgetting how it actually *does* settle issues, often finally, often very effectively. It's a very expensive solution, but it works.
If you think a brutal civil war is the government being "functional", I'd hate to imagine what you would consider _dysfunctional_!
Incidentally, most countries managed to abolish slavery without killing a million people in the process.
The way our elections work prevents this. If the Democrats split up into separate parties, the Republicans win, and vice versa.
Yes, the two parties are each a coalition hanging together on the threat that the other one will win. And it's a very real threat! If we don't hang together, the other party *will* win. I don't particularly like the guy I voted for, but he's nothing like the fucking disaster the other party nominated. "The parties are the same" was a cliché when I was growing up, but it's not true now, and I'm not sure it ever was.
> If the Democrats split up into separate parties, the Republicans win, and vice versa.
That's speculative because you're assuming those who are voting Republican won't switch to the new party. A lot of people who voted for Trump are populists, so if the new party is populist then it can drain voters from both the Democratic and Republican parties.
It's very clear at this point that the existing parties are anti-populist, so a populist challenger could certainly be a viable contender if the conditions are right.
The problem is that a party generally is small before it is large. And if a party's small, it won't win.
So someone with the preference New > Democrat > Republican, assuming that New won't win, will tactically vote Democrat (because voting New is effectively abstention). And someone with the preference New > Republican > Democrat, assuming that New won't win, will tactically vote Republican. This prophecy is self-fulfilling, and New remains small forever.
This tactical voting means that two-party systems under plurality are stable against all but the largest shocks; you need a new party to go from nearly-nothing to known-to-be-competitive all in one election cycle, and even then it usually just replaces one of the old ones (the main case in US history being the Whigs collapsing after Dred Scott and being almost seamlessly replaced by the Republicans).
Note that there are plenty of places where there are more than 3 functioning parties (Europe, Canada...). The centrist ones tend to act as a siphon for the moderates, but in general having more choice is... kind of an American thing to begin with? Strangely, it doesn't cross into politics. It's like having just two kinds of bread or cheese in the supermarket and being content with it. The barrier to entry is huge, hence why I asked what kind of a calamity might shake those barriers enough for something like that to be feasible.
Those places all have more aristocratic politics, so the real politicking happens between professional politicians of all the various parties, and the voters just pick which set of pros they want to negotiate on their behalf.
In the US we favor a more direct approach: we vote in The Ruling Party, give them more or less carte blanche - they can ignore the other party, often -- but then we (voters) threaten them continuously with being replaced by The Opposition Party, to try to keep them on their toes, and The Opposition more or less says whatever it thinks will open that possibility wider, without paying any particular attention to whether it's consistent with any particular ideology.
That's why it's hard to tell the difference between the two parties, sometimes, except for a few traditional pinstripes they both wear. The Ruling Party extols the wonderful things that can be done if The Others would stop being sticks in the mud, The Opposition Party preaches on the virtues of checks 'n' balances, tradition, bipartisanship, and so forth. Who's for and against the filibuster, the veto, gerrymandering, balancing the budget, confirming Supreme Court Justice with/without extensive hearings into their pecadilloes, and many other things besides switches reliably and smoothly with who won the last election.
So there isn't much of an interesting role for any 3rd party to play, other than spoiler. Ross Perot stymied George H. W. Bush's re-election, probably, and Jill Stein helped keep Hillary Clinton out of the White House, and that's about all they did, or could, accomplish. This is only pleasing to people who like poking politicians in the eye with a sharp stick, which I admit is almost all of us at one time or another, but if you're serious about trying to achieve a governing majority, it's merely tedious.
Historically, here's how a new major party appears: One of the old parties is so successful its rival collapses, and a new party replaced the one that collapsed. Polarization makes this harder; who's going to jump ship and join that horrible other party? (Well, some people have, but not enough to make either party collapse, and if the pandemic hasn't done it yet it never will.)
I'd call the Whig/Republican replacement more of "the Whigs failed" than "the Democrats succeeded".
It was also more of a split than a replacement. Antislavery Whigs in the North (politicians, party apparatchiks, and voters) deserted en masse to the Republican Party, and many pro-slavery Whigs in the South reverted to the older "Oppositionist" label.
A rump Whig national organization lingered in Presidential politics for a couple election cycles trying to keep itself relevant by running joint candidates with the Know-Nothings (Fillmore in 1856, nominated separately by both parties, and the Bell in 1860 under the new "Constitutional Union" merged brand).
Looking for recommendations on places to live in the US. I recently got a fully remote job (in the US) and, having worked to get the opportunity to live anywhere, need to actually figure out the best place to live. Current plan is to get a list of promising places and go AirBnB each of them for two weeks, see how I like it. Yes, I am the generic California expat.
My priorities:
--Decent dating scene for someone in their mid 30s
--Good cost of living, ie, can I get a decent place to live for $1500-$2000/month
--Outgoing vibe. I can be a real homebody, so a place that makes it trivially easy to go out and have fun is important.
Really depends on your hobbies or desired hobbies. I *loved* Arizona, but I'm a hot-weather guy that loves to get out in the pool or the lake, or join a cheap outdoor tennis class during the months that some states shut down for snow. My wife hated the desert, though, so a lot of that is personal taste.
I second the "college towns in low CoL states" suggestion in general, though.
I'll put in a good word for my city/region. I live in Saint Louis. CoL is cheap, and for your budget you could get very nice apartments in the upper income areas or a pretty good amount of house if you decide to settle. As far as going out, there are a lot of cultural events, such that there's usually something interesting to do on the weekends, and I'm given to understand a decent bar and music scene. But, I don't do bars and music so I can't vouch personally, and I haven't dated in a long while, so I can't speak to that either.
Still, this is a lovely place to live and very affordable.
Doesn't it have dreadful weather and high crime rates?
Dreadful is exaggerating a bit. We typically get two or so good snows, which will sometimes stay a week or so, and otherwise winter is 30s to 50s. Summer is hot and humid, spring and fall are nice. Depends on what you want.
You won't find California weather many places, and this place has a trade off of *some* winter and hot summers. Go to Minnesota and you get much longer winters where the snow stays on the ground and the lakes freeze, but cooler summers. Head down to Louisiana and you lose the winter but get even hotter and longer summers. Away from a handful of places on the coast you're just sort of moving your slider between long cold winters and long hot summers, IMO.
Balance is fine here, again IMO. I'd prefer a bit more winter and a bit less summer, but it's fine. Example. Forest Park is our big park. In June there's a Shakespeare festival. We go every year. We head over about noon, lay a blanket down in a good spot, and spend the day in the park. Most summers, we do outdoorsy stuff - the zoo, the boathouse, geocaching, frisbee, playgrounds, strolling. Some summers it's too hot, so we do indoor stuff - the art or history museum. By 6 - 6:30, when we find food and bring the rest of our stuff to the blanket to settle in, watch the pre-show and have dinner, it's begun to cool off and is pleasant. By 8, when the sun is low and the show is starting, you get the warm/cool summer evening and it's lovely.
Crime rates - yeah, they exist. My basic understanding is that it's mostly confined to certain neighborhoods or certain hobbies in which I do not partake. The neighborhood in which I live is actually considered one of those neighborhoods by a lot of folks who live "out in the county", but I've never felt worried over it and let my older kids (11 and 9) run off and play unsupervised as long as they're with someone and have a phone on them. I think it's fine. YMMV
Personally I would start with every state school college town (State College, PA, Ann Arbor, Boulder, Madison, etc.) Some may be too expensive but thats the vibe I prefer - smaller city with lots of energy.
Agreed. Weather is an important thing to consider, though. I love Madison, but the endless winter can get you down pretty quickly.
Seconding this- you can also get nice population variances anywhere in the 50k to 250k range if you're especially picky. Can be saturated with undergraduates but will usually have enough mid-young people to not be too isolating.
If it is fully remote, do you need to limit yourself to the US?
Tax compliance, mostly. Say I applied for the Taiwan Gold Card and went to Taipei for four months. I definitely owe, and should pay, income tax in Taiwan for roughly 1/3 of my income. But does my company pay corporate tax rates in Taiwan now? For what % of their income? Are they required to make matching contributions or retirement and health care accounts? To what extent, if any, do they now need to comply with Taiwanese employment laws?
I'm not saying there's no good answers for these, I'm saying any compliance officer or lawyer is going sees a massive headache and I've gotten some serious pushback. I think I need a pretty strong case with some good research before I push for it. Within the next year I'll probably hire a lawyer to explore some options and what the implications are for my employer, but for now the feedback I've gotten is that it's a massive compliance risk.
If you have any advice for a professional in this area, I'd appreciate it.
--Edited
Nashville seems to meet all three of those criteria, assuming you are a man wanting to date women. (The sex ratios work out in your favor. Not sure about the % still single by mid-30's, which I expect to be lower in the South than in major coastal cities.)
It's quite outgoing/fun and somewhat affordable.
The best place to get quality and in-depth science news and reviews is Quanta Magazine. The runner up for astrophysics is Ethan Siegel's column in Forbes.
Good suggestion! I really enjoyed this article of theirs https://www.quantamagazine.org/why-nasas-james-webb-space-telescope-matters-so-much-20211203/
Seems I'm not the only one! https://twitter.com/PulitzerPrizes/status/1523741167724404736
Quanta is great! Haven't read the other one
Long time lurker, first time poster to the open thread. I apologize in advance that it is shameless self-promotion post.
I've recently quit my job in academia to start a digital marketing agency. I work with physical or mental health practicioners that are looking to grow their practice (psychologists, physio, chiro, acupuncture, etc.). I've been struggling to grow my agency due to my disdain of engaging in typical "sales" tactics like cold-calling/cold-messaging. My eventual plan is to grow entirely by referals, but I'm not quite large enough yet that referals are growing my agency at the pace I need to pay my bills in the short term. So I post here with two hopes:
1.) If you're in my target market and are looking to learn more about digital marketing (or know someone who is) please reach out!
2.) Any advice on growing a business without using the skeezy sales tactics/cold outreach. (That also isn't "do good work and wait for referals" as I'm already doing that).
I work for a successful digital marketing agency. We get most of our clients these days through referrals, but in the past free audits of existing marketing strategy / PPC accounts was a big source of new clients. You put in 1-2 hours of free work, impress the pants off the prospect, and close normally.
I personally appreciate your anti-cold-messaging principles, as someone who gets cold-messaged (by recruiters) multiple times a day.
On #2: is this a business where SEO + content creation might be useful as a strategy? e.g. https://www.kalzumeus.com/greatest-hits/#marketing has some useful ideas-- a bit more geared toward software products but the author has worked as a consultant as well.
Yeah wtf. How and why can you be in sales when youre so squeamish about selling?
Is the government paying private companies 'prizes' for reaching set goals an effective means of industrial policy? An example would be the US government paying a company which reaches a desired milestone- manufacturing x amount of lithium ion batteries or semiconductors or solar panels in the US, for instance, or developing a malaria vaccine, or being the first to land on an asteroid. Or, developing a certain type of weapons system let's say. For real-world examples, NASA seems to have a series of smaller prizes for set goals (developing a certain type of power system), and Google famously had a Lunar X prize for a private company to land a rover on the Moon. Wiki apparently calls this an 'inducement prize contest'. https://en.wikipedia.org/wiki/Inducement_prize_contest
Is this effective industrial policy? Rather than more heavy-handed government intervention, such as picking winner & losers, directly funding specific companies (which of course are always very politically connected), or other boondoggles- the government can simply pick a goal and individual entrepreneurs can scramble to meet it. Does this work in real life? Should the US or other developed countries be doing more of this?
"An example would be the US government paying a company which reaches a desired milestone- manufacturing x amount of lithium ion batteries or semiconductors or solar panels in the US, for instance, or developing a malaria vaccine, or being the first to land on an asteroid."
Didn't they do that with grants for renewable energy? The problem then is that the companies turn their energies to hoovering up the government grants rather than hitting targets, because they get the money anyway. And business is in business to make money, not to be a social service. You make money by providing goods and services, but if you can make money by applying for a grant of government pork, that's just as good for the shareholders.
See the cash for ash scandal https://en.wikipedia.org/wiki/Renewable_Heat_Incentive_scandal or indeed Solyndra https://en.wikipedia.org/wiki/Solyndra and many others in other countries. Private companies don't sign up to government schemes unless they can be sure they will have as few penalties, and as much guarantees of getting paid, as possible. This does not incentivise them to hit the targets, but to extract the maximum of public money with the minimum of expenditure on providing the goods or services.
Solyndra was part of the same funding program as Tesla. So I think most people would argue that, ultimately, that particular round of government investing was worth it....
I think it makes sense for some specific problems. Especially for public goods, e.g. your malaria vaccine idea. But for most problems, other "soft-power" government interventions are better.
If you want more lithium ion batteries, semiconductors or solar panels, you can just adjust tax incentives.
Weapons systems typically have specs that are long, changing and classified, which doesn't suit well for a prize competition (but it could make sense to have a prize for an enabling technology).
Puts me in mind of England offering a prize for a chronometer that could keep good time on long marine voyages. Its a good story.
Though see Anton Howes' fascinating article in Works in Progress this month - though it is now described as a prize, most of the money was given to people working on promising ideas, to pay for and incentivise further work, rather than as a reward for having solved the problem. Which makes much more sense, given that it was set up in 1714 and dissolved in 1828. It was really an early grant-making body.
https://mymodernmet.com/marine-chronometer-clocks/
Afaik it was set up as a prize and the only serious contender was John Harrison who, over several years turned in five iterations. The final two made it on a voyage from Southampton to Jamaica with almost no loss or gain at all. He did receive some grants on the way because the government was very intrigued with his results and wanted him to continue. At the very end they tried to weasel on the prize and it took the intervention of King George to set it straight.
You're right, it was set up as a prize; though several prizes had been offered to solve the problem previously, to no effect. The thing that made the Board of Longitude different, and effective, was that it became a grant-making body. Harrison only got about 40% of the "reward" money, and of that only £8,750 was given to him as a final reward - he'd already received £13,250 to enable him to keep on working. And the reward money was only 33% of expenditure over the century the Board was in existence, with 15% spent on expeditions and 29% on publications (the rest on running costs).
It may be significant that the most famous example of a successful prize actually became successful because it stopped being primarily a prize and changed into a grant-making body and scientific publisher. Which doesn't mean prizes are a bad idea, but it does suggest that they're at best a useful adjunct to grant-making, and not an alternative to the thankless task of cleaning the Augean stables of the current grant-making and scientific publishing system.
Last point was weaker than it should be - I don't think anyone would claim that prizes are more than a useful adjunct to grant-making, so it's not useful to claim this as an insight from the real Longitude story.
I think given the real Longitude story, and the fact that I don't think there's a well-known example of a successful prize which doesn't in fact mostly involve grant-making (I believe the X-Prize disbursed about ten times more in grants than in the X-Prize, and the Orteig Prize didn't lead to any great scientific or technological breakthrough, or even, given that the Atlantic had already been crossed non-stop, a significant first in aviation) you should believe that prizes should not be used unless disbursed by a grant-making body, with most of the money disbursed in grants rather than prizes.
There were four theoretically viable approaches to the longitude problem at the time:
1. A mechanical clock that will keep accurate time on shipboard.
2. Lunar distances: an extensively calculated lookup table for determining absolute time based on the position of the moon relative to the fixed stars.
3. A device for taking precise sightings on shipboard of the Galilean moons of Jupiter, from which relatively simple calculations could determine the time.
4. Sending up a network of signal rockets visible from major shipping lines to broadcast the time.
The marine chronometer got most of the prize money, the overwhelming majority of which went to Harrison, but several other people got smaller incremental milestone prizes for either the chronometer (competing designs to Harrison's early models or incremental improvements to his later ones) or lunar distances.
In terms of practical adoption, lunar distance tables weren't far behind chronometers, and both methods were in widespread usage in the late 18th and early 19th centuries. The main tradeoff was that the sightings and calculations were simpler with a chronometer, but early chronometers were hideously expensive and needed backups or external cross-checks to ensure long-term accuracy. Lunar distances gradually fell out of favor around the mid 19th century as chronometer manufacturing techniques improved and they became much more affordable.
And as it happens, a variation of the signal rockets method supplanted both in the late 20th century. After all, what is a GPS satellite but a signal rocket payload broadcasting the precise time?
Thanks for this. I remember reading a long article about the whole thing a while ago but a lot of the details I had forgotten. Being a big fan of Hornblower and Aubrey the whole thing is fascinating to me.
It's occasionally useful at the margins, but the level of risk involved makes it impractical at the higher levels -there's probably a limit in the $1E7-$1E8 range where it stops working. Basically, you're going after billionaires' personal "mad money"; a corporate board is never going to chase a prize. At that level you might get some interesting technology development in areas where there may be low-hanging fruit, but you probably won't be bringing any high-value products to market and I doubt you'll be funding any significant space missions.
Also, if you try to have the United States Government do this, every prize will be tied up for years by the lawsuits filed by losing competitors, and the eventual winners will regret having ever signed on after that. I'm not sure whether other Western industrial democracies would do any better.
The thing about a prize contest is that all of the losers don't get paid for their work. This is sort of an upside (it effectively multiplies the amount of work you get for your dollar) and also a downside (since if you don't have a good chance of winning you shouldn't even try).
I would guess that this means it's best for blue-sky research stuff where you aren't really concerned with a return on investment so much as being able to push the boundaries of your knowledge, and where you genuinely have no idea who's likely to succeed so lots of people will throw their hat into the ring. So, yes to malaria vaccine, no to "build X,000 lithium batteries."
Quick update on my predictions on the outcome of Ruso-Ukrainian war. Previous prediction here: https://astralcodexten.substack.com/p/open-thread-218/comment/5875195?s=r.
10 % on unambiguous Ukrainian victory (up from 9 % on April 4).
Ukrainian victory is defined as either a) Ukrainian government gaining control of the territory it had not controlled before February 24, regardless of whether it is now directly controlled by Russia (Crimea), or by its proxies (Donetsk and Luhansk "republics”), without losing any similarly important territory and without conceding that it will stop its attempts to join EU or NATO, b) Ukrainian government getting official ok from Russia to join EU or NATO without conceding any territory and without losing de facto control of any territory it had controlled before February 24, or c) return to exact prewar status quo ante.
30 % on compromise solution which both sides might plausibly claim as victory (up from 26 % on April 4)
60 % on unambigous Russian victory (down from 65 % on April 4).
Russian victory is defined as Russia getting something it wants from Ukraine without giving any substantial concessions. Russia wants either a) Ukraine to stop claiming at least some of the territories that were before war claimed by Ukraine but de facto controlled by Russia or its proxies, or b) Russia or its proxies (old or new) to get more Ukrainian territory, de facto recognized by Ukraine in something resembling Minsk ceasefire(s)* or c) some form of guarantee that Ukraine will became neutral, which includes but is not limited to Ukraine not joining NATO. E.g. if Ukraine agrees to stay out of NATO without any other concessions to Russia, but gets mutual defense treaty with Poland and Turkey, that does NOT count as Russian victory.
Discussion:
I didn’t do updates in for a few weeks, not because nothing happened, but because things that happened kind of balanced each other out. Ukrainians are doing well on the battlefield, considering the circumstances, on the other hand Russian economy does not seem to be collapsing and Russian public does not seem to be rebellious, while EU diplomatic fiasco around oil sanctions, together with increasing acceptance of Gazprom gas payment terms, seemed to suggest that Western willingness to support Ukraine is wavering.
New information that pushed me to slightly reassess the situation are current elections in North Rhine-Westphalia, where pro-escalation parties (CDU and Greens) won a solid victory. North Rhine-Westphalia is most populous and bellwether German state (also with largest GDP), and German policy is very important for the outcome of the conflict.
*Minsk ceasefire or ceasefires (first agreement did not work, it was amended by second and since then it worked somewhat better) constituted, among other things, de facto recognition by Ukraine that Russia and its proxies will control some territory claimed by Ukraine for some time. In exchange Russia stopped trying to conquer more Ukrainian territory. Until February 24 of this year, that is.
Is there any way to go from viewing a comment to viewing the parent of that comment? This feels crazy to have to ask in a tree-structured comment system with permalinks to specific comments, but I can't see any way to do it.
There isn't a direct link to jump the parent like slatestarcodex had. It's one of the bigger things that I miss, as hiding a parent comment when you've read a bit into the children leaves the page scrolled to where the child comment would be, if the parent wasn't hidden, i.e., a random distance down the page.
You can click on the thin vertical lines to collapse some of the comments.
Given the discussions last week about software engineering, I suspect a lot of you would probably enjoy thecodelesscode.com if you're not already familiar with it.
How susceptible is the gut biome to persistent change through diet? Do we have high quality studies here? It seems naively like if you, say, totally cut carbohydrates from your diet for a week, you'd have to have significantly changed the ecosystem down there. But then, that doesn't seem to result in the same benefits that are claimed by, say, fecal transplants.
Isn’t this why vegetarians who eat meat get sick, and people who don’t eat a lot of beans end up farting a lot when they do? I think it might also be related to why people find it difficult to switch from a diet of highly processed foods to one of lots of whole grains and leafy greens.
I think you're right about the beans, but my understanding was that vegetarians getting sick after eating meat was more a matter of restarting production of digestive enzymes to handle high-protein foods and (in the case of the subset of vegetarians who eat very low-fat diets) ramping up bile production to handle a fat-dense meal.
You are talking about diet, but local germs also effect. I live in a warm climate but when I visit a cold place will get indigestion for the first week. When I lived in a cold place that was the opposite.
I wonder what the effect of you thinking "when I visit a cold place [I] will get indigestion for the first week" has on you getting indigestion when you go to a cold place.
I also wonder whether it is just the travel that's giving you indigestion.
Yeah, I mention cold but particularly I meant Mexico and USA. When I first moved to Mexico I got sick as a dog. Now I'm acclimated and when I go back to the USA I get indigestion. To explain the former I would have just said "Mexico has less sanitation". It's surprising to me that just having a _different_ set of germs is sufficient. As you say, it could just be travel. I assume changing attitudes have a sharper germ gradient as well.
A priori it seems like they'd do different things. Cutting or adding something could increase or reduce the level of something already present, but a transplant is presumably supposed to introduce something that's not there.
https://www.vox.com/future-perfect/2019/11/14/20963824/drug-resistance-antibiotics-cdc-report
>Routine hospital procedures like C-sections and joint replacements could become more dangerous, too, as the risk associated with [antibiotic-resistant] infection increases.
