Did anyone else know that 4/20 was Adolf Hitler's Birthday?
After seeing that the set of Ellen's show was basically a copy of the temple on Epstein island I think these folks might be twisted enough to have made up the story behind 420.
I always thought the original tale was kind of sketchy.
I knew it - because here in Austria, every once in a while, a right-wing politician is convicted for posting how much they like Eiernockerl (Hitler's favourite dish) on the 20th of April xD
Everyone's talking about using LLMs to generate code, but personally, I've found them useful for everything *else*.
In my ongoing hobby project, I haven't found Claude Code useful for *coding* because I have high standards for code quality and particular ways I want to do things and so on. However, it's very useful for *debugging*.
Last week as an experiment, I investigated a bug myself (which took me 21 minutes to root cause) and then gave the same bug to Claude Code (which took 10 minutes). Since then, I've switched to just popping any bugs I discover into Claude - it hasn't been stumped yet, and it's a great time saver.
On another note, a couple weeks ago, I thought of a math puzzle that I couldn't solve. I tried asking Claude, but it was completely useless. This week, I tried asking it again, and this time Claude identified it was similar to a problem I hadn't heard of, and googling that led me to a paper with a solution. There's random variation and prompt differences and so on, but I can't help suspect that the big difference was the 4.5 -> 4.6 shift.
When you ask Claude to find a bug, what input do you give it? Do you need to describe the bug in detail, or can it work from just the code and the error it generates (or expected vs observed behavior for bugs that don't generate errors)?
In one case, I literally just pasted in the error message (including the command to run the test case that reproduces it) and that was enough. LLMs are pretty good at figuring out implied commands.
That probably only works in cases where the output is an error that is clearly unintended though.
“Immediately after the incident, Kristi Noem said three men ambushed agents with a broom and a shovel, and she declared them guilty of attempted murder. The White House, DHS, Stephen Miller a host of conservative influencers amplified those allegations.
But the Justice Department yesterday dismissed the charges against the alleged ambushers, saying that newly discovered evidence is materially inconsistent with the initial allegations. Lyons said video evidence shows that the sworn testimony provided by two separate officers appears to have made untruthful statements”
Orwell couldn’t have known about a video camera in the hand of every citizen when he wrote about state “doublespeak.”
I’ve scrolled the Fox News website and this development was never mentioned.
"So rather than continuing to make the same points about AI, I’m going to try and get a little juice by offering Scott Alexander of Astral Codex Ten a bet. Scott is an AI enthusiast, a signatory of the AGI 2027 thing, and someone who has grown increasingly irritated by my existence. (Which appears to be partially, but far from entirely, due to my views on LLMs.) Alexander is also someone who has been deeply interested in the prediction markets scene, and that whole enterprise is based on the notion that people make better predictions when money is on the line. So I’m here to bet Scott $5,000 that AI will not meaningfully disrupt the economy in the next three years, winner to be certified and paid out on February 14th, 2029, under the specific victory conditions I list below."
Not commenting on the likelihood of the bet itself - there's a lot of conditions, Freddie proposes that even if one of them fails to come true he loses the bet and there's more than a few that might fall either way regardless of AI developments - but the part about how there's a lot of desire to performatively beat up on a comparatively small number of "AI skeptics" among the online talking head community rings rather true. Like, there are commenters here who seem to be personally affronted by the fact that Ed Zitron or Gary Marcus say what they say.
Gary Marcus is just an embarrassment at this point. And I say that as someone who used to be a vocal AI skeptic myself. The difference is that I changed my views in response to new evidence.
I believe this is the most important issue in the US. We do not have a real democrasy. We are told to vote for lesser evils every election cylce. Few if any are talking about it outside of election season.
The best solution is a system that optimizes for information while simple enough for most to use.
When corporations want your honest opinion, they use these methods. Not ranked choice (RCV). There is a reason that you might have heard about RCV. A lot of money was used to kill the movement by propping up the next most easiest system to manipulate and then making RCV illegal in some states. It was a campaign that had no interest in liberating us from not being represented.
While RCV is better than what we have, it's not the best. Two methods are better. Approval Voting (likes systems), and score voting (5-star ratings systems).
Star Voting is a score method that is the best we have. It starts with giving 0-5 stars to each candidate, as if they were a product review online. That is followed with a runoff where we take the top 2 candidates with the most stars and each citizen's vote goes to which of these top two candidates their ballot gave the most votes to.
This way you can show more information than approval voting. If Bernie Sanders ran for a third party and you like Bernie way more than you like Biden you can give him 5 stars and Biden only one, while giving any trash candidates you fear possibly winning, like Trump, 0.
You can be clear who you think is a lesser evil in this way without sacrificing your honest vote. Ballots are also determined at the local government level as well, so we can push for this change with relative ease.
The only problem is that the word isn't out yet. We need to get a major campaign underway to unite with approval voting advocates and any honest RCV advocates to unrig our voting system, whatever it takes.
No voting system will ever bring about democracy due to the problem of Rational Ignorance. Any system will invariably be dominated by well-organized blocs, media institutions, and demagogues. The only way we're ever going to have democracy that works is with sortition-based approaches. Not suggesting randomly selecting executives or anything like that of course, or even randomly selecting legislators.
In general though, if you convene a representative stratified sample of citizens to research and deliberate on a topic, they do a pretty good job. Each one is only one of a few dozen instead of one of millions, and they have skin in the game, since they have to live with the consequences of their decision, so they have a natural incentive to be invested in the process and try to do a good job. Something which elected officials and career civil servants often lack.
Any particular reason to support one of Approval Voting or STAR voting over the other? I haven't thought too much about this, but I think Approval Voting might be better in practice, simply because it's simpler to understand for people... I like STAR, but simplicity is itself a virtue. At least when talking to the public.
Many people feel the same way. My personal opinion is that star voting has much of the same function as approval voting, but it's more information. Being able to recognize that I really love Bernie Sanders over Biden is a huge distinction. And when people are trying to tell me to vote for lesser evils. What am I supposed to do with Biden? He's fucking trash. But people are voting for him because they're scared that Trump will win if you don't.
I don't approve Biden, but they'll manipulate the vote by shaming me for not approving Biden.This is all hypothetical of course. I'm not misremembering who was or will be on any ballots.
One thing voting system nerds really underestimate is the importance of simplicity and transparency. We already have people casting doubt on election outcomes as it is, just imagine how bad things would be if it took several weeks to decide the winner and it was the result of a blackbox algorithm that noone understands.
What advantageous property over Instant-Runoff do you believe STAR has? Because it seems to me at first glance like you're starting from there, and then adding back in the need for strategic voting from FPTP, usually considered an undesirable feature for a voting system.
Star Voting is different for a few reasons. You can put the same value for two candidates is one. There's ways to manipulate the vote and campaign for strtegic voting in RVC (IRC is effectively the same in this way, right?)
While you can't escape the spoiler effect, you can make it harder to explain how one would strategically vote to a population. Telling an average person "You must give me 5 stars and the rest 0" would result in political suicide. They would not understand why there could be a fear, and the fear is there for RCV becaise they can feel cynical that enough others will strategically vote so as to be afraid that your vote won't count if you vote honestly.
starvoting.org has more explanation in the education sections.
(The Assistant Axis: Situating and Stabilizing the Default Persona of Language Models)
They ask LLMs to roleplay a bunch of different kinds of characters, then do principal component analysis on these characters’ responses. A principal component of “is like the assistant character” / “is not like the assistant character” emerges.
What jumps out at me is that “trickster” is the anti-assistant character for 2 out of 3 models.
It jumps out at me because “trickster” is how I would describe the character DeepSeek R1 0528 usually adopts when talking to me.
Ok, fine. That guy is the waluigi of the honest, helpful, harmless assistant.
I wrote up a reaction to AI 2027’s grading of their own 2025 predictions [1].
Summary: what was good was trivial, and what was non-trivial was not good.
# Part 1. The good, but trivial
In 2024, everyone knew that agentic stuff was coming and that it was increasingly helpful for coding, because it was already happening. A check on Hacker News chatter at the time will make this clear. Even many of my coworkers at the relatively sleepy mutual life insurance company I work for knew what was going on. We were all talking about it.
# Part 2. The non-trivial and not good
The really stunning, jagged details of AI development beyond 2024 were not predicted at all.
In early 2026, AI sometimes performs at elite human levels on well-defined, formally verifiable computational tasks, primarily math and programming. When it succeeds, the speed and cost are often superhuman, making it economically valuable. Meanwhile, visual perception skills often remain stubbornly below toddler capabilities.
No shockingly giant, specifically agentic model worthy of being called “Agent-1” was created by anyone. There was only continued scaling of existing GPT-like LLMs. Nor is there any evidence that anything resembling Agent-1 is in the 2026 pipeline.
Agentic abilities remain largely limited to automating technical drudgery in math and coding. These abilities are provided through engineering scaffolding that encourages constant human supervision, which remains necessary to have any hope of a quality product.
The idea that frontier models are currently, or are about to become, Yudkowskian doom-agents with their own goals and drives, and interests in engaging in antisocial behaviors to achieve these goals, remains esoteric speculation without any substantial empirical or theoretical foundation. This speculation is no more substantiated today than it was at the end of 2024—or in the early 2000s, when Yudkowsky was first formulating it.
> In 2024, everyone knew that agentic stuff was coming and that it was increasingly helpful for coding, because it was already happening.
I don't think "everyone knew" that. IMO the first agents that were good enough to actually be sometimes useful didn't show up until spring 2025, and they didn't really get good until Opus 4.5. There's a big difference between HN hype and reality, and I suspect you're also misremembering the timelines a bit.
> The idea that frontier models are currently, or are about to become, Yudkowskian doom-agents with their own goals and drives, and interests in engaging in antisocial behaviors to achieve these goals, remains esoteric speculation without any substantial empirical or theoretical foundation.
"Substantial" must be doing a lot of work there, because we absolutely can see agents lie and cheat and so on when put in simulations where this is advantageous (notably VendingBench).
If you put these things together, you've got a poor man's coding agent.
The 2025 surprise was that RLVR pushed the frontier of reliable, useful multi-step reasoning and action farther than some of us might have expected, both on the math and the coding side. This could have been "predicted" with inside information, as the ideation and proof-of-concept demonstration surely long precede the publicly available proof, or it could have just been an optimistic guess based on what had already occurred. Personally I was already expecting rapid AI progress in formal reasoning after the publication of AlphaGeometry 1 in January 2024.
> we absolutely can see agents lie and cheat
Lying requires intentionality. Where is the evidence of intentionality? I have yet to see it. Regardless of the number of incidents where Yudkowsky and team have written tweets attributing intention to some AI behavior, they have yet to provide anything more than intuitive attributions. These are to me unpersuasive. There is no reason to expect LLMs to have goals, other than reflexively anthropomorphizing them because they deploy language in a human-meaningful way.
Which is not new. People have been doing that in movies since 2001: A Space Odyssey and HAL 9000.
I believe that LLMs follow given instructions and reason about how to follow the instructions. I do not think they have or are employing the theory of mind necessary to intentionally deceive.
How many times has a toaster won a solo victory in a proper game of Diplomacy, to Calhamer's rules and victory conditions?
Last I heard, the toasters were only winning a version modified to be just a territory-maximization game with a side order of negotiation. Which is exactly what I'd expect if someone whose LLM or whatever didn't have a theory of mind, wanted to be able to brag about it winning at Diplomacy.
Yes, they could win without a theory of mind that would generalize to the real world. Just like they can solve IMO problems but not do research mathematics at a human level.
Even an automated, deterministic engine for playing the solved game of Checkers could be interpreted as having a theory of mind. It considers what its opponent could do and plays accordingly. That doesn't mean it has a useful theory of mind for the real world.
I think making a human being is a greater accomplishment than anything one can do with a PhD, because the kid can get a PhD too plus there are a million other things the kid will do, the only real issue is that it is applied in a strangely sexist, women-only way, I mean, if we want to be real, for many men, fatherhood is their greatest accomplishment too. I am such a man, too, fatherhood is the only important thing I do. What else am I doing with my life? Sitting in an office developing code that makes rich people richer? But my kid now watches astronomy docus and writes short novels and that is amazing that I/we made such a being!
The kid will get the PhD 25 years later, so you should apply some temporal discount. ;)
> for many men, fatherhood is their greatest accomplishment too
I agree.
The bias against men is caused by men's contribution to family being less direct. Imagine a traditional family, where the man has a job and brings home money, and the woman stays at home and cooks and takes care of the children directly. Of course without the man's contribution this entire setting couldn't work. But it's tempting to say "she is with the kids while he is away", ignoring the part that his "being away" is the reason why she can be with the kids.
To steelman this position, mothers (who stay at home) often have greater influence on the kids than fathers. If your role is mostly about bringing money, you can make a difference in whether your children are starving or not, whether they can afford good schools and after-school activities, etc. But you have little impact on whether they are spoiled or well-behaved, whether they do their homework, etc. So when the people compliment the kids being not only "OK", but "high quality", it is mostly the mother who made the difference. The father can approve of all this, but would be mostly helpless if the mother started doing it wrong.
However, many families do not have the traditional setting "the man does his job and little else, and the woman is always with the kids". More likely, the man does his job *and* then also spends time with the kids, and the woman also has a job (maybe part-time) and takes care of the kids; and the kids spend a large part of their day at school.
Also, both in the traditional and the modern settings, there are weekends, when the father can also be with the kids all day long.
(Also, if you believe that genetics is more important than nurture, then the father's contribution becomes close to 50% regardless of everything else. Maybe except for athletic performance, where mitochondria seem to play an important role.)
What I keep thinking about, how can people who enjoy existing be adamantly “childfree” ? What if their parents were, too? Would they prefer that they themselves have not been born? Don’t they understand that all their accomplishments are their parents accomplishments, even if they were bad parents, because accomplishment requires existence first?
I am a much stricter, angrier person than either of my parents. I've nearly had panic attacks dogsitting. And those were well-behaved dogs, whereas a child is a crapshoot that you're stuck with for eighteen years.
There's over a 90% likelihood that if I had to raise a child they would get thrown through a window. I will count pre-kindergarten death by laceration as a negative accomplishment.
From the "unhelpful answers given by LLMs" department. I had asked DeepSeek which well-known books had their US copyright expire in 2026, and got...
"Journey to the End of the Night (Voyage au bout de la nuit) by Louis-Ferdinand Céline: A seminal, darkly comic, and stylistically revolutionary French novel. (French publication 1932, but the user specified "first publication elsewhere" counts if the translation was in 1930. However, its first book publication anywhere was 1932. So it doesn't strictly fit the "first book publication in 1930" rule, only the translation caveat if a 1930 translation existed, which it didn't. "
For decades, the image of a marijuana user was someone melted into a couch eating Doritos. That image is outdated. A growing body of research — and a massive shift in how athletes actually behave — tells a very different story.
A University of Colorado Boulder study found that 80% of cannabis users in legal states use marijuana shortly before or after exercise. Not sitting around. Exercising. And among those who combined cannabis with workouts, 70% said it increased enjoyment, 78% said it improved recovery, and 52% said it boosted motivation.
For me having a couple of tokes (not a ton but enough to get a head change) really made me aware of my physical body and positioning. IE I was able to focus more on my form.
I think it is helpful in sports. Better focus, more aware of what your body is doing. Even just distance running it’s easier to pick up energy draining faults in form.
This is so strange to think. There were populations in the past who consumed marijuana before battle... so much different from the typical "stoned" thing...
There is a stoner stereotype, think Spicoli in ‘Fast Times at Ridgemont High’, and like all stereotypes it has some nugget of truth. I’ve known a few people who stay ‘baked’ every day and sit on their sofa watching silly TV but it’s a small minority of people who use THC. In calibrated doses with a constructive attitude I think it is more than just a recreational drug.
Substack won’t let me edit posts on mobile today, and also swallowed one and sprinkled it in the void just as I was finishing. Have I mentioned lately that I hate Substack?
If you restrict voting to people with IQ 120 and above, what would change? I think there would be significant pros, but what would be the downsides? Can the bloc of people at that IQ and beyond be said to have unified preferences that would be disadvantageous to those at the lower tiers? Would there be needs of the lower tiers that would go unmet under such a system (what would those be)?
People with sub-120 IQ could presumably still run for office and win, so that's a wrinkle. And it doesn't take a genius to recognize that this IQ 120+ elite would be wise to not just stomp all over the interests of the unwashed masses, but use their outsized power to win over the hearts and persuade the feeble minds of the unwashed masses, so you get the disenfranchised citizens at least pulling in the desired direction instead of rioting in the streets. Right?
Hello me! This was tried, in the form of minimal wealth requirements, as money and IQ/education tends to correlate. Most democracies started like that. And that is precisely the problem. The interests of the poor who are predominantly under 120 will not be represented. This is why universal suffrage was introduced.
This is the No. 1 contradiction of politics as such: we want the smart to rule, not the stupid, but we also do want the wealthy to lord over the poor. Unfortunately, the (relatively) wealthy and (relatively) smart are the same people mostly.
The other big advantage of universal suffrage is that it provides legitimacy and reduces the risks of civil war and rebellion.
Democracy isn't about finding the best leaders (though it's better than most systems at that), but about finding leaders *that people won't rebel against*.
Does that logic extend to women's suffrage? There is almost no risk of them posing a genuine threat to society, and historically, there seems to be no big issues that arose with disenfranchising women. (Stability-wise, anyways.)
The correlation between intelligence and wealth depends on the society.
I mean, in any regime, being smarter is an advantage (ceteris paribus) for trying to get to the top. But *how much* of an advantage it is, that depends. Some societies are more flexible than others; sometimes the son of a tailor has to become a tailor unless something exceptional happens; sometimes the son of a tailor is free to choose any kind of school (except maybe the most expensive ones). There is no perfect meritocracy, successful parents can always provide some advantage to their children, but some societies are *more* meritocratic than others.
Another important factor is sexual behavior. Even if we assume that the society starts stratified by IQ, and people are likely to marry their equals, if the men at the top are likely to seduce or rape the women at the bottom, this will result in highly intelligent children born in poverty.
As a result of these two forces, in bad societies (from the liberal perspective) the correlation between intelligence and wealth will be lower, because stupid kids from powerful families will stay at the top (think Ramzan Kadyrov), and smart kids born at the bottom will stay at the bottom. On the other hand, in good societies, the people will gradually sort themselves by intelligence as the smart poor kids get a chance, albeit not perfectly.
(This was actually noticed and complained about by some leftists, because the side effect is that in the past it was relatively easier to organize the poor than it is now. That's because there were many talented people in the working class, who could use their talents for the leftist causes, whether by writing or speaking or organizing. Modern society allows these talented people to leave the working class, which in effect makes the working class weaker. So ironically, today when someone gives speeches about the oppression and the need to fight it, it is usually not a member of the working class, but some trust fund kid from an expensive university, enjoying their youthful left-wing phase before they inherit daddy's business empire.)
Just by definition that would be about 9 million people in a population of 300 million, probably better educated and better paid, with maybe some increased representation by Jews, Indians, and East Asians.
They would probably have some good ideas and be more able to parse complicated issues and think about the tradeoffs. On the other hand leadership would be completely unrepresentative and probably have a lot of cognitive biases around what everyday life looks like, how people feel about school and jobs, etc. Kind of like the elite discourse now, where the answer to all job problems is learning to code, becoming an entrepreneur, etc.
Just the usual stuff. Human sacrifice, dogs and cats living together, mass hysteria. Sorry, Carlos, my heart is just never in these never going to happen hypotheticals.
Well, the biggest downside is that you've essentially given whichever politicians control the testing and certification requirements carte blanche to choose their constituency however they want to. I'm sure that would end well...
Wouldn't you be disenfranchising about 90% of men at that point? It doesn't matter how much wealth is concentrated in the 10%, you can't go against the will of that many people without them eventually just deciding to kill you. Even military dictatorships need more support than that.
At least the perceived interests for sure. They will be somewhere between Scott's compassionate neoliberalism and full libertarianism because this benefits them. Even if it turns out to be in the interests of the rest, they will likely not see it, they like getting freebies.
If the poor are stupid, I guess they are screwed either way. Either they don't get a vote, and those who vote may ignore their interests. Or they get a vote, but will use it to vote for something shiny and stupid, which will hurt them, too.
But the *perceptions* matter. If they don't get a vote, they will be angry. If they get a vote... it depends on who manipulated them successfully, and what will he do for them. Could be something good. Could be something bad. Could be even trying to keep them angry on purpose, so that they keep voting for the same guy, as long as he can successfully keep blaming everything bad on someone else.
Okay, so i got my wise ass joke (stolen from Ghost Busters) out of the way I feel like I should give you a somewhat serious answer.
I think it would be more likely to consider the needs of the rest of the poor dumb schlubs with ordinary, run of the mill, intelligence if there were some correlation between intelligence and empathy. On that score I’m not hopeful.
If they're not going against them, what's even the point? The real value of democracy is just having an easy system of placating the majority so they don't kill each other. Dictatorship or oligarchy complicates this since you now don't have accurate data to work with, though it does have the potential upside that they might be better at catering to the population than whatever morons the public would have elected. Disenfranchising 90% of men not only removes the main upside of democracy, but also doesn't even solve the issue of populist politics. At that point, just go all the way. Don't half-ass it.
>Can the bloc of people at that IQ and beyond be said to have unified preferences that would be disadvantageous to those at the lower tiers? Would there be needs of the lower tiers that would go unmet under such a system (what would those be)?
Probably. Because those in the above-120 IQ group are unlikely to be very conversant with the interests of those at the lower tiers, especially at the lowest tiers. This proposal, like many similar ones, ignores the fact that much of politics is different groups vying for government to enact policies which serve their interests. https://www.thebritishacademy.ac.uk/blog/what-is-politics/
The first thing that would change would be the testing methods for determining who has IQ 120 and above. Suddenly the questions used to test high IQ would be directly tied to policy positions.
The second thing would be the voting bloc deciding they don't need to explain things to stupid people. This in turn means they don't actually explain it to themselves, and end up with all kinds of gaps in their worldview that are currently exposed by failing to explain them to the dummies.
The third thing would be people calling for increased restrictions when their policies don't win. If only 120 folks can vote, suddenly 120 votes are "the dumb ones," and we should cut them out of the process. This process loops until we have literally one person making all the decisions.
Heh. Appreciate your blunt cynicism and can't dispute any of it.
I think though that the loop would break. By the second or third loop, the above sequence sparks violent insurrection to a degree that the US has never yet experienced. Not just late-1960s or Jan 6 2001 levels, and not sectional like in 1861, but more like Russia-1917 level. And waaaay more nonwealthy Americans own guns than was true in 1917 Russia
Yeah, this is the theoretical pushback-free version. I'm mostly thinking about a previous argument where someone said "good in theory, bad in practice." To which I say, even in theory it's just dictatorship with extra steps.
Seems to me the IQ thing would never fly because it generate so much indignation and is hard to defend. Where’s the evidence it would have good results? Also, there is a good case to be made that everybody, including dumb people, has a right to influence things in the direction they want.
What about subject matter tests about the issues and candidates people will be voting on? About most of these there are simple, basic facts that one needs to know in order to vote for the kind of governing one wants.. In voting for president/vice president, you would be able to weed some people out just by asking whether this is the first, second of third term for the presidential candidate, and which of 5 possibilities is their running mate. You could weed out more by. picking 3 big issues about which the candidate has consistently said the same thing, and then for each issue giving extremely easy multiple choice questions. What’s he said about US tariffs? (a) they should be higher. (b) they should be lower. (c) they should be eliminated . And then about some contentious issues there are matters of fact you can test on. For example, abortion: What do embryos look like at 8 weeks? (a) [drawing of fully developed baby in miniature, 1” long]. (b) [drawing of actual embryo]. (c) they are too small to be seen without a microscope
I think a set of questions this easy, for which one must get a fairly high percent right, would weed out at least 10% of possible voters. And a good case can be made that if people do not know facts this basic they are not capable of voting for the candidate who will run things the way the person wants them run.
I sort of agree but I also worry that these selection tests will become majorly biased (even more so than IQ tests). As one example, I expect many conservatives to know a lot about the details of guns in ways that isn't actually that relevant for gun control, and liberals to know more about details of birth control, potheads to know more about the details of drugs, etc.
Even worse: The tests would probably become "Donald Trump won the 2020 election: true or false?", with the "correct" answer depending on which party was in charge.
For people interested, I wrote up a simple four-point case for AI catastrophe, that tries to be better than existing intros by being simple, clear, and relatively rigorous: https://linch.substack.com/p/simplest-case-ai-catastrophe
Where can I find the latest on voter ID arguments from a credible source? As a foreigner I always thought trying to make an argument that requiring IDs is bad to sound stupid. I understand the underlying dynamics, but the way Democrats message it, to me, sounds dumb. I saw a video of Kamala explaining that requiring IDs is bad because there aren’t any FedExes in rural areas where people can easily make copies of their IDs. Is the issue really the same as in the 80s? And why can’t Democrats at least make a sound argument that begins by conceding that IDs are a good thing?
In theory, there's no problem. *In practice*, it's likely a bad-faith effort to disenfranchise people who are suspected of leaning Democrat.
Though ironically, the Trump realignment means that it is *Republicans* who now depend on low-propensity voters, so if these laws are actually implemented fairly, it would result in a massive Democratic landslide.
I get that part. Why can’t Democrat politicians talk about it with clarity about the secondary (and most relevant) aspects of the conversation instead of saying that there aren’t fedex stores in rural areas? Is it that hard to have a short shpill that teaches instead of one liner that smells funny?
Politicians of any party are almost never capable of talking about *anything* with clarity. If you want sober minded policy analysis look to Substack, not press conferences.
The simplest argument against it is that it's a solution in search of a problem. To the best of my knowledge, every attempt that's ever been made to find voter fraud has come up with a number of cases indistinguishable from zero. And if you think about the incentives involved, it's really, really easy to see why: casting a fraudulent vote involves a non-trivial amount of work and risk for what's usually a completely trivial benefit.
When somebody is pushing this hard to solve a "problem" that doesn't actually exist, you should have a very strong suspicion that they have an ulterior motive.
In my circles there's a deep suspicion that there's large voter fraud related to mail-in voting and electronic voting machines and 'irregularities' that all point in one direction. Saying 'unproven' doesn't eliminate deep suspicion. And it's deep enough, and if even partly true, has consequences huge enough, that at least to me it surpasses any concern I have with disenfranchised having ID problems.
This is one of those problems that, if enough people think it exists, it exists. And, it might actually exist. Unproven in one direction also means unproven in the other.
If enough people think it exists, that is a problem, but not one solvable by changing how we vote. Collecting votes in a huge country is a complex process, and is little understood by most, so it will always be possible to sow doubt and impossible to prove to the public's satisfaction that the system is safe and secure. If a rumor started that many people's drinking water has been doctored or that new highway construction is being carried out using an inferior grade of asphalt that will crack and crumble in the next 18 months, how many people even know how to begin checking whether those things are true? And if the public believes they are true, how could anyone, even with very solid evidence, convince the public they are not true? What visible safeguards could be put in place that would convince the doubters there was no longer any danger?
Thinking about it... when I've voted, the voting booth check-in already has a list of registered voters in that district, and check you off when you give them your name. So, just by requiring everyone to confirm their name and checking them off as Voted, voter fraud is already reduced to easily-identified occurrences. You have to vote in an actual registered voter's name, and to do that without a double-check conflict you have to know them well enough to know they aren't voting themselves.
Note that the latter claims that voter ID laws disproportionately affect traditionally disenfranchised groups. If so, it seems reasonable to oppose them, in the absence of evidence that there is an actual fraud problem that they will solve.
Imo, this is a great example of how this "disproportionate effect" logic ends up paralysing you. Like, I understand there are some people in the US who are against ID ideologically, but Im not and I think neither are you, and from that perspective:
you can get by ok without ID -> poor people dont bother getting ID -> "we cant require ID for that, it would disproportionately affect the poor" -> you can get by ok without ID -> etc.
Its a lot like objecting to new technology because "only the rich will get it".
Except that the United States is not one of the societies where "you can get by OK without ID". Maybe if you live on an Indian reservation somewhere and basically never leave. And we should put some thought into how to make sure everyone on the rez gets to vote. But over the vast majority of the US socioeconomic landscape, no ID = extreme hardship. To the extent that many states explicitly grant drivers' licenses and state ID cards to known illegal immigrants, because the alternative is basically forcing them to become serious criminals if they want to feed their family.
So if there's a problem with people not having IDs, the solution pretty much has to be "how do we get those people IDs, stat?", not "OK, we've made it so they can vote, moving on..."
Well, Im not from there, but from what I hear, theres some complications around which kinds would count, and who has those, etc. Maybe Im confusing something, but it sounds like the entire existence of this discussion would make no sense if youre right.
>Its a lot like objecting to new technology because "only the rich will get it"
Except that, unlike voting, no one has a right to new technology.
Note also that I did not mean to imply what you seem to have inferred. I did not say that disproportionate effect is a trump card, but rather that disproportionate effect shifts the burden of proof. And probably increases it.
I'm switching phones from Apple to Samsung. Does anyone have any recommendations for essential settings to enable/disable or anything like that? I'm really not familiar with Android at all.
I just made this transition. Big things for me was unlocked bootloader, so I can put different ROMs on and root the phone. If you know any programming, it's basically a little linux computer at that point, and you can make it do whatever you want.
>> Part of that is because word order is very flexible, as Scott suggests, but another reason is that Chinese poetry isn't supposed to rhyme. Rhyming is just too easy, so it sounds as childlike as alliteration in English.
I don't understand how this comment got written. Neither of these claims is true. Word order is not flexible. (Classical Chinese is flexible in what part of speech a character might represent, which allows you a lot of freedom to throw a bunch of random characters in sequence and try to close-read your way to something coherent, but that's very different from word order being flexible.) And Chinese poetry is supposed to rhyme. Old Chinese poetry rhymed. Middle Chinese poetry rhymed. Modern Chinese poetry rhymes too.
Here's the introduction to Eric L. Hutton's translation of the Xunzi:
> The Xunzi contains numerous rhymed passages. A number of these are quotations from the Odes, which is an ancient collection of poems -- or more accurately, songs -- that the Xunzi treats as a repository of wisdom, and which it cites to illustrate and support its claims. Many other rhymed passages appear to be original to the Xunzi.
> The function of these rhymed sections and their significance for understanding the Xunzi are substantial issues that merit lengthy discussion more appropriate for an article or a book than this introduction, but I consider the presence of these rhymes a feature of the text that is sufficiently noteworth to deserve being reflected conspicuously in the translation.
There is a potential issue, noted further on in the introduction:
> The identification of the rhymes in the Chinese text requires detailed knowledge of ancient phonology that I lack, and so I have relied on published studies of rhymes in the Xunzi by other scholars. Since their analyses may have missed some of the rhyming passages, and since there is ongoing debate about how to reconstruct the sounds of ancient Chinese in the first place, I do not claim to have identified every instance of rhyme in the text [...]
Rhymes in ancient poetry are generally not obvious to the modern observer. We do not conclude that the poetry didn't rhyme. It did rhyme, but the nature of the writing system is such that recognizing rhymes takes a trained expert.
No such excuse is available for modern poetry. It rhymes and you can hardly avoid noticing this.
Is there a name for the sort of idea that does not quite work when you try to define things and pin it down, but also seems to capture something valuable at a higher level before you dig down.
I remember this from the concept of "elite overproduction" in Scott's review of the Peter Turchin book - it was ill defined and it was hard to pin down, but also wormed its way in.
I have found this with a few concepts that I have come across lately where a core idea seemed to have a useful insight, but trying to pin it down seemed to destroy it. But the feeling of useful insight remains
Has anyone worked out a way to work with these ideas and keep the insight despite the difficultly of making it specific?
"Deepity" is a word coined by Daniel Dennet for a saying that sounds like profound insight but is only true in a shallow or trivial way. E.g., "love is just chemicals in the brain" sounds like a profound criticism of the concept, but is only true in the sense that *everything* is chemicals in the brain, so you're really just criticizing the concept of "having thoughts" in general.
Similarly, I think "elite overproduction" sounds insightful because it conjures up images of entitled college students complaining they can't get jobs in underwater basketweaving, but if it's true at all, it's only in some vague sense that *every* college student is potentially an elite.
The point on pinning down Elite overproduction is scattered through, but the core criticism is
>Fourth, everything about the elite overproduction story confuses me. Who are “elites”? This category made sense in Secular Cycles, which discussed agrarian societies with a distinct titled nobility. But Turchin wants to define US elites in terms of wealth, which follows a continuous distribution. And if you’re defining elites by wealth, it doesn’t make sense to talk about “not enough high-status positions for all elites”; if you’re elite (by virtue of your great wealth), by definition you already have what you need to maintain your elite status. Turchin seems aware of this issue, and sometimes talks about “elite aspirants” – some kind of upper class who expect to be wealthy, but might or might not get that aspiration fulfilled. But then understanding elite overproduction hinges on what makes one non-rich-person person a commoner vs. another non-rich-person an “elite aspirant”, and I don’t remember any clear discussion of this in the book.
>Fifth, what drives elite overproduction? Why do elites (as a percent of the population) increase during some periods and decrease during others? Why should this be a cycle rather than a random walk?
The point about worming into your head in the the wrap up
>First, the concept of “elite overproduction” is one that worms its way into your head. It’s the sort of thing that was constantly in the background of Increasingly Competitive College Admissions: Much More Than You Wanted To Know. It’s the sort of thing you think about when a million fresh-faced college graduates want to become Journalists and Shape The Conversation and Fight For Justice and realistically just end up getting ground up and spit out by clickbait websites. Ages of Discord didn’t do a great job breaking down its exact dynamics, but I’m grateful for its work bringing it from a sort of shared unconscious assumption into the light where we can talk about it.
This is the sort of thing I mean - there is a really good core of an idea but operationalising it is difficult.
Other examples I will mention but I am not a good enough writer to explain might be the instinct behind DOGE - the feeling that we have too many people working on managing or recording how we work relative to people actually doing work - so compare number of bureaucrats or accountants or managers to say plumbers or construction workers. However when you actually try to work out what you can cut you find like DOGE did that you cut important things you care about.
Spinning the same idea from the other side of the political spectrum it might be the Abundance argument from Ezra Klein - can you actually make concrete suggestions about what you would do operationally different?
Or the recent "great feminisation" thesis. If you strip away the sexist culture war elements I think it can be boiled down to institutions with the aim of truth seeking vs conflict reduction, which feels like it might have something to it, but most criticisms of this come from looking at details which don't seem to support this mechanically.
> Spinning the same idea from the other side of the political spectrum it might be the Abundance argument from Ezra Klein - can you actually make concrete suggestions about what you would do operationally different?
There's lots of concrete suggestions if you actually listen to them. For example, ease up on zoning requirements and permitting requirements, ditch the Jones Act and tariffs, stop Everything Bageling and so on
For movies that work well as long as you don't think about the details, I've been using the term "Van Gogh narrative". (The Dark Knight being the main one.)
But I'm also of the opinion that an idea you can't define is one you don't understand, and it's dangerous to put any weight on it.
Round here the mantra is "make your beliefs pay rent in expected experiences". I suspect there's more to life but attention spans are short, if you want to sell an idea you probably do need to pin things down, unless you can express it in an art form.
Scott, what should someone do if their only good ideas come when they talk to their subminds that aren't the best at sanity? Does that say something about the world? Or is it an overdose of pride?
If the ideas are good who cares where they come from? I find that in order to think inventively about something I often need to let my mind get really flexible and reach into weird non-rational spaces. Afterwards I mess around with my haul and only keep the stuff that makes sense. But sometimes I end up keeping cleaned up versions of ideas that were ridiculous.
Thank you, I will try to manage the ideas better. I think it's possibly a lot of bad influences that I need to learn to separate out so I'm going to concentrate more on my family instead of the world.
Mad Coffee Science update: a few months ago I posted about a process I'd been developing for making coffee, using warm brewing temperatures (150-160ºF/65-70ºC) and agitation to approximate cold brew flavor profiles in a short amount of time. Several commenters (Aristocat, Performative Bafflement, Melvin, Metfora, and Fedaiken) chimed in to point out that I was using entirely the wrong type of beans for the desired flavor profile and recommended some brands and styles to try instead.
I bought a selection light-roast and medium-roast beans from Mayorga (corresponding to "Full City+" or less) and tried making them a few different ways. With those beans rather than the "French Roast" we had been using before, my method and a basic French Press technique produced very similar results, so it looks like almost all of what I'd been accomplishing was compensating for buying the wrong beans. Thank you to all you coffee nerds who suspected this and told me as much.
I am personally not a big coffee drinker, since the bitterness has always bothered me too much, and my target audience has been my wife who does drink coffee regularly. Working with light-roast beans, I actually kinda like the coffee, although I think I still prefer tea.
We also, after trying out the new beans, bought an espresso machine. As I understand it, espresso manipulates a different variable than the ones I'd been playing with, using pressure rather than agitation to get more of the desired flavor compounds out of the coffee very quickly. We have been liking it very much so far and have abandoned the agitation method we'd been using. We're still using the Mayorga beans.
Glad to hear you've found a better outcome! Tea is great! I'm just addicted to the rush of coffee lol
Ironically i used to have to have half and half and sugar to drink coffee, then I removed the sugar, then I removed the cream. I was very surprised to find that I came to like the bitterness. Frankly, its like i get used the bitterness so quickly that I start to taste all the other flavors.
I like the light roasts as stated before, but I'm sure their higher caffeine content may help my predilection for them!
About the bitterness of coffee: For me, coffee is right on the brink of being unpleasant, but gets kind of grandfathered in due to caffeine dependence. Oddly, my cats both hate the smell of it. Once I spilled a whole funnel of damp, used grounds on the floor and one of my cats used his paw to scrape the mess into a small centra area. The other one does a covering-up-poop routine on the kitchen counter if there’s some coffee spilled on it, piling more and more imaginary kitty litter onto it until the noise drives me crazy and I lift him off the counter and give him a distraction.
Always happy to see people trying out weird things with coffee, even if it doesn’t work out in the end. Definitely second (third? fourth?) using better coffee beans, something where you can taste the bean, not just the roast. If you’re used to dark roast and want to try something completely different, try a light “natural process” bean.
A little late to the show, but I wrote an analysis of the Pro Football Hall of Fame election and wanted to share it. I'm not a Pats fan but I'm irked by the process. I think most of the analysis erroneously supposes that people voted against Bill Belichick, but what happened was a likely outcome of people allocating scarce resources (votes).
---
“You know he got a chance he can win. White dudes like to do shit like that...vote for the wrong dude as a goof. They get drunk and shit and go like:
‘Let's vote for Jesse Jackson!’
‘I just voted for Jesse Jackson !’
And next day would be like this: ‘He f---in' won?’" – Eddie Murphy in Delirious, talking about Jesse Jackson running for President.
In POTUS elections, many people “vote their conscience”. Instead of supporting the crazy Republican, they vote for the Libertarian candidate. Or instead of supporting the crazy Democrat, they vote for the Green Party candidate. They know that their own vote won’t affect the outcome since the election is “safe” (or unwinnable) in their red or blue state, so might as well make a statement.
Or sometimes, as Mr. Murphy suggests, they’ll vote non-optimally for a goof.
So Bill Belichick didn’t get into the Pro Football Hall of Fame. Neither did Robert Kraft. Nor did two of the three Senior Players.
Background: A 50 person committee (mix of NFL people and media) has a slate of five finalists (each selected from separate semifinalist groups of 8-10): one Coach, one Contributor, and three Senior Players (last played no later than 1999; two of the three had their primes in the 1970s and the third in the 1980s, retiring after 1993 season). Each committee member votes for 3 of the 5. Anyone getting 80% is in (if no one gets 80%, the top vote getter is in). (The Hall specifies that it’s “top three getting 80%”, but it’s impossible for four candidates to each get 80%.) Last year, only one of the five was picked, and that was again the case this year.
Belichick had "spirit" issues, as he was directly implicated in one cheating scandal (Spygate) and was there for another (Deflategate), plus was innovative with finding loopholes in the rules. He was known to be gruff with the media. Other coaches or owners may have resented him. So that could explain some of the non-votes (at least 11 of the 50 did not vote for him).
So there are many possible reasons for not getting to 80%:
1. The Contributor was Bob Kraft, owner of the Patriots. One could imagine a voter saying, "i'm not going to use two of my three votes on candidates from one team." And some might actually think Kraft was more worthy, or was friends with Kraft but disliked Belichick and chose Kraft.
1a. Same logic, but “I’m not going to use two of my three votes on non-Players.”
2. It's not a given that the Senior Players will have another shot so voters may have prioritized voting for them. They might still be eligible but there are many other worthy candidates in the category. In fact, one voter has come forward and cited this factor One reason Bill Belichick isn’t first-ballot Hall of Famer. | Kansas City Star
3. Voters may have thought that Belichick was getting in regardless so they used their precious votes on other people they felt worthy. What’s it matter if he gets 94% or 96%? But bumping someone else from 78% to 80% is huge. The problem is when other voters act this way, too. (He f---in’ lost?!) I think this was true of some of the voters but no one is going to cop to that this year.
4. The method matters and the quality of the other candidates matters. This exact process was only put in place last year and resulted in only one of the five being elected. If all candidates are exactly equal, each one gets 60%, far from the 80% threshold. If there is one guarantee on the slate, then the other four only have a 50% starting point. (Conversely, if one of the five is clearly not qualified and no one votes for them, the baseline is 75%.) So it wouldn’t be surprising if the votes are split and one or even zero candidates are elected. If you had a Yes/No on each candidate, that would reduce this problem. (Voters might feel reluctant to flood the Hall with five new inductees so might vote No on the least qualified even if that candidate is above the bar.)
5. It's hard to compare across groups and eras. Was Bob Kraft a more impactful owner than Roger Craig was as a running back? How can you even start to quantify this? Even comparing players, was LC Greenwood the defensive lineman better than Ken Anderson the quarterback? Greenwood’s teams were more successful (Go Steelers!) but there are nine Steelers from that era (plus the coach and owner) in the Hall. There are two Bengals, one of whom came on after Anderson’s prime.
5. Then there’s jealousy and general dislike. Bill Polian of the long-time rival Colts was one of the voters and claims not to remember whether he voted for Belichick or not but was rumored to have spoken out against Belichick in the committee meeting (it’s a violation of the rules for anyone to reveal what happened in that meeting other than revealing their own votes if they wish). The Colts suffered often, occasionally unfairly, at the hands of the Patriots. To paraphrase Mencken, he couldn’t get his hand upon the collar of his superior, so he got his thumb into his eyes.
6. And, of course, the scandal(s). Belichick was implicated in Spygate (filming other team’s practices), which cost them a big fine and a top draft pick. He was coach during Deflategate which led to Tom Brady (who will certainly get in first ballot, though through a different voting process) being suspended for four games but Belichick was never himself implicated. And he never hesitated to exploit a loophole in the rules no matter how unfair it seemed. Plus he was a bit gruff and didn’t suffer fools in the media. But football has had an interesting relationship with rule-breakers, certainly when compared to baseball. Players have been suspended for steroid use without getting much flack (including possible GOAT Lawrence Taylor, who was first-ballot Hall inductee). (Take a look here List of suspensions in the NFL - Wikipedia and see how many of your favorite players are named.) Two Hall of Famers were even suspended for a year for gambling on football! I acknowledge that this should cost him some points, but his resume is just so far above the line that he’d had to go full OJ to fall below the line (and even then, I’m not sure; one of the Senior Player finalists last year killed his wife and himself but CTE had driven him mad). So football seems to be a little inconsistent on this one.
Then there are the ripple effects, at least if the current process holds. Mike Holmgrem (coach for the Packers and Seahawks) was a finalist last year but the Coach committee chose Belichick as its sole finalist this year. Presumably Belichick will be chosen again next year and get elected after serving his penance, so Holmgren and other top coaches Coach Blue-Ribbon Committee names 9 Semifinalists for Hall of Fame Class of 2026 | Pro Football Hall of Fame will get bumped. Additionally, next year’s Contributor and Senior Players will be on the ballot with a near-unanimous choice, leaving fewer available votes to get to 80%, leading to more of a backlog.
The scandals alone are not enough to explain the failure to elect, so most analyses which focus on just that are misguided. If they don’t fix the process, it’s just going to get worse.
Anyway, my thesis is that the reason for him not getting in is primarily the voting method, not the merits. Let's say you're creating a new all-sports Hall of Fame and your five candidates are Tom Brady, Babe Ruth, Michael Jordan, Wayne Gretzky, and Lionel Messi (or Pele, or Maradona). Each voter has to pick only three. I wouldn't be surprised if no one got 80% of the votes. I don't see how the Pro Football HoF continues to use this approach. Baseball has a similar problem with their Veterans Committee, where the players split the vote and often no one hits the threshold.
"Let's say you're creating a new all-sports Hall of Fame and your five candidates are Tom Brady, Babe Ruth, Michael Jordan, Wayne Gretzky, and Lionel Messi (or Pele, or Maradona). Each voter has to pick only three. I wouldn't be surprised..."
I'll just note that the baseball Hall of Fame did have this "too many worthy candidates" problem early on. First class was inducted in 1936. It included Babe Ruth, Honus Wagner, Ty Cobb, Walter Johnson and Christy Mathewson.
Omitted was ... Cy Young.
Not because he wasn't good enough but because of the rules there weren't enough open slots and he came in behind the first five guys. He got in later.
The reporting on this is annoyingly misleading. No one voted AGAINST any candidate unless their ballot had open slots that they didn't use. And I haven't read that this was the case here. Voters may have voted FOR other folks and then run out of votes. That isn't the same thing as a vote against.
He gets in at some point, but lets all not ignore the elephant in the room: The only thing B. Bellicose did that was truly superlative was stick his neck out for the greatest single player to ever do it.
Yes, the browns and win percentage and culture and bla bla bla, in sports you have an objective measure. If you count his rings minus Bradys rings, you got a big fat 0. If you count the wins minus bradys wins, he is sub .5.
He wasn't bad by any means, but he was replacement level. The thing is: the brady call was so fucking good and so not guaranteed I think he deserves to get in on that alone.
"If you count his rings minus Bradys rings, you got a big fat 0. If you count the wins minus bradys wins, he is sub .5."
If you count the rings of most NFL coaches with 3+ Super Bowl wins and then subtract the rings from the Hall of Fame QBs they had playing for them you tend to get a big fat 0.
Walsh - Montana is 0 (or -1).
Chuck Noll - Terry Bradshaw is 0.
Andy Reid - Patrick Mahomes is 0.
Maybe these guys don't belong in the Hall of Fame, but two of them are in and I expect that Andy Reid will be. I also expect that Belichick will be.
The only 3-time Super Bowl winning coach who did it with multiple QBs is Joe Gibbs. He did it with three non-Hof QBs: Joe Theismann, Doug Williams and Mark Rypien. I think there is a solid case to be made that Joe Gibbs is the best Super Bowl era head coach based on this alone, but I'm probably in the minority and winning a bunch of Super Bowls behind one QB does tend to get you in.
That's overstating things a bit, I'd say. (Disclaimer: I'm not a Pats fan but live here.) The defense was 1st in points allowed three times, 2nd another three times, top 10 all but six times, and never worse than 17th (they weren't quite as good in yards allowed). Special teams (as measured by DVOA) was above average every year from 1996-2020 under his rule. Brilliant defensive game plans twice against the Rams in SB to limit #1/#2 offense to small number of points (also as DC for Giants against Jim Kelly Bills, also #1 in points). 2/3 of his HC career (the beginning and the end) coincided with Brady so "minus Brady" doesn't mean as much as you think. (He is tied for most playoff wins as a Browns coach since 1990, however! But that's just one win.)
It occurred to me that every stable isotope of every element we know about (we have reason to believe nothing's stable over 100) seems to exist in the earth's crust - not counting stuff that's too light or unreactive to stay there and so it hangs out in the atmosphere. Or escapes, but we know it probably was in the earth at one point (e.g. helium).
It also occurred to me that this probably isn't something reliable throughout the universe. Such elements arise naturally from stellar explosions, which I believe aren't uniform - it's not the case that any star that explodes is going to create plenty of every stable isotope (right?).
To check this, I ended up asking ChatGPT. "Are there any stable isotopes not found in the earth's crust?" It did assert that there exist stable isotopes that are "essentially absent from the Earth's crust", and even supplied some reasons. As before: it's a noble gas (unreactive); it's very light (helium). It also suggested that some isotopes might not come from decay (no stellar explosion ever creates it), although it gave no examples; it also suggested some isotopes might sink into the mantle or core, or rise into space (fair, although "space" implies "very light", so ChatGPT is repeating itself somewhat).
It also gave examples. Other than the expected isotopes of helium, neon, and argon, it also gave tantalum-180 - a weird beast with a half-life of about 8 hours if made in a lab, but also occurs in nature and is _observed stable_. It doesn't go away. However, it's the rarest stable isotope we can find, which makes me wonder if it decays down to some baseline and then settles down. (ChatGPT didn't reveal much of this; it said "tantalum-180 is an oddity" and I found out the rest by looking it up specifically.)
Overall, while I expect obvious differences between chemistry in space and chemistry in an Earth-sized gravity well, the stuff that stays on Earth seems to exist roughly in the same quantity that it occurs in the solar system or between stars. In other words, stuff in the Milky Way seems to have shuffled enough that we're unlikely to see worlds where there's, say, practically no platinum. It seems every terrestrial planet will have basically everything, about as common or rare as on Earth.
Huh, interesting. 180m1^Ta is an isomer of 180^Ta. An isomer is an semi stable excited state of a nucleus. https://en.wikipedia.org/wiki/Nuclear_isomer I didn't know that there were isomers that were more stable than the low energy state. That's really neat.
I've known about that island for decades, yeah. I'm not a nuclear physicist, and wasn't quite obsessive enough as a kid to read more about why it's theorized, so I don't know if it's wishcasting, or based on known nuclear effects, or it's neutrino-style reverse theorizing, e.g. "if there _were_ such an island, here's what would have to be true, and what would be observable in experiments if so".
Even if it existed, I think about how much energy is pumped into heavy elements in order to produce stuff like tennessine, followed by a writeup reporting "we detected... seven atoms". I then bristle at the amount of energy that would be required to generate, say, a thimbleful of it.
Plus, tennessine is possibly _on_ that island... and its half-life is in the 22-51ms range. Not sure how much use we're gonna get out of that island, sadly.
I can just see it now, though - experiments producing a couple hundred atoms on that island suggest the existence of another island beyond that, and on we go, sailing atom-sized ships in search of a new archipelago.
While the elements involved are not exactly stable and the primary culprit is merely long lived, there is a significant counterexample to your notion that the galaxy is evenly mixed with all the elements distributed such that we will find them on Earth. There is a star known as Przybylski’s Star that has a spectrum showing exceptionally radioactive elements up to californium and einsteinum. These elements should not exist in detectable quantities unless there is a long lived source element that is decaying to produce them. Hence, there should be a superheavy long lived element that either does not exist on Earth or is such a rare siderophile that Earth's crust lacks a discoverable amount of it.
Clearly, that star is depleted in oxygen, iodine, and uranium.
Seriously, this is interesting; thanks for sharing it. I wonder how many other stars show that same abundance, and what specifically is giving rise to it.
As far as I know, Przybylski’s Star is the only star that has its particular blend of oddities. There is a subtype of stars known as Ap and Bp stars that levitate elements in their upper atmospheres, of which Przybylski's star is one of (even though by temperature Przybylski’s Star is F type). However, the rest of these stars do not show the radioactive elements that are associated with Przybylski’s Star.
My best guess for its formation is that Przybylski’s Star was originally part of a trinary star system with two closely orbiting neutron stars. They eventually collided and exploded, expelling vast quantities of neutron rich elements. Some of these elements accreted onto Przybylski’s Star. During the event the black hole would pick up a kick greater than the escape velocity of the trinary, leaving Przybylski’s Star behind.
Tantalum IS weird. The difference between the half-lives you mention is because you're mixing up Tantalum-180 (half-life ~8 hours, not particularly remarkable) with Tantalum-180m, called "metastable" but as you note, it has never been observed to decay, and has a measured half-life of at least 2.9×10^17 (290 quadrillion) years.
Yeah, I got that there's a 180 and a 180m, but I couldn't tell from the WP article what's materially different, and I was trying to dumb down even that difference for the readers by calling them "made in a lab" and "found in nature". And that's probably not as accurate.
I think that's a very poor gloss for the difference. 180m is an excited state of the nucleus, so a way to think of it might be as analogous to phosphorescence, but for nucleons instead of electrons.
So molecules can absorb energy (usually as a photon), enter an excited state (electron moves to an outer orbital), and then it usually immediately drops back down and releases that energy as light. This is fluorescence.
Under some conditions, for interesting reasons (see "forbidden transition"), the excited electron CAN'T drop back down to a lower state quickly, and is instead stuck there for much longer before relaxing to the ground state. This makes the excited state "metastable": there is a lower-energy state that can be reached, so you can't call it "stable", but the system can stay in this state longer than the term "unstable" would suggest.
So analogously, there is a ground state (regular Tantalum-180) to which the 180m isomer is supposed to relax to (and subsequently undergo radioactive decay), but that doesn't happen!
...and called their change "Improved article by simplifying technical content, removing primary sources, adding secondary reliable references, and enhancing clarity and neutrality on mechanistic interpretability".
I would call it "Removed everything that resembled math", but to each his own.
So be the change you want to see, and boldly (https://en.wikipedia.org/wiki/Wikipedia:Be_bold) restore it. If you are worried it will be controversial, leave a message on the talk page explaining your reasons.
But if I were, and if I did what you suggested, the likely outcome is that David Gerard (administrator of Wikipedia, known for his hostility towards the rationalist community) would immediately write to the talk page "he is one of *them*, so he has a conflict of interest" and revert my edits.
Also, in recent years, contributing to Wikipedia became much more difficult than it used to be. The kind of approach you suggest worked perfectly twenty years ago. I have created a few pages, just by writing what I knew at the moment... and maybe it was short and someone added a template "this article is a stub, someone please expand it", but later someone else expanded it and the template was removed, and the article was created.
These days, if you want to create an article, you need to create a draft in a separate namespace, write the draft of the article and submit it, and it will be judged by other editors whether it is good enough. And if something is missing, instead of publishing it with a request for improvement, the submission is simply rejected.
Example: https://en.wikipedia.org/w/index.php?title=Draft:B._B._Alston&oldid=1321253372 -- I know that this is low effort, but that's my point: 20 years ago it would be perfectly OK. I wrote some information, which is better than nothing (which is the current state), anyone else is free to expand on it. Now, I get rejected for "not showing significant coverage in reliable secondary sources". This is an author of popular stories for children that get translated and published on the opposite side of the planet, but is that good enough for Wikipedia? Ironically, Wikipedia already mentions him at https://en.wikipedia.org/wiki/Quick_Picks_for_Reluctant_Young_Adult_Readers which means that a list containing 100 popular books for kids is notable enough to be reproduced in its entirety, but the individual books or their authors are not. By the way, if you look at the edit history of the user who rejected the draft, you'll see that the history mostly consists of removing information from Wikipedia.
These days "deletionists" rule the Wikipedia. Writing content is low status. Removing or censoring it is high status.
> These days "deletionists" rule the Wikipedia. Writing content is low status. Removing or censoring it is high status.
"Had Wikipedia died in 2011, its achievements would have been immortal. Had it died in 2018, it would still have been a great site but flawed. But here we are in 2025. Alas, what can one say?" (from https://www.tracingwoodgrains.com/p/how-wikipedia-whitewashes-mao)
Well, my last Wikipedia adventure was an attempt to add a "some people think this topic might be a moral panic" that predictably got stampeded by the moral panic. So I don't really disagree.
Looks like David Gerard is no longer a Wikipedia administrator, though? His user page only mentions Rational Wiki, and his Wikipedia page says until 2025 [citation needed].
Almost twenty years ago, I went looking for an article about SLDNF - an algorithm used in logic programming. It didn't exist, so I did what research I had time for and wrote the first version. (That article is now generalized to SLD (Selected, Linear, Definite) - you can still see it today.)
Shortly after I published it, I received a private email from one of the other WPians, whose name I recognized from Talk pages on related logic programming articles, thanking me for having written it, and proceeding into a brief rant about how hard it was proving to submit similar articles due to the hawkish eyes of various other users with this or that axe to grind.
Axes to grind. About _logic programming_.
And again, this was almost twenty years ago. Wikipedia itself was only about 6 years old.
(I just searched for that email. Can't find it; possibly under a different email address. Really wish I'd saved it; it read like a weird one-off, and I didn't realize back then that it might be a signal of a wider trend.)
...as for Gerard, the most thorough source I know for what happened there is from fellow rat TracingWoodgrains. He wrote a piece on Gerard back in July 2024 (that also links to a lengthy earlier piece about Reliable Sources). There might be a later article, or a comment on the former one, that reports an update. https://www.tracingwoodgrains.com/p/reliable-sources-how-wikipedia-admin
So 420... people who live in a place where pot is practically legal, easy medical excuse or otherwise, what are your experiences? I live in a place where it is not, and I have two arguments against legalization. One is the awful smell, because a lot of people cannot be bothered to vape. The second is that the last thing people in 2026 need is an appetite enchancer. We really need appetite suppressants + exercise-boosters - maybe better legalize ephedrine. Why was actually ephedrine banned? Other than it being a meth precursor... but that's the kind of logic like closing lead mines if you don't want people shoot each other...
People forget how awful the status quo was 25 years ago. Peoples' entire lives were ruined over a freaking plant. I think a grudging acceptance (no ads, limits on THC content, rather thick taxes, enforcement on public use much like we do with drinking) is probably better than what we have in "420 friendly" states, but let's not forget that it could all end tomorrow: marijuana is illegal at the federal level and technically federal law is supposed to supercede state law, we are just ignoring that little inconvenience. So if you want to go back to the days of SWAT teams smashing down doors and shooting peoples' dogs because someone saw a weed plant through the windows, we could be back to that world uncomfortably quickly.
As a Seattleite, you get used to it. It's way better than a cloud of cigarette smoke. I am concerned about the health effects, but it seems likely to me that a switch from heavy street use of alcohol to heavy street use of THC would be a win for public safety.
I love the smell. But even if I didn’t, “some guy doesn’t like the smell” would be an insane reason to illegalize something, especially something in high demand that is relatively easy to transport & sell.
Unless…do you personally profit from black market activity? Crime? Violence? Widespread intergenerational antagonism between civilians and police? Then banning drugs might be something you’d like.
> But even if I didn’t, “some guy doesn’t like the smell” would be an insane reason to illegalize something, especially something in high demand that is relatively easy to transport & sell.
It's a very normal reason to illegalize consuming it in public, though. Think about "no durians" signs. Fruit is legal, and nobody wants to change that. But eating it in public is forbidden.
We mostly achieved the same thing with tobacco cigarettes; smoking them is legal as long as you're not in public, not inside an establishment open to members of the public, and, in the case where you want to smoke at home, you own your home outright.
Given how obvious it is that this can be done (we're doing it right now!), I don't really see why a smell of marijuana is viewed as some sort of necessary consequence of legalization.
My dad smoked medical marijuana for a long time, the secondhand smoke of which may or may not have affected my memory (his own memory was mostly gone by the end, but I think that was the blood problems messing with his head). But outside of smoking around kids, the only negative I've noticed is the smell. Some people go into the store and just leave a cloud of weed behind them, but are otherwise happy and pleasant to deal with.
Are you referring to it as an IQ suppressor because it reduces cognitive function while a person is high, or is there evidence for it lowering IQ in general, even while you are not actually inebriated?
I've heard of it having detrimental effects on overall IQ when a person uses it while their brain is in development, but I've not seen anything about this effect in an adult with a fully developed brain. Would certainly be interested to know if this is the case, though.
A lot of people say they have more insight, creativity, improved chess playing, ability to solve complex problems, etc. while stoned. I don’t know if they’d score higher on an iq test while stoned necessarily, but I can easily believe it improves cognition in some of these kinds of areas for some people.
I think set and setting probably have a lot to do with this.
I don't know what studies have been done but I think it's pretty generally accepted that regular pot use makes you dumber: bad memory, poor executive function, spacey. Basically the stoner stereotypes. I've gone through phases of pot use in my life and I can say unequivocally that it makes me significantly dumber even when not high. The deficit only goes away after several weeks clean. I used to listen to Loveline and Dr Drew would say that it typically takes 6 months for cognitive function to return to baseline.
My understanding is that if you smoke daily, your IQ and other cognitive measures are suppressed even when you're sober, and that it takes a few months of sobriety (or drastically reduced use) to return to baseline (though a few weeks will get you most of the way there).
I don’t think this is correct. Daily smokers develop a high tolerance. I know multiple-times-per day smokers who you’d never know they smoked if you didn’t see them do it.
Hmm I'm a pot smoker and beer drinker. One thing I learned when I went to 'Alcohol classes'*, was that while alcohol had a half life in your body of something like 6 hours. Pot (when smoked) only sticks around for ~2 hours. (Consumables are a whole different thing.)
*I got a DWI years ago and had to go to several classes telling me how to control it.
I smoke weed somewhat regularly and I would be completely in favor of a law regulating where people can smoke. I don't want to bother anyone with the smell (which I actually enjoy).
The appetite enhancing part, eh, I've yet to meet anyone who has gained significant weight due to weed
While you get hungrier while smoking it seems to get compensated by later appetite reduction - weed makes you feel hungry, but it doesn't *stop* you from feeling full.
While I am straight edge on that shit, I actually don't mind the smell at all. It's not like cigarettes, which are possible the worst smell you are likely to run into without making special effort.
It's mostly an aesthetic eyesore, as weed shops are some of the ugliest businesses in existence. It should be legal, but only sold by megacorporations with extremely bland branding.
The smell is awful, I avoid places I would otherwise like to go like the lovely park a block from my house.
I would prefer to keep it legal because I don't think people should be going to prison for it. But I would like to promote a social norm that smoking in certain places is unacceptable, much like we did with cigarettes. Bars & clubs, outside the front doors of restaurants, parks.
I'm in this exact same boat. I want it legalized; I also can't stand the smell. It makes me nauseous, much like my dad's reaction to cigarette smoke.
I suspected what Wanda Tinasky's paper seems to suggest: most people imprisoned officially for MJ were only there because the state had them to a more serious charge, but was willing to forgo that in exchange for an easier charge that still carried jail time. The idealist in me would prefer the books reflect the state's true objection to letting the accused go free, but I suspect (I don't have time to read the paper at the link) this current framework means saving a great deal of time and money required to prove the greater charge. If so, I recognize there's an important tradeoff being made here.
In a practical sense I think only a very very few ever did go to prison for it. People in jail for MJ possession are almost never there for simple possession: that's just what they pled to after e.g. being caught driving drunk with an unregistered handgun. The "no one should be in prison for having a joint in their pockets" trope is mostly a red herring.
The genie is out of the bottle now but I think legalizing it was a mistake.
"less than 1 percent of all State inmates had been convicted under a charge of only marijuana possession ... just 0.3 percent were first-time offenders."
>the last thing people in 2026 need is an appetite enchancer.
Well, some do and some don’t. I agree with you about banning ephedrine. And, yeah, the smell can sometimes be a bit of a pain. So are bus exhausts for that matter.
I have no problem with one, but I’m a nonstarter on the other two. Except: Standardized dosage is fine because it simplifies things. But only low dosage? Why not banish hard liquor? Weak beer only from now on. Pot is self administered. You can have one toke, or a quarter of a gummy; entirely up to the user. Pot is not worth creating another black market in this country.
I will say I did not use it during most of my working life (25–50). But as a 70 year-old with time on my hands and not much to do except think about things that I want to think about, I’m very happy to live in a place where pot is legal. Even for the years in which it was still illegal here in New York it was trivial to get it.
I live in a state where it's been practically legal for many years. I'm not a user myself but most of my friends are, and as far as I can tell they've universally switched to vaping THC products as opposed to using traditional bongs or joints. A few years ago I had a roommate who used quite heavily so I am aware of the smell, and said roommate did eventually have a brief stint in a psych ward for suicidal depression, though I would say weed was not the primary cause but merely an exacerbating substrate. That was just one person though, and none of my other friends have had anywhere near that kind of extreme reaction.
In spite of this, I'd say I'm still pro-legalization. My former roommate is already estranged from family and in difficult financial circumstances, and I can scarcely imagine how much worse it would have been had prison time been involved. I've also been to several parties where most of the attendees were vaping weed and those people appeared to be having the greatest of fun with little practical downside other than appetite and general long term worsening of memory. I don't want to deprive these people of something that improves their subjective experience.
I feel like the Android user interface is getting less user friendly year by year.
First, the constant changes: we have three Android phones in our family, each one is navigated differently. I feel helpless when my wife gives me her phone, she feels helpless when I give her mine. Recently, my mother-in-law's phone broke and she bought a new one... it took me an hour to just rearrange the five icons for her favorite applications on her main screen.
Things like "you have an application installed, but you do not have its icon on the screen, how to put it there" or "you want to remove this application's icon from the screen, but you do not want to uninstall it, only hide the icon" are done completely differently. The icons on the screen can be put in groups; good luck getting such icon *out* of the group, I guess you need to click on a very specific pixel on the screen, because I succeeded to do it in maybe 1 out of 10 attempts.
Second, the utter inscrutability of the controls, if you still can call them that way. Buttons and menus are out of fashion. Instead, you need to perform an undocumented magical movement with your fingers, and then something happens. If you put your finger somewhere and keep it there for three seconds, sometimes a secret menu opens, containing the thing you were looking for. If you swipe across the screen, another hidden menu appears. The exact details of course change: on one phone swiping your finger from the top of the screen to the bottom always opens the same menu; on another phone this leads to two different menus, depending on whether you did the swipe on the left side of the display, or the right side of it (once we even opened a third menu, but I have no idea how, we failed to replicate that).
The settings for what an application does could be in the application itself, on in the Settings application. If it's the latter, it could be at any place, like if it is about sound notifications, maybe you need to go to Sounds menu, or the Notifications menu, or maybe the Applications menu; and if the application is from Google, then maybe you need to go to Google menu first, and then it is either Sounds or Notifications or Applications or Settings or Extra Settings or god knows where.
Now of course the things that Google *wants* you to do (for example, to upload your private photos on their servers, under the pretense of backup) are trivial. Often you turn them on without actually wanting to, by clicking on a wrong pixel, or by a dialog appearing under your finger at the last millisecond, with the "I consent" checkbox already preselected, and of course clicking at almost any place on the screen closes the dialog and confirms your consent; there are only a few pixels that you have to click carefully to withdraw the consent you never gave in the first place. (Of course, there are no "yes" and "no" options; there is only "yes" and "not today, but keep nagging me tomorrow".) But if you want to do something Google *doesn't* want you to do (such as turn off uploading your private pictures on their servers, and remove the ones that were already uploaded when you clicked the wrong pixel), the only way to do that is to find a tutorial on YouTube that will carefully navigate you through the obscure menus, submenus, options, and gotchas.
Third, authentication. Long gone are the day when you could provide a username and password, and get an access to the service. For example, your current authentication options for Gmail (which already knows that it's you, by the way; you are already logged in to your Google account, but for some reason Gmail requires extra confirmation when you run it for the first time on your new phone) are either to send you a notification (does it mean SMS? something else? no idea. anyway, when I click on "send the notification now", nothing happens), or it tells you to navigate to a secret submenu in Settings, request a 12-digit code, and then type it into Gmail (the secret submenu actually shows you two 12-digit codes, apparently randomly generated each time, and it doesn't matter which one you type into Gmail, it always says "wrong code").
I spent my entire life using computers, and I have an Android phone for a decade, and I am completely lost when I try to do something on a newer phone. How is my 80 years old mother in law supposed to use it, I have no idea.
Is there some kind of "hallway testing" (when you grab a random person on the hallway and check if they can figure out how to do something in your user interface) ever done in Google? Or is it just "screw them, if they don't want to pay for Apple, they don't have an alternative anyway"?
I hear you and sympathize with everything you're saying. It's very much a "there should be a market based solution to this, and there isn't" sort of thing.
If you're tech savvy, you can do what I do, and get unlocked bootloader phones you can put custom ROM's on, then everything is the same for every phone you do that on. Just find one that everyone likes and stick with it.
> Third, authentication. Long gone are the day when you could provide a username and password, and get an access to the service.
OTP auths are so infuriatingly broken, and spreading like cancer everywhere, it really distresses me.
Like I would gladly pay some pretty startling amounts of money (well into the 5 figures) to never have to deal with them again. Is there any option like that anywhere? Does any company in the entire world offer to solve this pain point? Nope, the never-ending spread continues, with OTP rolling out to more and more of the entire world, forcing you to give up either privacy or get and track decoy phone numbers with meticulous record keeping.
I also want to pay a 5 or 6 figure "I'm not a moron" bond+insurance so I can skip any fake "cover your ass" liability paperwork you're forced to sign at every business everywhere, because those are spreading like cancer too. If I slip up and injure myself at a climbing gym or eating tacos or whatever? Comes out of my umbrella policy, business is covered.
The worst part is, all of that fake CYA paperwork literally does nothing - they're just immolating thousands of customer-hours purely in the service of entropy. When some idiot doesn't clip in on a self-belay wall and jumps off at the top? Guess what, no more self-belay in any climbing gym in the state, all that paperwork and thousands of wasted customer hours every year for decades was for nothing, purely in the service of entropy and waste.
Is everyone in your family using the same brand of Android phone? e.g. All Google Pixels? Cause different brands customize Android a *lot*, with Samsung being the worst offender IMO. I'll be moving back to Pixel or Moto after my current Samsung is retired because so many of the latter's UI choices are terrible.
tl;dr It's not *just* Android changing that's the problem.
> Is everyone in your family using the same brand of Android phone?
No they don't. I always choose the cheapest one (happened to be Motorola the last time I bought a phone), and it only has the basic functionality and I am happy with it. Others try to choose something more fancy, and somehow end up with tons of Chinese crap installed.
On Caplan's no right to your culture: what is your culture: your neighborhood, your town or your country? Because I can understand if people do not want a mosque to go up in their neighborhood or town, but not their whole country. I can understand that, because they want to co-own with like-minded people the public space around them. Life feels so much nicer when you do not only own your home, but your street or hood is also co-owned, that you just don't see suspicious faces there, and I don't mean it in a racist way, rather just few strangers and them all invited by a local. But this co-owned space does not need to be large, just the size of an average church says 250 people. So it is still okay to build a mosque one kilometer away.
Technically you are not allowed to discriminate, but I think in practice white people could easily make an agreement with non-white people that on the small local level discrimination is okay, they too want to co-own their spaces, they don't want their affordable neighborhoods gentrified, they don't need our Karens who object to people playing music at a Saturday afternoon and so on.
A country is just a bunch of neighborhoods though! Caplan's point is that your preferences in your comment ARE coercion, you are basically saying "I want to control where other people can go and what they can do." In the literal sense of "I do not want to see this kind of person on my street" -- sorry, you don't own your street! Anyone can walk or drive down it! That's what freedom is all about!
An individual does not own a street but collectively the citizens do. And citizens decide things by political procedure.
But Caplan as a libertarian is allergic to political procedures hence this stupid strawman.
Ask him if citizens collectively own the national territory or not. If not, how is national territory treated in his libertarianism. I think this question is never asked by Caplan, never mind answered,
If the citizens are deemed to collectively own public property, and if the citizens are entitled to regulate/dispose of their property however they like, then what's the objection to any sort of tyrrany?
What do you mean by tyranny in this context? One could theoretically rule as a tyrant and still ensure the citizens have perfect access to public property.
I'm not trying to be too specific about what sort of tyranny I'm thinking of (though preventing Jews from entering certain neighborhoods would probably qualify). I'm just arguing to Gian that, if your conception of democracy is that the citizens collectively are *owners* of the commons and may dispose of their property however they please, you can't really object to anything the government does, as long as it has democratic support. In particular, there are many kinds of tyrannical governance that you can no longer object to.
Individual property rights won't protect you, even in your own home, because the government will argue (correctly) that they hold allodial title over all land.
Point being, I think we need a notion of liberty that rejects the majority constraining the rights of minority groups without sound reasons (eg., if you limit freedom of movement for violent criminals, that would be less of a violation of liberty than if you limit freedom of movement of Jews or drag queens, and the limits being democratically decided doesn't make that much difference).
I think he did answer it in another book, a comic, about how open borders are a great idea. Which would mean that he believes nation-states shouldn't exist.
Reality is not scale invariant. So no, a country is *not* reducible to neighborhoods. A culture is not reducible to individuals. Scale matters.
Do I agree with local discrimination? Meh. Not really. But do I think it's inevitable? Sure. People who think alike live together. And that naturally excludes (or makes feel less welcome) those who think or act differently. In *all* directions.
This is an absolutely ridiculous definition of freedom because if I am free to stand in the middle of the street, then this either implies that either drivers of cars are also free to run me over or that drivers of cars are now not free to use the road.
Seems like these are the kinds of situations where a "reasonable person" standard becomes relevant. In the real world, it's possible for a motorist and a pedestrian to encounter one another on a quiet residential street and negotiate safe passage for both of them. Even if one or both of them refuses to give way, there are reasonable limits to acceptable actions that either of them can take in response.
But you can collectively own a street. If said collective didn't want certain people to move in, what right do the libertarians have to object? It's private property.
If you own the land, and the street, then sure! You have every right to stop someone from moving into your driveway, for example. Just don't ask me to plow it, resurface it, or enforce traffic laws on it with my tax money.
I live in a neighborhood that used to have a restrictive covenant against Jews; I'm not convinced that it would be a better neighborhood if that covenant were still legal and in effect.
Is that because of the concept of the restrictive covenant, or because you don't find the target of this particular covenant objectionable? What if the rule was that you can't use a house in that neighborhood as a brothel? Would you have the same objection?
I think whores and Jews are such radically different kinds of category, in ways *extremely* relevant to whether or not certain restrictions can justly be put on them, that I wonder if you're being serious. It's impossible to un-Jew yourself.
You can un-jew yourself any time you want. There are any number of religions that will welcome your conversion. It's actually easier than un-whoring yourself because you don't even lose your income.
Was walking my nephew's tight neighborhood the other day, and one house was blasting Mariachi music in the backyard. We could hear it at the end of the block and I'm thinking, "Dang, I wouldn't want to live within a 1/2 mile of them."
Similarly, if you have anybody smoking pot outdoors in your neighborhood, everybody else who lives around them will be breathing it too.
There is no neighborhood in the world where everybody likes what goes on. I have a friend who lives in a high class town nearby where the HOA's are like Nazis. Yep, it's super nice all the time, but you'd better be prepared for the strictness.
I think the standard libertarian solution is that if you want to impose your values on the entire city, you better make enough money so that you can *buy* the city, and then you can make the rules. Which in theory you could achieve by coordinating with other million people who want the same kind of city, but in practice, good luck coordinating that. Also, if one person in the middle of the city refuses to sell to you, and instead sells to someone who builds a really loud mosque there, or maybe even a small nuclear testing site, you are screwed.
I guess in libertopia, democracy is valid only if you join it explicitly.
If there is a city that says "if you buy a house here, you agree to follow the laws that the citizens passed democratically" and it's a part of your contract, and you sign that voluntarily when you buy the house, okay.
But if you already own a house in an existing city, and you didn't sign such contract when you bought it, they can't make you. (Well, in practice they can, but the libertarians would object.)
So in libertopia, you need to start with building a new city, and then selling the houses with a contract that says that you will follow the rules, and that you can only sell the house to a third party if they sign the same contract.
The existing cities... well, you could try to buy them house by house, hoping that one day you will own it all, and then you can enforce your rules... which obviously wouldn't happen, because as soon as you own 90% of the city and your intentions become clear, the remaining owners would increase their prices dramatically, knowing that their houses are precious hostages. Or you can accept that only 90% of the city will follow your culture, and hope that maybe some day in the future you will succeed to get the remaining 10%, or that the people who want other cultures will avoid a city that is 90% resistant against them.
An entire city is too much, I would try the 250 to 1000 people level, that is easier. Imagine it is not the free state project so it is not ran by people who are less able to organize things than bears :) first find a place already tending like that. I am sort of a culturally Catholic agnostic, so any place with a nice old church with colored windows will do. Get some kind of a mutual building society thing going. And so on. It is all doable, my plain question is what is the book really about, that people should not even try preserve the character of even a neighborhood, or that they should not object to people of different cultures living in their countries somewhere near them?
Yeah, the standard smaller-scale solution is a walled compound. And you will need an armored car to commute to work. (Your kids will be homeschooled, of course.)
Eh, I am making fun of the libertarians, but maybe in a society that actually works this way, there would be established convenient solutions, something like Kickstarter, where you could specify the kind of community you want to live in, and it would find you the right kind of neighbors, and when you all put the money in escrow, it would build the neighborhood. But still we do not have these kinds of solutions here and now. Again, the standard libertarian solution would be "then build it, and make a few millions or billions in the process", but I assume that you want to focus your life on doing something else, you just want a nice place to live.
Why do other people catch AI making so many mistakes than I do? I use Gemini free for everything, and it generally works, it makes a good gym plan, meal plan, found out much of the health issues from my alcohol abuse are mitigated by high-dosing B1 and magnesium, it answers all my history questions, it finds me a psychoanalysis of BDSM that matches my exact experiences, it turns a random chat into a documentation, it makes great SQL queries that automate a lot of complicated work, it can even get into black magic level stuff like CROSS APPLY, it even found an algorithm for calculating when is Easter when I wanted to compare working days across years, and it could even explain to me how information entropy relates to thermodynamic entropy. Is it possible that people use it as a google, with short simple questions, instead of a long discussion?
Also, other free AI to try? Free Claude seems worse than Gemini.
Guarding our boundaries is exhausting, and it feels like a big relief to surrender sometimes. But that paradoxically requires a partner who is both completely reliable and yet a little aggressive, because true surrender requires an actual "attacker". This was written in books in th 1970's but I forgot the author.
Ah thank you, that sounds very plausible! Makes me wonder why only an (estimated) fifth of the people is into it, and it seems impossible to change one's preference (those who like it can't make themselves stop liking it, and those who don't cannot learn to like it [or if they can, in fact they had been liking it all along?])
I think Scott wrote somewhere here that perhaps sexual fetishes are some "wires crossing" in the brain, like in the homunculus the feet are right next to the genitals, so it's easy to miswire a foot fetish.
I always thought it's simply some funny wiring in the brain, and all those reasons "why it feels good" are a posteriori rationalizations. Nevertheless, this reason sounds very logical.
Well, sorry for derailing, you were asking about something completely else...
I'm quite skeptical of AI, which may make a difference. Since I expect it to make mistakes I end up double checking everything it gives me, so naturally I catch every mistake it makes.
As far as how I use it: the rare times I use it I am usually using it like Google (although, usually I just Google it instead of asking an AI). The only other times I use it are to compile information into a written document. You know, "Take the information in the attached document and make a 500-1,500 word report on it in the style of a business analyst" or something like that. Saves me a bit of time, though I do have to double check all the facts to make sure it didn't hallucinate anything.
Generally I get errors from AI when trying to do something extremely specific where there is one and only one correct response.
The SQL stuff you're doing doesn't sound that hard, tbh, and describing CROSS APPLY as "black magic" tells me that you're working mostly at the level of solved problems that have extensive solutions and solutions to related issues in the training data.
Yes, this business software field is extremely repetitive. Every small business owner ever only wants to know how much money they earned per various dimensions. This is why Excel Pivot tables were such a hit, one could just shovel the transactions in and it supplied what was needed.
Using Gemini/ChatGPT/Claude for legal research, I found that all of them frequently cited nonexistent or irrelevant cases. I'd say about half the time I asked it for a summary of a case, upon reading the actual case myself, I'd find that the AI's summary was misleading or entirely wrong. However, some fraction of the cases they cited did actually exist, and were relevant to me, so the models were absolutely helpful overall (to me, as someone with no legal background) in preparing a defense, even accounting for all the irrelevant cases that ended up being time wasters.
I often go to ChatGPT with Bible study questions, with highly variable results. Sometimes it will spit out a useful analysis of a difficult verse; other times it fails to even produce the correct verse. Much of the variation, I suspect, comes down to which version of the model handles my query.
Asking Gemini and other models about software build issues, I've consistently found them to be worse than useless. They'll routinely describe a nonexistent Autotools feature/option, and then double down and claim that the feature exists but isn't documented, or some other nonsense, and I'll waste 10 minutes going down a fake rabbit hole.
I am driven to incandescent rage by people asking chat bots spiritual questions. I was with someone who was using gpt to get suggested meditations for doing the stations, and I was yelling MOTHER FUCKER THIS ISN'T A FUCKING PRAYER WHEEL! YOU GOTTA DO THAT SHIT YOURSELF OR IT DOESN'T COUNT! in my head the whole time.
Makes me wanna move inna woods and live up a tree like an og hermit.
It's actually a pretty optimal use case, if done responsibly. In the case of Christianity, LLM training data likely contain thousands of Bible translations. For every word in the Bible, millions of words of commentary and analysis have been written, and much of this is likely in the training data as well.
The key to getting useful results is to always be explicit about what perspective you want to hear, e.g., "What was Barth's view on _____?". Without any grounding, I've noticed that ChatGPT tends to default to theologically conservative positions on Bible-related questions.
I want you to know that I am now yelling MOTHER FUCKER THIS ISN'T A FUCKING PRAYER WHEEL! YOU GOTTA DO THAT SHIT YOURSELF OR IT DOESN'T COUNT! in my head at you now.
Outsourcing your theology and spiritual development to a soulless machine is allowed, but it is also something they would write about in 40k or trench crusade as a joke.
As with any other domain, I think it all comes down to how carefully the LLM is used. Asking a question to an LLM, accepting the answer uncritically, and moving on is indefensible regardless of the topic. Asking a question, relating the answer to what you already understand, accepting what seems right (but being very careful not to update your beliefs too far without looking at a primary source), and pausing to look into any claims that seems off - I think you're unlikely to internalize any wrong information with this approach. Or better yet, ask a question where you already have a hunch what the correct answer is, but you're not quite sure - if the LLM affirms what you were expecting, update your certainty by 10% and move on; otherwise, look into the matter further.
I don't think what they say is any less meaningful than what the Church or local preacher has to say about religion. Whether any of it is meaningful is up to you, but they are all just repeating the same talking points. Nothing uniquely bad about AI.
Asking a chatbot "Tell me about God" seems pretty bad, but it's definitely not valueless.
Ex: "Sufism seems a lot like neoplatonic strains of Christianity, which primary works should I read to see the most similar parts of each"?
Or like "I've seen different translations of Luke 17:21, help me understand how the Greek words were used in actual practice at the time they were written".
Maybe those don't count as spiritual questions for you, though.
It depends on the purpose of the question. Are you doing homework? Are you idly taking in facts that you will forget soon after?
Go nuts. If you don't care, I can get mad at you for not caring but that is seperate from getting mad for not trying.
Are you writing a book? Are you practicing your religion? Do you really, really care? In that case, we enter familiar territory: MOTHER FUCKER THIS ISN'T A FUCKING PRAYER WHEEL! YOU GOTTA DO THAT SHIT YOURSELF OR IT DOESN'T COUNT!
I am remarkably anti a machine doing my thinking for me though. If Butler is out there struggling, he needs to let me know is what I'm getting at.
I don't even think that the current dev. path will lead to agi or some shit, I would like it more if it would! I just hate talking to someone who then talks to the chat bot and looks at me like a dog that just shit on the carpet and expects me to be impressed.
How qualified are you to evaluate whether it’s making mistakes in all these areas? LLMs are really good at producing correct-sounding responses, so if you’re not an expert you may not catch them.
That said, they seem less mistake-prone now compared to a year ago.
I haven't tried it recently, but in my experiments, AIs failed to explain mathematics, specifically set theory. I guess it is a combination of (1) fewer people talk about it, so there is less text to learn from, and (2) the details in math are extremely important, it is not enough to get things approximately right.
More precisely, an AI could explain things more or less okay, but when asked for a specific proof, or to answer whether something is true or false, it did things like subtly changing a definition of a term in the middle of a proof.
For example, it proved something using "infinite sequences", but at the beginning of the proof, the words meant "sequences that are infinitely long (regardless of what they contain)", and at the end of the proof the words meant "sequences that contain infinite values (regardless of their length)". Both the first half and the second half of the proof looked correct; you had to pay attention to notice that the definition has changed.
It was funny when I asked the same question in two different browser tabs, and in one the AI proved the theorem, and in the other the AI found a counter-example. Upon closer inspection, *both* the proof and the counter-example were bullshit.
But I expect that if you ask something from high-school math, you mostly get solid results.
Oh my, I can actually load this page without massive performance issues! Gave up on the Commentariat a while back, maybe I'll participate or at least read a bit again.
> I think most of our useful social norms exist through a combination of divine grace and reciprocal communitarianism. To some degree they arise spontaneously and are preserved by the honor system. To another degree, they are stronger or weaker in different groups, and the groups that enforce them are so much more pleasant than the groups that don’t that people are willing to go along.
> The norm against malicious lies follows this pattern. Politicians lie, but not too much. Take the top story on Politifact Fact Check today. Some Republican claimed his supposedly-maverick Democratic opponent actually voted with Obama’s economic policies 97 percent of the time. Fact Check explains that the statistic used was actually for all votes, not just economic votes, and that members of Congress typically have to have >90% agreement with their president because of the way partisan politics work. So it’s a lie, and is properly listed as one. But it’s a lie based on slightly misinterpreting a real statistic. He didn’t just totally make up a number. He didn’t even just make up something else, like “My opponent personally helped design most of Obama’s legislation”.
...
> This would have been a more believable claim than the one he made. Because he showed his work, it was easy for me to debunk it. If he had just said it was in some obscure report, I wouldn’t have gone through the trouble. So why did he go the harder route?
> People know lying is wrong. They know if they lied they would be punished. More spontaneous social order miraculous divine grace. And so they want to hedge their bets, be able to say “Well, I didn’t exactly lie, per se.”
> And this is good! We want to make it politically unacceptable to have people say that Jews bake the blood of Christian children into their matzah. Now we build on that success.
In 2026, it seems like we have gone off the deep end. Certain politicians lie with reckless abandon -- not 'lie but if you squint its a misinterpretation', but rather 'this is a bald faced lie by any count'. See, for e.g., the immediate claims that Alex Pretti was a terrorist. Or, more mundane but perhaps more important, Trump continuing to claim that the economy is the bestest greatest economy in the whole wide universe.
A few misc thoughts and questions I've been pondering.
1) What the fuck happened? Scott posits that there are situations where suddenly entire groups 'become honest'. Is the opposite also possible? Can groups suddenly become liars all at once?
2) How deep has this sunk into the culture? A few folks have mentioned that an entire generation has basically been raised in a political environment characterized by some level of blatant lying. Have we screwed up this norm permanently? Is there a way to bring it back?
3) Which direction does causality go? Did we lose the norm and then get the circus, or get the circus and lose the norm?
I have a theory here that if Republicans are getting called liars by biased liberal "fact checkers" over technicalities like the example cited, then strategically it makes sense to just lie. What is the advantage of making an effort to stay very close to the truth if you're just going to be called a liar anyway?
Because lies mostly hurt your own side. The other side won't believe them anyway, but your followers will, and then they'll get confused and mad when reality fails to live up to their expectations.
> Fact Check explains that the statistic used was actually for all votes, not just economic votes, and that members of Congress typically have to have >90% agreement with their president because of the way partisan politics work. So it’s a lie, and is properly listed as one. But it’s a lie based on slightly misinterpreting a real statistic. He didn’t just totally make up a number.
"Why did lying become acceptable?" would imply a false premise - that lying was ever unacceptable. A brief survey of the incentive model should reveal that lying is always lucrative within certain limits. Poker players are among the most experienced in knowing precisely when lying does and does not work.
If some groups become spontaneous truthtellers, it's probably because reality is a natural limit on lying, and occasionally reality manifests on something that people (1) need in order to make important decisions and (2) had previously been assuming worked in some way it didn't. Usually, people either naturally conform to reality because it affects them directly (you can't pretend you're making soup by boiling air, but you can pretend indefinitely that erbium doesn't exist), or they already believed what was true all along and so there was no opportunity for "spontaneous truthtelling" to occur.
If that's the case, then some groups can tell lies indefinitely if they're capable of insulating themselves from important decisions for long periods of time. Specifically, if their immediate biological needs are met, they're free to think more abstractly, which enables them to think less about reality. This might be fine if biology's simply never going to play a role in their direct lives until the end of each respective one. OTOH, such people can share their ideas with people who are a little closer to reality, giving them a little tug away from it. The latter people share those ideas in turn, until real tension exists between stuff coming from the abstract thinkers and the people who have to grow actual food, build actual houses, and raise actual children.
If you truly want to go back to reality, it might depend on how high you've climbed into the abstract. You might not even realize how far up you are - the previous decade appears to have been coopted by an entire school of thought that not only attacked the very foundations of logic in order to try to redesign it in terms of power disparity, but also poisoned the well against the older system, so that its own converts would believe that _they_ were the reality-oriented. So for any value of you, you can't even start with the premise that you're the good guys.
The surest way to know now is probably to ask whether you can grow your own food and build your own house, and if not, how close you are to being able.
(If this sounds vague and darkly hinting, I apologize, but it's hard to help someone with a problem if they might be actively invested in prolonging it without realizing.)
> If this sounds vague and darkly hinting, I apologize, but it's hard to help someone with a problem if they might be actively invested in prolonging it without realizing
I think I have a good sense of where your politics lie, but I'm fairly certain you don't really get my politics very well. It's neither here nor there.
> A brief survey of the incentive model should reveal that lying is always lucrative within certain limits...If that's the case, then some groups can tell lies indefinitely if they're capable of insulating themselves from important decisions for long periods of time
I think this misses my ask. Scott writes extensively about how, even though it may be very lucrative to do so, people do not outright lie. They may misrepresent, omit, selectively mishear, but very rarely do people just make up numbers. He wrote about this in the linked post above and several times afterwards.
Quoting again, more selectively: "This would have been a more believable claim than the one he made. Because he showed his work, it was easy for me to debunk it. If he had just said it was in some obscure report, I wouldn’t have gone through the trouble. So why did he go the harder route?"
This is the critical piece. There were always people who could insulate themselves from consequences, but they did not lie outright despite their inability to do whatever trad-coded things you think are important. If your response is something like "Scott is wrong, pols and journos make shit up all the time, the person who showed his work in the example above is a one-off and this isn't a real phenomenon" then we are just having a different discussion. If you agree that this is a real phenomenon, I'm interested in understanding why, and why now.
> I think I have a good sense of where your politics lie, but I'm fairly certain you don't really get my politics very well. It's neither here nor there.
I might not get your politics, true. My sense of your politics is driven by the arguments I've seen you make here in the past, and I don't follow you closely enough to say I've seen all of them. Specifically, here, though, I think I have a good guess at who you're referring to as "certain politicians" who "lie with reckless abandon", and while I can readily concede on what I'm guessing the referent is, I suspect you're conspicuously leaving out critical examples. The rest of your politics, whatever they may be, can be left aside for this.
Moreover, you're not the first person to make this general lament about blatant lying, and not the first I've had exchanges with. Past experience tells me I have to be very delicate with how I couch my answer. I can't just flat say what I think is going on; if I do, you'll interpret it as motivated (cf. your opening remark about "my politics") and won't take it seriously. You might not take it seriously regardless, which means I'm really just talking to anyone else happening by and reading this. Either way, I have to be careful, and not make assertions resting on premises that only one side of the political aisle holds.
I get that you're asking "why so blatantly, and why now". I did respond to that: it "would imply a false premise - that lying was ever unacceptable". I then went into a brief discussion of how lying was always a tool in the belt; there are always cases where it's applicable. After that, yes, I went into a brief exploration of your question (1): "Scott posits that there are situations where suddenly entire groups 'become honest' [...] Can groups suddenly become liars all at once?". If that's not the main question you wanted answered, I apologize - but you did label it #1.
Plus, you had three sets of questions. That was going to require a lot of response, so I prioritized.
To answer "why so blatant, and why now", though, one kind of needs the underlying background on when and where lying is applicable, so my response still needed to be written.
As I said, groups can lie to each other about things that aren't that important to them. Moreover, what's literally said doesn't always matter as much as what's implied. If you and I are lifelong friends, say, with a history of chatting, there are things I could say that would appear false to an outsider, but you know what I really mean, what I'm really trying to say, and it'd be a reaffirmation of friendship. What I literally said doesn't matter. It might piss off outsiders, and *that's* what matters.
And that might be what's mattering here.
That's one possible explanation for what you're seeing. Hard for me to say for sure without more detail. And if so, there are multiple reasons for why one might do that, and multiple ways to misinterpret those reasons, and it's its own entire can of worms.
What a load of hogwash. The problem with this is that when Trump lies it’s not some sort of in joke between the likeminded to piss off common ‘outsiders’.
They are simply the lies of a man with no moral core that true believers lap up as truth.
BTW
turgid != lucid
You are a theatre guy, surely you are familiar with the line “Brevity is the soul of wit.
> As I said, groups can lie to each other about things that aren't that important to them. Moreover, what's literally said doesn't always matter as much as what's implied. If you and I are lifelong friends, say, with a history of chatting, there are things I could say that would appear false to an outsider, but you know what I really mean, what I'm really trying to say, and it'd be a reaffirmation of friendship. What I literally said doesn't matter. It might piss off outsiders, and *that's* what matters.
To restate back to you, the claim is
"The reason there is a marked shift in perceived incidence of pols saying outright lies is because they care less now about precision than they did to the past, and that in turn is because they are purposely trying to anger 'outsiders'"
If that's right, I agree that would explain what is going on, but framing it as an intentional strategic choice is...pretty grim
"The reason there is a marked shift in perceived incidence of pols saying outright lies is because they care less now about precision than they did to the past, and that in turn is because they are purposely trying to anger 'outsiders'"
To refine on this: caring less now about precision implies precision is always important. In truth, it probably isn't as important as your restatement implies.
It depends on what's being said. If you're a farmer and I'm the local town guy in charge of running things and I tell you what day is going to be best for planting the spring crop, that's probably very important to you and you're going to care whether I said Monday or Tuesday. If, OTOH, I say "Monsanto's going to screw us again like they do every year", and meanwhile there's ten or so hard-working trained biochemists in town who owe Monsanto their livelihoods and take offense and want everyone to know that my remark is wrong on the details, you and your fellow farmers probably only vaguely care about whether Monsanto really is doing something this time, and much more about the fact that if you're ever in a fight with Monsanto, I'm going to be on your side.
This is the sense I'm getting at with blatant lying: everyone has plenty to worry about, and prioritizes information that addresses their direct concerns and cares little about information that doesn't. So anyone on their side can lie about the latter stuff - everyone there knows it's a lie, and it's not important (to them). If it pokes the other side in the eye, even better; the unspoken implication about one's commitment is more important than whatever blah blah details.
I agree with your assertion that it's grim - it implies the conflict theory perspective is growing stronger, and I think that's bad. That said: so what? Those of us who agree also contain a large subset of people who still can't get away from that mindset and think that the solution is to get "the other side" to stop lying so much. That's not only the wrong solution, but it's feeding the problem. In reality, _both_ sides are lying... about details their own side doesn't care about as much. And telling truth about details they _do_ care about (and getting angry when the other side gets them wrong).
> If that's right, I agree that would explain what is going on, but framing it as an intentional strategic choice is...pretty grim
If lies are so powerful, then the only thing that can destroy a lie is a better one. Regardless of the honesty of the administration, they are going out of their way to destroy existing ones, like the lie that all people are equals, and that they have "God-given" rights.
I would be interested in someone working hard to develop a data-based analysis, but my perception is that:
(a) like most things 2026, the seeds and sprouts go back a long way
(b) the strength of this anti-lying/exaggeration norm wasn't really that strong and appears stronger through a nostalgic filter
One major reference point for both is the info from the Pentagon Papers and the conduct of the Vietnam War. My read of that evidence is that government sources consistently, knowingly lied to the US public about the prospects for the war for essentially the entire duration of the conflict. Iraq II was similarly filled with falsehoods, especially the reasons for the war and the level of planning for the post-war management of Iraq.
Reagan had enough deception and sleaze to support the nickname
'the Teflon president,' with Iran-Contra one of the most memorable. Clinton ... famously slippery with the truth. GW Bush: 'Mission Accomplished' and 'Brownie is doing a heck of a job' (on top of the Iraq II stuff...)
Here is a paper built around an analysis of a 1991-92 Congressional scandal:
For a silly example: one of the George Washington stories we told a lot was the cherry tree myth. It is a lie about not lying! Note: I don't recall my own children hearing this in school, so maybe this myth has been expunged?
At the same time, I empathize with your frustration and the sense that this has gotten worse. I just think it is a change in intensity and not a sudden break that needs to be explained.
Coordination on truth is easier than coordination on lies, because there is one possible truth but a thousand possible lies? Schelling points and all that. Scott does not seem to get that honesty is just Schelling points. It is so hard to keep a log of what you lied to whom when. The basic Schelling point is just telling the truth unless doing so would have massive consequences i.e. getting arrested.
Turns out that politicians can just utter whatever combination of syllables they feel like excreting from their mouth-holes in the moment, no-one really cares that their utterances fail to correspond to either reality or their previous utterances, and there are no negative consequences whatsoever to this strategy so long as the noises being spouted appeal to enough of the audience.
Theahura, this is what a color revolution looks like from the inside. As the pendulum reaches its turning point a little added force in each direction, left and right, soon has the sheep utterly discombobulated and attacking each other.
Those who have enough intelligence to be neither blue or red can either get popcorn and enjoy the show or be crushed by both sides.
Eventually the pendulum will loop the loop and reset itself, not long now I think..
Theahura, I apologise for my poor communication skills. An outside force has been using media influence to amplify any story that divides Americans and mute any story that unites them. Over time the two main forces, Republican and Democrat, drift further and further apart. Shakespeare wrote that a house divided against itself will not stand, meaning that when a great challenge comes along requiring a united effort the two sides are too busy fighting each other to meet the threat.
Americans do not benefit from being at each others throat. However Americas rivals find it much easier to out compete them.
But how will the pendulum continue to swing when one side is decisively destroyed? We may be looking at a permanent change within the country's power dynamics. That's going to make this country very difficult to live in for certain types of people, regardless of what choices they make.
Something, color revolutions (external pressure) start by hijacking a cultural revolution (internal pressure). A good example is the external pressure applied to Iran since the embassy siege in the 70s, it takes a long time.
The American led Western Empire has had the openly stated aim of regime change in China and Russia for a long time. What did they think the Chinese and Russians were going to do, just sit on their hands?
At the fall of the USSR Russia was on her knees, China had not yet received the goose that laid golden eggs, (the Wests industrial capacity) America was free to mould the world in any way they wished, they were masters of the universe.
How did it all go so horribly wrong? How is it possible to make so many catastrophically bad decisions in such a short time?
A little gentle pressure on the pendulum, first left then right, again and again and again until left and right are the ones pushing the pendulum and they destroy everything of beauty and value and the external force wins. This is what a color revolution is.
>In 2026, it seems like we have gone off the deep end. Certain politicians lie with reckless abandon -- not 'lie but if you squint its a misinterpretation', but rather 'this is a bald faced lie by any count'. See, for e.g., the immediate claims that Alex Pretti was a terrorist. Or, more mundane but perhaps more important, Trump continuing to claim that the economy is the bestest greatest economy in the whole wide universe.
At least one of us must be in a bubble, because I can easily steelman both of those as either true or directionally true misinterpretations?
"The economy is better than ever" seems a pretty plausible claim; we're not in a recession, stock market is the highest it's ever been, unemployment is pretty good, inflation is low, etc. etc.. There are obviously arguments against this and some of them I can buy, but...come on. Listen to literally any state of the union speechfrom the past 50 years, from presidents left and right.
"Alex Pretti is a terrorist" is a more complicated one. There's a moral argument I can see for assuming innocence until proven guilty (I tend towards this approach, but this also means I try to extend grace for the police side too!), but if it's between "Alex Pretti was a confirmed domestic terrorist" and "Alex Pretti was a saint and a martyr who never did nothing wrong and whose shooting was incontrovertibly unjustified", we know both are obviously false now, as of Feb 10 2026...right? The main sin here seems to be "judging the situation based on rumors or mistaken information before we had all the facts", which is obviously nothing new or limited to a single person or party or side.
The economy has some wiggle room, but by the ruling party's own standards it is dogshit. It is worse than it was when they inherited it on all metrics but total valuation, and you gotta admit that three companies passing the same billions around in a big circle is pretty alarming.
Re. point two: This is straight up not true. Everyone saw some guy scuffle with ice, then get held face first onto the ground and executed thusly. You can see with your eyes that even if dude was a terrorist, the people who killed him and the people who justified it had no way to know that at the time.
I think it's important to distinguish between misrepresentations and lies.
I generally agree with Scott that the media (and most people) rarely lie. Cf https://open.substack.com/pub/astralcodexten/p/sorry-i-still-think-i-am-right-about you can follow the links back to his other posts in the subject. Many people spin, but it's rare for someone to outright fabricate statements that aren't true. That feels different, and is what I'm referring to that seems unique in this era.
I short handed the economy one, but I think there are examples of trump literally making up numbers -- how much revenue saved from DOGE, for eg.
> I generally agree with Scott that the media (and most people) rarely lie.
Depends on how you're accounting "rarity".
Zvi Mowshowitz had a much more accurate take: there are topics on which the media rarely lies. There are others on which it lies brazenly nearly 100% of the time. It's just a question of media culture, and it's specific to every individual topic.
In that case, my model is something like this: Trump asks someone how much money DOGE saved, a few people give some answers, he takes the highest number and then tweets it out without checking it. NYT spends a week repeating that Trump is knowingly spreading capital-F Falsehoods. Conservatives don't trust NYT (for valid reasons) and honestly don't really care about the exact number in the first place, as long as it's directionally correct. Later, some lower-ranking official either softly walks it back or justifies the logic by which they came to this number or explains some nuance that was missing in the original Trump tweet. The cycle then repeats.
This feels different (and probably slightly more understandable) from "Trump and his cabinet meet in a dark, smoky room and decide on how best to intentionally deceive the American public" or "Trump is a pathalogical liar who just makes up baseless nonsense out of a hat".
Reading this again I'm actually more interested in the defense of 'alex pretti is a terrorist' as 'directionally true misinterpretations'. I know that a big part of the postmodern project was to significantly devalue the meaning of words, but there were also factual statements that were proven untrue on video -- "the gun was being brandished at agents" is I think the example that Noem went with.
I can maybe see this as "Noem just fired from the hip", and I agree that this is not the 'dark smoky room', but I also didn't specify that there had to be a dark smoky room to begin with. I was more curious about why there seemed to be a degradation in truth-telling writ large.
One reasonable answer is "more people are just willing to shoot from the hip more, which naturally increases the number of lies being told, where 'lie' is defined as 'saying something that is not true' instead of 'saying something with the intent to mislead'"
I think your last paragraph is true, and explains Noem's psychotic willingness to assert something that would later be shown to be obviously false.
But Paul Brinkley really got to the heart of the matter with his insightful comment elsewhere in the thread:
> As I said, groups can lie to each other about things that aren't that important to them. Moreover, what's literally said doesn't always matter as much as what's implied. If you and I are lifelong friends, say, with a history of chatting, there are things I could say that would appear false to an outsider, but you know what I really mean, what I'm really trying to say, and it'd be a reaffirmation of friendship. What I literally said doesn't matter. It might piss off outsiders, and *that's* what matters.
Administration officials can call Pretti a "domestic terrorist" without shame because the phrase doesn't have a consistent, consensus referent anyway, and they think 𝘢𝘭𝘭 𝘰𝘧 𝘵𝘩𝘦 𝘱𝘳𝘰𝘵𝘦𝘴𝘵𝘦𝘳𝘴 𝘢𝘳𝘦 𝘵𝘦𝘳𝘳𝘰𝘳𝘪𝘴𝘵𝘴, at least the ones engaging in disruptive tactics.
> at least the ones engaging in disruptive tactics.
Though notably even here, the words are doing a lot of eliding of meaning.
> the phrase doesn't have a consistent, consensus referent anyway
I think more generally, *no* words have a consistent consensus referent. This wasn't some novel truth, people were talking about signs and signifieds back in the 60s. So is it only now that pols realized they had the ability to really lean into it?
>Reading this again I'm actually more interested in the defense of 'alex pretti is a terrorist' as 'directionally true misinterpretations'.
That one is pretty easy. If we define terrorism as deliberate violence to achieve political ends, Alex engaged in terrorism when he kicked the officer's car 10 days before this and broke its tail light. Therefore, he was a known terrorist. Yes, I would agree it's playing a little fast and loose with definitions, but if e.g. hypothetically Noem was sued by Alex's family for slander or something, I think it's an argument that could work in court.
As for the "directionally true" part, it correctly challenges the notion that Alex was a completely innnocent, peaceful protester -- this is the signal from the noise.
>One reasonable answer is "more people are just willing to shoot from the hip more, which naturally increases the number of lies being told, where 'lie' is defined as 'saying something that is not true' instead of 'saying something with the intent to mislead'"
This framing seems broadly accurate to me.
---
On another note...I'm actually interested if we agree on the same set of baseline facts, since I have a more right-wing media diet than most and am probably personally to the right compared to most people here (I'm still a card carrying fence sitting grey tribe centrist though, I promise!). I'm probably wrong about some of these since I have not followed the story as closely as others.
the version of the story I know (from my own perspective):
* ICE and border patrol end up deploying in force to minnesota for complicated political reasons.
* a lady gets shot by ICE which causes a lot more protests occur after this. Most of them are peaceful, but some go too far and end up disrupting innocent bystanders or are violent or borderline violent.
* both Trump and Waltz engage in some dangerous and unproductive political brinksmanship.
* a guy named Alex Pretti is involved in some of these protests. He is definitely in the "violent or borderline violent" category. There is a video of him attacking a police vehicle during one of these protests and causing significant damage. Somehow he doesn't get thrown in prison for this.
* he later decides to show up to another protest in order to antagonize armed federal officers while carrying a concealed firearm. This is technically his right, but it's also a horribly, indefensibly stupid idea.
* the border patrol officers seem to recognize him
* the officers attempt to arrest him for unclear reasons.
* he resists arrest for unclear reasons. There is a confusing scuffle. One of the officers probably disarms him during the scuffle. Two of the officers suddenly panic and open fire on him. It's still unclear whether or not the cause of the panic was an accidental discharge from Alex's gun at this point.
* When all is said and done, alex is shot dead.
* the federal government irresponsibly claims the shooting was justified (before all the facts were known)
* the new york times irresponsibly suggests the shooting was probably unjustified (before all the facts were known)
* as of feb 10, 2026, it is still not obvious whether the shooting was legally justified or unjustified.
Ya this is interesting. I think just given what you've laid out, my understanding of the relevant facts is much less sympathetic to the government than yours. For added context, I'm not on any social media (no Twitter, Facebook, Reddit, insta) except for a bit on substack where I don't subscribe to anyone but mostly get matt yglesias and Nate silver on my feed. I subscribe to Scott, Matt Levine, and Ben Thompson. I read a lot of AI papers and skim Hacker News roughly daily. I don't trust most media outlets, but I will get and read links from folks in my friends discord.
With that in mind, here's my understanding of all of the relevant facts:
> ICE and CBP are deployed to Minneapolis for complicated political reasons. Notably, these reasons have basically nothing to do with actually solving illegal immigration (Minnesota in general and Minneapolis in particular have a very very small illegal immigrant population, being ~1500 miles from the closest southern border crossing).
> The local population does not want ICE there, including the entire local and state governing apparatus. Local protests begin basically immediately.
> ICE seemingly purposely inflames tensions. They weild guns indiscriminately, escalate rapidly in countless interactions, and are caught on video beating individuals at a moments notice. These videos spread, resulting in more protests that get more organized.
> The federal government issues memos and public statements indicating that ICE has immunity. ICE and the feds are brought before courts constantly, lose frequently, but continue to get more brazen. Videos circulate of ICE detaining US citizens, shooting priests, and detaining 5 year olds. School attendance plummets.
> ICE shoots a woman (I prefer the active voice here). She survives. The federal government immediately shuts down investigations and refuses to cooperate with the state. They tar the victim as a terrorist, claiming the agents feared for their life. Soon after, evidence is released showing ICE agents bragging about the shooting ("7 holes with 5 bullets"). No federal investigation occurs. This court case is ongoing.
> Soon after, ICE shoots a woman. She dies. By this point, protestors have organized to make sure that all ICE action has blanket video. Again the federal government stonewalls and claims the woman in question is a terrorist. The federal government opens investigations of the victim's widow (!). The ICE agent is not investigated, and the federal government refuses to share evidence with the state. Despite this, video leaks. It becomes immediately clear that regardless of what happened in the moment, ICE clearly and unilaterally escalated the encounter, and then later refused to allow the woman medical attention as she bled to death. The actual shooting is debated, but it is undoubtedly against training (at the least) and in my opinion from my own lying eyes state sanctioned murder.
> This leads to an increase in protests, and in the organization of those protests. I think very few of these protests are violent (if they were, the violence would be more specific and televised; it is neither. There is much more control and discipline than in LA). Local and state government are pressured by locals to do something about the actions of ICE. (I didn't follow the exact details of the brinkmanship you refer to, but I'm not surprised that it was happening in Minneapolis.)
> Alex Pretti gets involved around this time.
> He goes to at least one protest where he kicks out a tail light. He is beaten for this, but not arrested. (I do not think this makes him obviously a violent protestor -- there is very little indication that he was violent outside of this one interaction, and there is so much video coverage and his life under so much scrutiny that I would expect more evidence to surface. It hasn't.)
> Ten days later, Alex is at another protest. Here, events are captured basically entirely on video. The events as I see them:
- Alex and a woman are video taping an officer but are not otherwise being belligerent, intervening, or getting in the officers way. Notably, the officers are not doing a raid, so there is no action to intervene.
- an ICE agent confronts the woman, seemingly as she is walking away, and escalates aggressively by pepper spraying her. It does not seem like there is any obvious provocation. The woman falls to the ground.
- Alex gets in between the ICE agent and the woman, with just a phone in hand and is not violent or interacting with the agent beyond getting between the agent and the woman. The ICE agent pepper sprays him too. (I don't make any claims on whether the ice agent recognized Alex. I think it is far worse for ice if they did)
- several agents pile on top of Alex. One is seen on video hitting Alex on the head with a blunt object, possibly a pepper spray can.
- video shows an ice agent pulling a gun from Alex's holster. This is the first time the gun is seen. In that same video, it looks like the agent that fired the first shot sees that the gun has been removed from the holster. This agent moves another agent out of the way and then fires the first shot into Alexs back. (At this point I've seen enough videos to call this an execution. But even if we're being generous, ice does not get to hide behind the chaotic conditions they created as justification for their violence)
> In the immediate aftermath, the president and head of DHS release statements claiming Alex is a terrorist, the latter claiming that Alex brandished a gun with the intent to kill many agents. These statements are proven false by video. The NYT may claim some things also -- I don't follow the NYT, and I don't see why it matters what they claim as they are not the federal government.
> The federal government refuses to even release the names of those involved. They again stonewall the state, refusing to provide evidence or even allow the state to investigate the crime scene until after they have left.
> As of Feb 10, none of the ice agents involved in any of this are under investigation. In fact, there is no investigation at all, so there is no way to determine if the shootings were justified or lawful. This is because the apparatus responsible for this investigation has unilaterally decided not to do one.
I think the key principle here is that discussing the specifics of the shot outside of the context that ice created is misguided. ICE claims "ah there was so much chaos, how could we have known better!" But the reality is, ICE is the one with the agency -- that is why they are literally called agents. When I examine what I know of the facts, including your version of such, I come away thinking that it is ICEs choices that led to this outcome.
I can see myself read my own bias into your question and I think my answer might not be exactly what you were asking but when read "What the fuck happened?" in relation to lies I generally think "This another person on the left complaining about the right". Where as, as I only listen to left media, all I hear is spin and lies from the left and I'm surprised they don't see it.
A concrete and hopefully incontrovertible example is the New York Times writing about Scott and all the spin and lies that were in it.
I used that example because I hoped it was clear it was a lie from the left. But, every time I listen to almost any media I hear/read 95% spin and lies.
This is not to excuse the right. I don't listen to right media at all so I have no idea how bad it is there. My point is only that it's bad enough on the left to encourage escalation. The whole "pull the mote from you're own eye first".
If guessed wrong about where this was coming from, sorry in advance.
Possibly none of it (or essentially none - the media can certainly tell falsehoods by accident, if not on purpose).
One of the lessons I took from "The Media Very Rarely Lies" is one I already knew - it's very possible to imply all sorts of falsehoods without actually saying them (and that this matters - it makes the difference between grumbling and a lawsuit).
I think the part of the US that didn't like what corporate media was telling them, was also noticing that the media could say things without literally saying them, although they also often confused the two, asserted the media was lying, would get argued down ("they didn't actually literally say that"), get frustrated, and simply keep not believing that media, but also stop saying they were lying. People on the other side similarly confused this silence with belief.
I'm in Aerospace. I observe that in every case I've ever seen when the news touches on Aerospace, every single word is wrong, including "and" and "the." Based on that sample, I conclude that every word they say about any technical subject is equally likely to be correct.
Law and Economics are every bit as technical as is Aerospace. QED.
I can attest that most of what they write about law is wrong. And most is easy to get right.
On another note, a recent NYT an article re Iran, said, "While the government appears to have largely suppressed the protests, Mr. Memarian, the think tank analyst, warned of “a dam that could suddenly crack.” “It may harden public resolve, making the government more vulnerable if protests continue, rather than less,” he said."
The person in question is identified as "Omid Memarian, a senior Iran analyst at DAWN, a Washington-based organization focused on the Middle East.." Leaving aside that this guy is a journalist rather than a political scientist, the NY Times does not understand that they need to consult an expert on regime change, or government repression, or political violence, rather than an expert on Iran They do this all the time.
The Republican Party never really succeeded in making the case that the tax cuts for the wealthy were good for the average voter. That largely ruled out an honest approach where they would tell voters what they were going to do and explain why those policies would be good for voters. I think that pushed Republicans to see how much dishonesty they could get away with. What undermined the norm is that it turned out that dishonesty worked.
Read about the "Two Santa Claus Theory" by Jude Wanninski. He argued in the 70's that Republicans could get credit for fiscal conservatism and also stay popular by always cutting taxes and not cutting (popular) spending. The history of Republican administrations from Reagan to now shows that they took this theory to heart.
I think the plan was that by cutting taxes and not spending, they'd force the Democrats to be the bad guys and cut spending, but in practice it hasn't worked out that way.
> I think the plan was that by cutting taxes and not spending, they'd force the Democrats to be the bad guys and cut spending, but in practice it hasn't worked out that way.
Well, I think you're right about the conclusion. ("It hasn't worked out that way.") But my understanding of the theory is radically different from yours. The slogan was "starve the beast". The original idea was to cut spending; that's what they wanted to do. But when that proved to be politically impossible, "starve the beast" was the theory that if you cut taxes, spending would have to fall because the amount of money available to spend would be lower.
Once you've got a fiat currency, it's all the same money, fake or not -- it all comes from that magic printer in the Treasury that goes brrr, right? There's no economic law that forces people to stop spending just because their deficit is getting big.
The closest we got to cutting spending was the Clinton years, when federal spending grew at an average rate of 1.51% per year (after inflation), and we eliminated the Federal deficit. George W. Bush promised to run a balanced budget if elected, but when he won he cut taxes and increased spending at an average of 4.56% per year in his first term. (The average increase over his entire tenure was 5.62% per year, pushed up by spending to address the Great Recession during Bush’s final year.)
I somehow missed the “Two Santa Claus Theory,” so I thought Republicans actually cared about the deficit until I saw the lack of pushback against Bush.
Seems to me that a frequent mechanism that causes this is a weaponized "fallacy of gray". It works like this:
1) Notice that some other person/group is not perfect and did something wrong.
2) Therefore, you are morally entitled to do the same thing. Or a similar thing. Or a different thing. You are entitled to do ten times the same thing, or a hundred times the same thing. (The logic is that things are either forbidden or allowed, and now that we have established that it's clearly the latter, it doesn't make sense to argue quantitatively.) You are entitled to do the thing even to someone unrelated, or only related by a vague analogy, to the person or group who originally did the wrong thing. Your friends loudly support this interpretation.
3) "Hey, if *they* are allowed to do that, then so am I. Stop being unfair to me! This is a legitimate self-defense."
To some degree this is natural and many children do it. An important part of growing up is to unlearn this pattern of escalating every perceived offense. But most people still keep this in their hearts, only suppressed by the expected social disapproval.
So you need to get them loud social approval. Easy, if you own the media, or can manipulate social networks.
I think it's worse than that - I think more often than not 2 actually comes before 1. People decide on the terrible thing they want to do and their entitlement to do it first. *Then* they go looking for the "high tech lynching" of Clarence Thomas or the Lewinsky scandal or the 1876 election or whatever they can claim that the "other guys did first."
The elites destroying their republic aren't doing it because they were nice decent people who saw the sins of the other side and somehow overreacted. They're miserable bastards who'd be destroying their republic either way. In a world where Gore v. Bush hadn't happened, or where Mitch McConnell had given Merrick Garland an up or down vote, or whatever, they'd be doing the same things; they'd just be pulling different justifications out of the couch cushions for them. The desire to commit the wrong creates the need for the justification, and therefore a justification is found. If the best justification you can find is weird trash, then that's what you go with.
That's why the proffered justifications are often so wildly goofy and disconnected from what is allegedly being done "in response" to them. When people's best argument for the thing they want to do is clearly a turd, but they still fully want and intend to do the thing anyway anyway, it's no surprise that you suddenly see them throwing around full-throated turd-claims.
I think the people willing to do this have always been there, but the internet and social media have made the tactic more effective in the digital age than it was before. Easier access to information when you are trying to scrape up justifications, more hands to do the scraping since mass numbers of your tribe can get engaged online digging, and a pluralistic media environment that makes it easier to flood your tribe's media environment with those justifications.
There is one step missing. First you are entitled to do the same thing, but then you realize small wrongs are the same as big wrongs, so now you are entitled to do 100 times more.
What, why are small wrongs the same as big wrongs, this confuses the consequentalist? Because most people are not consequentalists, they are motivated by shame and judgement. If we say "Bob is a liar", it does not matter how big is the lie, it is largely the same judgement and shame.
This is why don't really believe in consequentalism, because it is incompatible with intuitions and social dynamics. Real life morality is playing the "who is the asshole?" game.
Then I guess the difference from the past is that thanks to internet, we can always find one person somewhere on the planet who broke a rule, and then we can share the article with all members of our tribe, and say: "see? if *they* can do it, so can we!"
In the past, I guess people would apply additional heuristics and e.g. only consider the breaking of the rule relevant if it happened in their city, if it was done by a person from the same social class, etc.
I've been building WagerAlpha — a tool that auto-pulls your Polymarket or Kalshi bets (via wallet address or API) and calculates real performance metrics: raw ROI/alpha, closing line value (CLV), bet Elo rating, category confidence scores, provisional EV, Sharpe/Sortino/Calmar, max drawdown, etc.Goal is to give sharps verifiable proof of edge instead of just claims, and later let high-performers create paid communities (platform takes 20% cut).Beta is live with full Elite access currently free. If you're trading prediction markets seriously and want to see your actual numbers (or just give feedback), link your bets here: https://wageralpha.io
Happy to answer questions or hear what features people would want. Not financial advice — just a tool I built for Polymarket/Kalshi tracking.
Can I enter the reference for someone else's account and see their stats?
Ideally, it would be nice to have the following:
(1) tagging bets into sub-portfolios: there are many different types of questions and it would be useful for a trader to identify whether they are better at some vs others. I understand tagging ex post opens the possibility of cherry picking in a way that muddies the performance reporting.
(2) a way to record the trader's private estimate of probability at the time of trading.
Lastly, a scaling trades up is a major concern. Something around market size and transaction costs is necessary to really know if a trader could handle an interesting amount of OPM.
Hi Joshua, thank you for the feedback I really appreciated it.
1. I made some updates based on your comments and now the bets are categorized based on category in the dashboard, I had the app autosort that instead of having people tag their own. I also added a way for traders to privately list their own confidence, however its hard to prove it was before with pulled API trades so I kept that field private for people to analyze their own predictions.
2. I was thinking the communities were more a way for strong prediction analysts to share their analysis and points of view than something like pure pick selling. Although I will see if there is some way to include market size in the analytics as well since I agree that an edge in highly traded deep market is both more difficult and valuable than an edge in a small and shallow market.
If I want to support Iranians in their resistance against the clerical regime, what would be the best charity to donate to?
I guess I'm most interested in charities whose explicit goal is to bring about the end of the mullah regime (although I guess they might not be openly stating that even just out of a concern for the safety of their employees).
Thanks for any recommendations.
I guess currently anyone who can help bring more Starlink systems to Iran would be great. Or weapons so that next time the people can actually shoot back at the revolutionary guards. I doubt anyone who can do that actually talks about it publicly on the internet. But that is exactly what makes it hard to support them.
Ha. I wouldn't actually mind sending money to Mossad if they could guarantee that they will only use my money to bring weapons and Starlink terminals to Iranian resistance.
Which do you think would be more damaging to the fabric of American society (such as it is): the dismantling of monolingualism – perhaps through the proliferation of things like the 2026 Superbowl half-time show – or the conquest of Greenland – and the concomitant shift in international relations?
Largish parts of America are already bilingual and I can usually talk to big companies in Spanish by pressing nine, so I'm not sure how explosive formal bilingualism would be. In Canada it's a permanent annoyance but people aren't stabbing each other over it.
It is well established that bilingual individuals benefit from cognitive advantages compared to folk fluent in only one language, especially as they get older. All the people around here desperate to try dubious interventions in the hope of even tiny increases to kids’ intelligence should be all for maximising exposure to multiple languages.
There's no creeping bilingualism. Virtue-signaling white and Jewish liberals put Bad Bunny on the sportsbowl as a middle-finger to normal Americans and also because of their sadomasochistic desire to sabotage the Democratic party's chances. In places like lots of Hispanics live there's little creeping bilingualism as Hispanics like English-speaking jobs, movies, cultural phenomena, etc.
The right's schtick of "only people who agree with me are Normal Americans" has always been dumb, but I think claiming that Normal Americans aren't interested in seeing a Grammy-winning, chart-topping artist at the Super Bowl is an especially dumb variety.
Like, come on man. You and I may not listen to Bad Bunny, but clearly a lot of people do! He did not go multiple-platinum solely because of latte-sipping West Coast elites.
I may not be normal, but I liked the halftime show. My complaint about there being no subtitles is I feel like I missed a huge part of the art by not being able to understand the lyrics.
>Normal Americans aren't interested in hearing a performance in a language they don't speak.
Why not? Plenty of people are interested in K-pop, J-pop, and J-rock despite not speaking Korean or Japanese. If we look to the past English was insanely popular in the European disco scene at a time when English fluency was much lower than it is today and most people listening to English music couldn't understand a single word. English is still popular in many Japanese music genres despite neither the artists nor the intended audience understanding any of it. It just sounds cool.
"Plenty of people" are not "normal Americans". I think Turok is probably correct that listening to music with foreign-language lyrics is a niche taste in the 21st century United States, even if it has been somewhat more popular in other times and places. Instrumental music is language-independent; if there is value in adding lyrics, it's mostly by way of the audience understanding the (poetic rather than literal) meaning of the lyrics.
Yes, some people do find other sorts of value in it. And there are plenty of venues - most of them, really - where music catering to a minority taste is appropriate. The Super Bowl, is probably not the right place for that unless you're deliberately trying to offend a large fraction of the audience - or parade a naked emperor in front of them to demonstrate you have the power to make them say "great clothes, man!"
So what is a "normal American"? The most average guy possible? If it doesn't raise eyebrows I'd classify it as normal. And someone telling you they listen to K-pop is not raising any eyebrows. Hell for a brief few weeks in 2012 a K-pop song was the most popular song in the country thanks to viral memetic spread.
> Instrumental music is language-independent; if there is value in adding lyrics, it's mostly by way of the audience understanding the (poetic rather than literal) meaning of the lyrics.
I strongly disagree on this point. The vocalization is an integral part of the sound of a song, it doesn't merely convey information the way subtitles would.
If you don't believe me, listen to an instrumental version of your favorite (non-instrumental) song, and read the lyrics in parallel; make sure you aren't "sounding out" the vocalization in your head. It will feel different than listening to the original version.
> sadomasochistic desire to sabotage the Democratic party's chances.
This seems contrived to get to your initial conclusion, and the current odds on prediction markets don't support this idea. It seems more likely that they're betting they've successfully replaced enough of the Anglophone population, and the "middle-finger to normal Americans" is their way of honestly signaling their allegiance to their new Hispanohablante voters.
I don’t speak Spanish but listen mostly to Spanish-language music on the radio. Aside from the fact that I very much enjoy Norteño, reggaeton, and other predominantly Spanish genres, an advantage of this is that when the ads come on they are easy to tune out.
Or if i don’t feel like tuning them out then the ads are interesting rather than annoying, because my brain can work on the puzzle of picking out whichever few words I recognize to piece together what they’re saying, and then ponder the similarities of what is being sold to the Spanish-speaking market segment where I live.
Thus the destruction of monolingualism makes my experience within American society better, not worse.
The US has always been multilingual. Before WW1 there were entire German speaking neighborhoods. In San Francisco, there's still an active German language Lutheran Church, remnant from the previously active German immigrant community. German, Polish, Italian speaking neighborhoods were ubiquitous. But they became white as western European immigration dropped off. And the same thing is happening to Latinos. It's immigration that keeps topping off the culture. But the older Latino generations are becoming white. America can handle Spanish on the Superbowl just fine. That isn't going to destroy us.
However invading Greenland, destroying the pax Americana, losing our dollar reserve currency -- That's an apocalypse.
To my understanding, the social category of "white person" is well less than a millennium old[1]. Humans have always come in a range of ethnicities and a range of appearances, but different cultures have drawn very different category boundaries through that space. Given that the category boundaries are partly artificial in the first place (and change with time), insisting that this or that group has *always* fit into the scheme in such and such a way seems...not especially defensible.
As a side note, I find it equal parts puzzling and frustrating when people act as though their own culture norms are perfectly universal and perfectly logical, even in contexts where they're being directly called into question.
[1] And before you raise any objections about who isn't or isn't "objectively" white, I'll invite you to consider that the same *individual* can have considerable variation in skin tone throughout their life due to things like health and sun exposure. Obviously different people have different skin tones, but the boundaries are anything but sharp.
Cultural Whiteness is trivially not about skin colour alone. When people talk about groups "Becoming white" they're talking about how the various fractured white ethnicities, (anglo saxons, Germans, Polish, Irish etc etc etc) became homogenised and accepted into mainstream america whereas before they had been seen as a dangerous out-group.
The american conception of whiteness doesn't easily translate over to europe, although over the last ten or fifteen years its been one of the biggest exports. American's would find the level of discrimination Polish people faced in europe in the 00's and 10's bizzare.
>Cultural Whiteness is trivially not about skin colour alone. When people talk about groups "Becoming white" they're talking about how the various fractured white ethnicities, (anglo saxons, Germans, Polish, Irish etc etc etc) became homogenised and accepted into mainstream america whereas before they had been seen as a dangerous out-group.
Clarifying is not a motte and bailey, but simply repeating "motte and bailey" is a thought terminating cliche. If you'd rather tilt at windmills than understand what people to the left of you actual think, thats your perogative though.
Depends on how you define "destroy". The European ethnic communities are all homogenized and assimilated. It's not unreasonable for people to think of the American social fabric having been destroyed after White Flight/the Great Migration. But it's not like everyone died of sadness and America just ended, just because terrible shit happened. In what way would invading Greenland/losing reserve currency be an apocalypse?
I think there's a reasonably good chance that the end result of a U.S. invasion of Greenland and the loss of reserve currency status for the dollar would be the United States of America ceasing to exist as a unified political entity. Relations between different parts of the country are already strained. Piling on what would likely be the least popular war in the history of the nation AND a great depression-level economic disruption would certainly touch of massive civil unrest. Do you think the country's leadership is capable of navigating such a challenging and delicate situation without causing it to spiral further? I don't.
Now, whether or not the U.S. falling apart counts as an "apocalypse" is very much a matter of perspective. But it would certainly be a much bigger, more abrupt upheaval than some internal migrations and generational cultural shifts.
Maybe apocalypse is a strong word. But to me it signals the loss of tangible hard power and a hard economic future for Americans. If we lose the ability to dictate the trade terms, sanctions, our ability to offload treasuries in our time of persistent deficits. People run to the dollar in times of world crisis, that keeps our borrowing costs low. Trump can only bluster about tariffs and deals because people NEED to trade with America. If we lose that 'need', then we got nothing. Invading Greenland pushes the world into a future where people don't NEED America.
I am long America. I want the SP500 to grow 14% every year for the next 20-30 years. Losing our reserve status will more directly threaten that, than Spanish on the Superbowl.
> the dismantling of monolingualism – perhaps through the proliferation of things like the 2026 Superbowl half-time show
"Americans are so funny. You've accidentally listened to one song in a foreign language, and now you think the sky is falling and the fabric of society will crumble. Meanwhile, large parts of the rest of the world have been listening to English music for 70+ years, and their non-English, monolingual society hasn't been 'dismantled'."
That would be my reply to you if I didn't know that you habitually post incendiary, intentionally baiting questions, packed with false dilemmas and unjustified assumptions. If you were really interested in a discussion about social cohesion, as opposed to just stirring the pot, you could have asked something like "Would a proliferation of foreign languages in American entertainment traditions – like the Superbowl half-time show – inflict damage on the social cohesion of American society?"
>and their non-English, monolingual society hasn't been 'dismantled'."
Well... it did. Have you heard how Berliners talk? It is terrible. "Wir suchen eine nice Location für unsere Wedding". Words like "cringe" or "sus" keep popping up in German or Hungarian. Even the Frogs are giving up, and they were always keen on keeping the French language "clean".
If you speak German well enough to judge how native speaker use the language, *surely* you should be aware that borrowing foreign words is such a common and longstanding German practice that German has a dedicated word for it:
Of course, the language we refer to as "German" has been fragmented into many dialects for as far back as we have written records of it. So complaining that foreign influences are going to "dismantle" German language or culture could never be anything but a profoundly unserious argument.
Many German foods are named after cities, presumably cities that someone thought - or wished to give the impression - that they were characteristic of. This is normal, analogous to a "Philly cheesesteak" or a "Chicago-style pizza".
All German city residency demonyms are named after cities.¹ The words are the same. This is even more normal; the suffix is even the same as in German's sister language, English. Berliner : Berlin :: New Yorker : New York.
¹ As far as I know. It's not out of the question that the name of the city and the name of the residents could differ, similarly to how we say the people of "Myanmar" are "Burmese".
In addition to what deusexmachina said: Berlin's problems don't stem from multilingualism. At best, it's another symptom of immigration without assimilation, but that has certainly not been caused by the broadcast of foreign language music.
How is the use of non-German words evidence of a society being dismantled?
The Kaiser's troops spoke of Detachements, Pelotons, Tranchees, Parapets, Lafettes, Fourage, Coups de Main, Retraits, and today people speak of cringe, sus, and "nice" things.
Why would a multilingual US be automatically a bad thing? Switzerland has 3 (technically 4) languages and works just fine. And English and Spanish are actually relatively close so easy for people to learn both. They are closer to each other than French is to German.
Conquest of Greenland would be a monumentally stupid idea that would hurt US the most, then Europe and could only help Russia and China. It would mean US burning all of its remaining soft power for ... ??? I don't know even know what, there is nothing to gain other than Trump's ego, everything else can be easily achieved under existing treaties. It would mean the collapse of NATO which would greatly weaken both US and Europe and in general pretty much the collapse of the current world order. The idea is so idiotic that hopefully even some of the less sycophant voices in the Trump administration seem to be steering Trump away from the idea. Just rename a few airports or something after him instead ...
Let's be honest here, on the canton level Switzerland is very monolingual. I was to Winterthur, a little project. Yes they had to pass some high school exam in French and Italian. Then they went on never using it ever.
Yes, that is probably fair. But I also think it is not a problem. The country still functions fine. I think it is mostly because Switzerland embraces strong federalism and cantons are fairly independent of each other in many ways. They are also much smaller than almost all US states and so it is fairly easy to move between them if you're not happy in one (you can still visit your friends and family in the previous canton in a matter of 1-2 hours).
On the other hand with bad policies you can get the way of Belgium which is not even a real country anyway :-P. In many ways it's two countries glued together into one and the linguistic and cultural differences make it worse. But if Belgium became more like Switzerland in its governance I think it could still be fine. And arguably Switzerland is in many ways more American (in the sense of its government living up to the ideas of the founders of the US) than the current USA.
Multilingualism works OK in Switzerland because France and Germany and Italy (or at least northern Italy) are similarly developed and wealthy countries. Similarly, Canada.
But the gap in GDP and HDI between Anglophone and Hispanophone parts of the Americas is enormous. The GDP of Latin America is something like $11K per capita.
> Multilingualism works OK in Switzerland because France and Germany and Italy (or at least northern Italy) are similarly developed and wealthy countries.
This doesn't feel like a complete analysis. Notably, while France, Germany, and northern Italy might be similary wealthy to each other, they're all much poorer than Switzerland.
I'm not sure what the mechanism you're hypothesizing is, but the large gap between Switzerland and its neighbors seems like it'd be relevant.
Multilingualism does not imply more immigration. One can imagine Spanish becoming more officialized, and immigration becoming more strict. So the GDP of other countries is not that relevant.
I can see issues for a country shifting from Mono- to Multi-lingual system. In fact, I don't think it is a good idea. But the fact that a country in South America both uses Spanish and has lower GDP than the US is not a relevant factor to consider.
Well, Ticino and Sicily are not quite so close (as you point out yourself), but ok.
So I guess the idea is that there would be too many immigrants from Latin America which is bad because?
I guess the point is one of the following - a) they bring crime or b) they will live of welfare?
I think both a) and b) would be reduced if the country were bilingual but I concede that b) is a fair point as long as you want to maintain the welfare state.
But I think the solution is to abolish or significantly reduce the welfare state. US managed to soak up millions of mostly poor people from Europe in the past without any long-term issues. Actually, the exact opposite happened as without them, US would never be the superpower it is today. And Latin Americans are culturally not all that far from Spaniards or Italians, hence neither from northern Americans. I don't see why, under the same conditions, their immigration would be less successful than that of the Irish or the Italians in the 19th century.
You can probably even keep the welfare state if the law said something like "anyone can come but you only get access to welfare after 10/15/20 years of working here"*. It would also be fair if it said "and you don't have to pay into the welfare fund while you are not yet eligible to collect its benefits" but that would never work politically. I think people would still be willing to come even so.
*legally, of course, but that is trivial when you give a work visa to anyone.
> You can probably even keep the welfare state if the law said something like "anyone can come but you only get access to welfare after 10/15/20 years of working here"
AIUI, immigrants are already banned from receiving most forms of welfare for five years.
Well, then figure out the right (politically and economically) number of years or even tie it to citizenship and it should work. 5 might be too low. 15 should be enough. Nobody is going to work for 15 years in hopes of living off welfare. You might scrape by for 5 years but not 15.
I dunno. His hypotheticals are pretty far from being inflammatory triggers . If he mainly wanted to stir the pot, seems like the thing to ask would be which would give the US the greatest chance of improved unity and goodwill: assassinating Trump or letting him and his staff continue with their agenda. (Although how many of us haven't asked ourselves that?)
The dark matter of the human psyche is an emotion which you could call something like "group-level humiliation". It's a feeling that you get when you're part of a group that have been humiliated and low-status throughout history, and it brings feelings of shame as well as resentment against history's winners. This feeling has motivated all sorts of acts throughout history, ranging from the merely irrational to the mass-murderous.
But the reason I call it dark matter is that I can't personally relate to it at all, I think a lot of us here can't, because our ancestors have always been winners, or close enough to winners. I can understand what it means to be humiliated on an individual level, but not what it means to be humiliated on a group level. I just don't get it, why do you _care_ that the so-and-sos beat your ancestors in a war three hundred years ago? I can only surmise that this emotion exists, and is powerful, from looking at the irrational actions of people in other groups.
There are a lot of these culturally transmitted values that are irrational and alien to middle-class Americans. There are social class-based distinctions where one is expected to show deference to those of higher class (and how to show deference is socially coded as well). There are honor and shame based distinctions where one is expected to respond to verbal insult (especially to one's manhood or family) with physical violence. "Why do you care about an ancestral ethnic grudge?" is as alien as "Why do you consider 'this person insulted my mother' as a reason to kill someone?". You could even think about norms about haggling or bribery or patronage as falling into the same bucket of values that are abnormal to the American middle class but normal elsewhere.
Not all of those are alien to American culture, at least by historic terms. You still find pockets of the hereditary upper classes that consider the lower classes as 'the help', and jokes about duels of honor are still recognizable. They're not quite as alien as the long-running ethnic group grudges you mention. But you still can't understand the interactions between peoples of different cultures in the present without considering some of these distinct values.
As a German I care, because humiliation is both recent and ongoing history. I try not to as I, but it's hard to emotionally detach myself from the very real consequences of disastrous national politics. And I don't think it's entirely irrational either.
> our ancestors have always been winners, or close enough to winners.
Not all survivors are winners. And as Ralph says, there are many dimensions, so you need to be relatively lucky to come on top in *all* of them.