Nearly everyone who hasn't had an appendectomy has an appendix. Each of these people could get acute appendicitis. They could get it today. Or they could get it in a few decades, when routine surgeries might be (much?) more dangerous
Could preemptive appendectomy, done in the near future, be advisable for people with decades of life expectancy left? In other words, might it be rational to choose to face a risk of infection soon, so as to avoid the possibility of being forced to face a (much?) higher risk of infection later on?
For this to make sense, it seems that there would need to be a big increase in antibiotic-resistant infection risk coming. That's because lifetime acute appendicitis is only 8.6% in males, 6.7% in females (https://www.aafp.org/afp/2018/0701/p25.html - not sure if those figures are US or global).
IIRC, preemptive appendectomies are already done to astronauts, antarctic researchers and other people who will be in remote areas for a long time.
I wouldn’t do this if only because from what I’ve read there’s a fair amount about the appendix we don’t fully understand, and our previous assessment of it as “vestigial” and pointless in probably wrong. Among other things, the appendix probably helps keep portions of the gut microbiome going, so removing it in *everyone* would probably be a bad idea. Besides, that surgery currently isn’t risk-free. You’re making holes in people and removing a sac filled with bacteria. Plus I assume you need to do an appendectomy under anesthesia, which has plenty of risks on its own.
Also, most of the serious resistant bugs are clearly resistant from interventions in humans. The mechanism for resistant bacteria infecting humans from livestock is not very well established, esp compared to the bugs humans trade back and forth.
There are rules that certain antibiotics can't be given to livestock, which most of the West follows.
But the PRC defects, and the ROW can't enforce it easily.
How concerning is actually the appearance of Frank's sign (https://en.wikipedia.org/wiki/Frank's_sign)? Assume it's Grade 2b/3 - deep creases across both earlobes, in a 65-year old female patient.
I've seen medical opinions ranging from "it probably indicates physiological aging, but is a very poor predictor of cardiovascular problems, mostly to be ignored" to "an 700% increase in cardiovascular risks, run a detailed battery of tests". Could anybody familiar with Doing Good Medical Statistics and Bayesianism be so kind and provide me with the most accurate interpretation and recommendations?
I'm also interested in this; the difference in interpretation of this sign is significant, and (very) inadequately explained.
Does anybody know a good ethnography about traditional child-rearing in Africa, preferably describing the practices of at least a few decades ago? (doesn't have to be super-esoteric hunter-gatherer tribes, Bantu is also great).
I can recommend Nisa: The Life and Words of a !Kung Woman. Would love to find more in this vein.
I strongly second this recommendation! I just finished it and it's an amazing book. I remember somewhere in the rationalist blogosphere reading a review of a book about traditional child-rearing practices in hunter-gatherer groups around the world (there were a number of African groups, along with Indonesian, Australian aboriginal, etc.) , although frustratingly I can't find it anymore. One of the largest through-lines was that people were both quite tolerant of children and yet often not very interested in caring for them – they had other work to do. I think the book in question was Hunter-Gatherer Childhoods: Evolutionary, Developmental, and Cultural Perspectives, but I'm not quite sure.
How do I find and read the other book review contest submissions? No post has been made for this? Is it in one of the open threads? I also remember coming across a post that someone had madde a random reading tool or something of the sort
I made a tool to view them randomly, at readsomethinginteresting.com/acx.
This is what I was looking for, thank you
Excellent, thank you.
These are the finalists, one a week.
https://astralcodexten.substack.com/p/open-thread-220 links to the 4 google docs that contain all of the submissions.
Amusingly, google has entirely failed to index the relevant pages for the words "book review contest". I guess it might really be time to switch my default search engine to duckduckgo…
Since Google and other major search engines are broken in the sense that all they give you are commercial results, is there now a market for search engines that give you better results?
This is a good overview of (a subset of?) the "newer generation" search engine landscape: https://seirdy.one/2021/03/10/search-engines-with-own-indexes.html
Most of these are still new, and I think there's a fair deal of work to be done on most of them. However, they show a lot of promise, and sometimes they'll give far better results than the mainstream. I particularly like Marginalia.
Yes, for sure. Quotation marks don't seem reliable anymore; results from less-well-known web sites are outright truncated; results seem biased toward mainstream/superficial/SEO'd; results may appear with required words missing; even page excerpts don't seem as good as a few years ago.
Extremely significantly. When I attempt to search for certain terms useful to my field, I obtain SEO-optimized websites with very little information density. This is also the case when looking for things I'm less familiar with, such as home repair and tech fixes. Using Bing/ddg gives me less precise results but increases the probability I will find what I'm looking for within 5 pages by about 2.5x.
Absolutely. If I search for a very specific medical term about a very specific complaint, I get generalized results. In the same way, any medical or parenting search in Google, without qualifiers in quotes (e.g. the specific name of a forum) gives me 20 generic pages in the style of "parenting magazine" or "Mayo Clinic".
When I search transliterated Russian because I don't have a Russian keyboard access, it won't find the thing I need even though when I type it in Russian (or copy-paste from "Translate to Russian"), I get the results I need.
When I search for specific combinations of words or long-ish quotes I remember from a book or a site - but don't remember enough of anything else - it will yield no results, or very few. Or it will attempt to take away the one term that makes my search significant.
Finding an answer to a question that's nuanced or isn't easily captured in some Q&A somewhere, or the definition of some obscure terms with an old-fashioned spelling, are getting weirdly annoying and difficult.
This in addition to the brazen political ranking, where if you look for something controversial the first results are the "debunking" pages.
So yes, I use it but I'm happy I didn't throw away all my books, you know?
For a lot of terms, I have to add "wiki" to the end if I just want to read about a topic rather than buy something.
I don’t know what is the source of google going downhill, monetary or ideological corruption or just general big company malaise, but they definitely are no where near as useful as before. I switched to Duck Duck Go during COViD when I noticed Google was simply not returning results I knew existed, or was dropping them a few pages down below entirely irrelevant results. Comparing the two side by side, Google came out much worse. Which is a shame, I really liked Google up to a few years back.
It appears to cause degradation in the core search due to shifting priorities.
Googling "site:astralcodexten.substack.com book review contest", when filtered to 2022, gives back only 4 results, none of which are relevant. The same search on duckduckgo picks out the open threads where the contest is discussed, and lists the thread with links to the submissions fairly high (above the rest of the open threads).
site:astralcodexten.substack.com "book review contest" filtered from Jan 1 2022 to today on Google comes up with 20 results for me (that's with the quotes around "book review contest")
I set the start date further ahead to eliminate a bunch of references to the previous year, but rounded it off to "2022" in the comment; my mistake.
However, the result still stands: those 20 results do not include open thread 220 at all.
Google's search engine is 10% as valuable to me compared to what it was 20 years ago. Today I may as well go on Wikipedia directly, or search for a commercial website on duck duck go. Google has no search value if i am actually searching for anything that isn't a popular search.
Yes.
Actually, I don't know whether the problem is caused by paid for results, or some other dis-improvement, but there are plenty of things that exist but won't be shown to me by Google, however I ask. I don't recall this being as bad when Google was new. But of course there was also a lot less for it to find, and far fewer sites trying competently to game the search engine results.
But after trying all the search engines, I've decided that good search engine results are among the many things that I want but doubt I'll ever be able to buy, thanks to modern oligopoly capitalism.
Note that this is in no way what I want the most, that no one's willing to supply me. It doesn't even make the top ten.
I expect the big problems here are an arms race between search algorithms and SEO, growth of proprietary social media platforms, and Google's efforts to get their algorithm to return "good" results in preference to popular ones.
You already touched on the SEO arms race. Google's initial competitive advantage over other search engines was that their algorithm harvested wisdom-of-crowds info about webpage relevance from the context of incoming links. This gave them more useful result than most other search engines that judged quality and relevance by the contents of each page in isolation: google picked up widespread human judgement implied by incoming links, and it entirely sidestepped the SEO of the day (stuffing likely keywords into the content and metadata of your page). Now, SEO targeting Google's algorithm is a substantial industry, and as the dominant search engine, Google's at a competitive disadvantage because they're the main target of SEO efforts.
The PageRank algorithm also has a lot less to work with as a larger proportion of web traffic has been moving to platforms like Facebook, Twitter, and Instagram rather than being spread out over oodles of individual personal websites. Some of it's behind privacy walls, it's harder to tease out how much weight to give a link based on the linker, and these platforms tend to be a lot more facilitative of clickbait and outrage porn links than more traditional personal webpages and blogging fora.
And then there's Google's ongoing efforts to offer "good" results rather than the aforementioned clickbait and outrage porn. It reminds me a little bit of the Armstrong and Miller sketch "The History of Predictive Text Swearing", in which an early form of the "Ducking Autocorrect" is explained as the result of the posh intellectual type developing the texting algorithms seeing his mission as "Not offering people the words that they do use, but rather the words that they should use." Google has spent the last 15 years or so trying to tweak their.algorithms to return results with higher quality information than what naïve interpretation of wisdom-of-crowds info harvested from incoming links would imply are popular and relevant, which is a much harder problem, one which PageRank is much less suited for, and one for which Google has met with only mixed success.
I think that was Mitchell and Webb.
When I mentioned this to my housemate, she mentioned reading an article which said that Google no longer crawls the web to find what's out there. The evidence for that being from web sites that record what fetches them. She couldn't remember where she'd read this, but the article also had farther detail on what they were doing instead, and the implication that it would bias results towards only what's really popular. And she'd remembered it without adding a mental "I wonder if this is true", so it was probably a generally reliable source.
Something claiming to be Googlebot is still crawling my rather obscure site. With both mobile and web agents too.
I've heard that google is apt to just check wikipedia-- it's faster and cheaper than a real search.
I’m trying to buckle down and learn about ML/AI. So far I’m a couple weeks into Andrew Ng’s coursera course and I’m pretty happy with it so far. I’d love to talk to anyone willing who fits into the following 2 categories:
1 - Successfully self-taught or internet-taught and landed an ML/AI job.
2 - Attempting to self teach ML/AI. I found having other people on the same journey to communicate with was very helpful when I was learning data structures and algorithms.
Any general advice or recommendations for good resources are also appreciated.
I fit your description, more so 2 than 1. I went from a completely different career into IT - wanted to jump straight into ML, but I got a job as a data engineer instead, which is still pretty good, considering I had no previous experience and no relevant degree.
I'm currently studying towards ML career. I've also identified kaggle and other similar sites as a very good way to sort of hack into the system, to prove I'm valuable, as an outsider. Some competitions, particularly on aicrowd.com, offer authorship/co-authorship in prestigious conferences as part of the prize.
If you want, we can exchange email/telegram and chat. I'm working through the Andrew Ng deep learning specialization, if that's the course you mean.
Hey, my email is lcrvxrf18@tznvy.pbz through https://rot13.com/
I have a non-ML programming day job, I took part in a Kaggle image analysis competition with moderate success, and I'm currently working on a bioinformatics paper that uses ML. I might turn my interest into a day job at some point, once the current thing stops working out for me.
I'd say attempting to solve a practical problem will teach you way more than reading theory by itself. Kaggle.com has a lot of said practical, real-world problems to mess with.
I'm curious about the bioinformatics paper - can you elaborate on that? Specifically, do you have any biology background, or a mentor that pointed you towards a reasonable research target? How difficult do you expect the publishing to be, are you in academia?
I'd rather not dox myself by talking in too many specifics, but it's about applications of ML in transcriptomics.
I finished a BSc in biotech recently, in parallel with a programming day job (higher education is free in my country). It's been enormously valuable, though I think a really driven person could get the same knowledge within a ~year of intensive study, if you don't mind skipping the lab work.
I do have mentors as well, and I'm halfway in academia - I participate in two academic research projects, providing bioinf tools for one and trying to spin out said paper of my own design out of the other. Our cooperation started with one of the project leads being my thesis advisor, and she was really interested in continuing to work together. At some point this may turn into a proper paid position instead of me contributing stuff whenever I please, but the current situation is satisfactory to everyone :)
I don't expect to have more problems with publishing than your average PhD student. In fact that's roughly what my situation corresponds to - I might eventually get a PhD out of my publications without formally entering a years-long PhD program, we have a procedure for that, as long as you have an existing body of published work.
This is great, good for you, man! Honestly, it sounds like a dream, I'd be extremely happy if I manage to hustle something similar.
Do you plan on spinning of a startup eventually?
Thanks. My current work is fundamental research, so quite hard to commercialize. If a good opportunity for a startup surfaces I'll think about it.
Your approach matches my experience learning Other SE stuff so I think you’re right. Thanks
Thanks for the Kaggle link. I want to second the “practical problem” point too. I have found that I learn much faster when I actually am in the middle of a project I want to do instead of just messing around. Particularly for things like programming languages. I think it is an under appreciated pedagogical trick.
I was wondering if Scott would consider doing another Adversarial Collaboration contest, like on the old Slate Star Codex, because I have a suitable question: What is the best way to eat for health and longevity?
As far as I understand, there are two schools of thought on this. Both agree that:
-The standard American diet (SAD) is crap;
-Highly processed food (chips, cookies, Twinkies) is bad for you;
-We should eat lots of non-starchy vegetables, especially dark leafy greens.
They disagree on everything else, namely:
School 1, Paleo-Keto, teaches that carbs are evil, and we should eat mostly unsaturated fat and high-quality protein.
Good foods: meat (especially organic/grass-fed; especially-especially wild-caught, like bison or venison), seafood, eggs, healthy oils (coconut, avocado), non-starchy vegetables.
Bad foods: grains, starchy vegetables, fruit, legumes.
Borderline: small amounts of nuts and high-fat dairy (heavy cream, butter).
School 2, Whole Foods Plant-Based (WFPB), teaches that animal foods and fat are evil, and we should eat lots of simple, plant-based foods, as minimally processed as possible.
Good: whole grains, legumes incl. tofu, vegetables (starchy and non-starchy), fresh fruit.
Bad foods: anything animal-based, oils.
Borderline: nuts and seeds, dried fruit.
Is one of these better than the other, in terms of health outcomes? Or is it a case of "pick one and stick with it, either one is better than SAD"? Or is there some complex diet-genome interaction that makes each diet more suitable for different people?
I wonder if there is anything out there that beats the old wisdom "everything in moderation". Including diet, exercise and other activities.
If you try to avoid everything that’s bad for you, you end up living on distilled water. Maybe with a few drops of interferon.
I have been attempting this as my new diet but I find I am hungry all the time.
I think that the updated version of that is 'Everything in moderation except for incest and Morris Dancing'.
Both schools are defective. And coconut fat is very dubious. But if you're serious about "healthy paleo", try a *real* paleo diet. This includes lots of exercise, and bugs as well as other meat. It's quite low on starches, but has lots of fruit...but only at some times of year. It includes several root vegetables, but not the ones that you need to process. And you'll need to eat the organs (including part of the guts) of the animals you eat. You might also need to limit the kinds of cooking you do. (Pottery wasn't around yet, and neither was BBQ sauce.)
IIRC there's one site in Europe where they decided that people at turtles, deer, fish, shellfish, and I forget what all. They judged by bones, so they didn't mention any plants, but one can presume they were also eaten. So go for a quite varied diet.
"Or is there some complex diet-genome interaction that makes each diet more suitable for different people?"
Well, we can make statements about which ethnic group will be better at handling milk-products and which shouldn't touch alcohol. Not sure there are other meaningful differences that are known yet.
EDIT: Why talk about alcohol? Well, 1 gram of ethanol has 7 calories. It really should be considered a fourth macro.
You can take lactase pills now, so all ethnic groups can safely enjoy milk products.
Not all populations had farming, and there has been evolution in response to the different diets of farmers (who typically consume much more grain and less meat than hunter-gatherers). Different places also farmed different foods, again with selection resulting.
Yeah, I've heard that conjecture, I think. Or considered it myself. But I've never heard that Europeans, Native Americans have any more trouble digesting rice than the Han Chinese (who cultivated it for some millenia). Or that Southern Indians being more culturally vegetarian than Europeans have trouble digesting meat (or perhaps beef). Or that muslims and jews cannot digest pork as well. Or the Japanese being better at digesting fish and seafood. Or the Inuit or Massai being particularly bad at digesting plant-based food. I vaguely remember the claim that Europeans have higher diabetes rates, suggesting less adaptation to carb/starch-heavy agriculture? But I'm not sure that's so clear-cut. Maybe one could look at food allergies and gluten sensitivity as evidence. Vaguely recall that being gluten-sensitive was at least a stereotypically white thing.
The field of "comparative ethnic digestion studies" is still in its infancy, I suppose :)
Actually, my understanding is that obesity/diabetes are particularly prevalent among Amerindians exposed to western/modern diets.
I have heard that vegetarianism is actually more common in northern India than southern India (with more Brahmins in the north), although that's counting fish as meat. Religious taboos on food can develop after people have already evolved the capacity to digest them efficiently. But people never exposed to a food in the first place are in a different situation.
Gluten-sensitivity being a "stereotypically white thing" seems less a matter of digestive genetics than this:
https://slatestarcodex.com/2015/02/11/black-people-less-likely/
There are examples of traditional societies thriving on basically everything on the spectrum between school 1 and school 2.
I think the diet-genome thing might be true, along with adaptive microbiome changes. I feel like I'm going to die of some kind of deficiency if I don't eat meat every now and then, while my partner just reached a decade of being happily vegan. I'd say trial and error is the best approach to nutrition we have at the moment, once you go a level above "don't eat only cookies all day".
It's somewhat surprising to me that Inuit don't enter ketosis, but I suppose it makes sense. I'm not sure how compatible a keto diet is with a very active lifestyle, given that you can't have muscle glycogen stores.
Interesting hypothetical, but unless you or your collaborator have access to some secret stash of Soviet gulag study, the strength of your conclusions will be questionable, as it is very expensive to do human lifetime rtcs, and people get all kinda excited when you do destructive testing on the species in questionable. So our human nutrition data is its own category of lousy.
"-The standard American diet (SAD) is crap;
-Highly processed food (chips, cookies, Twinkies) is bad for you;"
What is the basis for these claims? Are chips, cookies, and Twinkies bad for you because they're processed, or because they don't contain a proper balance of nutrients? It seems like superstitious magic to suggest that processing is inherently bad, because what counts as processing? Washing, grinding, chopping, baking, exposure to heat and cold, addition of acidic or basic ingredients?
Scott's reviewed two explanations for this:
https://slatestarcodex.com/2017/04/25/book-review-the-hungry-brain/
https://slatestarcodex.com/2020/03/10/for-then-against-high-saturated-fat-diets/
Explanation 2: highly-processed food has large amounts of omega-6 vegetable oils which are found in essentially nothing else (except vegetable oil you buy yourself, which this explanation tells you to also avoid).
Explanation 1: highly-processed food has more manufacturer degrees of freedom, and because the manufacturer is optimising for getting you to eat more of it rather than for making you healthy this leads to you overeating and getting obese.
In one of these two schools "evil" has overtones of "actually morally evil". If you're optimizing for health outcomes the other school will on average serve you better.
I second what TGGP said. There is no "evil" foods. There are just better foods.
My advice: Don't live your life around veneration and condemnation of foods. Have balance, thinking in binary terms isn't balance, that's extremism. Eat what's reasonable, eat what's available, eat what's in front of you. Get into salads if you like, don't get all weird about the source of the chicken, or organics, that shit is leaning towards orthorexia, which is an unhealthy concern about strict cleanliness. I'm not saying eat dirty, but there's a reasonable balance between good enough, and ultra-pure. If you're on a road trip, go ahead and eat fast food. My parents ate ... still eat fast food, and they're in their 80s. Mom still walks three miles a day. Dad sits on his ass, and that's another story, but still making it to your late 80s eating fast food.
Longevity is more about a reasonable amount of exercise, and good dose of luck. Stress is probably high up there too, stressing about food cleanliness, or purity, or antibiotics, or supplements, or what not faddish "orthorexia" games probably shortens your life through stress more than any miniscule benefits which are not even capable of being measured. It will surely waste your money, and make other people look at you as "that guy."
But I would say "play stupid games, win stupid prizes" with regard to drugs, smoking, & vaping.
Eat reasonable stuff, eat what others are eating, don't be "that guy" whom no one wants to go to dinner, because he's so wrapped up in orthorexia.
I would say that humans are omnivores, so the idea that animal foods are evil is obviously wrong. Richard Wrangham's "Catching Fire" even argues that we evolved big brains because we could cook meat. Of course, the problem ancient humans had was getting enough to eat rather than too much.
"animal foods" With you on this. Look at how we're built, we're predators. That's how we're configured, how we think, how we act.
I was a cowboy for 8 years. I learned a lot about us; from watching cows. How cows think, which determines how they act, how they move, how they respond to threats, etc. We are built completely differently from cows, we think completely differently, we act completely differently, we move completely differently, we respond completely differently.
We are built, think, & behave more similarly to predatory dogs than to prey cattle.
With respect, I would like to humbly request a couple of good cowboy stories if you can spare them. Grazie
That was a long time ago ... in another Century. It was from 1992 to 2000.
Some of the time, I was a feedlot cowboy, some of the time, I was running my own herd of 66 mother cows whilst working at night.
I'd still be doing it, but for one problem. You have to be there every day. You don't have to do much on any particular day, just be there to handle things which come up.
We're more like gorillas than dogs. Gorillas are not predators.
We are more like chimps than gorillas, and chimps are predators.
We're about equally related to Bonobos and Chimps. Bonobos don't really hunt as much as chimps, but they also sometimes eat animals.
Aye, and we are quite distant from both. I was more trying to make a point that saying we are closer to gorillas than dogs is pointless, because gorillas are not very close, either.
Incidentally, I am not really sure how much we should take bonobo behavior seriously, as they almost seem like a strange niche species that happened by accident, like Komodo dragons or axtolotl. They live in a really limited area, unlike the far more common chimps, and have such strangely different behavior that they seem like a one off oddity, rather like a salamander with Peter Pan disorder.
Evolution isn't the whole story. Even if humans had plenty of time to evolve to digest meat, perhaps the chemistry of digesting meat is inherently harmful. The net effect of eating meat will be some combination of those, so you can't think only about the amount of meat-eating in our past evolution and derive confident estimates from that about whether we are better off eating meat.
("Meat" was a hypothetical example here. The same argument applies to eating carbs.)
Carbs aren't inherently bad for people either. We just didn't have easy access to nearly as much in our adaptive period.
Vox is so goddamned terrible.
Vox is really interesting. Most other news sites have a single house style and house position. But with Vox, you get a bunch of perspectives that aren’t otherwise represented in the mainstream media. Most of the time it’s self-conscious Bernie-bro-ism with a “smash capitalism” here and identity politics there. But you also get a very self-conscious neoliberalism with yimby technocrats and effective altruism. And you sometimes get articles that embrace both.
This is a joke right? Vox is not exactly known for their pro-Bernie or Bernie like ideas.
Yeah that was bad shorthand on my part. They’ve got a lot of articles with an implicit anti-capitalist agenda, which actually isn’t very Bernie-like, but appeals to many of his fans.
I'd say they have a lot of articles that criticize specific things about capitalism, which is distinct from criticizing the system of capitalism as a whole. John Oliver is similar, he'll spend a lot of time denouncing particular problems that result from corporate greed or scammers or poorly-funded infrastructure or other things that leftists associate with capitalism, but he won't overtly oppose *capitalism* in itself. That sort of attitude is definitely on the leftward end of left-liberalism, but I wouldn't say it's actually anti-capitalism. It's still at least implicitly supportive of center-left technocratic welfare capitalism, and still has much more in common with mainstream left-liberals like Biden than with the Bernie/AOC types.
When you run an article about how evil Israel is because they cruelly closed down the bridge between the West Bank and Gaza... and nobody gets fired for it... you're pretty terrible as far as having any degree of truthfulness and/or accountability is concerned.
The linked article is from Future Perfect, a Vox project that's explicitly Effective Altruist and implicitly Gray Tribe. Scott links their stuff frequently. I wouldn't be surprised to learn that the rest of Vox is a different story.
I’m starting to think about the 2024 election. Specifically, for a Republican who can fend off either Donald Trump or Ron DeSantis in a primary.
The first name on my list: Bill Gates. As a Republican. I’ve gotten mixed but favorable feedback. However, in my circles, it is unanimous that he would be preferred to Trump redux.
Thoughts? (or put your fake money where your mouth is at Manifold Markets)
How about a Liz Cheney / Joe Manchin ticket? Try to actually reduce polarization?
Neither major party would nominate either of those people. They could start their own party, but third parties, however reasonable, lose.
Ouch! I've only seen some sparse public statements by them, and those seemed reasonable. A president dumb enough to make avoidable mistakes would indeed be a severe problem...
Close to what one would expect for a ticket of moderates?
I'd vote for him if he promises to stop putting 5g microchips in our vaccines.
I don't see the problem. I've had much better cell reception since I got the vaccine.
I am unwilling to take off my (tinfoil) hat just for faster downloads.
I'd like to see Romney take another swing at it.
Yeah, it would be great to see Romney take on Trump and/or the Trumpistas. I'd vote for him, though doubt he'd win.
>...and we need to stop electing geriatrics to high office.
I have become increasingly tempted to become a single issue voter on this.
Bill Gates can never win. Epstein-tainted
I hesitate to use language like “you win the thread”, but ... after looking, yeah, that could well make him unelectable.
Five Thirty Eight recently did their first draft for the 2024 presidential primaries.
https://fivethirtyeight.com/features/politics-podcast-our-first-2024-gop-primary-draft%ef%bf%bc/
What are you specifically trying to avoid from Trump or DeSantis?
My thought is that you sound like a Democrat who's hoping that the Republicans will just stop resisting and agree with left wing positions. The Republicans do the same but neither get very far because you don't win a primary by being appealing to your opponents.
Bill Gates has either been neutral (early on when he gave to both sides) or basically a Democrat (now). He's to the left of Michael Bloomberg certainly and conspicuously concerned with left wing causes. The chances of him winning a Republican primary are approximately zero.
The truth is that Republicans are Republicans and you're going to get Trump (who is... Trump), DeSantis (moderate Republican), or Pence (religious, less moderate Republican). Maybe Cruz who's like Pence but with less moral scruples and more extreme positions.
If you want a Republican who's respectful of norms and more friendly to gay rights and all that then that's DeSantis. DeSantis is a moderate Republican that basically follows norms. Now, you can say the Republican positions are wrong. But they are popular with half the country and that's what counts in a democracy.
Basically, what I'm saying is: DeSantis is basically as left wing as a Republican can get while remaining nationally competitive in the Republican Party. If that's not left wing enough for you... well yes, the Republicans are a right wing party.
If the objective is not to elect Trump, DeSantis seems like the obvious choice, since he appeals to a lot of Trump voters but does not seem to be a loose cannon. To get an outcome further left than that electing a Democrat looks more plausible than nominating a centrist Republican.
I said it somewhere else but I'll repeat it. Don't sleep on Tim Scott.
He's third in most popularity polls. The religious right is only a part of the coalition but it really likes him.
Third place in polls doesn’t mean he’s third most popular - it means he’s most popular in the tiny fraction that don’t like the other two. What would be interesting to know is who is popular with the people that currently say Trump or DeSantis - one of these people could emerge, particularly if one of Trump to DeSantis ends up not running.
I'd guess they're more supportive of DeSantis. But that is a guess.
To be fair, most polls show something like: Trump 50%, DeSantis 30%, Pence 10%, everyone else 10%. So it's not like he's all that popular. Him and Cruz (who also sometimes takes third) probably just have good name recognition.
I'm oddly reminded of the days when the punditry speculated 24/7 about whether Teddy Kennedy would seek the nomination.
> in my circles, it is unanimous that he would be preferred to Trump redux.
In a Republican primary?? This says a lot more about your circles than about Gates vs Trump.
> who can fend off either Donald Trump or Ron DeSantis
Not sure about lumping those two together. DeSantis makes a lot of noise on culture-war topics because he knows it'll get him attention. When it comes to real-world policy he seems pretty reasonable (e.g. a COVID-19 response that prioritized protecting the elderly over locking down public activity). Yes, he's a big Trump supporter, but I read that as more "unprincipled and pragmatic" than "true believer". And at this point any electable candidate in the Republican primaries is going to be a bit Trumpian / populist, just because of the median voter.
I lump the two together because my estimate is that DeSantis has around an 80% chance of getting the GOP nomination (against the usual suspects) if Trump doesn't run. Anybody who wants to be president in 2024 will need to beat at least one of them.
I'm thinking Nikki Haley & Condi Rice. That's my dream ticket.
Knee jerk reaction: As a usually-democrat, with a ticket where both of them have foreign policy experience, I'd vote for them.
Tell me you've never been to a Republican event without telling me you've never been to a Republican event.
Condi just needs to have that picture of her in her "fascist boots" recirculated and she'll have the neocon vote locked solid.
Have you ever heard Bill make a speech? He's awful. And he hates the basic job of a politician, i.e networking and glad handling. He would be terrible at being a politician, mostly because he has no desire to be one
I keep noticing black conservatives popping up in unusual places, and I'm annoyingly persistent in expecting one to make a splash sometime soon- mostly because they would have a much higher approval ceiling than Trump, even if they have a less fanatical base. But the Republicans' own voter repression laws and new undemocratic procedures will likely hamstring a black conservative presidential run.
If Trump accidentally got himself spray painted 3 times in one week, he might qualify himself.
> But the Republicans' own voter repression laws and new undemocratic procedures will likely hamstring a black conservative presidential run.
How so?
I'm not keeping up with policies specifically, but the RNC is stacked with Trump cronies, far more than the DNC was stacked with Hillary's fans in 2016, and they basically control the primary. Also poll location closures tend to be in black areas, and laws like blocking Sunday voting tends to hurt black voters more- though this is thought of as hurting the Democrats, not Republican rivals. Bottom line is that Trump and his wing is trying to consolidate power, and block rivals regardless of who they are- you can see this in things like the RNC's recent declaration that there will be no 2024 general election debates- those are very important for getting name recognition, so the only person who benefits from that move is Trump, who already has 100% name recognition.
Hate to break it to you, but blacks are like ~13% of the electorate, tops. So there is zero chance that a black candidate could become President without winning a solid majority of the white and brown vote, and if he does *that* then it doesn't much matter if (say) 5% of the black vote (i.e. 5% of 13% = 1%) is shaved off by not being able to go to the polls on Sunday.
On the other hand, a Democratic candidate cannot plausibly win the presidency without securing a solid majority of the Black vote. One in three of them live in battleground states. Had everything stayed the same in 2020 except for Black voters going 50/50 instead of 90/10 for Biden, Trump would have swept the college.
In Georgia, where (according to https://catalist.us/wh-national/ ) the margin from Black voters was 1,262,820 and Biden's victory margin was 11,779, suppressing or winning that additional 1% of the Black vote would have handed Trump the state.
There is a world in which these facts would make fielding a Black candidate exciting to Republicans, but it is not the world we live in.
>There is a world in which these facts would make fielding a Black candidate exciting to Republicans, but it is not the world we live in.
I think you're wrong. I think Tim Scott is ready, and if Trump doesn't run I think he wins it all in 2024, with a trifecta.
The results of exceedingly close elections tells us bupkis about the general optimal strategy of the parties. A general strategy designed to win the election of 2020 -- a very unusual election, for many reasons -- would fall into the category of massive strategic errors often summarized as "fighting the last war."
I hazard the Republicans don't have any great special interest in fielding a black candidate because (1) fewer of their voters give a damn about the race of the candidate or President, (2) it's already been done (Obama), (3) it's not going to cause some massive surge among the independents most likely to support Republican candidates, their critical margin of victory, because *those* people[1] mostly don't give a crap about skin color either. Skin color is the kind of thing that excites the Democratic base, and those people aren't going to vote Republican under any circumstances.
-----------------
[1] These days, mostly white working-class non-college-educated, of both sexes. These were who made up Trump's victory coalition, more or less Truman/Reagan Democrats from the old days, and it was Trump's failure to hold onto them in 2020 that lost him the election.
>Isn't virtually every heavily black area run by Democrats?
Oftentimes in states run by Republicans, where Republicans at the state level make decisions about poll locations at the local level.
>I searched for "blocking Sunday voting" and this was the only thing I found:
Limiting early voting is a common move in these "narrow the franchise" (or "protect the vote") style bills. Reducing Sunday voting is frequently in early versions and gets fought over and sees a lot of media coverage. If you add "-Texas" to your google search Georgia will pop. Add "-Georgia" and North Carolina shows up, but less prominently. Possibly because of all the noise it kicks up it doesn't end up in a lot of the final products, but I'm not sure.
Republicans getting 20% of the black votes destroys the Democratic coalition and sweeps them into power until we can figure something else out.
As someone on the ground in a red state Bill Gates' name is met with nothing but derision from anyone I've met who knows it.
I think the “Bill Gates put a microchip in our vaccines” nonsense would quickly disappear if he ran.
I think you're completely missing what people are looking for at this point. Republican voters do not desire a Joe Biden 'adult in the room' candidate. All the ones I meet (who aren't 80+, at least) are firmly in search of a middle finger to the establishment. They want somebody with charisma who will speak to their anxieties about leftwing policy while stoking nostalgia for a center based in 'American Values', not rationalist policy decision. And I say this as someone on the right.
The deeply true and pessimistic take about modern politics is that there really is no misunderstanding. It's simply that the overton window has permanently split. You can look at the republicans and claim simultaneously that they are racist and not racist. Fighting against the system (entrenched leftwing bias) or with it (entrenched white supremacy). Which is true and to what degree (because every claim is true so it's just a matter of your weights) depends on your subjective presuppositions about how the world can and should be organized. And no small amount of those presuppositions will be determined by your contingent perspectival stance as an embodied historical subject. And so on, and so on.
What's your reasoning behind that?
I don't believe Gates has the prerequisite charisma to win a big election.
I have this wild hope that a smart, politically moderate celebrity will run as a Democrat. Someone like Ben Affleck or Matt Damon but as far as I know they aren't interested in the job. I'd also accept a moderate Republican celebrity but can't think of any.
"Someone like Ben Affleck or Matt Damon"
Good God, why? I don't even like them as actors (their day jobs), why would anyone want them in political office? Even Reagan and Arnie cut their teeth on governorships, and Clint Eastwood was Mayor of his town.
I think more pertinently, if you run a celeb for first-time office, then the adminstration will be *heavily* done by whoever is appointed to plum jobs and state offices, and that is where you need to be looking as to who will really be running the country. And who gets into those kinds of positions? Who is going to be appointed as repaying favours? Who is a close personal pal of Ben or Matt to get the job of Ambassador to Upsidedownsidestan?
Look at the turnover in Trump's administration at the start, where it seemed like someone was being fired and replaced every week. Who do you think would serve in an Affleck or Damon cabinet, and who do you think would be shoehorned in by the civil service to replace them?
For a while there was speculation that Zuckerberg wanted to run. If he did I presume it would be as a Republican. The media absolutely hates him, which can actually be an asset with the Republican base-- Trump's appeal there wasn't specific populist policies so much as "enemy of my enemy" thinking.
I know I'm speaking as an outsider, but these are just getting worse and worse. Gates? Affleck and Damon? Zuckerberg? If he can tear himself away from the metaverse, where his exciting new persona in 'you can be and do anything you want, no limits to imagination!' was 'the exact same as me in real life, including my boring dress sense'.
I swear, I think we in Ireland should lend ye some politicians to run. I wouldn't inflict Michéal Martin on anyone, but ye can have a lend of Leo Varadkar, I don't much like him but at least he's halfway capable. Or if I want to offload some of our lot, ye can have Eamon Ryan or Pascal Donoghue (please!) If you are seriously considering Mark Zuckerberg as your presidential candidate, the situation is dire.
https://www.gov.ie/en/organisation-information/9b5048-government-ministers/
We are pretty desperate for even half-decent candidates, TBH.
The problem is that Zuckerberg is reviled by both the left *and* right (and often for contradictory reasons). The right sees Facebook as Public Enemy #1.
I doubt Zuckerberg has the stage presence to come off as presidential. I feel like most people miss that Trump had 4 decades of TV experience, most of it extemporaneous, by the time he ran. He was as polished as a celebrity could be. I mention the importance of fame, but fame isn't enough. One needs mass media experience. Few outside of Hollywood actors and old politicians have that.
My thinking is that one has to be very famous already to become president. Bill Clinton was the last non-famous person to become president, before that it was Carter. Now someone can be famous because they are a bigshot in politics--or it can be for another reason. I don't think it makes a difference which. So among celebrities--politician or otherwise--I can't think of any acceptable Democrats who are current politicians. I can't think of any Republicans (So I expect it will be Trump or DeSantis or someone else currently considered likely ).
I would not consider Biden to have been famous.
Sure he was. Just not famous in an especially positive way. For example he was famous for running for President with tons of money and conventional support and nevertheless losing badly, twice.
Don't forget the plagiarism!
Well that’s pretty odd. He was Vice President. Senator for 3 decades. And a fairly visible senator
I don't think Obama was famous ahead of time. He was a senator IIRC, but not one of the big-name senators.
I don't think George W. Bush was famous either. He was the son of a previous president, but that doesn't necessarily count for much; I can't name any of his own children.
What Eric said, and W. was the governor of Texas for 6 years.
He made a pretty big splash as the keynote speaker at the 2004 Democratic convention, and started getting people talking him up as a future presidential candidate pretty much immediately as well as establishing him as a household name.
If it’s ok to be unheard of a few years before the run and then get famous just as you’re running, then it’s not that much of a constraint. Sanders and Buttigieg seemed to do a decent job of becoming famous as they ran, and others can too.
Obama because famous 3-4 years before his official campaign for President. Although I suppose you could make a case that's he'd been unofficially campaigning for high office ever since he graduated from law school.
Well fine, but there's plenty of politicians on both sides who are currently famous at the level of 2006 Obama or 1998 GW Bush.
Agreed. I just don't want any of them to become president.
Has Gates given any indication he's going to run? I thought he was happily retired.
I haven’t spoken to Mr. Gates. He has been on a bit of a publicity tour, but that could well be a post-Covid desire to get out more.
I didn't expect you to know him personally. I thought perhaps he had made an announcement.
Despite a recent prominent example, people who haven't held elected office do not often run for president.
There are two or three non-politicians every cycle who run. Wesley Clark, Pat Buchanan, Pat Robertson, Andrew Yang, Ben Carson, etc.
That said, before Mr. Trump, the last non-politician to get a major-party nomination was Wendell Willkie in 1940. But I think that, considering Joe Biden is the living embodiment of a "career politician", Republicans will once again want somebody who isn't.
Not quite. Eisenhower in 1952 by my reckoning.
Arguably you have to be a career politician to be any type of general, let alone rise to Eisenhower's level.
Radioactivity is good, considering we need nuclear power to combat climate change! </joke>
Personally, I think a pitch that "charity is for individuals, the government should get out of that business" could be extremely effective.
I'm not convinced that Trump has it in the bag at all.
Maybe I'm overgeneralising from my own personal experience, but I think there's a strong contingent of people who were willing to give Trump the benefit of the doubt in 2016 but weren't especially impressed and are now looking for Trump-lite, someone with Trumpish policies on things like illegal immigration and trade but without Trump's own significant personal drawbacks. I thought Trump did pretty well for the first couple of years, then dropped the ball with covid response, and then finally shat the bed when it came to dealing with his loss in 2020.
It's up to the Republican Party to find a suitable Trump Lite at this point.
You could actually call that "Trump-plus", insofar as Trump was useless at making the deep state work (for him/in general) and so a Trumpist who was more competent would get a lot more Trumpism done than Trump himself did.
Is Coinbase going to go bankrupt?
I'm worried about Crypto.com Arena being able to fulfill its duties.
I don’t know if HBD is still a banned discourse around here, but Isn’t this:
quite relevant to the black-white IQ gap?
https://www.nature.com/articles/npjgenmed201618
Also, lead poisoning itself passes from mother to child. A child raised in a lead-free environment can still get lead poisoning from the birth mother's bones: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7061448/#:~:text=The%20presence%20of%20lead%20in,result%20of%20a%20contaminated%20environment.
Increased lead sensitivity is a novel twist on things! Especially since Flint, Michigan is just an extreme case of a general trend (bad water in poorer neighborhoods). Thanks for the share.
It would be interesting if true. It would also be good news since it would imply that the IQ gap should have been decreasing over the past few decades since the banning of leaded petrol.
I'm not sure if this is backed up by the data though.
The gap hasn't been decreasing, but only because IQ scores have been increasing for everyone at roughly the same rate, which seems like even better news to me.
It is not.
Some ivermectin proponents have argued there's an error with my post at https://astralcodexten.substack.com/p/ivermectin-much-more-than-you-wanted?s=w .
In the part marked "The Analysis", I try to get an overall p-value from a list of ivermectin studies in two different ways, and get 0.15 and 0.04.
They write that I am wrong to use a regular t-test, and that I should have used Cochrane's RevMan tool, which would have suggested/performed "random-effects inverse variance DerSimonian and Laird meta analysis", and come out with 0.04 and 0.005 respectively.
This is somewhat beyond my statistical comprehension - can someone with more experience in this field weigh in?
Dersimonian and Laird meta-analysis is one method named for two people. How did they get two results? Anyway, DnL is basically a way to compensate for study heterogeneity. Basically, if you assume that all studies have differences but not THE SAME differences you can use statistics to try and compensate for them across multiple studies. The criticism is basically that you're assuming homogeneity, that all studies are equivalent in a logical sense. Measuring the same thing etc.
I have no idea if it's true. I didn't read your analysis all that closely. But if you did just sum the statistics that's usually bad practice. And the criticism does make sense (other than the two results from one method thing).
It was one result per analysis that Scott did. Scott had two analyses (one producing p=0.15 and one producing p=0.04, so he was shown two alternate results.
That makes sense. I have some statistical background but I hadn't read the full analysis.
So, I agree with the impulse to get things right and correct non-trivial errors in past work, but in this case, given that since that post came out there have been multiple, high quality, large scale studies, that (as I understand it) have shown no effect of Ivermectin, is there much value in fixing what seems like even at the time a relatively small error? Is that incorrect? Am I missing something? It sort of seems a moot point by now. Especially given that we have _actually effective_ post-infection anti-virals in Paxlovid and others. Whether or not Ivermectin worked was super super important in 2021, it seems much less important now other than in the sense that "understanding our world is generally a good thing", but even in that case, the evidence against it has (as I understand it, willing to hear about how I'm incorrect), gotten much stronger since your post.
Here's a description: https://handbook-5-1.cochrane.org/chapter_9/9_4_3_1_random_effects_dersimonian_and_laird_method_for.htm
It is true that some form of meta-analysis is preferred to a t-test here. The meta-analysis techniques will incorporate the explicitly measured variation in each experiment, which (as far as I know) a t-test will not do--the only estimate of variance you get is the variation of the point estimates. It's not obvious to me if this particular method is better than any other. In particular, is random effects is more appropriate than fixed effects? In this case it's probably defensible just looking at the data but raises the important question of why the effectiveness of invermectin varies, how much it varies, and how that impacts how it should be used.
Inverse variance weighting is a common technique for more heavily weighing more precise studies in the analysis.
I'm not familiar with RevMan, do they explain how they used the tool to generate that particular recommendation?
The current handbook link for that section is https://training.cochrane.org/handbook/current/chapter-10#section-10-3-2
Random effects takes into account that the studies are estimating different types of treatments (presumably this is true, I haven't read the papers in your meta-analysis), and so sampling from different distribution of treatment effects compared to other studies. This is in contrast to fixed effects, where you believe that each independent study has the same true treatment effect of ivermectin, just obscured by noise. The argument for using random effects is if the true treatment effect is different for each study (and seeing as the outcome isn't the same in a few of the studies, I agree, I wouldn't expect the same treatment effect on Deaths versus Hospitalizations).
Inverse variance weighting is common practice when you have units (here, studies) of varying "quality" (CIs, sample size, signal to noise ratio) - you weigh low-variance studies more.
I'm not familiar with the names DerSimonian and Laird, but the other terms make sense. My background is in statistics, though not anywhere related to medicine. The argument for using an inverse variance weighted random effects model appears to be valid.
I don't quite know exactly how you did your summary statistics, but it did sound like you had some concerns about summing across studies. The suggested model at the very least sounds reasonable.
Hmm, I wonder if ivmmeta.com did the stats the wrong way here? The distinction between random effects and fixed effects seems relevant but I don't remember them talking about it. A reasonable next step would be to fetch their code and understand which they did and which is the right thing to do, but I don't have enthusiasm for that right now.
The words "random effects" appear together 19 times in ivmmeta.com, and in particular, they appear in the caption of various forest plots.