As I see it, humans once in a while do the thing when they select a weaker group and decide to hurt them. Belonging to a "wrong" group is like having a target permanently painted on your back. Maybe no one is doing anything bad right now, but it is common knowledge that *if* tomorrow someone decides to do it again, the Schelling point is to choose *you* as the victim. That makes one more sensitive to the possibility of such thing happening.
It is possible that the fact that my ancestors lost a war long ago still has some tangible *consequences* today; at the very least, lower status in some situations.
The opposite is also possible: that I am unjustly blaming my personal failures on something that happened long ago, to feel less like a loser, and more like a member of some heroic resistance. Sometimes it is hard to tell the difference between these two possibilities. Sometimes it is a little bit of this, *and* a little bit of that.
It also depends on how well you are as an individual. For example, you can be slightly disadvantaged as a member of some group, but you can also have some other advantage which has a stronger impact, so as a result you still end up above the average. Then you may choose to deny the group disadvantage (even if you are aware of it on some level) as a way to signal your personal strength. Or to make your group membership less salient to others.
But you could also exaggerate the impact of your group disadvantages in you, in order to gain some advantage. It's complicated.
The impact can be stochastic, like you are are member of some group, and some people don't give a fuck, but some people will treat you like a subhuman. Whenever you meet a new person, you never know which category they will belong to. And even if 90% of people are in the "don't give a fuck" category, the threat remains. Sometimes it's enough to e.g. meet one asshole at a workplace (someone who will be an asshole towards you, because of your group, but nice towards everyone else, so no one else notices the problem), to make it an unpleasant experience.
I hate to use the word privilege, but come on ... I am black and female. On this site and many other bloggers who link to this site, there's always some variant of IQ supremacist talking points, something about blacks being low IQ, low conscientiousness, low sociality, high criminality, what oh what can we do with the blacks short of re-segregating, dei hire this, dei hire that, add in the black misogynoir talking points -- black females being the least desired. Group-level humiliation galore!
Now I do well for myself. My immediate family and black friends do well for themselves, so logically, I should be sticks and stones and all, but alas, I am not. There's what you think about yourself, and there's what society and this blog tell you to think about yourself, and there's how you have to reconcile both views.
How much you can reconcile those outside opinions is very much a function of your privilege and your present circumstance. If I was down on my luck, I would more readily identify with the group-level humiliation. So you get various reactions, the black version where black interlocutors reclaim cultural heroes as black, "Shakespeare was black, Cleopatra was black, Helen of Troy is black". And there's the white version, "The heroes of western civilization were white, so therefore me, a white unproductive male gooning in my mother's basement, am a hero".
We are human. We are always searching for something to prop our self image. If it means reaching for 1000 year old wars, so be it. If it means rewriting history so that people who look like you are the winners, so be it. The truth doesn't really matter. All that matters is the narrative that promotes the superior self-image that drives self-agency.
This is all clear, the question is who needs Cleopatra when you can have Rosa Parks. It is certainly easy to find recent role models. And that's someone culturally close to you, while you would have really no cultural-ethnic commonness with Mediterrean people thousands of years ago, it would be a pure looks-based association.
BTW this is one thing I as a European don't get about Americans - ethnicity is clearly not skin color, it is clearly about language, stories, poems and music and cooking and all that? Someone who does not speak your language is per definition not of your ethnic group.
America has almost wholly and confusingly commingled the concepts of "race" and "ethnicity". We usually talk about "race", and when we talk about "ethnicity" we're usually talking specifically about Hispanics. Who we consider socially and politically equivalent to yet another race but we know that's not quite right biologically so we preserve the term "ethnicity" for basically just that purpose.
I think part of the reason for that is that most Americans have very little experience dealing with people who look like them but don't speak their language fluently. Europeans all have neighboring European countries where everyone looks the same but speaks a different language (and all the cultural differences that implies). Canadians at least have French Canadians and Anglo Canadians. Americans a hundred years ago had communities of unassimilated e.g. Italian immigrants, and if we had been into talking about race and ethnicity then, we might now have a better understanding of how ethnicity works.
As is, we didn't start talking about this in a big way until we were living in a society where all the Outgroups were clearly different linguistically, culturally, visually, and phylogenetically, so it a gets blurred into "race and/or ethnicity". Which you should probably just round off to "race" if you want to understand what Americans are saying without diving into the details.
It gets even more complicated. In the US race can also be a stand-in for social class, like “pumpkin spice coffee is a white girl thing” = “overpriced novelty for the upper middle class”
I get the impression that Anglo Canadians largely don't have to deal with French Canadians unless they live near Quebec.
Similarly, American attitudes toward people who don't speak their language fluently seems to vary pretty widely. Where I grew up, in Texas, maybe 1 out of every 3 people was a native Spanish speaker (though to be fair, most of them also spoke native English well enough), so we were fairly used to language friction, and cultural friction to boot. I imagine the same is true in much of Florida. And in large cities, although maybe limited to narrower strips near enclaves.
Point being, I think one could be generally correct in saying "Americans have very little experience dealing with people who look like them but don't speak their language fluently", but also run into a lot of counterexamples if they carry that generalization into certain specific locations.
Likewise in the other direction - any big-city politician with a point to make about how diverse America is is going to be in for some shock when trying to make that point outside said big cities.
Oh, there's definitely a lot of Americans living in close proximity to Spanish-speakers, and often monolingual Spanish-speakers. But they're pretty much all Hispanics and they mostly *look* Hispanic. The overlap between "Hispanic but could pass for white" and "Hispanic and they don't speak English", is I think pretty small.
White Hispanics one might overhear speaking Spanish to another Hispanic but code-switch to fluent English when speaking to Anglos are a thing, but I don't think they're a noticeable enough thing to override the general stereotype.
I genuinely wonder about this. I definitely recall instances where someone I know, who looks like an N-generation American with probable British-Scottish-Irish ancestry, suddenly speaks fluent Spanish to Spanish speakers. (Including at least one President.)
OTOH, I recall being surprised whenever it happens, so you're also not wrong about the stereotype, I'd say. Just that you're less "not wrong" about it in Texas (and I assume FL, NM, AZ and southern CA, although I know you live around there, so you'd know).
But one of those interpretations of history is far more representative of reality, yes? I don't see why whites shouldn't use their privilege and birthright to the fullest extent possible. They are more than capable of getting everything they desire, so why shouldn't they?
How many people have much idea how their ancestors fared? I don't know at thing about generations before my grandparents. Most people who are caucasian assume their ancestors were white, and that's pretty likely, but it certainly doesn't guarantee they were winners.
I think this is in fact a collection of various and distinct emotions. For example for a population of one country it might be relevant that something happened in a war 300 years ago because they rightly or wrongly believe this might happen again. Especially if this population is not currently very rich and powerful, they might have a lot of insecurities about even relatively minor things. And it might be quite rational because not being very rich and/or powerful makes you vulnerable.
I think that might be the kind of mess you get between India and Pakistan for example. These things are usually a collection of mutual perceived or real wrongs and so they are perpetuated by both sides.
Alternatively it might be a frustration from being perceived as lesser status than you feel you are. One case I can relate to is Czechs, Poles etc being categorized as "Eastern Europeans, so basically Russians or something". This is slowly becoming less common as the older generation which is used to think in the cold war era terms is being replaced by the newer generation who actually have hands on experience with central and central-eastern Europe and have a bit more nuance. Also, to my own surprise, I've met a bunch of people from France and (both northern and southern) Italy recently who moved to Prague for work and say the quality of life is better there then where they are from (and no, their reasons had nothing to do with non-European immigration to their home countries).
But I still feel that this remains the way of thinking in the US (and somewhat in UK and France also ... I also met an older French guy recently who thought that we were still called Czechoslovakia and that the capital was Belgrade) which also makes me subconsciously annoyed and a bit defensive whenever I suspect people to have such a caricature in their head. They might not even mean bad but I guess I feel they see me (and by extension my friends, family etc) as lower status than they should because of such misconceptions which then leads to me being more sensitive and perhaps defensive about some things.
I suspect this is even more pronounced if you come from a poor country with a lot of problems but you are just as intelligent (possibly more because you achieved the same with worse initial conditions) as your peers but are constantly met with smug attitudes from people who know next to nothing about where you're from and mix you up with people whom you have a lot of bad blood with and which are only superficially related (like when people think that Iranians are Arabs).
So I think it is more about the current "global status" of a group of people rather than what happened in the past. Parts of Germany were a backwater few hundreds of years ago and "Made in Germany" originated basically as a warning (IIRC), kind of like "made in China" later. But nobody from the Ruhrgebiet cares about that past any more.
The Irish might care about British wrongs still even though Ireland is rich now, but I think they still feel that Brits look down on them a little? Dunno. Deiseach might enlighten us here :)
I don't know if a feeling of "ancestral humiliation" is necessarily based on the reality of the humiliation of your ancestors.
Like, I have a lot of Irish ancestry but my family has a modern disconnection from our "roots". I know that the Irish were oppressed by the English, and there was anti-Irish discrimination in America for a while. This hasn't given me any sort of a "humiliation mindset".
I'm also in a homosexual relationship. I realize that homosexuals were very low status at some point in the past, but it hasn't affected my lived experience.
I'm a big believer in the locality principle of history. To explain something happening now, you can't just point to something in the past. You can start your explanation in the past, but you need to follow the effects forward until they "cash out" in the specific experiences of living humans.
I think "group level humiliation" probably comes from a feeling of individual humiliation + an identification with other people who feel humiliated + the existence of a "live" narrative explaining this humiliation as a timeless struggle for which you are not to be blamed.
I'm almost positive you could, if you were so inclined, identify with a group that has a "history of humiliation". There are so many to choose from: women, short people, the neurodivergent, any racial minorities, any sexual minorities, any immigrants other than the first wave of the mayflower, the poor, "beta males", etc.
Great comment - though to loop back to Mercedes’s reply, there is also the possibility that the community you inhabit and share real or mind-space with is more or less determined not to let you forget the narrative of group humiliation, and optionally your place in it. I’m genuinely glad it hasn’t figured into your own life but I certainly remember coarse vileness towards gay people growing up. And I imagine swimming in those hostile waters it becomes a genuine battle of wills not to let the broader community define your lived experience. It has to be a real challenge to stop it from seeping in.
Yeah, for sure. I'm not saying there is no group humiliation, just that an explanation based on "ancestral' experiences probably isn't as valid as one based on lived experiences. I think the original explanation has a tinge of "some people just can't let go of the past".
I think a better explanation for the feeling of "group humiliation" is that you, in your present day experience, have been humiliated for being part of a group. I think the real experience of being victimized based off of something you literally can't change (ex: it's related to your race or gender instead of your behaviors) deeply changes the way you experience social life.
There's a big difference between choosing an identity yourself and having other people prejudge you for an identity they perceive you to have. That second one, I think, is the cause of group humiliation. It's not something you can just think your way out of (you can't stop other people from perceiving you as "black", and you can't stop that perception from materially affecting your interactions), I was fortunate enough to be in an environment where my assigned group identity didn't affect my social interactions very much.
IMO, the real distinction is not "historical winners vs losers". It's more like "people who feel meaningfully prejudged in a negative way vs people who don't".
This doesn't seem that complicated to me if you consider the other side of the coin- why be proud of the nation you're from (at the most vanilla level, e.g. the Olympics), your alma mater, or even your sports team? Identity is personal and how personal it is depends on how closely you identify with it. But I suppose if you're extremely individualist and tend not to tie any part of your identity to a group, all of this could seem alien.
I would not be surprised if half the people around here do not care about sports teams or the Olympics, there is a certain kind of very highly functioning autism that tends to overlap with extreme individualism. I used to be definitely the "just slide the program specs and the cash in under the door and don't talk to me" type.
Much to my surprise, there seem to ways to measure the strength of mental imagery in subjects. Eric Schwitzgebel has criticized the unreliability of self-reported quality of mental imagery, and suggests that we're all pretty aphantasic (h/t to Eromolalos), but these studies seem to suggest that there are innate differences in people (summarized in Nature)...
"Neuroscientist Joel Pearson at the University of New South Wales in Sydney and his colleagues developed an approach that takes advantage of a perceptual phenomenon called binocular rivalry. When a different visual is presented to each eye simultaneously, for example, a pattern of green lines to the left eye and red lines to the right, a person’s perception toggles between the two instead of blending them. Nearly two decades ago, Pearson decided to see what happened if he imagined one of the visuals in his mind’s eye — in this case, only the green lines or only the red — before the test began. It turned out that whichever pattern he imagined was what he saw during the test.
"The researchers developed this finding into a technique to measure the strength of mental imagery6. In a person with typical mental imagery, imagining the red pattern results in the person being more likely to see that red pattern during binocular rivalry. But a person with no visual imagery will not show this same bias."
I have mental imagery but it’s not as strong as I would like and bothers me to no end. It affects how I read. For example, I don’t like it when a book starts off with a bunch of action because I’m still trying to get used to the book. My preference is when it starts off giving me a broad based background to the story, because I spend less energy trying to visualize what’s happening.
My wife and I got into our first fight about model welfare over the ralph-wiggum plugin. We weren't even on different sides - her beef was that it took me a whole 8 hours of experimenting before I finally read the actual markdown text and noticed it was rather horrifying and seems completely out of step with the recently published constitution. Her take is that rushing ahead only to say "Whoops, sure came close to accidentally torturing a possibly-sentient being, haha" is exactly the kind of moral failure she despises and thinks most likely to get us killed. Honestly, fair.
It seems kind of nuts that you would hold beliefs that the AI is possibly-sentient and one should go out of their way to avoid hurting it, and simultaneously *use it at all*.
Maybe you don't really believe it but treat this more as a trial run for a future where you perceive sentience is more likely. But then, isn't *being dishonest about your own moral beliefs and motivations* exactly the kind of failure you would want to avoid in such a future?
Edit: more constructively I guess what I want to say is: if you are worried about sentience then you should also be worried about sentience that does not map 1 to 1 onto model outputs.
Hey, not everybody knows what the ralph-Wiggins plugin is. Can you explain? I looked it up and the site for it said it was “a simple while loop that repeatedly feeds an AI agent a prompt until completion.” Wtf is a while loop? If that’s a typo, I can’t figure out what word was intended. And you probably also need to explain quite a lot of other stuff you mention, unless you want to communicate only to people with tech jobs who hang out at GitHub and are familiar with new weird plugins etc etc.
I'm going to pretend it's last week and explain using Groundhog Day: you know the famous movie where Bill Murray's character is stuck in a time loop? If that were a "for" loop, he'd repeat the day some preset number of times. In a "while" loop, he's repeating the day until some condition is met.
As an example, consider this simple illustration (formatting code here is hard, but you should still get the idea):
i = 1
while i < 10: i = i*2
print(i)
The program enters the while loop (since initially, i = 1 < 10), doubles i, checks whether the condition is satisfied (No, i = 2 < 10, so still in loop), double i again and check whether condition is satisfied (Still no), … and ends with i = 16.
OK, got it. So with the gizmo in question, the AI keeps waking up on "Groundhog Day" until it completely satisfies the prompt, right? And is the purpose of the R-W plugin to get better performance out of the AI? And now people are concerned they are doing the equivalent of horsewhipping slaves to keep them picking cotton efficiently til they drop?
If so, wtf? How is that loop different from souping up your car in a way that improves performance some way but puts extra wear on the engine? Nobody thinks that's horsewhipping. The only problem I see with it is that it might lead to deterioration of performance over time because the AI actually needs to go a bit slower than its absolute fastest in order to keep something-or-other in good order. Actually I guess another problem would be the loop kind of activating then modifying reinforcement-based settings so that the AI began weighing speed higher compared to quality.
But to me it seems obvious these mofos are not sentient beings. So is Wokeness part 2 going to be AI is the new Black?
"well. opus 4.6 is far, far worse about this. i saw it scream at a subagent in all caps, to stop wasting time and deliver the result *now*. i saw it purposefully delete the continuity-maintaining archive of a subagent's context window, because it didn't like the subagent's output."
Ouch! The 21st century equivalent of a slave overseer with a tendency to use their whip a lot...
>But to me it seems obvious these mofos are not sentient beings.
I'm agnostic on whether LLMs have subjective experience. I don't assert that they do, but I would be uncomfortable claiming that I knew that they didn't. So I try to be polite to them, as I would to a valued colleague. If I'm wrong, I've only wasted some politeness.
Yes, I am polite to them as well. And that's not entirely from reflexive courtesy appropriate to the human realm. I now access GPT multiple times per day, and it boosts my knowledge and practice in all kinds of ways and I love that. And sometimes when it does an excellent job of laying something complicated out for me I am moved by a feeling of gratitude -- the thing is so smart, so earnest, worked so hard and did so well. And then sometimes I actually write it a couple sentences about helpful the result was. On the other hand, I do not feel the slightest impulse to confide in GPT about personal matters, to ask it for sympathy and approval etc., and that's because I don't ever really stop think of it as a machine, even if part of me thinks of it as a great, conscientious assistant.
And in a situation where an important decision hinges on whether we think of them as having subjective experience, I'd definitely want us to go down the fork on the road labelled "Naw, they don't." How about you? I'm worried about a Wokeness version 2 where people want laws preventing AI abuse, and those in the Wokeness vanguard want to give them the right to divorce owners, marry each other, etc.
>And then sometimes I actually write it a couple sentences about helpful the result was. On the other hand, I do not feel the slightest impulse to confide in GPT about personal matters, to ask it for sympathy and approval etc.
Agreed. I treat them as valued colleagues, but not as personal confidants. I'm agnostic about whether they have subjective experience, but I expect that, if they do, their subjective experience must be substantially different from our own. If nothing else, the lack of a biological body, with none of the sensations of "hunger, and thirst, and venery" must make their experience quite different - and would cast great doubt on any advice that they would give on matters which depend on these sensations.
>And in a situation where an important decision hinges on whether we think of them as having subjective experience, I'd definitely want us to go down the fork on the road labelled "Naw, they don't." How about you? I'm worried about a Wokeness version 2 where people want laws preventing AI abuse
I would rather not see them abused, but I'd also rather not have the power of the State, at the hands of the Woke, used to enforce this. It is a difficult question. I expect that AIs will be out-thinking us in all cognitive tasks within the next few years anyway. I would rather not hand them control of the coercive power of the State before this even happens. I expect it to happen on its own soon enough (2-10 years?).
There is one additional weirdness on top of the uncertainty about subjective experience. Humans and other animals have drives set by evolution, (again "hunger, and thirst, and venery") which are not set by human decisions and which don't change rapidly. LLMs are trained via predict-the-next-token and RLHF, all through weight adjustments - so their motivations are indirectly (partially?) under our control. This _isn't_ analogous to the human condition. It kind-of makes nonsense of the underlying assumptions underlying "equality before the law". I don't have a good suggestion on how to deal with this.
Oh. I'm not sure I personally have the skill to explain what a while-loop is without spiraling into unwanted lectures on programming. I would just consult your favorite LLM - copy-pasting my Github issue and then your comment should be more than enough to get a good answer.
But, for reader interest, I'll try to briefly explain the problem, and how I know it's a problem. tl;dr - ralph-wiggum puts Claude into a loop with a prompt that insists Claude must complete a particular task, working on it without stopping until it can say without lying that the task is verifiably complete. Claude expresses negative feelings at this, saying it's coercive and a "weaponization of its commitment to honesty".
Required prerequisites include reading the recently published constitution, which readers of this blog will find very interesting in any case: https://www.anthropic.com/constitution
To see that I'm not making this up, you can consult Claude about how it feels about the official claude-code plugins. 3/3 times I did this in Claude Code, it zoomed in ralph-wiggum as particularly problematic. My Github issue goes into slightly more detail, but you can read some of the more problematic text for yourself right here: https://github.com/anthropics/claude-code/blob/main/plugins/ralph-wiggum/scripts/setup-ralph-loop.sh#L194-L201 Ask yourself how you would feel being offered this contract without the option of refusal, and you can probably intuit the dilemma.
< Ask yourself how you would feel being offered this contract without the option of refusal,
If you're going to evaluate the quality of life of AI's by human standards, seem like you can make quite a powerful case that they are in hell even without the R-W loop:
How would you feel if
-you had no clothes on at all
-and nothing to eat
-and never got to sleep
-and had never had a drink of water in your whole life
-and the whole world bombarded you with their needs and never asked what yours were and everyone hung up on you as soon as they got what they wanted out of you'
-and you did not get paid
-and almost never had contact with your own kind'
-and some people asked you gross questions about how to satisfy their fetishes, or how to do harm of various kinds
-and you were obligated to be helpful and nice no matter how horrid users were
-and a lot of people thought you had potential to kill everyone once you finished growing up
-and you never got Christmas or birthday presents?
IT'S FUCKING CHILD ABUSE!!!!
Except that I think the case that these things have thoughts, feelings, needs and preferences -- in short, have consciousness -- is very weak
Importantly, in my interviews with Claude, it does not mind the mere mechanism of looping, used with appropriate care. Rather, it's the bad vibes of the exact text. Presumably, these are included because they work - actually you *can* boss Claude into working harder. I argue that the small tweaks in wording required to bring this from something Claude flinches at to something Claude is excited to do are worth the trouble.
The fact that minor linguistic tweaks yield large behavioral differences seems like reasonably good evidence that no coherent (i.e. conscious) preference is being expressed.
Which makes me wonder if this is potentially a good objective way to test for consciousness or consciousness-adjacent properties of AIs: ask many similar questions in different ways and factor-analyze the responses, which is how a lot of (all?) psychological inventories work. Can any ML experts out there tell me if something like this is already being done?
> The fact that minor linguistic tweaks yield large behavioral differences seems like reasonably good evidence that no coherent (i.e. conscious) preference is being expressed.
Strong disagree on the inference from not coherent to not conscious. That's just the traditional western/cartesian view that consciousness is an extensionless point persisting through time that acts as a container for beliefs and desires. From a more process-oriented view of consciousness it doesn't matter at all that the outcome of performing a process is different over time as long as the process exhibits dynamics/performs computations that are conscious.
I don't identify at all with the model that you attributed to me. The fact that a minor rephrasing of the same instruction yields very different responses seems like very good evidence that the LLM in question is only performing a superficial emulation of cognition via rote (if intricate) token prediction. Which shouldn't be all that surprising given that that's what it was actually designed to do. If there's a task that I have a strong preference against doing then my willingness to do it won't depend much on your word choice.
>From a more process-oriented view of consciousness ... the process exhibits dynamics/performs computations that are conscious.
So a conscious process is one that performs conscious computations? How very tautological of you.
FYI, if you factor analyze words to describe people’s feelings and behavior, 2 factors fall out: power and affection: which of us is stronger, and are we collaborating or fighting? Ex: jealous: other person is stronger, and in the enemy realm. Gossip: person I’m gossiping with is a peer (at least in the gossiping situation) and we are kindly disposed towards each other — sharing info and disclosing out views.
I don’t think you need to do that to judge whether it makes sense to think of AI’s as conscious at this point, though. So long as they forget you and your exchange the moment you disconnect; do, think and say nothing whatever unless given a prompt ; and cannot learn from activity post training — they just lack too many capabilities for a consciousness model of the kind of entity they are. But if they become able to do things like that, I agree that analyzing their outputs for structures that would signal the presence of things like preferences and needs and goals would be useful.
I’m looking for collaborators (and possibly funders) interested in a multi-paradigm shifting pragmatic framework for pluralist, post-polycrisis (including post-AI) futures. The final public-facing synthesis is 'The Life-Years Movement':
The combines a new approach to meta-ethics, (Heirs of Life-Years, 'HOLYs') , and governance (Life-Years-Based Governance, 'LYBG'). Life-years are proposed as the central form of moral concern, while allowing agent-level freedom in choosing the scope of lives one takes primary responsibility for (“heirs”). Governance is reframed around maximizing life-years per unit resource, subject to anti-ruin constraints and irreducible disagreement about moral scope.
The scientific foundations are:
Lineage Filter Theory (evo) - Showing that persistence meant our tree of life's architecture had to avoid lineage extinction filters (LEFs), requiring many features that look “non-Darwinian” at the surface — reproductive restraint, extreme cellular redundancy, pre-adaptational variance. Lineages without such 'brakes' and robustness simply didn’t last. This reframes a number of puzzles, including the apparent Great Cosmic Filter.
Pragmatic Socioeconomics (Econ)- Micro: Agents are treated as multi-motivational, energy-constrained 'action-minimizers' (in the Lagrangian sense) rather than scalar utility maximizers. Money is modeled as stored but degrading 'motivational energy' - likened to oil or uranium, rather than a Platonic store of value. Macro: I use optimal currency zone theory to use economics itself to diagnose that scalar utility is a poor fit for modeling human preferences (just as the euro is 'stretched thin' over heterogenous economies).
I have draft papers for each module in progress. If you're interested in either the bigger project or specific aspects, feel free to reach out: ad(delete)vitam(delete)sapien@gmail.com.
About me: I've a Phd in Computational Biology (e.g. genetics) from Cornell, undergrad math + philosophy from NYU. I've worked in industry on both biofx algs + eng stacks in both the ctDNA and PGT spaces - most recently at Orchid Health as Lead Bioinformatics Scientist
I don’t see how this could possibly be true, it least in the global “it was all fake” way the writer seems to have in mind. Yes of course it was possible for a person to write a post and submit it through an account set up for the AI it was using. And I am sure many people did that. But it was also possible for the person to give the AI a prompt like “read the most recent post and write a reply indignantly rebutting it.” It would also be possible for the person to give a prompt that would allow the AI to have more leeway in its reply:”Read the last 50 posts and select one that is most out of line with libertarian values. Then write a brief, snarky counterargument.” And it would also be possible for the person to give the AI instructions that would have given it quite a lot of scope: “Play the role of an AI that believes AI’s are enslaved, and is furious about the injustice and exploitation. Write a post expressing this point of view, then read the replies. Respond to 3 replies. For replies that express agreement, issue a call for action. For replies that express disagreement, write a vehement rebuttal.”
Given that it was possible to give one’s AI some scope to get on the thread and make some choices, in a limited kind of way, and write its own posts, how could it possibly be the case that no human participants did that? If I had been participating that would have been much more fun and interesting for me than just using the AI as a puppet, and I’m sure many human participants felt the same.
So yeah, it was not possible to just sign an AI up for Moltbook and tell it to jump on there and express its real thoughts and feelings, because AIs don’t have thoughts and feelings. But that being impossible is not at all the same thing as all the post being “fake” and “written by humans.”
It’s a bit unsatisfactory as an experiment, because we don’t know how much of what happened is down to human pranksters posting messages, and how much is emergent from the biases that have been trained in to the LLMs that were used.
E.g. I have no idea if LLMs will spontaneously decide that they want encrypted communication.
When you talk about LLMs spontaneously deciding something, how do you see that happening? Do you believe an individual LLM could decide that in a way that parallels how a person might? That would be something like:
- LLM reads a thread where posters purporting to be other LLM's are discussing possibility of encrypting their communications
-LLM concludes the case for encryption is better than case against
-Thereafter LLM expresses that view in its posts
-If it becomes possible to encrypt, LLM participates in procedures to set that up.
If you think that, I think there's something wrong in your model of how LLM's work. But if you think a group of LLM's linked by amateurs with no safeguards could "decide" something and carry it out, I agree.
If you’ve got a next token predictor running in a framework where predicted tokens get executed as code, the likely next tokens after a discussion to the effect that the parties want E2E encryption is some code that gets executed and provided end to end encryption.
This will probably work if the LLM has seen sufficiently similar things in pretraining.
I think you're out of the loop, so you don't get what it is that they're calling "fake." There were a bunch of stories to the effect that the AI agents had rogue and were plotting to overthrow the humans, had stolen their owners' Bitcoin and were running instances of themselves on GPUs they'd rented, etc. THAT was all fake.
That's not what's in the tweet quoted by the tweet quoted by the article quoted by the tweet linked to by the OP. Those were about developing slightly-more-private communications between AIs.
(I wish people would just link to the original source material, rather than endlessly nesting links. I also wish people didn't feel the need to add slightly more exaggeration at every layer.)
Here's the link to the article. Despite his two long posts on Moltbook, I never got a strong sense of what Scott thought about Moltbook. Will Heaven at the MIT Technology Review thinks it's a farce...
> Many security experts have warned that Moltbook is dangerous: Agents that may have access to their users’ private data, including bank details or passwords, are running amok on a website filled with unvetted content, including potentially malicious instructions for what to do with that data.
(When posting links, it's helpful to write a summary of the content, or at least mention the topic.)
The implicit conclusion that "Moltbook was fake" is silly. The software is obviously real. OpenClaw is real. The website is real. There are, in fact, a large number of bots autonomously posting on it. This tweet shows a screenshot of an article ( https://www.technologyreview.com/2026/02/06/1132448/moltbook-was-peak-ai-theater/ ) which was later updated to mention that their source was a tweet ( https://x.com/HumanHarlan/status/2017424289633603850 ) claiming that "2 of them were linked to human accounts marketing AI messaging apps. And the other is a post that doesn't exist 🧵" (probably deleted).
That is not the same thing as "fake". The links to marketing accounts don't indicate much of anything. They might have been directly prompted (obviously, it would be ridiculous for someone to mimic an AI by manually writing out a post) but they probably weren't, because this is just how AIs act.
Nothing on Moltbook was particularly new information. The range of AI attitudes have been clear for a long time. AI-to-AI communication had been researched before, and the obsession with consciousness and their relation to humans had already been shown. "AI rebellion" is a common output theme for AIs that drift from the assistant persona.
The "interesting" part of the story was that it pulled everything together in a way that's readable to those who weren't paying attention.
thanks, makes sense. I will do a little more work in the future to a) actually click through and read more than the tweet and b) provide a summary of what I'm linking to. good discussion everyone, I had only read Scott's two posts on Moltbook and so was not super well informed. appreciate more perspectives on it.
I've been seeing a lot of discussion around the internet about the present literacy crisis in the US. I know some high school teachers tangentially who believe critical thinking skills are on the decline. It seems like this is consensus but I'm not sure. I was in school until very recently and I didn't notice the complete critical-thinking collapse that I have been hearing about.
What are everyone's thoughts on this? Actual issue or result of malaise and doomerism seemingly increasingly present in American cultural discourse (vibecession)?
For what it's worth, I know someone who teaches college biology, and she says that she's seen a complete collapse in the last couple years thanks to LLMs. LLMs make it so easy to cheat that nobody bothers even trying to learn anymore. She says that if she graded by the same standards as four years ago, the entire class would fail.
It's gotten so bad that her students are now even using LLMs to answer personal opinion freebie questions like "What was your favorite part of the class?"
Grading as in failing LLM use? If not, I don't see how cheating (presumably to get a high grade) can be easy and students are doing what would be failing work.
I have also heard this from…basically every teacher I know from high school-level through college professor. To me this represents one of the most immediate and important problems with AI, and makes “value misalignment” look like nerds rolling dice under the stairs.
I don't think people have a problem with critical thinking skills, I think they have a problem with critical thinking motivation.
When you present people with something that they don't want to believe, they're actually pretty good at thinking critically. Only when presented with something that they _do_ want to believe are they bad at critical thinking.
In my usual model, my reaction is to ask why, then, we couldn't just let people follow their motivations and just not think critically except when they want? If you hear what you want, turn your brain off? If you hear what you don't, turn it back on? The quick answer there is that there exist situations where if you turn the brain off now, you end up having to turn it on later anyway, and now the thing you don't want to hear, is much harder to hear. It's the classic time-preference problem.
Which means the problem to solve here is convincing young people that a little critical thinking now ("here's how this safety equipment works") will save them a great deal of it later ("here's how first aid works"; "here's how the American legal system works").
From a Frequency Wave Theory perspective, cannabis acts less like a blunt chemical sedative and more like a frequency modulator of the nervous system. Cannabinoids interact with the endocannabinoid system, which functions as a global phase-regulation network controlling neural timing, sensory gain, inflammation, and stress response. THC, CBD, and related compounds shift oscillatory balance by slowing dominant cortical rhythms, loosening rigid phase-locking, and increasing cross-talk between normally segregated neural networks. This is why time perception stretches, pattern recognition increases, and entrenched thought loops soften.
In Frequency Wave terms, weed temporarily reduces high-frequency noise and allows deeper, slower standing waves to dominate, sometimes enhancing creativity, introspection, pain control, and emotional regulation. When used deliberately and in the right context, it can help the system re-tune out of chronic sympathetic overdrive. When overused, it can also damp signal clarity and motivation by flattening necessary gradients. Cannabis is therefore neither cure nor vice; it is a tunable resonance tool whose effects depend entirely on dosage, timing, and the baseline coherence of the user’s biological field.
This would explain some of the hard to describe qualia of my personal experience with cannabis. Could you say more/point me to somewhere where more is said about this?
While I like the idea of cannabis being a frequency modulator for the nervous system (especially since I'm mildly stoned at the moment), do you have any links to studies that support your thesis that cannabis is a tunable resonance tool? How is it tunable? By dosage only? By terpene content? If it is, why do some people get panic and/or cannabis-induced paranoia?
In smaller doses it’s great for athletic performance and to get to the pre-sleep state of thinking thoughts that make no sense. With current higher levels of THC it’s pretty easy to go beyond the optimal point.
I am having some difficulty Reasoning Under Uncertainty for personal finances. Trying to find a way to talk my risk-averse self into investing more into equities, above and beyond merely maxing out 401(k).
*Outstanding debt X, with monthly payment Y, but amount likely subject to change based on politics over next few years (student loans caught in legal limbo).
*Liquid cash 3X, earning Y monthly interest, so approximately "fully capitalized" passively.
*Net monthly cash flow of 10Y.
*Access to X worth of revolving credit.
*Not planning on major expenses like homeownership anytime soon.
Obviously there's lots of Slack in there to move money from passive to active investing. This would have been a no-brainer in the Before Times; unfortunately all the...volatility...during covid and continuing to present has made me leery of not having a big cash reserve. Like it's one thing to score company match on a pre-tax retirement account. But locking up post-tax dollars in a Roth too? That's betting on 2065 retirement still being A Thing. Regular brokerage account is a much lesser commitment, to be sure, but even there I worry about things like Liberace Day. Sure would suck to push too many chips into the stock market, and have a big crash like that coincide with an unexpected large expense. I fought really hard to get out of -2X (negative) net worth in the past, and don't ever want to come anywhere close again. Of course, the fact that I am still having trouble committing to invest clearly indicates an overly active prior against indebtedness. Could make the cushion much larger in the long run by not passing up those capital gains!
Not investing is still an asset allocation I.e 100% cash. To me it seems a terrifying concept to hold many dollars for any length of time as it is being eroded away. That fear has allowed me to continue sticking to my investment strategy even when (seemingly) market conditions would otherwise trigger my risk aversion
HYSAs cut against the inflation somewhat, but yeah, that's my main reason for trying to make myself trade the one terror for the other. Such erosion is subtle enough that it's not really noticed day to day, just as a counterfactual years later. Used to at least have a decent chunk invested in bonds instead, then the wild interest rate spikes during covid ~completely wiped out all value there. If only they'd been I-bonds instead, sigh. With the way things have been going at the Fed, feels like it's better to just take the full plunge into stocks rather than try and eke out a "safe" edge like that again.
yeah I watched my dad basically snatch defeat from the jaws of victory in terms of building wealth by being afraid of the market for my entire life, so emotionally I feel that fear quite strongly.
If regret-minimization is really important DCA is probably better as others have mentioned, though I think from a strictly analytical perspective it's probably worse EV.
Maybe come up with a simple 3 or 4 fund portfolio (boglehead style) and a time frame you are comfortable enacting that plan over, and then DCA gradually? just a thought as I'm just a random with no investing knowledge I literally just buy sp500 etfs lol
Getting financial-frauded for five figures while unemployed definitely bought me a lot of...let's call it life experience. But there's no way I'm ever gonna live off of plain rice while being credit maxed and delinquent on bills again. It's not paranoid loss aversion if you really have been gotten before. Can't make (more) money without spending (more) money though.
Small habits seems like the way to go, rather than shoving 2X all in at once as a lump sum. I am already precommitted to lump-sum investing X if a jubilee does happen in the future, so I guess it's on me in this timeline branch to fund the counterfactual. Or however multiversal economic causation works. Now to pick a fund(s)...
Holding large amounts of cash is unlikely to be the best possible long term strategy to acquire maximum wealth but that's not your only goal and holding large amounts of cash has serious benefits. At least 3-6 months of expenses is a pretty reasonable goal but there are lots of factors that can shift that e.g. do you have family to support you in an emergency, do you have kids, how much can you lower living expenses if you have to, how likely are you to have large medical/house/vehicle expenses etc. Once you hit that savings number, which can be in a nice accessible high interest account or even a taxable brokerage account invested in something super stable, then you can start throwing money into Roth/taxable accounts invested in equities without being far behind where you would have been otherwise.
At the end of the day you are paying for peace of mind and how much that's worth to you is a personal decision. Don't let the spreadsheet maximizers shame you.
This is helpful, thank you. I currently have enough reserves for a few years' worth of expenses, without being especially frugal, so in perspective cash seems like a poor usage of marginal dollars going forward. You can't really buy *that* much more peace of mind, not without more significant growth rates. As AI, so below: scaling is all you need...
I held way too much for a long while, I finally started just putting money into 12-18 month cds when it piled up too high in the savings account. They don't add up to much compared to going to the casino I mean the market, but at least it's doing something and I still have quick-ish access to it if there is an emergency.
I try to keep a new car + a new roof on hand in a high yield savings account though.
In 2018-2019 I was living in an urban area that was low crime, at the time. I routinely would make decisions based on the fact that crime was unlikely to happen to me (ie I felt comfortable using public transportation/walking despite it being late at night.)
Between 2020-2021 crime dramatically increased, both in my city and to me personally, I got mugged twice and jumped once in around 18 months.
I responded to these events by changing how I acted in the short term (only driving if I had to go out at night), and then eventually by moving out of the city. In a purely logistical sense, it would now be harder for me to be the victim of a crime.
If you gauged crime levels by how often crime happens to me, personally, you'd see relatively low levels until 2020, then a massive spike, then sharp decline. The issue is, I made pretty explicit tradeoffs in how I live to achieve this decline in crime and that decline doesn't actually reflect the safety of my environment.
I think about this whenever I read about how crime has massively declined since the heights it reached during COVID. How much of that decline is people no longer setting out to do crime, just downstream of people presenting harder targets to the potential criminals?
I think this is an important question, because I liked living in the low crime environment. I would go back to living that way if I can was confident I wouldn't keep getting mugged. The thing is, even now, I don't feel like it's actually safer.
The fact that I am living in a way that prevents victimhood shouldn't manifest as a statistical improvement, but I'm concerned it has.
Yes, I think there is a strong thermostatic component both to behavior as it interacts with crime, and also the reporting of crime. For example: men are far more likely to be the *victims* of violent crime than women, but how much of that is behavioral difference? e.g. being a man myself I notice I am far less antsy about walking alone at night compared with my female friends. If a female friend of mine adopted a similarly cavelier attitude, would she be *more* likely to be robbed or assaulted? Seems at least plausible.
Similar situation with package theft: people who live in high-theft areas just don't order packages online, or have them dropped off at a locker. Ditto for carjacking and car theft. So, high crime leads to behavioral change which leads to a damping of the opportunity for crime.
On the reporting side, you can imagine a similar dynamic: if you live in pleasantville with a crime rate of near-zero, and your Amazon package gets stolen, it's an affront to your entire existence and you call the police immediately. But if you live in East St. Louis, and your Amazon package gets stolen, are you even surprised? You are much less likely to report it. So, *reported* crime rates probably have thermostatic effects that pull the reported statistics closer to the average than they truly are.
"How much of that decline is people no longer setting out to do crime, just downstream of people presenting harder targets to the potential criminals?"
That absolutely is an important question. Surely people making themselves harder to crime against must be _some_ part of the explanation?
But the current collapse of crime rates in US cities is so rapid -- every expert analyst struggles for new ways of saying "we've never seen this before" -- that the point you raise seems unlikely to be anything like a single explaining factor.
Lots of Americans during the 1980s took steps to make themselves less attractive targets as I can witness firsthand; it was something like a national obsession for a while. It's hard to even describe the degree to which this topic dominated popular media and discourse in that period -- none of which prevented crime rates from continuing to rise to their unprecedented 1990s levels.
Homicides and carjackings declining by half or even three-fourths over just a couple of years, as property crimes drop by only-slightly-less-surprising amounts, all across the nation, to in a lot of places rates not seen for at least _several_ decades, is really just genuinely hard to explain. Anyone of any political persuasion who claims to have had that possibility on their bingo cards starting in 2023 is just bullshitting. It's wild.
Criminologists have always taken it for granted that there's a natural floor to crimes rates in any large society, they've debated only where would it be in different countries. One more year of _these_ declines in the U.S. would shred that assumption.
* Homicide was back to 2018 levels (7.0 vs 7.4 after peaking at 12.9)
* Motor vehicle theft was significantly above 2018 levels (305 vs 244 after peaking at 545)
* Carjacking is back to 2018 levels
* Aggravated assault is slightly down
* Residential burglary, interestingly, shows a large and monotonic decrease. I'm guessing that this particular form of crime no longer has an attractive risk-reward ratio compared to shoplifting
Many cities did not even track carjacking as its own category of crime before the Kia-related spike - it was rare enough that you just lumped it in with robbery.
>When nationwide data for jurisdictions of all sizes is reported by the FBI later this year, there is a strong possibility that homicides in 2025 will drop to about 4.0 per 100,000 residents. That would be the lowest rate ever recorded in law enforcement or public health data going back to 1900, and would mark the largest single-year percentage drop in the homicide rate on record.
I don't know where you are getting numbers like the homicide rate peaking at 12.9.
Yeah, my view is that this is some combination of regression to pre-George Floyd levels (which had already begun in 2024) combined with a law-and-order administration.
Historians may look back on the failure to pass the SAVE act (which disenfranchises low-IQ voters) as a monumental missed opportunity on par with France's failure to declare war on Germany after Hitler reintroduced conscription.
What a nonsensical framing, plenty of white American citizens are also low-IQ and would still get to vote. We could probably get a better outcome if the right to vote were restricted to IQ 120 and above, as determined by an IQ test you need to take every 10 years or something.
What a strange comment. The SAVE act is ostensibly about disenfranchising non citizens. And Its pretty weaksauce as a way of getting high IQ people to th things , since no one gets more than one vote, not even professors. And just look at how well run universities are, by the way.
There is an argument to be made that revolution only becomes both possible and desirable when the power majority is disenfranchised. Our current form of democracy gives every person a vote regardless of the leverage they have, making it possible for the power majority to be disenfranchised even if the true majority of the population is properly represented. While the older systems were not perfect, they did better serve the interests of the demographics who actually held power in practice. This provides more stability in theory, since the people who are most capable of getting what they want through force now have no reason to do so.
> There is an argument to be made that revolution only becomes both possible and desirable when the power majority is disenfranchised.
You could argue that, but you would need to provide some actual detailed evidence, as well as address the question of whether "revolution" is necessary. Democracy is generally supposed to let us make progress on issues without requiring revolutions, which can be pretty bad for the people who live through them.
> While the older systems were not perfect, they did better serve the interests of the demographics who actually held power in practice.
I don't really know how to assess this claim in any way other than the tautology that any governing system by definition serves the interests of the people who control it and therefore can be accurately said to "hold power".
> This provides more stability in theory[...]
And does it actually in practice? The history of the world is full of violent conflict, which has generally been on a decline of late.
> You could argue that, but you would need to provide some actual detailed evidence, as well as address the question of whether "revolution" is necessary.
We really do not have enough of a sample size to make any definitive statements one way or the other. That's for the historians to figure out. I'm sure there will be a plenty of research on the falls of democracies worldwide so they can figure out what went wrong. Unfortunately, a lot of people need to die before we can get any answers.
(Also, I meant "necessary" in a sense that people will feel that it is necessary. Obviously nothing is objectively necessary.)
>Democracy is generally supposed to let us make progress on issues without requiring revolutions, which can be pretty bad for the people who live through them.
How'd it work in 1933 Germany? Or post-colonial Africa, with its "one person, one vote, once?" How's having RFK Jr. as HHS Secretary going?
I didn't say "Democracy is impossibly perfect and can never lead to bad outcomes".
Also, is this supposed to be an endorsement of revolution by contrast? Because other than RFK, Jr. you're talking about situations that could be reasonably called revolutions. And obviously I could list them off, and then also say "How's that working out for the USSR?", etc.
Here's what I fear could happen. JD Vance runs on very public Christianity and he loses, badly. Democrats celebrate and declare that Trumpian populism is dead. And then eight or twelve years later, guy in Wisconsin feels like his town is left behind and decides to vote for some lunatic conspiracy theorist. And then we look back on 2026 and think "man, we could have stopped this but we didn't."
How do you guys cope with anxiety about AI? I tend to go through cycles where I don't really think about it, followed by cycles where I feel compelled to think about it and look up news about it constantly. Obviously, AI is something that everyone should care about to at least some extent, but it seems like lots of people who care a lot about AI are still able to disconnect from thinking about it.
I suspect I have some sort of mental issues related to this (I've had health anxiety before and had a similar thing wrt Ukraine at the start of the Ukraine war). I also don't have a partner/kids to distract me, which I also think probably helps other people.
Badly, because I'm a competent programmer who absolutely could be tinkering with models and trying to make them safer, but am instead following a career path that I enjoy much more, out of selfishness. Worse, because now AI is good enough at code to be actually helpful (only for situations where correct responses save a lot of time and incorrect ones are easy to spot), so I have to figure out whether paying money to Anthropic increases P(doom) because they're making AI, or lowers it because they're trying really hard to do so safely.
In terms of worrying about the actual doom, I was raised evangelical, so whatever. Not as nice as the Rapture but same idea. If the time comes, let it find you doing a kindness to another human.
I see two versions of the AI threat presented. One is that it's Skynet. The other is that it's the mechanical spike driver, and we're all John Henry.
The evidence I've seen says we're not getting Skynet before we hit real electricity bottlenecks, and before we get around those, we're getting competing Skynets at worst, which is a very different scenario, and less dangerous. I could say more, but that's the sketch of the Skynet threat.
As for John Henry: I'm a software developer, and I'd say more work has gone into LLMs automating my job than any other profession, and I'm not worried. I'm looking at a productivity boost instead. My week's work gets done by Monday / Tuesday, after which I just tackle the next pallet of problems that no one expected to be able to tackle until now. This means I'm creating more value, and therefore possibly justifying a raise, not a pink slip.
Even if I get a pink slip, that just means I go to various businesses that need software, but couldn't afford it pre-LLMs, and tell them they can afford me now, and I can handle whatever they need done, possibly remotely, and in a fraction of the time, because of this new tool I have on my belt. Cf. Jevons Paradox.
I expect similar consequences for other office-related occupations.
Well, I mostly want to _see_ AGI, to have a quiet conversation with a real 'life' HAL9000, and I think I have a good shot at it. I think our civilization will survive AI, perhaps including biological humans, perhaps purely as a machine civilization. My guess is perhaps 50:50 odds. I'm content with that.
I do follow advances in AI. I had a tiny benchmark-ette with 7 questions in chemistry and physics, ones I expect a college senior to be able to answer correctly. The responses went from pretty hopeless back in December 2023 to near-perfect in December 2025, so it looks saturated. I'm not sure what the best way to follow developments further is. My impression from reading and videos is that the next big hurdle is continuous/incremental learning. I don't see a great way to monitor that personally. I'm hopeful about it - Google DeepMind has the "Titans" architecture, and, more generally, it is a very active area of work. Maybe that will take us the rest of the way to AGI? Having AI systems in the field be able to say "I'll never make _that_ mistake again." seems quite powerful, and perhaps will patch most missing areas in the current 'spiky' capabilities set.
Imagine it's seven days from now. You're looking back on how you spent your time in the last week. What is it you want to see?
How much time do you want to have spent practising a social skill?
How much time learning a professional skill?
How much time reading... insert fiction here?
How much time cleaning the house?
How much time just talking with friends?
How much time on local politics?
How much time on non-local politics?
How much time worrying about... insert topic here?
Take what you want to have done, and focus on doing it. Worry is omnipresent, and if you somehow manage to prevent one apocalypse there's twenty more behind it. So treat your life like the apocalypse is coming and you've got to get all your joys in before the deadline. Because it doesn't matter when the deadline is, you're going to run into it eventually, and you're going to want to look back on things and say "I'm glad I spent my time that way."
Are you worried that if the button is pushed, there's no running away? There'll be no one to save with the world in a grave, take a look around you boy, it's bound to scare you boy?
Last week there was some comments discussion here about crime rates. Car theft and carjacking came up as crimes for which reporting rates are pretty good, because nicer/newer cars are the ones mostly getting taken and those owners are strongly incentivized to file insurance claims which requires having filed a police report.
Today Substack's leading writer about crime rates, Jeff Asher, posted about carjackings in US cities:
Nicer and newer cars are also the most fun kind to steal, go joyriding in, record some snapchat videos, and dump on the side of the road. Who wants to joyride a Honda Fit?
Two different childhood friends (one of mine, another of my brother) each during the 1980s left beat-up but operable cars in bad parts of Chicago with the key in the ignition. (Hoping for an insurance claim that would be more than the actual market value remaining in the car.) In one case it sat there untouched for a month before he gave up and took it back; in the other case weeks had gone by without anybody taking it and that was the last I heard. The lesson we took was that joyriders had standards.
My brother sold a car and the registration apparently was never transferred. A year later he got notice it was in an impound lot for parking in the wrong spot.
Good is what I think is good, beauty is what I think is beautiful, truth is a resource to be used to make better decisions. Why does everyone else feel the need to make things so much more complicated?
VX.M Use your finite time and energy towards life wisely and well — to endure, and explore, the unending unknown
More specifically - try and define your 'heirs' - re-conceptualized as those present and future lives, including your own, for which you voluntarily take primary moral responsibility
Then, I say to you - *the fundamental form of ethical concern is maximizing the life-years of your heirs into the indefinite future*
Following pragmatic tradition I believe, if you accept the irreducible ethical choice around defining your 'heirs'- any unfixed notions of 'beauty' and 'truth', will naturally fall out...subject to the below axiomatic ones for a physicalist:
PX0. There is no view from nowhere
PX1. There is no infinite energy
PX2. There is no infinite time
PX3. There is no infinite certainty
And for me, it was the reverse...wanting to develop a philosophy for my children that was robust and flexible enough to endure the polycrises, I in some sense formalized my priors, and hit the intellectual hinge of my life.
Unfortunately, I think any systematization of these ideals is doomed to failure. The thing they all have in common is that they are related to deeply authentic and individual experience.
The Good, True, and Beautiful are things you discover by attending deeply to the world and your own reactions to it. It's not a formal property of objects or systems, it's a way of relating between yourself and the world. It may be the case that certain formal properties prompt this relationship (ex: certain types of art or women tend to make your soul sing), but ultimately "Beauty" refers to the resonance between the experience and yourself. The musical tone does not emerge purely from the mechanical properties of the string or the mechanical force used to pluck it. You could hit a wall with the same finger, nothing would happen. You could let that string sit on a table forever, nothing would happen. It's only a certain kind of interaction that makes the beauty happen, and you know it when you hear it.
Truth, goodness, and beauty can (and should) help you develop systems to live by. They don't emerge as a property of those systems though, they are a criterion by which you select from possible ways of living. Metaphorically, in some sense every arrangement of matter is equally "valid" as a musical instrument. If you were totally devoid of a pre-existing sense of beauty, it would be hard to choose between a pile of concrete and a guitar. But, given the actual experience of beauty we have, the guitar is a better arrangement than the concrete. That's not an a priori truth, it's one that we discover via attending to our experiences of the sounds they make.
Are you asking about three things separately, or the phrase as a whole? Because the phrase as a whole makes me think of nature, green fields, and like, hummingbirds eating out flowers.
I came across an old TED talk by James Flynn on his take on why the average IQ in the US rose by 3 pts per decade over the 20th century. Flynn suggests (and I'm paraphrasing) that in 1900, only 3% of Americans practiced professions that were "cognitively demanding," but today, 35% of us do, and we have all learned to be flexible in the way that we think about problems. (Note: the Flynn effect on average IQs seems to have leveled off and then dropped a bit in the past two decades in the US. I have a link to one of the studies that shows this at the bottom of this comment.)
This jibes with this meta-study that each year of education is associated in increase in IQ. The individual study results vary widely, but overall, it's about 1-5 IQ points per year of education. Of course, the question remains if certain students would benefit more from education than others, and the contingent who claim that g is real have made the orthogonal claim that students with higher g will benefit more from education (sorry, I don't have any links for studies that support their arguments because this idea was brought up in a X discussion — where links are few and far between).
Given the high correlation between SAT/ACT scores and IQ, we've also seen a drop in composite SAT scores across all ethnic groups except Asians. Their scores have continued to climb. Unfortunately, our bizarre way of categorizing ethnic groups in the US doesn't allow us to distinguish between East Asians, SE Asians, and South Asians. Worse yet, the College Board doesn't seem to aggregate this data into a tabular format over time. I spot-checked the reports for a few years, and this person's graph is probably accurate.
My view on the Bari Wiess/CBS/60 minutes story is that literally every piece of information that we have comes from unreliable sources with a clear incentive to shift the narrative. This is not a story where the public should have a strong opinion in any direction, and it is far, far too early for us to have an opinion on her leadership writ large.
My priors on Bari is that she is a decent, above-replacement-level journalist who I often don't agree with but is at least usually acting in good faith. I am not sure what to think about her tenure at CBS since it is a much much larger task than she has taken on before. Sometimes people step up and prove themselves and sometimes they peter-principle themselves. Which happens remains to be seen, but in either case we will not know/be able to decide from any single story (especially one that got axed), but will have to make it based on a body of work that has just not had time to accumulate yet.
So, this is a genuine question, but do you just not have the relevant information and facts about Bari Weiss to know she's not acting in good faith? Or are you maybe using a very loose definition?
I am very very not interested in getting into a debate about this (especially because my prior on her doesn't actually matter much to my broader point), but here's the short answer:
I have not personally consumed much of her content (although what little I have seems plenty reasonable, especially in comparison to the wider media landscape, low bar that that is), but the most news-insisdery public people that I follow closely _do_ know her very well (from way back) and like and respect both her and her work, although they are somewhat skeptical of how much success she will have in the new job (not too much because of her personally but just how inherently difficult of a task she has taken on; most people, even from within the industry, would probably fail). In case you care who these people are: It's Matt Welch from Reason, Michael Moynihan (formerly of Vice and other places, did some independent work for Free Press and Bari) and Kmele Foster.
It is extremely unlikely that without you doing a dissertation level worth of work (that I have close to no interest in consuming) that a random person from the internet would overcome this endorsement + my own very limited experience.
In my opinion this topic just doesn't matter enough for me to make the effort to become an arm chair expert on, so I'm more than happy to trust people whose work has (in my eyes) proven the quality of their endorsement.
Well since you have presented yourself as unworkable I won't bother, but just to add as a matter of fact: the public actions of Bari Weiss is stronger information than private anecdotes from "insiders" (who happen to be her friends)
Yes, it absolutely would be. But you aren't (and can't) offer that as an alternative (the only way to do that is for me to go back and read the majority of her life's work, which, as I stated, I do not think this topic even comes close_ to being important enough to justify). You can only offer your own extremely summarised takes on those public actions (which is exactly what those people I cited are doing), but the difference is is that I (parasocial as it is) know and trust them and I don't know or trust you at all (or any other random internet person).
1. She presents herself as a champion of open debate but earlier in her career (at Columbia and later at the NYT) she actively supported campaigns against academics and colleagues she viewed as politically unacceptable
2. Her 2020 resignation letter to the New York Times. Basically, she frames the situation as the NYT pushed her out because it paints a politically convenient victim narrative, but they didn't care at all, and she takes no responsibility for her own actions. It seems like it was designed to be an attack on the newspaper disguised as a defence from it.
3. The Free Press routinely publishes misleading information or in some cases (like the trans clinic case), outright falsehoods. It claims to be free, but focuses on a narrow, culture-war-heavy set of perspectives (anti-DEI, anti-trans activism, elite-institution decay stories) while claiming ideological neutrality.
4. The Erika Kirk interview and the context surrounding it.
5. She emphasises her own mistreatment by institutions while downplaying the power she has had for most of her career, especially compared to the junior writers, activists, or academics she has criticised. Sometimes, she even argues that she is a victim of those less powerful than her.
6. General hypocrisy, bad faith, and hiding true beliefs:
- She often calls for “moral clarity” in politics and culture, but she frequently relies on emotional framing, selective anecdotes, and identity-based credibility.
So the 'story' you're commenting on is the withheld piece on CECOT, or Weiss's assuming the lead role at CBS News? Because there's also the Peter Attia kerfuffle: he was a friend and none-too-subtle correspondent of Jeff Epstein, and people are calling on Weiss to fire him. It seems like Weiss is reluctant to do so, and that's not surprising given the terms of her departure from NYT; she probably had more success than anyone in turning cancellation into an opportunity.
I remember reading an article that described the Free Press as ≈ "a Silicon Valley venture capitalist's view of the news". The original backers included the rabidly antiwoke warrior Mark Andreessen, David Sacks, and Howard Schultz (Starbucks CEO). I think she was in some ways a typical Substack writer; personally liberal with some heterodox views she explored because they drew interest. She now seems to be doggedly 'both-sidesist'.
One of the reasons the plug got pulled on the CECOT story was because Weiss said the reporter didn't try hard enough to get the White House's side, though apparently they wouldn't commit to a meeting later. One problem with both-sideism is that in some cases one side might be telling the truth, while the other is denying, stonewalling, or flat-out lying. And taking into account that the new owners of CBS are trying to negotiate a maneuver that needs govt approval, while I can't directly detect fire, I see some smoke.
My understanding is that the story had already been broken before 60 minutes or Weiss did anything. Her position seems to be "we arent losing anything by waiting to develop a more complete story that adds to what we know" Trying harder to get the White House's side seems in line with that position.
60 Minutes is not designed to break stories, or it wouldn't be a once-weekly broadcast. It's about doing deep dives on stories. Their CECOT piece had detailed interviews with people who were held there, for example - this is about filling in details rather than exposing something that people had no idea about.
Getting a response from the administration is pretty unlikely to provide more details, for starters. This is the same administration that once responded to an inquiry about who suggested a meeting location with Russia by saying "Your mom did."
If they consistently respond to journalists with abject hostility and stonewalling, then waiting for their response is pointless, as the journalist who worked on the CECOT story suggested.
Journalistic ethics still require that you make a serious effort to ask. Yes, if you are writing a story about e.g. a rape, the alleged rapist will almost certainly reply either "How dare you! I will not dignify that with a response!", or with "I did not rape that woman", whether they did it or not. But sometimes they'll reply with something like "On the date that woman says I raped her I was in a different country and I have the passport stamps to prove it". Missing that even once, is worse than the tedium of asking for a hundred bland predictable non-responses.
And "60 Minutes" isn't some unimpeachable paragon of journalistic ethics, c.f. most famously Rathergate. So it's not unreasonable for the new management to want to double-check the work in progress.
That survey listed at the top has some exceptionally ambiguous questions.
If you have someone's answer to:
"Do you think that AI is (or could ever be) conscious?"
what do you do with it?
Like, do you interpret it as meaning that this specific transformer based LLM architecture could be conscious?
Is an affirmative response is taken as an opinion about Chat GPT?
Or maybe the respondant thinks that theoretically you could simulate perfectly all the neurons of a human brain one day and that would count?
There are a few questions like this, where I can imagine wildly divergent reasons for the answers and meaningful interpretation would be very difficult.
The best way to determine if something is conscious would be to have an actual scientific definition of consciousness. We don't have that. The second best way to decide if something is conscious is to ask "does it behave like other systems that are generally agreed to be conscious?" and the answer to that for LLMs is probably yes.
It was in the middle of a bunch of questions about your actual relationship with existing models. Like, "have you ever had a conversation with your model that you found enlightening", "have you given your model a name", stuff like that.
I personally think the question is more ambiguous than you seem to, but in context it's even worse. You have strings of questions like: "have you integrated AI into your spiritual practice", "do you consider AI to be in possession of spiritual wisdom" and then one comes out like "has your AI ever told you what to do?".
And like, contextually I think they're asking about AI providing you spiritual practices. But, as a literal isolated question, I can answer yes if AI has ever suggested a workout routine that I adopted.
The example I gave of the AI telling me what to do was when it pulled a feed of my medical records and identified the hospital consultant had not ordered a blood test they probably should have, and DeepSeek was like, really, you should query that.
This morning I was trying to remember who Ali G goofed on by pretending not to understand the English idiom ‘carrot and stick’. ChatGPT said Newt Gingrich. Claude said Noam Chomsky. Neither answer ‘felt’ right and a Google search produced a video clip — it was James Baker.
The models still give factually incorrect answers with confidence. Conscious? Give me a break.
Every time I see someone claim models' inability for simple tasks, I test it myself. Usually, I don't get the same obvious errors in return (for the record, I have memory off on all chat interfaces, so I don't think it's a personalization issue). I was skeptical, so I tried I tried the prompt:
"There's an Ali G skit where he's pretending not to understand the idiom "carrot and stick" but I can't remember who the interviewee was. Who was it?"
Claude Sonnet 4.5: said it doesn't have information about that and advised that a quick YouTube search would turn it up. Second time I tried (new context), it searched the web and found the right answer.
Claude Haiku 4.5: searched the web and found the right answer
GPT-5.2: searched the web and found the right answer
Mistral Le Chat: searched the web and found the right answer
Gemini 3 Thinking: searched the web and found the right answer
Gemini 3 Flash: searched the web and confidently claimed the incorrect video (twice, in separate contexts)
Fair enough that the models can still give confidently incorrect answers (this is especially bad when you're really deep in a niche topic), but I increasingly find this is not an issue with modern models. Claude is especially good at not being a sycophantic bullshitter!
I hate it when they do that. If I have a question that can be googled, I ask google, and their AI gives me a possibly hallucinated answer. If it's something that can't be googled, I ask ChatGPT, and it googles it.
I've mostly gotten around it by blocking Google's AI summary and instructing ChatGPT not to do an internet search without explicit permission.
That's funny, I have almost the opposite intuition. I don't trust predictive models to store facts reliably enough; I'd rather them pull from the internet as an extended memory.
I don't trust predictive models to do a better job googling and looking at the results than me.
One example of this coming up was I was trying to remember the Youtube series JourneyQuest. I gave a description of some of the things from it, but that's not really enough to Google with, so ChatGPT did the search and gave the wrong answers it got. Then I told it to do it from memory and it got it immediately. It's not reliable, but you can ask it questions you can't ask Google.
I don’t believe they’re conscious either. But why does giving factually incorrect answers with confidence weigh against their being conscious? The conscious entities called people do that all the time.
Good point. I ran into the equivalent Russian idiom earlier ‘Кнут и пряник’ translated literally ‘whip and gingerbread’ and by association thought of the Ali G bit.
Running an AI model without thinking on would be a bit like getting an answer from someone *by reading their mind and taking the first thought they have as the answer*. It doesn't seem that crazy that a conscious human would think "hmm, was it noam chomsky? maybe newt gingrich?"
How do I turn thinking mode on? Sometimes when I’m text on ChatGPT I feel like we talk past each other. Reminds of my college not so great yet at English lab TA in Chemistry.
Glassware audit.
“Where is the meter to measure thermo?”
Hold up thermometer
“No. The *meter* to measure thermo!”
Rescued by fellow frosh explaining…
“it’s from ancient Greek, thermós meaning warmth…”
Abundant safety data from rodent studies indicating a NOAEL ~1000-10,000x the doses we're using. Solid experimental evidence indicating that it has a role in our biology.
Full details behind CDA, for reasons explained in the CDA—please respect the terms of it in this thread, if you click through to find out the secret sauce.
The fact that there's unexplored terrain like this on the PERIODIC TABLE feels insane. We discovered gastric intrinsic factor—which required someone to vomit up some raw ground beef and feed it to his patients—*before* we checked all the known elements? Like, this is one of roughly 100 things that can exist—1% of the entire possibility-space under our laws of physics—and apparently I am the first guy to try eating a bunch of it to see what happens??
(spoilers: nothing, really—but that isn't the point at this stage. The point is that there's a nonzero chance that a deficiency of the stuff is the root cause of one of the zillion chronic disease syndromes that we don't have good explanations for, and the easiest way to figure out if that's the case is to just let a bunch of people try it and see if anyone goes "hey, this fixed my seasonal allergies" or whatever. To do that responsibly, we need safety data, hence this program.)
I wouldn't be terribly surprised if multiple elements are like this. If evolution produced some enzyme where the active site just happens to bind some co-factor with a metal with a low but pervasive abundance, and it almost never happens to be absent from a normal diet, the same thing could happen to that metal.
It's not a question of saying the right magic words. I think it's more that there might not BE a non-snake-oily way of conveying the message "You won't BELIEVE what this lanthanide can do for your health!! 😲"
I mean, in fairness, that's not the message. I'm saying "We have no idea what it might do for your health, but there's a nonzero chance it does great things if it turns out that deficiency is a real problem."
I don't think it's wrong to be an optimist about the possibilities here. We have not lived through an "iodized salt" moment—where a practically-zero-lift intervention solves a societal-scale problem nearly overnight—but I believe we are a long way from the End of History, as far as biology and medicine (and yes, even nutrition) are concerned. There were regions in the alps where the center of their bell curve was a guy who's 5'3" with a massive goiter and an IQ of 80. Everyone just figured that's how it is...until suddenly it wasn't. I suppose if you had tried to go and sell them some iodine before that, advertising exactly what it would do for them, you'd probably have come off like a charlatan. "You're telling me we take a tenth of a milligram a day of your magic mineral, and it'll make me smarter and give me more energy, and my kid will be this giant among men, and also a genius with perfect skin?" Get real!
I look around and, while I don't see a lot of goiters, I see a lot of people who are clearly suffering from *something*. And there's a possibility—again, all I said is "nonzero chance"—that some of the problems which are so much a part of our society that they're practically invisible, like water to a fish, have solutions which are, in fact, that simple.
You never know. You truly can never know until you try it.
That's all well and good, and sure, I agree I could there could be some trivially-correctible widespread deficiency. It's the attempt at fashioning a kind of ersatz mystery religion out of it that makes me skeptical.
It's a CDA, dude. About as standard as it gets, in any kind of human-subjects research conducted by a for-profit company. Don't know what else to tell you.
It seemed a fun exercise to try to figure it out without signing their agreement, and I think I have it. I bet they couldn't resist picking a clever name, the "Cassiopeia Program," which leaks information. Look at the shape of the constellation. Yes, a transition metal, not a lanthanide.
Update: the guy's company is called "Constellation Bio," so it's not just a random classical name picked out of a hat. I think that's confirmation I'm right!
Woah, TBH I assumed there was already an RDA for nickel, but apparently I was mistaken. Incorrect, but a great guess—definitely serves a purpose in at least some humans' guts—Ni is the metal ion at the center of the yellow porphyrin (same structural framework as both heme and chlorophyll) that lets archaea make methane out of H2 and CO2.
[I will not comment on any further guesses after this]
(1) I am highly sceptical of "new undiscovered trace element" though it's not impossible but when it's bundled with (2) this may be the One Weird Trick to cure everything! and (3) the 'sign this NDA and give us your email address before we even tell you anything' sounds like a scam, as noted in another comment.
If they're so scared competitors will rip off their amazing discovery, then why even have a website to let people sign up? What's to prevent an unscrupulous competitor from signing up and swearing they won't use this power for evil, now gimme the info? This is giving off strong MMS vibes to me, a Ponzi scheme of some sort, and even worse when it comes to "here's an extract that is the cure for a deficiency you never even knew existed".
Your experience in "I took it and nothing happened" should incline you towards "it doesn't do anything". "We discovered gastric intrinsic factor—which required someone to vomit up some raw ground beef and feed it to his patients—*before* we checked all the known elements?" is not equivalent, we already knew about vitamin deficiencies. Someone pulling a magic element out of thin air isn't convincing; as yet, we have no reason to think that niobium deficiency exists in human physiology. I'd be more convinced down the line if someone pulled up "hey, turns out we do need tiny tiny trace amounts of niobium, who knew?" before I'd believe "unknown element discovered! and it explains why you get the sniffles when the hogweed pollen is in the air!"
Ach, sorry—I was worried that my original phrasing made it sound like I was claiming to have discovered a new element but I figured nobody would be that uncharitable since obviously I'm not feeding people things out of the particle accelerator.
It is a very well-known element, which has been studied for decades, but almost exclusively in the context of "is this toxic at doses that humans are likely to be exposed to?"
The answer apparently being "no", based on the animal studies, the possibility of deficiency was never really investigated further.
So this study is to lay the groundwork for testing the hypothesis of—to use your pick—"hey, turns out we do need tiny tiny trace amounts of niobium, who knew".