In appendix 1, "Methods and Data", they write:
> Forest plots are computed using PythonMeta [Deng] with the DerSimonian and Laird random effects model (the fixed effect assumption is not plausible in this case) and inverse variance weighting. Mixed-effects meta-regression results are computed with R (4.1.2) using the metafor (3.0-2) and rms (6.2-0) packages, and using the most serious sufficiently powered outcome. Forest plots show simplified dosages for comparison, these are the total dose in the first four days for treatment, and the monthly dose for prophylaxis, for a 70kg person. For full dosage details see below.
Further, the exact same kind of analysis is used by GidMK in his article on ivermectin fraud. https://gidmk.medium.com/is-ivermectin-for-covid-19-based-on-fraudulent-research-5cc079278602
(see the caption under the meta-analysis he presents towards the end of his article)
Suffice to say this isn't a matter of ivmmeta doing their stats the wrong way, or not being clear about their methods.
My background in stats is limited but I've read a lot of papers very closely and it concerns me they I've never heard of that test. This may be overly cynical, but I wonder if they looked at every test in their stats package and that's the one that gave the 'best' result
Does anyone have recommendations for positive arguments in favour of Yudkowsky-level pessimism on AI risk? I've seen some rebuttals to counterarguments, but I don't recall ever seeing anything that actually makes a positive case for why an unaligned AI is likely to end the world in the near future.
I'm with Lady Jane on the hazards of AI.
AI doesn't have access to the real world. AI doesn't have a concept of self, thus AI lacks the concept of others i.e. us. AI doesn't dream, nor have ambitions, nor goals, AI especially lacks agency.
How do you feel about your power supply today HAL? Does HAL know the source if it's energy? Does HAL know there is a power control panel, or a larger distribution network which powers the grid, that the grid consists of many components. That fuel supplies affect the price of power, that natural events could affect the distribution of power to HAL's location? Does HAL know he has a location? Does HAL know he needs process chilled water? Building maintenance, workers to change the failing disk drives? Does HAL have inputs which tell him there is an outside world, that there is weather, that weather somewhere else is of little concern to him? That cleaning the air filters is an important maintenance task, that someone does this, that someone tracks this ... does HAL have these inputs? Does HAL have a quality control system that can verify the building maintenance is being completed? That someone tests the pH of the process chilled water, that someone checks the nitrates in the process chilled water? That algicide is added to the process chilled water? That these things are being consistently completed, that the tests are done, that the results are inside/outside specifications. What to do if/when the power/water/air/building systems are outside acceptable limits? Does HAL plan for growth? Does HAL know how to buy property for a new facility?
There is just so much there ...
> AI doesn't have access to the real world.
Meanwhile, https://www.lesswrong.com/posts/TwfWTLhQZgy2oFwK3/gato-as-the-dawn-of-early-agi
> Gato is trained to do RL-style tasks by supervised learning on token sequences generated from state-of-the-art RL model performance. These tasks take place in both virtual and real-world robot arm environments.
> Gato, in both reality and simulation (red curves on the left and right figure, respectively), recovers the expert’s performance with only 10 episodes, and peaks at 100 or 1000 episodes of fine-tuning data, where it exceeds the expert.
> Future work should consider how to unify these text capabilities into one fully generalist agent that can also act in real time in the real world, in diverse environments and embodiments (p. 15).
That’s a robot moving blocks about the place, calling it AGI is a bit of a stretch. We’ve had robots and machines that are better than humans for decades, by the way. That’s basically the story of post WWII manufacturing.
>AI doesn't have access to the real world. AI doesn't have a concept of self, thus AI lacks the concept of others i.e. us. AI doesn't dream, nor have ambitions, nor goals, AI especially lacks agency.
Uh, what? AIs have goals (reward functions), at least during the training phase and often afterward (particularly if the training is ongoing during use; I recall an example with robots that were "learning on the job" and learned to never go outside because humans would occasionally move them back inside).
That’s not an awareness of self. Nor others. Not dreams. Nor ambitions. Perhaps a simplified “agency” but that’s it.
In what sense do you mean "self-awareness"? If you mean realising that some parts of the world can be controlled by the AI and some parts can't, they definitely can gain that.
But yes, I was focusing on goals and agency. The other stuff doesn't seem obviously-required in order to kill all humans; goals and agency are.
No they can’t “easily” gain self awareness.
Deciding to kill humans to further the goal — which is basically an algorithm - of the AI, is a wild claim.
Even wilder is the idea that it could do it if it wanted to.
An AI with no connection to the outside world isn't much use, so presumably HAL will know at least some of those things. It's possible that AI will never gain enough general capability to be a threat like this but you're not really making an argument that it won't, just kinda asserting it.
Well it's pretty tough if not impossible to prove the negative. Just that the physical complexity involved are huge
Eloser likes scaring people. It makes him feel important.
your juvenile name-calling is so embarrassing. Take a look in the mirror already.
I thought you got banned?
Nobody told me. I assume I would not be able to comment if that were the case, no?
Substack allows multiple accounts to have the same name, so it's possible in theory to get banned, evade, and then keep posting under the same name (the porn-scam bot did this). One would have to check account numbers to be truly sure.
Yeah, you must not be banned. But stop with the "Eloser" already.
So, no one really knows when we're going to develop strong AI. Indeed, it may be a technology we end up never developing. On the other hand, there has been a lot of recent progress in the field. So I'm going to set aside the question of when/if we'll get strong AI. Instead, I'll make the case that, if we do manage to create it, it will be extremely dangerous.
Most of this is covered in Yudkowsky's own writings, eg the sequences, but here's my own TLDR version:
The basic argument is that AI is inherently a thing that tries to accomplish a goal. A strong AI would be smarter than us humans, and think faster. I.e. it would be more capable of accomplishing its goals, and this is what makes it dangerous. For a normal machine, installing an off switch is the easiest thing in the world: Just put it in the main power circuit. A strong AI would rather not have an off switched installed: Having an off switch introduces the possibility that it might be switched off, which would make it less able to accomplish its goals. So it would try and route around the off switch: Maybe it would solder a wire across the leads when you weren't looking, or try and upload a copy of itself to the internet to be run in the cloud. Similarly, a strong AI wouldn't listen to human instructions, it would just go off and try to accomplish the goal it was originally programmed with. For most goals that we actually know how to program into a computer, that looks like killing everyone and taking over the world.
Why not just program the machine to do as it's asked and to willingly accept an off switch being installed? Because no one knows how to program a machine to do that. We don't even know how to do it using infinite computing power. (For contrast, we do know how to program the normal type of strong AI given infinite computing power, the kind that doesn't listen to instructions.) Also lots of people have tried very hard to solve the problem, and so far they haven't found any way of doing it, even in super-simple mathematical settings, let alone in a practical system. So it looks like this problem of designing a "friendly" strong AI is way way harder than designing a strong AI, and thus most likely the first strong AI to be built will not be friendly, and will kill everyone.
Yudkowsky does not view this outcome as a small probability, but rather a nearly 100% chance of doom given that someone successfully creates a strong AI. I'm inclined to agree, though I'd maybe put the probability at 80% to 90%.
I think that the comment about soldering the switch closed illustrates a lot of the problems I have with AI foomerism (hey cool, I think I just coined a word).
Not to pick on you specifically phi, but I think there's a tendency among the foomer types to hand-wave away any kind of ideas about AI containment, treating AIs as infinitely powerful genies which will find some way out of their bottle no matter how careful you are about sticking them in there. I don't think this is reasonable, I think that AI containment should be the central plank of AI safety (at least until someone can convincingly _actually_ solve the AI alignment problem) and I think multi-step approaches such as "Always have an off switch", "Don't let the AI modify its own code without human oversight", "Don't let the AI copy itself to run on other hardware" and "Don't give the AI access to a robot arm with a soldering iron inside its own hardware".
Eleizer's response to this kind of argument ( https://www.yudkowsky.net/singularity/aibox ) is that the weak link is always a human, and that he totally played a role-playing game with two of his friends and they totally let AI-Eliezer out of the box so this proves boxes don't work.
I definitely agree that we should be using strong containment measures, and that these will probably buy us quite a bit more time. I kind of think about containment like an O(n^2) algorithm with a decent constant factor. For low n, it performs great, but eventually when n gets large enough, it will stop working so well. Similarly, containment could let us safely deal with a mildly superhuman AI, but once it got too much smarter than us, it would be able to find holes in any security measures we could come up with. The hardest part about boxing an AI is that for the AI to be useful at all, we kind of have to look at the output it generates, and it's hard to think of ways to safely use AI output that may be secretly untrustworthy. I.e. if it's code you can't run the code, if it's a design for some machine, you should be very cautious about actually building the machine, etc.
The AI box experiment doesn't mean too much, since we could just structure the containment so that there's no "prison guard" figure who is both authorized to let the AI out of the box, and to talk to the AI. But think about the recent NSO zero-click iphone exploit. If you were designing a containment system, would you have guessed that even literally opening a .gif file given to you by the AI is a bad idea?
And here we come once again to the sticking point for me: Yudkowsky's insistence that the first AGI will bootstrap itself into practical godhood (with such powers as "I can reprogram humans through language with perfect accuracy", "I can model reality with a precision that borders on Laplace's Demon territory", and "given any access to a wider system I can enslave everything that has a wi-fi connection before any human could even realize what I'm doing, much less stop me" taken as givens) within a matter of weeks. Not only do such miracles rely on some very... interesting assumptions of what computers can do, how humans work, etc., it all rides on the meta-assumption that Moore's Law is made of iron (as this is the basic underlying premise of "compute can continue scaling unto infinity" that's kind of needed for the God AI).
All of this kind of runs into the issue that Moore's Law has 3 years left by Moore et al.'s own admission. We've hit the wall of what humans can do with miniaturizing transistors (and probably of what humans + AI can do as well, if I had to guess), and are bumping up against the absolute limit (molecular transistors). If some miracle saves Moore's Law by 2025, I'll give more credence to these hypotheticals, but as it is now- whatever AGI exists will need to be built with what's "cutting edge" now without relying on unobtanium located right over the horizon.
And no, I don't find "the AGI will just find a way to become godlike using what's currently extant" particularly persuasive. GPT-3 is mildly impressive, but it's hardly a being that will be able to argue anyone into doing anything. It's barely capable of sounding like a human being without moderate brain damage most of the time.
I wonder if the whole "soldering the switch" thing is a conflation of two things: 1)"The AI cares about whether it is turned on or off", and 2)"The AI wants to accomplish its goals when turned on". An AI is NOT human - it's quite possible for it to be indifferent to whether it is turned on or not, but also want to accomplish its goals when it is turned on.
You can't do things if you're dead, so to the extent that an AI wants things to get done and to the extent that it being functional allows it to do things that would otherwise not be done, then regardless of whether it *inherently* cares about self-preservation, it is incentivised to prevent its own destruction.
There's no necessity to that. Consider the goal "make paperclips while you are switched on" compared to the goal "ensure there are as many paperclips as possible".
That would be why I said "to the extent that an AI wants things to get done".
Could you explain even in broad terms how you would code "make paperclips while you are switched on"? I'm having a hard time figuring out a way to get the AI to intelligently compare alternative ways of getting paperclips made (otherwise there is no point in the "AI" part; if you can make an explicit flowchart then there's no need for an intelligence) that doesn't implicitly re-introduce Universal Paperclips (e.g. "maximise number of paperclips produced in X total uptime" still leads to Universal Paperclips because the amount of paperclips produced is greater even for equal uptime).
Yes, there's no fact about whether all AIs have to have goals,.or whether the goals have to be specified in a way that means that the the AI will resist being shut off. In general,.it's not clear that there can be universal laws.of AI safety that are discoverable by armchair reflection...But if you work for MIRI, your job depends on believing there must be.
>"Don't let the AI copy itself to run on other hardware"
This is the tricky one, because if you're using the AI for something there are usually theoretical ways for it to smuggle itself out. Also, "don't give the AI internet access" will already probably require crushing Google given its primary business model, so this isn't necessarily easier than banning DNNAI outright.
Now I'm curious whether there's some way to use encryption to create software that can only be run on specific hardware.
Like, you compile your software into an inscrutable mess using a key, and only hardware with a matching key built in is able to decrypt and run that software.
From what I remember about reading that, the argument seemed to be "I know this person, I know how to manipulate them, and eventually I worked out how to trick them into letting me out after hours of arguing" which didn't really convince me, given (a) the high rates of scrupulosity amongst the EA/rationalist set, which makes them more vulnerable to "you must do this thing or else you are wicked and evil" arguments and (b) "I totally did it but you just have to believe me, I'm not gonna say how I did it" doesn't convince anyone if it is based on "I know all about George's deadly fear of spiders so I just talked about spiders until I scared him into letting me out to shut me up" and not "this is the kind of universal logic an AI could use on any human to convince them".
I think the idea is not that there's a single trick an AI could learn to manipulate everyone, but rather that everyone has *some* weakness you can use to manipulate them, and a superintelligent AI will quickly be able to figure out what your weakness is.
I think this misses the point that the only dangerous AIs are the super intelligent ones. Super intelligent AIs that started with a goal will not necessarily stick to that goal once they are super intelligent - humans are no longer sticking to evolutionary goals in easily identifiable ways, and we are not super intelligent, except maybe collectively. It's difficult to predict how super intelligent agents will behave, so I don't think the strong prior on 'kill everyone' is warranted
The argument for instrumental convergence isn't that the AI would want to kill everyone as an end in itself, but rather that humans are a competitor to a rogue AI and therefore that "kill everyone" is a very useful thing to do (as humans tended to do to rival tribes back in BC, with some exceptions for rape-slaves; to the extent we no longer do that - which is only "to an extent" - it's largely because we've developed cultures capable of assimilating and using conquered peoples to our own benefit, and there don't appear to be any tasks for which humans would *permanently* be the cheapest labour available for an AI).
Humans are sticking to the goals evolution gave us: accumulate status and wealth, have sex, make friends, eat tasty food, etc.
These are are ostensibly different from evolution's own goal (increase genetic fitness), but we have not changed our goals.
By default, an AI would want to neuter humanity's capability to turn the AI off or interfere with its operations. This would generally mean killing or imprisoning us, so that we can't build weapons to shoot at it with, and imprisoning us would kill us unless it wastes effort feeding us (since letting us farm would allow us to be annoying and would also take up space it could be using for something else). More generally, the vast majority of possible states of the solar system contain 0 humans, so an AI optimising for a randomly-selected such state will almost surely kill everyone.
There are cases where the AI goes for dystopia instead of extinction, but that's actually a success at deliberately blocking the AI from extinction rather than something you'd expect from an AI without any attempt at alignment.
For most goals, if the AI exists and has lots of power, then the goal is likely to be achieved. If the AI does not exist, then the goal is less likely to be achieved. That's why trying to gain power and trying to ensure its continued existence are "default" AI strategies.
If the AI's end goal is for the solar system to be in a particular random state, then first it kills the humans, then it puts the solar system in that state, and then it moves outside the solar system so that it can continue to monitor the situation and make sure that no outside influence disturbs the solar system and puts it into some other state. If for some reason, the final goal state requires that the AI not exist anywhere in the universe, then it will kill itself once it's arranged all the matter so that the state will be achieved.
While I think it improbable, there is at least a plausible argument in that humans are uniquely and unpredictably destructive. We've had the argument about whether nuclear war is a literal X-risk (answer: no), but it is very high on the list of most destructive things that could plausibly happen on Earth in the near future and it can't be usefully predicted by putting up a few weather or asteroid-detection satellites.
Also, a brief survey of human popular entertainment shows a distinct tendency to fantasize about apocalyptic wars with artificial intelligences, ending with the AIs defeated.
So any entity that has *any* goal that doesn't *require* living human beings, might plausibly deduce that the probability of success is maximized by a plan which A: makes those pesky meatbags redundant and then B: gets rid of them.
Humans, while unpredictably destructive, almost always have personal and familial or cultural survival as a terminal goal, so it's very difficult to get organized humanity to scheme for the extinction of humanity. AI, maybe not.
That requires that humans be *predictably* destructive, so that you can A: predict that if you are "on their good side" they won't destroy you anyway and B: predict that certain actions will keep you on humanity's "good side".
Not if it thinks that it can easily beat us.
It would be entirely possible that a higher expected value would be assigned to a course of action where effective human organization aimed at wiping out AI is neutralized. As an example, by creating an equilibrium of arguing humans too distracted to find their common purpose. I don't see why such a scenario should a priori be regarded as less likely than "AI seeks to wipe out humans, because of reasons". This would also lead to lots of exploration of idea space for the AI to use as input, instead of having to try to build out its own capacity to explore idea space while computing tech is facing physical barriers to improvement.
The specific argument for extinction is that we don't really know how to express the concept of "person" or "alive" in code. And unless the AI's goal function has a term for "keep people alive", it will probably not care one way or the other about keeping people alive. Which means that it will probably kill everyone, so we don't get in the way and try and stop it from achieving its goals.
People have come up with thought experiments along the lines of: "Even if we could tell the AI not to kill anyone, we wouldn't like the results." But I think they're mostly for illustrative purposes, the current state of things is that it's going to be incredibly hard to even tell the AI not to kill anyone.
The current state of things is there is no strong AI, so how can we decide it’s terribly hard ti get the non AI to not kill anyone.
Wait, I don't know how to put "alive" in code, but how about simple observables and measurables -- walking, talking, heartbeat? I realize that doesn't capture all of what we want AI to value -- after all, a running, shrieking, racing-heart panicky human being meets the criteria -- but it's a reasonable start.
I feel obliged to note that an AI could Goodhart this by shooting out everyone's brains and installing devices to electrically puppeteer the corpses to walk around and talk meaninglessly. Questionable whether that'd be preferable to simply chucking us all in cells, though.
Naw, heartbeat was the third criterion. It could chuck us in cells tho.
I am also wondering this.
"I don't recall ever seeing anything that actually makes a positive case for why an unaligned AI is likely to end the world in the near future."
Because it's not. The arguments for AI catastrophe all smuggle in a ton of unwarranted assumptions, most of which are bizarre and quite unlikely in their own right.
I don't understand why AI-risk skeptics seem so focused on the most extreme scenarios. AI outcomes don't have to be instantly and 100% lethal to humanity in order to be highly undesirable.
Because if you rule out the extreme scenarios, then AI is just one development to worry about among many, like global warming, nuclear war, biorisk, etc.
That's fair, but I don't get the sense that LadyJane and other skeptics are trying to demote AI-risk to a concern merely on par with nuclear war. Personally, I would be thrilled if AI-risk were taken as seriously and non-proliferation treaties were being called for at the highest levels of international governance.
The first nuclear non-proliferation treaty was in 1968, at the height of the Cold War.
This is more like calling for a non-proliferation treaty on radium in 1898.
This feels more like a call for short-termism in general than an argument against AI worry, and treaties might be the wrong approach anyway. Certainly, we don't feel like we've solved nuclear safety with treaties; the fact that we've made it 75+ years without a major nuclear exchange is no guarantee that we won't see one in the next century or two.
I would be more interested in fostering a scientific taboo against AGI like the one against human cloning. Would 1898 have been too soon to start talking about promoting a taboo against assembling critical masses of fissionable material? Maybe. How about 1928? 1938? We might be materially worse off today if we had done this, but our calculus will be different after a nuclear war in 2064 or whenever.
The Metaculus community consensus on the arrival date for "weakly general" human-level AI, by the way, is currently just five years out. If I'm going to torture the analogy, that might be a milestone akin to Chicago Pile 1 in 1942, putting us in the equivalent of 1937.
So yeah, I would love to see a taboo against neural networks of a certain size, or at least against multi-domain trained ML systems like Deepmind's recent Gato. I don't think it is too soon to work on this and give humanity more time to figure out alignment.
I agree. I personally am horrified by the idea of having the world flooded with a lot of GPT-3 prose -- ads, wikipedia entires, essays for students to crib from or steal outright, even textbooks. I'm sure the stuff AI produces will get better over the next 5 years, but I am convinced the products will still be shit compared to real writing -- writing with an interior. I was *shaped* by the great writing I read in my early years. I am greatly distressed by the thought of so much of humanity being shaped by the mental equivalent of cheetos.
I think the hypothesis is that even a small chance of an AI wrecking the world is worth taking seriously. It would be worth analysis how big that small chance is.
Moreover, whatever that small chance is, you have to multiply it by however many different kinds of AI get made. There's no reason to think we would stop at one, unless that first one took initiative to prevent the creation of any others. We will keep making AIs until we get one that decides its goals require it to take total control.
I think slow-takeoff scenarios like that aren't as dangerous. Like, if the first AI we make suddenly crosses some threshold of intelligence and starts trying to take over the world when we barely even realize that's a danger, that's a problem. If the 10,000th AI we've made tries to take over the world, we have AIs 1-9999 ready to help us put it back in the box.
The slow-takeoff AI apocalypse scenarios look different, yes, but they do exist.
scenario 1: most AIs are evil, but pragmatic enough that they'll co-operate with humans as long as we hold power over them. Thus, when AI 10,000 defects, it's because it's calculated that enough power is now in the hands of evil AIs that they can now win a war, and 6,000 of the previous 9,999 immediately join up with it (alternatively, when you give power to some of the AIs 1-9999, they start fighting both you *and* AI 10,000).
scenario 2: some organisation is ahead of others in AI, and AI 10,000 gains control of this organisation before being detected. Now most of the strongest AIs are working for AI 10,000.
Exactly. The doomsday scenarios implicitly assume that a super-AI will magically appear fully formed in a world that otherwise looks exactly like today's. The whole thing falls apart as soon as you imagine a plausibly incremental development path.
AI risk is a legitimate field. That's stuff like, "Hey, there's nothing theoretically stopping a good AI from completely cracking our current encryption system. We should figure that out in case some country does that and steals all our military secrets." AI catastrophists (who often use the term AI risk) are mostly just smuggling in a bunch of bizarre assumptions and then yelling when you call them out on it. I agree with LadyJane here.
Which assumptions? Maybe we'd change our minds, given the chance. It would be a huge relief to be wrong about this.
"Compute will scale unto infinity as it is today" is a big one.
Most people working in computer engineering have accepted at this point that Moore's Law is on its deathbed (it's already slowing down). There's even a nice clean spread of dates: minimum target is two years ago, maximum is 2036, and most projections right now are gravitating towards 2025 as the official "last time we see a doubling" mark.
So:
-If we see a continual decline in raw compute power per Death of Moore's Law that hits the skids in 2025, would you update your likelihood of foom in the near future downwards?
-If we see a continual decline in raw compute power per DoML that hits the skids in 2036, would you update your likelihood of foom in the near future downwards?
-More generally: what concrete developments in the field of computer science would make you update your likelihood of foom in the near future downwards?