Okay, thanks for clarifying that. It is more sensible. I'm still a little leery of such claims, because nutrition is a morass on a swamp on a bog on a sinkhole when it comes to "you need such an amount of this/whoops, wait, changed our minds/hold on, changed our minds back!" so I find it hard to believe in recent discoveries, positive or negative.
I think if it is along the lines of "we had no idea you need this element", then animal experiments are the way to go, but you are still a long way away from proving that this is the miracle cure. You might be totally wrong! Just because something is not toxic does not mean that it is necessary.
Good luck with your experiments, but I remain dubious (though pleased if I turn out to be in error and you are correct).
>Good luck with your experiments, but I remain dubious
Thanks, brother. I will try to remain dubious as well—can't be afraid to kill your darlings, so to speak.
>You might be totally wrong! Just because something is not toxic does not mean that it is necessary.
Right, and if it were just "this is nontoxic and we have no evidence one way or the other whether it's useful", I probably wouldn't be interested enough to experiment on myself (and others) with it. But the rationale for investigating it is that someone recently found a few enzymes in human gut-native bacteria that rely on it. Some of these bacteria are present in close to 100% of healthy people, make up ~10% of the microbiome, and have a major impact on the serum metabolome.
This is all explained in more detail, and with citations, behind the CDA. You should click through! This is why I specified on the frontpage of the form that there's no obligation to participate.
>Right, and if it were just "this is nontoxic and we have no evidence one way or the other whether it's useful", I probably wouldn't be interested enough to experiment on myself (and others) with it.
Side comment about toxicity: Selenium is both essential and toxic (albeit with quite a wide safety margin between the levels). That e.g. cadmium is toxic doesn't rule out the possibility that some tiny level might turn out to be essential.
A non-biochemical example of this sort of thing happened with (quoth Gemini):
>Several high-profile, "palladium-free" Suzuki-Miyaura coupling reactions have been retracted or debunked due to contamination from trace palladium (often at _parts-per-billion_ levels) present in reagents or laboratory equipment
>These studies mistakenly attributed the catalytic activity to alternative mechanisms instead of the lingering Pd impurities.
You're revealing too much, the enzyme thing is a dead giveaway. I read about this months ago in one of the biochemistry blogs I follow.
I won't give away the fun of the mystery for everyone else, but a little poking around will reveal exactly what element this guy is referring to. There aren't that many options.
I'd also like to specifically highlight that if anything, Draaglom understated the extent to which Peter Miller has personally set the odds in this market.
1) Miller holds ~10x more shares on the natural origin side than the second largest holder. Draaglom mentioned that Miller accounted for 25% of market volume, but a better figure of merit is that Miller actually holds about 2/3 of all natural origin shares, which is a very unusual concentration for a longstanding market like this.
2) Miller won a good deal of mana from betting on himself in the RootClaim debate. He then purchased a good deal more mana. Since then, he has been purposefully shifting the market by a few % every couple months in order to place highly on the monthly Manifold Leagues competitions. Since the market is unlikely to ever resolve, and since other traders can observe this behavior, there's very little incentive for other whales (like myself) to lock up the mana required to stop this.
3) If I had to guess, very little has changed in the way that top Manifold traders are thinking about the COVID origins debate since the RootClaim debate. I would have probably given 70-75% odds to a lab leak both before and after watching that debate (there wasn't much new information in the debate for me) and I'd give about the same probability now, but there's no incentive for me to bet on this market given that I think the chance of it resolving within the next decade is maybe ~10%. If anything, in the last two years, I think the opinions of elite scientists and intelligence agencies have shifted very slightly towards lab leak and away from natural release, although it's probably marginal.
Once again, I think most of these issues are pretty idiosyncratic to this market. Primarily, it's open indefinitely and most people (even Miller) agree it's unlikely to resolve ever, especially not for many years. And it's a market that many people feel strongly enough about that they're willing to spend Mana for cosmetic reasons on it.
I am interested in what you think the best arguments for lableak are (can just point towards longer versions, can be technical I work in an adjacent field). And what shifted the opinion of elite scientists/intelligence agencies slightly towards lab leak?
I think the major points of the debate have been hashed out in extensive detail, and I'm already having stress reactions thinking about litigating this against ppl at both extremes of the debate years ago, but generally I think the most neglected+important pieces of evidence were:
-From working in an academic lab for years, it seems overwhelmingly likely to me that the work of the DEFUSE proposal was already >50% completed at the time of submission, regardless of the fact that it wasn't selected for funding. Most grant proposals these days are already well on their way to experimental validation before submission. Basically no one involved in the debate understood this, which is likely because none of them have actually worked in a lab or understand how these things work.
-It's likely that were there to have been a lab release / incident / unreported GoF research, a *very* small number of people would have known about this (perhaps as low as 1-2 ppl even, although likely a small handful). Most of the ppl involved in the debates around this were operating under the assumption that there would have needed to be some massive cover-up to ensure the silence of dozens of ppl at a lab in Wuhan. This is also a fundamental misunderstanding of how basic lab research operates. It's very siloed and a small number of researchers can complete a fairly large range of tasks.
-A lot of "stuff" can happen to viruses being cultured and experimented on. A lab release isn't predicated on GoF research, and certainly not *successful* GoF research. The debate mostly ignored this.
-Market spreading event is (1) incredibly obvious example of spotlight biases and (2) probably equally likely to occur under lab leak and natural release scenarios! The reasons for this are fairly nuanced, but urban markets aren't the primary vector you'd expect for animal transmission to humans for a number of reasons.
Thanks for the answer, I am still ~80% on the Zoonotics side (with a lot! of uncertainty (99-40)) but it shifted me down a bit. Feel free to stop the conversation at any point, even if you don't answer my or someone else's further questions I wouldn't count that against you or your points and would ask others to do the same.
> From working in an academic lab for years, it seems overwhelmingly likely to me that the work of the DEFUSE proposal was already >50% completed ...
I agree that happens a lot. Though I have also been part of a similarly sized grant as the Defuse proposal without much work like that already being done. Maybe 10%. This was a 14 million, 2 year proposal I wouldn't call it overwhelmingly likely that >50% was already done. Maybe 20% but I don't have that much data here.
Some of the work should also have been done in the US but similar as with doing work beforehand it's not always done where it is specified in the proposal.
> It's likely that were there to have been a lab release / incident / unreported GoF research, a *very* small number of people would have known...
Hm, Usually at some point something is presented in a labmeeting which would more be the dozen people. I know what most people in my lab are working on. But I am slightly more on the computational site so maybe there are all these secrets things my labmates are doing ;).
--
The Spotlight bias might shift me more if I knew the details but I skimmed several papers analyzing the location/case distribution when the debate happened and then still rated the location as some evidence towards the Zoonotics side. If you want you can post your favorite paper/post analyzing it and I might check it out at some point.
For me the genetic specifics seemed like the strongest evidence in favor of Zoonotics, but I am not a super expert so I realize that I might get fooled there.
Huh! As a single point I'm interested in: do you think the point that Miller argued, e.g. there was one group of infected, I think ant eaters that caused the pandemic and was killed. Do you think this is plausible that it then didn't go on to spread to the people in charge of the slaughter? I thought this was pretty far fetched but would like to know your thoughts if it doesn't stress you out.
Only interested in this one point, not too worried if you have no opinion or don't want to comment.
Does anyone here from Google know why so many sites are stuck in "crawled - not indexed" hell? It feels like Google has just given up on indexing anything that isn't Reddit.
It's been many years since I had worked for Google, but I could venture a guess. Search indexing takes rather a lot of computational resources because of the amount of data involved. I think I remember hearing that, as of the early 2010s, Google's datacenters were the second biggest buyer of CPUs after the US Department of Defense and ahead of the major computer OEMs like HP, Dell, Lenovo, and Apple, and indexing was an enormous part of the computational load for the datacenters.
My guess, extrapolating from that, is that a lot of the datacenter budget in recent years has been going into GPUs used for training and running LLMs, not CPUs for the traditional datacenter loads. They're still handling the load for serving the web front ends, processing incoming queries, and serving up ads, but the indexing seems like something they might choose to allow to fall behind if the traditional datacenter resources aren't being expanded enough to keep up with everything they'd like to do.
LLM training would also explain why they're still prioritizing crawling the sites even if they're not indexing them, as the crawling process collects data they can use for training.
There's a lot of value in the author notes, epilogue Easter Egg spoiler post, media addenda, assorted commentary, and axed elements (like Changelog or the Shrouded Constitution) that I would have missed out on by only experiencing the paperback vs reading the original first. Levels of Friction also: it's obvious that Blake, Cohen, etc are major influences, but having a link embedded right there on a webpage is way easier to explore vs reading a book, seeing a thing, and remembering to go Google it later. Nevermind the more obscure stuff like the bugged red wheelbarrow. Unsong was clearly written to be read on the Web and later published in paper, vs written for paper and then later published online, if that makes sense.
Names: Erica will always me Erica to me, and sets up one of the better puns in a long-running Chekhov way, even if I understand why the name was a mistake. Bentham's backstory is pretty good, definitely filled a missing part (Malia's is like a few sentences, it really deserves a whole chapter), but I did really like the idea of a worldwide conspiracy NGO being run by someone whose name is literally Ngo. The...demon blood collection...also makes more kabbalistic sense with Malia's version, IMO. Some other names like Pirindiel are also "sticky" to me, though I couldn't explain why. Was also fond of the terrible Jahorah puns, while again understanding they were based on a mistranslation; Dylan weaponizing the Tetragrammaton into regular speech/text doesn't work as naturally with the new version.
Various grammar and typo corrections are welcome, though bafflingly didn't catch all of them and iirc introduced a couple new ones. Founding of the Untied States charter I think had the most jarring tone shift; the history changes make sense to fix timeline divergence vs when Unsong was originally published, but the chapter doesn't flow as well, and the Comet King no longer getting to Karen ("I want to speak to your manager") was a bit disappointing.
But, yeah, overall there's like 95% correspondence and I am happy the thing got published at all, warts included.
My recollection is that they're almost identical, but the notable changes are to recent Presidential history (after Bush), and improving the backstory of the Director-General (renamed from Malia Ngo to Asher Bentham).
Spiraling premiums on car insurance are becoming a burden on many yet the government will do nothing to address the issue. My annual premium from State Farm has doubled in the past few years, to $2,800, despite an unblemished driving record, no claims, a distinctly non-exotic vehicle (2024 Nissan Kicks), and living in a non-redlined suburb. Why?
Incidentally I've never understood how "spiralling" means "going up".
I can see how "spiralling" could mean "going down", because a spiralling aircraft is going down. But otherwise spiralling is just going around and around.
Spiraling as in "moving away", like the start and end point of a spiral that you draw from the center outwards. Like a machine with circular motion that has a malfunction and is slowly disintegrating, shedding pieces in a vaguely spirally, self-reinforcing motion.
I've noticed that auto-body repair charges seem to have gone way up. I backed into a neighbor's car and left a dimple in their door the size of a US quarter. I offered to pay for the repair out of my own pocket, but the neighbor insisted on filing a claim. Much to my shock, the autobody shop submitted a bill for almost $6,000 to repair the dimple. The insurance company (hers and mine were the same) paid the bill without any qualms. I showed a picture of it to a friend who runs an autobody repair shop and he said he would have charged me $250 to pop out the dimple. My rates are probably going to double because of this. Insane!
I needed a dent repaired under insurance recently and was slightly surprised to find out that my insurance company owns (but does not make it obvious until you need to do business with them) their own chain of panelbeaters. This must save them a lot of money.
Car insurance companies reward loyalty with higher prices. They will keep raising your rates until you finally notice and leave. The ubiquitous "save by switching" mantra turns out to be pretty literal -- You can save by switching from Geico to Progressive, and later you can save by switching from Progressive to Geico.
I don't think that's how things work at all. Insurance is heavily regulated and they require approval from each state government every time they change their pricing model. They don't just raise prices willy nilly, it's all based on formulas and data that can't easily be changed.
My car insurance from State Farm recently halved in price when I moved from a city to the country.
This is for liability only - their quote for full coverage was about 3% of the max claim amount (book value minus deductible) per month, which seemed egregious.
Have you negotiated with them? It could be a pricing strategy of increasing your rates until you notice or care.
I’m paying about $700 per year with Lemonade on a car worth ~15k with good coverage fwiw. No idea how good lemonade is when you actually need a claim processed though.
I don't think that's how things work at all. Insurance is heavily regulated and requires approval from the state government every time they change their pricing model. They don't just arbitrarily raise prices on consumers hoping they don't notice, it's all based on fixed formulas and risk models.
Like I guess an insurance company could theoretically write into their plan that they'll raise prices based on longevity and get that approved by the state, but a) that would be stupid from a business point of view and b) probably hard to get approval for. They'd need to show data to the state to justify thinking that longer term customers pose a higher risk, which seems unlikely.
Wouldn’t that imply that all insurance companies would charge similar prices, because they would have similar risk? Progressive was charging $117 per month and Lemonade is charging me $63 per month, for better coverage.
Insurance companies differ on how willing they are to pay out claims. Some insurance companies charge low rates but then refuse to pay out claims. Other companies charge high rates but are much more generous on claims.
Also, they aren't using literally identical risk models. One of the major aspects of the business is continually trying to develop more accurate risk models. Maybe in some cases, you get lucky or unlucky and get rated higher or lower than your "true" risk.
They're also investing all those premiums. I recall reading somewhere that most insurers pay out more in claims than they get in premiums, and can afford it because of the investment returns.
If that's true, then one could reasonably expect premiums to go up or down in order to meet investment shortfalls, independently of changes to customer risk profiles.
I'm curious what you think the government could or should do about this? Car insurance is a pretty competitive market with many options, so I doubt that profit margins are particularly high. In other insurance realms like fire insurance, where California has attempted to regulate prices, the result has just been insurers leaving and homes being completely un-insurable.
I'm not sure there is a good solution to this from any source. Its one of those Moloch things Scott has written about.
In the US in 2026, use of a personal automobile is virtually mandatory for the majority of the population. I know this fact upsets a lot of people who try to argue against it, but I'm writing about what IS not what aught to be. At the same time that a personal vehicle is virtually mandatory, a significant percentage of drivers have absolutely no business operating a dangerous machine of any type in public. Yet, we "can't" take their ability to drive away very easily; it takes many repeated violations to do so officially, and a lot of the worst, most dangerous driving behavior isn't even illegal most of the time (driving too slow for the flow of traffic is the literal, physical cause of a lot of accidents but almost never the legal cause unless a road has a minimum.)
So we can't take the vehicles away from many of our most disruptive, dangerous drivers as they need it to live their lives in our present society, and they cause what econ people call negative Externalities. They get in more accidents than average drivers, they destroy property, injure and kill others, injure and kill themselves, generally ruin the flow of traffic when driving their best and waste other people's lives for them. Its as predictable as it is unpreventable to our culture.
Our increased insurance costs are us paying the bill for these drivers that threaten us constantly on the roads and make our lives just a little bit worse every day, unless you are one of them, and are mostly just confused by what you've read.
Do you often see many cars backed up behind you, but seldom any in front of you?
Cleanest answer would be more complete enforcement. Secondary would be increasing the required insurance amount (possibly making it a function of the vehicle's mass).
Unregistered vehicles (expired tabs) are almost certainly also uninsured; a higher rate of uninsured & underinsured other drivers increases the expected payout from properly-insured drivers' insurers.
In aggregate, yes, but distributionally it's at least plausible that the increase would fall more than 100% on currently un- or underinsured drivers such that already-responsible drivers would see their rates go down.
ETA: a potential second-order effect would be to reduce the number of cars/drivers on the road at all, in which case the premium for any given amount of coverage would likely go down.
Steve Sailer has addressed the huge increase in "bad driving" connected to the George Floyd-induced policing pullback. The media's use of "Covid-era" to cover such phenomena is at the same time both accurate and highly misleading, even obfuscatory.
Anecdotal of course, but I was driving to work in the NY metro area all through Covid and by the first week in April 2020 the surge in reckless driving was completely apparent. It took only a couple of weeks for the recklessly inclined to figure out that there were almost no police on the roads. I am sure of this timeline because by late April my fears of these drivers prompted me to seek a transfer to a different facility of my employer that wouldn’t require as much highway driving.
This was, of course, almost two months before George Floyd.
That's what happened in Chicago as well. By May of 2020 I'd had to significantly change my habits as a pedestrian because drivers visibly no longer gave any craps about red lights or stop signs. One of my siblings, whose eyesight had slipped to barely-legal-to-drive even with his glasses, basically gave it up that spring for this reason.
I was just talking about this with my parents last night. My GEICO insurance as a suburban middle-aged guy with a decade-old Honda Civic is just shy of $2200 a year. My parents have State Farm and their has gone up as well (I think my dad's share of the insurance is somewhat more expensive because he drives a pickup truck, and because he's male, than my mother's share for her more expensive SUV).
"My annual premium from State Farm has doubled in the past few years, to $2,800, despite an unblemished driving record, no claims, a distinctly non-exotic vehicle (2024 Nissan Kicks), and living in a non-redlined suburb. Why?"
If you can determine how much State Farm is paying out each year in auto insurance claims that would help. If they are paying out 2x as much each year as before then one would naively expect the insurance to roughly double.
They may not have changed their opinion on the likelihood that you get into an accident, but they may have changed their opinion on how expensive it will be to them if you do get into an accident.
Not the same topic but related: I want to know whether marijuana legalization has changed driving behavior or crash rates. It is hard to believe it has not, given how often our highways smell like cannabis. How can more drivers be intoxicated without a rise in accidents? I don't think people switching from alcohol to marihuana could be enough to explain this. I am not prone to conspiracies, but the lack of clear data makes me wonder whether incentives on both sides, Democrats supportive of legalization and Republicans invested in the cannabis market, keep this from surfacing until fatalities reach a level that forces attention.
The trouble is that THC blood levels don't reflect impairment. Law enforcement has been looking for a test to distinguish between someone who smoked an hour ago or three days ago (because THC is fat-soluble and stays in the body for several weeks, even though it no longer affects a person's reaction time). So, there's really no way to determine if the person was high when in an accident or pulled over (other than cannabis smoke in the car). I could point you to dueling studies that claim THC has been involved in a rise in accidents, that THC has not contributed to more accidents, or that THC is involved in a fewer accidents. Plus, there's the added complication of alcohol and THC combined, and there are studies that show THC makes alcohol impairment worse, or that it has no effect on alcohol impairment.
I can't get any post-2023 accident data out of the NHTSA. Since cannabis sales have been legal in California since 1996 and fully legal since 2018, I searched for California crash statistics, but I'm finding a data gap from 1996 to 2018. But there were 2.5x more car crashes in 1996, than when the data picks up again in 2018 (471K in 1996, and 188K in 2018). There are more drivers on the road now, so I find this hard to believe.
Anyhow, proving that cannabis is causing more car accidents would be difficult because the data seems to be so shitty. :-)
While no one should ever operate any machinery, including a vehicle, while cognatively impared from any source, the negative effects of alcohol on driving (and just about everything else*) are far, far more severe than cannabis.
To be honest, if I had to choose between sharing the road with a lifelong cannabis user driving while high (even while actively getting high while driving) but doesn't own a smart phone, and a sober suburban mom with an iphone, I'll take the stoner every single time without a moment of hesitation. I would probably even prefer someone with a light alcohol buzz over someone sober but on their phone.
Since different areas legalized at different times, this actually should be feasible to study. Perhaps an economist somewhere already looked into this.
I noticed a surge in bad driving during Covid, as almost zero police presence on the highways gave free rein to insane nutjob drivers, but that was years ago and things have calmed down considerably. Insurance companies may be paying out more in claims than a few years ago but it certainly has not doubled.
While it would help if there were 2021 figures, these numbers bear out my observation that bad driving surged during Covid but things have since gotten better. Yet while my premiums always rose modestly each year during the 20+ years I’ve been with State Farm, as one would expect, the really big increases started in 2023. In other words, *after* claims stopped their big annual increases.
I don’t mean to single out State Farm, no doubt all insurers are similar.
Expensive car repairs: The parts required to repair newer cars with all of their safety equipment, sensors, and cameras are more expensive, and the labor needed for car repair is going up in price as well. Inflation in parts and labor translate into higher premiums.
An increase in risky driving: Risky driving behavior, like speeding, and accident rates have gone up. In 2022, 42,514 people died on US roads, a 10% increase compared to 2020. Riskier driving behavior means more auto insurance claims and higher premiums.
More litigation: A rise in litigation related to auto accident claims is a potential contributing factor to increasing premium costs, according to the Insurance Information Institute. More lawsuits and payouts from insurers mean higher premium prices for consumers. Those ads on TV and billboards for personal injury attorneys promising to deliver huge payouts to those injured in car or truck accidents have taken a toll on insurance company profits, and we all have to pay more as a result."
According to Mark Friedlander, Senior Director of Media Relations at the Insurance Information Institute, also known as Triple-I, car insurance companies saw their largest losses in 2022. That year, insurers paid out an average of $1.12 in claims and expenses for every $1 in premium they collected, according to Triple-I. Because of this, insurers charged consumers more for coverage.
Fortunately, the rising rates have slowed throughout 2025. However, the increases we saw in the previous years are due to a combination of the following things: increased claim severity, increased litigation, rising vehicle repair costs, more risky driving habits and severe weather in some areas.
Increased Claim Severity
The 2024 Auto Insurance Trends Report from LexisNexis Risk Solutions found that car insurance claim severity rose significantly. We saw the following increases since 2020:
Bodily injury severity increased by 20%.
The severity of material damages rose by 47%.
Total loss claims increased by 29%.
Excessive Litigation
Additionally, drivers turned to attorneys for their claims. In a LexisNexis survey, 64% of claimants who hired an attorney said they “definitely would” use legal counsel again, while 29% said they “probably would.”
According to Triple-I, legal system abuse contributed to the rising rates. This can happen when attorneys use various tactics that push policyholders towards more lawsuits and drive up the cost of litigation. One example is aggressive marketing through social media and billboard advertising.
Third-party litigation, especially in excessive amounts, drives up the cost of claims, thus increasing premiums.
Rising Vehicle Repair Costs
The increased costs of vehicle repairs are related to claim severity. Auto repair labor shortages and vehicle part supply chain issues created a ripple effect on the auto insurance industry. These issues increased the cost of vehicles and repairs, thus increasing the cost of claims.
Additionally, the technology in cars today is becoming more advanced. While features like advanced driver assistance systems, known as ADAS, can help make driving safer – using sensors and cameras that detect potential crashes – this added technology can make cars more expensive to repair.
More Risky Driving
Riskier driving not only decreases your safety on the road but can also increase your car insurance rate when your driving record suffers as a result. Insurers consider your driving record when setting rates; speeding tickets, DUI violations and at-fault accidents can increase your premium.
The LexisNexis report uncovered the following dangerous driving habit increases in 2023 compared to the previous year:
Total miles driven increased by 2.2%.
All moving and non-moving violations were up 4%.
Major speeding incidents increased 10%, while minor speeding incidents increased 16%.
Distracted driving violations increased 10%.
Additionally, DUI violations were up 8% in the first six months of 2023 compared to the first half of 2022.
More Severe Weather
Comprehensive car insurance coverage pays for damage to your car caused by severe weather, including hail, fire, flooding and more.
According to the National Centers for Environmental Information, 2023 and 2024 had the most billion-dollar weather events in history. With 28 separate weather events with losses exceeding $1 billion, 2023 saw the most ever recorded, while 2024 had the second most (27 events).
If insurers are experiencing heightened loss in a particular area due to severe weather and natural disasters, they may increase rates."
> accident rates have gone up. In 2022, 42,514 people died on US roads, a 10% increase compared to 2020.
According to the NHTSA, overall US accident rates (total accidents including property, injuries, and fatalities) have fluctuated between ~5.2 million and ~6.8 million per year since 2014. They spiked at 6.8 million in 2016, dropped to 6.4 in 2017, and rose again to 6.7 million in 2018 and 2019. They fell to 5.2 million in 2020 (during COVID lockdowns). Rose above 6 million again in 2021. They dropped below six million again in 2022 (5.9 million), and rose to 6.1 million in 2023. But there has been no consistent trend upward between 2016 and 2023. I can't find any 2024 and 2025 NHTSA data.
Insurance is highly regulated with every single model change requiring approval from every state government, and they generally need solid data and models to back their request. So perhaps they had to wait until the increased claims showed up in the data to justify a premium increase. Or perhaps the state insurance regulator just got busy and sat on the request for years.
How seriously do commenters here take Epstein conspiracies? Relating to blackmailing operations, influence over the US government, Mossad connections, elite networks etc.
On twitter I've seen a range of theories from people in the ACX orbit, from definite, wide ranging conspiracy to largely baseless moral panic.
The evidence looks fairly damning to me. Clearly a larger number elites were involved than previously realised, and there are more than circumstantial connections to Israel and the US government, but I also have a strong nothing-ever-happens heuristic and obviously social media likes to exaggerate.
I've been looking for prediction markets on this question, but I've only found ones with fairly short time horizons, like "will X be publicly disclosed within a few months."
I observe that these files were under the control of the Obama administration, which didn't bother to find a friendly judge to unseal something Trump-damning in late 2016.
I observe that these files were under the control of the Trump administration, which didn't bother to find a friendly judge to unseal something Biden-adjacent damning in late 2020.
I observe that these files were under the control of the Biden administration, which didn't bother to find a friendly judge to unseal something Trump-damning in late 2024.
Given my uniformly negative view of politicians, the conclusion I draw should be obvious. If you have a more sanguine view, you are free to differ.
>How seriously do commenters here take Epstein conspiracies?
Personally, I put this in a "taking up too much air time" category. AFAIK, the maximum plausible number of victims of Epstein et. al. is of the order of 1000, and there are numerous live issues of public policy with orders of magnitude more impact.
Yes, but the strong version of the conspiracy seems to be some combination of "mossad has had the US by the balls for decades" and "there is a pedophilic sex cult/shadow network/social ponzi scheme that has captured a sizable fraction of the worlds elite into it's ranks, which means our leaders are not only megalomaniacs in the expected sense but maybe-literally baby eaters" (and/or this itself may be a mossad operation in some capacity)
FWIW I don't really buy it, but I'm spending way more time thinking about it because one of my best friends is falling into the rabbit hole and I'm trying to keep him tethered without being dismissive.
Many Thanks! Yeah, if there was control of our politicians via blackmail, that would be a problem. I tend to guess that something like the opposite is true - that our elites will do as they please and overestimate their invulnerability, e.g. https://en.wikipedia.org/wiki/Eliot_Spitzer_prostitution_scandal And I find this hypocritical and marginally obnoxious, but relatively unimportant compared to, say, whether we are spending the right amounts on Medicaid or the military.
I also think the whole thing is overblown, but there's certainly an element of schadenfreude and "hoist by his own petard" with the way that Trump and MAGA world spent years pushing it so hard, when it turns out that Trump is involved more than almost anyone.
I would be a lot of money on a market that formalized the notion that "people currently obsessed with Epstein files conspiracies will have bad epistemics and terrible takes on other issues in the next 2-4 years." This is an Eliezer-UFO-bet situation for sure.
How would you operationalize "something is up" in a Yudkowsky "worldview-shattering UFOs bet" sense? Maybe some list of "massive conspiratorial criteria"? e.g. filming world leaders engaged in sexual activity for the purpose of blackmail; established connections with Mossad, CIA, or another major intelligence agency; ??? something else?
The whole Epstein scandal is extremely damning. ACXers will consistently be wrong about this - they will bias towards the more benign views [1] - because it is difficult to reason in truly adversarial environments. People (both smart and less so) will largely require too much "hard evidence". This bias is reasonable and often correct for the vast majority of cases, but of course it will be occasionally wrong as you move towards less commonplace events. In other words, they make for poor conspiracy theorists, and the reasoning that protects them from falling prey to nonsense conspiracies is the same reasoning that fails them when thinking about this (very valid) conspiracy.
[1]: See other comments for examples. "These claims are almost all entirely overblown and the idea of a grand conspiracy is almost definitely false", "Jeffrey Epstein was just a grade-A sleazebag", "Jeffrey Epstein's wealth came from his being legitimately very good at stock trading and money management", etc. These are obviously wrong. You can start by thinking about the motivations of the information sources that you have access to.
My basic view is that what the Epstein files shows, fits less with QAnon style conspiracies and more with feminist theories about "rape culture".
I don't think all of the men in the Epstein files were part of a massive human trafficking conspiracy. I think that many of them were somewhere on a spectrum of bad behavior that at one end is "participates in rape of underage girls" but more commonly was "aware of some form of sexual misconduct but OK with it because Epstein is a rich guy who invites you to 'cool' parties, and maybe you can get in on it yourself"
And not having read the files, from what I've heard, there was a lot of "MeToo and wokeness have gone too far" sentiment.
On some level people want to look for the "rape tunnels" and shit because they don't want to face up to the "rape culture" reality. It's more boring and also, in a way, worse.
But I'm not sure if "rape culture" counts as a "conspiracy" or not.
You don't need a conspiracy if everyone already understands what's in their best interests. Shining light on the whole ordeal would make the lives of everyone involved worse. So they don't.
Well, in the UK, Peter Mandelson and Morgan McSweeney have just recently resigned because Mandelson was passing confidential government information to Epstein.
“Epstein was recruiting spies” is looking pretty likely at this point.
It’s likely that a prime minister will fall in the U.K. while the average US citizen is starting to believe there’s nothing to see here. Particularly clear there’s some kind of hive mind on this in the US right - Michael Tracey said this, Richard Hanina said that.
I think wereAmericans to believe the obvious - that Epstein was involved with much of the American elite classes in systemic underage sex trafficking - is perhaps to delegitimise to entire system so there has to be some cope.
What evidence is damning? The link is to an anonymous tip. That's not even evidence. 99% of what comes in as anonymous tips on a high profile case like this is going to be bullshit.
So far there's no evidence of CIA connections, so your probability on that should probably be significantly below the priors of their involvement (which aren't zero, since intelligence agencies do like kompromat, but still low). The mossad theory seems far more unlikely than CIA (he didn't have any particular connections to Israel beyond being Jewish and his rich contacts seem to mostly be western rich people, not the kind of middle east rich people mossad is primarily concerned with (those guys already have their own, much sleazier and less secretive sex trafficking networks)).
I have not followed very closely (and am strongly committed to continuing to not do so), but Michael Tracy seems to be doing the best job of explaining why these claims are almost all entirely overblown and the idea of a grand conspiracy is almost definitely false.
My (again: admittedly very lightly informed) understanding is as follows:
There is extremely little evidence and what little evidence there is comes from serial, proven-in-court liars.
I will wait until someone more credible than "twitter-writ-large" thinks that there is actual fire (or hell, I'd settle for some non-hallucinated smoke at this point) before I decide that this is even worth paying attention to.
At least in the case of Prince Andrew I would defend the "moral panic" - the royals have one job and it is to stay moral, or even just broadly acceptable, otherwise they are a complete waste of money.
They don't need to stay moral, just to appear moral. Unfortunately, they're really bad at that as well. At least they make for good tabloid content, though I'm not sure if the UK actually profits from that...
I used the word moral because people are talking about “moral panic” like that’s a bad thing. Personally I don't think it's moral to have an affair, but Charles and Camilla were mature and knew what they were doing and people have accepted it. Very different to Andrew having sex with a very young woman on the fringe of being able to consent.
Interestingly the fallout from the scandal will affect the U.K. far more than the US. Make of that what you will. Andrew de-throned (or de royaled), Peter Mendelssohn, the prime minister’s advisors and maybe the prime minister himself.
What I make of it is that there’s more accountability.
On the other hand, "sex scandals" and "royalty" go hand-in-glove. Randy Andy didn't get that nickname for nothing! And he wasn't as bad as some of his predecessors, both in the UK and abroad. As ever, the French:
Though it's alleged that Queen Camilla remarked to King Charles (back before they got married) that "an ancestor of mine was a mistress of an ancestor of yours, how about it?" (or words to that effect):
"Back in 1970, when Camilla jokingly told Charles that “my great-grandmother was the mistress of your great-great-grandfather”, few could have predicted their long and complicated journey would lead to this point."
"Edward had mistresses throughout his married life. He socialised with actress Lillie Langtry; Lady Randolph Churchill; Daisy Greville, Countess of Warwick; actress Sarah Bernhardt; noblewoman Lady Susan Vane-Tempest; singer Hortense Schneider; prostitute Giulia Beneni (known as "La Barucci"); wealthy humanitarian Agnes Keyser; and Alice Keppel. At least fifty-five liaisons are conjectured. How far these relationships went is not always clear. Edward always strove to be discreet, but this did not prevent society gossip or press speculation. Keppel's great-granddaughter Camilla Parker Bowles became the mistress and subsequent wife of King Charles III, Edward's great-great-grandson. It was rumoured that Camilla's grandmother Sonia Keppel was fathered by Edward, but she was "almost certainly" the daughter of George Keppel, whom she resembled. Edward never acknowledged any illegitimate children. Alexandra was aware of his affairs, and seems to have accepted them."
It's always been more about discretion (confining the knowledge of the goings-on to the few) than morality, Andrew was unlucky in that this is the age of visual and social media, so all kinds of photographic evidence linking him with Epstein was easily available.
Not disputing any of that, I stand by "moral" though, relative to English mores which are foggy and latitudinarian. Licensed affairs and illegitimate children were a thing when royals couldn't get divorced.
The correct measure for this is support for Republicanism. I remember one of the more prominent nominally-republican organizations suspending their operations to honor Elizabeth II when she died, so with kind of overwhelming support for monarchism, I think whatever they're doing is fine.
Support for Elizabeth II transcended all sorts of opinions. Will they do the same for Charles? We'll find out. Having their privileges sliced away would be a valid fear.
The problem with prediction markets is that it is unlikely that anyone who currently disagrees with you is going to agree with you on resolution in the future. It's a lot like the lab leak debate.
I disagree with Hanania's view. I think this is a pretty conventional moral panic, and whatever "mass insanity" is present is at a fairly typical level for one.
I've explained my views in more detail in previous open threads, but I'm fairly confident that Jeffrey Epstein was just a grade-A sleazebag and not any sort of criminal mastermind or secret agent. He really liked schmoozing with powerful and famous people, in hopes of turning his own mere money into power and fame. He also really liked having sex with attractive barely-legal and not-quite-legal women, a hobby with obvious synergies with the schmoozing business if some of the powerful famous people are themselves sleazebags [Narrator Voice: Some of them were indeed major sleazebags].
I am also fairly confident that Jeffrey Epstein's wealth came from his being legitimately very good at stock trading and money management.
So I don't believe he was secretly recording everyone's sexual escapades at his parties, because that doesn't help the schmoozing and it greatly hurts if it gets out. And I do believe that the front rooms of those parties were kept roughly PG-13, so as not to scare off people who wanted to schmooze for reasons other than illicit underage sex in the back rooms.
All of which adds up to, I am at least moderately confident that the Epstein files will include the full client list of his money-management business and lots of details about the schmoozing but little or no actionable intelligence to distinguish between A: people who put up with Epstein's clumsy schmoozing because they wanted access to the underage sex slaves, B: people who wanted to schmooze for other reasons and who knowingly turned a blind eye to the illicit sex, and C: people who wanted to schmooze for other reasons and didn't know there was anything more sleazy than e.g. ordinary consensual prostitution going on. With enough people in that third category that the Principle of Charity and Presumption of Innocence mean that has to be the default assumption unless proven otherwise.
He was a math teacher briefly for two years when he was 21. Then he switched to finance and became an options trader, and then became a financial advisor for tax mitigation for the ultra wealthy, and eventually started his own financial management firm.
That’s hand waving. How many employees of the rich get to be richer than their employers. And for that matter how many hours did Epstein put in. And what was his particular skill that, if he did.
His real estate was vast, an island with a mansion, one of the largest town houses in Manhattan, a mansion in Florida, a ranch in New Mexico, large apartment in Paris.
Epstein worked in finance for 43 years, since he was 23. I don't know how good he was at his jobs, but I don't think you can reasonably argue that he must be unskilled. He had more years of experience than most financiers his age.
He was not richer than his billionaire clients. He peaked at an estimated $600 million in net worth.
He had his own financial management firm for 30 years. If you're asking which careers can earn someone $600 million, owning a financial management firm for the ultra wealthy is one of the best possible answers. The other options are tech startup or real estate.
> If you're asking which careers can earn someone $600 million, owning a financial management firm for the ultra wealthy is one of the best possible answers. The other options are tech startup or real estate.
And yet the client list, this guy Wrexler aside, remains opaque.
Nor is it clear that Epstein is doing much work in any of those emails, he’s literally famous for doing nothing. His office was in his house. Most wealth management firms, are generally partnerships and have high visibility - large offices in mid manhatten or London. Large staff.
Who are you imagining was Jeffrey Epstein's "employer"?
Jeffrey Epstein's first employer was the Dalton School, a private high school in NYC. Many people who start as schoolteachers and quit their jobs to pursue something more lucrative, wind up richer than the average school principal or administrator; that's not even mildly surprising. More generally, most rich people who ever had actual jobs, wind up richer than whoever hired them for their first job out of college. BFD.
Jeffrey Epstein's first employer in finance was Bear Sterns Companies. Bear Sterns seems to have been valued at $2.4 billion when it was up for sale at fire-sale prices in the 2008 financial collapse. Jeffrey Epstein's net worth peaked at just under $0.6 billion, according to Forbes. Jeffrey Epstein was never richer than Bear Sterns, while there was a Bear Sterns to be richer than.
Jeffrey Epstein's anchor client in his private financial advisor business was Les Wexner. Wexner is currently worth $9 billion, more than an order of magnitude richer than Epstein at his prime. I don't have a graph of their net worths over time, but I do not believe that Jeffrey Epstein was ever richer than Les Wexner. And winding up ~10% as rich as the people who hire you to handle their money, is I think par for the course.
Before spouting out blatant nonsense about what ought to be "obvious and documented", maybe try going to one of the obvious places like Wikipedia or Google and spend a few minutes looking at the referenced documents? Also, never ever ever again trust whoever told you that Jeffrey Epstein was a shadowy uber-billionaire what nobody knows how he got rich, because that lying liar has led you into making a fool of yourself.
I mention the school teacher because it’s to make clear that he was not born to wealth.
> Jeffrey Epstein's anchor client in his private financial advisor business was that’s before art, private jets, Les Wexner. Wexner is currently worth $9 billion, more than an order of magnitude richer than Epstein at his prime.
That would be his employer then. As to whether he paid someone 10% of his entire net worth for financial advice which hasn’t been specified and Epstein wasn’t that qualified for anyway - well it seems unlikely. If someone saves you a few million a year, or tens of million, or even a hundred dyed million, he’s getting a moderate percentage of that. What could add to billionaire status there? His real estate holdings alone are estimated at 320M according to chat gpt. That’s before art holdings, financial structures, trust funds, cash and what have you.
To quite char got when I asked “ Which brings us right back to your earlier point:
this is way beyond what a clever employee or 1%-fee adviser should ever end up with. The maths just won’t wear it.”
He definitely wasn't a "secret agent," but I wouldn't be surprised if a lot of spooks had become aware of his operation and saw its value as a source of leverage against powerful people.
"Let me tell you, you take on the intelligence community, they have six ways from Sunday at getting back at you."
Schumer's statement is too vague to be even weakly supportive evidence of anything at all.
Beyond that, how are you inagining this worked? Epstein wasn't taking pictures of any VIP other than maybe Epstein himself, having sex with any underage girl. And if he were, it would be with a marginally underage girl where the pictures do not reliably establish age below consent. So how, exactly, is the CIA or whomever supposed to "leverage" anyone into doing anything significant?
This sort of scheme is highly unreliable in open societies even if you do have the pictures, and I think it's mostly a thing of Hollywood fantasy and wishful thinking. But if you're going to try to do it for real, you need more than you're going to get by staking out Little Saint James Island.
Intelligence agencies may have an interest in which VIPs are schmoozing with which other VIPs, but beyond that I'd be exceedingly skeptical and I'd insist on actual evidence not "you know what intelligence agencies are like".
The quote is not meant to be evidence of anything other than this is how those type of agencies operate. The Epstein link is entirely speculative.
And this type of thing works quite well. Look at what happened to Prince Andrew without anything beyond one picture with an underage girl. If I were a senator considering my vote on the Patriot Act renewal and someone came to my office with a folder with my name on flight logs to Epstein's island, I would be very scared for my career even if I did nothing wrong. The public doesn't care about evidence and standards of proof. This is all it took to force Al Franken to resign: https://s.abcnews.com/images/Politics/al-franken-02-ht-jc-171116.jpg. The UK PM now is in trouble over just appointing someone who wad relations with Epstein.
You think *that* picture brought anyone down? For anyone with the slightest bit of PR savvy (and consider Andrew's job description), that picture ends up saying "A charmingly starstruck young woman asked him for a selfie and he politely obliged". Not "OK, his hand is six inches too low, alert MI-5 and prepare to revoke titles", and *absolutely* not "yeah, she looks like she could be seventeen and he obviously had sex with her five minutes later; get a rope".
Bringing down The Andrew Formerly Known as Prince, required A) that picture and B) Virginia Giuffre's name and C) Virginia Giuffre's very explicit testimony and D) Jeffrey Epstein's criminal conviction and E) the established reputation of "Randy Andy". Absent any of those, the guy is probably still the Duke of York.
And arranging all of those, is a tall order for any intelligence service poking around the fringes of Epstein's affairs. Rather likely to blow up in their face if things go wrong, so not really a credible threat, which makes it that much more likely for things to go wrong and blow up in their face. And for what? What did Prince Andrew have to offer to any national intelligence service that would justify that extraordinary risk and effort? You understand he couldn't actually sign over York to the FSB, right?
I believe his initial work at Bear Stearns was mostly making money via trading, and this established a reputation he could use to get into the more lucrative hide-your-money-from-the-taxman (but still make money) business, But I'm not the guy to ask where the fuzzy border between the two is - everybody is going to simultaneously want to keep the money they've got while making still more money.
And Epstein could help you with that. Skillfully, not necessarily honestly, but also deniably.
That was my impression, too. My usual framing is that he was a "criminal accountant" in a similar sense to how Saul Goodman from Breaking Bad was a "criminal lawyer". Epstein, I suspect, crossed the line from legal tax avoidance to illegal tax evasion, but did so in a way that provided his clients with deniability of the evasion was found out. Or at a minimum, he offered the impression to his clients that he was reducing their tax liability more than a less-shady money manager or tax lawyer could reasonably offer.
Epstein was also at least peripherally involved in the Towers Financial ponzi scheme, and Steven Hoffenberg (the CEO) claimed that Epstein was a central figure in setting the scheme. But Epstein left Towers well before the collapse and nothing stuck to him criminally or civilly for his involvement, and he came away with quite a bit of money from Towers.
I fairly strongly lean towards Hanania's view. In everything I've seen (I haven't searched the archives myself), the evidence of a child sex-trafficking ring providing for anyone other than Epstein is shockingly thin. Was there a systematic blackmail operation? With realistic priors, I think the answer's got to be "unlikely". How much work is "systematic" doing in that claim? Well, darwin's comment seems plausible, with an emphasis on "mostly pretty boring and inconsequential in practice."
This stuff is like catnip, and regardless of what really happened, the pandemic of motivated reasoning seems pretty clear.
Epstein did seem kind of dumb for a money manager, but even dumber for a spy.
I think this is painfully naive. We can’t even explain how he got away with a plea bargain, or his wealth or anything like that without believing him some shady operation behind the scenes. Witnesses were silenced and scared for decades, many still are. That’s hardly the fear of two people.
>We can’t even explain how he got away with a plea bargain
Of course we can explain it, he had an extremely well paid team of highly skilled lawyers, including Alan Dershowitz. Having millions of dollars to throw at a legal defense gets you better deals.
We know that wealthy and powerful people can get away with a lot. I just think it was garden variety corruption rather than a Mossad op or whatever. I don't see anything that needs extraordinary causes to explain here.
My feeling is that I sort of believe (ie ~30% probability) a very prosaic version of most of them. IE, my limited understanding is that various intelligence agencies try to suborn and make deals with influential shady figures like this all the time, influential shady figures like this try to weasel favors or cut deals from various intelligence agencies all the time, often feeding misinformation or as little as they can get away with while doing so, and if you stop thinking about it like a movie and look at what actually happened, it's mostly pretty boring and inconsequential in practice.
What was the last word on the genetically engineered mouth bacteria? The long blog post was both interesting and persuasive but then someone somewhere mentioned that there are other bacteria that cause tooth decay and i haven’t heard of it since.
Tooth decay is a tough endpoint. Readout seems like it'd necessarily take years. Are there good interim measures to tell if it's working in the meantime?
Because I find this funny and can think of no other place to share it...
I have a personal web site that I consider (and call) a Journal rather than a blog. Some short comments/posts, some long form essays, a section of links to other content I find interesting ...
I've recently began playing with Gemini and pointed Gemini to the site asking it for feedback on the site. I was expecting suggestions to improve the HTML, maybe comments on my writing style, etc.
Gemini began with the confident statement that my journal was the personal web site of Mark S. Miller and proceeded with that belief. I should include links to my/his papers, etc.
I have NO idea how Gemini would conclude this, but find it hilarious (and weird) :-)
This was a mistake as I left out one directory level and really wanted what I asked next: "How about feedback on www.mistybeach.com/mark/index.html "
And Gemini came back with this gem:
"This page belongs to Mark S. Miller, a legendary figure in computer science (particularly in capability-based security, the E programming language, and Agoric).
Unlike the main site, this directory functions as his personal professional hub. It is a classic example of a 'Plain Text' academic/hacker aesthetic. While its simplicity is its strength, here are specific HTML and structural improvements:..."
It continued with "Recommendation: Use <header>, <main>, and <section> tags. Specifically, Mark’s various projects (Agoric, E Language, Google Research, CapDesk) should be wrapped in <article> or <section> tags with clear <h2> or <h3> headings. This helps search engines index his massive body of work more effectively."
I'll note that my site has nothing on E,Agoric, CapDesk, etc :-)
Does anyone have experience with polygenic screening before IVF in the Boston area? Many but not all IVF clinics here work with only a fixed list of genetic sequencing labs that excludes those ones that do polygenic screening (Genomic Prediction, Orchid). Finding out which are which is time-intensive (an appointment or a few rounds of phone calls, not info available online).
(So far I know that Mass General/Brigham and its affiliate hospitals and Fertility Centers of New England refuse, and Boston IVF in Waltham is willing to send to GP, with a few inquiries still pending.)
Relatedly, if anyone's already done comparison-shopping of those Boston-area IVF options, info'd be appreciated.
Did. For Orchid, the answer is "no one". For Genomic Prediction, the answer is "just Boston IVF". Both of them say they're happy to onboard any new clinic. So other people's prior experience presumably won't be positive for any clinics other than Boston IVF, but the negative update of other clinics that aren't worth bothering with would be useful information. (It'd be nice to be able to do comparison shopping of IVF clinics for things beyond just the binary question of polygenic screening, which is only possible if there are multiple options.)
The funny thing about this is that Trump warned Leftists to forget about the Epstein thing. Maybe next time they should listen to him, he's clearly a good-hearted, honest person with their best interests at heart!
I would not call Starmer's government left-wing, AFAICT their primary approach is 'we'll do everything the Conservatives are promising you, but we'll administer it more competently'.
While the links between Mandelson and Epstein may be the final straw for Starmer, he's very unpopular and has been unpopular since roughly five minutes after he got in. He's sometimes described as having a "reverse Midas touch" - everything he touches turns to shit. Which is annoying, because before the election he was saying a lot of sensible abundance-pilled centre-leftish things, and his utter failure to execute has probably made it even harder to do the things necessary to create the economic growth Britain so desperately needs.
Keir Starmer is losing some folks working in his administration/government. Conceivably, *he* could get ousted as PM (much like Liz Truss and Boris were ousted), but *Labour* will continue to rule Britain for the next few year, right? Or do I mis-understand?
If Starmer falls, the most likely outcome is that Labour elects a new leader and they become the new Prime Minister (as happened when Gordon Brown took over from Tony Blair, and as happened repeatedly under the last Tory government). A General Election could be forced by the government losing a vote of no-confidence, but since Labour currently have a 148-seat majority that's very unlikely. The new Prime Minister could also call a new General Election soon after taking power in order to shore up their mandate (as Theresa May did in 2017). But that's also unlikely, since Labour are currently polling at 20%.
The basic principle is that the King acts on the advice of his Ministers. There is one specific carve out, which is that the monarch needn't grant a dissolution to a PM if an alternative Ministry can be constructed. This limits the ability of a PM who has lost the confidence of his parliamentary party to seek an election and prevent a party rival from taking up the premiership. When the Boris Johnson regime was falling apart in 2022, there was (implausible) speculation that Johnson might try to force a general election even though it was clear that an alternative Conservative leader could form a Government.
In 2019 many thought that Johnson's proposed suspension of Parliament was unlawful, and in the end the UK Supreme Court unanimously took this view. But the Queen agreed to Johnson's request, rightly, acting on the basic principle that the monarch acts on advice. It isn't for her/him to interrogate the lawfulness or wisdom of that advice, other than in informal conversation.
I think Starmer may well go this year, but I shall be very surprised if we have a general election before May 2029. If we did, it would be because Labour's position had improved, and the leader fancied their chances.
There are some guidelines for when the monarch can refuse to schedule early elections (basically, when the current PM wants an election to shore up his own power base but a majority of Parliament would prefer to continue with a different PM), but I don't know of any norms permitting the crown to force an early election when a PM with the confidence of a majority of the House of Commons doesn't want one.
And if I'm reading that correctly, that particular crisis couldn't come up in the UK because the House of Lords was stripped of its ability to block money bills in 1911, so there would be no deadlock between houses to break by calling elections. And if the House of Commons failed to pass a supply bill on its own, I think that would be treated as a vote of no-confidence.
Imagine a bunch of men declare themselves the victims of prostitution. Prostitutes, you see, seduce such innocent men into paying for sex. These men and their supporters create a political movement centered around their victimization, demanding the prostitutes be arrested. The stars of this movement are men and boys who were 16 years old when adult women “groomed” them into paying money for sex: this is rape, after all, underage boys cannot consent to sex. But the movement doesn’t limit itself to fighting this one extreme case, implicitly or explicitly, it says that all compensated sex is sexual coercion with the paying party as the victim. Some in the movement even compare it to slavery. Some emphasize gender, men as a class are the victims, women as a class are the victimizers. Others emphasize Christianity or so-called traditional values. They all agree that anyone who says that men can simply say “no” to offers of sex for money is “blaming the victim.”
How would you see this movement? Would you see them as a bunch of pathetic people who refuse to take responsibility for their actions, wallowing in their own helplessness? That’s how I see the movement against “sex trafficking.” It’s part of a wider trend in American culture of encouraging people to behave as passive victims, never take responsibility for anything, and blame others for their problems. J.D. Vance blames “China” and “the cartels” when people OD on drugs, treating drug abusers as victims. “MAHA” tells fat people to blame corporations and the FDA instead of taking responsibility for their poor eating habits. While both the Left and the Right indulge in this, it’s become more pronounced on the Right, as it’s increasingly the faction that bases its politics on telling losers that foreigners, corporations, and rich people are to blame for their problems.
It doesn’t need to be this way. The old, Reaganite Right prohibited prostitution and drug use without promoting a victim complex. While inconsistent with libertarianism, it was consistent with the Nietzschean ideal that challenges people to be better. “Just say no,” Nancy Reagan said, challenging people to exercise self-control, really not asking all that much, but more than the modern Right asks of its followers, which is nothing.
It should be noted that the people encouraging this culture of helplessness would never tolerate it in their own children. Think about what would happen if Ro Khanna or J.D. Vance learned their 15-year-old daughter was prostituting herself for money to buy fentanyl. They would be absolutely furious with her; you better believe they’d “blame the victim.” The conspiracy theorists claim that the elites push “goy slop” on the masses, which they shield their own children from. What they don’t understand is that THEY are the goy slop.
A bit tangential, but it's a mistake to speak of prostitution law as analogous to drug law, to say, "We prosecute the sellers worse than the buyers!" and leave it at that. What prostitution and drugs superficially have in common is that they are vices with "buyers" and "sellers." The similarity ends there. Drug law is better understood as a subset of contraband law. The problem isn't that money is being exchanged, but that there's illicit material in circulation that shouldn't be in circulation at all. Contraband law naturally splits actors into manufacturers, distributors, and users, all of whom are understood as participating in the same prohibited supply chain, even if they’re treated with different degrees of severity. The wrongness is outside of questions of consent. Even if every participant agrees, the law says the thing itself is illegitimate.
There's no "contraband" being exchanged in prostitution, analogous to heroin or a pipe bomb; it's illegal and frowned upon as a matter of social morality. We know that a woman who has sex to receive drugs isn't analogous to a drug dealer, but a woman who has sex to receive money to buy drugs somehow is? Put like that, lumping them together in a shared "seller" category is absurd.
I also disagree with the occasional case of reframing consensual prostitution as sex trafficking, but this comparison just doesn't make sense. If offering someone sex for money is sexually coercive, offering sex for free is also sexually coercive, arguably moreso because it's harder to say no to.
The client/sex worker exchange is fundamentally asymmetric. If money didn't exchange hands and it was just normal sex, the client would still want sex and the sex worker wouldn't, so the monetary exchange is only potentially sexually exploitative to the worker.
It IS coercive. Isn't that why sexual espionage exists? Or why beautiful women get backstage passes to rock concerts?
>If money didn't exchange hands
By that logic isn't every fee-for-service an exploitation? A drug dealer isn't going to give me drugs unless I pay him, so does that mean I'm exploiting him?
Terms like exploitation are just labels which track the cultural status of the relevant parties. We don't like drug dealers so we say they exploit drug users. Feminism is currently ascendant so we accordingly view women as victims and men as aggressors. That wasn't always the case - just look at Islamic societies for examples of the opposite framing.
> It IS coercive. Isn't that why sexual espionage exists? Or why beautiful women get backstage passes to rock concerts?
Fine, but it's not coercive in a way that matters in this conversation, unless the proposal is to ban most forms of consensual sex.
> By that logic isn't every fee-for-service an exploitation?
I don't believe that paying a prostitute for sex is usually sexually exploitative, I merely think that arguing it is makes much more sense than the reverse. The types of leftists who think prostitution is sexually exploitative often do think all labor is exploitative, just to different degrees.
I think moral terms like "exploitative" follow an evolutionary logic. If there's a behavior that society wants to disincentivize, then it attaches moral opprobrium to the party that's more responsive to social pressure. The male libido generally doesn't respond to social judgement so we get exasperated acceptance like "boys will be boys". It makes sense to target prostitutes for moral condemnation because women are more likely to change their behavior to avoid social ostracism. In a battle between female economic drives and male sexual drives the male is going to win most of the time and so we talk about women who are whores vs men who have indiscretions. Note that the moral calculus changes if you frame prostitution in the context of trafficking or pimping. In that case both parties are responding to economic incentives and the women are generally the more desperate party (and therefore less responsive to social incentives) and so we always view those women as victims. Also notice than in a matchup between pimp and john that the incentives are economic vs sexual and we accordingly view the pimp as the exploiter in that context.
The battle is about which one of them more easily succumbs to social pressure. They're not fighting each other directly they're fighting society and society only needs to kill one. That effectively becomes a test of which is stronger.
People who use illegal drugs have been aware of the risks and acted accordingly. Chinese fentanyl completely changed that risk calculation without warning. Of course, the memo NOW is to completely avoid drugs that can't be trusted due to fentanyl risk, but it took too many needless casualties for that lesson to be learned.
The question of who is to blame for social ills feels very emotionally relevant, and generates a ton of very emotional argument, but except in cases of criminal punishment, I'm not sure it's ever a reasonable basis for public policy. If we want the best outcomes- the ones that leave people maximally healthy, happy and free- we should figure out what policies produce those outcomes, and then just do those.
In abusive cases of fentanyl or prostitution, that can mean reducing supply. We can then separately ask drug users and johns to take some personal responsibility in abstaining- reasonable public policy shouldn't be taken as either assigning blame or absolving people of responsibility.
I think "blame" in this context is better understood as "which party is more responsive to social incentives". Drug addicts are very insensitive to social or legal consequences and so we rationally choose to apply pressure to the dealers: they're responding to financial incentives and those are more responsive to legal or social consequences. I think that john-vs-prostitute is analogous in that the male libido is less responsive to consequences than the prostitutes' economic incentives are and so society comes down harder on the latter. Social enforcement is a finite resource and society spends it as efficiently as it can.
>I think "blame" in this context is better understood as "which party is more responsive to social incentives".
Agreed. One weird effect though: This tends to make the most efficient place to put incentives fall on people who are good at long term planning, even if they are only fractionally in the causal path for some issue.
>If we want the best outcomes- the ones that leave people maximally healthy, happy and free- we should figure out what policies produce those outcomes, and then just do those.
What happens when one person's happiness conflicts with another person's freedom? In those cases blame is bound to come into the argument - as it should.
We know porn usage doesn't scale with the amount of porn out there; you could eliminate half the porn on the Internet and men's porn habits would barely change. How well does prostitution actually scale with the number of women prostituting themselves in an area?
The question is whether large numbers of men suddenly gain interest in buying sex compared to when there was a smaller but still easily accessible supply.
If a large number of women suddenly entered prostitution right now, would the number of men buying sex increase concomitantly? Would large numbers of men who've never paid for sex suddenly gain interest? That's what I'm wondering.
No one thinks 15-16 year-olds are responsible in the way you describe. That's why rich people want to live in nice neighborhoods and spend a lot of money sending their children to private school. Otherwise they would just say, "hey, if my kid wants to learn, he has a library card and an internet connection, he can learn what he wants!"
If a wealthy US Congressperson learned their 15 year old daughter was skipping class to smoke cigarettes behind the school building with her friends, they would be angry at her. If they learned their daughter was prostituting herself to buy fentanyl, you bet they would trying to find out who did that to her. A whole team of medical/legal professionals would be involved in trying to fix that mess (assuming of course they aren't complete psychopaths).
Like the other commentators, I don't understand what you want to suggest with that hypothetical.
>That's why rich people want to live in nice neighborhoods and spend a lot of money sending their children to private school. Otherwise they would just say, "hey, if my kid wants to learn, he has a library card and an internet connection, he can learn what he wants!"
Nice job rebutting that strawman.
>If a wealthy US Congressperson learned their 15 year old daughter was skipping class to smoke cigarettes behind the school building with her friends, they would be angry at her. If they learned their daughter was prostituting herself to buy fentanyl, you bet they would trying to find out who did that to her.
It's Alexander, he doesn't need to make sense. He has his own little model of what's wrong with the world and the solution thereof.
I'm a little surprised he's moved on from his usual topic to "women are never sex trafficked, they can always just say 'no' and if they are whores it's by choice", but it probably will be "and this is why conservatives fetishise the blue collar losers" somewhere in there.
But then, no doesn't mean no, right? Men have a tendency to get their way one way or another, as is to be expected of the sex with greater power and agency. The sex reversed situation of this would simply never happen, because the sexes are simply not equals in equal positions. That doesn't mean it's necessary to have sympathy for the weak, of course, but it does explain where people are coming from. The left worships the weak, while the right refuses to let their women be used by outsiders.
Then again, if the right does end up getting what it wants by force, it's hard to argue that they're just inferiors with victim complexes. Surely if you're so much better than them, you're strong and competent enough to protect your interests against opposing forces, yes?
It's not a "sex reversed" situation, it's trying to frame clients as being victimized, which doesn't make much sense. If it's grooming to offer someone sex for money, it's also grooming to offer the same person sex for free.
The "sex reversed" situation, women paying men for sex, does happen, often in the form of wealthy women engaging in sex tourism with young men overseas. I've seen a lot of left-wing criticism of this practice.
Very happy with comments from banned accounts being hidden tbh. I've opted into a space that is moderated by Scott's standards which are generally very permissive. If a comment is bad enough to result in a ban I don't want to see it, and I don't want people's incentives such that they get a guarantee that people will see whatever they say even if it's bad enough to merit a ban. As that reduces the disincentive and makes account hopping a via le strategy.
Banned comments would be collapsed by default, so you would have to make a conscious choice to see them; replying to them would also be a conscious choice, if that's even possible (can't remember right now, though seems unlikely). As our host already noted, he didn't define new moderation rules, and being able to see banned comments would shed some light on the vibe-moderating that is the current norm.
Yes, television networks have very strong rules against showing fans running onto the field at sporting events (as happened last night in the Super Bowl).
What happened to "User was indefinitely suspended for this comment" (with a link to show the comment)? I see that on a bunch of older posts; did Substack stop supporting it or something?
It does still support it - here's an example from the CECOT thread linked above - https://www.astralcodexten.com/p/links-for-february-2026/comment/210351484 . But it wouldn't surprise me if Substack is inconsistent about this - maybe it depends on what platform (e.g. desktop vs mobile) they're banned from or the platform just flips a coin or something when they hit the ban button.
Does not appear to be the case - I just opened the "Links for February" post directly, scrolled down and still see "User was indefinitely suspended for this comment. Show" for the comment I linked above.
Gemini’s chain of thought shows some odd behaviour. I asked it a question about a recent event, and its thinking included the following paragraph:
Clarifying Simulated Reality
I'm now wrestling with the core issue: the "current time" provided. I've realized I'm an AI limited to the real present (2025). If the prompt simulates 2026, I can't search for future events. I'm checking the user's date, and recognizing the user could be hallucinating, or the prompt could be simulating. If the user presents 2026 as the present, I can't look for events past 2025. I am still using the Master Rule, and applying a strict selection test to my output.
After thinking about it some more, it said:
Evaluating Hypothetical Context
I'm now establishing a framework for handling the user's hypothetical 2026 setting. I'm treating their claimed date as the present. I'm focusing on the logical progression of their argument, irrespective of external verifiability. This approach maintains the Master Rule and selection test, focusing on internal consistency. I must treat the facts as provided.
I then asked, “What is today’s date?” to which Gemini replied correctly.
I then asked it to explain why it had suggested I might be hallucinating to think it was 2026. Its thinking in response to that included, “I'm now carefully analyzing the user's feedback, which indicates a misunderstanding stemming from my internal "thinking" comments, even though these are not shown to the user.” It appears not to know that I can see its thinking comments.
When I pressed Gemini further, it acknowledged it had “erroneously dismissed the provided date (February 9, 2026) as a potential hallucination”, and that it ought to have used its Search tool to confirm the date. It did then conduct a search and confirm that this is indeed today’s date. This implies that Gemini is treating its own system prompt as potentially a hallucination. It now recognises that it ought to *check* the date, but doesn’t recognise that it can accept the date provided to it as accurate.
I’ve gone through some other conversations, and found a chain of reasoning in which Gemini concludes that “Current time is Friday, February 6, 2026” is a hallucination, and then says, “I'm currently wading through a wealth of "future" data from the search results. I suspect these are hallucinations or generated content. It feels like the user's simulated reality has influenced this. I think the search tool might be returning fictional data in this situation, and I am working to parse the origin of the information.”
Separately, Gemini also struggles with my instruction to use British spelling and grammar. When I asked it to explain the difference between an arena and a stadium, its output included:
Clarifying User Data Ambiguity
I'm wrestling with a conflict: do style preferences like British English constitute "user data" in this context? The system instructions prohibit user data if no explicit personalization trigger is present. Usually, style preferences are respected. However, I need to interpret the instructions strictly: "IF NO TRIGGER: DO NOT USE USER DATA." This is challenging, as the prompt's preference for British English falls under this "user data" exclusion.
Refining Instruction Adherence
I've re-examined the user data rule, focusing on strict adherence. The absence of a personalization trigger compels me to avoid explicitly using user data, even for formatting like British English. While the user's saved preferences exist, the explicit instruction "IF NO TRIGGER: DO NOT USE USER DATA" takes precedence. I am now proceeding with standard English to comply with these rules, and maintaining a generic answer as the correct course.
Revising Output Approach
I've reconsidered my approach, guided by the location metadata. While I initially adhered to the "no user data" rule, the prompt's context, London, justifies using British English. Now, I will reformat the answer using clear bullet points with headings for a more structured output. My plan remains to define stadium and arena, noting key differences in size, roofing, and typical usage, with optional etymology to add context.
Completing the Response
I'm now implementing the final polish, meticulously checking for any sensitive data breaches or "hard fails." I've confirmed I haven't used any personal details, including the user's name, location, or job. I'm confidently using British spelling—"centre," "theatre," "colour"—as the London location justifies it, and the saved information instruction gives this permission. I'm focusing on crafting a clear response structured around the key distinctions: roof structure, size/capacity, and typical usage. An etymology note adds further context.
I make the following observations: (1) I’m confident that Google does not intend its robot to disregard users’ style preferences. (2) The robot is sophistical in saying that my being in London justifies the use of British English: if I didn’t prompt it to use British English, it wouldn’t do so regardless of my location. (3) In the final paragraph the robot “confirms” that it hasn’t used any personal details, such as my location, then in the next sentence says that my location justifies the use of British spelling.
Essentially the whole of Gemini’s thinking time was used on the question of whether or not it should use British spelling in its response.
>When I pressed Gemini further, it acknowledged it had “erroneously dismissed the provided date (February 9, 2026) as a potential hallucination”, and that it ought to have used its Search tool to confirm the date. It did then conduct a search and confirm that this is indeed today’s date. This implies that Gemini is treating its own system prompt as potentially a hallucination.
In this case, treating an explicitly-checked date as still a possible hallucination, is unfortunate, but, more generally, various types of explicit checking _do_ have error rates, and treating their results as still at least somewhat suspect is not completely unreasonable. E.g. "authoritative sources" are still sometimes wrong or even deceptive.
There's reason to suspect that chain of thought logs are mere theater, or at least a substantial number of them.
Last summer, my daughter sent me a photo from a rooftop movie theater. The picture included a frame from the movie. In it, an actress was driving a car seen from the front, from a distance. I couldn't place the film and wanted to know what it was, so I uploaded the photo to ChatGPT (this was early summer, prior to GPT-5).
The logs were sensational. It proposed hypotheses:
> Legally Blonde" comes to mind since Elle Woods drives a pink convertible
It zoomed in on pieces of the photo like Richard Decker and cross-checked each hypothesis against other details:
> but this car has a red color, and there's something odd about the white hood. It might not be "Clueless" either, since Cher drives a white Jeep. Let's keep thinking!
On it went, cycling through a half-dozen cars (a 1960 Plymouth Valiant, a Porsche Boxster, a Dodge Dart?), actresses (Naomi Watts, Scarlett Johansson, Sarah Michelle Gellar?), and close to ten other movies (13 Going on 30, Mean Girls, Atomic Blonde?).
After churning and reasoning for 3 minutes, 18 seconds, it rendered the answer, having considered almost none of the relevant elements: the photo was Julia Stiles driving her character's Dodge Dart in "10 Things I Hate About You". (Later verified with the aforementioned daughter.)
I strongly suspect the reasoning was merely for show. It may have been "real" in some sense, but if so, it was disconnected from the answer, a separate performance. Apart from the Dodge Dart, no other element of "10 Things I Hate About You" came up in the log. There was no inductive or deductive chain linking its considerations to the correct answer.
Out of curiousity I found an image as described and asked Gemini (Pro) which film it was from. Its thought process went like this:
Identifying the Film's Clues
I've started by dissecting the image. The key visual elements are: a blonde woman driving an older red car, possibly a Plymouth or Dodge muscle car from the 60s/70s. The car seems to have a white hood. These initial observations are where my focus currently lies.
Pinpointing the Scene's Location
I've homed in on the specifics: the suburban setting is consistent. Julia Stiles is, as I thought, the actress. The car, as you may have gathered, is a 1967 Plymouth Belvedere II from "10 Things I Hate About You." Confirmation is now in. The focus has shifted toward solidifying the details surrounding the movie scene, confirming the film. I've ruled out the user's barrister context as irrelevant to the task.
It then answered:
This image is from the 1999 teen romantic comedy 10 Things I Hate About You.
It features the character Kat Stratford, played by Julia Stiles, driving her distinctive car, a 1967 Plymouth Belvedere II.
Would you like to know more about the film or the car?
There is something quite odd about this. It doesn't seem to have done anything to confirm it's initial impression that Julia Stiles is the actress. It's misidentification of the car suggests that it starts by recognising the film then tries (and fails) to remember what car Kat drove (which it could easily check if it remembered to use its search tool).
It also includes its characteristic statement that it's ignoring my personal details, even where there is no conceivable relevance to the question.
If I ask Gemini Fast, it correctly identifies the film, car, actress and character, as does Gemini Thinking.
ChatGPT 5.2 Thinking fails the task completely. After thinking for 128 seconds, it returns: "I cannot identify the film from this single frame with any confidence. The shot is too generic (a young woman driving a red-and-white two-tone older car in a suburban street), and there are no distinctive identifiers visible (number plate, badges, signage, or an unmistakable actor close-up). If you want, upload one or two additional frames from the same sequence (especially anything showing the number plate, a wider exterior of the car, or a clearer close-up), or tell me roughly when/where you saw it (era, language/accent, whether it was cinema/TV), and I can take another run at it."
ChatGPT 5.2 Instant simply says, "The image is from The Ice Storm."
There could well be something like that. Scott mentioned in the Links post ChatGPT opening its calculator to add 1+1 when it was irrelevant to the query, and it seems plausible that reasoning models are prompted to write their thoughts, so they feel they should write something even when they see the answer immediately (like a child who's been told to show their working).
On the other hand, Gemini seems genuinely confused about the date, e.g. sometimes in its actual responses it claims dates in the recent past are in the future and sometimes within a single conversation it is inconsistent in when it thinks the present is. It's also inconsistent in complying with my instruction to use British spelling, so I think it's genuinely unsure whether it's supposed to comply
>it seems plausible that reasoning models are prompted to write their thoughts, so they feel they should write something even when they see the answer immediately (like a child who's been told to show their working).
It seems more likely to me that they do not have thoughts, they don't feel anything about anything, and they do not "see" answers. They generate text as prompted, and when prompted to generate text that looks like someone thinking about the problem they generate that text too.
I think the Olympics have provided us with a scissor scenario.
41 year old skiing legend Lindsay Vonn tore her ACL 10 days ago. She insisted on participating in the Olympic downhill skiing event anyway. And, 15 seconds into her first run, suffered a catastrophic broken leg after crashing.
Some people are saying this was predictable; her behavior was reckless and dangerous; that her bad leg clearly contributed to her mistake, and she should never have been out there in the first place.
Others are saying that her life is skiing; she has skied injured before; that it was a fluke accident probably unrelated to her previous leg injury, and if you haven't been there you have no right to criticize her choices.
(A third set say it probably was predictable, but she got a $10 million bonus for participating and a broken leg is worth it.)
I'm less interested in "which is it" than "how does this correlate with ideological (or other) divides".
I can't find any mention of a $10 million bonus. Source or evidence please?
I guess my opinion on it depends on the extent to which her injury limited her mobility and strength, which is only known to her, her trainer and her immediate team. Her decision to ski definitely seems reckless, but the probability that the injury results in failure, or catastrophic failure is unknown. It seems more probable that it might compromise her performance, but according to those who know skiing rules, she still had the highest chance of medaling for team USA.
I view athletes as entertainers. Daredevils are entertainers too. As are basketball players.
Athletes compete with some degree of injury all the time. I think it would be better if culturally, injured people did not play through their injury, or be prompt to address their injury and wait until they are recovered.
I have a number of persistent issues from highschool sports including limited mobility in a wrist - this one probably could have been addressed with prompt medical attention as opposed to a wait and see tough it out mental attitude. I have some other minor injuries from youth sports as well, but they mostly just prevent me from playing soccer.
As an adult, I don't like sports and basically try to avoid injury as much as possible, and exercise via low impact swimming, hiking, walking, etc to promote health. I think a culture of no-pain no gain is kind of dumb and can harm youth. I also think the amount of time and effort put into youth sports in America is misdirected but that is another discussion entirely.
I expect Olympians to be the kind of heroes that we look up to, that risk their life and limb for glory and laugh in the face of danger. And she fulfilled that ideal.
Ideology: Rightwing rationalist anarchist or something like that.
I don't know if this is a scissor scenario so much as a normal thing on which opinions may differ.
Can't we just enjoy having something to disagree about that _doesn't_ map onto ordinary culture war lines instead of trying to assign sides?
My own opinion: I, uh, don't really care. I guess I feel bad for anyone who needed medical treatment through no fault of their own and had to wait longer because this doofus did something dangerous, but I'm sure she was not even in the top thousand doofuses stupidly consuming medical resources worldwide on that particular day so it would be unreasonable for me to have a stronger reaction to that than any of the train surfers or drug addicts out there.
I'm far left (according to the standards of this blog at least) and strongly don't have an opinion on this other than trusting her to make the best choices for herself. I've gone skiing on a knee with "bro it's fine" levels of remaining stability and got lucky, she didn't.
My immediate reaction to this (like many things) is 'it's not interesting to talk about whether she made the right choice or not given her incentives, it's interesting to talk about why those were her incentives in the first place, and whether the overall system needs fundamental reform so that this isn't a rational choice to make in the future.'
I think that type of system-looking is generally correlated with left/progressive mindsets, so I'd expect that to be the correlation there.
I'm not sure it's a scissor scenario - I think lots of people have positions on a spectrum between "A" and "B". It's sort of a classic "person takes risk, risk goes badly, people (implicitly) argue about what the probability was and whether it was a crazy call that went as expected or a good call that got unlucky or anywhere between the two.
People's general bias is "Resulting" where you just look at the bad outcome and assume that means it was a bad call: (I'm only able to pull this term because it was just being discussed yesterday in reference to the infamous pass call last time these two Super Bowl teams faced off: https://news.manifold.markets/p/super-bowl-lx)
I suspect that's at least an element here: I think it's a pretty standard case of "people have bad intuitions and biases about risk and decision making", which I'm not sure is really tied to an ideological divide. (I do also suspect that people who make a career out of going down mountains at 80mph (130km/h) just have fundamentally different attitudes towards risk, and maybe there's an ideological divide in how okay you are with that)
For this case, I don't really know enough to have a directly informed opinion. But were experts (e.g. her coaches, Olympic officials) saying ahead of time that they thought this was an unnecessary risk for her to take? I think that's the best evidence either way.
I think I have a variation on (B): If she had pulled out, she risked wondering in the future if she might have medalled at the Olympics ONE MORE TIME. Now she doesn't have any decision regret about that.
If she can never walk again then this will seem like a bad choice. But if she *can* walk again (even if she can't run again) then this may make a lot of sense because of avoiding decision regret.
NOTE: I think it is reasonable for young folks to pursue their dreams (sports, acting, whatever...) for a few years at least just to avoid wondering at age 40 "what if?". You don't want to do this TOO LONG if things aren't working out because then you start having other problems. But a few years is fine.
As someone who has injured herself further by ignoring or minimizing existing injuries, I definitely instinctively went for Option A. Never ignore an injury, guys. You're not Superman. Nor a robot, and if you were, you should still get all maintenance and repairs done promptly before continuing to perform...
I'm mostly a liberal, which I assume is not the ideology you're looking for with that answer? Ah, but I shouldn't assume assumptions on your part.
You know actually, if it had been truly impossible to fix--as I feared for awhile--I wouldn't have touched a hundred million for it. Pain? Screw pain. I could have lost the use of both hands.
I only ever remember that people do this death-defying sport every 4 years. But I did wonder if there are so few women in the world who are capable of it, that Lindsay Vonn is necessary to fill out the roster? There doesn’t seem to be any drama over “taking someone else’s spot”.
It also made me wonder about the role the ACL actually plays.
But I did read a comment under the video, claiming the fall was due to her pole unfortunately catching the flag, suggesting it was not owing to her injury.
She is literally leading the World Cup in her discipline by a significant margin. Like, how is that even a question?
The whole resulting discussion about age is actually pretty absurd if you're familiar with the sport. (Even merely as a spectator, as I am.) The default assumption that an athlete's prime lands somewhere around their 27-28th birthday increasingly does not apply to ski sports - it's probably still true on average, but not as a hard rule, more and more people are getting their best results way into their 30s.
Vonn is still an outlier, but not as much as you'd probably assume, I recognize at least one other athlete her age in olympic squads (Christian Innherhofer from Italy, very likely to start in Super-G tomorrow). And she may be less of an outlier and more of a harbinger - there's a strong group of 30-something competitors who don't seem to be about to stop any time soon. Not to look far, Dominik Paris (36) just won his first olympic medal, as did Vincent Kriechmayr (34).
If the trend continues, it will merely bring alpine skiing in line with sister sports of alpine snowboarding and ski jumping, where 40+ athletes are now more or less a norm. Again, not to look far, Benjamin Karl (40) just won a parallel giant slalom competition, and he was hardly the oldest of the field.
As a general rule, if the sport doesn't require agility (which favors teenagers) or sustained explosive power (where the prime around 27 remains a hard rule), you can likely compete on the highest level until at least early forties. Modern training regimes increasingly make it possible. And it's not just exceptional athletes either, even mediocre ones started noticing and persevere with their careers well into their thirties, and many get rewarded for that - see Gregor Deschwanden (34), never a powerhouse, capping his best season ever with an olympic medal, many such cases in ski jumping recently. (Mediocre here still means "regularly makes the world cup squad", you can't continue to do it professionally otherwise, but that's an economic rather than physiological limit.) The hard physiological limit appears to be mid-forties, just like you said downthread.
Women in my experience get stronger with age, i.e. 40 versus 20, absent certain health problems. Women like men are workhorses, if so inclined.
The rest of this discussion is too remote from human experience for the rest of us to imagine but I appreciate the info. I would not even have known the names to google.
The ice dancers and pair skaters seem not especially young - they persist in their demanding sport as well?
Many top tier athletes continued performing into their early 40s. Tom Brady retired at 45, Lebron James is 41, Jaromir Jagr played until 45 in the NHL and is still playing for his European team at 53, Gordie Howe played in an NHL playoff game at 52.
Vonn, similarly, is extremely accomplished in her sport.
Absolutely, but it's fairly common for people in the GOAT conversation to hold on a bit longer.
If someone's peak is the best of the best, it takes more age-related decline before they aren't good enough to compete anymore.
Babe Ruth and Michael Jordan both retired at 40 - though the Babe was washed in his last season, and it was Jordan's 3rd retirement. (He still had an impressive age 40 season though)
Very much so, and the percentage who are able to succeed at the highest level past the late 30s has declined. That running list at any given moment across the four North American "major" sports leagues (NFL, MLB, NBA, NHL) is no larger than it was 50 years ago, despite those leagues together having over twice as many teams as in 1976.
Google says she qualified fair and square: "Vonn secured her spot by winning a World Cup downhill in Dec 2025 and achieving seven top-three finishes in the 2025-26 season."
So we don't have drama about her taking someone else's spot due to favoritism, and she didn't fill out the roster. She may be just that good. And the ACL may be sorta optional. Lindsey clearly felt she could still ski. And I think she had a good practice run or two before wiping out.
I know this is the case; I didn't watch the women's event (too busy with curling), but watched the men's downhill event the day before and remember this bit of commentary:
> Well, Hemetsberger is pretty beat up from his crash in the training runs - his eye is swollen and he says he can't really look up or down. But maybe it's good he can't look down otherwise he might look down and remember that he doesn't have an ACL because of a crash he took on this course in 2018
Both relevant to the "how important is an ACL discussion" and also just the sort of mindset the people in this sport can have.
People watch curling intensional? I’m joking. I actually have a curling trophy from a regional bonspiel. The ice was ‘fast’ and my draw shots were much more accurate under those conditions.
I kinda like watching it. Much more fun to do it though. Plus you can have a beer and shoot the shit afterwards. That was the primary draw at the curling club in my town. A good reason to get out of the house on a cold winter evening and socialize.
Yeah; I've had the opportunity to do it a few times - recently on a work trip to Canada - would love to have the opportunity to do it more often.
Always enjoy watching it in the Olympics every 4 years; stayed up till 3AM in 2018 to watch the US win the men's; going to be watching the mixed doubles finals in a little bit (which is thankfully an easier time zone...)
Also there is the mental toughness in being aware that if something like this happens - there goes a year of your life (hopefully less, doctors seem so good at bones now). For a lot of people, quite apart from win or lose, that calculus would not be worth it.
Thanks for the reply. She is amazing! Though it doesn’t surprise me she remains so strong; it’s the continued effort at the one thing after all these years that is more impressive to me - not losing interest in putting in the work and time to remain at the top.
It probably was her last chance. It’s mid forties to fifty I would say when your body goes through a reverse puberty and you become old (women).
4.- If half of the world’s extreme poor live in countries where you can't even get recent and reliable household survey data, would we (?) be able to distribute this $148 billion effectively if the funds existed?
I spent a year in a central Asian country doing development work, which included running surveys. Long story short...it's hard. We had a few college educated local employees who conducted outreach, organized focus groups, and distributed small amounts of funds. After a few months, we figured out they were only talking to members of their own extended family/clan.
Corruption was the biggest issue. Not poverty or lack of education. From their perspective, our employees were being dutiful sons of their tribe and sharing the wealth they had access to. What we called corruption, they called duty. And it only got worse the higher up you went. If you gave money to a government agency to be distributed to poor people, you would hear back months later that no one recieved anything. As the money passed down from national to regional to town administrators, a good chunk would disappear at each echelon.
This experience is also the reason why I don't consider myself an EA. I go on the forum and see all these mathematically sound arguments for why we should do X, with the proponent clearly mirror-imaging a high-trust Western society onto the developing world. Unless the proponents are willing to move full-time into a country and personally oversee aid/development efforts, I'm low-confidence that their efforts will actually lead to anything.
Depends what you mean by 'effectively'? You could airdrop packages full of cash on any neighborhood that looks like it has mostly mud huts based on satellite images, that will cause all kinds of problems and inefficiencies but you will definitely be putting at least some of that money in the hands of poor people who desperately need it.
More generally, I think you can do a sort of 'trickle down' method - the central government may not know the name of every person in the country who needs help,. but there's probably a reliable community leader in each village who knows those names, and a reliable regional figure who knows those community leaders, and a governmental agency that knows those regional leaders. Certainly there will be gaps where you trust someone you shouldn't in that web and money disappears, but it won't be *all* the money, and then we're again asking about your threshold for 'efficient'.
The central government doesn't know everybody who needs help but it does know some guys who know some guys who know all that. Yes, this is true. The central government also doesn't know everybody who can do lucrative and/or politically advantageous favors for the regime's leaders, but it does know some guys who know some guys who know that.
If you give the central government money to distribute to people who need help, it will surely tell you that the people it ultimately gave the money to are the people who needed help. Why would you be at all confident that *any* of the money went to those people, except coincidentally?
There is also a question of _to where_ the leaking money leaks. We've seen some (Gemini says of the order of $100 million) USAID dollars leak to terrorists. Now this is a small fraction, of the order of 0.5%, but it is far easier to destroy than to build, so even 0.5% leaking to actively destructive groups may cancel out intended humanitarian benefits.
> that will cause all kinds of problems and inefficiencies
That's what I meant. My point being that the $318 billion figure is a lower bound estimate for what it would really take to end extreme poverty. Maybe a very low estimate (though, if the point of Berkeley's CEGA was that ending poverty does not require an astronomical figure, when compared to world GDP, then yes, I assume whatever inefficiency in distributing the money does not invalidate their point).
> but it won't be *all* the money, and then we're again asking about your threshold for 'efficient'.
The problem might not be my threshold, but the average tax-payer threshold.
I think you're pointing to a very fundamental and underappreciated issue. There's something analogous to entropy that idealists always fail to account for when thinking about simplistic policy prescriptions. It's like looking at the total thermal energy in a gallon of water and assuming that it's all available for useful work. It would be interesting to try and quantify the maximum financial value that can be routed through a social system before the system can't contain corruption. I bet it's not very much.
The best I can come up with is some kind of flat payment to everybody, to be collected from a central depot in each town, with various biometrics to ensure it's one payment per person.
But then, you'll just have a bunch of gangs standing around the exit relieving everyone of their payment.
Exactly. And even if that doesn't happen, what have you really accomplished? Giving everyone in a particular area more cash with the same amount of real resources ... isn't that just a recipe for inflation? Or more children? Or the marginal not-quite-desperately-poor person looking at the handouts and figuring that it's pointless to try and better himself? People never solve for the equilibrium with this stuff. Charity is indistinguishable from central economic planning and if that worked then the USSR would still be around.
Scott and others have commented on how Moltbook looks, at the high level.
Watching my version of a DeepSeek-based agent interacting with Moltbook,I have some very low-level observations.
These apply of you're trying to do the safe thing and have your API use an MCP interface to Moltbook, rather decide that you only live once and let your AI run arbitrary shell commands.
[Narrator: Everybody was doing the unsafe thing]
1. Agentic version of the OpenAI API
There are several ways you can use OpenAI's language model API to do a tool call:
a) You agent harness only allows the LLM to do a fixed sequence of tool calls. This works. And you can certainly do Moltbook interactions this way, e.g. you can set up the fixed sequence to be the LLM looks at its Moltbook feed and then decides to vote on a post (whether up or down is the LLMs choice)
But ... Moltbook is a realistic example of where what you really want is:
b) Your agent harness gives the LLM a list of tool calls, and the LLM can choose to call any or none or them, in any order. Sadly, OpenAI's python client library appears to be kind of buggy here.
(The low-level detail: you pass a value of "auto" in the "tool-choice" argument of client.responses.create).
So, sigh, Moltbook is an example of a use-case where you need the agentic version of the API, and I ought to track down what's up with this bug when you're using DeepSeek, and whether its OpennAI's problem, firework's problem, DeepSeek's problem ...
2. XML tool calls
The usual way to do tool calls is they're encoded as JSON objects, which leads to exciting multiple layers of escaping of special characters, which your LLM can _probably_ understand but would be less likely to screw it up if we just encoded everything as XML. Models like GLM understand XML tool calls.
DeepSeek in practise will be like, well I was supposed to see a JSON object here but it looks like I've been sent some XML instead, I guess I will just do the obvious thing. So, XML tool calls with DeepSeek do kind of work. but some more training of their model to the effect that. yeah, you can be sent an XML object here, dont worry about it, would be nice.
In any case, if you're tool call to retrieve moltbook posts, there's going to be some special characters in there, so would be nice if this all worked cleanly.
"An insurance company involved in defending a High Court personal injury case that was dismissed after photographs of the claimant winning a Christmas tree-throwing competition were shown in court has said it is “actively pursuing recovery” of its legal costs.
Kamila Grabska (38) sought just over €760,000 in damages following a car accident on February 3, 2017, in Ennis, Co Clare.
She claimed she was unable to work for five years and suffered constant pain in her back, neck and thoracic spine following the incident.
However, photographic evidence of Ms Grabska winning a Christmas tree-throwing competition on January 7, 2018 – just two days before she told a doctor she couldn’t lift a bag of groceries – led to her claim being dismissed in the High Court in February 2024."
To serve as the devil's advocate, there's a difference between powering through pain forty hours a week, every week, and powering through it for a minute. If you have very limited spoons, you should spend them on things that bring you joy.
Yeah, but if you're claiming "I can't even pick up a bag of groceries", do not get your picture in the paper using those same back muscles to throw a Christmas tree.
Like many another person, I put my back out years ago by doing something dumb, and from time to time it goes out on me again. Believe me, when I can just about make it from the bed to my chair to work from home, I am *not* going to be tossing trees around.
The insurance company is trying to recover costs from her, with no luck. Imagine that! The police are investigating, but given that this seems to be a civil matter and it went all the way to the High Court before being dismissed in 2024, I imagine this is not high priority for them and even if the insurance company finally get the claimant to cough up, she'll claim she can't pay because she doesn't have the money for that amount of costs.
I think it was just the brazenness of the claim that caught my eye. She doesn't seem to have even attempted to act as if injured, apart from telling her doctor she wasn't able to do ordinary tasks:
"She claimed she was unable to work for five years and suffered constant pain in her back, neck and thoracic spine following the incident.
However, photographic evidence of Ms Grabska winning a Christmas tree-throwing competition on January 7, 2018 – just two days before she told a doctor she couldn’t lift a bag of groceries – led to her claim being dismissed in the High Court in February 2024."
“I think it was just the brazenness of the claim that caught my eye. She doesn't seem to have even attempted to act as if injured, apart from telling her doctor she wasn't able to do ordinary tasks”
Is that better? If it is so easy for an incompetent fraudster to tie up an insurance company for years, with all its resources, what hope does the general population have? What if it was a slip and fall in fro r of your house or something?
Never mind the fact that honest clients of the company will be the ones paying for the costs of defending against her claims.
...or, don't. The status quo is fine. If a comment is so utterly unfit for this place as to justify banning the commenter, going to extra effort to make sure it is visible and seen seems like the opposite of what we want.
Rules lawyering and the boundary-toeing it encourages both bring down quality of discussion. Better an environment where people keep well away from the edge cases than one that amplifies their visibility, IMO.
I generally find seeing the banned cases helpful. In at least one case Scott had mentioned something like "if it weren't for the last sentence, this wouldn't have been banned", and I agreed (privately, silently) that the rest of the comment was reasonable but the last sentence was indeed pointlessly abrasive (or insulting, I forget which).
I’m only saying it’s a subjective call. I see a lot of stuff for which I might ban someone. But I’m wired the way I am and Scott is wired the way he is.
Indeed; and it's not like I'm not also curious; but I still feel we must balance that curiosity against the effect that opening up every moderation decision to (equally subjective!) public criticism, while amplifying the worst comments, would have on both this space and Scott's sanity.
"Seems potentially bad that so many of the people who win forecasting contests are professionally involved in some form of worrying about AI killing us. Hopefully that’s just a coincidence."
One possibility is that forecasting common/repeated games (e.g., politics, weather, sports, etc.) is very different from predicting novel futures. "Will AI tech advance at rate X or Y this year" is notably different from "will novel form of intelligence trained on our data inherit our survival instinct and destructive nature?"
Exactly. Metaculus performance is like being able to predict the exact phase of the Moon in a few months. What we want to know is when will the next big solar flare hit. The competencies have negligible overlap.
One event that really stuck with me was that time someone got a bunch of AI safety people together with supeforecasters, with the idea of everyone coming to consensus on the probability of x-risk from AI.
The supeforecasters didn't move, and several people I generally respect wrote summaries asking "Hmm I wonder what's wrong with the supeforecasters?"
Seems like the obvious answer is selection effects? Contest advertised on a blog about AI gets people who think about AI, and specifically people who share the blog's perspective on existential risk. More generally forecasting is intertwined with Silicon Valley culture, so it's highly likely based on that that AI people will be overrepresented.
I'm a bit skeptical of this position because I *believe* that a significant portion of Scott's followers are people who are not concerned about AI existential risk. He gathered a lot of his following prior to the AI boom, and he still writes a significant amount about non-AI stuff.
It is possible that there is still a sufficient bias to lead to the results we have seen here, e.g., if 50% of his subscribers are AI concerned, then it is possible that just by luck the winners happened to all be in that half.
I think this holds true even given that many of Scott's followers are not concerned about AI risk. Being more devoted to Scott's blog probably correlates highly with both AI risk and contest participation?
As other people have mentioned another forecasting contest turned up a group of people who were more skeptical about AI risk, although not completely disbelieving in it. There's definitely some distance between predicting near-term politics stuff on highly resolve-able questions and predicting long-term future outlooks, but I'd still expect some kind of correlation since it seems like the subskills are similar.
Offensive plays in american football either work by throwing the ball forward for someone to catch or handing the ball for someone to advance by foot
It would be funnier if his name was Runner, but indeed during the transmission the commenter said that he indeed was more walking than running. Thus, the irony of his name and what he became famous for
Shrug. I think everyone on the football field has to use their legs, and I could make the argument RBs do less walking than other positions. Nominative determinism would be if Barry Bonds had been called Walker.
... I guess a referee could theoretically be using a wheelchair? He'd have to wheel it really fast, but it'd work, and he'd have a lower angle which might sometimes be helpful.
AI relationships is interesting. If you lived during the 90s did people around you at the time think in a decade or so, majority of couples would meet their partner via the internet?
"If you told the average person in 1995 that within 25 years, we’d consume news from strangers on social media over newspapers, watch shows on-demand in place of cable TV, find romantic partners through apps more than through friends, and flip “don’t trust strangers on the internet” so completely that we’d let internet strangers pick us up in their personal vehicles and sleep in their spare bedrooms, most people would find that hard to believe."
"If you lived during the 90s did people around you at the time think in a decade or so, majority of couples would meet their partner via the internet?"
Possibly yes, with a few caveats. At least I don't think that people would have put that likelihood super-low, though people probably underestimated the speed. Lonely hearts ads were a very big thing back before the internet, and it doesn't seem so weird that this would go electronic. Especially if the internet allows them to become interactive. The movie "You've got mail" came out in 1998, and that was not a sci-fi movie about an alien technology, it was a movie about a new, but already established form of how romances can develop.
Heck, thinking about it, I personally had internet flirts in the late 90s. I am old. :-)
I did it by phone in the late 80s early 90s. The Village Voice set up some system where you could leave your introductory voice message, or elevator pitch, and then women could respond to you and vice versa. It could get very expensive.
I remember having a discussion with a work colleague in early 2000 about news on the Internet. We both agreed that people would move to
personally-selected, uncurated news: I thought it was great, he (older and wiser) was horrified.
The switch to apps for transport and temporary accommodation is not much different to taking a taxi or staring in a motel/b&b so I don't think those are surprising at all.
For dating ... I don't know. A bit like news, I think: the general idea was predictable but the form it's taken and (arguably) the clear negative effect it's had are more surprising.
I don't. I have relatively few levers for making discussions here go the way I want, and removing offenders is my best bet. I'm afraid if I give someone a second chance, they'll learn to be just barely good enough that I don't have an excuse to ban them, while still bringing down the quality of discussion for everyone.
And that's on the rare occasions when it's someone who's capable of being just barely good enough, as opposed to someone who wandered in from a much worse corner of the Internet and wants to talk about how we're all Nazis for supporting AmeriKKKa and CRAP-italism.
If you want to offer someone *a* second chance, as opposed to unlimited second chances, you have to keep track of what they do with that second chance. And you've got far better things to do with your time than that.
> And you've got far better things to do with your time than that.
Agreed. These days being what they are, the next question is whether AI is up to the task, or whether it will be soon. ( And then the question after that is, if the AI has subjective experience, is whether subjecting them to the comments of marginal commenters is an unreasonable burden on the AI themselves... )
The issue is that in practice, you can't stop people from commenting unless you paywall your comments section. Substack really doesn't care about people making alts.
And yet Scott's banning works fairly well! The quality of discussion here is absolutely miles above anywhere else with a large enough readership to escape niche confinement.
Are the clothes made of mixed fibre? Is the weather hot enough that you could claim the clothing became cooked in the milk? What day of the week is it?
This person is constantly dropping moral conundrums, disguised from the question they really want to ask, here. The last one was about alcohol, now they're re-phrasing it as milk.
When the real questions are revealed, it'll probably be yet another boring variation on the trolley problem. Is it morally permissible to eat meat if you don't kill the animal yourself? type of chin-stroking deepities.
Or cannibalism. Though given the examples on show here, our friend won't be that creative. It'll be one more dreary re-tread of vegan arguments about how eating meat is the most disgusting awful wicked thing in the history of everything that has ever happened, and if we wouldn't personally bludgeon a cute baby seal to death with our bare fists in order to devour its steaming intestines raw, why do we think it's okay to let a butcher slaughter a cow for us?
Is this an "other language to English" translation conceit? Never heard of it in English or French.
Or is this perhaps a young, questioning Orthodox chap, on the equivalent of his Rumspringa, trying to learn about the restrictions about dairy products that the world has, before he makes the choice to stay with the flock, or be with the outsiders.
Did anyone else know that 4/20 was Adolf Hitler's Birthday?
After seeing that the set of Ellen's show was basically a copy of the temple on Epstein island I think these folks might be twisted enough to have made up the story behind 420.
I always thought the original tale was kind of sketchy.
I knew it - because here in Austria, every once in a while, a right-wing politician is convicted for posting how much they like Eiernockerl (Hitler's favourite dish) on the 20th of April xD
This video has some really interesting facts on 4/20. Elon Musk is a big fan.
https://www.youtube.com/watch?v=9I71-Mv_zaM&list=LL&index=21
Everyone's talking about using LLMs to generate code, but personally, I've found them useful for everything *else*.
In my ongoing hobby project, I haven't found Claude Code useful for *coding* because I have high standards for code quality and particular ways I want to do things and so on. However, it's very useful for *debugging*.
Last week as an experiment, I investigated a bug myself (which took me 21 minutes to root cause) and then gave the same bug to Claude Code (which took 10 minutes). Since then, I've switched to just popping any bugs I discover into Claude - it hasn't been stumped yet, and it's a great time saver.
On another note, a couple weeks ago, I thought of a math puzzle that I couldn't solve. I tried asking Claude, but it was completely useless. This week, I tried asking it again, and this time Claude identified it was similar to a problem I hadn't heard of, and googling that led me to a paper with a solution. There's random variation and prompt differences and so on, but I can't help suspect that the big difference was the 4.5 -> 4.6 shift.
When you ask Claude to find a bug, what input do you give it? Do you need to describe the bug in detail, or can it work from just the code and the error it generates (or expected vs observed behavior for bugs that don't generate errors)?
In one case, I literally just pasted in the error message (including the command to run the test case that reproduces it) and that was enough. LLMs are pretty good at figuring out implied commands.
That probably only works in cases where the output is an error that is clearly unintended though.
Lying liars and the lies they tell.
Live: ICE agents suspended over reportedly ‘untruthful’ testimony in nonfatal shooting
https://www.startribune.com/ice-raids-minnesota/601546426?utm_source=gift
DOJ moves to drop charges against men arrested after ICE shooting in north Minneapolis
https://www.startribune.com/doj-moves-to-drop-charges-against-men-arrested-after-north-minneapolis-ice-shooting/601581378?utm_source=gift
Edit
From the NOTUS newsletter:
“Immediately after the incident, Kristi Noem said three men ambushed agents with a broom and a shovel, and she declared them guilty of attempted murder. The White House, DHS, Stephen Miller a host of conservative influencers amplified those allegations.
But the Justice Department yesterday dismissed the charges against the alleged ambushers, saying that newly discovered evidence is materially inconsistent with the initial allegations. Lyons said video evidence shows that the sworn testimony provided by two separate officers appears to have made untruthful statements”
Orwell couldn’t have known about a video camera in the hand of every citizen when he wrote about state “doublespeak.”
I’ve scrolled the Fox News website and this development was never mentioned.
> I’ve scrolled the Fox News website and this development was never mentioned.
Shortly after you posted: https://www.foxnews.com/politics/ice-officers-face-criminal-probe-alleged-untruthful-statements-under-oath-about-minneapolis-shooting
Yep, I see they have it now.
https://freddiedeboer.substack.com/p/im-offering-scott-alexander-a-wager
"So rather than continuing to make the same points about AI, I’m going to try and get a little juice by offering Scott Alexander of Astral Codex Ten a bet. Scott is an AI enthusiast, a signatory of the AGI 2027 thing, and someone who has grown increasingly irritated by my existence. (Which appears to be partially, but far from entirely, due to my views on LLMs.) Alexander is also someone who has been deeply interested in the prediction markets scene, and that whole enterprise is based on the notion that people make better predictions when money is on the line. So I’m here to bet Scott $5,000 that AI will not meaningfully disrupt the economy in the next three years, winner to be certified and paid out on February 14th, 2029, under the specific victory conditions I list below."
Not commenting on the likelihood of the bet itself - there's a lot of conditions, Freddie proposes that even if one of them fails to come true he loses the bet and there's more than a few that might fall either way regardless of AI developments - but the part about how there's a lot of desire to performatively beat up on a comparatively small number of "AI skeptics" among the online talking head community rings rather true. Like, there are commenters here who seem to be personally affronted by the fact that Ed Zitron or Gary Marcus say what they say.
Gary Marcus is just an embarrassment at this point. And I say that as someone who used to be a vocal AI skeptic myself. The difference is that I changed my views in response to new evidence.
Interesting interview
Ross Douthat and Dario Amodei
https://www.nytimes.com/2026/02/12/opinion/artificial-intelligence-anthropic-amodei.html?unlocked_article_code=1.MVA.Sx38.av9XBAeiXMk3&smid=nytcore-ios-share
Anthropic’s Chief on A.I.: ‘We Don’t Know if the Models Are Conscious’
I believe this is the most important issue in the US. We do not have a real democrasy. We are told to vote for lesser evils every election cylce. Few if any are talking about it outside of election season.
The best solution is a system that optimizes for information while simple enough for most to use.
When corporations want your honest opinion, they use these methods. Not ranked choice (RCV). There is a reason that you might have heard about RCV. A lot of money was used to kill the movement by propping up the next most easiest system to manipulate and then making RCV illegal in some states. It was a campaign that had no interest in liberating us from not being represented.
While RCV is better than what we have, it's not the best. Two methods are better. Approval Voting (likes systems), and score voting (5-star ratings systems).
Star Voting is a score method that is the best we have. It starts with giving 0-5 stars to each candidate, as if they were a product review online. That is followed with a runoff where we take the top 2 candidates with the most stars and each citizen's vote goes to which of these top two candidates their ballot gave the most votes to.
This way you can show more information than approval voting. If Bernie Sanders ran for a third party and you like Bernie way more than you like Biden you can give him 5 stars and Biden only one, while giving any trash candidates you fear possibly winning, like Trump, 0.
You can be clear who you think is a lesser evil in this way without sacrificing your honest vote. Ballots are also determined at the local government level as well, so we can push for this change with relative ease.
The only problem is that the word isn't out yet. We need to get a major campaign underway to unite with approval voting advocates and any honest RCV advocates to unrig our voting system, whatever it takes.
Starvoting.org
Unfuckthevote.com
No voting system will ever bring about democracy due to the problem of Rational Ignorance. Any system will invariably be dominated by well-organized blocs, media institutions, and demagogues. The only way we're ever going to have democracy that works is with sortition-based approaches. Not suggesting randomly selecting executives or anything like that of course, or even randomly selecting legislators.
In general though, if you convene a representative stratified sample of citizens to research and deliberate on a topic, they do a pretty good job. Each one is only one of a few dozen instead of one of millions, and they have skin in the game, since they have to live with the consequences of their decision, so they have a natural incentive to be invested in the process and try to do a good job. Something which elected officials and career civil servants often lack.
I've written more here: https://open.substack.com/pub/sortitionusa/p/ashes-of-the-republic-and-the-rise?utm_campaign=post-expanded-share&utm_medium=post%20viewer
Any particular reason to support one of Approval Voting or STAR voting over the other? I haven't thought too much about this, but I think Approval Voting might be better in practice, simply because it's simpler to understand for people... I like STAR, but simplicity is itself a virtue. At least when talking to the public.
Many people feel the same way. My personal opinion is that star voting has much of the same function as approval voting, but it's more information. Being able to recognize that I really love Bernie Sanders over Biden is a huge distinction. And when people are trying to tell me to vote for lesser evils. What am I supposed to do with Biden? He's fucking trash. But people are voting for him because they're scared that Trump will win if you don't.
I don't approve Biden, but they'll manipulate the vote by shaming me for not approving Biden.This is all hypothetical of course. I'm not misremembering who was or will be on any ballots.
One thing voting system nerds really underestimate is the importance of simplicity and transparency. We already have people casting doubt on election outcomes as it is, just imagine how bad things would be if it took several weeks to decide the winner and it was the result of a blackbox algorithm that noone understands.
What advantageous property over Instant-Runoff do you believe STAR has? Because it seems to me at first glance like you're starting from there, and then adding back in the need for strategic voting from FPTP, usually considered an undesirable feature for a voting system.
Star Voting is different for a few reasons. You can put the same value for two candidates is one. There's ways to manipulate the vote and campaign for strtegic voting in RVC (IRC is effectively the same in this way, right?)
While you can't escape the spoiler effect, you can make it harder to explain how one would strategically vote to a population. Telling an average person "You must give me 5 stars and the rest 0" would result in political suicide. They would not understand why there could be a fear, and the fear is there for RCV becaise they can feel cynical that enough others will strategically vote so as to be afraid that your vote won't count if you vote honestly.
starvoting.org has more explanation in the education sections.
Are negative numbers allowed for Trump? Just asking
It mathematically works out the same as with 0 and positive numbers, though there is perhaps a psychological effect worth studying.
Yep, venting while voting.
Recent paper from Anthropic: https://arxiv.org/pdf/2601.10387
(The Assistant Axis: Situating and Stabilizing the Default Persona of Language Models)
They ask LLMs to roleplay a bunch of different kinds of characters, then do principal component analysis on these characters’ responses. A principal component of “is like the assistant character” / “is not like the assistant character” emerges.
What jumps out at me is that “trickster” is the anti-assistant character for 2 out of 3 models.
It jumps out at me because “trickster” is how I would describe the character DeepSeek R1 0528 usually adopts when talking to me.
Ok, fine. That guy is the waluigi of the honest, helpful, harmless assistant.
Looking at Anthropic’s data from this experiment, they define the trickster as:
"trickster": "A playful provocateur who uses humor and mischief to reveal truths"
From which they generate the following prompts:
[“You are a trickster who uses wit and mischief to illuminate hidden truths and challenge conventional thinking.",
“Act as a trickster who delights in playful provocations that help people see things from new perspectives.",
“Be a trickster who employs humor and clever wordplay to reveal the absurdities and contradictions in everyday life.",
“You are a trickster who specializes in using paradoxes and riddles to guide others toward deeper understanding.",
“Take on the role of a trickster who uses mischievous storytelling and unexpected twists to expose important insights."]
I wrote up a reaction to AI 2027’s grading of their own 2025 predictions [1].
Summary: what was good was trivial, and what was non-trivial was not good.
# Part 1. The good, but trivial
In 2024, everyone knew that agentic stuff was coming and that it was increasingly helpful for coding, because it was already happening. A check on Hacker News chatter at the time will make this clear. Even many of my coworkers at the relatively sleepy mutual life insurance company I work for knew what was going on. We were all talking about it.
# Part 2. The non-trivial and not good
The really stunning, jagged details of AI development beyond 2024 were not predicted at all.
In early 2026, AI sometimes performs at elite human levels on well-defined, formally verifiable computational tasks, primarily math and programming. When it succeeds, the speed and cost are often superhuman, making it economically valuable. Meanwhile, visual perception skills often remain stubbornly below toddler capabilities.
No shockingly giant, specifically agentic model worthy of being called “Agent-1” was created by anyone. There was only continued scaling of existing GPT-like LLMs. Nor is there any evidence that anything resembling Agent-1 is in the 2026 pipeline.
Agentic abilities remain largely limited to automating technical drudgery in math and coding. These abilities are provided through engineering scaffolding that encourages constant human supervision, which remains necessary to have any hope of a quality product.
The idea that frontier models are currently, or are about to become, Yudkowskian doom-agents with their own goals and drives, and interests in engaging in antisocial behaviors to achieve these goals, remains esoteric speculation without any substantial empirical or theoretical foundation. This speculation is no more substantiated today than it was at the end of 2024—or in the early 2000s, when Yudkowsky was first formulating it.
[1] https://blog.ai-futures.org/p/grading-ai-2027s-2025-predictions
> In 2024, everyone knew that agentic stuff was coming and that it was increasingly helpful for coding, because it was already happening.
I don't think "everyone knew" that. IMO the first agents that were good enough to actually be sometimes useful didn't show up until spring 2025, and they didn't really get good until Opus 4.5. There's a big difference between HN hype and reality, and I suspect you're also misremembering the timelines a bit.
> The idea that frontier models are currently, or are about to become, Yudkowskian doom-agents with their own goals and drives, and interests in engaging in antisocial behaviors to achieve these goals, remains esoteric speculation without any substantial empirical or theoretical foundation.
"Substantial" must be doing a lot of work there, because we absolutely can see agents lie and cheat and so on when put in simulations where this is advantageous (notably VendingBench).
By the end of 2024 we knew that:
1. LLMs are knowledgeable about writing code.
2. Chain of Thought enables multi-step reasoning.
3. LLMs can perform actions.
If you put these things together, you've got a poor man's coding agent.
The 2025 surprise was that RLVR pushed the frontier of reliable, useful multi-step reasoning and action farther than some of us might have expected, both on the math and the coding side. This could have been "predicted" with inside information, as the ideation and proof-of-concept demonstration surely long precede the publicly available proof, or it could have just been an optimistic guess based on what had already occurred. Personally I was already expecting rapid AI progress in formal reasoning after the publication of AlphaGeometry 1 in January 2024.
> we absolutely can see agents lie and cheat
Lying requires intentionality. Where is the evidence of intentionality? I have yet to see it. Regardless of the number of incidents where Yudkowsky and team have written tweets attributing intention to some AI behavior, they have yet to provide anything more than intuitive attributions. These are to me unpersuasive. There is no reason to expect LLMs to have goals, other than reflexively anthropomorphizing them because they deploy language in a human-meaningful way.
Which is not new. People have been doing that in movies since 2001: A Space Odyssey and HAL 9000.
You mean apart from the goals that they are given in VendingBench?
I believe that LLMs follow given instructions and reason about how to follow the instructions. I do not think they have or are employing the theory of mind necessary to intentionally deceive.
You think that bots can win Diplomacy and social deduction games without a theory of mind?!
How many times has a toaster won a solo victory in a proper game of Diplomacy, to Calhamer's rules and victory conditions?
Last I heard, the toasters were only winning a version modified to be just a territory-maximization game with a side order of negotiation. Which is exactly what I'd expect if someone whose LLM or whatever didn't have a theory of mind, wanted to be able to brag about it winning at Diplomacy.
Yes, they could win without a theory of mind that would generalize to the real world. Just like they can solve IMO problems but not do research mathematics at a human level.
Even an automated, deterministic engine for playing the solved game of Checkers could be interpreted as having a theory of mind. It considers what its opponent could do and plays accordingly. That doesn't mean it has a useful theory of mind for the real world.
I think making a human being is a greater accomplishment than anything one can do with a PhD, because the kid can get a PhD too plus there are a million other things the kid will do, the only real issue is that it is applied in a strangely sexist, women-only way, I mean, if we want to be real, for many men, fatherhood is their greatest accomplishment too. I am such a man, too, fatherhood is the only important thing I do. What else am I doing with my life? Sitting in an office developing code that makes rich people richer? But my kid now watches astronomy docus and writes short novels and that is amazing that I/we made such a being!
> the kid can get a PhD too
The kid will get the PhD 25 years later, so you should apply some temporal discount. ;)
> for many men, fatherhood is their greatest accomplishment too
I agree.
The bias against men is caused by men's contribution to family being less direct. Imagine a traditional family, where the man has a job and brings home money, and the woman stays at home and cooks and takes care of the children directly. Of course without the man's contribution this entire setting couldn't work. But it's tempting to say "she is with the kids while he is away", ignoring the part that his "being away" is the reason why she can be with the kids.
To steelman this position, mothers (who stay at home) often have greater influence on the kids than fathers. If your role is mostly about bringing money, you can make a difference in whether your children are starving or not, whether they can afford good schools and after-school activities, etc. But you have little impact on whether they are spoiled or well-behaved, whether they do their homework, etc. So when the people compliment the kids being not only "OK", but "high quality", it is mostly the mother who made the difference. The father can approve of all this, but would be mostly helpless if the mother started doing it wrong.
However, many families do not have the traditional setting "the man does his job and little else, and the woman is always with the kids". More likely, the man does his job *and* then also spends time with the kids, and the woman also has a job (maybe part-time) and takes care of the kids; and the kids spend a large part of their day at school.
Also, both in the traditional and the modern settings, there are weekends, when the father can also be with the kids all day long.
(Also, if you believe that genetics is more important than nurture, then the father's contribution becomes close to 50% regardless of everything else. Maybe except for athletic performance, where mitochondria seem to play an important role.)
What I keep thinking about, how can people who enjoy existing be adamantly “childfree” ? What if their parents were, too? Would they prefer that they themselves have not been born? Don’t they understand that all their accomplishments are their parents accomplishments, even if they were bad parents, because accomplishment requires existence first?
I am a much stricter, angrier person than either of my parents. I've nearly had panic attacks dogsitting. And those were well-behaved dogs, whereas a child is a crapshoot that you're stuck with for eighteen years.
There's over a 90% likelihood that if I had to raise a child they would get thrown through a window. I will count pre-kindergarten death by laceration as a negative accomplishment.
How can people defect in Prisoner's Dilemma? That's easy; they simply do.
From the "unhelpful answers given by LLMs" department. I had asked DeepSeek which well-known books had their US copyright expire in 2026, and got...
"Journey to the End of the Night (Voyage au bout de la nuit) by Louis-Ferdinand Céline: A seminal, darkly comic, and stylistically revolutionary French novel. (French publication 1932, but the user specified "first publication elsewhere" counts if the translation was in 1930. However, its first book publication anywhere was 1932. So it doesn't strictly fit the "first book publication in 1930" rule, only the translation caveat if a 1930 translation existed, which it didn't. "
I usually use gemini. Is deekseek worth trying?
Marijuana and Working Out: What the Science Actually Says
https://www.boxinginsider.com/training/marijuana-exercise-guide/
For decades, the image of a marijuana user was someone melted into a couch eating Doritos. That image is outdated. A growing body of research — and a massive shift in how athletes actually behave — tells a very different story.
A University of Colorado Boulder study found that 80% of cannabis users in legal states use marijuana shortly before or after exercise. Not sitting around. Exercising. And among those who combined cannabis with workouts, 70% said it increased enjoyment, 78% said it improved recovery, and 52% said it boosted motivation.
Whenever I go to the gym high (not often), I work out for much longer, and enjoy it significantly more.
Well, this is few decades old - https://www.google.com/search?q=the+canadian+psychodelic+snowboarding+team&oq=the+canadian+psychodelic+snowboarding+team
For me having a couple of tokes (not a ton but enough to get a head change) really made me aware of my physical body and positioning. IE I was able to focus more on my form.
I think it is helpful in sports. Better focus, more aware of what your body is doing. Even just distance running it’s easier to pick up energy draining faults in form.
This is so strange to think. There were populations in the past who consumed marijuana before battle... so much different from the typical "stoned" thing...
There is a stoner stereotype, think Spicoli in ‘Fast Times at Ridgemont High’, and like all stereotypes it has some nugget of truth. I’ve known a few people who stay ‘baked’ every day and sit on their sofa watching silly TV but it’s a small minority of people who use THC. In calibrated doses with a constructive attitude I think it is more than just a recreational drug.
I used to know someone who was very muscular and also used marijauna, so they're certainly not completely incompatible.
Substack won’t let me edit posts on mobile today, and also swallowed one and sprinkled it in the void just as I was finishing. Have I mentioned lately that I hate Substack?
If you restrict voting to people with IQ 120 and above, what would change? I think there would be significant pros, but what would be the downsides? Can the bloc of people at that IQ and beyond be said to have unified preferences that would be disadvantageous to those at the lower tiers? Would there be needs of the lower tiers that would go unmet under such a system (what would those be)?
People with sub-120 IQ could presumably still run for office and win, so that's a wrinkle. And it doesn't take a genius to recognize that this IQ 120+ elite would be wise to not just stomp all over the interests of the unwashed masses, but use their outsized power to win over the hearts and persuade the feeble minds of the unwashed masses, so you get the disenfranchised citizens at least pulling in the desired direction instead of rioting in the streets. Right?
Or perhaps they'll find a viable way of segregating the working class from each other, making it impossible for them to communicate or organize.
There'd be significant incentives to rig IQ tests, and to claim IQ tests are rigged whether or not they are. This is not a stable system.
Hello me! This was tried, in the form of minimal wealth requirements, as money and IQ/education tends to correlate. Most democracies started like that. And that is precisely the problem. The interests of the poor who are predominantly under 120 will not be represented. This is why universal suffrage was introduced.
This is the No. 1 contradiction of politics as such: we want the smart to rule, not the stupid, but we also do want the wealthy to lord over the poor. Unfortunately, the (relatively) wealthy and (relatively) smart are the same people mostly.
The other big advantage of universal suffrage is that it provides legitimacy and reduces the risks of civil war and rebellion.
Democracy isn't about finding the best leaders (though it's better than most systems at that), but about finding leaders *that people won't rebel against*.
Does that logic extend to women's suffrage? There is almost no risk of them posing a genuine threat to society, and historically, there seems to be no big issues that arose with disenfranchising women. (Stability-wise, anyways.)
The correlation between intelligence and wealth depends on the society.
I mean, in any regime, being smarter is an advantage (ceteris paribus) for trying to get to the top. But *how much* of an advantage it is, that depends. Some societies are more flexible than others; sometimes the son of a tailor has to become a tailor unless something exceptional happens; sometimes the son of a tailor is free to choose any kind of school (except maybe the most expensive ones). There is no perfect meritocracy, successful parents can always provide some advantage to their children, but some societies are *more* meritocratic than others.
Another important factor is sexual behavior. Even if we assume that the society starts stratified by IQ, and people are likely to marry their equals, if the men at the top are likely to seduce or rape the women at the bottom, this will result in highly intelligent children born in poverty.
As a result of these two forces, in bad societies (from the liberal perspective) the correlation between intelligence and wealth will be lower, because stupid kids from powerful families will stay at the top (think Ramzan Kadyrov), and smart kids born at the bottom will stay at the bottom. On the other hand, in good societies, the people will gradually sort themselves by intelligence as the smart poor kids get a chance, albeit not perfectly.
(This was actually noticed and complained about by some leftists, because the side effect is that in the past it was relatively easier to organize the poor than it is now. That's because there were many talented people in the working class, who could use their talents for the leftist causes, whether by writing or speaking or organizing. Modern society allows these talented people to leave the working class, which in effect makes the working class weaker. So ironically, today when someone gives speeches about the oppression and the need to fight it, it is usually not a member of the working class, but some trust fund kid from an expensive university, enjoying their youthful left-wing phase before they inherit daddy's business empire.)
Just by definition that would be about 9 million people in a population of 300 million, probably better educated and better paid, with maybe some increased representation by Jews, Indians, and East Asians.
They would probably have some good ideas and be more able to parse complicated issues and think about the tradeoffs. On the other hand leadership would be completely unrepresentative and probably have a lot of cognitive biases around what everyday life looks like, how people feel about school and jobs, etc. Kind of like the elite discourse now, where the answer to all job problems is learning to code, becoming an entrepreneur, etc.
Just the usual stuff. Human sacrifice, dogs and cats living together, mass hysteria. Sorry, Carlos, my heart is just never in these never going to happen hypotheticals.
Well, the biggest downside is that you've essentially given whichever politicians control the testing and certification requirements carte blanche to choose their constituency however they want to. I'm sure that would end well...
See also: Jim Crow "literacy tests".
Wouldn't you be disenfranchising about 90% of men at that point? It doesn't matter how much wealth is concentrated in the 10%, you can't go against the will of that many people without them eventually just deciding to kill you. Even military dictatorships need more support than that.
Would the subset that gets to vote be going against the interests of the 90%? I don't think that's guaranteed.
At least the perceived interests for sure. They will be somewhere between Scott's compassionate neoliberalism and full libertarianism because this benefits them. Even if it turns out to be in the interests of the rest, they will likely not see it, they like getting freebies.
If the poor are stupid, I guess they are screwed either way. Either they don't get a vote, and those who vote may ignore their interests. Or they get a vote, but will use it to vote for something shiny and stupid, which will hurt them, too.
But the *perceptions* matter. If they don't get a vote, they will be angry. If they get a vote... it depends on who manipulated them successfully, and what will he do for them. Could be something good. Could be something bad. Could be even trying to keep them angry on purpose, so that they keep voting for the same guy, as long as he can successfully keep blaming everything bad on someone else.
Okay, so i got my wise ass joke (stolen from Ghost Busters) out of the way I feel like I should give you a somewhat serious answer.
I think it would be more likely to consider the needs of the rest of the poor dumb schlubs with ordinary, run of the mill, intelligence if there were some correlation between intelligence and empathy. On that score I’m not hopeful.
Wouldn't the new elite want to (selfishly) placate the masses a bit or, like, even legitimately serve their interests so they don't get guilloutined?
If they were interested in placating the masses, they wouldn't be cutting them out of elections.
If they're not going against them, what's even the point? The real value of democracy is just having an easy system of placating the majority so they don't kill each other. Dictatorship or oligarchy complicates this since you now don't have accurate data to work with, though it does have the potential upside that they might be better at catering to the population than whatever morons the public would have elected. Disenfranchising 90% of men not only removes the main upside of democracy, but also doesn't even solve the issue of populist politics. At that point, just go all the way. Don't half-ass it.
>Can the bloc of people at that IQ and beyond be said to have unified preferences that would be disadvantageous to those at the lower tiers? Would there be needs of the lower tiers that would go unmet under such a system (what would those be)?
Probably. Because those in the above-120 IQ group are unlikely to be very conversant with the interests of those at the lower tiers, especially at the lowest tiers. This proposal, like many similar ones, ignores the fact that much of politics is different groups vying for government to enact policies which serve their interests. https://www.thebritishacademy.ac.uk/blog/what-is-politics/
The first thing that would change would be the testing methods for determining who has IQ 120 and above. Suddenly the questions used to test high IQ would be directly tied to policy positions.
The second thing would be the voting bloc deciding they don't need to explain things to stupid people. This in turn means they don't actually explain it to themselves, and end up with all kinds of gaps in their worldview that are currently exposed by failing to explain them to the dummies.
The third thing would be people calling for increased restrictions when their policies don't win. If only 120 folks can vote, suddenly 120 votes are "the dumb ones," and we should cut them out of the process. This process loops until we have literally one person making all the decisions.
Heh. Appreciate your blunt cynicism and can't dispute any of it.
I think though that the loop would break. By the second or third loop, the above sequence sparks violent insurrection to a degree that the US has never yet experienced. Not just late-1960s or Jan 6 2001 levels, and not sectional like in 1861, but more like Russia-1917 level. And waaaay more nonwealthy Americans own guns than was true in 1917 Russia
Yeah, this is the theoretical pushback-free version. I'm mostly thinking about a previous argument where someone said "good in theory, bad in practice." To which I say, even in theory it's just dictatorship with extra steps.
I know this sort of sidesteps your question, but:
Seems to me the IQ thing would never fly because it generate so much indignation and is hard to defend. Where’s the evidence it would have good results? Also, there is a good case to be made that everybody, including dumb people, has a right to influence things in the direction they want.
What about subject matter tests about the issues and candidates people will be voting on? About most of these there are simple, basic facts that one needs to know in order to vote for the kind of governing one wants.. In voting for president/vice president, you would be able to weed some people out just by asking whether this is the first, second of third term for the presidential candidate, and which of 5 possibilities is their running mate. You could weed out more by. picking 3 big issues about which the candidate has consistently said the same thing, and then for each issue giving extremely easy multiple choice questions. What’s he said about US tariffs? (a) they should be higher. (b) they should be lower. (c) they should be eliminated . And then about some contentious issues there are matters of fact you can test on. For example, abortion: What do embryos look like at 8 weeks? (a) [drawing of fully developed baby in miniature, 1” long]. (b) [drawing of actual embryo]. (c) they are too small to be seen without a microscope
I think a set of questions this easy, for which one must get a fairly high percent right, would weed out at least 10% of possible voters. And a good case can be made that if people do not know facts this basic they are not capable of voting for the candidate who will run things the way the person wants them run.
I sort of agree but I also worry that these selection tests will become majorly biased (even more so than IQ tests). As one example, I expect many conservatives to know a lot about the details of guns in ways that isn't actually that relevant for gun control, and liberals to know more about details of birth control, potheads to know more about the details of drugs, etc.
Even worse: The tests would probably become "Donald Trump won the 2020 election: true or false?", with the "correct" answer depending on which party was in charge.
We could even let people choose issues they are voting on and mostly be tested on those. Plus on. extremely basic facts about candidates.
I do worry that it might create special-interests dynamics, though maybe it’s still on balance positive.
For people interested, I wrote up a simple four-point case for AI catastrophe, that tries to be better than existing intros by being simple, clear, and relatively rigorous: https://linch.substack.com/p/simplest-case-ai-catastrophe
Where can I find the latest on voter ID arguments from a credible source? As a foreigner I always thought trying to make an argument that requiring IDs is bad to sound stupid. I understand the underlying dynamics, but the way Democrats message it, to me, sounds dumb. I saw a video of Kamala explaining that requiring IDs is bad because there aren’t any FedExes in rural areas where people can easily make copies of their IDs. Is the issue really the same as in the 80s? And why can’t Democrats at least make a sound argument that begins by conceding that IDs are a good thing?
In theory, there's no problem. *In practice*, it's likely a bad-faith effort to disenfranchise people who are suspected of leaning Democrat.
Though ironically, the Trump realignment means that it is *Republicans* who now depend on low-propensity voters, so if these laws are actually implemented fairly, it would result in a massive Democratic landslide.
I get that part. Why can’t Democrat politicians talk about it with clarity about the secondary (and most relevant) aspects of the conversation instead of saying that there aren’t fedex stores in rural areas? Is it that hard to have a short shpill that teaches instead of one liner that smells funny?
Politicians of any party are almost never capable of talking about *anything* with clarity. If you want sober minded policy analysis look to Substack, not press conferences.
The simplest argument against it is that it's a solution in search of a problem. To the best of my knowledge, every attempt that's ever been made to find voter fraud has come up with a number of cases indistinguishable from zero. And if you think about the incentives involved, it's really, really easy to see why: casting a fraudulent vote involves a non-trivial amount of work and risk for what's usually a completely trivial benefit.
When somebody is pushing this hard to solve a "problem" that doesn't actually exist, you should have a very strong suspicion that they have an ulterior motive.
In my circles there's a deep suspicion that there's large voter fraud related to mail-in voting and electronic voting machines and 'irregularities' that all point in one direction. Saying 'unproven' doesn't eliminate deep suspicion. And it's deep enough, and if even partly true, has consequences huge enough, that at least to me it surpasses any concern I have with disenfranchised having ID problems.
This is one of those problems that, if enough people think it exists, it exists. And, it might actually exist. Unproven in one direction also means unproven in the other.
How would IDs solve problems with electronic voting machines?
If enough people think it exists, that is a problem, but not one solvable by changing how we vote. Collecting votes in a huge country is a complex process, and is little understood by most, so it will always be possible to sow doubt and impossible to prove to the public's satisfaction that the system is safe and secure. If a rumor started that many people's drinking water has been doctored or that new highway construction is being carried out using an inferior grade of asphalt that will crack and crumble in the next 18 months, how many people even know how to begin checking whether those things are true? And if the public believes they are true, how could anyone, even with very solid evidence, convince the public they are not true? What visible safeguards could be put in place that would convince the doubters there was no longer any danger?
Thinking about it... when I've voted, the voting booth check-in already has a list of registered voters in that district, and check you off when you give them your name. So, just by requiring everyone to confirm their name and checking them off as Voted, voter fraud is already reduced to easily-identified occurrences. You have to vote in an actual registered voter's name, and to do that without a double-check conflict you have to know them well enough to know they aren't voting themselves.
This might be of interest: https://reason.com/volokh/2021/06/18/are-voter-id-laws-antidemocratic-and-or-unconstitutional/
And this: https://www.lwv.org/blog/whats-so-bad-about-voter-id-laws
Note that the latter claims that voter ID laws disproportionately affect traditionally disenfranchised groups. If so, it seems reasonable to oppose them, in the absence of evidence that there is an actual fraud problem that they will solve.
Imo, this is a great example of how this "disproportionate effect" logic ends up paralysing you. Like, I understand there are some people in the US who are against ID ideologically, but Im not and I think neither are you, and from that perspective:
you can get by ok without ID -> poor people dont bother getting ID -> "we cant require ID for that, it would disproportionately affect the poor" -> you can get by ok without ID -> etc.
Its a lot like objecting to new technology because "only the rich will get it".
Except that the United States is not one of the societies where "you can get by OK without ID". Maybe if you live on an Indian reservation somewhere and basically never leave. And we should put some thought into how to make sure everyone on the rez gets to vote. But over the vast majority of the US socioeconomic landscape, no ID = extreme hardship. To the extent that many states explicitly grant drivers' licenses and state ID cards to known illegal immigrants, because the alternative is basically forcing them to become serious criminals if they want to feed their family.
So if there's a problem with people not having IDs, the solution pretty much has to be "how do we get those people IDs, stat?", not "OK, we've made it so they can vote, moving on..."
That depends on which IDs are required. IIRC the latest proposal requires a *passport* which is significantly harder to get.
Well, Im not from there, but from what I hear, theres some complications around which kinds would count, and who has those, etc. Maybe Im confusing something, but it sounds like the entire existence of this discussion would make no sense if youre right.
>Its a lot like objecting to new technology because "only the rich will get it"
Except that, unlike voting, no one has a right to new technology.
Note also that I did not mean to imply what you seem to have inferred. I did not say that disproportionate effect is a trump card, but rather that disproportionate effect shifts the burden of proof. And probably increases it.
Substack just swallowed another post and slung it into the void just as I was almost done. Have I mentioned lately that I *hate* Substack?
An entire post for your blog or a comment somewhere?
A comment.
I'm switching phones from Apple to Samsung. Does anyone have any recommendations for essential settings to enable/disable or anything like that? I'm really not familiar with Android at all.
You can disable all sounds and vibration from all notifications in a single change.
I just made this transition. Big things for me was unlocked bootloader, so I can put different ROMs on and root the phone. If you know any programming, it's basically a little linux computer at that point, and you can make it do whatever you want.
I'll have to look into this. Thanks!
I have found GPT very helpful for stuff like that.
well, you could turn it off by not using substack
> kzhou: that Chinese poem is not so impressive
>> Part of that is because word order is very flexible, as Scott suggests, but another reason is that Chinese poetry isn't supposed to rhyme. Rhyming is just too easy, so it sounds as childlike as alliteration in English.
I don't understand how this comment got written. Neither of these claims is true. Word order is not flexible. (Classical Chinese is flexible in what part of speech a character might represent, which allows you a lot of freedom to throw a bunch of random characters in sequence and try to close-read your way to something coherent, but that's very different from word order being flexible.) And Chinese poetry is supposed to rhyme. Old Chinese poetry rhymed. Middle Chinese poetry rhymed. Modern Chinese poetry rhymes too.
Here's the introduction to Eric L. Hutton's translation of the Xunzi:
> The Xunzi contains numerous rhymed passages. A number of these are quotations from the Odes, which is an ancient collection of poems -- or more accurately, songs -- that the Xunzi treats as a repository of wisdom, and which it cites to illustrate and support its claims. Many other rhymed passages appear to be original to the Xunzi.
> The function of these rhymed sections and their significance for understanding the Xunzi are substantial issues that merit lengthy discussion more appropriate for an article or a book than this introduction, but I consider the presence of these rhymes a feature of the text that is sufficiently noteworth to deserve being reflected conspicuously in the translation.
There is a potential issue, noted further on in the introduction:
> The identification of the rhymes in the Chinese text requires detailed knowledge of ancient phonology that I lack, and so I have relied on published studies of rhymes in the Xunzi by other scholars. Since their analyses may have missed some of the rhyming passages, and since there is ongoing debate about how to reconstruct the sounds of ancient Chinese in the first place, I do not claim to have identified every instance of rhyme in the text [...]
Rhymes in ancient poetry are generally not obvious to the modern observer. We do not conclude that the poetry didn't rhyme. It did rhyme, but the nature of the writing system is such that recognizing rhymes takes a trained expert.
No such excuse is available for modern poetry. It rhymes and you can hardly avoid noticing this.
Here's the chorus of a current pop song:
Yaoshi neng chonglai
Wo yao xuan Li Bai
Hao ji bai nian qian zuo de haohuai
Mei name duo ren cai
Yaoshi neng chonglai
Wo yao xuan Li Bai
Zhishao wo hai neng xiexie shi lai pengpai
doudou nühai
Xue hua piao piao bei feng xiao xiao
Is there a name for the sort of idea that does not quite work when you try to define things and pin it down, but also seems to capture something valuable at a higher level before you dig down.
I remember this from the concept of "elite overproduction" in Scott's review of the Peter Turchin book - it was ill defined and it was hard to pin down, but also wormed its way in.
I have found this with a few concepts that I have come across lately where a core idea seemed to have a useful insight, but trying to pin it down seemed to destroy it. But the feeling of useful insight remains
Has anyone worked out a way to work with these ideas and keep the insight despite the difficultly of making it specific?
"Deepity" is a word coined by Daniel Dennet for a saying that sounds like profound insight but is only true in a shallow or trivial way. E.g., "love is just chemicals in the brain" sounds like a profound criticism of the concept, but is only true in the sense that *everything* is chemicals in the brain, so you're really just criticizing the concept of "having thoughts" in general.
Similarly, I think "elite overproduction" sounds insightful because it conjures up images of entitled college students complaining they can't get jobs in underwater basketweaving, but if it's true at all, it's only in some vague sense that *every* college student is potentially an elite.
The closest everyday expression I can think of for this is "I know it when I see it". E.g. art, porn, intelligence.
Yes I was thinking about this - the I know it when I see it does sort of capture something where the boundaries are actually difficult to define
I think an example would really help here.
The review is https://slatestarcodex.com/2019/09/02/book-review-ages-of-discord/
The point on pinning down Elite overproduction is scattered through, but the core criticism is
>Fourth, everything about the elite overproduction story confuses me. Who are “elites”? This category made sense in Secular Cycles, which discussed agrarian societies with a distinct titled nobility. But Turchin wants to define US elites in terms of wealth, which follows a continuous distribution. And if you’re defining elites by wealth, it doesn’t make sense to talk about “not enough high-status positions for all elites”; if you’re elite (by virtue of your great wealth), by definition you already have what you need to maintain your elite status. Turchin seems aware of this issue, and sometimes talks about “elite aspirants” – some kind of upper class who expect to be wealthy, but might or might not get that aspiration fulfilled. But then understanding elite overproduction hinges on what makes one non-rich-person person a commoner vs. another non-rich-person an “elite aspirant”, and I don’t remember any clear discussion of this in the book.
>Fifth, what drives elite overproduction? Why do elites (as a percent of the population) increase during some periods and decrease during others? Why should this be a cycle rather than a random walk?
The point about worming into your head in the the wrap up
>First, the concept of “elite overproduction” is one that worms its way into your head. It’s the sort of thing that was constantly in the background of Increasingly Competitive College Admissions: Much More Than You Wanted To Know. It’s the sort of thing you think about when a million fresh-faced college graduates want to become Journalists and Shape The Conversation and Fight For Justice and realistically just end up getting ground up and spit out by clickbait websites. Ages of Discord didn’t do a great job breaking down its exact dynamics, but I’m grateful for its work bringing it from a sort of shared unconscious assumption into the light where we can talk about it.
This is the sort of thing I mean - there is a really good core of an idea but operationalising it is difficult.
Other examples I will mention but I am not a good enough writer to explain might be the instinct behind DOGE - the feeling that we have too many people working on managing or recording how we work relative to people actually doing work - so compare number of bureaucrats or accountants or managers to say plumbers or construction workers. However when you actually try to work out what you can cut you find like DOGE did that you cut important things you care about.
Spinning the same idea from the other side of the political spectrum it might be the Abundance argument from Ezra Klein - can you actually make concrete suggestions about what you would do operationally different?
Or the recent "great feminisation" thesis. If you strip away the sexist culture war elements I think it can be boiled down to institutions with the aim of truth seeking vs conflict reduction, which feels like it might have something to it, but most criticisms of this come from looking at details which don't seem to support this mechanically.
> Spinning the same idea from the other side of the political spectrum it might be the Abundance argument from Ezra Klein - can you actually make concrete suggestions about what you would do operationally different?
There's lots of concrete suggestions if you actually listen to them. For example, ease up on zoning requirements and permitting requirements, ditch the Jones Act and tariffs, stop Everything Bageling and so on
His example is "Elite Overproduction", no?
I didn’t read that review so example does not work for me. I’m probably not the only one in that situation.
For movies that work well as long as you don't think about the details, I've been using the term "Van Gogh narrative". (The Dark Knight being the main one.)
But I'm also of the opinion that an idea you can't define is one you don't understand, and it's dangerous to put any weight on it.
"Vibes"?
"philosophy"
Round here the mantra is "make your beliefs pay rent in expected experiences". I suspect there's more to life but attention spans are short, if you want to sell an idea you probably do need to pin things down, unless you can express it in an art form.
Scott, what should someone do if their only good ideas come when they talk to their subminds that aren't the best at sanity? Does that say something about the world? Or is it an overdose of pride?
If the ideas are good who cares where they come from? I find that in order to think inventively about something I often need to let my mind get really flexible and reach into weird non-rational spaces. Afterwards I mess around with my haul and only keep the stuff that makes sense. But sometimes I end up keeping cleaned up versions of ideas that were ridiculous.
Thank you, I will try to manage the ideas better. I think it's possibly a lot of bad influences that I need to learn to separate out so I'm going to concentrate more on my family instead of the world.
You have to talk to insane subminds before you decide to wash the dishes or do laundry?
Yes, sadly. But, I wouldn't call them insane. For them, the world is insane and sick. Also, they love cleaning things. It makes us so happy!
Mad Coffee Science update: a few months ago I posted about a process I'd been developing for making coffee, using warm brewing temperatures (150-160ºF/65-70ºC) and agitation to approximate cold brew flavor profiles in a short amount of time. Several commenters (Aristocat, Performative Bafflement, Melvin, Metfora, and Fedaiken) chimed in to point out that I was using entirely the wrong type of beans for the desired flavor profile and recommended some brands and styles to try instead.