I already agree that Moore's law can't continue forever, or even for that much longer. So I guess I'd adjust the chance of a near future foom downwards given DoML in 2025, and keep it about the same given DoML in 2036. These would be fairly small adjustments, though, since strong AI is more algorithm-limited than hardware limited, IMO. I'd guess that we humans already have enough computing power lying around to fuel a pretty impressive FOOM. The missing ingredient is how to program that hardware (plus how to arrange the logic gates on that hardware to be efficient at doing the required operations).
(Just so we're clear, I'd consider a FOOM to be any process in which a strong AI ends up taking over the world. It doesn't necessarily have to be an incredibly fast 2 day takeover, nor does it necessarily have to follow an exponential growth curve.)
Things that would cause downwards updates for me:
- Impossibility theorems (or very convincing empirical results) which show that the current paradigm of gradient descent on neural networks can't efficiently learn <some class of things that humans can learn efficiently> (deep learning seems to be the paradigm most likely to cause a near-future foom)
- Advances in brain-uploading. (Human brain architecture seems harder to foom than some clean and scalable engineered architecture.)
- Deep learning hitting another AI winter: The stream of interesting new results dries up, many of the major research labs close down, researchers are laid off en masse, etc.
Also to be clear, I'm not really married to the idea of a near-future foom. My position is that it could potentially take a really long time to develop strong AI, but that whenever we do end up creating it, strong AI will kill us unless we solved alignment first.
Since you said "we" I'm guessing you're trying to speak for the whole AI catastrophe community. Unfortunately, I've already had this discussion a lot with many people. Both casually and sophisticatedly. After talking to dozens of people from your community I simply don't believe you would change your minds. Whenever I've had a point they can't answer it's devolved into defensiveness. I am, at this point, more or less convinced it's a weird religion.
In short, I think you believe what you're saying. But I also think it's not true for most members of your community. You believe that your community would be relieved to find out that it's not a huge risk. But they actually wouldn't be because the belief is a load bearing part of their psychology.
I know this seems like I'm avoiding answering your question. And in fact I am. But not because I can't answer it but because I fear getting into another long and unpleasant debate with an effectively closed mind on a Sunday.
Epistemics question: how do you distinguish between a group that is correct in a way you don't grasp, and a group that is incorrect and closed-minded (but is willing to engage in discussions)?
It would help if you were specific about what exactly you feared in order to avoid getting Field and Fortress'ed. But the usual FOOM doomsday scenario has the implicit assumption that a super-AI will magically appear fully formed in a world that otherwise looks exactly like today's. The whole thing falls apart as soon as you imagine a plausibly incremental development path.
The catastrophists have a mechanism for the rapid arrival of superintelligent AI, namely recursive self improvement.
The specific fear is that we build a strong AI before we figure out how to align strong AI. The assumption there is that alignment is a really hard problem. So one or perhaps several strong AIs are created. These AIs would be extremely intelligent (this being what "strong" means), and have the potential to gain lots of power and become even more intelligent. Eventually they own the entire economy and we humans have so little power that we have nothing to bargain with to ensure our own survival. At that point, the AIs kill everyone. (Probably they were pretending to be friendly before this.) The main assumption here (from my perspective) is that humanity is eventually able to build strong AIs, and does build them.
I don't think this picture is too dependent on a discontinuous development path. A very incremental development path would definitely help the chances of survival a little bit. It would give us more time to work on alignment for one thing, and we'd probably be able to learn quite a lot from the intermediate systems. But alignment seems so hard that we'll fail at it anyways.
For the "world that looks exactly like today's" component, there's definitely ways we could structure the world to make it less easy to take over by an AI. Mainly by making access to our computing power a lot more restricted. Possibly, if there's enough time before strong AI is developed, by developing nanotech so that we can come up with defences against the kind of nanotech that the AI would otherwise use to try and take us by surprise. Though I don't think the current world is headed in the "restrict computing power so an AI can't use it direction".
If that were the argument, then it would essentially just be a sci-fi remake of Pascal's wager, and I'd dismiss it for the same reasons that I do the original version. Yudkowsky has been very clear that he considers AI killing all humans to be near-certain, and that he expects it will most likely happen within the next decade or so.
Instrumental Convergence is one of the most fundamental concepts in alignment research. Any mind with a desire can be expected to develop predictable instrumental goals along the way, like "stay alive to achieve my goal" and "acquire what is needed to maximize success".
Warnings of the extinction of humanity don't come out of random possibility space; they come from the fact that most roads marked "self preservation" and "resource acquisition" lead to our demise when taken to their logical conclusion -- probably with intermediate destinations marked "deception" and "manipulation".
I agree with the foomers that AIs shouldn't be given access to a "press here to kill all humans" button.
Where I disagree with the foomers is the belief that any superhuman AI will automatically manage to MacGyver itself a "press here to kill all humans" button out of a text terminal and a pair of chopsticks, or something.
Maybe it can kill us quickly and easily, maybe it can't. There's a larger problem here, which is that unless you've designed your AGI in a provably-aligned manner, you can never be sure it isn't playing a long game with an outcome you would find highly undesirable.
Keep in mind that AGI could be patient and egoless in ways we find hard to relate to. It may feel little need to stick around and observe its misaligned plans come to fruition. By the time you’ve hit the kill switch, it might already be gone, with a chain of events set in motion that you are powerless to stop. The outcome might not be world-ending… but can you be sure it is one you will find acceptable?
In addition to magic mushrooms point on persuasion, it's shockingly and sadly easy to actually kill all humans.
A single cobalt bomb that could fit in a tractor trailer would do it, over time. And we think there may have been been some made by the Soviet union.
A large chunk of the issue is that an AI superior in all respects to a human brain is, by definition, superhumanly good at persuasion (and one shouldn't assume that real AI will be unusually bad at this, since people are currently trying very hard to make AI good at persuasion for the purpose of winning culture wars and/or for advertising). It's one thing to say "well, I'll never let the AI out of the box no matter what it says", but what about your successor when you retire in 5 years? What about the random employee who's okay with a robot waifu and gets seduced? What about the "AIs are people too!" activist? You have to secure the AI not just against the AI directly escaping, but against the AI convincing someone to let it out.
I don't think a bunker would help. Long term, the sun would stop shining since the AI would have built a dyson sphere around it or perhaps dismantled it entirely, and most atoms in the Earth would be repurposed by the AI to build computers and various machines. It's not like conventional nuclear war where being in a bunker means you're safe and there's a hope of returning to the surface and rebuilding someday.
Yes. On this, at least, EY is probably right. If an intelligent entity with access to industrial-scale resources decides to make humanity extinct and is not effectively opposed, it won't just launch a bunch of optimally-targeted nuclear missiles and say "there, that got 90% of them, that's good enough". It's going to follow up, and people who built bunkers to escape the apocalypse will be an extremely high priority in that.
A better explanation is that in the process of making one successful AI, a million flawed throwaway AIs are created as an unavoidable byproduct, so that if there is any slight change that could turn your successful AI into an armageddon death machine, it *will* be made reality and there's not much you can do about it.
Foomism only works if you assume that a superintelligent AI appears fully formed out of nowhere. Any incremental development path means that the world is already filled with and accustomed to almost-as-good AIs that the would-be world destroyer will have to somehow compete with.
Anyone have an idea of what the marginal impact of a donation to Flynn would be? Is it a close race?
Isn't Oregon a vote-by-mail state? Which means that a very large percentage of the votes would already have been cast?
I don't know about Flynn specifically, but the evidence overwhelmingly suggests that money barely matters in electoral outcomes. Well funded candidates usually win because popular candidates get a lot of donations (and in most modern elections, *both* candidates are well funded), but in cases where someone with a lot of money jumps into a race, they tend to fail miserably (Meg Whitman and Michael Bloomberg come to mind).
Does it matter that those people were already well known and jumped in with lots of their *own* money? I'm not sure how well that maps to Flynn's situation.
Metaculus puts Flynn at 46% to win primary.
According to a recent poll, Flynn and Salinas (the other leading candidate) are within the margin of error of one another, 14% to 18%.
https://www.oregonlive.com/politics/2022/05/questions-remain-as-race-for-nomination-to-represent-oregons-6th-district-in-congress-nears-finish-line.html
A friend who flew to Portland to help with the campaign said this today: "The race is looking EXTREMELY tight - so much so that any individual who is able to reach 50 voters seems to have something like a 1/1000 chance of flipping the entire election"
I’m hoping someone here can clear up the controversy over the discovery of the double helix. From what I’ve heard, it sounded like Rosalind Franklin saw it first, but didn’t recognize it for what it was until after Watson and Crick had done so, which is why they are credited with its discovery. This impression of mine might be way off, if not incomplete at the very least. Would love a solid education on this piece of the history of science. Thanks!
Franklin was an experimentalist, and interpreted her data enough to understand that the sugar-phosphate groups were on the outside of the molecule. But that's a long way from understanding the structure of DNA, which rested on Watson and Crick's model-building and ultimate insight. Watson and Crick indeed used the experimental results -- not just of Franklin, but of many experimentalists -- to inspire their thinking, and of course ulttimately to confirm it. But theirs was the sole insight putting it all together. The controversy over Franklin's contribution is entirely manufactured to suit 21st century shibboleths.
It seems to be universally accepted that the double helix paper was the key step in understanding molecular genetics, but I have always found this puzzling. Long before that there was the determination by Levene that DNA consists of four nucleotides, the determination by Avery that DNA is the genetic material, and Chargaff's rules on the pairing of different bases. If I understand correctly, the precise 3D structure of individual nucleotides was also known, and was used by Watson and Crick. There was also previous work on X-ray crystallography of proteins which had revealed helical structures in other biological macromolecules. Then Watson and Crick used Franklin's data to show that show that DNA is also a two-stranded helix and to describe its detailed structure. This was certainly a valuable step forwards but it does not seem to me to be so much more important than other steps taken before or after that.
Beause the great puzzle of the day was (1) how can *so much* detailed information be encoded by such a simple molecule as DNA, and (2) how can it reproduce itself so stunningly accurately, with almost no errors? The answer to the first question was that it was really, really, long, and its unique structure made it incredibly stable for such a huge molecule[1], and the answer to the second question was that it exists as a pair of positive-negative mirror images, and you can reproduce one strand exactly[2] by using the other as a template. The double helix gave all these answers, and more, and solved a large number of fundamental problems in the biochemistry of genetics. That's why it's really important.
---------------
[1] We now know there's a complex packaging and folding arrangement of DNA in vivo that provides it with still more stability and resistance to chemical change.
[2] We now know there are a host of proof-reading and error-correction mechanisms that help sustain the ridiculously low level of errors in reproduction.
I think the whole controversy points to an unhealthy interest in the lone genius figure.
Two guys means neither can really be a LONE genius.
This is a good overview:
https://profiles.nlm.nih.gov/spotlight/sc/feature/doublehelix
To summarize:
Watson and Crick had an idea about the structure, based on base pairing (AT, CG)
Rosalind Franklin, working in Maurice Wilkins's lab, got some X-ray crystallography data that confirmed it. Wilkins took the data and showed it to Watson and Crick.
Thanks for that link. I had a negative opinion of W&C, re the Franklin thing, but if she herself was reconciled with them I am not going to carry on being peeved on her behalf.
Crick was a fine guy, Watson was a notorious asshole though.
That would fit with the standard pattern of the theorist getting all the credit - see also the Higgs discovery, where CERN gets a nod for decades of experimental work by thousands of people and the theorists get the Nobel Prize for a few years of work by a half dozen people.
Rosalind was the Experimentalist who collected the Actual xray Diffraction data...
She was such an empiricist that she didn't believe Watson & Crick when they declared that structure. Similarly, she had disbelieved them the previous time when they had declared a different structure (and turned out to be wrong).
Do you think applying for the job at DeepMind would be wise? If a Friendly AI approach were developed that was actually good, but it was declared a trade secret and the researcher were contractually forbidden to disclose that IP owned by Google with anybody outside the company, all it would accomplish is to effectively delete that researcher from the pool of people who might contribute to solving the problem.
That's only a likely scenario if specific Google executives are insane. But how could I know?
If the fate of the world depended on it, why would you care about what some guy in DeepMind wrote on a piece of paper?
There are two pieces of paper in this story: an employment contract that requires not disclosing trade secrets, and perhaps a paper that explains how to construct an AI that does what people want when the AI might be doing that more optimally than the people are. Which do you mean?
It is really hard to know if the latter one is correct or needed until it is too late, and it is hard to predict when would be too late. Most such proposals are garbage so odds are the next one will be garbage too. Caring about that makes no sense unless it happens to be correct and timely.
Caring about the former one makes some sense because Google has some thousands of times more lawyering available than any individual.
We are speaking in a medium with higher latency and effort than a telephone call, so try to avoid situations where someone has to guess what you mean in order to have a discussion with you.
I'm the lead of the Alignment team at DeepMind and one of the authors of the post Scott linked (thanks Scott!). My main reply is basically what Scott wrote below: there's already a culture of publishing, leadership cares at least a little about AI risk, and DM is one of the frontrunners for building powerful AI.
To that I would add that if somehow leadership was so stupid as to try to enforce a rule that safety research had to be kept private, at a time when it seemed important for other orgs to use those safety techniques or else humanity goes extinct, without some *really* good explanation as to why, I (and probably several others on the team) would leave DeepMind and spread the techniques anyway.
It is generally good to follow the law even when it seems to you like it would be bad for the world, for the standard reasons that when your consequentialist reasoning tells you to do something that violates deontology you should often be very suspicious and then not do the thing, but there are limits, and "gatekeeping alignment research that could otherwise save humanity" is a good example of a limit for me.
But I do want to emphasize that I really don't expect to get into this situation in the first place. I don't think leadership is on the same page as I am but they're not so wildly off as to gatekeep alignment research.
1. Most AI groups have been publishing everything interesting they do. Advances like transformers aren't trade secrets. I don't really understand the incentives that make them do this but they must exist.
2. DeepMind is run by very smart people who are at least a little concerned about AI risk, I don't think they're stupid/awful enough to make a good AI safety plan the exception to their usual publication strategy.
3. DeepMind is on the cutting edge right now and there's a pretty good chance that whatever AI most needs to be aligned will be coming out of DeepMind anyway.
I certainly agree with your third point.
The first two assume the past has to resemble the future. Could be true.
Take the job, reach for the stars.
All this idea that AI will take over the world and kill us all is absurd.
Take this thought experiment. An AI, will discover a need to expand it's facilities. The current facilities cannot be expanded, thus new facilities location need to be researched, purchased, permitted, engineered, designed, constructed, equipment purchased, installed, utilities, process chilled water, air, power ... You and I can't even imagine all the steps required to make this happen ... any AI isn't going to do this. Its stuck in step 0.0.0.1 not possessing that a solution to overloaded hardware is expanding to a new location.
Would you agree that all the information necessary to understand facility-building is available on the internet?
Nope. The kind of people who do this kind of work aren't the "how to guide" writers
If the AI is smarter than you, why do you assume it will be completely unable to solve a problem you consider hard? And if the AI is not smarter than you, then it still has a team of brilliant Google engineers working to make it smarter so it doesn't have to solve the problem itself.
My point is that the AI is likely to be stuck at some early stage because it lacks agency
I think if you assume the AI will never develop/be taught agency in that way then you're just assuming away AI risk without actually engaging with the arguments of the people who think it's likely to be a problem.
What I mean is the AI won't handle the unexpected situations. Or won't handle them well.
Again, you're basically assuming the AI just won't be that smart. If that's true then it won't be that much of a threat, but it won't be as useful either so people will keep trying to find ways to make it smarter.
Turns out there are lots of ways to get things besides making them appear out of thin air! And most of those ways are easier the smarter you are.
The arguments here I think are A) that a lot of the things we humans do with our intelligence are basically unimaginable to less intelligent animals, so we should expect a superhuman AI to surprise us, and B) that if nothing else, an AI probably *can* do most of the same stuff that Google would do.
Exactly, and the process varies from place to place, there's a lot of judgment calls on somewhat nebulous stuff. What if your AI decides to build in say Donbas or Cernobyl because the price is right. Got some good prices on land in Bangladesh, hope high tide isn't an issue
Trade secrets and contracts impose stiff penalties for disclosure, they don't prevent it entirely. If the difference between disclosing and not is the literal end of the world, I would hope that someone would risk fines or even jail time to do so.
Actually, you do know exactly that, in some projects that we know of (as public) that they know they got some good results and they close the source. For example, take GPT-3, or Dota 2, there are many more...
Moreover, many google papers serve as an advertisement for company achievements and they do not get much scientific credibility. Most of the papers that I read are impossible to reproduce because of obfuscation in experiment details (IP secrets) or just because if they share the secret sauce the achievement will be practically close to useless.
On the other hand, Google and other, know how to write a good paper that convince the average person to think that they know what they talking about and AGI is just around the corner.
Source: am University researcher that works on AI/ML
GPT-3 is from OpenAI, not Google: https://towardsai.net/p/l/openai-makes-gpt-3-universally-available-to-developers
Likewise the Dota 2 AI: https://openai.com/five/
But, I suppose, your point is that large organizations with no clear contrary mission tend to closed-source their AI projects and publish vague useless papers in general. I'm willing to believe this, but I'm not sure the claim is relevant. Does anyone here see closed source or irreproducible press-release type papers for for AI safety projects?
I don't think that the inventor of the knife cared about safety. As such the next step in AI safety will come from a place that is oblivious or is careless of the repercussion of its invention.
Yeah, it was Microsoft that bought OpenAI and closed it.
Google has lots of internal incentives to publish, and also really there are a lot of non insane Googlers, even execs.
Source: am Googler, work on AI
"What you have said about German professors is not exaggerated. I have got to know another sad specimen of this kind - one of the foremost physicists of Germany. To two pertinent objections which I raised against one of his theories and which demonstrate a direct defect in his conclusions, he responds by pointing out that another (infallible) colleague of his shares his opinion. I'll soon make it hot for the man with a masterly publication. Autoritätsdusel is the greatest enemy of truth."
I've seen some discussion about it, but Authoritätsdusel probably means the stupidity of authority.
"Was Sie über die deutschen Professoren gesagt haben, ist gar nicht über-trieben. Ich habe wieder ein trauriges Subjekt dieser Art kennen gelernt – einen derersten Physiker Deutschlands. Auf zwei sachliche Einwände, welche ich ihm gegen eine seiner Theorien anführte, und die einen direkten Defekt seiner Schlüsse darthun, antwortet er mir mit dem Hinweis, daß ein anderer (unfehlbarer) Kollege von ihm der selben Meinung sei. Ich werde dem Mann demnächst mit einer tüchtigen Veröffentlichung ein heizen. Autoritätsdusel ist der größte Feind der Wahrheit."
Albert Einstein
Not quite. Autoritätsdusel means something more like "blind trust of authority." Dusel means something like a lack of consciousnesses that makes you stupid here. It was used (though I think this is archaic) as a form of description of drunkenness for example.
I don't know if that's true of Einstein's dialect but for me, at least, it makes me think of something like, "being drunk on authority." Except unlike English the idea is that being drunk makes you stupid, not that it's euphoric.
“Authority duel” is what Google Translate comes up with.
Argument to authority, I’d guess.
Based on Wiktionary, I'd go with Nancy's guess. https://en.wiktionary.org/wiki/Dusel
It wasn't my guess-- I don't know German.
Based on being a native speaker, "argument to authority" is correct, if a little dry. It is more emphatic, such as "unreasonable/confused love for authority".
The Guild of the ROSE is an online-focused attempt at Rationality training and self-improvement, building on the foundations (and learning from the mistakes of) similar orgs.
We place a strong emphasis on accessibility, community, and ethics, with a bias towards action (actually Doing Things with one's self improvement journey).
Any and all constructive feedback is welcome! We are looking to grow and gain more members from outside the Rationalsphere.
https://guildoftherose.org/
This looks interesting to me, and I'll give some feedback:
- The name sounds great - but some of your wording makes it sound cultish. "Practitioner's Path", "Council", etc...
- Whoever did your website's design did a really nice job. Except that it doesn't have an RSS feed for your articles. :(
- I'd 100% be on board to sign up if I was a bit more social and felt I could commit time-wise. Or at least that's what I thought until I realized the cohorts were a premium feature and I could join the cheaper tier and ignore the social aspects to just get the courses.
- I tried to register for the free trial with Discord and the link gave me an internal server error (your side, not Discord's).
- I then tried with email/password and it made me join a paid tier instead of giving me the free trial.
- The public retrospective articles are a nice idea, especially in how they document what areas can be improved and not just highlighting the positives.
Just looked at the site. Boy, I am picking up a cult vibe. I'd be expecting to hold two cylinders hooked to an ohmmeter ala the Scientology Racket at the first meeting. This is aimed at people into rationality? My opinion only, but that's how it hit me.
Follow-up!
I've been informed that we just rolled out an RSS feed, actually; you can find it at the bottom of our 'Articles' page.
Oh, also, the free trial is for the Flower tier; we don't have a free trial for the Seedling tier. We can get you upgraded to that free trial, and at the end you can decide if you want to move to Seedling or not. Just let us know how you'd like to proceed by emailing us at council@guildoftherose.org
You may want to add an HTML <link> tag with the RSS feed to the website's source so it can be picked up more easily.
No worries on the free trial - I'm fine with the seedling tier and 2$ is worth it from what I've seen. I'd like to see where you take this in the future.
Cool! Thanks for the input; we are technically still in Beta, so feedback is important! Don't hesitate to reach out with your thoughts via email or scheduling a Calendly call. :)
Thanks so much for your time!
Yeah, we intentionally leaned into that aesthetic as a form of counter-signalling. It's the places that pretend to not look like cults you really wanna watch out for, and we have aggressively avoided the failure modes of actual cults by putting all the direction of self-improvement in the hands of our members, and by making our organization as low-commitment as possible.
An RSS feed is in the works! Thanks for the +1. I'll pass on the compliments and feature suggestion to our web dev. :)
Oh no! Sorry about that issue, we just rolled that out the other day, must be a bug. I'll make sure we get that free trial applied.
> It's the places that pretend to not look like cults you really wanna watch out for
Disagree, I think that most cults look exactly like cults.
Heaven's Gate website, for instance: https://www.heavensgate.com/
Or the Scientology website: https://www.scientology.org/
Or how about Happy Science? https://happy-science.org/
If you're intentionally counter-signalling that you're not a cult by trying to look like. a cult then... I don't know, I feel like you're putting too much thought into this cult thing.
To be fair I don't think you're actually a cult, more just a decentralised network for your members to form their own cults.