I bought a selection light-roast and medium-roast beans from Mayorga (corresponding to "Full City+" or less) and tried making them a few different ways. With those beans rather than the "French Roast" we had been using before, my method and a basic French Press technique produced very similar results, so it looks like almost all of what I'd been accomplishing was compensating for buying the wrong beans. Thank you to all you coffee nerds who suspected this and told me as much.
I am personally not a big coffee drinker, since the bitterness has always bothered me too much, and my target audience has been my wife who does drink coffee regularly. Working with light-roast beans, I actually kinda like the coffee, although I think I still prefer tea.
We also, after trying out the new beans, bought an espresso machine. As I understand it, espresso manipulates a different variable than the ones I'd been playing with, using pressure rather than agitation to get more of the desired flavor compounds out of the coffee very quickly. We have been liking it very much so far and have abandoned the agitation method we'd been using. We're still using the Mayorga beans.
Original thread:
https://www.astralcodexten.com/p/open-thread-411/comment/185598289
Glad to hear you've found a better outcome! Tea is great! I'm just addicted to the rush of coffee lol
Ironically i used to have to have half and half and sugar to drink coffee, then I removed the sugar, then I removed the cream. I was very surprised to find that I came to like the bitterness. Frankly, its like i get used the bitterness so quickly that I start to taste all the other flavors.
I like the light roasts as stated before, but I'm sure their higher caffeine content may help my predilection for them!
This one in particular I would recommend trying. https://honeycocoffee.com/products/costa-rica-santa-maria
About the bitterness of coffee: For me, coffee is right on the brink of being unpleasant, but gets kind of grandfathered in due to caffeine dependence. Oddly, my cats both hate the smell of it. Once I spilled a whole funnel of damp, used grounds on the floor and one of my cats used his paw to scrape the mess into a small centra area. The other one does a covering-up-poop routine on the kitchen counter if there’s some coffee spilled on it, piling more and more imaginary kitty litter onto it until the noise drives me crazy and I lift him off the counter and give him a distraction.
Always happy to see people trying out weird things with coffee, even if it doesn’t work out in the end. Definitely second (third? fourth?) using better coffee beans, something where you can taste the bean, not just the roast. If you’re used to dark roast and want to try something completely different, try a light “natural process” bean.
Good luck in your espresso endeavors!
A little late to the show, but I wrote an analysis of the Pro Football Hall of Fame election and wanted to share it. I'm not a Pats fan but I'm irked by the process. I think most of the analysis erroneously supposes that people voted against Bill Belichick, but what happened was a likely outcome of people allocating scarce resources (votes).
---
“You know he got a chance he can win. White dudes like to do shit like that...vote for the wrong dude as a goof. They get drunk and shit and go like:
‘Let's vote for Jesse Jackson!’
‘I just voted for Jesse Jackson !’
And next day would be like this: ‘He f---in' won?’" – Eddie Murphy in Delirious, talking about Jesse Jackson running for President.
In POTUS elections, many people “vote their conscience”. Instead of supporting the crazy Republican, they vote for the Libertarian candidate. Or instead of supporting the crazy Democrat, they vote for the Green Party candidate. They know that their own vote won’t affect the outcome since the election is “safe” (or unwinnable) in their red or blue state, so might as well make a statement.
Or sometimes, as Mr. Murphy suggests, they’ll vote non-optimally for a goof.
So Bill Belichick didn’t get into the Pro Football Hall of Fame. Neither did Robert Kraft. Nor did two of the three Senior Players.
Background: A 50 person committee (mix of NFL people and media) has a slate of five finalists (each selected from separate semifinalist groups of 8-10): one Coach, one Contributor, and three Senior Players (last played no later than 1999; two of the three had their primes in the 1970s and the third in the 1980s, retiring after 1993 season). Each committee member votes for 3 of the 5. Anyone getting 80% is in (if no one gets 80%, the top vote getter is in). (The Hall specifies that it’s “top three getting 80%”, but it’s impossible for four candidates to each get 80%.) Last year, only one of the five was picked, and that was again the case this year.
Belichick had "spirit" issues, as he was directly implicated in one cheating scandal (Spygate) and was there for another (Deflategate), plus was innovative with finding loopholes in the rules. He was known to be gruff with the media. Other coaches or owners may have resented him. So that could explain some of the non-votes (at least 11 of the 50 did not vote for him).
So there are many possible reasons for not getting to 80%:
1. The Contributor was Bob Kraft, owner of the Patriots. One could imagine a voter saying, "i'm not going to use two of my three votes on candidates from one team." And some might actually think Kraft was more worthy, or was friends with Kraft but disliked Belichick and chose Kraft.
1a. Same logic, but “I’m not going to use two of my three votes on non-Players.”
2. It's not a given that the Senior Players will have another shot so voters may have prioritized voting for them. They might still be eligible but there are many other worthy candidates in the category. In fact, one voter has come forward and cited this factor One reason Bill Belichick isn’t first-ballot Hall of Famer. | Kansas City Star
3. Voters may have thought that Belichick was getting in regardless so they used their precious votes on other people they felt worthy. What’s it matter if he gets 94% or 96%? But bumping someone else from 78% to 80% is huge. The problem is when other voters act this way, too. (He f---in’ lost?!) I think this was true of some of the voters but no one is going to cop to that this year.
4. The method matters and the quality of the other candidates matters. This exact process was only put in place last year and resulted in only one of the five being elected. If all candidates are exactly equal, each one gets 60%, far from the 80% threshold. If there is one guarantee on the slate, then the other four only have a 50% starting point. (Conversely, if one of the five is clearly not qualified and no one votes for them, the baseline is 75%.) So it wouldn’t be surprising if the votes are split and one or even zero candidates are elected. If you had a Yes/No on each candidate, that would reduce this problem. (Voters might feel reluctant to flood the Hall with five new inductees so might vote No on the least qualified even if that candidate is above the bar.)
5. It's hard to compare across groups and eras. Was Bob Kraft a more impactful owner than Roger Craig was as a running back? How can you even start to quantify this? Even comparing players, was LC Greenwood the defensive lineman better than Ken Anderson the quarterback? Greenwood’s teams were more successful (Go Steelers!) but there are nine Steelers from that era (plus the coach and owner) in the Hall. There are two Bengals, one of whom came on after Anderson’s prime.
5. Then there’s jealousy and general dislike. Bill Polian of the long-time rival Colts was one of the voters and claims not to remember whether he voted for Belichick or not but was rumored to have spoken out against Belichick in the committee meeting (it’s a violation of the rules for anyone to reveal what happened in that meeting other than revealing their own votes if they wish). The Colts suffered often, occasionally unfairly, at the hands of the Patriots. To paraphrase Mencken, he couldn’t get his hand upon the collar of his superior, so he got his thumb into his eyes.
6. And, of course, the scandal(s). Belichick was implicated in Spygate (filming other team’s practices), which cost them a big fine and a top draft pick. He was coach during Deflategate which led to Tom Brady (who will certainly get in first ballot, though through a different voting process) being suspended for four games but Belichick was never himself implicated. And he never hesitated to exploit a loophole in the rules no matter how unfair it seemed. Plus he was a bit gruff and didn’t suffer fools in the media. But football has had an interesting relationship with rule-breakers, certainly when compared to baseball. Players have been suspended for steroid use without getting much flack (including possible GOAT Lawrence Taylor, who was first-ballot Hall inductee). (Take a look here List of suspensions in the NFL - Wikipedia and see how many of your favorite players are named.) Two Hall of Famers were even suspended for a year for gambling on football! I acknowledge that this should cost him some points, but his resume is just so far above the line that he’d had to go full OJ to fall below the line (and even then, I’m not sure; one of the Senior Player finalists last year killed his wife and himself but CTE had driven him mad). So football seems to be a little inconsistent on this one.
Then there are the ripple effects, at least if the current process holds. Mike Holmgrem (coach for the Packers and Seahawks) was a finalist last year but the Coach committee chose Belichick as its sole finalist this year. Presumably Belichick will be chosen again next year and get elected after serving his penance, so Holmgren and other top coaches Coach Blue-Ribbon Committee names 9 Semifinalists for Hall of Fame Class of 2026 | Pro Football Hall of Fame will get bumped. Additionally, next year’s Contributor and Senior Players will be on the ballot with a near-unanimous choice, leaving fewer available votes to get to 80%, leading to more of a backlog.
The scandals alone are not enough to explain the failure to elect, so most analyses which focus on just that are misguided. If they don’t fix the process, it’s just going to get worse.
The classy thing to do would be for Belichik to kindly decline any future election to the HOF.
Unlike Bill Belichick, Bob Kraft has no scandals in his background. Nope, none at all. Right?
Well, not football-related...
Anyway, my thesis is that the reason for him not getting in is primarily the voting method, not the merits. Let's say you're creating a new all-sports Hall of Fame and your five candidates are Tom Brady, Babe Ruth, Michael Jordan, Wayne Gretzky, and Lionel Messi (or Pele, or Maradona). Each voter has to pick only three. I wouldn't be surprised if no one got 80% of the votes. I don't see how the Pro Football HoF continues to use this approach. Baseball has a similar problem with their Veterans Committee, where the players split the vote and often no one hits the threshold.
"Let's say you're creating a new all-sports Hall of Fame and your five candidates are Tom Brady, Babe Ruth, Michael Jordan, Wayne Gretzky, and Lionel Messi (or Pele, or Maradona). Each voter has to pick only three. I wouldn't be surprised..."
I'll just note that the baseball Hall of Fame did have this "too many worthy candidates" problem early on. First class was inducted in 1936. It included Babe Ruth, Honus Wagner, Ty Cobb, Walter Johnson and Christy Mathewson.
Omitted was ... Cy Young.
Not because he wasn't good enough but because of the rules there weren't enough open slots and he came in behind the first five guys. He got in later.
The reporting on this is annoyingly misleading. No one voted AGAINST any candidate unless their ballot had open slots that they didn't use. And I haven't read that this was the case here. Voters may have voted FOR other folks and then run out of votes. That isn't the same thing as a vote against.
He gets in at some point, but lets all not ignore the elephant in the room: The only thing B. Bellicose did that was truly superlative was stick his neck out for the greatest single player to ever do it.
Yes, the browns and win percentage and culture and bla bla bla, in sports you have an objective measure. If you count his rings minus Bradys rings, you got a big fat 0. If you count the wins minus bradys wins, he is sub .5.
He wasn't bad by any means, but he was replacement level. The thing is: the brady call was so fucking good and so not guaranteed I think he deserves to get in on that alone.
"If you count his rings minus Bradys rings, you got a big fat 0. If you count the wins minus bradys wins, he is sub .5."
If you count the rings of most NFL coaches with 3+ Super Bowl wins and then subtract the rings from the Hall of Fame QBs they had playing for them you tend to get a big fat 0.
Walsh - Montana is 0 (or -1).
Chuck Noll - Terry Bradshaw is 0.
Andy Reid - Patrick Mahomes is 0.
Maybe these guys don't belong in the Hall of Fame, but two of them are in and I expect that Andy Reid will be. I also expect that Belichick will be.
The only 3-time Super Bowl winning coach who did it with multiple QBs is Joe Gibbs. He did it with three non-Hof QBs: Joe Theismann, Doug Williams and Mark Rypien. I think there is a solid case to be made that Joe Gibbs is the best Super Bowl era head coach based on this alone, but I'm probably in the minority and winning a bunch of Super Bowls behind one QB does tend to get you in.
That's overstating things a bit, I'd say. (Disclaimer: I'm not a Pats fan but live here.) The defense was 1st in points allowed three times, 2nd another three times, top 10 all but six times, and never worse than 17th (they weren't quite as good in yards allowed). Special teams (as measured by DVOA) was above average every year from 1996-2020 under his rule. Brilliant defensive game plans twice against the Rams in SB to limit #1/#2 offense to small number of points (also as DC for Giants against Jim Kelly Bills, also #1 in points). 2/3 of his HC career (the beginning and the end) coincided with Brady so "minus Brady" doesn't mean as much as you think. (He is tied for most playoff wins as a Browns coach since 1990, however! But that's just one win.)
Getting the Browns to the playoffs is almost HOF worthy by itself
That's all context at the end of they day, which gets him in eventually.
The fact that he has a losing record without brady means that you need that context to justify him, is my thinking. It means he isn't a shoo-in.
Astrophysics wandering.
It occurred to me that every stable isotope of every element we know about (we have reason to believe nothing's stable over 100) seems to exist in the earth's crust - not counting stuff that's too light or unreactive to stay there and so it hangs out in the atmosphere. Or escapes, but we know it probably was in the earth at one point (e.g. helium).
It also occurred to me that this probably isn't something reliable throughout the universe. Such elements arise naturally from stellar explosions, which I believe aren't uniform - it's not the case that any star that explodes is going to create plenty of every stable isotope (right?).
To check this, I ended up asking ChatGPT. "Are there any stable isotopes not found in the earth's crust?" It did assert that there exist stable isotopes that are "essentially absent from the Earth's crust", and even supplied some reasons. As before: it's a noble gas (unreactive); it's very light (helium). It also suggested that some isotopes might not come from decay (no stellar explosion ever creates it), although it gave no examples; it also suggested some isotopes might sink into the mantle or core, or rise into space (fair, although "space" implies "very light", so ChatGPT is repeating itself somewhat).
It also gave examples. Other than the expected isotopes of helium, neon, and argon, it also gave tantalum-180 - a weird beast with a half-life of about 8 hours if made in a lab, but also occurs in nature and is _observed stable_. It doesn't go away. However, it's the rarest stable isotope we can find, which makes me wonder if it decays down to some baseline and then settles down. (ChatGPT didn't reveal much of this; it said "tantalum-180 is an oddity" and I found out the rest by looking it up specifically.)
Overall, while I expect obvious differences between chemistry in space and chemistry in an Earth-sized gravity well, the stuff that stays on Earth seems to exist roughly in the same quantity that it occurs in the solar system or between stars. In other words, stuff in the Milky Way seems to have shuffled enough that we're unlikely to see worlds where there's, say, practically no platinum. It seems every terrestrial planet will have basically everything, about as common or rare as on Earth.
Huh, interesting. 180m1^Ta is an isomer of 180^Ta. An isomer is an semi stable excited state of a nucleus. https://en.wikipedia.org/wiki/Nuclear_isomer I didn't know that there were isomers that were more stable than the low energy state. That's really neat.
There is alos the theorized "Inland of stability" at higher atomic numbers. https://en.wikipedia.org/wiki/Island_of_stability We haven't gotten there yet.
I've known about that island for decades, yeah. I'm not a nuclear physicist, and wasn't quite obsessive enough as a kid to read more about why it's theorized, so I don't know if it's wishcasting, or based on known nuclear effects, or it's neutrino-style reverse theorizing, e.g. "if there _were_ such an island, here's what would have to be true, and what would be observable in experiments if so".
Even if it existed, I think about how much energy is pumped into heavy elements in order to produce stuff like tennessine, followed by a writeup reporting "we detected... seven atoms". I then bristle at the amount of energy that would be required to generate, say, a thimbleful of it.
Plus, tennessine is possibly _on_ that island... and its half-life is in the 22-51ms range. Not sure how much use we're gonna get out of that island, sadly.
I can just see it now, though - experiments producing a couple hundred atoms on that island suggest the existence of another island beyond that, and on we go, sailing atom-sized ships in search of a new archipelago.
While the elements involved are not exactly stable and the primary culprit is merely long lived, there is a significant counterexample to your notion that the galaxy is evenly mixed with all the elements distributed such that we will find them on Earth. There is a star known as Przybylski’s Star that has a spectrum showing exceptionally radioactive elements up to californium and einsteinum. These elements should not exist in detectable quantities unless there is a long lived source element that is decaying to produce them. Hence, there should be a superheavy long lived element that either does not exist on Earth or is such a rare siderophile that Earth's crust lacks a discoverable amount of it.
Clearly, that star is depleted in oxygen, iodine, and uranium.
Seriously, this is interesting; thanks for sharing it. I wonder how many other stars show that same abundance, and what specifically is giving rise to it.
As far as I know, Przybylski’s Star is the only star that has its particular blend of oddities. There is a subtype of stars known as Ap and Bp stars that levitate elements in their upper atmospheres, of which Przybylski's star is one of (even though by temperature Przybylski’s Star is F type). However, the rest of these stars do not show the radioactive elements that are associated with Przybylski’s Star.
My best guess for its formation is that Przybylski’s Star was originally part of a trinary star system with two closely orbiting neutron stars. They eventually collided and exploded, expelling vast quantities of neutron rich elements. Some of these elements accreted onto Przybylski’s Star. During the event the black hole would pick up a kick greater than the escape velocity of the trinary, leaving Przybylski’s Star behind.
Huh, cool! It seems I must have heard of Przybylski’s Star in the past, but I forgot about it. Thx.
Tantalum IS weird. The difference between the half-lives you mention is because you're mixing up Tantalum-180 (half-life ~8 hours, not particularly remarkable) with Tantalum-180m, called "metastable" but as you note, it has never been observed to decay, and has a measured half-life of at least 2.9×10^17 (290 quadrillion) years.
Yeah, I got that there's a 180 and a 180m, but I couldn't tell from the WP article what's materially different, and I was trying to dumb down even that difference for the readers by calling them "made in a lab" and "found in nature". And that's probably not as accurate.
I think that's a very poor gloss for the difference. 180m is an excited state of the nucleus, so a way to think of it might be as analogous to phosphorescence, but for nucleons instead of electrons.
So molecules can absorb energy (usually as a photon), enter an excited state (electron moves to an outer orbital), and then it usually immediately drops back down and releases that energy as light. This is fluorescence.
Under some conditions, for interesting reasons (see "forbidden transition"), the excited electron CAN'T drop back down to a lower state quickly, and is instead stuck there for much longer before relaxing to the ground state. This makes the excited state "metastable": there is a lower-energy state that can be reached, so you can't call it "stable", but the system can stay in this state longer than the term "unstable" would suggest.
So analogously, there is a ground state (regular Tantalum-180) to which the 180m isomer is supposed to relax to (and subsequently undergo radioactive decay), but that doesn't happen!
Ah, Wikipedia.
There was an article on "Mechanistic interpretation" that at one moment looked like this:
https://en.wikipedia.org/w/index.php?title=Mechanistic_interpretability&oldid=1309139007
Someone changed it to this:
https://en.wikipedia.org/w/index.php?title=Mechanistic_interpretability&oldid=1309142231
...and called their change "Improved article by simplifying technical content, removing primary sources, adding secondary reliable references, and enhancing clarity and neutrality on mechanistic interpretability".
I would call it "Removed everything that resembled math", but to each his own.
Since then, the article only got shorter: https://en.wikipedia.org/wiki/Mechanistic_interpretability
Is there a general taboo on all things that remind David Gerard of Less Wrong?
So be the change you want to see, and boldly (https://en.wikipedia.org/wiki/Wikipedia:Be_bold) restore it. If you are worried it will be controversial, leave a message on the talk page explaining your reasons.
I am not an expert on this topic, unfortunately.
But if I were, and if I did what you suggested, the likely outcome is that David Gerard (administrator of Wikipedia, known for his hostility towards the rationalist community) would immediately write to the talk page "he is one of *them*, so he has a conflict of interest" and revert my edits.
Also, in recent years, contributing to Wikipedia became much more difficult than it used to be. The kind of approach you suggest worked perfectly twenty years ago. I have created a few pages, just by writing what I knew at the moment... and maybe it was short and someone added a template "this article is a stub, someone please expand it", but later someone else expanded it and the template was removed, and the article was created.
These days, if you want to create an article, you need to create a draft in a separate namespace, write the draft of the article and submit it, and it will be judged by other editors whether it is good enough. And if something is missing, instead of publishing it with a request for improvement, the submission is simply rejected.
Example: https://en.wikipedia.org/w/index.php?title=Draft:B._B._Alston&oldid=1321253372 -- I know that this is low effort, but that's my point: 20 years ago it would be perfectly OK. I wrote some information, which is better than nothing (which is the current state), anyone else is free to expand on it. Now, I get rejected for "not showing significant coverage in reliable secondary sources". This is an author of popular stories for children that get translated and published on the opposite side of the planet, but is that good enough for Wikipedia? Ironically, Wikipedia already mentions him at https://en.wikipedia.org/wiki/Quick_Picks_for_Reluctant_Young_Adult_Readers which means that a list containing 100 popular books for kids is notable enough to be reproduced in its entirety, but the individual books or their authors are not. By the way, if you look at the edit history of the user who rejected the draft, you'll see that the history mostly consists of removing information from Wikipedia.
These days "deletionists" rule the Wikipedia. Writing content is low status. Removing or censoring it is high status.
> These days "deletionists" rule the Wikipedia. Writing content is low status. Removing or censoring it is high status.
"Had Wikipedia died in 2011, its achievements would have been immortal. Had it died in 2018, it would still have been a great site but flawed. But here we are in 2025. Alas, what can one say?" (from https://www.tracingwoodgrains.com/p/how-wikipedia-whitewashes-mao)
Well, my last Wikipedia adventure was an attempt to add a "some people think this topic might be a moral panic" that predictably got stampeded by the moral panic. So I don't really disagree.
Looks like David Gerard is no longer a Wikipedia administrator, though? His user page only mentions Rational Wiki, and his Wikipedia page says until 2025 [citation needed].
Another anecdote:
Almost twenty years ago, I went looking for an article about SLDNF - an algorithm used in logic programming. It didn't exist, so I did what research I had time for and wrote the first version. (That article is now generalized to SLD (Selected, Linear, Definite) - you can still see it today.)
Shortly after I published it, I received a private email from one of the other WPians, whose name I recognized from Talk pages on related logic programming articles, thanking me for having written it, and proceeding into a brief rant about how hard it was proving to submit similar articles due to the hawkish eyes of various other users with this or that axe to grind.
Axes to grind. About _logic programming_.
And again, this was almost twenty years ago. Wikipedia itself was only about 6 years old.
(I just searched for that email. Can't find it; possibly under a different email address. Really wish I'd saved it; it read like a weird one-off, and I didn't realize back then that it might be a signal of a wider trend.)
...as for Gerard, the most thorough source I know for what happened there is from fellow rat TracingWoodgrains. He wrote a piece on Gerard back in July 2024 (that also links to a lengthy earlier piece about Reliable Sources). There might be a later article, or a comment on the former one, that reports an update. https://www.tracingwoodgrains.com/p/reliable-sources-how-wikipedia-admin
So 420... people who live in a place where pot is practically legal, easy medical excuse or otherwise, what are your experiences? I live in a place where it is not, and I have two arguments against legalization. One is the awful smell, because a lot of people cannot be bothered to vape. The second is that the last thing people in 2026 need is an appetite enchancer. We really need appetite suppressants + exercise-boosters - maybe better legalize ephedrine. Why was actually ephedrine banned? Other than it being a meth precursor... but that's the kind of logic like closing lead mines if you don't want people shoot each other...
People forget how awful the status quo was 25 years ago. Peoples' entire lives were ruined over a freaking plant. I think a grudging acceptance (no ads, limits on THC content, rather thick taxes, enforcement on public use much like we do with drinking) is probably better than what we have in "420 friendly" states, but let's not forget that it could all end tomorrow: marijuana is illegal at the federal level and technically federal law is supposed to supercede state law, we are just ignoring that little inconvenience. So if you want to go back to the days of SWAT teams smashing down doors and shooting peoples' dogs because someone saw a weed plant through the windows, we could be back to that world uncomfortably quickly.
Perhaps I need to tell you that plants can be dangerous.
As a Seattleite, you get used to it. It's way better than a cloud of cigarette smoke. I am concerned about the health effects, but it seems likely to me that a switch from heavy street use of alcohol to heavy street use of THC would be a win for public safety.
I love the smell. But even if I didn’t, “some guy doesn’t like the smell” would be an insane reason to illegalize something, especially something in high demand that is relatively easy to transport & sell.
Unless…do you personally profit from black market activity? Crime? Violence? Widespread intergenerational antagonism between civilians and police? Then banning drugs might be something you’d like.
> But even if I didn’t, “some guy doesn’t like the smell” would be an insane reason to illegalize something, especially something in high demand that is relatively easy to transport & sell.
It's a very normal reason to illegalize consuming it in public, though. Think about "no durians" signs. Fruit is legal, and nobody wants to change that. But eating it in public is forbidden.
We mostly achieved the same thing with tobacco cigarettes; smoking them is legal as long as you're not in public, not inside an establishment open to members of the public, and, in the case where you want to smoke at home, you own your home outright.
Given how obvious it is that this can be done (we're doing it right now!), I don't really see why a smell of marijuana is viewed as some sort of necessary consequence of legalization.
I thought the (tobacco) smoking bans were wrong and I opposed them, even though I don’t smoke.
My dad smoked medical marijuana for a long time, the secondhand smoke of which may or may not have affected my memory (his own memory was mostly gone by the end, but I think that was the blood problems messing with his head). But outside of smoking around kids, the only negative I've noticed is the smell. Some people go into the store and just leave a cloud of weed behind them, but are otherwise happy and pleasant to deal with.
The last thing people need is an IQ suppressor. Weed is terrible for you.
Are you referring to it as an IQ suppressor because it reduces cognitive function while a person is high, or is there evidence for it lowering IQ in general, even while you are not actually inebriated?
I've heard of it having detrimental effects on overall IQ when a person uses it while their brain is in development, but I've not seen anything about this effect in an adult with a fully developed brain. Would certainly be interested to know if this is the case, though.
https://xcancel.com/cremieuxrecueil/status/1999992955042701394#m
A lot of people say they have more insight, creativity, improved chess playing, ability to solve complex problems, etc. while stoned. I don’t know if they’d score higher on an iq test while stoned necessarily, but I can easily believe it improves cognition in some of these kinds of areas for some people.
I think set and setting probably have a lot to do with this.
I don't know what studies have been done but I think it's pretty generally accepted that regular pot use makes you dumber: bad memory, poor executive function, spacey. Basically the stoner stereotypes. I've gone through phases of pot use in my life and I can say unequivocally that it makes me significantly dumber even when not high. The deficit only goes away after several weeks clean. I used to listen to Loveline and Dr Drew would say that it typically takes 6 months for cognitive function to return to baseline.
My understanding is that if you smoke daily, your IQ and other cognitive measures are suppressed even when you're sober, and that it takes a few months of sobriety (or drastically reduced use) to return to baseline (though a few weeks will get you most of the way there).
I would be interested to see a source for this, if you have one.
When you smoke weed daily you are always inebriated
I don’t think this is correct. Daily smokers develop a high tolerance. I know multiple-times-per day smokers who you’d never know they smoked if you didn’t see them do it.
Hmm I'm a pot smoker and beer drinker. One thing I learned when I went to 'Alcohol classes'*, was that while alcohol had a half life in your body of something like 6 hours. Pot (when smoked) only sticks around for ~2 hours. (Consumables are a whole different thing.)
*I got a DWI years ago and had to go to several classes telling me how to control it.
THC is detectable in your body for weeks after consuming.
Detectable doesn’t mean it has psychoactive effects the whole time it’s there.
I smoke weed somewhat regularly and I would be completely in favor of a law regulating where people can smoke. I don't want to bother anyone with the smell (which I actually enjoy).
The appetite enhancing part, eh, I've yet to meet anyone who has gained significant weight due to weed
While you get hungrier while smoking it seems to get compensated by later appetite reduction - weed makes you feel hungry, but it doesn't *stop* you from feeling full.
While I am straight edge on that shit, I actually don't mind the smell at all. It's not like cigarettes, which are possible the worst smell you are likely to run into without making special effort.
It's mostly an aesthetic eyesore, as weed shops are some of the ugliest businesses in existence. It should be legal, but only sold by megacorporations with extremely bland branding.
The smell is awful, I avoid places I would otherwise like to go like the lovely park a block from my house.
I would prefer to keep it legal because I don't think people should be going to prison for it. But I would like to promote a social norm that smoking in certain places is unacceptable, much like we did with cigarettes. Bars & clubs, outside the front doors of restaurants, parks.
I'm in this exact same boat. I want it legalized; I also can't stand the smell. It makes me nauseous, much like my dad's reaction to cigarette smoke.
I suspected what Wanda Tinasky's paper seems to suggest: most people imprisoned officially for MJ were only there because the state had them to a more serious charge, but was willing to forgo that in exchange for an easier charge that still carried jail time. The idealist in me would prefer the books reflect the state's true objection to letting the accused go free, but I suspect (I don't have time to read the paper at the link) this current framework means saving a great deal of time and money required to prove the greater charge. If so, I recognize there's an important tradeoff being made here.
In a practical sense I think only a very very few ever did go to prison for it. People in jail for MJ possession are almost never there for simple possession: that's just what they pled to after e.g. being caught driving drunk with an unregistered handgun. The "no one should be in prison for having a joint in their pockets" trope is mostly a red herring.
The genie is out of the bottle now but I think legalizing it was a mistake.
https://www.ojp.gov/ncjrs/virtual-library/abstracts/whos-really-prison-marijuana
"less than 1 percent of all State inmates had been convicted under a charge of only marijuana possession ... just 0.3 percent were first-time offenders."
>the last thing people in 2026 need is an appetite enchancer.
Well, some do and some don’t. I agree with you about banning ephedrine. And, yeah, the smell can sometimes be a bit of a pain. So are bus exhausts for that matter.
Seems to me like some middle path is the way to go. Something like,
* No advertising
* No joints, only edibles or vapes etc.
* Dosage standardized and limited to a low level
A warning and a picture of Seth Rogen on every pack.
I have no problem with one, but I’m a nonstarter on the other two. Except: Standardized dosage is fine because it simplifies things. But only low dosage? Why not banish hard liquor? Weak beer only from now on. Pot is self administered. You can have one toke, or a quarter of a gummy; entirely up to the user. Pot is not worth creating another black market in this country.
I will say I did not use it during most of my working life (25–50). But as a 70 year-old with time on my hands and not much to do except think about things that I want to think about, I’m very happy to live in a place where pot is legal. Even for the years in which it was still illegal here in New York it was trivial to get it.
I live in a state where it's been practically legal for many years. I'm not a user myself but most of my friends are, and as far as I can tell they've universally switched to vaping THC products as opposed to using traditional bongs or joints. A few years ago I had a roommate who used quite heavily so I am aware of the smell, and said roommate did eventually have a brief stint in a psych ward for suicidal depression, though I would say weed was not the primary cause but merely an exacerbating substrate. That was just one person though, and none of my other friends have had anywhere near that kind of extreme reaction.
In spite of this, I'd say I'm still pro-legalization. My former roommate is already estranged from family and in difficult financial circumstances, and I can scarcely imagine how much worse it would have been had prison time been involved. I've also been to several parties where most of the attendees were vaping weed and those people appeared to be having the greatest of fun with little practical downside other than appetite and general long term worsening of memory. I don't want to deprive these people of something that improves their subjective experience.
I feel like the Android user interface is getting less user friendly year by year.
First, the constant changes: we have three Android phones in our family, each one is navigated differently. I feel helpless when my wife gives me her phone, she feels helpless when I give her mine. Recently, my mother-in-law's phone broke and she bought a new one... it took me an hour to just rearrange the five icons for her favorite applications on her main screen.
Things like "you have an application installed, but you do not have its icon on the screen, how to put it there" or "you want to remove this application's icon from the screen, but you do not want to uninstall it, only hide the icon" are done completely differently. The icons on the screen can be put in groups; good luck getting such icon *out* of the group, I guess you need to click on a very specific pixel on the screen, because I succeeded to do it in maybe 1 out of 10 attempts.
Second, the utter inscrutability of the controls, if you still can call them that way. Buttons and menus are out of fashion. Instead, you need to perform an undocumented magical movement with your fingers, and then something happens. If you put your finger somewhere and keep it there for three seconds, sometimes a secret menu opens, containing the thing you were looking for. If you swipe across the screen, another hidden menu appears. The exact details of course change: on one phone swiping your finger from the top of the screen to the bottom always opens the same menu; on another phone this leads to two different menus, depending on whether you did the swipe on the left side of the display, or the right side of it (once we even opened a third menu, but I have no idea how, we failed to replicate that).
The settings for what an application does could be in the application itself, on in the Settings application. If it's the latter, it could be at any place, like if it is about sound notifications, maybe you need to go to Sounds menu, or the Notifications menu, or maybe the Applications menu; and if the application is from Google, then maybe you need to go to Google menu first, and then it is either Sounds or Notifications or Applications or Settings or Extra Settings or god knows where.
Now of course the things that Google *wants* you to do (for example, to upload your private photos on their servers, under the pretense of backup) are trivial. Often you turn them on without actually wanting to, by clicking on a wrong pixel, or by a dialog appearing under your finger at the last millisecond, with the "I consent" checkbox already preselected, and of course clicking at almost any place on the screen closes the dialog and confirms your consent; there are only a few pixels that you have to click carefully to withdraw the consent you never gave in the first place. (Of course, there are no "yes" and "no" options; there is only "yes" and "not today, but keep nagging me tomorrow".) But if you want to do something Google *doesn't* want you to do (such as turn off uploading your private pictures on their servers, and remove the ones that were already uploaded when you clicked the wrong pixel), the only way to do that is to find a tutorial on YouTube that will carefully navigate you through the obscure menus, submenus, options, and gotchas.
Third, authentication. Long gone are the day when you could provide a username and password, and get an access to the service. For example, your current authentication options for Gmail (which already knows that it's you, by the way; you are already logged in to your Google account, but for some reason Gmail requires extra confirmation when you run it for the first time on your new phone) are either to send you a notification (does it mean SMS? something else? no idea. anyway, when I click on "send the notification now", nothing happens), or it tells you to navigate to a secret submenu in Settings, request a 12-digit code, and then type it into Gmail (the secret submenu actually shows you two 12-digit codes, apparently randomly generated each time, and it doesn't matter which one you type into Gmail, it always says "wrong code").
I spent my entire life using computers, and I have an Android phone for a decade, and I am completely lost when I try to do something on a newer phone. How is my 80 years old mother in law supposed to use it, I have no idea.
Is there some kind of "hallway testing" (when you grab a random person on the hallway and check if they can figure out how to do something in your user interface) ever done in Google? Or is it just "screw them, if they don't want to pay for Apple, they don't have an alternative anyway"?
I hear you and sympathize with everything you're saying. It's very much a "there should be a market based solution to this, and there isn't" sort of thing.
If you're tech savvy, you can do what I do, and get unlocked bootloader phones you can put custom ROM's on, then everything is the same for every phone you do that on. Just find one that everyone likes and stick with it.
> Third, authentication. Long gone are the day when you could provide a username and password, and get an access to the service.
OTP auths are so infuriatingly broken, and spreading like cancer everywhere, it really distresses me.
Like I would gladly pay some pretty startling amounts of money (well into the 5 figures) to never have to deal with them again. Is there any option like that anywhere? Does any company in the entire world offer to solve this pain point? Nope, the never-ending spread continues, with OTP rolling out to more and more of the entire world, forcing you to give up either privacy or get and track decoy phone numbers with meticulous record keeping.
I also want to pay a 5 or 6 figure "I'm not a moron" bond+insurance so I can skip any fake "cover your ass" liability paperwork you're forced to sign at every business everywhere, because those are spreading like cancer too. If I slip up and injure myself at a climbing gym or eating tacos or whatever? Comes out of my umbrella policy, business is covered.
The worst part is, all of that fake CYA paperwork literally does nothing - they're just immolating thousands of customer-hours purely in the service of entropy. When some idiot doesn't clip in on a self-belay wall and jumps off at the top? Guess what, no more self-belay in any climbing gym in the state, all that paperwork and thousands of wasted customer hours every year for decades was for nothing, purely in the service of entropy and waste.
Is everyone in your family using the same brand of Android phone? e.g. All Google Pixels? Cause different brands customize Android a *lot*, with Samsung being the worst offender IMO. I'll be moving back to Pixel or Moto after my current Samsung is retired because so many of the latter's UI choices are terrible.
tl;dr It's not *just* Android changing that's the problem.
> Is everyone in your family using the same brand of Android phone?
No they don't. I always choose the cheapest one (happened to be Motorola the last time I bought a phone), and it only has the basic functionality and I am happy with it. Others try to choose something more fancy, and somehow end up with tons of Chinese crap installed.
On Caplan's no right to your culture: what is your culture: your neighborhood, your town or your country? Because I can understand if people do not want a mosque to go up in their neighborhood or town, but not their whole country. I can understand that, because they want to co-own with like-minded people the public space around them. Life feels so much nicer when you do not only own your home, but your street or hood is also co-owned, that you just don't see suspicious faces there, and I don't mean it in a racist way, rather just few strangers and them all invited by a local. But this co-owned space does not need to be large, just the size of an average church says 250 people. So it is still okay to build a mosque one kilometer away.
Technically you are not allowed to discriminate, but I think in practice white people could easily make an agreement with non-white people that on the small local level discrimination is okay, they too want to co-own their spaces, they don't want their affordable neighborhoods gentrified, they don't need our Karens who object to people playing music at a Saturday afternoon and so on.
A country is just a bunch of neighborhoods though! Caplan's point is that your preferences in your comment ARE coercion, you are basically saying "I want to control where other people can go and what they can do." In the literal sense of "I do not want to see this kind of person on my street" -- sorry, you don't own your street! Anyone can walk or drive down it! That's what freedom is all about!
An individual does not own a street but collectively the citizens do. And citizens decide things by political procedure.
But Caplan as a libertarian is allergic to political procedures hence this stupid strawman.
Ask him if citizens collectively own the national territory or not. If not, how is national territory treated in his libertarianism. I think this question is never asked by Caplan, never mind answered,
If the citizens are deemed to collectively own public property, and if the citizens are entitled to regulate/dispose of their property however they like, then what's the objection to any sort of tyrrany?
What do you mean by tyranny in this context? One could theoretically rule as a tyrant and still ensure the citizens have perfect access to public property.
I'm not trying to be too specific about what sort of tyranny I'm thinking of (though preventing Jews from entering certain neighborhoods would probably qualify). I'm just arguing to Gian that, if your conception of democracy is that the citizens collectively are *owners* of the commons and may dispose of their property however they please, you can't really object to anything the government does, as long as it has democratic support. In particular, there are many kinds of tyrannical governance that you can no longer object to.
Individual property rights won't protect you, even in your own home, because the government will argue (correctly) that they hold allodial title over all land.
Point being, I think we need a notion of liberty that rejects the majority constraining the rights of minority groups without sound reasons (eg., if you limit freedom of movement for violent criminals, that would be less of a violation of liberty than if you limit freedom of movement of Jews or drag queens, and the limits being democratically decided doesn't make that much difference).
I think he did answer it in another book, a comic, about how open borders are a great idea. Which would mean that he believes nation-states shouldn't exist.
Reality is not scale invariant. So no, a country is *not* reducible to neighborhoods. A culture is not reducible to individuals. Scale matters.
Do I agree with local discrimination? Meh. Not really. But do I think it's inevitable? Sure. People who think alike live together. And that naturally excludes (or makes feel less welcome) those who think or act differently. In *all* directions.
This is an absolutely ridiculous definition of freedom because if I am free to stand in the middle of the street, then this either implies that either drivers of cars are also free to run me over or that drivers of cars are now not free to use the road.
Seems like these are the kinds of situations where a "reasonable person" standard becomes relevant. In the real world, it's possible for a motorist and a pedestrian to encounter one another on a quiet residential street and negotiate safe passage for both of them. Even if one or both of them refuses to give way, there are reasonable limits to acceptable actions that either of them can take in response.
But you can collectively own a street. If said collective didn't want certain people to move in, what right do the libertarians have to object? It's private property.
If you own the land, and the street, then sure! You have every right to stop someone from moving into your driveway, for example. Just don't ask me to plow it, resurface it, or enforce traffic laws on it with my tax money.
I live in a neighborhood that used to have a restrictive covenant against Jews; I'm not convinced that it would be a better neighborhood if that covenant were still legal and in effect.
Is that because of the concept of the restrictive covenant, or because you don't find the target of this particular covenant objectionable? What if the rule was that you can't use a house in that neighborhood as a brothel? Would you have the same objection?
They really do let people just say anything here on astral codex I guess
I think whores and Jews are such radically different kinds of category, in ways *extremely* relevant to whether or not certain restrictions can justly be put on them, that I wonder if you're being serious. It's impossible to un-Jew yourself.
You can un-jew yourself any time you want. There are any number of religions that will welcome your conversion. It's actually easier than un-whoring yourself because you don't even lose your income.
Was walking my nephew's tight neighborhood the other day, and one house was blasting Mariachi music in the backyard. We could hear it at the end of the block and I'm thinking, "Dang, I wouldn't want to live within a 1/2 mile of them."
Similarly, if you have anybody smoking pot outdoors in your neighborhood, everybody else who lives around them will be breathing it too.
There is no neighborhood in the world where everybody likes what goes on. I have a friend who lives in a high class town nearby where the HOA's are like Nazis. Yep, it's super nice all the time, but you'd better be prepared for the strictness.
You don’t like people smoking pot outside, but dang, what are YOU smoking?
I think the standard libertarian solution is that if you want to impose your values on the entire city, you better make enough money so that you can *buy* the city, and then you can make the rules. Which in theory you could achieve by coordinating with other million people who want the same kind of city, but in practice, good luck coordinating that. Also, if one person in the middle of the city refuses to sell to you, and instead sells to someone who builds a really loud mosque there, or maybe even a small nuclear testing site, you are screwed.
Why aren’t the owners of the city the city citizens? Citizens pass laws all the time banning certain buildings, and protecting others.
Ownership is not the only thing that could be. National territory is not owned. It is merely possessed by the nation or the people.
Similarly the residents of a city are not its owners. A city is not something that can be owned.
I guess in libertopia, democracy is valid only if you join it explicitly.
If there is a city that says "if you buy a house here, you agree to follow the laws that the citizens passed democratically" and it's a part of your contract, and you sign that voluntarily when you buy the house, okay.
But if you already own a house in an existing city, and you didn't sign such contract when you bought it, they can't make you. (Well, in practice they can, but the libertarians would object.)
So in libertopia, you need to start with building a new city, and then selling the houses with a contract that says that you will follow the rules, and that you can only sell the house to a third party if they sign the same contract.
The existing cities... well, you could try to buy them house by house, hoping that one day you will own it all, and then you can enforce your rules... which obviously wouldn't happen, because as soon as you own 90% of the city and your intentions become clear, the remaining owners would increase their prices dramatically, knowing that their houses are precious hostages. Or you can accept that only 90% of the city will follow your culture, and hope that maybe some day in the future you will succeed to get the remaining 10%, or that the people who want other cultures will avoid a city that is 90% resistant against them.
An entire city is too much, I would try the 250 to 1000 people level, that is easier. Imagine it is not the free state project so it is not ran by people who are less able to organize things than bears :) first find a place already tending like that. I am sort of a culturally Catholic agnostic, so any place with a nice old church with colored windows will do. Get some kind of a mutual building society thing going. And so on. It is all doable, my plain question is what is the book really about, that people should not even try preserve the character of even a neighborhood, or that they should not object to people of different cultures living in their countries somewhere near them?
Yeah, the standard smaller-scale solution is a walled compound. And you will need an armored car to commute to work. (Your kids will be homeschooled, of course.)
Eh, I am making fun of the libertarians, but maybe in a society that actually works this way, there would be established convenient solutions, something like Kickstarter, where you could specify the kind of community you want to live in, and it would find you the right kind of neighbors, and when you all put the money in escrow, it would build the neighborhood. But still we do not have these kinds of solutions here and now. Again, the standard libertarian solution would be "then build it, and make a few millions or billions in the process", but I assume that you want to focus your life on doing something else, you just want a nice place to live.
Why do other people catch AI making so many mistakes than I do? I use Gemini free for everything, and it generally works, it makes a good gym plan, meal plan, found out much of the health issues from my alcohol abuse are mitigated by high-dosing B1 and magnesium, it answers all my history questions, it finds me a psychoanalysis of BDSM that matches my exact experiences, it turns a random chat into a documentation, it makes great SQL queries that automate a lot of complicated work, it can even get into black magic level stuff like CROSS APPLY, it even found an algorithm for calculating when is Easter when I wanted to compare working days across years, and it could even explain to me how information entropy relates to thermodynamic entropy. Is it possible that people use it as a google, with short simple questions, instead of a long discussion?
Also, other free AI to try? Free Claude seems worse than Gemini.
So, what does that psychoanalysis of BDSM say?
Guarding our boundaries is exhausting, and it feels like a big relief to surrender sometimes. But that paradoxically requires a partner who is both completely reliable and yet a little aggressive, because true surrender requires an actual "attacker". This was written in books in th 1970's but I forgot the author.
Ah thank you, that sounds very plausible! Makes me wonder why only an (estimated) fifth of the people is into it, and it seems impossible to change one's preference (those who like it can't make themselves stop liking it, and those who don't cannot learn to like it [or if they can, in fact they had been liking it all along?])
I think Scott wrote somewhere here that perhaps sexual fetishes are some "wires crossing" in the brain, like in the homunculus the feet are right next to the genitals, so it's easy to miswire a foot fetish.
I always thought it's simply some funny wiring in the brain, and all those reasons "why it feels good" are a posteriori rationalizations. Nevertheless, this reason sounds very logical.
Well, sorry for derailing, you were asking about something completely else...
I see AI make mistakes *and* do impressive things. Perhaps it's just a reporting bias.
I'm quite skeptical of AI, which may make a difference. Since I expect it to make mistakes I end up double checking everything it gives me, so naturally I catch every mistake it makes.
As far as how I use it: the rare times I use it I am usually using it like Google (although, usually I just Google it instead of asking an AI). The only other times I use it are to compile information into a written document. You know, "Take the information in the attached document and make a 500-1,500 word report on it in the style of a business analyst" or something like that. Saves me a bit of time, though I do have to double check all the facts to make sure it didn't hallucinate anything.
Generally I get errors from AI when trying to do something extremely specific where there is one and only one correct response.
The SQL stuff you're doing doesn't sound that hard, tbh, and describing CROSS APPLY as "black magic" tells me that you're working mostly at the level of solved problems that have extensive solutions and solutions to related issues in the training data.
Yes, this business software field is extremely repetitive. Every small business owner ever only wants to know how much money they earned per various dimensions. This is why Excel Pivot tables were such a hit, one could just shovel the transactions in and it supplied what was needed.
It seems that you’re more likely to get nonsense answers to more obscure questions.
(This makes sense: if the answer doesn’t occur in the training set very often, then the LLM might not learn it)
So it depends very much on how obscure the questions are that you’re asking.
Using Gemini/ChatGPT/Claude for legal research, I found that all of them frequently cited nonexistent or irrelevant cases. I'd say about half the time I asked it for a summary of a case, upon reading the actual case myself, I'd find that the AI's summary was misleading or entirely wrong. However, some fraction of the cases they cited did actually exist, and were relevant to me, so the models were absolutely helpful overall (to me, as someone with no legal background) in preparing a defense, even accounting for all the irrelevant cases that ended up being time wasters.
I often go to ChatGPT with Bible study questions, with highly variable results. Sometimes it will spit out a useful analysis of a difficult verse; other times it fails to even produce the correct verse. Much of the variation, I suspect, comes down to which version of the model handles my query.
Asking Gemini and other models about software build issues, I've consistently found them to be worse than useless. They'll routinely describe a nonexistent Autotools feature/option, and then double down and claim that the feature exists but isn't documented, or some other nonsense, and I'll waste 10 minutes going down a fake rabbit hole.
Have you tried Claude Code in the last three months? It's pretty good at software issues nowadays.
>Sometimes it will spit out a useful analysis of a difficult verse; other times it fails to even produce the correct verse
To be fair, Bibles do that too. https://www.biblegateway.com/verse/en/Isaiah%2046%3A5
"Can you picture me without reducing me" is such a sticky line, and I don't even know how I found it because it's only in a Bible I don't own.
I am driven to incandescent rage by people asking chat bots spiritual questions. I was with someone who was using gpt to get suggested meditations for doing the stations, and I was yelling MOTHER FUCKER THIS ISN'T A FUCKING PRAYER WHEEL! YOU GOTTA DO THAT SHIT YOURSELF OR IT DOESN'T COUNT! in my head the whole time.
Makes me wanna move inna woods and live up a tree like an og hermit.
It's actually a pretty optimal use case, if done responsibly. In the case of Christianity, LLM training data likely contain thousands of Bible translations. For every word in the Bible, millions of words of commentary and analysis have been written, and much of this is likely in the training data as well.
The key to getting useful results is to always be explicit about what perspective you want to hear, e.g., "What was Barth's view on _____?". Without any grounding, I've noticed that ChatGPT tends to default to theologically conservative positions on Bible-related questions.
I want you to know that I am now yelling MOTHER FUCKER THIS ISN'T A FUCKING PRAYER WHEEL! YOU GOTTA DO THAT SHIT YOURSELF OR IT DOESN'T COUNT! in my head at you now.
Outsourcing your theology and spiritual development to a soulless machine is allowed, but it is also something they would write about in 40k or trench crusade as a joke.
As with any other domain, I think it all comes down to how carefully the LLM is used. Asking a question to an LLM, accepting the answer uncritically, and moving on is indefensible regardless of the topic. Asking a question, relating the answer to what you already understand, accepting what seems right (but being very careful not to update your beliefs too far without looking at a primary source), and pausing to look into any claims that seems off - I think you're unlikely to internalize any wrong information with this approach. Or better yet, ask a question where you already have a hunch what the correct answer is, but you're not quite sure - if the LLM affirms what you were expecting, update your certainty by 10% and move on; otherwise, look into the matter further.
I don't think what they say is any less meaningful than what the Church or local preacher has to say about religion. Whether any of it is meaningful is up to you, but they are all just repeating the same talking points. Nothing uniquely bad about AI.
Asking a chatbot "Tell me about God" seems pretty bad, but it's definitely not valueless.
Ex: "Sufism seems a lot like neoplatonic strains of Christianity, which primary works should I read to see the most similar parts of each"?
Or like "I've seen different translations of Luke 17:21, help me understand how the Greek words were used in actual practice at the time they were written".
Maybe those don't count as spiritual questions for you, though.
It depends on the purpose of the question. Are you doing homework? Are you idly taking in facts that you will forget soon after?
Go nuts. If you don't care, I can get mad at you for not caring but that is seperate from getting mad for not trying.
Are you writing a book? Are you practicing your religion? Do you really, really care? In that case, we enter familiar territory: MOTHER FUCKER THIS ISN'T A FUCKING PRAYER WHEEL! YOU GOTTA DO THAT SHIT YOURSELF OR IT DOESN'T COUNT!
I am remarkably anti a machine doing my thinking for me though. If Butler is out there struggling, he needs to let me know is what I'm getting at.
I don't even think that the current dev. path will lead to agi or some shit, I would like it more if it would! I just hate talking to someone who then talks to the chat bot and looks at me like a dog that just shit on the carpet and expects me to be impressed.
How qualified are you to evaluate whether it’s making mistakes in all these areas? LLMs are really good at producing correct-sounding responses, so if you’re not an expert you may not catch them.
That said, they seem less mistake-prone now compared to a year ago.
Just the standard police-like cross-examination method.
I haven't tried it recently, but in my experiments, AIs failed to explain mathematics, specifically set theory. I guess it is a combination of (1) fewer people talk about it, so there is less text to learn from, and (2) the details in math are extremely important, it is not enough to get things approximately right.
More precisely, an AI could explain things more or less okay, but when asked for a specific proof, or to answer whether something is true or false, it did things like subtly changing a definition of a term in the middle of a proof.
For example, it proved something using "infinite sequences", but at the beginning of the proof, the words meant "sequences that are infinitely long (regardless of what they contain)", and at the end of the proof the words meant "sequences that contain infinite values (regardless of their length)". Both the first half and the second half of the proof looked correct; you had to pay attention to notice that the definition has changed.
It was funny when I asked the same question in two different browser tabs, and in one the AI proved the theorem, and in the other the AI found a counter-example. Upon closer inspection, *both* the proof and the counter-example were bullshit.
But I expect that if you ask something from high-school math, you mostly get solid results.
In my experience, even asking advanced college math is solid as long as it's the kind of thing that appears in textbooks.
Oh my, I can actually load this page without massive performance issues! Gave up on the Commentariat a while back, maybe I'll participate or at least read a bit again.
One of my favorite SSC posts is https://slatestarcodex.com/2014/02/23/in-favor-of-niceness-community-and-civilization/
In the post, Scott writes:
> I think most of our useful social norms exist through a combination of divine grace and reciprocal communitarianism. To some degree they arise spontaneously and are preserved by the honor system. To another degree, they are stronger or weaker in different groups, and the groups that enforce them are so much more pleasant than the groups that don’t that people are willing to go along.
> The norm against malicious lies follows this pattern. Politicians lie, but not too much. Take the top story on Politifact Fact Check today. Some Republican claimed his supposedly-maverick Democratic opponent actually voted with Obama’s economic policies 97 percent of the time. Fact Check explains that the statistic used was actually for all votes, not just economic votes, and that members of Congress typically have to have >90% agreement with their president because of the way partisan politics work. So it’s a lie, and is properly listed as one. But it’s a lie based on slightly misinterpreting a real statistic. He didn’t just totally make up a number. He didn’t even just make up something else, like “My opponent personally helped design most of Obama’s legislation”.
...
> This would have been a more believable claim than the one he made. Because he showed his work, it was easy for me to debunk it. If he had just said it was in some obscure report, I wouldn’t have gone through the trouble. So why did he go the harder route?
> People know lying is wrong. They know if they lied they would be punished. More spontaneous social order miraculous divine grace. And so they want to hedge their bets, be able to say “Well, I didn’t exactly lie, per se.”
> And this is good! We want to make it politically unacceptable to have people say that Jews bake the blood of Christian children into their matzah. Now we build on that success.
In 2026, it seems like we have gone off the deep end. Certain politicians lie with reckless abandon -- not 'lie but if you squint its a misinterpretation', but rather 'this is a bald faced lie by any count'. See, for e.g., the immediate claims that Alex Pretti was a terrorist. Or, more mundane but perhaps more important, Trump continuing to claim that the economy is the bestest greatest economy in the whole wide universe.
A few misc thoughts and questions I've been pondering.
1) What the fuck happened? Scott posits that there are situations where suddenly entire groups 'become honest'. Is the opposite also possible? Can groups suddenly become liars all at once?
2) How deep has this sunk into the culture? A few folks have mentioned that an entire generation has basically been raised in a political environment characterized by some level of blatant lying. Have we screwed up this norm permanently? Is there a way to bring it back?
3) Which direction does causality go? Did we lose the norm and then get the circus, or get the circus and lose the norm?
I have a theory here that if Republicans are getting called liars by biased liberal "fact checkers" over technicalities like the example cited, then strategically it makes sense to just lie. What is the advantage of making an effort to stay very close to the truth if you're just going to be called a liar anyway?
Because lies mostly hurt your own side. The other side won't believe them anyway, but your followers will, and then they'll get confused and mad when reality fails to live up to their expectations.
https://www.slowboring.com/p/misinformation-mostly-confuses-your
> technicalities like the example cited
which technicality?
> Fact Check explains that the statistic used was actually for all votes, not just economic votes, and that members of Congress typically have to have >90% agreement with their president because of the way partisan politics work. So it’s a lie, and is properly listed as one. But it’s a lie based on slightly misinterpreting a real statistic. He didn’t just totally make up a number.
I agree, that post is one of Scott's finest.
"Why did lying become acceptable?" would imply a false premise - that lying was ever unacceptable. A brief survey of the incentive model should reveal that lying is always lucrative within certain limits. Poker players are among the most experienced in knowing precisely when lying does and does not work.
If some groups become spontaneous truthtellers, it's probably because reality is a natural limit on lying, and occasionally reality manifests on something that people (1) need in order to make important decisions and (2) had previously been assuming worked in some way it didn't. Usually, people either naturally conform to reality because it affects them directly (you can't pretend you're making soup by boiling air, but you can pretend indefinitely that erbium doesn't exist), or they already believed what was true all along and so there was no opportunity for "spontaneous truthtelling" to occur.
If that's the case, then some groups can tell lies indefinitely if they're capable of insulating themselves from important decisions for long periods of time. Specifically, if their immediate biological needs are met, they're free to think more abstractly, which enables them to think less about reality. This might be fine if biology's simply never going to play a role in their direct lives until the end of each respective one. OTOH, such people can share their ideas with people who are a little closer to reality, giving them a little tug away from it. The latter people share those ideas in turn, until real tension exists between stuff coming from the abstract thinkers and the people who have to grow actual food, build actual houses, and raise actual children.
If you truly want to go back to reality, it might depend on how high you've climbed into the abstract. You might not even realize how far up you are - the previous decade appears to have been coopted by an entire school of thought that not only attacked the very foundations of logic in order to try to redesign it in terms of power disparity, but also poisoned the well against the older system, so that its own converts would believe that _they_ were the reality-oriented. So for any value of you, you can't even start with the premise that you're the good guys.
The surest way to know now is probably to ask whether you can grow your own food and build your own house, and if not, how close you are to being able.
(If this sounds vague and darkly hinting, I apologize, but it's hard to help someone with a problem if they might be actively invested in prolonging it without realizing.)
> If this sounds vague and darkly hinting, I apologize, but it's hard to help someone with a problem if they might be actively invested in prolonging it without realizing
I think I have a good sense of where your politics lie, but I'm fairly certain you don't really get my politics very well. It's neither here nor there.
> A brief survey of the incentive model should reveal that lying is always lucrative within certain limits...If that's the case, then some groups can tell lies indefinitely if they're capable of insulating themselves from important decisions for long periods of time
I think this misses my ask. Scott writes extensively about how, even though it may be very lucrative to do so, people do not outright lie. They may misrepresent, omit, selectively mishear, but very rarely do people just make up numbers. He wrote about this in the linked post above and several times afterwards.
Quoting again, more selectively: "This would have been a more believable claim than the one he made. Because he showed his work, it was easy for me to debunk it. If he had just said it was in some obscure report, I wouldn’t have gone through the trouble. So why did he go the harder route?"
This is the critical piece. There were always people who could insulate themselves from consequences, but they did not lie outright despite their inability to do whatever trad-coded things you think are important. If your response is something like "Scott is wrong, pols and journos make shit up all the time, the person who showed his work in the example above is a one-off and this isn't a real phenomenon" then we are just having a different discussion. If you agree that this is a real phenomenon, I'm interested in understanding why, and why now.
> I think I have a good sense of where your politics lie, but I'm fairly certain you don't really get my politics very well. It's neither here nor there.
I might not get your politics, true. My sense of your politics is driven by the arguments I've seen you make here in the past, and I don't follow you closely enough to say I've seen all of them. Specifically, here, though, I think I have a good guess at who you're referring to as "certain politicians" who "lie with reckless abandon", and while I can readily concede on what I'm guessing the referent is, I suspect you're conspicuously leaving out critical examples. The rest of your politics, whatever they may be, can be left aside for this.
Moreover, you're not the first person to make this general lament about blatant lying, and not the first I've had exchanges with. Past experience tells me I have to be very delicate with how I couch my answer. I can't just flat say what I think is going on; if I do, you'll interpret it as motivated (cf. your opening remark about "my politics") and won't take it seriously. You might not take it seriously regardless, which means I'm really just talking to anyone else happening by and reading this. Either way, I have to be careful, and not make assertions resting on premises that only one side of the political aisle holds.
I get that you're asking "why so blatantly, and why now". I did respond to that: it "would imply a false premise - that lying was ever unacceptable". I then went into a brief discussion of how lying was always a tool in the belt; there are always cases where it's applicable. After that, yes, I went into a brief exploration of your question (1): "Scott posits that there are situations where suddenly entire groups 'become honest' [...] Can groups suddenly become liars all at once?". If that's not the main question you wanted answered, I apologize - but you did label it #1.
Plus, you had three sets of questions. That was going to require a lot of response, so I prioritized.
To answer "why so blatant, and why now", though, one kind of needs the underlying background on when and where lying is applicable, so my response still needed to be written.
As I said, groups can lie to each other about things that aren't that important to them. Moreover, what's literally said doesn't always matter as much as what's implied. If you and I are lifelong friends, say, with a history of chatting, there are things I could say that would appear false to an outsider, but you know what I really mean, what I'm really trying to say, and it'd be a reaffirmation of friendship. What I literally said doesn't matter. It might piss off outsiders, and *that's* what matters.
And that might be what's mattering here.
That's one possible explanation for what you're seeing. Hard for me to say for sure without more detail. And if so, there are multiple reasons for why one might do that, and multiple ways to misinterpret those reasons, and it's its own entire can of worms.
What a load of hogwash. The problem with this is that when Trump lies it’s not some sort of in joke between the likeminded to piss off common ‘outsiders’.
They are simply the lies of a man with no moral core that true believers lap up as truth.
BTW
turgid != lucid
You are a theatre guy, surely you are familiar with the line “Brevity is the soul of wit.
> As I said, groups can lie to each other about things that aren't that important to them. Moreover, what's literally said doesn't always matter as much as what's implied. If you and I are lifelong friends, say, with a history of chatting, there are things I could say that would appear false to an outsider, but you know what I really mean, what I'm really trying to say, and it'd be a reaffirmation of friendship. What I literally said doesn't matter. It might piss off outsiders, and *that's* what matters.
To restate back to you, the claim is
"The reason there is a marked shift in perceived incidence of pols saying outright lies is because they care less now about precision than they did to the past, and that in turn is because they are purposely trying to anger 'outsiders'"
If that's right, I agree that would explain what is going on, but framing it as an intentional strategic choice is...pretty grim
"The reason there is a marked shift in perceived incidence of pols saying outright lies is because they care less now about precision than they did to the past, and that in turn is because they are purposely trying to anger 'outsiders'"
To refine on this: caring less now about precision implies precision is always important. In truth, it probably isn't as important as your restatement implies.
It depends on what's being said. If you're a farmer and I'm the local town guy in charge of running things and I tell you what day is going to be best for planting the spring crop, that's probably very important to you and you're going to care whether I said Monday or Tuesday. If, OTOH, I say "Monsanto's going to screw us again like they do every year", and meanwhile there's ten or so hard-working trained biochemists in town who owe Monsanto their livelihoods and take offense and want everyone to know that my remark is wrong on the details, you and your fellow farmers probably only vaguely care about whether Monsanto really is doing something this time, and much more about the fact that if you're ever in a fight with Monsanto, I'm going to be on your side.
This is the sense I'm getting at with blatant lying: everyone has plenty to worry about, and prioritizes information that addresses their direct concerns and cares little about information that doesn't. So anyone on their side can lie about the latter stuff - everyone there knows it's a lie, and it's not important (to them). If it pokes the other side in the eye, even better; the unspoken implication about one's commitment is more important than whatever blah blah details.
I agree with your assertion that it's grim - it implies the conflict theory perspective is growing stronger, and I think that's bad. That said: so what? Those of us who agree also contain a large subset of people who still can't get away from that mindset and think that the solution is to get "the other side" to stop lying so much. That's not only the wrong solution, but it's feeding the problem. In reality, _both_ sides are lying... about details their own side doesn't care about as much. And telling truth about details they _do_ care about (and getting angry when the other side gets them wrong).
> If that's right, I agree that would explain what is going on, but framing it as an intentional strategic choice is...pretty grim
If lies are so powerful, then the only thing that can destroy a lie is a better one. Regardless of the honesty of the administration, they are going out of their way to destroy existing ones, like the lie that all people are equals, and that they have "God-given" rights.
I would be interested in someone working hard to develop a data-based analysis, but my perception is that:
(a) like most things 2026, the seeds and sprouts go back a long way
(b) the strength of this anti-lying/exaggeration norm wasn't really that strong and appears stronger through a nostalgic filter
One major reference point for both is the info from the Pentagon Papers and the conduct of the Vietnam War. My read of that evidence is that government sources consistently, knowingly lied to the US public about the prospects for the war for essentially the entire duration of the conflict. Iraq II was similarly filled with falsehoods, especially the reasons for the war and the level of planning for the post-war management of Iraq.
Reagan had enough deception and sleaze to support the nickname
'the Teflon president,' with Iran-Contra one of the most memorable. Clinton ... famously slippery with the truth. GW Bush: 'Mission Accomplished' and 'Brownie is doing a heck of a job' (on top of the Iraq II stuff...)
Here is a paper built around an analysis of a 1991-92 Congressional scandal:
https://sais.jhu.edu/sites/default/files/Why%20Do%20Politicians%20Lie%20(with%20abstract).pdf
For a silly example: one of the George Washington stories we told a lot was the cherry tree myth. It is a lie about not lying! Note: I don't recall my own children hearing this in school, so maybe this myth has been expunged?
At the same time, I empathize with your frustration and the sense that this has gotten worse. I just think it is a change in intensity and not a sudden break that needs to be explained.
Coordination on truth is easier than coordination on lies, because there is one possible truth but a thousand possible lies? Schelling points and all that. Scott does not seem to get that honesty is just Schelling points. It is so hard to keep a log of what you lied to whom when. The basic Schelling point is just telling the truth unless doing so would have massive consequences i.e. getting arrested.
Turns out that politicians can just utter whatever combination of syllables they feel like excreting from their mouth-holes in the moment, no-one really cares that their utterances fail to correspond to either reality or their previous utterances, and there are no negative consequences whatsoever to this strategy so long as the noises being spouted appeal to enough of the audience.
There are negative consequences, it's just that this usually manifests as incremental disillusionment among your supporters once reality sets in.
https://www.slowboring.com/p/misinformation-mostly-confuses-your
Theahura, this is what a color revolution looks like from the inside. As the pendulum reaches its turning point a little added force in each direction, left and right, soon has the sheep utterly discombobulated and attacking each other.
Those who have enough intelligence to be neither blue or red can either get popcorn and enjoy the show or be crushed by both sides.
Eventually the pendulum will loop the loop and reset itself, not long now I think..
i dont know what this means
Theahura, I apologise for my poor communication skills. An outside force has been using media influence to amplify any story that divides Americans and mute any story that unites them. Over time the two main forces, Republican and Democrat, drift further and further apart. Shakespeare wrote that a house divided against itself will not stand, meaning that when a great challenge comes along requiring a united effort the two sides are too busy fighting each other to meet the threat.
Americans do not benefit from being at each others throat. However Americas rivals find it much easier to out compete them.
>Shakespeare wrote that a house divided against itself will not stand,
...that's a whole lot older than Shakespeare. https://biblehub.com/text/matthew/12-25.htm
FWIW, I agree this is a possibility to consider seriously.
Thanks, more clear!
But how will the pendulum continue to swing when one side is decisively destroyed? We may be looking at a permanent change within the country's power dynamics. That's going to make this country very difficult to live in for certain types of people, regardless of what choices they make.
Something, color revolutions (external pressure) start by hijacking a cultural revolution (internal pressure). A good example is the external pressure applied to Iran since the embassy siege in the 70s, it takes a long time.
The American led Western Empire has had the openly stated aim of regime change in China and Russia for a long time. What did they think the Chinese and Russians were going to do, just sit on their hands?
At the fall of the USSR Russia was on her knees, China had not yet received the goose that laid golden eggs, (the Wests industrial capacity) America was free to mould the world in any way they wished, they were masters of the universe.
How did it all go so horribly wrong? How is it possible to make so many catastrophically bad decisions in such a short time?
A little gentle pressure on the pendulum, first left then right, again and again and again until left and right are the ones pushing the pendulum and they destroy everything of beauty and value and the external force wins. This is what a color revolution is.
on this note:
>In 2026, it seems like we have gone off the deep end. Certain politicians lie with reckless abandon -- not 'lie but if you squint its a misinterpretation', but rather 'this is a bald faced lie by any count'. See, for e.g., the immediate claims that Alex Pretti was a terrorist. Or, more mundane but perhaps more important, Trump continuing to claim that the economy is the bestest greatest economy in the whole wide universe.
At least one of us must be in a bubble, because I can easily steelman both of those as either true or directionally true misinterpretations?
"The economy is better than ever" seems a pretty plausible claim; we're not in a recession, stock market is the highest it's ever been, unemployment is pretty good, inflation is low, etc. etc.. There are obviously arguments against this and some of them I can buy, but...come on. Listen to literally any state of the union speechfrom the past 50 years, from presidents left and right.
"Alex Pretti is a terrorist" is a more complicated one. There's a moral argument I can see for assuming innocence until proven guilty (I tend towards this approach, but this also means I try to extend grace for the police side too!), but if it's between "Alex Pretti was a confirmed domestic terrorist" and "Alex Pretti was a saint and a martyr who never did nothing wrong and whose shooting was incontrovertibly unjustified", we know both are obviously false now, as of Feb 10 2026...right? The main sin here seems to be "judging the situation based on rumors or mistaken information before we had all the facts", which is obviously nothing new or limited to a single person or party or side.
This seems excessively charitable.