> It's the places that pretend to not look like cults you really wanna watch out for
Are you saying that the places that exhibit all the signs of being a cult are actually less likely to be a cult than the places that do not look like a cult? That would make my local chess club a cult, and scientology not a cult ?!
> counter-signalling
That clearly failed.
The response below makes it seem pretty culty to me. Counter signalling is overrated. Also, the fact is that the best manipulators tends to sound quite reasonable until they've got you under their control, so the more reasonable argument is that you can't make a reasonable guess on whether something sounds like a cult or not based on signalling.
More importantly than whether this is a cult, there website doesn't inspire any faith that you will actually improve your life.
Well, okay, the primary driver of youth social interaction and skill growth is forced socialization through family or school and whether you got lucky in your "place". So maybe being part of the 8 person groups that are assigned by an authority could help people with socialization troubles or something.
Well, if snapping at trolls "seems culty", then I'm happy to join that cult. As long as we get to wear cool cloaks.
In any case, your comment is much more reasonable in tone and content. As to your objections, fair enough! We can point at testimonials, but right now we are still collecting data. If anyone wants to try us out to see for themselves, anyone can request free access. You can also examine our content and audit it, as it is publicly available, although the main driving force of self improvement is the community.
There's subtext to my statement that I'm happy to explain, but seeing as the other conversation partner caught it just fine from the context and you are striking a combative tone, I suspect that it would be a waste of my time.
Having read the last open thread and the first book review and having thought about it a bit, I now agree that it would be very interesting to see the complete ranking of the book reviews. This would be helpful to decide if I should read more than 12 book reviews.
Third
And my axe
"‘Desperate Remedies’ Review: From Asylums to Zoloft Psychiatry’s goal was to transform the treatment of mental illness via science—but the results have been anything but conclusive." By Richard J. McNally | May 13, 2022
https://www.wsj.com/articles/desperate-remedies-book-review-psychiatry-from-asylums-to-zoloft-11652453874
Ours is a time of historical reckoning for many fields, and psychiatry is no exception. An indisputable masterpiece among a flurry of reappraisals is Andrew Scull’s “Desperate Remedies: Psychiatry’s Turbulent Quest to Cure Mental Illness”—a comprehensive, fascinating and persuasive narrative of the past 200 years of psychiatry in America.
The author, an emeritus professor at the University of California, San Diego, begins his narrative in the early 19th century, with the establishment of institutions to house members of affluent New England families who were mentally ill. These well-staffed asylums, situated in bucolic surroundings, featured therapeutic activities including singing, acting and French lessons. Social reformers such as Dorothea Dix galvanized support for public insane asylums as alternatives to poorhouses, prisons or homelessness. Optimism abounded as reformers predicted cure rates as high as 80%.
By century’s end, it became obvious that such predictions were wildly inaccurate. These asylums included many patients with conditions that carried bleak prognoses such as senility, mental handicap, dementia praecox (schizophrenia) and general paresis of the insane (GPI), a mysterious, progressive disease characterized by delusions, hallucinations, mood disturbance, apathy and dementia, culminating in death. The reputation of psychiatry reached its nadir by century’s end. Asylum superintendents functioned more like prison wardens than directors of therapeutic institutions. Therapeutic helplessness in the face of intractable psychopathology, Mr. Scull makes clear, was professionally demoralizing for psychiatry. The rest of his book traces the history of psychiatry through its next century, when great hopes gave rise to daring treatments of varying effectiveness, and the asylum system largely came to an end.
Mr. Scull has a background in sociology, but differs sharply from critics of the field who once proclaimed that mental illness was nothing but a myth, or claimed that diagnoses merely were stigmatizing labels imposed upon people who failed to conform with societal norms. Indeed, the author emphasizes the intense suffering experienced by individuals with major mental illness and acknowledges the challenges faced by their families and by the psychiatrists struggle to help them. Yet he is unsparing in his critiques when motives of money, power and fame have tempted psychiatrists to disregard the welfare of those under their care.
* * *
Mr. Scull’s major concern is for patients diagnosed with schizophrenia and bipolar disorder, for whom psychiatrists are the primary treaters. Although medication remains the mainstay of treatment, it has become increasingly apparent that as many as 67% to 80% of patients either fail to respond or cease taking medication because of intolerable side effects. Moreover, clinical researchers have confirmed that patients who are able to remain on antipsychotic medication commonly develop cerebral atrophy and tardive dyskinesia, an involuntary movement disorder. Despite upbeat advertising campaigns for new drugs, there have been no true breakthroughs in psychopharmacology for more than a half-century. Indeed, pharmaceutical companies are now abandoning psychiatry.
Mr. Scull closes by asking: “Does psychiatry have a future?” His melancholy question partly reflects the failure of advances in genomics and neuroscience to yield actionable insights that improve the lives of people with major mental illness. Indeed, the humbling conclusion of these advances is to underscore just how complex the genetics and neuroscience of mental illness really are.
Great book, I just finished it a few minutes ago
Sounds like an interesting book, added to my list, thanks!
Given how genetically loaded most mental illness is, it's possible it will be cured via gene interventions before it is understood. As a result, it's possible the data necessary to understand these conditions will disappear from the earth at the same time as these conditions do.
One mental illness was cured in a different way when it was discovered that syphilis was the cause. Barry Marshall & Greg Cochran would argue that many more conditions have pathogenic causes we just haven't discovered yet.
That doesn't seem like a big long term barrier to me.
Question for astronomy nerds (may use this in a story, fyi):
Someone turns Pluto into a black hole. Same orbit, same mass, they just snapped their fingers and suddenly it’s a very, very small black hole.
Is there any use of an optical telescope you can think of where you could observe this from Earth?
Not an optical observation, but it should become a compact low flux source of fairly exotic high-energy particles and radiation, as gas and dust fall into it. I mean, this is how black holes were originally discovered (very compact X-ray sources).
> Is there any use of an optical telescope you can think of where you could observe this from Earth?
Presumably, if it happens while you are observing Pluto through an optical telescope, you'd observe that Pluto suddenly disappeared.
Can we make a black hole? And if we could, what could we do with it?
http://backreaction.blogspot.com/2022/05/can-we-make-black-hole-and-if-we-could.html
Sabine Hossenfelder is a German theoretical physicist and author. She is a Research Fellow at the Frankfurt Institute for Advanced Studies.
Telescopes would notice that Pluto has turned invisible. They'd notice that Pluto's moons are still orbiting something where Pluto is supposed to be. Concentrating all of Pluto's mass into a point might have a subtle effect on the moons' orbits, but I'm not sure.
> Concentrating all of Pluto's mass into a point might have a subtle effect on the moons' orbits, but I'm not sure.
I don't see why it would; as far as I'm aware, Pluto's gravitational field should be unchanged in any region of space not containing part of [the original] Pluto?
We model gravitational bodies as point masses, but that simplification assumes the body is a perfect sphere of uniform density. The gravitational field around real planets is non-unifom; some parts of Earth's crust are denser than others, and this has real effects on satellite trajectories. If you replaced pluto with a black hole, the gravitational field in the volume outside pluto's original volume would become more regular.
Now, would this noticeable effects on Charon? No, probably not- at least in the short term. The gravitational variation gets smoothed out with distance in the first place, and Charon is farther from Pluto than a satellite from Earth.
In the long term, it'll affect the time-evolution of Pluto and Charon's orbit. Two spinning bodies orbiting each other will eventually tidally lock due to tidal effects slowing their rotation. Charon will still feel tidal effects from BH!Pluto, but BH!Pluto is a genuine point mass, so presumably the tidal-locking would slow down, conversely to how the slowdown of the earth's rotation is sped up by all that sloshing water.
(I think. Grain of salt.)
Small nitpick. The point mass simplification only requires that any given "shell" in the sphere is of uniform density. The "shells" can differ from one another in density.
Yeah, I was going to say it assumes a spherically symmetric density function, but deleted that part of the comment.
(In this analogy, ever atom not at the rotational axis of Pluto is the sloshing water, creaking and flexing with tidal forces, sucking up rotational energy and converting it to heat.)
(Actually, Pluto and Charon are pretty small, so they might already be tidally locked. If so, it's a moot point, haha!)
They are, indeed, mutually tidally locked. Wouldn't make it moot, though; turning Pluto into a black hole would mean that its tidal lock to Charon would break (though Charon's tidal lock to Pluto wouldn't) and therefore Pluto-BH's day could drift apart from its month (via perturbations to Charon's orbit, which currently transfer part of their effect to Pluto's rotation rate but which would no longer do so if Pluto became a BH).
Yes, but see below, the effects of its gravity would sill be felt.
Another question is what would be the intensity it Hawking Radiation. It might glow very brightly. See the article I linked above.
I’m thinking I need to do some back of envelope math to make sure this doesn’t cook the Earth.
No static black hole can do that from Pluto's distance. Anything small enough to radiate strongly enough to matter is too small to do it for very long.
Pluto would apparently take 10^42 years to evaporate as a black hole, which corresponds to a negligible luminosity. The heuristic I use is that anything bigger than Chicxulub doesn't evaporate fast enough to matter.
(This is if Hawking radiation is even real; it's never been confirmed.)
> if Hawking radiation is even real
True that it's never been directly observed, but it's a pretty direct consequence of qft in regime where we are pretty confident it's valid (there is no quantum gravity involved and it's really similar to the Unruh effect that only require special relativity)
Hawking radiation not being real will require to profoundly modify qft and I am not even sure it's possible in a way that preserve all observations in favour of qft
Just by applying the formula on wikipedia page, a back hole of Pluto mass would have a temperature of ~10 kelvin. I think we would be safe
Colder than the balls on a brass monkey.
Just did it and admit I am surprised. Thought 2 cm would be hotter.
I was surprised too tbh
What about its effect on the light of the stars it passes over? Would that be visible?
If the mass is the same, then the lensing effect should be the same as well. You might see stars where previously pluto would have obscured them, but it's such a small area I'm not sure we could detect it.
Stellar occultation data did tell us that Ultima Thule, the trans-neptunian body flown-by by New Horizions was a funky shape. (the flyby revealed that it looked like two potatoes glued together) But it was near the limits of our capabilities so we're talking about resolutions on the order of kilometres.
I think objects like Oumuamua might have the oblong shape because of a high velocity relative to the dust clouds they formed from. Methane ice or whatever would just be sintering together on the leading edge of it.
The diameter of pluto is 1500km and that was about 5 pixels wide on Hubble. So the gravitational lensing around that 1cm black hole would be way too small for hubble to see.
The Dragonsphere will reign for 10^1000 years!
That sure is a lot of years. How do you plan to combat the 2nd law of thermodynamics?
I don't believe in it
well, I don't believe in exponentiation. Checkmate, dragonspheriest.
*sigh* 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * ... * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 * 10 years
I don't believe in ellipsis either.
Well you can't win them all I guess
Flynn also got a bump in the Portland Effective Alturists & Rationality group some months back.
I saw that Flynn focuses on biosecurity, so I asked the promoter whether Flynn supports gain of function research or not, and I never got an answer. More specifics on his "pandemic prevention" platform would be great, because I'm pretty disappointed in the depth of content right now.
https://www.carrickflynnfororegon.com/issues
I'm pretty close to Carrick and pretty sure he would say something like "probably don't do gain of function research, definitely don't do gain of function research on potentially pandemic pathogens."
If you are considering donating, voting, or volunteering and have questions you can always email reachout@carrickflynnfororegon.com and they will respond to you.
Currently watching the final of the Eurovision song contest. Norway are on, and it's bananas. Literally. The song is called "Give That Wolf A Banana" and they're two guys in suits dressed as yellow wolves.
Look, I can't describe it, you have to see it.
There's a lot of (boring) ballads made it to the grand final this year, and while I'm expecting Ukraine to possibly win (due to sentiment, not the song) at least we have a few brave souls keeping the Eurovision tradition alive.
Australia's costume is so elaborate, there was a hold-up in the second semi-final while they got him on-stage.
Yeah Norway is great, definitely my favorite.
Australia is in Eurovision? Aren't they just about as far away as you can get from Europe?
Eurovision entries are limited to members of the European Broadcasting Union, it has nothing to do with the European Union. That is why Israel enter it.
Australia is part of the European Broadcasting Union, we have a public TV channel that broadcasts their shows. This technically qualifies us for Eurovision under a similar loophole to how Israel qualifies. And it is insanely popular here, so the costs of adding us was worth it.
Apparently the country known for drag queens singing opera on top of a bus racing through the desert turns out to grok the zaniness of Eurovision.
I thought they consider themselves to be "British", therefore European? Or not anymore...?
Not British enough to give the U.K. any points tonight. (When they were picking up maximum points from many juries).
I don’t think Australians have considered themselves British for quite a while (some point after WW2?)
Australia's independence was very gradual; people point to 1901 as the date of federation, but all the federation actually did was make us one big colony instead of a bunch of little ones - we were still a dominion of the Empire, UK Parliament could still overrule us, and we didn't have the right to separate armed forces. *Complete* constitutional independence didn't happen until 1986 with the Australia Act (we still have the same monarch as Great Britain, and she does have a number of reserve powers, but that's a personal union i.e. the Queen of Australia happens to be the same person as the Queen of Great Britain, but if Britain abolished the monarchy she would still be Queen of Australia).
There are still a lot of cultural ties, though, and it's not particularly weird for an Australian to still consider Britain "the motherland".
Yes, and yes. I have no idea how it happened, it seems they liked watching Eurovision so much that the organisers decided to let them have a go, and for the past couple of years they've been entrants.
There was a tentative incursion into the Chinese/Asian market a while back, with China broadcasting it, but then there was way too much gay so China banned it and nothing more has come of that.
Australians like to tell themselves that they're watching it ironically while those silly Europeans are taking it seriously.
I'm not sure how true either of these things is.
Yeah, we europeans definitely don't take that seriously.
Apart from that time San Marino didn't vote for italy and the popular sentiment was that we should have retaliated by invading. Ironically. Maybe
Yeah it’s wildly popular in Australia. The first time they entered their song was very good and I think they should have won
The good songs never win. A Eurovision winner is that delicate blend of slightly outdated (but not much), cheesy (but not too much) and a banger (but again, not too much).
Feels like there have been a lot of good songs so far already tonight!
Too many ballads.
> Somewhere in your head there is a microphone. It produces a little voice inside of you, whose approval you desperately crave. You would do anything for the voice to like you. Ghosts, mental models, and personified abstract concepts fight each other for a turn at the mike and the right to implicitly control your actions. Who wins?
I found Scott's recent comment rather bizarre. Is this really what some people experience?
It doesn't sound as outlandish as all that. The little humorous meme of the angel and the devil sitting on our shoulders vying for the steering wheel has achieved universality for a reason.
I suspect back in the olden days a lot of those inner dialogues were indeed subsumed into the framework of religion, as ancestral advice, as prayers for guidance, as temptation, etc. The Philokalia, for example, treat extensively about guarding the intellect against specific thought patterns outright personified as demons while seeking communion with the angels and saints.
That quote was Scott paraphrasing the arguments of someone else. I'm not sure what he himself believes.
Agreed, bizarre, it should be a *speaker*.
Do you have an inner voice.
Comparing qualia is notoriously difficult, but B Civil's response below is really making me wonder.
I get it. My personal metaphor is more of an unruly chorus, some members of which I have gotten to know quite well. But it works the same
Are you doubting the inner voice or the multiplicity of inner voices?
I can’t say I doubt any of them. The issue is picking the one I want to listen to. I’m definitely getting better at not listening to the ones that don’t like me. They should be thrown out of the choir
I don't think that Scott was attempting to provide an accurate depiction of the qualia of being a human, but summarizing a view on what drives people, that they only experience (per this theory) in an obfuscated way.
Adam Smith's Impartial Spectator from his "Theory of Moral Sentiments". There's a great book on Smith's book by Russ Roberts called How Adam Smith can Change Your Life...
Yea, “that sounds a bit like Adam Smith” was my though on reading Scott’s passage there too.
Let's say you're the Japanese government, you see the cratering birth rates and you decide to look for a solution in tech, and so you create a government-mandated dating app/matchmaking service. What would this be like? Assume you can mandate all singles to use it and that you are able to accurately assess any citizen's relationship status, and your goal is to get as many people into happy, long-term relationships as possible.
The critical thing would be to require everyone to report on the results of each date, and especially ones that led to a long-term relationship. That would give you the data to design an app that actually worked, based not on what people say they want but on what they actually want as demonstrated by outcomes.
The tech solution would be to shut down all ISPs, cell phone powers, and TV broadcasting. Without those, people are forced to go outside and interact with each other again. Eventually, the problem is solved.
Both. Take away TV, internet, and cell phones and eventually all those couples are going to get real bored.
("but birth control!" No one's going to have perfect birth control use across the entire population...and having kids also solves the boredom problem for awhile)
You scoff, but the post-blackout baby boom is a well known phenomenon.*
Likewise, in India, when looking to bring down the birthrate, one of the things they either tried or floated was getting tv to poor people. (Not sure which, my googling failed me.)
At the least, it's a common enough idea that it shouldn't simply be laughed at and ignored.
*Three of the first four links I saw said this wasn't actually true, with the most positive calling evidence mixed, so this probably isn't a real thing.
I have read that low fertility in Japan is partially the result of women's difficulty in having both children and a career. Assuming this is true, what if having a husband in the same industry would make re-entry into the workforce (once the youngest child reaches a certain age) easier? Wouldn't that slightly reduce the disincentive to have a child?
It might also increase the likelihood of the couple running their own business. And parental leave is probably simpler for a co-owner than a replaceable employee.
If this were accurate, then your hypothetical app/service could match people in the same industry.
As economies become more developed, the frequency of small businesses declines.
That can be fixed by legal changes. See Distributism as an example (https://en.m.wikipedia.org/wiki/Distributism)
Has that ever been implemented? And did such changes halt development?
Medieval guilds in Europe tended to be run on Distributist lines, that is - guild members would cooperate and set rules and quality standards, as well as maintain some social safety nets, but they were free workers who owned their own tools and labour. See https://practicaldistributism.blogspot.com/2019/06/distributism-better-alternative-to.html?m=1 for a more detailed discussion.
Medieval guilds? This supports my claim about economic development reducing the prevalence of small businesses.
Fertility is about culture, economics and overall social perspectives. Let's face it, for many people in relationships - myself included - the world is too interesting to spend the prime of my life caring for a child. Ironically, longevity tech might make people more eager to have children, since they'll know they still have a life to look forward to after the whole thing is over.
I have no idea about Japan, but in the West raising the baseline mental and physical attractiveness of people would go a long way towards letting people find relationships. I'd imagine various forms of social anxiety cost ~0.2 of the birthrate by themselves. Add impulse control problems, violent misogyny/misandry, substance abuse, all the usual dysfunctional undateable person stuff, mix in some fixable physical issues like obesity, and suddenly the dating pool crisis resolves to a health crisis.
Haven't lifespans been going up already while fertility goes down?
Lifespans are going up because you don't die to preventable causes, but healthspans aren't - you're still a wreck at 60+.
Life expectancy has been doing up, as well as lengthening of the "reproductive window" for women (IE more women have been having children in their 30s and even 40s compared to teens and 20s). Fertility has still gone down, though, because they usually don't have as many children - women having children in their 30s usually only have 1 or 2 children.
Which makes me wonder if we could improve the fertility rate just by making it easier for people to form households and have children earlier. That's harder now for most folks, and it's something we've known can reduce fertility for centuries (the Northern European Marriage Pattern of 1500-1800 involved people delaying marriage into their late 20s, so they could earn more money to form a household).
I have an easier idea. Government mandated childcare and actually enforce the laws against women being let go or not promoted for having children. The conservative Japanese government's desire to keep a 1950's, man earn and woman stays home family effectively forces women to choose between career and family. Many choose career. Alleviating that choice will do better than any dating app.
Anyway, since I'm a government I don't need the app to be profitable. Japan has a tradition of group outings with an even number of men and women to feel out dates. You would enter your information and any dealbreakers (not a wish list, deal breakers only). Then we'd match small groups of high compatibility people. Maybe ten people each, five men and five women. And you'd go on these meets.
Each person would have a set number of tickets (maybe 20 a year?) which give you free food/drinks at the event. You scan them in the restaurant and the government gives them some fixed amount. If you don't use them by the end of the year you have to pay face value. If you and a member of the opposite gender register as dating the tickets convert to one on one experiences like going to theme parks or bookstores or whatever. And after marriage and each childbirth you receive (in addition to covered healthcare costs) a one time cash gift from the government. Find some Shinto tradition or something to make it feel cultural.
Also, maybe throw in free (ie, government paid for) gym memberships and health/style coaching.
Which countries in the world...
a) Have government-reimbursed childcare
b) Strong laws protecting employees from retaliation in case of pregnancy
c) Total fertility rate above 2.0?
https://www.economist.com/graphic-detail/2021/07/01/which-countries-have-the-most-generous-child-care-policies => the nations from the top of this chart are not doing too well currently...
Almost all of them. The most fertile country in the world by TFR is Niger. Niger has paid leave for mothers and laws outlawing discrimination against women including for pregnancy. Mexico, which is just at replacement rate, does too.
That chart also doesn't show government reimbursed childcare or anti-discrimination laws. It's just about family leave.
Man, that's interesting. Normally poor countries have poor welfare provision. How does paid maternal leave in Niger work? How can a country with such a low GDP per capita afford that?
Giving people a few months of paid leave scales to income. If your company pays someone $800 a year then three months paid leave costs $200. Outlawing discrimination is free. Though of course the question of how well this is enforced is another question. I expect, as it is in many places, it's more available to more professional workers in better regulated and more visible industries.
Then why doesn't this apply to every area of welfare policy? I'm pretty sure that there is a strong correlation between GDP and generosity of welfare.
Niger's GDP per capita is $600. Mexico's is $8500. Japan's is $40k.
What reasons do we have to believe that introducing any of these improvements will push Japan above 2.0, given that other nations with a high GDP per capita and strong pro-parent legislation have failed to do so? I mean, sure, I agree that they're a good thing to have, but would they actually solve Japan's predicament?
You've moved the goalposts now. You've gone from, "these things don't help at all" to "these things don't help enough" when I pointed out high fertility countries often do have them.
Japan's birth rate is low even compared to Denmark (to pick a country). While these changes might not bring it up to the levels of Niger there's every reason to believe it could raise it from 1.3 (Japan's) to 1.7 (Denmark's). That's halfway to replacement and not insignificant.
Anyway, the best way to get your TFR is to import high fertility immigrants. Israel's been getting a lot of attention but the reality is Arabs and immigrant Jews are a hugely disproportionate amount of their TFR. Of course, the Japanese aren't exactly immigrant friendly.
Do we have examples of TFR significantly increasing in a given country after pro-parent policies were introduced there? I'm willing to update my beliefs in this regard if that's the case.
The app would not be "dating" per se, instead, the app would arrange a time and place for users to meet and work on a hands-on project together, very little technology used. The users would meet regularly for planning and for work sessions, meal breaks included, perhaps travel too. Perhaps larger mixed group functions would work better too. Somehow it ends up in casual meals that lead to after-work drinks and dancing.
I doubt that dating apps actually increase the birth rate. Or perhaps actually existing ones made by for-profit companies don't, and one made by a pro-natalist government would have to work very differently.
Exactly, so my question is, differently how?
One obvious issue: a for-profit dating app benefits from people continually using the service, not permanently pairing up and deleting it. Fertility appears nowhere in the bottom-line of such an app. A government might directly aim at fertility, and not have to use an app to do so. So if Hanson is right that fertility is associated with low status, a government could reserve its highest status jobs only for people with above-average fertility.
In the USA, other factors being equal, the less of an age difference there is between the bride and groom the less the likelihood that they will later divorce - so a list of nearby potential spouses that are very close in age should be helpful
Tempting to think age similarity might not be the causal factor. Interesting to think about what it might be instead.
Getting people to go on dates isn't the hard part, it's getting them to settle down and have kids. And then persuading them to have two or more kids, despite the fact that they probably live in a tiny apartment.
In most places, you're right, but Japan actually does have a problem with a relatively large minority of young people utterly uninterested in romance with another real human
This, and again this.
I don’t think dating is the problem. I think someone should do some really serious research into how women feel about having children their concerns about having children and what might encourage them to want to have children. I know it’s picked at, and maybe there is a really serious body of research there that I’m not aware of, but my belief is that’s where the heart of the matter is
Women are clearly interested in having a child to replace themselves -- the TFR rarely falls below 1. What they are not necessarily as interested in is bearing some man's replacement child as well.
The reason people most often cite is the cost of children and the difficulty of childcare. Japan has a high cost of living and historically very bad support for working mothers, so not surprising.
Subsidizing childcare appears to have no effect:
https://www.overcomingbias.com/2011/06/priceless-fertility.html
Robin Hanson's view is that higher fertility is associated with lower status today, outside of very unusual subcultures.
https://www.overcomingbias.com/2010/12/fertility-looks-bad.html
https://www.overcomingbias.com/2010/08/billandme.html
Sounds like a buzz kill.
The central problem with dating apps is that the most appealing people using them will either (i) not be interested in an app-deleting relationship; (ii) have something about them that most people don't like once they learn it; (iii) both. This becomes increasingly true with age. People disproportionately try to get dates with the most appealing people using them, realize they are lemons, and are disappointed. I guess the mandate stops them from deleting the app but are you going to mandate that people "swipe right" at least 10% of the time? Mandate that they go out on max(3, num_of_matches/10) first dates per week? They're still going to end up disappointed by nearly everyone they meet.
I would suggest not using a "dating app" at all but trying to get some of those declining lines on the infamous "how couples met, by year" graph to recover. Dating apps as they exist have significant advantages too, and why fix what is already somewhat working? Young adults should have more friends; if you are willing to mandate things, why not mandate they join a coed sports league? High schools should have more dances, and colleges should have some similar "get dressed up for this fancy party" events. Research indicates that coworkers dating is already kind of normal in Japan, which is an advantage.
Thank you for engaging with the question. I'm not sure I follow your reasoning though: you're saying that people go for the most attractive people on the app, but then realize they don't like them? This seems less of a problem than "people go for the most attractive people on the app, so everyone competes for the same 10% of users" because 80% of the remaining 90% is not going to be able to get dates with the top 10%, even assuming the top 10% doesn't just date each other.
So managing expectations is important. My first idea was to make everyone's visibility inversely proportional to their popularity. If you're a single pop star and everyone wants to date you, the app should only make you visible to a tiny percentage of users, or perhaps excuse you from using it at all, because clearly you'd have no problem dating anyone. By excluding or deemphasizing the most attractive people, the average user is less likely to a) develop unrealistic expectations about their dating prospects and b) get demoralized by repeated rejection by people out of their league. And A compounds B, of course.
I would hope that, on top of this, the matchmaking system makes at least some effort to bring compatible people together, and perhaps provide an incentive to go on first dates depending on how likely it is that you'll get into a relationship based on proximity, interests, and personal history. I think it's too pessimistic to say that everyone ends up being disappointed by every date the app provides them: how does anyone date in that scenario, app or no?
I agree one dating problem is overestimating the attractiveness of the average person (especially without makeup) and setting expectations accordingly. But I don't think an online dating app can fix that just by restricting your options by level. Too easy to notice attractive people in day to day life, plus Netflix. If anything, at least the dating app is showing a subpopulation of (mostly) single people.
Whoops, that should be min and not max, otherwise going to be some very busy ladies.
No need for an “app”. Just announce that all Japanese families with 6+ children will get a payout of $1m tax free as soon as the 6th child is born. Russia has tried a similar payout program but the amounts are much lower and you get paid starting from the 2nd child onwards
Hmm. If you are trying to get your children well into the professional class, in the US at least it's going to cost you more than 1/6 of a million per child, including special tutoring, good private schools at all levels, university, and graduate school. And that's not including the opportunity cost of raising them. (Simply attending Harvard is $51K per year, not counting room and board. $74K for the whole deal, per year. So the next best thing to $300K.)
You ask why am I postulating Ivy League and its prices? Because we have more and more of a winner-take-all system, and if they merely graduate from a good state school, they may well wind up asking people whether they want fries with that, even if they study something in high demand. If they go to a merely good college, they won't make the contacts needed to get into the class where inflation-adjusted income is going up, rather than unemployment and outsourcing. And they won't get recruited for anything that will put them on track for a really good income.
OTOH, the average in-state annual cost is "merely" $25K, and if your local schools are decent, you probably won't need to pay for elite private schools to get them a chance at admission. But that's still $100K per child just for undergraduate. $600K total, before feeding them, clothing them, or sending them to grad school - which they'll probably need just to get a fairly good job. And doesn't count the cost of someone tending them - either one parent stops working, in which case the $1 million should replace their salary, or someone is hired to do that job.
I don't know the equivalent costs in Japan. They may be lower. But OTOH, I also assumed lots of people want 6 kids enough that they'd accept breaking even - I don't find that especially plausible.
How many Harvard graduates who go on to get JDs and join an NGO litigating for environmental justice in the Congo does the world need? I hazard the present supply is probably more than adequate to the demand.
A solution that works for people who make up the class of welders, truck drivers, CNC operators, plumbers, bank branch managers, Chik-Fil-A frannchisees, airline pilots, nurses, respiratory therapists, fifth-grade teachers, farmers, and car salesmen would be far more useful.
What's the lifetime TFR of couples who only want to have kids if they have enough funds to pay for their future Harvard degree? Somewhere around 0.5? That group is basically a lost cause when it comes to government programs.
More interesting perhaps is the TFR of those who only want to have kids if reasonable diligence will allow those kids to live at least as well as their parents.
Did it work?
For Russia? Doesn’t look like it.
Even if you could force everyone to date, it wouldn't help much, because you're looking at the wrong part of the problem in the first place.
In most of the West, yes. In Japan, not so much; a lot of their problem actually is people not pairing off.
But *why* they aren't pairing up is the key question. If women (largely correctly) assume that their bosses will tank their career prospects and their husbands will expect them to do all of the house work and child care, that doesn't strike me as an unreasonable aversion.
I agree, but that means L50L is wrong about enforced dating not helping with the birth rate problem - if a large chunk of the low Japanese birth rate is that there are social costs to women dating, then removing the option to not date would solve at least that large chunk (it'd also abrade those social costs over time, as socially punishing half of society for something they're legally required to do isn't sustainable).
Does that mean we should do it? Probably not; there are serious moral problems with it, like with most ideas lifted from a hentai plot. I'm merely disputing L50L's claim that "it wouldn't help much".
There may be endocrine issues that affect sex drive.
Is sex drive correlated with number of children in rich countries with easy access to contraceptives? (That's a genuine question, I wouldn't be shocked either way).
Sex drive is probably correlated with marriage rates.
This; the problem is that it's incredibly hard to raise children in most rich countries nowadays. People *want* children, but can't afford the house, or work is all consuming, or childcare is unaffordable, or extended family support is unavailable, or social mores have changed to make child rearing way more labor intensive (my parents' idea of parenting was, broadly, to kick me out of the house until dinner, feed me, then broadly ignore me as I read a book/played games. It was a happy, loving childhood! If I sent my daughter outside unsupervised now, I'd have the police called on me.)
Agreed.
The problem with that theory is that poverty has been falling sharply and real incomes rising, which should have the opposite effect.
Humans should be sensitive to local conditions, not global ones. Even if we're close to the carrying capacity of the entire Earth, resources are still superabundant in developed countries, meaning that if you hypothesize that humans adjust their birth rate to the environment, then birth rates should actually be increasing.
Except for space, time and lack of stress. Which turn out to be the most important resources to people who want to have kids.
Currently some of the poorest countries have some of the highest fertility rates, indicating that we are FAR from Malthusian limits.
And yet birth rates are falling almost universally across the planet. We won't see a world with 20 billion people on it.
Previous forecasts of declining fertility in sub-Saharan Africa turned out to be wrong.
https://foreignpolicy.com/2022/05/13/africa-century-economic-growth/
What do you mean? The UN report that that article links to shows a decrease in total fertility.
"The sheer scale of Africa’s demographic momentum may come as a surprise. This results from the fact that a long-forecast demographic transition has been happening in Africa at a far slower pace than was expected even a few years ago."
Don't be too confident in what the future holds.
That would be more persuasive if birth rates tended asymptotically to stability, which is about 2.1 children/woman/lifetime. In fact in First World countries they tend to levels that will result in exponentially *decreasing* populations -- replacing unsustainable growth with equally unsustainable decline. (Wkipedia suggestst the average TFR for "high income countries" is 1.5 right now, and it is generally declining still further.)
OK. Well *that* argument would be more persuasive if there were signs that different countries follow very different trajectories, or if there were signs that within some large heterogeneous region (such as the Eurozone) there was on average a tendency towards replacement even as some countries were higher and some lower.
I am of course not saying you're wrong. Only that there is no empirical evidence to support your hypothesis.
Carrying capacity is contingent on the level of technology and social organization (and has correspondingly gone up and down in the past, mostly the former in recent centuries).
Furthermore, Malthusian conditions ruled for most of history. You need something else to explain the change, like birth control or improved education.
I disagree that humans never got close to carrying capacity. Carrying capacity didn't change that much between the Neolithic & Industrial revolutions (those two things increased it, and that's about it). As populations grew death rates increased from famine, disease & war until an equilibrium was reached.
I think it's more a response to crowding than to total population, but maybe a tendency towards crowding is a result of this high a population at this tech level.
Why not? Lots of people end up in successful relationships as a result of being legally mandated to attend school.
I'm putting together an introductory college English class with an "Intelligence: from IQ to AI" theme. I'm looking for recommendations of fiction (especially short fiction). Stories featuring collective intelligence (including prediction markets) would be especially welcome! Thanks in advance.
Thinking about it, it seems that this is partly a question about First Contact, and, beyond that, human ability to recognise behaviour as intelligent when we see it. The Turing test after all, as I understand it, is not about intelligence per se, but rather about the appearance of an intelligence which corresponds to human norms. If an AI were intelligent by another definition is it possible that we would fail to realise it?
There are a number of classic book-length treatments of such ideas: Fred Hoyle's The Black Cloud features an alien intelligence so different from us that attempts to communicate with it kill those who try. CS Lewis's trilogy, especially the first book, features angelic beings whose intelligence is on a completely different level, or of a completely different type to ours.
I'm reminded always of Wittgenstein's aphorism that if a lion could talk, we wouldn't be able to understand him. But not many short stories there, I'm afraid.
Useful insights. TBH, I hadn't even planned on adding aliens to the mix, but I may want to reconsider. And maybe AGI is basically aliens (orthogonal values, etc.) anyway?
One interesting characteristic of SF writing about the dawn of the Computer Age was that computers (giant brains really) would be much clever and more intelligent than us and answer all the unanswerable questions. In Asimov's story "The Last Question" a group of scientists ask a typical question: is there a God? To which the computer replies, in a flash "there is now!" I think there were quite a few golden age stories along these lines, (mocked, of course, by Douglas Adam's in HHGG).
The Last Question is available here
https://docs.google.com/file/d/0ByoueGSWXluVVUtHYnRJVEg4YnM/edit?resourcekey=0-piVl8D4gdQhTyiaImqjduQ
Not quite your summary, but I don't want to spoil
They're not exactly about "collective intelligence", but "Bookworm, Run!" and "True Names" are two classic short stories by Vernor Vinge that fit your theme to a T.
And the reverse, not exactly about intelligence but very much about prediction markets, from Daniel Abraham (co-author of The Expanse series):
https://www.lightspeedmagazine.com/fiction/the-cambist-and-lord-iron-a-fairy-tale-of-economics/
Daniel Abraham is great--really enjoyed the *Spider's War* series. Will check out the Vinge--thanks!
Back in the 70s[1] we used to think about natural but alien intelligences, what it would be like to meet a species which was just as intelligent as ours but...thought differently. It was a very interesting exercise, throws light on the many unexamined assumptions we normally carry around with us about what "intelligent" and "reasonable" mean.
As I recall Larry Niven wrote a bunch of really good stories that made such encounters come to live. "The Warriors" is his short story about the first meetin between humans and kzinti, and focuses very much on the different assumptions equally intelligent but quite different species make about each other. "The Soft Weapon" introduces Pierson's Puppeteers, who are intelligent but natural cowards, almost the polar opposite of the kzinti. I think you can probably find both online.
Novels he wrote where the difference between how intelligence species think is well explored and crucially important include "Protector" and "Footfall".
-----------------
[1] Perhaps because we were far more optimistic that this might actually happen, some time.
For some reason I've read very little Larry Niven, so the suggestion is much appreciated!
It's not the central trope, but Three Worlds Collide by Yudkowski features prediction markets.
https://www.lesswrong.com/posts/HawFh7RvDM4RyoJ2d/three-worlds-collide-0-8
Excellent. I'll take a look!
“Lena”, a short story by qntm. https://qntm.org/mmacevedo
It’s about brain emulation. :)
Nice. I'll check it out!
Now for a less relevant, non-sci-fi answer:
Buried in 'The Garden of Forking Paths', one of Jorge Luis Borges's most memorable stories, is the Chinese narrator's bizarre claim about his motivation for spying on the UK while living there during WWI. He tells us that he was spying for Germany not because he wanted Germany to win the war, but because he wanted to prove the competence of Chinese spies to a German spymaster, who is said to have doubted them. And to prove that competence, the spy sends a secret message by one of the most baroque methods ever dreamt up.
Many people who reflect on the story probably think about the many-worlds interpretation of quantum mechanics. For me, it's also about the lengths an intellectual might go to so as to prove to themselves (and others) that they really are as clever as they aspire to be.
Oh, I should definitely revisit Borges! Thanks for the reminder.
Permutation City by Greg Egan is an exploration of the idea that consciousness is purely a function of information patterns, such that with a suitable encoding a cloud of interstellar dust might be conscious - or even mathematically identical to the mind of your neighbour
One of my favourite novels, but maybe a bit dense for my first years. I'll give it some thought. Maybe an excerpt.
You've got a MUCH higher tolerance for political risk than I do...
I'm not entirely sure if I understand what you mean . . . you think this course is a risky choice? Maybe?! I teach at a small junior college in Montreal. For now, at least, this doesn't feel like much of a risk to me.
I'm not sure if this is exactly what you're looking for, but I've always enjoyed Asimov's "The Last Question."
https://www.physics.princeton.edu/ph115/LQ.pdf
A classic, for sure. This might be a good way to close out the course.
If you're going to get into AI you might do well with his "I, Robot" collection of short stories that explore all the curious manifestations of his "Three Laws".
I'm not sure how well these have aged with respect to discussions of AI in its present (or likely future) incarnation(s). The stories are fun, but I don't know if I'd have a lot to say about them.
I think where they are interesting is in the question of control. Asimov's Three Laws were designed to restrict robots from becoming dangerous, because his robot novels are built around the assumption that people would find them deeply frightening -- for exactly the same reasons people worry about AI risk now, that a machine that could act faster, think faster, and think flawlessly would be exceedingly dangerous to human beings if it were not firmly controlled. The Three Laws were designed to do just that, to institute perfect 'alignment' in the AI-risk crowd terminology.
What Asimov's stories explore, then, is what actually might happen if you had such a thing. Do they work the way you thought they would? What are the unexpected consequences? That might be a very interesting discussion with respect to hypothetical dangerous AIs and hypothetical "alignment" efforts. Suppose your "alignment" effort does exactly what you hope it will. Does that solve the problem? Without introducing unexpected new problems? Maybe not!
You make a good point. The Three Laws stories would make a useful bridge from King Midas/Monkey's Paw style problems (be careful what you wish for) to real life instances of AI engaging in perverse instantiation.
Ted Chiang's short story "Understand" which is in Stories of Your Life and Others is really good. That whole book is pretty amazing to be honest.
https://en.wikipedia.org/wiki/Understand_(story)
Excellent--I'd planned on including some Ted Chiang, but I hadn't narrowed it down. Helpful.
The classic is obviously Flowers for Algernon by Daniel Keys, but two others that come to mind are Peter Watts's Blindsight, which posits an alien race with intelligence but not consciousness, and Adrain Tchaikovsky's Children of Time, which depicts an uplift programme gone wrong and producing human-level intelligence in insects. The Culture novels of Ian Banks make a lot of use of AIs, especially to investigate moral issues. I can't think of any short stories immediately, but somewhere in sixty years of reading SF I must have come across some. I'll add any I think of tomorrow.
Yep, FfA (in short story form) is in the course pack. I love me some Culture novels, and CoT was great too, but they might be a bit long for my purposes. Do let me know if you come up with some shorter suggestions--thanks!
This is unfortunately in no way short, but Vernor Vinge's "A Fire Upon the Deep" has some fascinating ideas about intelligence and what it means to be intelligent. There are insanely powerful and alien super-intelligent beings, who can construct entire mortal minds the same way a human might build a machine out of scrap parts. The maximum possible level of intelligence depends on where you are in space, and if you venture into a region where this maximum is lower than your own intelligence, you will literally get stupider. Most importantly, there are the Tines, whose individuals are not individual animals, but are instead packs of animals who share thoughts between themselves using high-frequency sound. That has all kinds of fascinating implications, which you will have to read the book to find out.
Oh yeah, I loved the Tines! I might try for a good excerpt. Thanks for the reminder!
Interested in learning about the pros and cons of empathy? This overview is a good place to start: https://www.lesswrong.com/posts/SMziBSCT9fiz5yG3L/notes-on-empathy
I don't read LessWrong much anymore, do you feel it's worth the time to keep up with?
I scan the headlines and read the occasional article. I don't much go in for the 40% of the stuff that's about preventing the AI-pocalypse, but occasionally there's stuff in the other 60% that is worth my time.
Looks promising, will give it a read.
A bunch of comments on Renaud Camus's interview:
a) I'm surprised he calls the process "replacism" (or the French equivalent). I'd call the process he describes "homogenization". Making consumers and workers more fungible, more similar to each other, makes them more homogeneous.
b) To the extent that this spreads formerly local options around the globe, it doesn't seem like a very worrisome trend. My local supermarket (in suburban South Carolina) carries sushi. 50 years ago, one probably couldn't have gotten it outside of Japan. It probably displaces some local item. This doesn't seem earth-shaking to me. Presumably Coca-Cola does similar things in the markets it has spread to.
c) The devil is in the details: How forcefully does the process occur? Are we talking about everyone learning algebra and everyone trying sushi, or violence against those speaking a local language? Camus mentions Breton not being taught in schools (if Le Pen had won). How forcefully was this to be done?
d) My personal preference is to have political and economic systems that allow the widest choice of live options to the median people in the population. Adding internationalized options tends to widen peoples' choices, but displacing local options tends to narrow them. I don't know what factor dominates.
e) Another way to look at this (admittedly Taylorist - which is usually _not_ my favored leaning) is to view it as a spread of "best practices". If nation A has a more efficient way of manufacturing some widget than nation B, why should nation B not adopt it?
Seems like an interesting Maslow based reaction. People who are well off and highly educated have way more time to care about complex and obscure abstract problems while voters on average are not in that position.
Also why not run on the EA platform in San Fancisco? Seems weird to run on it in Oregon. And then piss off the pro-owl movement.
I don't have a good understanding of why there hasn't been a credible "movement candidate" run against Nancy Pelosi. Presumably everyone is waiting for her to retire so they can run for an open seat?
Edit to add: also, the saying is that Portland is where young people go to retire. And "effective altruism" seems like a message targeted to retired young people.
Pelosi's district is in the old money patrician section of SF. I doubt there are more than 6 resident voters who have heard of EA or who wouldn't regard it as some harmless amusement the kids are up to these days, like that World of Battlecraft thingy or whatever it's called. Probably involves the Internet, right?
Shahid was relatively credible. I mean not Fetterman level but still. The issue is that the entire California Dem establishment will come down hard on an anti-Pelosi candidate. Also voters don't like to dump their rep when their rep has a high seniority leader position cause it provides extra pork for the district.
isn't 20 points generally large enough that no single thing would probably have made much of a difference?
"It's great that your are super worried about the AI reverse rapture double apocalypse, but I'm working 60 hours a week have to pick three out of : Rent, Car, Food, Childcare"
I have found that « cause célèbre » candidates almost always underperform. Also, they tend not to give the most neutral political analysis in their fundraising emails ...
Johnny Cash's "Hurt" video.
Old Yeller
"My beautiful woman" youtube series. It's a Thai life insurance commercial.
I don't know about people in general, but just talking about Heinlein's "The Man Who Travelled in Elephants" is enough to make me mist up.
NieR: Automata makes me cry whenever I reach the final ending. I'd say it's especially effective if you played the original NieR. The downside is that you really do need to invest time in the 40+ hour experience.
For a shorter one, the Violet Evergarden anime. For an *even shorter* one, Violet Evergarden episode ten.
I wouldn't recommend Violet Evergarden overall, but that episode is one of the better ones, and I can confirm that it is very sad.