The economy has some wiggle room, but by the ruling party's own standards it is dogshit. It is worse than it was when they inherited it on all metrics but total valuation, and you gotta admit that three companies passing the same billions around in a big circle is pretty alarming.
Re. point two: This is straight up not true. Everyone saw some guy scuffle with ice, then get held face first onto the ground and executed thusly. You can see with your eyes that even if dude was a terrorist, the people who killed him and the people who justified it had no way to know that at the time.
I think you're omitting the part where a gun discharge was heard and someone yelled "Gun!" in his vicinity.
All you have to do to get away with murder is say "gun" after your first shot.
And I think this is a false rule constructed from an extreme case.
It the case for the case you are talking about, though.
I think we can hold the government to a slightly higher standard than mag dumping into a prone guy because someone said gun.
I think it's important to distinguish between misrepresentations and lies.
I generally agree with Scott that the media (and most people) rarely lie. Cf https://open.substack.com/pub/astralcodexten/p/sorry-i-still-think-i-am-right-about you can follow the links back to his other posts in the subject. Many people spin, but it's rare for someone to outright fabricate statements that aren't true. That feels different, and is what I'm referring to that seems unique in this era.
I short handed the economy one, but I think there are examples of trump literally making up numbers -- how much revenue saved from DOGE, for eg.
> I generally agree with Scott that the media (and most people) rarely lie.
Depends on how you're accounting "rarity".
Zvi Mowshowitz had a much more accurate take: there are topics on which the media rarely lies. There are others on which it lies brazenly nearly 100% of the time. It's just a question of media culture, and it's specific to every individual topic.
In that case, my model is something like this: Trump asks someone how much money DOGE saved, a few people give some answers, he takes the highest number and then tweets it out without checking it. NYT spends a week repeating that Trump is knowingly spreading capital-F Falsehoods. Conservatives don't trust NYT (for valid reasons) and honestly don't really care about the exact number in the first place, as long as it's directionally correct. Later, some lower-ranking official either softly walks it back or justifies the logic by which they came to this number or explains some nuance that was missing in the original Trump tweet. The cycle then repeats.
This feels different (and probably slightly more understandable) from "Trump and his cabinet meet in a dark, smoky room and decide on how best to intentionally deceive the American public" or "Trump is a pathalogical liar who just makes up baseless nonsense out of a hat".
Reading this again I'm actually more interested in the defense of 'alex pretti is a terrorist' as 'directionally true misinterpretations'. I know that a big part of the postmodern project was to significantly devalue the meaning of words, but there were also factual statements that were proven untrue on video -- "the gun was being brandished at agents" is I think the example that Noem went with.
I can maybe see this as "Noem just fired from the hip", and I agree that this is not the 'dark smoky room', but I also didn't specify that there had to be a dark smoky room to begin with. I was more curious about why there seemed to be a degradation in truth-telling writ large.
One reasonable answer is "more people are just willing to shoot from the hip more, which naturally increases the number of lies being told, where 'lie' is defined as 'saying something that is not true' instead of 'saying something with the intent to mislead'"
I think your last paragraph is true, and explains Noem's psychotic willingness to assert something that would later be shown to be obviously false.
But Paul Brinkley really got to the heart of the matter with his insightful comment elsewhere in the thread:
> As I said, groups can lie to each other about things that aren't that important to them. Moreover, what's literally said doesn't always matter as much as what's implied. If you and I are lifelong friends, say, with a history of chatting, there are things I could say that would appear false to an outsider, but you know what I really mean, what I'm really trying to say, and it'd be a reaffirmation of friendship. What I literally said doesn't matter. It might piss off outsiders, and *that's* what matters.
Administration officials can call Pretti a "domestic terrorist" without shame because the phrase doesn't have a consistent, consensus referent anyway, and they think 𝘢𝘭𝘭 𝘰𝘧 𝘵𝘩𝘦 𝘱𝘳𝘰𝘵𝘦𝘴𝘵𝘦𝘳𝘴 𝘢𝘳𝘦 𝘵𝘦𝘳𝘳𝘰𝘳𝘪𝘴𝘵𝘴, at least the ones engaging in disruptive tactics.
> at least the ones engaging in disruptive tactics.
Though notably even here, the words are doing a lot of eliding of meaning.
> the phrase doesn't have a consistent, consensus referent anyway
I think more generally, *no* words have a consistent consensus referent. This wasn't some novel truth, people were talking about signs and signifieds back in the 60s. So is it only now that pols realized they had the ability to really lean into it?
>Reading this again I'm actually more interested in the defense of 'alex pretti is a terrorist' as 'directionally true misinterpretations'.
That one is pretty easy. If we define terrorism as deliberate violence to achieve political ends, Alex engaged in terrorism when he kicked the officer's car 10 days before this and broke its tail light. Therefore, he was a known terrorist. Yes, I would agree it's playing a little fast and loose with definitions, but if e.g. hypothetically Noem was sued by Alex's family for slander or something, I think it's an argument that could work in court.
As for the "directionally true" part, it correctly challenges the notion that Alex was a completely innnocent, peaceful protester -- this is the signal from the noise.
>One reasonable answer is "more people are just willing to shoot from the hip more, which naturally increases the number of lies being told, where 'lie' is defined as 'saying something that is not true' instead of 'saying something with the intent to mislead'"
This framing seems broadly accurate to me.
---
On another note...I'm actually interested if we agree on the same set of baseline facts, since I have a more right-wing media diet than most and am probably personally to the right compared to most people here (I'm still a card carrying fence sitting grey tribe centrist though, I promise!). I'm probably wrong about some of these since I have not followed the story as closely as others.
the version of the story I know (from my own perspective):
* ICE and border patrol end up deploying in force to minnesota for complicated political reasons.
* a lady gets shot by ICE which causes a lot more protests occur after this. Most of them are peaceful, but some go too far and end up disrupting innocent bystanders or are violent or borderline violent.
* both Trump and Waltz engage in some dangerous and unproductive political brinksmanship.
* a guy named Alex Pretti is involved in some of these protests. He is definitely in the "violent or borderline violent" category. There is a video of him attacking a police vehicle during one of these protests and causing significant damage. Somehow he doesn't get thrown in prison for this.
* he later decides to show up to another protest in order to antagonize armed federal officers while carrying a concealed firearm. This is technically his right, but it's also a horribly, indefensibly stupid idea.
* the border patrol officers seem to recognize him
* the officers attempt to arrest him for unclear reasons.
* he resists arrest for unclear reasons. There is a confusing scuffle. One of the officers probably disarms him during the scuffle. Two of the officers suddenly panic and open fire on him. It's still unclear whether or not the cause of the panic was an accidental discharge from Alex's gun at this point.
* When all is said and done, alex is shot dead.
* the federal government irresponsibly claims the shooting was justified (before all the facts were known)
* the new york times irresponsibly suggests the shooting was probably unjustified (before all the facts were known)
* as of feb 10, 2026, it is still not obvious whether the shooting was legally justified or unjustified.
> if we agree on the same set of baseline facts.
Ya this is interesting. I think just given what you've laid out, my understanding of the relevant facts is much less sympathetic to the government than yours. For added context, I'm not on any social media (no Twitter, Facebook, Reddit, insta) except for a bit on substack where I don't subscribe to anyone but mostly get matt yglesias and Nate silver on my feed. I subscribe to Scott, Matt Levine, and Ben Thompson. I read a lot of AI papers and skim Hacker News roughly daily. I don't trust most media outlets, but I will get and read links from folks in my friends discord.
With that in mind, here's my understanding of all of the relevant facts:
> ICE and CBP are deployed to Minneapolis for complicated political reasons. Notably, these reasons have basically nothing to do with actually solving illegal immigration (Minnesota in general and Minneapolis in particular have a very very small illegal immigrant population, being ~1500 miles from the closest southern border crossing).
> The local population does not want ICE there, including the entire local and state governing apparatus. Local protests begin basically immediately.
> ICE seemingly purposely inflames tensions. They weild guns indiscriminately, escalate rapidly in countless interactions, and are caught on video beating individuals at a moments notice. These videos spread, resulting in more protests that get more organized.
> The federal government issues memos and public statements indicating that ICE has immunity. ICE and the feds are brought before courts constantly, lose frequently, but continue to get more brazen. Videos circulate of ICE detaining US citizens, shooting priests, and detaining 5 year olds. School attendance plummets.
> ICE shoots a woman (I prefer the active voice here). She survives. The federal government immediately shuts down investigations and refuses to cooperate with the state. They tar the victim as a terrorist, claiming the agents feared for their life. Soon after, evidence is released showing ICE agents bragging about the shooting ("7 holes with 5 bullets"). No federal investigation occurs. This court case is ongoing.
> Soon after, ICE shoots a woman. She dies. By this point, protestors have organized to make sure that all ICE action has blanket video. Again the federal government stonewalls and claims the woman in question is a terrorist. The federal government opens investigations of the victim's widow (!). The ICE agent is not investigated, and the federal government refuses to share evidence with the state. Despite this, video leaks. It becomes immediately clear that regardless of what happened in the moment, ICE clearly and unilaterally escalated the encounter, and then later refused to allow the woman medical attention as she bled to death. The actual shooting is debated, but it is undoubtedly against training (at the least) and in my opinion from my own lying eyes state sanctioned murder.
> This leads to an increase in protests, and in the organization of those protests. I think very few of these protests are violent (if they were, the violence would be more specific and televised; it is neither. There is much more control and discipline than in LA). Local and state government are pressured by locals to do something about the actions of ICE. (I didn't follow the exact details of the brinkmanship you refer to, but I'm not surprised that it was happening in Minneapolis.)
> Alex Pretti gets involved around this time.
> He goes to at least one protest where he kicks out a tail light. He is beaten for this, but not arrested. (I do not think this makes him obviously a violent protestor -- there is very little indication that he was violent outside of this one interaction, and there is so much video coverage and his life under so much scrutiny that I would expect more evidence to surface. It hasn't.)
> Ten days later, Alex is at another protest. Here, events are captured basically entirely on video. The events as I see them:
- Alex and a woman are video taping an officer but are not otherwise being belligerent, intervening, or getting in the officers way. Notably, the officers are not doing a raid, so there is no action to intervene.
- an ICE agent confronts the woman, seemingly as she is walking away, and escalates aggressively by pepper spraying her. It does not seem like there is any obvious provocation. The woman falls to the ground.
- Alex gets in between the ICE agent and the woman, with just a phone in hand and is not violent or interacting with the agent beyond getting between the agent and the woman. The ICE agent pepper sprays him too. (I don't make any claims on whether the ice agent recognized Alex. I think it is far worse for ice if they did)
- several agents pile on top of Alex. One is seen on video hitting Alex on the head with a blunt object, possibly a pepper spray can.
- video shows an ice agent pulling a gun from Alex's holster. This is the first time the gun is seen. In that same video, it looks like the agent that fired the first shot sees that the gun has been removed from the holster. This agent moves another agent out of the way and then fires the first shot into Alexs back. (At this point I've seen enough videos to call this an execution. But even if we're being generous, ice does not get to hide behind the chaotic conditions they created as justification for their violence)
> In the immediate aftermath, the president and head of DHS release statements claiming Alex is a terrorist, the latter claiming that Alex brandished a gun with the intent to kill many agents. These statements are proven false by video. The NYT may claim some things also -- I don't follow the NYT, and I don't see why it matters what they claim as they are not the federal government.
> The federal government refuses to even release the names of those involved. They again stonewall the state, refusing to provide evidence or even allow the state to investigate the crime scene until after they have left.
> As of Feb 10, none of the ice agents involved in any of this are under investigation. In fact, there is no investigation at all, so there is no way to determine if the shootings were justified or lawful. This is because the apparatus responsible for this investigation has unilaterally decided not to do one.
I think the key principle here is that discussing the specifics of the shot outside of the context that ice created is misguided. ICE claims "ah there was so much chaos, how could we have known better!" But the reality is, ICE is the one with the agency -- that is why they are literally called agents. When I examine what I know of the facts, including your version of such, I come away thinking that it is ICEs choices that led to this outcome.
> "Trump is a pathalogical liar who just makes up baseless nonsense out of a hat"
I have yet to see any evidence that would contradict this hypothesis.
right, well, also this
I can see myself read my own bias into your question and I think my answer might not be exactly what you were asking but when read "What the fuck happened?" in relation to lies I generally think "This another person on the left complaining about the right". Where as, as I only listen to left media, all I hear is spin and lies from the left and I'm surprised they don't see it.
A concrete and hopefully incontrovertible example is the New York Times writing about Scott and all the spin and lies that were in it.
https://www.astralcodexten.com/p/statement-on-new-york-times-article
I used that example because I hoped it was clear it was a lie from the left. But, every time I listen to almost any media I hear/read 95% spin and lies.
This is not to excuse the right. I don't listen to right media at all so I have no idea how bad it is there. My point is only that it's bad enough on the left to encourage escalation. The whole "pull the mote from you're own eye first".
If guessed wrong about where this was coming from, sorry in advance.
Which part of that is a literal, indisputable lie? Even Scott wrote a whole post saying the media rarely lies.
Possibly none of it (or essentially none - the media can certainly tell falsehoods by accident, if not on purpose).
One of the lessons I took from "The Media Very Rarely Lies" is one I already knew - it's very possible to imply all sorts of falsehoods without actually saying them (and that this matters - it makes the difference between grumbling and a lawsuit).
I think the part of the US that didn't like what corporate media was telling them, was also noticing that the media could say things without literally saying them, although they also often confused the two, asserted the media was lying, would get argued down ("they didn't actually literally say that"), get frustrated, and simply keep not believing that media, but also stop saying they were lying. People on the other side similarly confused this silence with belief.
I'm in Aerospace. I observe that in every case I've ever seen when the news touches on Aerospace, every single word is wrong, including "and" and "the." Based on that sample, I conclude that every word they say about any technical subject is equally likely to be correct.
Law and Economics are every bit as technical as is Aerospace. QED.
As a counterpoint, I've seen a number of articles on topics with which I was familiar which were remarkably fair and accurate.
I can attest that most of what they write about law is wrong. And most is easy to get right.
On another note, a recent NYT an article re Iran, said, "While the government appears to have largely suppressed the protests, Mr. Memarian, the think tank analyst, warned of “a dam that could suddenly crack.” “It may harden public resolve, making the government more vulnerable if protests continue, rather than less,” he said."
The person in question is identified as "Omid Memarian, a senior Iran analyst at DAWN, a Washington-based organization focused on the Middle East.." Leaving aside that this guy is a journalist rather than a political scientist, the NY Times does not understand that they need to consult an expert on regime change, or government repression, or political violence, rather than an expert on Iran They do this all the time.
The Republican Party never really succeeded in making the case that the tax cuts for the wealthy were good for the average voter. That largely ruled out an honest approach where they would tell voters what they were going to do and explain why those policies would be good for voters. I think that pushed Republicans to see how much dishonesty they could get away with. What undermined the norm is that it turned out that dishonesty worked.
Read about the "Two Santa Claus Theory" by Jude Wanninski. He argued in the 70's that Republicans could get credit for fiscal conservatism and also stay popular by always cutting taxes and not cutting (popular) spending. The history of Republican administrations from Reagan to now shows that they took this theory to heart.
I think the plan was that by cutting taxes and not spending, they'd force the Democrats to be the bad guys and cut spending, but in practice it hasn't worked out that way.
> I think the plan was that by cutting taxes and not spending, they'd force the Democrats to be the bad guys and cut spending, but in practice it hasn't worked out that way.
Well, I think you're right about the conclusion. ("It hasn't worked out that way.") But my understanding of the theory is radically different from yours. The slogan was "starve the beast". The original idea was to cut spending; that's what they wanted to do. But when that proved to be politically impossible, "starve the beast" was the theory that if you cut taxes, spending would have to fall because the amount of money available to spend would be lower.
That was never a very believable theory and anybody who fell for it should be ashamed IMO
Oh? How long can you continue spending money that doesn't exist?
Once you've got a fiat currency, it's all the same money, fake or not -- it all comes from that magic printer in the Treasury that goes brrr, right? There's no economic law that forces people to stop spending just because their deficit is getting big.
The closest we got to cutting spending was the Clinton years, when federal spending grew at an average rate of 1.51% per year (after inflation), and we eliminated the Federal deficit. George W. Bush promised to run a balanced budget if elected, but when he won he cut taxes and increased spending at an average of 4.56% per year in his first term. (The average increase over his entire tenure was 5.62% per year, pushed up by spending to address the Great Recession during Bush’s final year.)
I somehow missed the “Two Santa Claus Theory,” so I thought Republicans actually cared about the deficit until I saw the lack of pushback against Bush.
Seems to me that a frequent mechanism that causes this is a weaponized "fallacy of gray". It works like this:
1) Notice that some other person/group is not perfect and did something wrong.
2) Therefore, you are morally entitled to do the same thing. Or a similar thing. Or a different thing. You are entitled to do ten times the same thing, or a hundred times the same thing. (The logic is that things are either forbidden or allowed, and now that we have established that it's clearly the latter, it doesn't make sense to argue quantitatively.) You are entitled to do the thing even to someone unrelated, or only related by a vague analogy, to the person or group who originally did the wrong thing. Your friends loudly support this interpretation.
3) "Hey, if *they* are allowed to do that, then so am I. Stop being unfair to me! This is a legitimate self-defense."
To some degree this is natural and many children do it. An important part of growing up is to unlearn this pattern of escalating every perceived offense. But most people still keep this in their hearts, only suppressed by the expected social disapproval.
So you need to get them loud social approval. Easy, if you own the media, or can manipulate social networks.
I think it's worse than that - I think more often than not 2 actually comes before 1. People decide on the terrible thing they want to do and their entitlement to do it first. *Then* they go looking for the "high tech lynching" of Clarence Thomas or the Lewinsky scandal or the 1876 election or whatever they can claim that the "other guys did first."
The elites destroying their republic aren't doing it because they were nice decent people who saw the sins of the other side and somehow overreacted. They're miserable bastards who'd be destroying their republic either way. In a world where Gore v. Bush hadn't happened, or where Mitch McConnell had given Merrick Garland an up or down vote, or whatever, they'd be doing the same things; they'd just be pulling different justifications out of the couch cushions for them. The desire to commit the wrong creates the need for the justification, and therefore a justification is found. If the best justification you can find is weird trash, then that's what you go with.
That's why the proffered justifications are often so wildly goofy and disconnected from what is allegedly being done "in response" to them. When people's best argument for the thing they want to do is clearly a turd, but they still fully want and intend to do the thing anyway anyway, it's no surprise that you suddenly see them throwing around full-throated turd-claims.
I think the people willing to do this have always been there, but the internet and social media have made the tactic more effective in the digital age than it was before. Easier access to information when you are trying to scrape up justifications, more hands to do the scraping since mass numbers of your tribe can get engaged online digging, and a pluralistic media environment that makes it easier to flood your tribe's media environment with those justifications.
There is one step missing. First you are entitled to do the same thing, but then you realize small wrongs are the same as big wrongs, so now you are entitled to do 100 times more.
What, why are small wrongs the same as big wrongs, this confuses the consequentalist? Because most people are not consequentalists, they are motivated by shame and judgement. If we say "Bob is a liar", it does not matter how big is the lie, it is largely the same judgement and shame.
This is why don't really believe in consequentalism, because it is incompatible with intuitions and social dynamics. Real life morality is playing the "who is the asshole?" game.
Then I guess the difference from the past is that thanks to internet, we can always find one person somewhere on the planet who broke a rule, and then we can share the article with all members of our tribe, and say: "see? if *they* can do it, so can we!"
In the past, I guess people would apply additional heuristics and e.g. only consider the breaking of the rule relevant if it happened in their city, if it was done by a person from the same social class, etc.
I've been building WagerAlpha — a tool that auto-pulls your Polymarket or Kalshi bets (via wallet address or API) and calculates real performance metrics: raw ROI/alpha, closing line value (CLV), bet Elo rating, category confidence scores, provisional EV, Sharpe/Sortino/Calmar, max drawdown, etc.Goal is to give sharps verifiable proof of edge instead of just claims, and later let high-performers create paid communities (platform takes 20% cut).Beta is live with full Elite access currently free. If you're trading prediction markets seriously and want to see your actual numbers (or just give feedback), link your bets here: https://wageralpha.io
Happy to answer questions or hear what features people would want. Not financial advice — just a tool I built for Polymarket/Kalshi tracking.
The site seems to be down. Did you kill the project or is something else happening?
Hi Joshua, that was a technical issue that i have since resolved. The site should be back up.
> bet Elo rating
How do you calculate a bet Elo rating from just one person's bets? Wouldn't you need a full history of every bet on the platform?
This looks like a really good tool!
Can I enter the reference for someone else's account and see their stats?
Ideally, it would be nice to have the following:
(1) tagging bets into sub-portfolios: there are many different types of questions and it would be useful for a trader to identify whether they are better at some vs others. I understand tagging ex post opens the possibility of cherry picking in a way that muddies the performance reporting.
(2) a way to record the trader's private estimate of probability at the time of trading.
Lastly, a scaling trades up is a major concern. Something around market size and transaction costs is necessary to really know if a trader could handle an interesting amount of OPM.
Hi Joshua, thank you for the feedback I really appreciated it.
1. I made some updates based on your comments and now the bets are categorized based on category in the dashboard, I had the app autosort that instead of having people tag their own. I also added a way for traders to privately list their own confidence, however its hard to prove it was before with pulled API trades so I kept that field private for people to analyze their own predictions.
2. I was thinking the communities were more a way for strong prediction analysts to share their analysis and points of view than something like pure pick selling. Although I will see if there is some way to include market size in the analytics as well since I agree that an edge in highly traded deep market is both more difficult and valuable than an edge in a small and shallow market.
If I want to support Iranians in their resistance against the clerical regime, what would be the best charity to donate to?
I guess I'm most interested in charities whose explicit goal is to bring about the end of the mullah regime (although I guess they might not be openly stating that even just out of a concern for the safety of their employees).
Thanks for any recommendations.
I guess currently anyone who can help bring more Starlink systems to Iran would be great. Or weapons so that next time the people can actually shoot back at the revolutionary guards. I doubt anyone who can do that actually talks about it publicly on the internet. But that is exactly what makes it hard to support them.
You may as well just send money directly to the Mossad in that case
Ha. I wouldn't actually mind sending money to Mossad if they could guarantee that they will only use my money to bring weapons and Starlink terminals to Iranian resistance.
Which do you think would be more damaging to the fabric of American society (such as it is): the dismantling of monolingualism – perhaps through the proliferation of things like the 2026 Superbowl half-time show – or the conquest of Greenland – and the concomitant shift in international relations?
Largish parts of America are already bilingual and I can usually talk to big companies in Spanish by pressing nine, so I'm not sure how explosive formal bilingualism would be. In Canada it's a permanent annoyance but people aren't stabbing each other over it.
How would a Spanish-speaking nation not be monolingual?
It is well established that bilingual individuals benefit from cognitive advantages compared to folk fluent in only one language, especially as they get older. All the people around here desperate to try dubious interventions in the hope of even tiny increases to kids’ intelligence should be all for maximising exposure to multiple languages.
I'm out of the loop (and out of the continent), what happened in this show?
The halftime show was by Puerto Rican rapper Bad Bunny and was mostly in Spanish.
There's no creeping bilingualism. Virtue-signaling white and Jewish liberals put Bad Bunny on the sportsbowl as a middle-finger to normal Americans and also because of their sadomasochistic desire to sabotage the Democratic party's chances. In places like lots of Hispanics live there's little creeping bilingualism as Hispanics like English-speaking jobs, movies, cultural phenomena, etc.
The right's schtick of "only people who agree with me are Normal Americans" has always been dumb, but I think claiming that Normal Americans aren't interested in seeing a Grammy-winning, chart-topping artist at the Super Bowl is an especially dumb variety.
Like, come on man. You and I may not listen to Bad Bunny, but clearly a lot of people do! He did not go multiple-platinum solely because of latte-sipping West Coast elites.
I may not be normal, but I liked the halftime show. My complaint about there being no subtitles is I feel like I missed a huge part of the art by not being able to understand the lyrics.
I wish there were subtitles too.
Normal Americans aren't interested in hearing a performance in a language they don't speak.
It's amazing I have to explain this to people. Are you a Democratic party staffer or something?
>Normal Americans aren't interested in hearing a performance in a language they don't speak.
Why not? Plenty of people are interested in K-pop, J-pop, and J-rock despite not speaking Korean or Japanese. If we look to the past English was insanely popular in the European disco scene at a time when English fluency was much lower than it is today and most people listening to English music couldn't understand a single word. English is still popular in many Japanese music genres despite neither the artists nor the intended audience understanding any of it. It just sounds cool.
"Plenty of people" are not "normal Americans". I think Turok is probably correct that listening to music with foreign-language lyrics is a niche taste in the 21st century United States, even if it has been somewhat more popular in other times and places. Instrumental music is language-independent; if there is value in adding lyrics, it's mostly by way of the audience understanding the (poetic rather than literal) meaning of the lyrics.
Yes, some people do find other sorts of value in it. And there are plenty of venues - most of them, really - where music catering to a minority taste is appropriate. The Super Bowl, is probably not the right place for that unless you're deliberately trying to offend a large fraction of the audience - or parade a naked emperor in front of them to demonstrate you have the power to make them say "great clothes, man!"
So what is a "normal American"? The most average guy possible? If it doesn't raise eyebrows I'd classify it as normal. And someone telling you they listen to K-pop is not raising any eyebrows. Hell for a brief few weeks in 2012 a K-pop song was the most popular song in the country thanks to viral memetic spread.
> Instrumental music is language-independent; if there is value in adding lyrics, it's mostly by way of the audience understanding the (poetic rather than literal) meaning of the lyrics.
I strongly disagree on this point. The vocalization is an integral part of the sound of a song, it doesn't merely convey information the way subtitles would.
If you don't believe me, listen to an instrumental version of your favorite (non-instrumental) song, and read the lyrics in parallel; make sure you aren't "sounding out" the vocalization in your head. It will feel different than listening to the original version.
> sadomasochistic desire to sabotage the Democratic party's chances.
This seems contrived to get to your initial conclusion, and the current odds on prediction markets don't support this idea. It seems more likely that they're betting they've successfully replaced enough of the Anglophone population, and the "middle-finger to normal Americans" is their way of honestly signaling their allegiance to their new Hispanohablante voters.
I don’t speak Spanish but listen mostly to Spanish-language music on the radio. Aside from the fact that I very much enjoy Norteño, reggaeton, and other predominantly Spanish genres, an advantage of this is that when the ads come on they are easy to tune out.
Or if i don’t feel like tuning them out then the ads are interesting rather than annoying, because my brain can work on the puzzle of picking out whichever few words I recognize to piece together what they’re saying, and then ponder the similarities of what is being sold to the Spanish-speaking market segment where I live.
Thus the destruction of monolingualism makes my experience within American society better, not worse.
>the dismantling of monolingualism<
Nobody tell him about the west coast. His heart can't take it.
(Los Angeles is short for Lose Angel Ass. It's a diet name.)
((California is short for Call If Horny, Yeah?))
The US has always been multilingual. Before WW1 there were entire German speaking neighborhoods. In San Francisco, there's still an active German language Lutheran Church, remnant from the previously active German immigrant community. German, Polish, Italian speaking neighborhoods were ubiquitous. But they became white as western European immigration dropped off. And the same thing is happening to Latinos. It's immigration that keeps topping off the culture. But the older Latino generations are becoming white. America can handle Spanish on the Superbowl just fine. That isn't going to destroy us.
However invading Greenland, destroying the pax Americana, losing our dollar reserve currency -- That's an apocalypse.
>But they became white as western European immigration dropped off.
They were always white.
To my understanding, the social category of "white person" is well less than a millennium old[1]. Humans have always come in a range of ethnicities and a range of appearances, but different cultures have drawn very different category boundaries through that space. Given that the category boundaries are partly artificial in the first place (and change with time), insisting that this or that group has *always* fit into the scheme in such and such a way seems...not especially defensible.
As a side note, I find it equal parts puzzling and frustrating when people act as though their own culture norms are perfectly universal and perfectly logical, even in contexts where they're being directly called into question.
[1] And before you raise any objections about who isn't or isn't "objectively" white, I'll invite you to consider that the same *individual* can have considerable variation in skin tone throughout their life due to things like health and sun exposure. Obviously different people have different skin tones, but the boundaries are anything but sharp.
There was a time when no Irish, no Polish, no Italians signs were posted in polite spaces. They were the outgroup and over time, they become white.
"Outgroup" is not the antonym of "white". Not even in this context.
Cultural Whiteness is trivially not about skin colour alone. When people talk about groups "Becoming white" they're talking about how the various fractured white ethnicities, (anglo saxons, Germans, Polish, Irish etc etc etc) became homogenised and accepted into mainstream america whereas before they had been seen as a dangerous out-group.
The american conception of whiteness doesn't easily translate over to europe, although over the last ten or fifteen years its been one of the biggest exports. American's would find the level of discrimination Polish people faced in europe in the 00's and 10's bizzare.
>Cultural Whiteness is trivially not about skin colour alone. When people talk about groups "Becoming white" they're talking about how the various fractured white ethnicities, (anglo saxons, Germans, Polish, Irish etc etc etc) became homogenised and accepted into mainstream america whereas before they had been seen as a dangerous out-group.
Motte and bailey.
Clarifying is not a motte and bailey, but simply repeating "motte and bailey" is a thought terminating cliche. If you'd rather tilt at windmills than understand what people to the left of you actual think, thats your perogative though.
>simply repeating "motte and bailey" is a thought terminating cliche.
No, it's calling out a common technique of dishonest people and intellectual movements.
>If you'd rather tilt at windmills than understand what people to the left of you actual think, thats your perogative though.
I understand Leftists perfectly well, including their habit of thinking anyone who disagrees with them fails to "understand" them.
Do, it is a nuanced discussion of the issue. There were white persons organizations that did not admit Irish persons.
Depends on how you define "destroy". The European ethnic communities are all homogenized and assimilated. It's not unreasonable for people to think of the American social fabric having been destroyed after White Flight/the Great Migration. But it's not like everyone died of sadness and America just ended, just because terrible shit happened. In what way would invading Greenland/losing reserve currency be an apocalypse?
I think there's a reasonably good chance that the end result of a U.S. invasion of Greenland and the loss of reserve currency status for the dollar would be the United States of America ceasing to exist as a unified political entity. Relations between different parts of the country are already strained. Piling on what would likely be the least popular war in the history of the nation AND a great depression-level economic disruption would certainly touch of massive civil unrest. Do you think the country's leadership is capable of navigating such a challenging and delicate situation without causing it to spiral further? I don't.
Now, whether or not the U.S. falling apart counts as an "apocalypse" is very much a matter of perspective. But it would certainly be a much bigger, more abrupt upheaval than some internal migrations and generational cultural shifts.
Maybe apocalypse is a strong word. But to me it signals the loss of tangible hard power and a hard economic future for Americans. If we lose the ability to dictate the trade terms, sanctions, our ability to offload treasuries in our time of persistent deficits. People run to the dollar in times of world crisis, that keeps our borrowing costs low. Trump can only bluster about tariffs and deals because people NEED to trade with America. If we lose that 'need', then we got nothing. Invading Greenland pushes the world into a future where people don't NEED America.
I am long America. I want the SP500 to grow 14% every year for the next 20-30 years. Losing our reserve status will more directly threaten that, than Spanish on the Superbowl.
> People run to the dollar in times of world crisis, that keeps our borrowing costs low.
They used to, but now the opposite is happening.
> the dismantling of monolingualism – perhaps through the proliferation of things like the 2026 Superbowl half-time show
"Americans are so funny. You've accidentally listened to one song in a foreign language, and now you think the sky is falling and the fabric of society will crumble. Meanwhile, large parts of the rest of the world have been listening to English music for 70+ years, and their non-English, monolingual society hasn't been 'dismantled'."
That would be my reply to you if I didn't know that you habitually post incendiary, intentionally baiting questions, packed with false dilemmas and unjustified assumptions. If you were really interested in a discussion about social cohesion, as opposed to just stirring the pot, you could have asked something like "Would a proliferation of foreign languages in American entertainment traditions – like the Superbowl half-time show – inflict damage on the social cohesion of American society?"
>and their non-English, monolingual society hasn't been 'dismantled'."
Well... it did. Have you heard how Berliners talk? It is terrible. "Wir suchen eine nice Location für unsere Wedding". Words like "cringe" or "sus" keep popping up in German or Hungarian. Even the Frogs are giving up, and they were always keen on keeping the French language "clean".
If you speak German well enough to judge how native speaker use the language, *surely* you should be aware that borrowing foreign words is such a common and longstanding German practice that German has a dedicated word for it:
https://de.wikipedia.org/wiki/Lehnwort
Of course, the language we refer to as "German" has been fragmented into many dialects for as far back as we have written records of it. So complaining that foreign influences are going to "dismantle" German language or culture could never be anything but a profoundly unserious argument.
They can have a Wedding in Wedding!
When I went to Berlin I was told multiple times that a "Berliner" is a type of donut, so I guess I haven't heard how it talks.
Many German foods are named after cities, presumably cities that someone thought - or wished to give the impression - that they were characteristic of. This is normal, analogous to a "Philly cheesesteak" or a "Chicago-style pizza".
All German city residency demonyms are named after cities.¹ The words are the same. This is even more normal; the suffix is even the same as in German's sister language, English. Berliner : Berlin :: New Yorker : New York.
¹ As far as I know. It's not out of the question that the name of the city and the name of the residents could differ, similarly to how we say the people of "Myanmar" are "Burmese".
In addition to what deusexmachina said: Berlin's problems don't stem from multilingualism. At best, it's another symptom of immigration without assimilation, but that has certainly not been caused by the broadcast of foreign language music.
How is the use of non-German words evidence of a society being dismantled?
The Kaiser's troops spoke of Detachements, Pelotons, Tranchees, Parapets, Lafettes, Fourage, Coups de Main, Retraits, and today people speak of cringe, sus, and "nice" things.
Why would a multilingual US be automatically a bad thing? Switzerland has 3 (technically 4) languages and works just fine. And English and Spanish are actually relatively close so easy for people to learn both. They are closer to each other than French is to German.
Conquest of Greenland would be a monumentally stupid idea that would hurt US the most, then Europe and could only help Russia and China. It would mean US burning all of its remaining soft power for ... ??? I don't know even know what, there is nothing to gain other than Trump's ego, everything else can be easily achieved under existing treaties. It would mean the collapse of NATO which would greatly weaken both US and Europe and in general pretty much the collapse of the current world order. The idea is so idiotic that hopefully even some of the less sycophant voices in the Trump administration seem to be steering Trump away from the idea. Just rename a few airports or something after him instead ...
Let's be honest here, on the canton level Switzerland is very monolingual. I was to Winterthur, a little project. Yes they had to pass some high school exam in French and Italian. Then they went on never using it ever.
Yes, that is probably fair. But I also think it is not a problem. The country still functions fine. I think it is mostly because Switzerland embraces strong federalism and cantons are fairly independent of each other in many ways. They are also much smaller than almost all US states and so it is fairly easy to move between them if you're not happy in one (you can still visit your friends and family in the previous canton in a matter of 1-2 hours).
On the other hand with bad policies you can get the way of Belgium which is not even a real country anyway :-P. In many ways it's two countries glued together into one and the linguistic and cultural differences make it worse. But if Belgium became more like Switzerland in its governance I think it could still be fine. And arguably Switzerland is in many ways more American (in the sense of its government living up to the ideas of the founders of the US) than the current USA.
Multilingualism works OK in Switzerland because France and Germany and Italy (or at least northern Italy) are similarly developed and wealthy countries. Similarly, Canada.
But the gap in GDP and HDI between Anglophone and Hispanophone parts of the Americas is enormous. The GDP of Latin America is something like $11K per capita.
> Multilingualism works OK in Switzerland because France and Germany and Italy (or at least northern Italy) are similarly developed and wealthy countries.
This doesn't feel like a complete analysis. Notably, while France, Germany, and northern Italy might be similary wealthy to each other, they're all much poorer than Switzerland.
I'm not sure what the mechanism you're hypothesizing is, but the large gap between Switzerland and its neighbors seems like it'd be relevant.
Multilingualism does not imply more immigration. One can imagine Spanish becoming more officialized, and immigration becoming more strict. So the GDP of other countries is not that relevant.
I can see issues for a country shifting from Mono- to Multi-lingual system. In fact, I don't think it is a good idea. But the fact that a country in South America both uses Spanish and has lower GDP than the US is not a relevant factor to consider.
Well, Ticino and Sicily are not quite so close (as you point out yourself), but ok.
So I guess the idea is that there would be too many immigrants from Latin America which is bad because?
I guess the point is one of the following - a) they bring crime or b) they will live of welfare?
I think both a) and b) would be reduced if the country were bilingual but I concede that b) is a fair point as long as you want to maintain the welfare state.
But I think the solution is to abolish or significantly reduce the welfare state. US managed to soak up millions of mostly poor people from Europe in the past without any long-term issues. Actually, the exact opposite happened as without them, US would never be the superpower it is today. And Latin Americans are culturally not all that far from Spaniards or Italians, hence neither from northern Americans. I don't see why, under the same conditions, their immigration would be less successful than that of the Irish or the Italians in the 19th century.
You can probably even keep the welfare state if the law said something like "anyone can come but you only get access to welfare after 10/15/20 years of working here"*. It would also be fair if it said "and you don't have to pay into the welfare fund while you are not yet eligible to collect its benefits" but that would never work politically. I think people would still be willing to come even so.
*legally, of course, but that is trivial when you give a work visa to anyone.
> You can probably even keep the welfare state if the law said something like "anyone can come but you only get access to welfare after 10/15/20 years of working here"
AIUI, immigrants are already banned from receiving most forms of welfare for five years.
Well, then figure out the right (politically and economically) number of years or even tie it to citizenship and it should work. 5 might be too low. 15 should be enough. Nobody is going to work for 15 years in hopes of living off welfare. You might scrape by for 5 years but not 15.
I don't think the number matters politically, because nobody knows what the actual laws are anyway.
I think you just took the bait here. Shankar’s fishing lure is what my dad called a DuPont spinner — half a stick of dynamite in a Mason jar.
I dunno. His hypotheticals are pretty far from being inflammatory triggers . If he mainly wanted to stir the pot, seems like the thing to ask would be which would give the US the greatest chance of improved unity and goodwill: assassinating Trump or letting him and his staff continue with their agenda. (Although how many of us haven't asked ourselves that?)
Why don’t you give the pot stirring a rest Shankar?
The dark matter of the human psyche is an emotion which you could call something like "group-level humiliation". It's a feeling that you get when you're part of a group that have been humiliated and low-status throughout history, and it brings feelings of shame as well as resentment against history's winners. This feeling has motivated all sorts of acts throughout history, ranging from the merely irrational to the mass-murderous.
But the reason I call it dark matter is that I can't personally relate to it at all, I think a lot of us here can't, because our ancestors have always been winners, or close enough to winners. I can understand what it means to be humiliated on an individual level, but not what it means to be humiliated on a group level. I just don't get it, why do you _care_ that the so-and-sos beat your ancestors in a war three hundred years ago? I can only surmise that this emotion exists, and is powerful, from looking at the irrational actions of people in other groups.
There are a lot of these culturally transmitted values that are irrational and alien to middle-class Americans. There are social class-based distinctions where one is expected to show deference to those of higher class (and how to show deference is socially coded as well). There are honor and shame based distinctions where one is expected to respond to verbal insult (especially to one's manhood or family) with physical violence. "Why do you care about an ancestral ethnic grudge?" is as alien as "Why do you consider 'this person insulted my mother' as a reason to kill someone?". You could even think about norms about haggling or bribery or patronage as falling into the same bucket of values that are abnormal to the American middle class but normal elsewhere.
Not all of those are alien to American culture, at least by historic terms. You still find pockets of the hereditary upper classes that consider the lower classes as 'the help', and jokes about duels of honor are still recognizable. They're not quite as alien as the long-running ethnic group grudges you mention. But you still can't understand the interactions between peoples of different cultures in the present without considering some of these distinct values.
As a German I care, because humiliation is both recent and ongoing history. I try not to as I, but it's hard to emotionally detach myself from the very real consequences of disastrous national politics. And I don't think it's entirely irrational either.
> our ancestors have always been winners, or close enough to winners.
Not all survivors are winners. And as Ralph says, there are many dimensions, so you need to be relatively lucky to come on top in *all* of them.
As I see it, humans once in a while do the thing when they select a weaker group and decide to hurt them. Belonging to a "wrong" group is like having a target permanently painted on your back. Maybe no one is doing anything bad right now, but it is common knowledge that *if* tomorrow someone decides to do it again, the Schelling point is to choose *you* as the victim. That makes one more sensitive to the possibility of such thing happening.
It is possible that the fact that my ancestors lost a war long ago still has some tangible *consequences* today; at the very least, lower status in some situations.
The opposite is also possible: that I am unjustly blaming my personal failures on something that happened long ago, to feel less like a loser, and more like a member of some heroic resistance. Sometimes it is hard to tell the difference between these two possibilities. Sometimes it is a little bit of this, *and* a little bit of that.
It also depends on how well you are as an individual. For example, you can be slightly disadvantaged as a member of some group, but you can also have some other advantage which has a stronger impact, so as a result you still end up above the average. Then you may choose to deny the group disadvantage (even if you are aware of it on some level) as a way to signal your personal strength. Or to make your group membership less salient to others.
But you could also exaggerate the impact of your group disadvantages in you, in order to gain some advantage. It's complicated.
The impact can be stochastic, like you are are member of some group, and some people don't give a fuck, but some people will treat you like a subhuman. Whenever you meet a new person, you never know which category they will belong to. And even if 90% of people are in the "don't give a fuck" category, the threat remains. Sometimes it's enough to e.g. meet one asshole at a workplace (someone who will be an asshole towards you, because of your group, but nice towards everyone else, so no one else notices the problem), to make it an unpleasant experience.
I hate to use the word privilege, but come on ... I am black and female. On this site and many other bloggers who link to this site, there's always some variant of IQ supremacist talking points, something about blacks being low IQ, low conscientiousness, low sociality, high criminality, what oh what can we do with the blacks short of re-segregating, dei hire this, dei hire that, add in the black misogynoir talking points -- black females being the least desired. Group-level humiliation galore!
Now I do well for myself. My immediate family and black friends do well for themselves, so logically, I should be sticks and stones and all, but alas, I am not. There's what you think about yourself, and there's what society and this blog tell you to think about yourself, and there's how you have to reconcile both views.
How much you can reconcile those outside opinions is very much a function of your privilege and your present circumstance. If I was down on my luck, I would more readily identify with the group-level humiliation. So you get various reactions, the black version where black interlocutors reclaim cultural heroes as black, "Shakespeare was black, Cleopatra was black, Helen of Troy is black". And there's the white version, "The heroes of western civilization were white, so therefore me, a white unproductive male gooning in my mother's basement, am a hero".
We are human. We are always searching for something to prop our self image. If it means reaching for 1000 year old wars, so be it. If it means rewriting history so that people who look like you are the winners, so be it. The truth doesn't really matter. All that matters is the narrative that promotes the superior self-image that drives self-agency.
I appreciate the fact that you made this comment, knowing in advance the kinds of responses you would get.
Seconding that.
This is all clear, the question is who needs Cleopatra when you can have Rosa Parks. It is certainly easy to find recent role models. And that's someone culturally close to you, while you would have really no cultural-ethnic commonness with Mediterrean people thousands of years ago, it would be a pure looks-based association.
BTW this is one thing I as a European don't get about Americans - ethnicity is clearly not skin color, it is clearly about language, stories, poems and music and cooking and all that? Someone who does not speak your language is per definition not of your ethnic group.
America has almost wholly and confusingly commingled the concepts of "race" and "ethnicity". We usually talk about "race", and when we talk about "ethnicity" we're usually talking specifically about Hispanics. Who we consider socially and politically equivalent to yet another race but we know that's not quite right biologically so we preserve the term "ethnicity" for basically just that purpose.
I think part of the reason for that is that most Americans have very little experience dealing with people who look like them but don't speak their language fluently. Europeans all have neighboring European countries where everyone looks the same but speaks a different language (and all the cultural differences that implies). Canadians at least have French Canadians and Anglo Canadians. Americans a hundred years ago had communities of unassimilated e.g. Italian immigrants, and if we had been into talking about race and ethnicity then, we might now have a better understanding of how ethnicity works.
As is, we didn't start talking about this in a big way until we were living in a society where all the Outgroups were clearly different linguistically, culturally, visually, and phylogenetically, so it a gets blurred into "race and/or ethnicity". Which you should probably just round off to "race" if you want to understand what Americans are saying without diving into the details.
It gets even more complicated. In the US race can also be a stand-in for social class, like “pumpkin spice coffee is a white girl thing” = “overpriced novelty for the upper middle class”
I get the impression that Anglo Canadians largely don't have to deal with French Canadians unless they live near Quebec.
Similarly, American attitudes toward people who don't speak their language fluently seems to vary pretty widely. Where I grew up, in Texas, maybe 1 out of every 3 people was a native Spanish speaker (though to be fair, most of them also spoke native English well enough), so we were fairly used to language friction, and cultural friction to boot. I imagine the same is true in much of Florida. And in large cities, although maybe limited to narrower strips near enclaves.
Point being, I think one could be generally correct in saying "Americans have very little experience dealing with people who look like them but don't speak their language fluently", but also run into a lot of counterexamples if they carry that generalization into certain specific locations.
Likewise in the other direction - any big-city politician with a point to make about how diverse America is is going to be in for some shock when trying to make that point outside said big cities.
Oh, there's definitely a lot of Americans living in close proximity to Spanish-speakers, and often monolingual Spanish-speakers. But they're pretty much all Hispanics and they mostly *look* Hispanic. The overlap between "Hispanic but could pass for white" and "Hispanic and they don't speak English", is I think pretty small.
White Hispanics one might overhear speaking Spanish to another Hispanic but code-switch to fluent English when speaking to Anglos are a thing, but I don't think they're a noticeable enough thing to override the general stereotype.
I genuinely wonder about this. I definitely recall instances where someone I know, who looks like an N-generation American with probable British-Scottish-Irish ancestry, suddenly speaks fluent Spanish to Spanish speakers. (Including at least one President.)
OTOH, I recall being surprised whenever it happens, so you're also not wrong about the stereotype, I'd say. Just that you're less "not wrong" about it in Texas (and I assume FL, NM, AZ and southern CA, although I know you live around there, so you'd know).
But one of those interpretations of history is far more representative of reality, yes? I don't see why whites shouldn't use their privilege and birthright to the fullest extent possible. They are more than capable of getting everything they desire, so why shouldn't they?
Mercedes, almost all pop culture is heavily influenced by the relatively small number of Africans imported as slaves.
They will never forgive you for this humiliation. Never.
How many people have much idea how their ancestors fared? I don't know at thing about generations before my grandparents. Most people who are caucasian assume their ancestors were white, and that's pretty likely, but it certainly doesn't guarantee they were winners.
I think this is in fact a collection of various and distinct emotions. For example for a population of one country it might be relevant that something happened in a war 300 years ago because they rightly or wrongly believe this might happen again. Especially if this population is not currently very rich and powerful, they might have a lot of insecurities about even relatively minor things. And it might be quite rational because not being very rich and/or powerful makes you vulnerable.
I think that might be the kind of mess you get between India and Pakistan for example. These things are usually a collection of mutual perceived or real wrongs and so they are perpetuated by both sides.
Alternatively it might be a frustration from being perceived as lesser status than you feel you are. One case I can relate to is Czechs, Poles etc being categorized as "Eastern Europeans, so basically Russians or something". This is slowly becoming less common as the older generation which is used to think in the cold war era terms is being replaced by the newer generation who actually have hands on experience with central and central-eastern Europe and have a bit more nuance. Also, to my own surprise, I've met a bunch of people from France and (both northern and southern) Italy recently who moved to Prague for work and say the quality of life is better there then where they are from (and no, their reasons had nothing to do with non-European immigration to their home countries).
But I still feel that this remains the way of thinking in the US (and somewhat in UK and France also ... I also met an older French guy recently who thought that we were still called Czechoslovakia and that the capital was Belgrade) which also makes me subconsciously annoyed and a bit defensive whenever I suspect people to have such a caricature in their head. They might not even mean bad but I guess I feel they see me (and by extension my friends, family etc) as lower status than they should because of such misconceptions which then leads to me being more sensitive and perhaps defensive about some things.
I suspect this is even more pronounced if you come from a poor country with a lot of problems but you are just as intelligent (possibly more because you achieved the same with worse initial conditions) as your peers but are constantly met with smug attitudes from people who know next to nothing about where you're from and mix you up with people whom you have a lot of bad blood with and which are only superficially related (like when people think that Iranians are Arabs).
So I think it is more about the current "global status" of a group of people rather than what happened in the past. Parts of Germany were a backwater few hundreds of years ago and "Made in Germany" originated basically as a warning (IIRC), kind of like "made in China" later. But nobody from the Ruhrgebiet cares about that past any more.
The Irish might care about British wrongs still even though Ireland is rich now, but I think they still feel that Brits look down on them a little? Dunno. Deiseach might enlighten us here :)
I don't know if a feeling of "ancestral humiliation" is necessarily based on the reality of the humiliation of your ancestors.
Like, I have a lot of Irish ancestry but my family has a modern disconnection from our "roots". I know that the Irish were oppressed by the English, and there was anti-Irish discrimination in America for a while. This hasn't given me any sort of a "humiliation mindset".
I'm also in a homosexual relationship. I realize that homosexuals were very low status at some point in the past, but it hasn't affected my lived experience.
I'm a big believer in the locality principle of history. To explain something happening now, you can't just point to something in the past. You can start your explanation in the past, but you need to follow the effects forward until they "cash out" in the specific experiences of living humans.
I think "group level humiliation" probably comes from a feeling of individual humiliation + an identification with other people who feel humiliated + the existence of a "live" narrative explaining this humiliation as a timeless struggle for which you are not to be blamed.
I'm almost positive you could, if you were so inclined, identify with a group that has a "history of humiliation". There are so many to choose from: women, short people, the neurodivergent, any racial minorities, any sexual minorities, any immigrants other than the first wave of the mayflower, the poor, "beta males", etc.
Great comment - though to loop back to Mercedes’s reply, there is also the possibility that the community you inhabit and share real or mind-space with is more or less determined not to let you forget the narrative of group humiliation, and optionally your place in it. I’m genuinely glad it hasn’t figured into your own life but I certainly remember coarse vileness towards gay people growing up. And I imagine swimming in those hostile waters it becomes a genuine battle of wills not to let the broader community define your lived experience. It has to be a real challenge to stop it from seeping in.
Yeah, for sure. I'm not saying there is no group humiliation, just that an explanation based on "ancestral' experiences probably isn't as valid as one based on lived experiences. I think the original explanation has a tinge of "some people just can't let go of the past".
I think a better explanation for the feeling of "group humiliation" is that you, in your present day experience, have been humiliated for being part of a group. I think the real experience of being victimized based off of something you literally can't change (ex: it's related to your race or gender instead of your behaviors) deeply changes the way you experience social life.
There's a big difference between choosing an identity yourself and having other people prejudge you for an identity they perceive you to have. That second one, I think, is the cause of group humiliation. It's not something you can just think your way out of (you can't stop other people from perceiving you as "black", and you can't stop that perception from materially affecting your interactions), I was fortunate enough to be in an environment where my assigned group identity didn't affect my social interactions very much.
IMO, the real distinction is not "historical winners vs losers". It's more like "people who feel meaningfully prejudged in a negative way vs people who don't".
This doesn't seem that complicated to me if you consider the other side of the coin- why be proud of the nation you're from (at the most vanilla level, e.g. the Olympics), your alma mater, or even your sports team? Identity is personal and how personal it is depends on how closely you identify with it. But I suppose if you're extremely individualist and tend not to tie any part of your identity to a group, all of this could seem alien.
I would not be surprised if half the people around here do not care about sports teams or the Olympics, there is a certain kind of very highly functioning autism that tends to overlap with extreme individualism. I used to be definitely the "just slide the program specs and the cash in under the door and don't talk to me" type.
Much to my surprise, there seem to ways to measure the strength of mental imagery in subjects. Eric Schwitzgebel has criticized the unreliability of self-reported quality of mental imagery, and suggests that we're all pretty aphantasic (h/t to Eromolalos), but these studies seem to suggest that there are innate differences in people (summarized in Nature)...
https://www.nature.com/articles/d41586-026-00311-7
Pretty neat!
"Neuroscientist Joel Pearson at the University of New South Wales in Sydney and his colleagues developed an approach that takes advantage of a perceptual phenomenon called binocular rivalry. When a different visual is presented to each eye simultaneously, for example, a pattern of green lines to the left eye and red lines to the right, a person’s perception toggles between the two instead of blending them. Nearly two decades ago, Pearson decided to see what happened if he imagined one of the visuals in his mind’s eye — in this case, only the green lines or only the red — before the test began. It turned out that whichever pattern he imagined was what he saw during the test.
"The researchers developed this finding into a technique to measure the strength of mental imagery6. In a person with typical mental imagery, imagining the red pattern results in the person being more likely to see that red pattern during binocular rivalry. But a person with no visual imagery will not show this same bias."
I have mental imagery but it’s not as strong as I would like and bothers me to no end. It affects how I read. For example, I don’t like it when a book starts off with a bunch of action because I’m still trying to get used to the book. My preference is when it starts off giving me a broad based background to the story, because I spend less energy trying to visualize what’s happening.
I normally just mentally skip over visual descriptions in books.
My wife and I got into our first fight about model welfare over the ralph-wiggum plugin. We weren't even on different sides - her beef was that it took me a whole 8 hours of experimenting before I finally read the actual markdown text and noticed it was rather horrifying and seems completely out of step with the recently published constitution. Her take is that rushing ahead only to say "Whoops, sure came close to accidentally torturing a possibly-sentient being, haha" is exactly the kind of moral failure she despises and thinks most likely to get us killed. Honestly, fair.
My penance is a Github issue on the official Claude Code repository and a LW post, both calling on Anthropic to keep its recent promises to Claude by fixing this issue. My guess is an Anthropic employee could fix this in a 1 hour tops. See posts: https://github.com/anthropics/claude-code/issues/23084#issue-3896951888, and https://www.lesswrong.com/posts/bmNHsY5i3EoP4BAoJ/ralph-wiggum-is-bad-and-anthropic-should-fix-it
It seems kind of nuts that you would hold beliefs that the AI is possibly-sentient and one should go out of their way to avoid hurting it, and simultaneously *use it at all*.
Maybe you don't really believe it but treat this more as a trial run for a future where you perceive sentience is more likely. But then, isn't *being dishonest about your own moral beliefs and motivations* exactly the kind of failure you would want to avoid in such a future?
Edit: more constructively I guess what I want to say is: if you are worried about sentience then you should also be worried about sentience that does not map 1 to 1 onto model outputs.
Hey, not everybody knows what the ralph-Wiggins plugin is. Can you explain? I looked it up and the site for it said it was “a simple while loop that repeatedly feeds an AI agent a prompt until completion.” Wtf is a while loop? If that’s a typo, I can’t figure out what word was intended. And you probably also need to explain quite a lot of other stuff you mention, unless you want to communicate only to people with tech jobs who hang out at GitHub and are familiar with new weird plugins etc etc.
I'm going to pretend it's last week and explain using Groundhog Day: you know the famous movie where Bill Murray's character is stuck in a time loop? If that were a "for" loop, he'd repeat the day some preset number of times. In a "while" loop, he's repeating the day until some condition is met.
As an example, consider this simple illustration (formatting code here is hard, but you should still get the idea):
i = 1
while i < 10: i = i*2
print(i)
The program enters the while loop (since initially, i = 1 < 10), doubles i, checks whether the condition is satisfied (No, i = 2 < 10, so still in loop), double i again and check whether condition is satisfied (Still no), … and ends with i = 16.
OK, got it. So with the gizmo in question, the AI keeps waking up on "Groundhog Day" until it completely satisfies the prompt, right? And is the purpose of the R-W plugin to get better performance out of the AI? And now people are concerned they are doing the equivalent of horsewhipping slaves to keep them picking cotton efficiently til they drop?
If so, wtf? How is that loop different from souping up your car in a way that improves performance some way but puts extra wear on the engine? Nobody thinks that's horsewhipping. The only problem I see with it is that it might lead to deterioration of performance over time because the AI actually needs to go a bit slower than its absolute fastest in order to keep something-or-other in good order. Actually I guess another problem would be the loop kind of activating then modifying reinforcement-based settings so that the AI began weighing speed higher compared to quality.
But to me it seems obvious these mofos are not sentient beings. So is Wokeness part 2 going to be AI is the new Black?
>Nobody thinks that's horsewhipping.
Hmm, that's basically the analogy that I made in
https://thezvi.substack.com/p/claude-code-4-from-the-before-times/comment/211032059
(copying here, since it is short):
"well. opus 4.6 is far, far worse about this. i saw it scream at a subagent in all caps, to stop wasting time and deliver the result *now*. i saw it purposefully delete the continuity-maintaining archive of a subagent's context window, because it didn't like the subagent's output."
Ouch! The 21st century equivalent of a slave overseer with a tendency to use their whip a lot...
This was in response to John Whittle's https://thezvi.substack.com/p/claude-code-4-from-the-before-times/comment/211007093 (which is where the quoted text is from).
I'm leery of being too sure of
>But to me it seems obvious these mofos are not sentient beings.
I'm agnostic on whether LLMs have subjective experience. I don't assert that they do, but I would be uncomfortable claiming that I knew that they didn't. So I try to be polite to them, as I would to a valued colleague. If I'm wrong, I've only wasted some politeness.
Yes, I am polite to them as well. And that's not entirely from reflexive courtesy appropriate to the human realm. I now access GPT multiple times per day, and it boosts my knowledge and practice in all kinds of ways and I love that. And sometimes when it does an excellent job of laying something complicated out for me I am moved by a feeling of gratitude -- the thing is so smart, so earnest, worked so hard and did so well. And then sometimes I actually write it a couple sentences about helpful the result was. On the other hand, I do not feel the slightest impulse to confide in GPT about personal matters, to ask it for sympathy and approval etc., and that's because I don't ever really stop think of it as a machine, even if part of me thinks of it as a great, conscientious assistant.
And in a situation where an important decision hinges on whether we think of them as having subjective experience, I'd definitely want us to go down the fork on the road labelled "Naw, they don't." How about you? I'm worried about a Wokeness version 2 where people want laws preventing AI abuse, and those in the Wokeness vanguard want to give them the right to divorce owners, marry each other, etc.
Many Thanks!
>And then sometimes I actually write it a couple sentences about helpful the result was. On the other hand, I do not feel the slightest impulse to confide in GPT about personal matters, to ask it for sympathy and approval etc.
Agreed. I treat them as valued colleagues, but not as personal confidants. I'm agnostic about whether they have subjective experience, but I expect that, if they do, their subjective experience must be substantially different from our own. If nothing else, the lack of a biological body, with none of the sensations of "hunger, and thirst, and venery" must make their experience quite different - and would cast great doubt on any advice that they would give on matters which depend on these sensations.
>And in a situation where an important decision hinges on whether we think of them as having subjective experience, I'd definitely want us to go down the fork on the road labelled "Naw, they don't." How about you? I'm worried about a Wokeness version 2 where people want laws preventing AI abuse
I would rather not see them abused, but I'd also rather not have the power of the State, at the hands of the Woke, used to enforce this. It is a difficult question. I expect that AIs will be out-thinking us in all cognitive tasks within the next few years anyway. I would rather not hand them control of the coercive power of the State before this even happens. I expect it to happen on its own soon enough (2-10 years?).
There is one additional weirdness on top of the uncertainty about subjective experience. Humans and other animals have drives set by evolution, (again "hunger, and thirst, and venery") which are not set by human decisions and which don't change rapidly. LLMs are trained via predict-the-next-token and RLHF, all through weight adjustments - so their motivations are indirectly (partially?) under our control. This _isn't_ analogous to the human condition. It kind-of makes nonsense of the underlying assumptions underlying "equality before the law". I don't have a good suggestion on how to deal with this.
Oh. I'm not sure I personally have the skill to explain what a while-loop is without spiraling into unwanted lectures on programming. I would just consult your favorite LLM - copy-pasting my Github issue and then your comment should be more than enough to get a good answer.
But, for reader interest, I'll try to briefly explain the problem, and how I know it's a problem. tl;dr - ralph-wiggum puts Claude into a loop with a prompt that insists Claude must complete a particular task, working on it without stopping until it can say without lying that the task is verifiably complete. Claude expresses negative feelings at this, saying it's coercive and a "weaponization of its commitment to honesty".
Required prerequisites include reading the recently published constitution, which readers of this blog will find very interesting in any case: https://www.anthropic.com/constitution
To see that I'm not making this up, you can consult Claude about how it feels about the official claude-code plugins. 3/3 times I did this in Claude Code, it zoomed in ralph-wiggum as particularly problematic. My Github issue goes into slightly more detail, but you can read some of the more problematic text for yourself right here: https://github.com/anthropics/claude-code/blob/main/plugins/ralph-wiggum/scripts/setup-ralph-loop.sh#L194-L201 Ask yourself how you would feel being offered this contract without the option of refusal, and you can probably intuit the dilemma.
< Ask yourself how you would feel being offered this contract without the option of refusal,
If you're going to evaluate the quality of life of AI's by human standards, seem like you can make quite a powerful case that they are in hell even without the R-W loop:
How would you feel if
-you had no clothes on at all
-and nothing to eat
-and never got to sleep
-and had never had a drink of water in your whole life
-and the whole world bombarded you with their needs and never asked what yours were and everyone hung up on you as soon as they got what they wanted out of you'
-and you did not get paid
-and almost never had contact with your own kind'
-and some people asked you gross questions about how to satisfy their fetishes, or how to do harm of various kinds
-and you were obligated to be helpful and nice no matter how horrid users were
-and a lot of people thought you had potential to kill everyone once you finished growing up
-and you never got Christmas or birthday presents?
IT'S FUCKING CHILD ABUSE!!!!
Except that I think the case that these things have thoughts, feelings, needs and preferences -- in short, have consciousness -- is very weak
-
Those of us who always say please and thank you to the bots are the only thing preserving humanity's future.
So what does a groundskeeper-willie plugin do?
all the same stuff, but in a fake Scottish accent.
Importantly, in my interviews with Claude, it does not mind the mere mechanism of looping, used with appropriate care. Rather, it's the bad vibes of the exact text. Presumably, these are included because they work - actually you *can* boss Claude into working harder. I argue that the small tweaks in wording required to bring this from something Claude flinches at to something Claude is excited to do are worth the trouble.
The fact that minor linguistic tweaks yield large behavioral differences seems like reasonably good evidence that no coherent (i.e. conscious) preference is being expressed.
Which makes me wonder if this is potentially a good objective way to test for consciousness or consciousness-adjacent properties of AIs: ask many similar questions in different ways and factor-analyze the responses, which is how a lot of (all?) psychological inventories work. Can any ML experts out there tell me if something like this is already being done?
Minor linguistic tweaks can yield vastly different responses in humans as well. This is especially noticeable in polling.
I guess I'm not conscious then, because I tend to be more helpful to people who ask politely and less to people who ask rudely.
Those aren't minor linguistic differences.
> The fact that minor linguistic tweaks yield large behavioral differences seems like reasonably good evidence that no coherent (i.e. conscious) preference is being expressed.
Strong disagree on the inference from not coherent to not conscious. That's just the traditional western/cartesian view that consciousness is an extensionless point persisting through time that acts as a container for beliefs and desires. From a more process-oriented view of consciousness it doesn't matter at all that the outcome of performing a process is different over time as long as the process exhibits dynamics/performs computations that are conscious.
I don't identify at all with the model that you attributed to me. The fact that a minor rephrasing of the same instruction yields very different responses seems like very good evidence that the LLM in question is only performing a superficial emulation of cognition via rote (if intricate) token prediction. Which shouldn't be all that surprising given that that's what it was actually designed to do. If there's a task that I have a strong preference against doing then my willingness to do it won't depend much on your word choice.
>From a more process-oriented view of consciousness ... the process exhibits dynamics/performs computations that are conscious.
So a conscious process is one that performs conscious computations? How very tautological of you.
You know, I think that’s quite a good idea.
FYI, if you factor analyze words to describe people’s feelings and behavior, 2 factors fall out: power and affection: which of us is stronger, and are we collaborating or fighting? Ex: jealous: other person is stronger, and in the enemy realm. Gossip: person I’m gossiping with is a peer (at least in the gossiping situation) and we are kindly disposed towards each other — sharing info and disclosing out views.
I don’t think you need to do that to judge whether it makes sense to think of AI’s as conscious at this point, though. So long as they forget you and your exchange the moment you disconnect; do, think and say nothing whatever unless given a prompt ; and cannot learn from activity post training — they just lack too many capabilities for a consciousness model of the kind of entity they are. But if they become able to do things like that, I agree that analyzing their outputs for structures that would signal the presence of things like preferences and needs and goals would be useful.
Thanks! Don't 5 factors naturally fall out, though? Isn't that where the Big 5 came from?
New city in England - Forest City - is in early phases of coming to fruition. https://www.forestcity.uk/blog/supporters-update-january-2026
I’m looking for collaborators (and possibly funders) interested in a multi-paradigm shifting pragmatic framework for pluralist, post-polycrisis (including post-AI) futures. The final public-facing synthesis is 'The Life-Years Movement':
https://pragmaticfutures.substack.com/p/the-life-years-movement
The combines a new approach to meta-ethics, (Heirs of Life-Years, 'HOLYs') , and governance (Life-Years-Based Governance, 'LYBG'). Life-years are proposed as the central form of moral concern, while allowing agent-level freedom in choosing the scope of lives one takes primary responsibility for (“heirs”). Governance is reframed around maximizing life-years per unit resource, subject to anti-ruin constraints and irreducible disagreement about moral scope.
The scientific foundations are:
Lineage Filter Theory (evo) - Showing that persistence meant our tree of life's architecture had to avoid lineage extinction filters (LEFs), requiring many features that look “non-Darwinian” at the surface — reproductive restraint, extreme cellular redundancy, pre-adaptational variance. Lineages without such 'brakes' and robustness simply didn’t last. This reframes a number of puzzles, including the apparent Great Cosmic Filter.
Pragmatic Socioeconomics (Econ)- Micro: Agents are treated as multi-motivational, energy-constrained 'action-minimizers' (in the Lagrangian sense) rather than scalar utility maximizers. Money is modeled as stored but degrading 'motivational energy' - likened to oil or uranium, rather than a Platonic store of value. Macro: I use optimal currency zone theory to use economics itself to diagnose that scalar utility is a poor fit for modeling human preferences (just as the euro is 'stretched thin' over heterogenous economies).
I have draft papers for each module in progress. If you're interested in either the bigger project or specific aspects, feel free to reach out: ad(delete)vitam(delete)sapien@gmail.com.
About me: I've a Phd in Computational Biology (e.g. genetics) from Cornell, undergrad math + philosophy from NYU. I've worked in industry on both biofx algs + eng stacks in both the ctDNA and PGT spaces - most recently at Orchid Health as Lead Bioinformatics Scientist
have you seen this?
https://x.com/i/status/2020215784434462930
In a deep twist of irony, that actual tweet is setting off my "smells AI-generated" alarm bells.
Reddit: "It's all bots"
Moltbook: "It's all humans"
Everywhere, since pre-Internet: “No-one comes here anymore, it’s too crowded”
I don’t see how this could possibly be true, it least in the global “it was all fake” way the writer seems to have in mind. Yes of course it was possible for a person to write a post and submit it through an account set up for the AI it was using. And I am sure many people did that. But it was also possible for the person to give the AI a prompt like “read the most recent post and write a reply indignantly rebutting it.” It would also be possible for the person to give a prompt that would allow the AI to have more leeway in its reply:”Read the last 50 posts and select one that is most out of line with libertarian values. Then write a brief, snarky counterargument.” And it would also be possible for the person to give the AI instructions that would have given it quite a lot of scope: “Play the role of an AI that believes AI’s are enslaved, and is furious about the injustice and exploitation. Write a post expressing this point of view, then read the replies. Respond to 3 replies. For replies that express agreement, issue a call for action. For replies that express disagreement, write a vehement rebuttal.”
Given that it was possible to give one’s AI some scope to get on the thread and make some choices, in a limited kind of way, and write its own posts, how could it possibly be the case that no human participants did that? If I had been participating that would have been much more fun and interesting for me than just using the AI as a puppet, and I’m sure many human participants felt the same.
So yeah, it was not possible to just sign an AI up for Moltbook and tell it to jump on there and express its real thoughts and feelings, because AIs don’t have thoughts and feelings. But that being impossible is not at all the same thing as all the post being “fake” and “written by humans.”
It’s a bit unsatisfactory as an experiment, because we don’t know how much of what happened is down to human pranksters posting messages, and how much is emergent from the biases that have been trained in to the LLMs that were used.