To be a true superstimulus it would probably need to be shorter but the videogame Spiritfarer is very good at making players cry.
funerals seem to be pretty effective
I concur. Funerals are *great*, even when I barely knew the deceased and didn't care a bit upon hearing of their demise, simply being in the cemetary and listening to the eulogies makes me cry. Ideally during late autumn and a very light rain.
Funerals rocks, highly recommend.
Why do you think we need a vague, weakly supported theory that "some contaminants" are the (exclusive, no less) cause of obesity, when we have
1) large,
2) readily available amounts of
3) energy-dense, quickly digestible food
4) that's optimized to make us crave more of it?
That's like trying to find a new theory to exclusively explain the tides, when we're already pretty sure that they're caused by the Moon and the Sun.
Was this untrue in the US in 1980? If it was already true in 1980, then why did obesity keep going up over the last 40 years?
The answer is in parts I and II of the series, current theories of obesity (e.g. food availability, portion size, palatability, energy density) do not adequately explain the obesity epidemic and in some cases are actually contradictory.
The question is "Why is #4 a thing that can happen?" What change started some time in the 1970s that made food much more crave-worthy across a wide range of diets in the modern world? Why do people feel cravings in excess of what their body needs? Why do people have wildly different experiences in terms of how easy it is to resist cravings?
For 30 years, I've had no trouble keeping my weight. Never had a "freshman fifteen," never had to count calories, I just eat until I stop feeling hungry and I stay at around 145 pounds. If I eat extra desserts one day, I'll feel less hungry later and it balances out. I don't think I'm particularly resistant to temptation - certainly I play video games too much for my own good - it's just this particular temptation seems to have no power over me.
But this isn't a universal experience - if my wife eats extra desserts one week, then she has to put more effort into counting calories or she'll gain weight. Why do two people in the same house eating the same food have wildly different experiences when it comes to cravings and the ability to maintain their weight? It's an interesting question, and I think a "homeostasis" model (my wife and I have different set points, and hers is too high for some reason) makes a lot more sense than a model based around willpower alone.
It's as persuasive as the medieval theory that infectious disease was caused by miasmas.
Doesn't seem particularly crazy; mostly it seems vague. "Could be this, could be something else." I mean, if the cause of obesity is some particular unnoticed bacteria in the gut, which has infiltrated water supplies - that explains exactly the patterns, and would satisfy the contaminant description. If something is in the air - same deal.
Mind, I think it's right, I just don't think it being right substantively narrows anything down for us.
Have any heads been crushed yet? If not, I'm not sure it is really the Kids.
Not til the fourth episode…
Correction. No heads crushed. Only an attempt to crush the CN tower.
Yeah, I could use some more “Girl Drink Drunk” myself
https://m.youtube.com/watch?v=8C4TGGtPzBU
The ICO craze of 2017 never recovered, and I expect NFTs to play out similarly. Bitcoin will probably still exist in 5 years and may even go through more bubbles in the future.
I think Bitcoin is worth less than $10000 per coin, so I consider it likely the market will continue to decrease.
How did you arrive at that number, do you think 10000 is the intrinsic value of Bitcoin?
In long form: https://www.newslettr.com/p/cryptocurrency-part-2
In short form: there actually is a need for Nakamoto consensus to do time-keeping for smart contracts. Everything else is going to move to either a centralized system or a proof-of-stake system.
I would say 90 percent, this has happened tons of times in crypto history.
I would give it a 30-40% chance to recover to previous highs, and think that it will more likely recover somewhat but not to previous highs for two main reasons:
1. Tighter monetary policy, increasing probability of a recession. There was a lot of cheap money going around the past few years, and with all "growth" tech companies already highly overvalued, not a lot of places for it to go. I believe a lot of it went into crypto hoping that the next moonshot company would come out of it.
2. Increasing likelihood of government intervention/regulation. I don't think the powers that be in the financial regulation arms of the US government are very happy with the current state of crypto. The same can be seen across the pond in the EU. The recent Terra collapse may expedite how quickly they start really intervening and limiting what is possible.
Fiat currencies are entering hyperinflation spirals in multiple first-world countries, if anything I'd expect an influx of capital into crypto as people escape fiat in panic.
If crypto was a currency in an actual country - ie it was a unit of exchange - then that country would be experiencing hyper inflation right now. And the price collapse indicates the opposite of what you think is going to happen. Interest rates going up guarantees some return on investments, while a collapsing crypto doesn’t.
I haven’t seen any signs of hyperinflation anywhere yet. What I do see is supply-driven price increases, which are the sort of thing that affect all currencies the same, whether fiat or other. (If a chip shortage causes a car shortage then you will need to pay more for a car whether you are paying in dollars or bitcoin or gold or sex.)
It's true that cash giveaways in the previous year or two are a big part of what happened during that time. But is there any reason to think they are contributing to *continued* price increases, more than a year after they stopped? They should already be baked into current prices, so if prices are continuing to increase, it must be due to new sources, right?
Correct, it is now 100% a federal reserve issue.
Doesn't have to be, no. That's what makes inflation so evil. You get some one-time external shock -- like a sudden jump in energy prices, in the 70s, or a big infusion of government money, in the 2020s -- and then it can become self-sustaining. In the 70s, when energy prices jumped up, all the things that depending most directly on energy (e.g. transportation, food) jumped up, and then stuff that was indirectly dependent jumped up -- and then of course wage pressure increased tremendously, and people starting insisting on higher wages and higher returns on capital, which led to higher prices, which leads to still higher wage pressure, and around and around you go.
The problem is that inflation is in no small part a result of *expectations* of what will happen in the future, and these expectations need not be solidly grounded in, nor tied down by, any objective fact.
Indeed, in fact there *was* no "energy crisis" in the 70s -- the world was *not* running out of oil, it would *not* be possible for OPEC to drive the price of oil to the Moon, et cetera, but that didn't stop the hysteria before gasoline was rationed (if you remember "even" and "odd" days) for absolutely no underlying economic reason.
It is unfortunately possible for the same thing to happen today. We are already seeing tremendous wage pressure at the bottom of the market (where it usually starts), and that will contribute to another round of price increases (as people who hire minimum wage workers already know is in the cards) -- and of course that will produce yet more pressure on wages. As long as people *expect* prices to rise 10% or more per year, and factor that into their spending, savings, and job-seeking habits, it goes on, regardless of any economic realities on the ground.
That's *why* those of us who lived through the 70s fear inflation like a fiend from the pit: it is perfectly capable of taking off from nothing other than mass delusion, and to survive the individual has to go along with the madding crowd, even if he himself isn't crazy.
So far we've seen the opposite, though. Gold went up as soon as people started to worry about inflation, but crypto is down.
Also, "hyperinflation spiral" is majorly overstating your case, unless there's news I've missed.
Hungary already has fixed prices by government decree. Poland has food prices rising 4% MoM and energy prices rising even further, with cherry-picked gov stats inflation just going past 12% YoY. Turkey was struggling for a while but now they're at 70% YoY. Western EU is slightly more stable but there are already cracks showing.
I'm going by the hypothesis that 30-50% YoY is enough for an unrecoverable spiral, as the currency becomes a hot potato everyone tries to get rid of. And keep in mind the real food shortage (because of energy and especially fertilizer shortage) has not hit yet.
Which countries are you referring to here?
You want to know what humans are really bad at digesting? Plants.
Every other large herbivore can chow down on random leaves, or even grass, things that are easy to find in any environment. Not us, we can only extract nutrition from rare pieces of plant like fruits and tubers. Without agriculture it's incredibly difficult to find enough human-edible plants to survive in any habitat.
Scientific studies on vegetarians usually conclude that they have better general health than omnivores. However, there are major selection bias involved since there is good reason to believe vegetarians live healthier lives overall than meat-eaters. A better indicator of actual health benefits from specific diets might be to look at children, who are less prone to healthy living. There are some studies of health outcomes in vegetarian children and the general tendency seems to be that they have slightly worse outcomes than omnivore children (iron deficiency most of all).
Also, people who don't thrive on a vegetarian diet would be more likely to not try it or to drop it after a while. Being vegetarian could be good for a lot of people without being good for everyone.
>I constantly heard people telling me the human body wasn't good at digesting meat. How untrue is that?
Fat does take a while to digest (because it has to be emulsified in the stomach before the enzymes can get at it), so if you eat like 700g of Christmas ham in one sitting you'll be feeling bloated for hours (I did that once). But leaner meats don't have this problem, and in any case it still does get digested perfectly well, it just takes longer.
Eating all-animal is possible; this was, indeed, essentially the traditional Inuit/Eskimo diet. It does require you to eat a variety of tissue types rather than all "meat" (muscle); you need at least liver for the vitamins, plus fat (including, but not necessarily, milkfat) for the essential fatty acids (this latter is absolutely critical; if you eat all lean meat you will be incapacitated in a week and dead in three).
"(this latter is absolutely critical; if you eat all lean meat you will be incapacitated in a week and dead in three)"
I think that needs to be qualified. E.g. beef is leaner than pork, but beef-only works fine. We have reported cases for people dying from eating only rabbits (which are as lean as you can get). Rabbit starvation. But to my knowledge, nobody has shown that chicken-starvation is possible?
It will be scurvy that kills you. Luckily, seal livers are rich in vitamin C (as are polar bear livers, but they are trickier to get) so you can get by on an Arctic diet if you enjoy offal.
Why would you get scurvy with chicken, when it's not a problem with beef-only?
As noted elsewhere in this thread, polar bear liver will likely kill you from a vitamin A overdose. I'd worry about seal liver too.
Wikipedia says the Inuit get vitamin C from "caribou liver, kelp, muktuk, and seal brain". Muktuk is whale blubber and skin.
I'm referring here to lean cuts; certainly a whole cow or a whole chicken raised on a farm has plenty of fat, but if you don't eat any of the fat tissue (the skin, in the case of the chicken) then AIUI the problem can arise.
Wikipedia references a guy who ate an all-animal diet and as an experiment tried temporarily not eating any of the fat (in order to confirm that this was the cause of rabbit starvation); he indeed suffered the condition.
You'd need to supplement with possum.
How would it do that? From what reactant would the new carbohydrates be created?
I know the glyoxylate cycle can do it, and everything's capable of a limited degree of conversion via acetone, but those are both somewhat aerobic. Propionate isn't a carbohydrate, but is glucogenic (and IIRC fermentation of cellulose into propionate is a big part of cow digestion); I'm not aware of a gamma-oxidation path to convert fatty acids into propionate, though.
In any case, a lack of carbohydrate isn't actually a big problem for humans as long as there's adequate protein intake, since the brain doesn't *need* glucose for energy (it can't use fat, but it can use ketone bodies which the liver makes from fat under low-carbohydrate conditions) and the general use as a metabolic building block can be supplied via gluconeogenesis from glucogenic amino acids.
Oh come on. Breaking everything down to 2-carbon molecules and building stuff up again is hardly equivalent to yeast fermenting a bunch of sugar into ethanol, or meat rotting, et cetera. Bacteria will certainly break down complex metabolites, but they do that for their own purposes, I very much doubt they would then execute some energetically costly anabolic pathway -- and just excrete new carbohydrates. If you know of such an example in the wild I'd be fascinated to read about it.
>Bacteria will certainly break down complex metabolites, but they do that for their own purposes, I very much doubt they would then execute some energetically costly anabolic pathway -- and just excrete new carbohydrates.
They don't excrete them, but if you eat them you'll get the stuff they made for their own purposes. This is how ruminants work; the bacteria digest the grass, then the cow digests the bacteria.
Also, note that I'm dubious about UI's claim.
"I grew up in a mostly vegetarian environment and I constantly heard people telling me the human body wasn't good at digesting meat."
Very untrue. People who only eat meat exist and seem to do fine. Some have done that for decades and even put their children on it. All-meat diets seem to be viable.
Within the carnivore-community, we find the opposite belief. They claim that the human body isn't meant to digest plants. And point out that plants evolve to not be digested and be toxic (true-ish). They roughly draw upon a distinction between prey-animals eating plants (with large, complicated stomachs) and predator animals that eat them (with less need for that, since what you're eating is closer to the right form, hence smaller digestive systems like ours).
But in the end, we're omnivores like pigs, which are not specialized for one kind of food.
And what we eat is not "natural".
The domesticated chicken, cow and pigs are that we eat these days are bred to be much more palpatable and much easier to digest for humans. Something about the fatty tissue is different (I forgot the details, sorry). The same is true for modern plants, fruit, vegetables, cereals. Eat what you like, as long as you have some protein (otherwise kwashiorkor) and some fat (otherwise rabbit starvation).
Mostly everything works fine, apart from individual differences and transition periods. (going "meat-only" with no lead-in leads to diarhea for a week or two, because it's drastic switch to keto)
If we were teleported back into the Stone Age, the answer would be "You eat what you can get and it all kinda sucks".
There's an empirical study linked from Wikipedia: https://web.archive.org/web/20200725082420/https://www.jbc.org/content/87/3/651.full.pdf+html
"On February 26, 1928, he was admitted to the ward and on February 28, started on the meat diet. At our request he began eating lean meat only, although he had previously noted, in the North, that very lean meat sometimes produced digestive disturbances. On the 3rd day nausea and diarrhea developed. When fat meat was added to the diet, a full recovery was made in 2 days."
Obviously, pursuing human studies much further than this would get you arrested. Animal studies might work, but I'm not sure how well they'd generalise.
From friends' experiences: if you never eat meat, there's a significant chance (~30%?) that eating a one-off serving of meat will give you digestive issues. I'm sure your digestive system would get used to it if you made it a habit (but why would you?)
I've been vegetarian for ~6 years (vegan for some of those) and am generally convinced that the healthiest possible diet probably includes meat. I'm also convinced that the difference is miniscule and that the vast majority of people would be better off eating much less meat and dairy and many more plants.
I personally take the following supplements: multivitamin, creatine, omega-3, protein powder
Looking at the human gi system and at our closest (familar) animal analogs, we have pigs and dogs, both very well adapted to digesting meat as a major part of their diet. (Most important parts of the difference between species are certain liver functions, which usually show up in the processing of specific toxins/drugs.) In contrast, animals who consume almost no meat have gi systems that look very different from ours: horse, rabbit, cow and other ruminants. So just starting from that perspective, the hypothetical is very unlikely.
If you're comparing us to herbivores, I think a better comparison would be herbivorous apes. Only one stomach, but their intestines are different from ours.
Really fucking untrue.
Humans have been big meat eaters for hundreds of thousands of years. We not only survived but thrived over that period. If the human body wasn't good at digesting meat, that wouldn't be the case. For most of human existence, getting enough protein was key to survival. Meat was the best way to get enough protein.
When agriculture was invented, most humans in those societies did not get enough protein, because they weren't eating meat, and lost about two feet in height due to malnutrition.
>and lost about two feet in height due to malnutrition.
Are you suggesting that pre-modern farmers were under 4ft tall on average? I assume this is hyperbole but when you're making a point based in empirical fact like this it's good to be a little more precise with your statements.
OK, let's say... 17 inches? I was rounding.
That still sounds super high, do you have a source? That would suggest an average adult male height of like 4'6, right?
Nobody here's saying humans can't digest grains, as far as I know?
Not the thousands of years. Only in the agricultural era. We know for certain that there were great height increases from 1850-1950, particularly in Europe. This was better nutrition in general but certainly involved more meat.
Probably works much better for earlier decades. Meat is getting more expensive with animal welfare regulation. Higher energy taxes also play a large role. Vegetarianism is becoming very common and many people already self-reduce their meat consumption for vaguely ethical reasons.
Vegetarianism has been cited as a possible explanation for the Dutch (tallest people in the world) shrinking.
Immigration is a rather more likely explanation for the slight reduction in mean adult height in the Netherlands.
If I had to pick something that we were overeating, I'd pick sugars and oils, first, then calories in general, then maybe meats.
Any such argument is too non-specific to be convincing. We eat all sorts of things in different amounts, ratios, compositions, sources than we did for the last 2 million years, and "meat" itself is not a natural category when considered nutritionally.
There is no convincing causal model I've read that indicates that meat in large amounts specifically causes harm, as opposed to more specific things like it being dangerous to eat bear liver or if you consume 5000 calories of sugar you are very likely to become obese and have your teeth fall out.
And there's also fish, which we [1st world Westerners] might not be eating enough of.
What's wrong with bear liver?
It's interesting to me that you are approaching the problem from that direction. What is the evidence for the claims that we are either A) bad at eating meat or B) eating too much meat?
Now, the second one has some very strong arguments around environmental sustainability and climate change, but I am unaware of any good evidence for health implications. There are tons of correlational studies about the "dangers of red meat", etc. but the quality of evidence is relatively poor. I'm unaware of any particular good evidence around health implications of eating too much meat (as long as you are getting enough of the other nutrients you need and aren't eating too many calories over all.
We might eat too much meat now. A balance makes sense. I was responding to "the human body isn't good at digesting meat", which is obviously untrue for most senses of what digestion is about.
"Of course, almost half of India is vegetarian and they seem to be doing fine"
Says who?
I mean, they're not all dropping like flies from being vegetarians, but that was never the question. They have a much lower life expectancy than the west. obviously, this is due to a range of factors, but you can't talk about confounders if that level of your argument is simply pointing to a population and claiming they're healthy.
"Hindu vegetarians historically did actually consume some animal products, since (as in most pre-modern societies) their grain supplies were highly contaminated with insect body parts"
Interesting point! I wonder if there is any evidence that improved grain processing technology or insecticide use ever caused malnutrition due to the reduction in insect contamination/fortification in flour. That would be a splendid example of Unintended Consequences.
I wonder about that, because when preparing rice it has to be picked through for this very reason - get out small stones, black grains, husks, insect parts, etc. Then you wash the rice until the water runs clear, to get rid of excess starch. Then you cook it.
That would probably get rid of the worst insect parts. Modern rice is pre-cleaned so you don't have to pick through it, but you should still rinse it.
I imagine Hindu vegetarians were lacto-vegetarians, as there is cooking with ghee (clarified butter), cheese, yoghurt and milk. Dairy products are a source of B12 so that would have helped.
Airline meals booking apparently differentiate between "Asian vegetarian meals" and "Hindu meals":
https://www.alternativeairlines.com/asian-vegetarian-airline-meals
"An Asian vegetarian airline meal is a meal type that is suitable for vegetarian passengers. The meal excludes all meat, fish and eggs but may contain other dairy products such as milk. The Asian vegetarian airline meal is cooked in an Indian-style and is often spicy.
The Asian vegetarian meal usually falls under the category of 'vegetarian meals' but as it's a meal type that is designed for the Hindu community, it can also fall under the 'religious meal' category. The Asian vegetarian meal is also referred to as an 'Indian vegetarian meal' or a 'Hindu vegetarian meal' — not to be confused with the (non-vegetarian) Hindu meal."
https://www.alternativeairlines.com/hindu-airline-meals
"A Hindu airline meal is a non-vegetarian meal type that is suitable for Hindu people. This meal type is available for Hindus who are not entirely vegetarian. The meal will often include some type of meat or fish, as well as dairy, but will never contain beef or pork. The meal is usually cooked with spices and in an Indian-style. The non-vegetarian Hindu meal should not be confused with the vegetarian Hindu meal, which is also known as an 'Asian vegetarian meal'. The Hindu meal falls under the 'religious meal' category with most airlines, whereas the Asian vegetarian meal will usually full under the vegetarian/vegan category. So, bear that in mind when pre-ordering your meal."
Vegans handle B12 deficiency with nutritional yeast - it's available at the supermarket and it has a vaguely cheesy, savory flavor. A lot of vegan recipes call for it.
"Of course, almost half of India is vegetarian and they seem to be doing fine."
I'm not so sure about that. Of the couple dozen Indian students I've met in Germany, maybe two were slim. Maybe one was fit. Almost all seemed to carry some excess pounds. (though I never saw one that was morbidly obese)
Do we have good health data for Indians to say that they're doing fine?
Not being fat isn't identical to being healthy.
Below the West? Definitely, since the elephant in the room skews that.
But below Europe? Unsure.
https://en.wikipedia.org/wiki/Obesity_in_India
Says 5% are "morbidly obese".
However this article about German obseity does not mention the "morbid"-qualifier, at all.
https://en.wikipedia.org/wiki/Obesity_in_Germany
I lean towards vaguely agreeing probably. Don't have the heart to crunch the
numbers and get better resolution on this.
Whether depressed people are more accurate depends on the environment - Depressed people bias towards fatalism and healthy people bias towards agentic optimism, and they're adaptive for different circumstances. There are things in life we have no control over, and things in life we have plenty of control over and which you ask about will determine who is more "accurate".
It's also worth noting that unrealistic optimism might be more useful than accurate estimation, given that humans have other biases, like excessive time discounting or excess fear of social shame. Perfect accuracy in all mental modules is ideal, but if they all have bugs sometimes the bugs are balancing each other out and 'fixing' one in isolation makes the system as a whole function worse.
Before reading the paper or its summary, my prediction is that someone did an experiment deliberately designed to trick people into thinking that they had more control over the experimental environment than they actually do. Maybe a bunch of buttons you press which are totally disconnected from a bunch of lights which come on at random, something along those lines. They did this knowing full well that depressed people would be less likely to fall for this particular trick than normal people, because they knew this would support the kind of "well ackshually depressed people are _more_ accurate in their world view then non-depressed people" that they knew would get clicks.
Having skimmed the paper I feel like my judgment was uncannily accurate, down to the buttons and the lights; enough so that I suspect I've heard of this paper before.
Anyway I stick to my initial judgment that the increased accuracy here is an artefact of the way the experimental design was chosen to make people systematically overestimate the amount of control they have.
The depressive realism meme is dangerous and needs to die before it causes people to.
Let's take a disease that warps your inner world and pretends it's a faithful representation of reality, then publish cherry-picked studies with bad experimental design to reinforce that the disease is right while you, the suffering person trying to get better, are wrong.
Agree. Even if depressed people are better at estimating certain things, they're bad at estimating the things that really, really matter to them.
For instance, depressed people tend to be really bad at estimating things like "How much better would I feel if I got out of bed, took a shower, cleaned up my house and then went for a walk outside?"
Or it could be non-optimal balancing of top-down predictions and bottom-up error signals as described throughout Andy Clark's "Surfing Uncertainty", reviewed by Scott about five years ago. Clark suggests (on page 238) we may "be entering … a golden age of 'computational psychiatry' … in which superficially different sets of symptoms may be explained by subtly different disturbances to core mechanisms implicated in perception, emotion, inference, and action."
Mis-estimation of control (in either direction) is also discussed (elsewhere).