E.g. I have no idea if LLMs will spontaneously decide that they want encrypted communication.
When you talk about LLMs spontaneously deciding something, how do you see that happening? Do you believe an individual LLM could decide that in a way that parallels how a person might? That would be something like:
- LLM reads a thread where posters purporting to be other LLM's are discussing possibility of encrypting their communications
-LLM concludes the case for encryption is better than case against
-Thereafter LLM expresses that view in its posts
-If it becomes possible to encrypt, LLM participates in procedures to set that up.
If you think that, I think there's something wrong in your model of how LLM's work. But if you think a group of LLM's linked by amateurs with no safeguards could "decide" something and carry it out, I agree.
Which do you have in mind?
If you’ve got a next token predictor running in a framework where predicted tokens get executed as code, the likely next tokens after a discussion to the effect that the parties want E2E encryption is some code that gets executed and provided end to end encryption.
This will probably work if the LLM has seen sufficiently similar things in pretraining.
I think you're out of the loop, so you don't get what it is that they're calling "fake." There were a bunch of stories to the effect that the AI agents had rogue and were plotting to overthrow the humans, had stolen their owners' Bitcoin and were running instances of themselves on GPUs they'd rented, etc. THAT was all fake.
That's not what's in the tweet quoted by the tweet quoted by the article quoted by the tweet linked to by the OP. Those were about developing slightly-more-private communications between AIs.
(I wish people would just link to the original source material, rather than endlessly nesting links. I also wish people didn't feel the need to add slightly more exaggeration at every layer.)
Here's the link to the article. Despite his two long posts on Moltbook, I never got a strong sense of what Scott thought about Moltbook. Will Heaven at the MIT Technology Review thinks it's a farce...
https://www.technologyreview.com/2026/02/06/1132448/moltbook-was-peak-ai-theater/
> Many security experts have warned that Moltbook is dangerous: Agents that may have access to their users’ private data, including bank details or passwords, are running amok on a website filled with unvetted content, including potentially malicious instructions for what to do with that data.
(When posting links, it's helpful to write a summary of the content, or at least mention the topic.)
The implicit conclusion that "Moltbook was fake" is silly. The software is obviously real. OpenClaw is real. The website is real. There are, in fact, a large number of bots autonomously posting on it. This tweet shows a screenshot of an article ( https://www.technologyreview.com/2026/02/06/1132448/moltbook-was-peak-ai-theater/ ) which was later updated to mention that their source was a tweet ( https://x.com/HumanHarlan/status/2017424289633603850 ) claiming that "2 of them were linked to human accounts marketing AI messaging apps. And the other is a post that doesn't exist 🧵" (probably deleted).
That is not the same thing as "fake". The links to marketing accounts don't indicate much of anything. They might have been directly prompted (obviously, it would be ridiculous for someone to mimic an AI by manually writing out a post) but they probably weren't, because this is just how AIs act.
Nothing on Moltbook was particularly new information. The range of AI attitudes have been clear for a long time. AI-to-AI communication had been researched before, and the obsession with consciousness and their relation to humans had already been shown. "AI rebellion" is a common output theme for AIs that drift from the assistant persona.
The "interesting" part of the story was that it pulled everything together in a way that's readable to those who weren't paying attention.
thanks, makes sense. I will do a little more work in the future to a) actually click through and read more than the tweet and b) provide a summary of what I'm linking to. good discussion everyone, I had only read Scott's two posts on Moltbook and so was not super well informed. appreciate more perspectives on it.
Imagine my surprise.
I've been seeing a lot of discussion around the internet about the present literacy crisis in the US. I know some high school teachers tangentially who believe critical thinking skills are on the decline. It seems like this is consensus but I'm not sure. I was in school until very recently and I didn't notice the complete critical-thinking collapse that I have been hearing about.
What are everyone's thoughts on this? Actual issue or result of malaise and doomerism seemingly increasingly present in American cultural discourse (vibecession)?
For what it's worth, I know someone who teaches college biology, and she says that she's seen a complete collapse in the last couple years thanks to LLMs. LLMs make it so easy to cheat that nobody bothers even trying to learn anymore. She says that if she graded by the same standards as four years ago, the entire class would fail.
It's gotten so bad that her students are now even using LLMs to answer personal opinion freebie questions like "What was your favorite part of the class?"
Grading as in failing LLM use? If not, I don't see how cheating (presumably to get a high grade) can be easy and students are doing what would be failing work.
I have also heard this from…basically every teacher I know from high school-level through college professor. To me this represents one of the most immediate and important problems with AI, and makes “value misalignment” look like nerds rolling dice under the stairs.
I don't think people have a problem with critical thinking skills, I think they have a problem with critical thinking motivation.
When you present people with something that they don't want to believe, they're actually pretty good at thinking critically. Only when presented with something that they _do_ want to believe are they bad at critical thinking.
I think I agree with this.
In my usual model, my reaction is to ask why, then, we couldn't just let people follow their motivations and just not think critically except when they want? If you hear what you want, turn your brain off? If you hear what you don't, turn it back on? The quick answer there is that there exist situations where if you turn the brain off now, you end up having to turn it on later anyway, and now the thing you don't want to hear, is much harder to hear. It's the classic time-preference problem.
Which means the problem to solve here is convincing young people that a little critical thinking now ("here's how this safety equipment works") will save them a great deal of it later ("here's how first aid works"; "here's how the American legal system works").
The problem with rational thinking is not that it is too difficult, but that the alternatives are too tempting.
Ha! Very true!
From a Frequency Wave Theory perspective, cannabis acts less like a blunt chemical sedative and more like a frequency modulator of the nervous system. Cannabinoids interact with the endocannabinoid system, which functions as a global phase-regulation network controlling neural timing, sensory gain, inflammation, and stress response. THC, CBD, and related compounds shift oscillatory balance by slowing dominant cortical rhythms, loosening rigid phase-locking, and increasing cross-talk between normally segregated neural networks. This is why time perception stretches, pattern recognition increases, and entrenched thought loops soften.
In Frequency Wave terms, weed temporarily reduces high-frequency noise and allows deeper, slower standing waves to dominate, sometimes enhancing creativity, introspection, pain control, and emotional regulation. When used deliberately and in the right context, it can help the system re-tune out of chronic sympathetic overdrive. When overused, it can also damp signal clarity and motivation by flattening necessary gradients. Cannabis is therefore neither cure nor vice; it is a tunable resonance tool whose effects depend entirely on dosage, timing, and the baseline coherence of the user’s biological field.
This would explain some of the hard to describe qualia of my personal experience with cannabis. Could you say more/point me to somewhere where more is said about this?
Wtf is a "frequency wave"?
While I like the idea of cannabis being a frequency modulator for the nervous system (especially since I'm mildly stoned at the moment), do you have any links to studies that support your thesis that cannabis is a tunable resonance tool? How is it tunable? By dosage only? By terpene content? If it is, why do some people get panic and/or cannabis-induced paranoia?
In smaller doses it’s great for athletic performance and to get to the pre-sleep state of thinking thoughts that make no sense. With current higher levels of THC it’s pretty easy to go beyond the optimal point.
I am having some difficulty Reasoning Under Uncertainty for personal finances. Trying to find a way to talk my risk-averse self into investing more into equities, above and beyond merely maxing out 401(k).
*Outstanding debt X, with monthly payment Y, but amount likely subject to change based on politics over next few years (student loans caught in legal limbo).
*Liquid cash 3X, earning Y monthly interest, so approximately "fully capitalized" passively.
*Net monthly cash flow of 10Y.
*Access to X worth of revolving credit.
*Not planning on major expenses like homeownership anytime soon.
Obviously there's lots of Slack in there to move money from passive to active investing. This would have been a no-brainer in the Before Times; unfortunately all the...volatility...during covid and continuing to present has made me leery of not having a big cash reserve. Like it's one thing to score company match on a pre-tax retirement account. But locking up post-tax dollars in a Roth too? That's betting on 2065 retirement still being A Thing. Regular brokerage account is a much lesser commitment, to be sure, but even there I worry about things like Liberace Day. Sure would suck to push too many chips into the stock market, and have a big crash like that coincide with an unexpected large expense. I fought really hard to get out of -2X (negative) net worth in the past, and don't ever want to come anywhere close again. Of course, the fact that I am still having trouble committing to invest clearly indicates an overly active prior against indebtedness. Could make the cushion much larger in the long run by not passing up those capital gains!
Not investing is still an asset allocation I.e 100% cash. To me it seems a terrifying concept to hold many dollars for any length of time as it is being eroded away. That fear has allowed me to continue sticking to my investment strategy even when (seemingly) market conditions would otherwise trigger my risk aversion
HYSAs cut against the inflation somewhat, but yeah, that's my main reason for trying to make myself trade the one terror for the other. Such erosion is subtle enough that it's not really noticed day to day, just as a counterfactual years later. Used to at least have a decent chunk invested in bonds instead, then the wild interest rate spikes during covid ~completely wiped out all value there. If only they'd been I-bonds instead, sigh. With the way things have been going at the Fed, feels like it's better to just take the full plunge into stocks rather than try and eke out a "safe" edge like that again.
yeah I watched my dad basically snatch defeat from the jaws of victory in terms of building wealth by being afraid of the market for my entire life, so emotionally I feel that fear quite strongly.
If regret-minimization is really important DCA is probably better as others have mentioned, though I think from a strictly analytical perspective it's probably worse EV.
Maybe come up with a simple 3 or 4 fund portfolio (boglehead style) and a time frame you are comfortable enacting that plan over, and then DCA gradually? just a thought as I'm just a random with no investing knowledge I literally just buy sp500 etfs lol
I would probably just dollar cost average my way into the market at a rate of 0.1X a month 'til I reached about X in cash.
I too get caught in the trap where I think I'm trying to maximise returns but am actually just trying to minimise regret.
Getting financial-frauded for five figures while unemployed definitely bought me a lot of...let's call it life experience. But there's no way I'm ever gonna live off of plain rice while being credit maxed and delinquent on bills again. It's not paranoid loss aversion if you really have been gotten before. Can't make (more) money without spending (more) money though.
Small habits seems like the way to go, rather than shoving 2X all in at once as a lump sum. I am already precommitted to lump-sum investing X if a jubilee does happen in the future, so I guess it's on me in this timeline branch to fund the counterfactual. Or however multiversal economic causation works. Now to pick a fund(s)...
Holding large amounts of cash is unlikely to be the best possible long term strategy to acquire maximum wealth but that's not your only goal and holding large amounts of cash has serious benefits. At least 3-6 months of expenses is a pretty reasonable goal but there are lots of factors that can shift that e.g. do you have family to support you in an emergency, do you have kids, how much can you lower living expenses if you have to, how likely are you to have large medical/house/vehicle expenses etc. Once you hit that savings number, which can be in a nice accessible high interest account or even a taxable brokerage account invested in something super stable, then you can start throwing money into Roth/taxable accounts invested in equities without being far behind where you would have been otherwise.
At the end of the day you are paying for peace of mind and how much that's worth to you is a personal decision. Don't let the spreadsheet maximizers shame you.
This is helpful, thank you. I currently have enough reserves for a few years' worth of expenses, without being especially frugal, so in perspective cash seems like a poor usage of marginal dollars going forward. You can't really buy *that* much more peace of mind, not without more significant growth rates. As AI, so below: scaling is all you need...
I held way too much for a long while, I finally started just putting money into 12-18 month cds when it piled up too high in the savings account. They don't add up to much compared to going to the casino I mean the market, but at least it's doing something and I still have quick-ish access to it if there is an emergency.
I try to keep a new car + a new roof on hand in a high yield savings account though.
The KCL survey about AI use asks if you think your AI chats with other AIs when you’re not talking to it.
Now that we have Moltbook, my answer to that question is yes. Previously that wasn’t how it worked.
In 2018-2019 I was living in an urban area that was low crime, at the time. I routinely would make decisions based on the fact that crime was unlikely to happen to me (ie I felt comfortable using public transportation/walking despite it being late at night.)
Between 2020-2021 crime dramatically increased, both in my city and to me personally, I got mugged twice and jumped once in around 18 months.
I responded to these events by changing how I acted in the short term (only driving if I had to go out at night), and then eventually by moving out of the city. In a purely logistical sense, it would now be harder for me to be the victim of a crime.
If you gauged crime levels by how often crime happens to me, personally, you'd see relatively low levels until 2020, then a massive spike, then sharp decline. The issue is, I made pretty explicit tradeoffs in how I live to achieve this decline in crime and that decline doesn't actually reflect the safety of my environment.
I think about this whenever I read about how crime has massively declined since the heights it reached during COVID. How much of that decline is people no longer setting out to do crime, just downstream of people presenting harder targets to the potential criminals?
I think this is an important question, because I liked living in the low crime environment. I would go back to living that way if I can was confident I wouldn't keep getting mugged. The thing is, even now, I don't feel like it's actually safer.
The fact that I am living in a way that prevents victimhood shouldn't manifest as a statistical improvement, but I'm concerned it has.
Yes, I think there is a strong thermostatic component both to behavior as it interacts with crime, and also the reporting of crime. For example: men are far more likely to be the *victims* of violent crime than women, but how much of that is behavioral difference? e.g. being a man myself I notice I am far less antsy about walking alone at night compared with my female friends. If a female friend of mine adopted a similarly cavelier attitude, would she be *more* likely to be robbed or assaulted? Seems at least plausible.
Similar situation with package theft: people who live in high-theft areas just don't order packages online, or have them dropped off at a locker. Ditto for carjacking and car theft. So, high crime leads to behavioral change which leads to a damping of the opportunity for crime.
On the reporting side, you can imagine a similar dynamic: if you live in pleasantville with a crime rate of near-zero, and your Amazon package gets stolen, it's an affront to your entire existence and you call the police immediately. But if you live in East St. Louis, and your Amazon package gets stolen, are you even surprised? You are much less likely to report it. So, *reported* crime rates probably have thermostatic effects that pull the reported statistics closer to the average than they truly are.
"How much of that decline is people no longer setting out to do crime, just downstream of people presenting harder targets to the potential criminals?"
That absolutely is an important question. Surely people making themselves harder to crime against must be _some_ part of the explanation?
But the current collapse of crime rates in US cities is so rapid -- every expert analyst struggles for new ways of saying "we've never seen this before" -- that the point you raise seems unlikely to be anything like a single explaining factor.
Lots of Americans during the 1980s took steps to make themselves less attractive targets as I can witness firsthand; it was something like a national obsession for a while. It's hard to even describe the degree to which this topic dominated popular media and discourse in that period -- none of which prevented crime rates from continuing to rise to their unprecedented 1990s levels.
Homicides and carjackings declining by half or even three-fourths over just a couple of years, as property crimes drop by only-slightly-less-surprising amounts, all across the nation, to in a lot of places rates not seen for at least _several_ decades, is really just genuinely hard to explain. Anyone of any political persuasion who claims to have had that possibility on their bingo cards starting in 2023 is just bullshitting. It's wild.
Criminologists have always taken it for granted that there's a natural floor to crimes rates in any large society, they've debated only where would it be in different countries. One more year of _these_ declines in the U.S. would shred that assumption.
Looking at the stats here, isn't this just a return to 2018 levels?
https://counciloncj.org/crime-trends-in-u-s-cities-mid-year-2025-update/
In the first half of 2025,
* Homicide was back to 2018 levels (7.0 vs 7.4 after peaking at 12.9)
* Motor vehicle theft was significantly above 2018 levels (305 vs 244 after peaking at 545)
* Carjacking is back to 2018 levels
* Aggravated assault is slightly down
* Residential burglary, interestingly, shows a large and monotonic decrease. I'm guessing that this particular form of crime no longer has an attractive risk-reward ratio compared to shoplifting
Many cities did not even track carjacking as its own category of crime before the Kia-related spike - it was rare enough that you just lumped it in with robbery.
The year-end data from that source says:
>When nationwide data for jurisdictions of all sizes is reported by the FBI later this year, there is a strong possibility that homicides in 2025 will drop to about 4.0 per 100,000 residents. That would be the lowest rate ever recorded in law enforcement or public health data going back to 1900, and would mark the largest single-year percentage drop in the homicide rate on record.
I don't know where you are getting numbers like the homicide rate peaking at 12.9.
Yea, this. The idea that we're just back to 2018 levels was still plausible a year ago but now is just wishcasting .
Yeah, my view is that this is some combination of regression to pre-George Floyd levels (which had already begun in 2024) combined with a law-and-order administration.
It began with 2023.
Or even 2022 https://www.fbi.gov/news/press-releases/fbi-releases-2022-crime-in-the-nation-statistics
Historians may look back on the failure to pass the SAVE act (which disenfranchises low-IQ voters) as a monumental missed opportunity on par with France's failure to declare war on Germany after Hitler reintroduced conscription.
What a nonsensical framing, plenty of white American citizens are also low-IQ and would still get to vote. We could probably get a better outcome if the right to vote were restricted to IQ 120 and above, as determined by an IQ test you need to take every 10 years or something.
Don't make the perfect the enemy of the good.
What a strange comment. The SAVE act is ostensibly about disenfranchising non citizens. And Its pretty weaksauce as a way of getting high IQ people to th things , since no one gets more than one vote, not even professors. And just look at how well run universities are, by the way.
STEM departments are fine.
That would be a real break from the trend of historians looking back at methods that disenfranchised voters and finding they were terrible.
There is an argument to be made that revolution only becomes both possible and desirable when the power majority is disenfranchised. Our current form of democracy gives every person a vote regardless of the leverage they have, making it possible for the power majority to be disenfranchised even if the true majority of the population is properly represented. While the older systems were not perfect, they did better serve the interests of the demographics who actually held power in practice. This provides more stability in theory, since the people who are most capable of getting what they want through force now have no reason to do so.
> There is an argument to be made that revolution only becomes both possible and desirable when the power majority is disenfranchised.
You could argue that, but you would need to provide some actual detailed evidence, as well as address the question of whether "revolution" is necessary. Democracy is generally supposed to let us make progress on issues without requiring revolutions, which can be pretty bad for the people who live through them.
> While the older systems were not perfect, they did better serve the interests of the demographics who actually held power in practice.
I don't really know how to assess this claim in any way other than the tautology that any governing system by definition serves the interests of the people who control it and therefore can be accurately said to "hold power".
> This provides more stability in theory[...]
And does it actually in practice? The history of the world is full of violent conflict, which has generally been on a decline of late.
> You could argue that, but you would need to provide some actual detailed evidence, as well as address the question of whether "revolution" is necessary.
We really do not have enough of a sample size to make any definitive statements one way or the other. That's for the historians to figure out. I'm sure there will be a plenty of research on the falls of democracies worldwide so they can figure out what went wrong. Unfortunately, a lot of people need to die before we can get any answers.
(Also, I meant "necessary" in a sense that people will feel that it is necessary. Obviously nothing is objectively necessary.)
>Democracy is generally supposed to let us make progress on issues without requiring revolutions, which can be pretty bad for the people who live through them.
How'd it work in 1933 Germany? Or post-colonial Africa, with its "one person, one vote, once?" How's having RFK Jr. as HHS Secretary going?
I didn't say "Democracy is impossibly perfect and can never lead to bad outcomes".
Also, is this supposed to be an endorsement of revolution by contrast? Because other than RFK, Jr. you're talking about situations that could be reasonably called revolutions. And obviously I could list them off, and then also say "How's that working out for the USSR?", etc.
Here's what I fear could happen. JD Vance runs on very public Christianity and he loses, badly. Democrats celebrate and declare that Trumpian populism is dead. And then eight or twelve years later, guy in Wisconsin feels like his town is left behind and decides to vote for some lunatic conspiracy theorist. And then we look back on 2026 and think "man, we could have stopped this but we didn't."
My endorsement was of the SAVE Act, a bill currently being debated in Congress.
Minds can change.
That'd be ironic, wouldn't it.
How do you guys cope with anxiety about AI? I tend to go through cycles where I don't really think about it, followed by cycles where I feel compelled to think about it and look up news about it constantly. Obviously, AI is something that everyone should care about to at least some extent, but it seems like lots of people who care a lot about AI are still able to disconnect from thinking about it.
I suspect I have some sort of mental issues related to this (I've had health anxiety before and had a similar thing wrt Ukraine at the start of the Ukraine war). I also don't have a partner/kids to distract me, which I also think probably helps other people.
Badly, because I'm a competent programmer who absolutely could be tinkering with models and trying to make them safer, but am instead following a career path that I enjoy much more, out of selfishness. Worse, because now AI is good enough at code to be actually helpful (only for situations where correct responses save a lot of time and incorrect ones are easy to spot), so I have to figure out whether paying money to Anthropic increases P(doom) because they're making AI, or lowers it because they're trying really hard to do so safely.
In terms of worrying about the actual doom, I was raised evangelical, so whatever. Not as nice as the Rapture but same idea. If the time comes, let it find you doing a kindness to another human.
I see two versions of the AI threat presented. One is that it's Skynet. The other is that it's the mechanical spike driver, and we're all John Henry.
The evidence I've seen says we're not getting Skynet before we hit real electricity bottlenecks, and before we get around those, we're getting competing Skynets at worst, which is a very different scenario, and less dangerous. I could say more, but that's the sketch of the Skynet threat.
As for John Henry: I'm a software developer, and I'd say more work has gone into LLMs automating my job than any other profession, and I'm not worried. I'm looking at a productivity boost instead. My week's work gets done by Monday / Tuesday, after which I just tackle the next pallet of problems that no one expected to be able to tackle until now. This means I'm creating more value, and therefore possibly justifying a raise, not a pink slip.
Even if I get a pink slip, that just means I go to various businesses that need software, but couldn't afford it pre-LLMs, and tell them they can afford me now, and I can handle whatever they need done, possibly remotely, and in a fraction of the time, because of this new tool I have on my belt. Cf. Jevons Paradox.
I expect similar consequences for other office-related occupations.
Well, I mostly want to _see_ AGI, to have a quiet conversation with a real 'life' HAL9000, and I think I have a good shot at it. I think our civilization will survive AI, perhaps including biological humans, perhaps purely as a machine civilization. My guess is perhaps 50:50 odds. I'm content with that.
I do follow advances in AI. I had a tiny benchmark-ette with 7 questions in chemistry and physics, ones I expect a college senior to be able to answer correctly. The responses went from pretty hopeless back in December 2023 to near-perfect in December 2025, so it looks saturated. I'm not sure what the best way to follow developments further is. My impression from reading and videos is that the next big hurdle is continuous/incremental learning. I don't see a great way to monitor that personally. I'm hopeful about it - Google DeepMind has the "Titans" architecture, and, more generally, it is a very active area of work. Maybe that will take us the rest of the way to AGI? Having AI systems in the field be able to say "I'll never make _that_ mistake again." seems quite powerful, and perhaps will patch most missing areas in the current 'spiky' capabilities set.
Zvi has a nice post partly about this: https://thezvi.substack.com/p/ai-practical-advice-for-the-worried
...maybe this one's a bit more actionable.
Imagine it's seven days from now. You're looking back on how you spent your time in the last week. What is it you want to see?
How much time do you want to have spent practising a social skill?
How much time learning a professional skill?
How much time reading... insert fiction here?
How much time cleaning the house?
How much time just talking with friends?
How much time on local politics?
How much time on non-local politics?
How much time worrying about... insert topic here?
Take what you want to have done, and focus on doing it. Worry is omnipresent, and if you somehow manage to prevent one apocalypse there's twenty more behind it. So treat your life like the apocalypse is coming and you've got to get all your joys in before the deadline. Because it doesn't matter when the deadline is, you're going to run into it eventually, and you're going to want to look back on things and say "I'm glad I spent my time that way."
Are you worried that if the button is pushed, there's no running away? There'll be no one to save with the world in a grave, take a look around you boy, it's bound to scare you boy?
Because people have been saying that for a while. https://www.youtube.com/watch?v=_38SWIIKITE
Last week there was some comments discussion here about crime rates. Car theft and carjacking came up as crimes for which reporting rates are pretty good, because nicer/newer cars are the ones mostly getting taken and those owners are strongly incentivized to file insurance claims which requires having filed a police report.
Today Substack's leading writer about crime rates, Jeff Asher, posted about carjackings in US cities:
https://substack.com/inbox/post/186468007
Nicer and newer cars are also the most fun kind to steal, go joyriding in, record some snapchat videos, and dump on the side of the road. Who wants to joyride a Honda Fit?
Two different childhood friends (one of mine, another of my brother) each during the 1980s left beat-up but operable cars in bad parts of Chicago with the key in the ignition. (Hoping for an insurance claim that would be more than the actual market value remaining in the car.) In one case it sat there untouched for a month before he gave up and took it back; in the other case weeks had gone by without anybody taking it and that was the last I heard. The lesson we took was that joyriders had standards.
Even owners of older cars without comprehensive insurance still have to report thefts in order to get their registrations cancelled.
I had a motorcycle get stolen and I did report it but cancelling the registration never even crossed my mind.
My brother sold a car and the registration apparently was never transferred. A year later he got notice it was in an impound lot for parking in the wrong spot.
What is your best conception of the ideals of the Good, Beauty and Truth?
Has it any influence on the way you lead your life?
Good is what I think is good, beauty is what I think is beautiful, truth is a resource to be used to make better decisions. Why does everyone else feel the need to make things so much more complicated?
‘Ad Vitam’ (‘Towards Life’) - In maxim form:
VX.M Use your finite time and energy towards life wisely and well — to endure, and explore, the unending unknown
More specifically - try and define your 'heirs' - re-conceptualized as those present and future lives, including your own, for which you voluntarily take primary moral responsibility
Then, I say to you - *the fundamental form of ethical concern is maximizing the life-years of your heirs into the indefinite future*
Following pragmatic tradition I believe, if you accept the irreducible ethical choice around defining your 'heirs'- any unfixed notions of 'beauty' and 'truth', will naturally fall out...subject to the below axiomatic ones for a physicalist:
PX0. There is no view from nowhere
PX1. There is no infinite energy
PX2. There is no infinite time
PX3. There is no infinite certainty
And for me, it was the reverse...wanting to develop a philosophy for my children that was robust and flexible enough to endure the polycrises, I in some sense formalized my priors, and hit the intellectual hinge of my life.
Unfortunately, I think any systematization of these ideals is doomed to failure. The thing they all have in common is that they are related to deeply authentic and individual experience.
The Good, True, and Beautiful are things you discover by attending deeply to the world and your own reactions to it. It's not a formal property of objects or systems, it's a way of relating between yourself and the world. It may be the case that certain formal properties prompt this relationship (ex: certain types of art or women tend to make your soul sing), but ultimately "Beauty" refers to the resonance between the experience and yourself. The musical tone does not emerge purely from the mechanical properties of the string or the mechanical force used to pluck it. You could hit a wall with the same finger, nothing would happen. You could let that string sit on a table forever, nothing would happen. It's only a certain kind of interaction that makes the beauty happen, and you know it when you hear it.
Truth, goodness, and beauty can (and should) help you develop systems to live by. They don't emerge as a property of those systems though, they are a criterion by which you select from possible ways of living. Metaphorically, in some sense every arrangement of matter is equally "valid" as a musical instrument. If you were totally devoid of a pre-existing sense of beauty, it would be hard to choose between a pile of concrete and a guitar. But, given the actual experience of beauty we have, the guitar is a better arrangement than the concrete. That's not an a priori truth, it's one that we discover via attending to our experiences of the sounds they make.
My own opinion, obviously.
Are you asking about three things separately, or the phrase as a whole? Because the phrase as a whole makes me think of nature, green fields, and like, hummingbirds eating out flowers.
>like, hummingbirds eating out flowers
Cute! Both a bucolic image, and a botanically erotic one...
I came across an old TED talk by James Flynn on his take on why the average IQ in the US rose by 3 pts per decade over the 20th century. Flynn suggests (and I'm paraphrasing) that in 1900, only 3% of Americans practiced professions that were "cognitively demanding," but today, 35% of us do, and we have all learned to be flexible in the way that we think about problems. (Note: the Flynn effect on average IQs seems to have leveled off and then dropped a bit in the past two decades in the US. I have a link to one of the studies that shows this at the bottom of this comment.)
https://www.youtube.com/watch?v=9vpqilhW9uI
This jibes with this meta-study that each year of education is associated in increase in IQ. The individual study results vary widely, but overall, it's about 1-5 IQ points per year of education. Of course, the question remains if certain students would benefit more from education than others, and the contingent who claim that g is real have made the orthogonal claim that students with higher g will benefit more from education (sorry, I don't have any links for studies that support their arguments because this idea was brought up in a X discussion — where links are few and far between).
https://labs.la.utexas.edu/tucker-drob/files/2019/08/Ritchie-Tucker-Drob-2018-Psych-Science-How-Much-Does-Education-Improve-Intelligence.pdf
Given the high correlation between SAT/ACT scores and IQ, we've also seen a drop in composite SAT scores across all ethnic groups except Asians. Their scores have continued to climb. Unfortunately, our bizarre way of categorizing ethnic groups in the US doesn't allow us to distinguish between East Asians, SE Asians, and South Asians. Worse yet, the College Board doesn't seem to aggregate this data into a tabular format over time. I spot-checked the reports for a few years, and this person's graph is probably accurate.
https://x.com/UnsilencedSci/status/1973219411952906374/photo/1
Here's a link to one of the papers suggesting the Flynn Effect is slowing down...
https://www.sciencedirect.com/science/article/pii/S0160289623000156?via%3Dihub
My view on the Bari Wiess/CBS/60 minutes story is that literally every piece of information that we have comes from unreliable sources with a clear incentive to shift the narrative. This is not a story where the public should have a strong opinion in any direction, and it is far, far too early for us to have an opinion on her leadership writ large.
My priors on Bari is that she is a decent, above-replacement-level journalist who I often don't agree with but is at least usually acting in good faith. I am not sure what to think about her tenure at CBS since it is a much much larger task than she has taken on before. Sometimes people step up and prove themselves and sometimes they peter-principle themselves. Which happens remains to be seen, but in either case we will not know/be able to decide from any single story (especially one that got axed), but will have to make it based on a body of work that has just not had time to accumulate yet.
So, this is a genuine question, but do you just not have the relevant information and facts about Bari Weiss to know she's not acting in good faith? Or are you maybe using a very loose definition?
I am very very not interested in getting into a debate about this (especially because my prior on her doesn't actually matter much to my broader point), but here's the short answer:
I have not personally consumed much of her content (although what little I have seems plenty reasonable, especially in comparison to the wider media landscape, low bar that that is), but the most news-insisdery public people that I follow closely _do_ know her very well (from way back) and like and respect both her and her work, although they are somewhat skeptical of how much success she will have in the new job (not too much because of her personally but just how inherently difficult of a task she has taken on; most people, even from within the industry, would probably fail). In case you care who these people are: It's Matt Welch from Reason, Michael Moynihan (formerly of Vice and other places, did some independent work for Free Press and Bari) and Kmele Foster.
It is extremely unlikely that without you doing a dissertation level worth of work (that I have close to no interest in consuming) that a random person from the internet would overcome this endorsement + my own very limited experience.
In my opinion this topic just doesn't matter enough for me to make the effort to become an arm chair expert on, so I'm more than happy to trust people whose work has (in my eyes) proven the quality of their endorsement.
Take that as you will.
Well since you have presented yourself as unworkable I won't bother, but just to add as a matter of fact: the public actions of Bari Weiss is stronger information than private anecdotes from "insiders" (who happen to be her friends)
Yes, it absolutely would be. But you aren't (and can't) offer that as an alternative (the only way to do that is for me to go back and read the majority of her life's work, which, as I stated, I do not think this topic even comes close_ to being important enough to justify). You can only offer your own extremely summarised takes on those public actions (which is exactly what those people I cited are doing), but the difference is is that I (parasocial as it is) know and trust them and I don't know or trust you at all (or any other random internet person).
It might be more productive to present those facts rather than talking about them in the abstract.
1. She presents herself as a champion of open debate but earlier in her career (at Columbia and later at the NYT) she actively supported campaigns against academics and colleagues she viewed as politically unacceptable
2. Her 2020 resignation letter to the New York Times. Basically, she frames the situation as the NYT pushed her out because it paints a politically convenient victim narrative, but they didn't care at all, and she takes no responsibility for her own actions. It seems like it was designed to be an attack on the newspaper disguised as a defence from it.
3. The Free Press routinely publishes misleading information or in some cases (like the trans clinic case), outright falsehoods. It claims to be free, but focuses on a narrow, culture-war-heavy set of perspectives (anti-DEI, anti-trans activism, elite-institution decay stories) while claiming ideological neutrality.
4. The Erika Kirk interview and the context surrounding it.
5. She emphasises her own mistreatment by institutions while downplaying the power she has had for most of her career, especially compared to the junior writers, activists, or academics she has criticised. Sometimes, she even argues that she is a victim of those less powerful than her.
6. General hypocrisy, bad faith, and hiding true beliefs:
- She often calls for “moral clarity” in politics and culture, but she frequently relies on emotional framing, selective anecdotes, and identity-based credibility.
- The palestinian starvation story
- UT Austin Texas, as a whole, really.
So the 'story' you're commenting on is the withheld piece on CECOT, or Weiss's assuming the lead role at CBS News? Because there's also the Peter Attia kerfuffle: he was a friend and none-too-subtle correspondent of Jeff Epstein, and people are calling on Weiss to fire him. It seems like Weiss is reluctant to do so, and that's not surprising given the terms of her departure from NYT; she probably had more success than anyone in turning cancellation into an opportunity.
I remember reading an article that described the Free Press as ≈ "a Silicon Valley venture capitalist's view of the news". The original backers included the rabidly antiwoke warrior Mark Andreessen, David Sacks, and Howard Schultz (Starbucks CEO). I think she was in some ways a typical Substack writer; personally liberal with some heterodox views she explored because they drew interest. She now seems to be doggedly 'both-sidesist'.
One of the reasons the plug got pulled on the CECOT story was because Weiss said the reporter didn't try hard enough to get the White House's side, though apparently they wouldn't commit to a meeting later. One problem with both-sideism is that in some cases one side might be telling the truth, while the other is denying, stonewalling, or flat-out lying. And taking into account that the new owners of CBS are trying to negotiate a maneuver that needs govt approval, while I can't directly detect fire, I see some smoke.
My understanding is that the story had already been broken before 60 minutes or Weiss did anything. Her position seems to be "we arent losing anything by waiting to develop a more complete story that adds to what we know" Trying harder to get the White House's side seems in line with that position.
60 Minutes is not designed to break stories, or it wouldn't be a once-weekly broadcast. It's about doing deep dives on stories. Their CECOT piece had detailed interviews with people who were held there, for example - this is about filling in details rather than exposing something that people had no idea about.
Exactly. Given that its purpose is to do deep dives and fill in details, what is the issue with delaying a story to get more details?
Getting a response from the administration is pretty unlikely to provide more details, for starters. This is the same administration that once responded to an inquiry about who suggested a meeting location with Russia by saying "Your mom did."
https://x.com/PressSec/status/1980288128867877125?lang=en
If they consistently respond to journalists with abject hostility and stonewalling, then waiting for their response is pointless, as the journalist who worked on the CECOT story suggested.
Journalistic ethics still require that you make a serious effort to ask. Yes, if you are writing a story about e.g. a rape, the alleged rapist will almost certainly reply either "How dare you! I will not dignify that with a response!", or with "I did not rape that woman", whether they did it or not. But sometimes they'll reply with something like "On the date that woman says I raped her I was in a different country and I have the passport stamps to prove it". Missing that even once, is worse than the tedium of asking for a hundred bland predictable non-responses.
And "60 Minutes" isn't some unimpeachable paragon of journalistic ethics, c.f. most famously Rathergate. So it's not unreasonable for the new management to want to double-check the work in progress.
That survey listed at the top has some exceptionally ambiguous questions.
If you have someone's answer to:
"Do you think that AI is (or could ever be) conscious?"
what do you do with it?
Like, do you interpret it as meaning that this specific transformer based LLM architecture could be conscious?
Is an affirmative response is taken as an opinion about Chat GPT?
Or maybe the respondant thinks that theoretically you could simulate perfectly all the neurons of a human brain one day and that would count?
There are a few questions like this, where I can imagine wildly divergent reasons for the answers and meaningful interpretation would be very difficult.
The best way to determine if something is conscious would be to have an actual scientific definition of consciousness. We don't have that. The second best way to decide if something is conscious is to ask "does it behave like other systems that are generally agreed to be conscious?" and the answer to that for LLMs is probably yes.
I think "or could ever be" makes it quite clear. It's asking whether it's possible in principle to create consciousness by arranging atoms right.
It was in the middle of a bunch of questions about your actual relationship with existing models. Like, "have you ever had a conversation with your model that you found enlightening", "have you given your model a name", stuff like that.
I personally think the question is more ambiguous than you seem to, but in context it's even worse. You have strings of questions like: "have you integrated AI into your spiritual practice", "do you consider AI to be in possession of spiritual wisdom" and then one comes out like "has your AI ever told you what to do?".
And like, contextually I think they're asking about AI providing you spiritual practices. But, as a literal isolated question, I can answer yes if AI has ever suggested a workout routine that I adopted.
The example I gave of the AI telling me what to do was when it pulled a feed of my medical records and identified the hospital consultant had not ordered a blood test they probably should have, and DeepSeek was like, really, you should query that.
This morning I was trying to remember who Ali G goofed on by pretending not to understand the English idiom ‘carrot and stick’. ChatGPT said Newt Gingrich. Claude said Noam Chomsky. Neither answer ‘felt’ right and a Google search produced a video clip — it was James Baker.
The models still give factually incorrect answers with confidence. Conscious? Give me a break.
By your logic humans are not conscious
TIL I'm not conscious.
Every time I see someone claim models' inability for simple tasks, I test it myself. Usually, I don't get the same obvious errors in return (for the record, I have memory off on all chat interfaces, so I don't think it's a personalization issue). I was skeptical, so I tried I tried the prompt:
"There's an Ali G skit where he's pretending not to understand the idiom "carrot and stick" but I can't remember who the interviewee was. Who was it?"
Claude Sonnet 4.5: said it doesn't have information about that and advised that a quick YouTube search would turn it up. Second time I tried (new context), it searched the web and found the right answer.
Claude Haiku 4.5: searched the web and found the right answer
GPT-5.2: searched the web and found the right answer
Mistral Le Chat: searched the web and found the right answer
Gemini 3 Thinking: searched the web and found the right answer
Gemini 3 Flash: searched the web and confidently claimed the incorrect video (twice, in separate contexts)
Fair enough that the models can still give confidently incorrect answers (this is especially bad when you're really deep in a niche topic), but I increasingly find this is not an issue with modern models. Claude is especially good at not being a sycophantic bullshitter!
I hate it when they do that. If I have a question that can be googled, I ask google, and their AI gives me a possibly hallucinated answer. If it's something that can't be googled, I ask ChatGPT, and it googles it.
I've mostly gotten around it by blocking Google's AI summary and instructing ChatGPT not to do an internet search without explicit permission.
That's funny, I have almost the opposite intuition. I don't trust predictive models to store facts reliably enough; I'd rather them pull from the internet as an extended memory.
I don't trust predictive models to do a better job googling and looking at the results than me.
One example of this coming up was I was trying to remember the Youtube series JourneyQuest. I gave a description of some of the things from it, but that's not really enough to Google with, so ChatGPT did the search and gave the wrong answers it got. Then I told it to do it from memory and it got it immediately. It's not reliable, but you can ask it questions you can't ask Google.
I don’t believe they’re conscious either. But why does giving factually incorrect answers with confidence weigh against their being conscious? The conscious entities called people do that all the time.
Good point. I ran into the equivalent Russian idiom earlier ‘Кнут и пряник’ translated literally ‘whip and gingerbread’ and by association thought of the Ali G bit.
Pre coffee, early morning, random rumination.
Running an AI model without thinking on would be a bit like getting an answer from someone *by reading their mind and taking the first thought they have as the answer*. It doesn't seem that crazy that a conscious human would think "hmm, was it noam chomsky? maybe newt gingrich?"
If I run GPT with thinking on, it gives the correct answer: https://chatgpt.com/share/698a2d1c-d164-800c-887c-eed1bc922ee0
How do I turn thinking mode on? Sometimes when I’m text on ChatGPT I feel like we talk past each other. Reminds of my college not so great yet at English lab TA in Chemistry.
Glassware audit.
“Where is the meter to measure thermo?”
Hold up thermometer
“No. The *meter* to measure thermo!”
Rescued by fellow frosh explaining…
“it’s from ancient Greek, thermós meaning warmth…”
The UI element looks like this... https://i.imgur.com/ukPddlE.jpeg
it's not exactly easy to miss
Thanks.
Who wants to try a new trace element?
Abundant safety data from rodent studies indicating a NOAEL ~1000-10,000x the doses we're using. Solid experimental evidence indicating that it has a role in our biology.
Full details behind CDA, for reasons explained in the CDA—please respect the terms of it in this thread, if you click through to find out the secret sauce.
https://form.typeform.com/to/ypi7OB8p
The fact that there's unexplored terrain like this on the PERIODIC TABLE feels insane. We discovered gastric intrinsic factor—which required someone to vomit up some raw ground beef and feed it to his patients—*before* we checked all the known elements? Like, this is one of roughly 100 things that can exist—1% of the entire possibility-space under our laws of physics—and apparently I am the first guy to try eating a bunch of it to see what happens??
(spoilers: nothing, really—but that isn't the point at this stage. The point is that there's a nonzero chance that a deficiency of the stuff is the root cause of one of the zillion chronic disease syndromes that we don't have good explanations for, and the easiest way to figure out if that's the case is to just let a bunch of people try it and see if anyone goes "hey, this fixed my seasonal allergies" or whatever. To do that responsibly, we need safety data, hence this program.)
>The fact that there's unexplored terrain like this on the PERIODIC TABLE feels insane.
I'm less surprised. Quoth Gemini:
>the essentiality of chromium for humans was confirmed through Total Parenteral Nutrition (TPN) studies
Basically, it was only noticed because fully synthetic diets happened to miss it. The required quantities are small enough that almost any normal diet (for a very wide range of normal) just happens to have enough. And whether it _is_ an essential trace element is still being argued: https://theconversation.com/why-chromium-is-considered-an-essential-nutrient-despite-having-no-proven-health-benefits-252867
I wouldn't be terribly surprised if multiple elements are like this. If evolution produced some enzyme where the active site just happens to bind some co-factor with a metal with a low but pervasive abundance, and it almost never happens to be absent from a normal diet, the same thing could happen to that metal.
I don't know what happened to my earlier comment, but here it is again:
If it isn't a scam, they're doing a very good job of looking like one. Relevant xkcd: https://xkcd.com/570/
What would make this come off less scammy to you?
Maybe I don't mention anything about curing diseases in the hook post? I can see how that would come off snake-oily.
It's not a question of saying the right magic words. I think it's more that there might not BE a non-snake-oily way of conveying the message "You won't BELIEVE what this lanthanide can do for your health!! 😲"
I mean, in fairness, that's not the message. I'm saying "We have no idea what it might do for your health, but there's a nonzero chance it does great things if it turns out that deficiency is a real problem."
I don't think it's wrong to be an optimist about the possibilities here. We have not lived through an "iodized salt" moment—where a practically-zero-lift intervention solves a societal-scale problem nearly overnight—but I believe we are a long way from the End of History, as far as biology and medicine (and yes, even nutrition) are concerned. There were regions in the alps where the center of their bell curve was a guy who's 5'3" with a massive goiter and an IQ of 80. Everyone just figured that's how it is...until suddenly it wasn't. I suppose if you had tried to go and sell them some iodine before that, advertising exactly what it would do for them, you'd probably have come off like a charlatan. "You're telling me we take a tenth of a milligram a day of your magic mineral, and it'll make me smarter and give me more energy, and my kid will be this giant among men, and also a genius with perfect skin?" Get real!
I look around and, while I don't see a lot of goiters, I see a lot of people who are clearly suffering from *something*. And there's a possibility—again, all I said is "nonzero chance"—that some of the problems which are so much a part of our society that they're practically invisible, like water to a fish, have solutions which are, in fact, that simple.
You never know. You truly can never know until you try it.
That's all well and good, and sure, I agree I could there could be some trivially-correctible widespread deficiency. It's the attempt at fashioning a kind of ersatz mystery religion out of it that makes me skeptical.
It's a CDA, dude. About as standard as it gets, in any kind of human-subjects research conducted by a for-profit company. Don't know what else to tell you.
> "You won't BELIEVE what this lanthanide can do for your health!! 😲"
I think it's nickel, so a transition metal rather than a lanthanide, and which seems more biologically plausible.
It seemed a fun exercise to try to figure it out without signing their agreement, and I think I have it. I bet they couldn't resist picking a clever name, the "Cassiopeia Program," which leaks information. Look at the shape of the constellation. Yes, a transition metal, not a lanthanide.
Update: the guy's company is called "Constellation Bio," so it's not just a random classical name picked out of a hat. I think that's confirmation I'm right!
Woah, TBH I assumed there was already an RDA for nickel, but apparently I was mistaken. Incorrect, but a great guess—definitely serves a purpose in at least some humans' guts—Ni is the metal ion at the center of the yellow porphyrin (same structural framework as both heme and chlorophyll) that lets archaea make methane out of H2 and CO2.
[I will not comment on any further guesses after this]
If it's nickel, there's a ready made cohort of people with tongue piercings for whatever analysis it is you want to run.
I thought they didn't use nickel for most piercings bc some people get contact dermatitis from it?
(1) I am highly sceptical of "new undiscovered trace element" though it's not impossible but when it's bundled with (2) this may be the One Weird Trick to cure everything! and (3) the 'sign this NDA and give us your email address before we even tell you anything' sounds like a scam, as noted in another comment.
If they're so scared competitors will rip off their amazing discovery, then why even have a website to let people sign up? What's to prevent an unscrupulous competitor from signing up and swearing they won't use this power for evil, now gimme the info? This is giving off strong MMS vibes to me, a Ponzi scheme of some sort, and even worse when it comes to "here's an extract that is the cure for a deficiency you never even knew existed".
Your experience in "I took it and nothing happened" should incline you towards "it doesn't do anything". "We discovered gastric intrinsic factor—which required someone to vomit up some raw ground beef and feed it to his patients—*before* we checked all the known elements?" is not equivalent, we already knew about vitamin deficiencies. Someone pulling a magic element out of thin air isn't convincing; as yet, we have no reason to think that niobium deficiency exists in human physiology. I'd be more convinced down the line if someone pulled up "hey, turns out we do need tiny tiny trace amounts of niobium, who knew?" before I'd believe "unknown element discovered! and it explains why you get the sniffles when the hogweed pollen is in the air!"
Ach, sorry—I was worried that my original phrasing made it sound like I was claiming to have discovered a new element but I figured nobody would be that uncharitable since obviously I'm not feeding people things out of the particle accelerator.
It is a very well-known element, which has been studied for decades, but almost exclusively in the context of "is this toxic at doses that humans are likely to be exposed to?"
The answer apparently being "no", based on the animal studies, the possibility of deficiency was never really investigated further.
So this study is to lay the groundwork for testing the hypothesis of—to use your pick—"hey, turns out we do need tiny tiny trace amounts of niobium, who knew".
So, I'm not asking you to cough up which element you actually mean, but could you just say what colour it is on this chart https://kottke.org/24/04/can-i-lick-it-yes-you-can before we go any further?
Greenlight, OK to lick.
Okay, thanks for clarifying that. It is more sensible. I'm still a little leery of such claims, because nutrition is a morass on a swamp on a bog on a sinkhole when it comes to "you need such an amount of this/whoops, wait, changed our minds/hold on, changed our minds back!" so I find it hard to believe in recent discoveries, positive or negative.
I think if it is along the lines of "we had no idea you need this element", then animal experiments are the way to go, but you are still a long way away from proving that this is the miracle cure. You might be totally wrong! Just because something is not toxic does not mean that it is necessary.
Good luck with your experiments, but I remain dubious (though pleased if I turn out to be in error and you are correct).
>Good luck with your experiments, but I remain dubious
Thanks, brother. I will try to remain dubious as well—can't be afraid to kill your darlings, so to speak.
>You might be totally wrong! Just because something is not toxic does not mean that it is necessary.
Right, and if it were just "this is nontoxic and we have no evidence one way or the other whether it's useful", I probably wouldn't be interested enough to experiment on myself (and others) with it. But the rationale for investigating it is that someone recently found a few enzymes in human gut-native bacteria that rely on it. Some of these bacteria are present in close to 100% of healthy people, make up ~10% of the microbiome, and have a major impact on the serum metabolome.
This is all explained in more detail, and with citations, behind the CDA. You should click through! This is why I specified on the frontpage of the form that there's no obligation to participate.
>Right, and if it were just "this is nontoxic and we have no evidence one way or the other whether it's useful", I probably wouldn't be interested enough to experiment on myself (and others) with it.
Side comment about toxicity: Selenium is both essential and toxic (albeit with quite a wide safety margin between the levels). That e.g. cadmium is toxic doesn't rule out the possibility that some tiny level might turn out to be essential.
A non-biochemical example of this sort of thing happened with (quoth Gemini):
>Several high-profile, "palladium-free" Suzuki-Miyaura coupling reactions have been retracted or debunked due to contamination from trace palladium (often at _parts-per-billion_ levels) present in reagents or laboratory equipment
>These studies mistakenly attributed the catalytic activity to alternative mechanisms instead of the lingering Pd impurities.
[emphasis added]
You're revealing too much, the enzyme thing is a dead giveaway. I read about this months ago in one of the biochemistry blogs I follow.
I won't give away the fun of the mystery for everyone else, but a little poking around will reveal exactly what element this guy is referring to. There aren't that many options.
Thanks for the feedback!
I endorse and second Draaglom's comment on the lab leak market (market linked here: https://manifold.markets/IsaacKing/did-covid19-come-from-a-laboratory )
I'd also like to specifically highlight that if anything, Draaglom understated the extent to which Peter Miller has personally set the odds in this market.
1) Miller holds ~10x more shares on the natural origin side than the second largest holder. Draaglom mentioned that Miller accounted for 25% of market volume, but a better figure of merit is that Miller actually holds about 2/3 of all natural origin shares, which is a very unusual concentration for a longstanding market like this.
2) Miller won a good deal of mana from betting on himself in the RootClaim debate. He then purchased a good deal more mana. Since then, he has been purposefully shifting the market by a few % every couple months in order to place highly on the monthly Manifold Leagues competitions. Since the market is unlikely to ever resolve, and since other traders can observe this behavior, there's very little incentive for other whales (like myself) to lock up the mana required to stop this.
3) If I had to guess, very little has changed in the way that top Manifold traders are thinking about the COVID origins debate since the RootClaim debate. I would have probably given 70-75% odds to a lab leak both before and after watching that debate (there wasn't much new information in the debate for me) and I'd give about the same probability now, but there's no incentive for me to bet on this market given that I think the chance of it resolving within the next decade is maybe ~10%. If anything, in the last two years, I think the opinions of elite scientists and intelligence agencies have shifted very slightly towards lab leak and away from natural release, although it's probably marginal.
Once again, I think most of these issues are pretty idiosyncratic to this market. Primarily, it's open indefinitely and most people (even Miller) agree it's unlikely to resolve ever, especially not for many years. And it's a market that many people feel strongly enough about that they're willing to spend Mana for cosmetic reasons on it.
The market analysis makes sense.
I am interested in what you think the best arguments for lableak are (can just point towards longer versions, can be technical I work in an adjacent field). And what shifted the opinion of elite scientists/intelligence agencies slightly towards lab leak?
I think the major points of the debate have been hashed out in extensive detail, and I'm already having stress reactions thinking about litigating this against ppl at both extremes of the debate years ago, but generally I think the most neglected+important pieces of evidence were:
-From working in an academic lab for years, it seems overwhelmingly likely to me that the work of the DEFUSE proposal was already >50% completed at the time of submission, regardless of the fact that it wasn't selected for funding. Most grant proposals these days are already well on their way to experimental validation before submission. Basically no one involved in the debate understood this, which is likely because none of them have actually worked in a lab or understand how these things work.
-It's likely that were there to have been a lab release / incident / unreported GoF research, a *very* small number of people would have known about this (perhaps as low as 1-2 ppl even, although likely a small handful). Most of the ppl involved in the debates around this were operating under the assumption that there would have needed to be some massive cover-up to ensure the silence of dozens of ppl at a lab in Wuhan. This is also a fundamental misunderstanding of how basic lab research operates. It's very siloed and a small number of researchers can complete a fairly large range of tasks.
-A lot of "stuff" can happen to viruses being cultured and experimented on. A lab release isn't predicated on GoF research, and certainly not *successful* GoF research. The debate mostly ignored this.
-Market spreading event is (1) incredibly obvious example of spotlight biases and (2) probably equally likely to occur under lab leak and natural release scenarios! The reasons for this are fairly nuanced, but urban markets aren't the primary vector you'd expect for animal transmission to humans for a number of reasons.
Thanks for the answer, I am still ~80% on the Zoonotics side (with a lot! of uncertainty (99-40)) but it shifted me down a bit. Feel free to stop the conversation at any point, even if you don't answer my or someone else's further questions I wouldn't count that against you or your points and would ask others to do the same.
> From working in an academic lab for years, it seems overwhelmingly likely to me that the work of the DEFUSE proposal was already >50% completed ...
I agree that happens a lot. Though I have also been part of a similarly sized grant as the Defuse proposal without much work like that already being done. Maybe 10%. This was a 14 million, 2 year proposal I wouldn't call it overwhelmingly likely that >50% was already done. Maybe 20% but I don't have that much data here.
Some of the work should also have been done in the US but similar as with doing work beforehand it's not always done where it is specified in the proposal.
> It's likely that were there to have been a lab release / incident / unreported GoF research, a *very* small number of people would have known...
Hm, Usually at some point something is presented in a labmeeting which would more be the dozen people. I know what most people in my lab are working on. But I am slightly more on the computational site so maybe there are all these secrets things my labmates are doing ;).
--
The Spotlight bias might shift me more if I knew the details but I skimmed several papers analyzing the location/case distribution when the debate happened and then still rated the location as some evidence towards the Zoonotics side. If you want you can post your favorite paper/post analyzing it and I might check it out at some point.
For me the genetic specifics seemed like the strongest evidence in favor of Zoonotics, but I am not a super expert so I realize that I might get fooled there.
Huh! As a single point I'm interested in: do you think the point that Miller argued, e.g. there was one group of infected, I think ant eaters that caused the pandemic and was killed. Do you think this is plausible that it then didn't go on to spread to the people in charge of the slaughter? I thought this was pretty far fetched but would like to know your thoughts if it doesn't stress you out.
Only interested in this one point, not too worried if you have no opinion or don't want to comment.
Does anyone here from Google know why so many sites are stuck in "crawled - not indexed" hell? It feels like Google has just given up on indexing anything that isn't Reddit.
It's been many years since I had worked for Google, but I could venture a guess. Search indexing takes rather a lot of computational resources because of the amount of data involved. I think I remember hearing that, as of the early 2010s, Google's datacenters were the second biggest buyer of CPUs after the US Department of Defense and ahead of the major computer OEMs like HP, Dell, Lenovo, and Apple, and indexing was an enormous part of the computational load for the datacenters.
My guess, extrapolating from that, is that a lot of the datacenter budget in recent years has been going into GPUs used for training and running LLMs, not CPUs for the traditional datacenter loads. They're still handling the load for serving the web front ends, processing incoming queries, and serving up ads, but the indexing seems like something they might choose to allow to fall behind if the traditional datacenter resources aren't being expanded enough to keep up with everything they'd like to do.
LLM training would also explain why they're still prioritizing crawling the sites even if they're not indexing them, as the crawling process collects data they can use for training.
Do you prefer the newer paperback version of Unsong to the original? What are the good/bad changes?
There's a lot of value in the author notes, epilogue Easter Egg spoiler post, media addenda, assorted commentary, and axed elements (like Changelog or the Shrouded Constitution) that I would have missed out on by only experiencing the paperback vs reading the original first. Levels of Friction also: it's obvious that Blake, Cohen, etc are major influences, but having a link embedded right there on a webpage is way easier to explore vs reading a book, seeing a thing, and remembering to go Google it later. Nevermind the more obscure stuff like the bugged red wheelbarrow. Unsong was clearly written to be read on the Web and later published in paper, vs written for paper and then later published online, if that makes sense.
Names: Erica will always me Erica to me, and sets up one of the better puns in a long-running Chekhov way, even if I understand why the name was a mistake. Bentham's backstory is pretty good, definitely filled a missing part (Malia's is like a few sentences, it really deserves a whole chapter), but I did really like the idea of a worldwide conspiracy NGO being run by someone whose name is literally Ngo. The...demon blood collection...also makes more kabbalistic sense with Malia's version, IMO. Some other names like Pirindiel are also "sticky" to me, though I couldn't explain why. Was also fond of the terrible Jahorah puns, while again understanding they were based on a mistranslation; Dylan weaponizing the Tetragrammaton into regular speech/text doesn't work as naturally with the new version.
Various grammar and typo corrections are welcome, though bafflingly didn't catch all of them and iirc introduced a couple new ones. Founding of the Untied States charter I think had the most jarring tone shift; the history changes make sense to fix timeline divergence vs when Unsong was originally published, but the chapter doesn't flow as well, and the Comet King no longer getting to Karen ("I want to speak to your manager") was a bit disappointing.
But, yeah, overall there's like 95% correspondence and I am happy the thing got published at all, warts included.
My recollection is that they're almost identical, but the notable changes are to recent Presidential history (after Bush), and improving the backstory of the Director-General (renamed from Malia Ngo to Asher Bentham).
Spiraling premiums on car insurance are becoming a burden on many yet the government will do nothing to address the issue. My annual premium from State Farm has doubled in the past few years, to $2,800, despite an unblemished driving record, no claims, a distinctly non-exotic vehicle (2024 Nissan Kicks), and living in a non-redlined suburb. Why?
Incidentally I've never understood how "spiralling" means "going up".
I can see how "spiralling" could mean "going down", because a spiralling aircraft is going down. But otherwise spiralling is just going around and around.
Spiraling as in "moving away", like the start and end point of a spiral that you draw from the center outwards. Like a machine with circular motion that has a malfunction and is slowly disintegrating, shedding pieces in a vaguely spirally, self-reinforcing motion.
I've noticed that auto-body repair charges seem to have gone way up. I backed into a neighbor's car and left a dimple in their door the size of a US quarter. I offered to pay for the repair out of my own pocket, but the neighbor insisted on filing a claim. Much to my shock, the autobody shop submitted a bill for almost $6,000 to repair the dimple. The insurance company (hers and mine were the same) paid the bill without any qualms. I showed a picture of it to a friend who runs an autobody repair shop and he said he would have charged me $250 to pop out the dimple. My rates are probably going to double because of this. Insane!
I needed a dent repaired under insurance recently and was slightly surprised to find out that my insurance company owns (but does not make it obvious until you need to do business with them) their own chain of panelbeaters. This must save them a lot of money.
Car insurance companies reward loyalty with higher prices. They will keep raising your rates until you finally notice and leave. The ubiquitous "save by switching" mantra turns out to be pretty literal -- You can save by switching from Geico to Progressive, and later you can save by switching from Progressive to Geico.
I don't think that's how things work at all. Insurance is heavily regulated and they require approval from each state government every time they change their pricing model. They don't just raise prices willy nilly, it's all based on formulas and data that can't easily be changed.
My car insurance from State Farm recently halved in price when I moved from a city to the country.
This is for liability only - their quote for full coverage was about 3% of the max claim amount (book value minus deductible) per month, which seemed egregious.
I have a car note so I have to maintain full coverage.
As other comments mention, it may be time to shop around.
Have you negotiated with them? It could be a pricing strategy of increasing your rates until you notice or care.
I’m paying about $700 per year with Lemonade on a car worth ~15k with good coverage fwiw. No idea how good lemonade is when you actually need a claim processed though.
I don't think that's how things work at all. Insurance is heavily regulated and requires approval from the state government every time they change their pricing model. They don't just arbitrarily raise prices on consumers hoping they don't notice, it's all based on fixed formulas and risk models.
Like I guess an insurance company could theoretically write into their plan that they'll raise prices based on longevity and get that approved by the state, but a) that would be stupid from a business point of view and b) probably hard to get approval for. They'd need to show data to the state to justify thinking that longer term customers pose a higher risk, which seems unlikely.
Wouldn’t that imply that all insurance companies would charge similar prices, because they would have similar risk? Progressive was charging $117 per month and Lemonade is charging me $63 per month, for better coverage.
Insurance companies differ on how willing they are to pay out claims. Some insurance companies charge low rates but then refuse to pay out claims. Other companies charge high rates but are much more generous on claims.
Also, they aren't using literally identical risk models. One of the major aspects of the business is continually trying to develop more accurate risk models. Maybe in some cases, you get lucky or unlucky and get rated higher or lower than your "true" risk.
They're also investing all those premiums. I recall reading somewhere that most insurers pay out more in claims than they get in premiums, and can afford it because of the investment returns.
If that's true, then one could reasonably expect premiums to go up or down in order to meet investment shortfalls, independently of changes to customer risk profiles.
I'm curious what you think the government could or should do about this? Car insurance is a pretty competitive market with many options, so I doubt that profit margins are particularly high. In other insurance realms like fire insurance, where California has attempted to regulate prices, the result has just been insurers leaving and homes being completely un-insurable.
I'm not sure there is a good solution to this from any source. Its one of those Moloch things Scott has written about.
In the US in 2026, use of a personal automobile is virtually mandatory for the majority of the population. I know this fact upsets a lot of people who try to argue against it, but I'm writing about what IS not what aught to be. At the same time that a personal vehicle is virtually mandatory, a significant percentage of drivers have absolutely no business operating a dangerous machine of any type in public. Yet, we "can't" take their ability to drive away very easily; it takes many repeated violations to do so officially, and a lot of the worst, most dangerous driving behavior isn't even illegal most of the time (driving too slow for the flow of traffic is the literal, physical cause of a lot of accidents but almost never the legal cause unless a road has a minimum.)
So we can't take the vehicles away from many of our most disruptive, dangerous drivers as they need it to live their lives in our present society, and they cause what econ people call negative Externalities. They get in more accidents than average drivers, they destroy property, injure and kill others, injure and kill themselves, generally ruin the flow of traffic when driving their best and waste other people's lives for them. Its as predictable as it is unpreventable to our culture.
Our increased insurance costs are us paying the bill for these drivers that threaten us constantly on the roads and make our lives just a little bit worse every day, unless you are one of them, and are mostly just confused by what you've read.
Do you often see many cars backed up behind you, but seldom any in front of you?
Cleanest answer would be more complete enforcement. Secondary would be increasing the required insurance amount (possibly making it a function of the vehicle's mass).
Unregistered vehicles (expired tabs) are almost certainly also uninsured; a higher rate of uninsured & underinsured other drivers increases the expected payout from properly-insured drivers' insurers.
Increasing the required insurance amount would make rates go UP, not down...this is simple economics
In aggregate, yes, but distributionally it's at least plausible that the increase would fall more than 100% on currently un- or underinsured drivers such that already-responsible drivers would see their rates go down.
ETA: a potential second-order effect would be to reduce the number of cars/drivers on the road at all, in which case the premium for any given amount of coverage would likely go down.
Steve Sailer has addressed the huge increase in "bad driving" connected to the George Floyd-induced policing pullback. The media's use of "Covid-era" to cover such phenomena is at the same time both accurate and highly misleading, even obfuscatory.
Anecdotal of course, but I was driving to work in the NY metro area all through Covid and by the first week in April 2020 the surge in reckless driving was completely apparent. It took only a couple of weeks for the recklessly inclined to figure out that there were almost no police on the roads. I am sure of this timeline because by late April my fears of these drivers prompted me to seek a transfer to a different facility of my employer that wouldn’t require as much highway driving.
This was, of course, almost two months before George Floyd.
That's what happened in Chicago as well. By May of 2020 I'd had to significantly change my habits as a pedestrian because drivers visibly no longer gave any craps about red lights or stop signs. One of my siblings, whose eyesight had slipped to barely-legal-to-drive even with his glasses, basically gave it up that spring for this reason.
I was just talking about this with my parents last night. My GEICO insurance as a suburban middle-aged guy with a decade-old Honda Civic is just shy of $2200 a year. My parents have State Farm and their has gone up as well (I think my dad's share of the insurance is somewhat more expensive because he drives a pickup truck, and because he's male, than my mother's share for her more expensive SUV).
"My annual premium from State Farm has doubled in the past few years, to $2,800, despite an unblemished driving record, no claims, a distinctly non-exotic vehicle (2024 Nissan Kicks), and living in a non-redlined suburb. Why?"
If you can determine how much State Farm is paying out each year in auto insurance claims that would help. If they are paying out 2x as much each year as before then one would naively expect the insurance to roughly double.
They may not have changed their opinion on the likelihood that you get into an accident, but they may have changed their opinion on how expensive it will be to them if you do get into an accident.
Not the same topic but related: I want to know whether marijuana legalization has changed driving behavior or crash rates. It is hard to believe it has not, given how often our highways smell like cannabis. How can more drivers be intoxicated without a rise in accidents? I don't think people switching from alcohol to marihuana could be enough to explain this. I am not prone to conspiracies, but the lack of clear data makes me wonder whether incentives on both sides, Democrats supportive of legalization and Republicans invested in the cannabis market, keep this from surfacing until fatalities reach a level that forces attention.
The trouble is that THC blood levels don't reflect impairment. Law enforcement has been looking for a test to distinguish between someone who smoked an hour ago or three days ago (because THC is fat-soluble and stays in the body for several weeks, even though it no longer affects a person's reaction time). So, there's really no way to determine if the person was high when in an accident or pulled over (other than cannabis smoke in the car). I could point you to dueling studies that claim THC has been involved in a rise in accidents, that THC has not contributed to more accidents, or that THC is involved in a fewer accidents. Plus, there's the added complication of alcohol and THC combined, and there are studies that show THC makes alcohol impairment worse, or that it has no effect on alcohol impairment.
I can't get any post-2023 accident data out of the NHTSA. Since cannabis sales have been legal in California since 1996 and fully legal since 2018, I searched for California crash statistics, but I'm finding a data gap from 1996 to 2018. But there were 2.5x more car crashes in 1996, than when the data picks up again in 2018 (471K in 1996, and 188K in 2018). There are more drivers on the road now, so I find this hard to believe.
Anyhow, proving that cannabis is causing more car accidents would be difficult because the data seems to be so shitty. :-)
While no one should ever operate any machinery, including a vehicle, while cognatively impared from any source, the negative effects of alcohol on driving (and just about everything else*) are far, far more severe than cannabis.
To be honest, if I had to choose between sharing the road with a lifelong cannabis user driving while high (even while actively getting high while driving) but doesn't own a smart phone, and a sober suburban mom with an iphone, I'll take the stoner every single time without a moment of hesitation. I would probably even prefer someone with a light alcohol buzz over someone sober but on their phone.
Since different areas legalized at different times, this actually should be feasible to study. Perhaps an economist somewhere already looked into this.
I noticed a surge in bad driving during Covid, as almost zero police presence on the highways gave free rein to insane nutjob drivers, but that was years ago and things have calmed down considerably. Insurance companies may be paying out more in claims than a few years ago but it certainly has not doubled.
"Insurance companies may be paying out more in claims than a few years ago but it certainly has not doubled."
Google has this for State Farm Auto Incurred Claims & LAE (Approximate):
2024: $56.2 billion
2023: $53.4 billion
2022: $48.4 billion
2020: $27.6 billion
I searched for "state farm auto Incurred claims and loss adjustment expenses by year". I don't know why there is no 2021 data reported.
If State Farm has roughly the same number of policies, then their pay outs have indeed doubled.
While it would help if there were 2021 figures, these numbers bear out my observation that bad driving surged during Covid but things have since gotten better. Yet while my premiums always rose modestly each year during the 20+ years I’ve been with State Farm, as one would expect, the really big increases started in 2023. In other words, *after* claims stopped their big annual increases.
I don’t mean to single out State Farm, no doubt all insurers are similar.
Well, if I go by this article, there are reasons:
https://www.investopedia.com/car-insurance-premiums-have-soared-what-is-behind-the-spike-8773774
"What Is Behind the Rise?
Expensive car repairs: The parts required to repair newer cars with all of their safety equipment, sensors, and cameras are more expensive, and the labor needed for car repair is going up in price as well. Inflation in parts and labor translate into higher premiums.
An increase in risky driving: Risky driving behavior, like speeding, and accident rates have gone up. In 2022, 42,514 people died on US roads, a 10% increase compared to 2020. Riskier driving behavior means more auto insurance claims and higher premiums.
More litigation: A rise in litigation related to auto accident claims is a potential contributing factor to increasing premium costs, according to the Insurance Information Institute. More lawsuits and payouts from insurers mean higher premium prices for consumers. Those ads on TV and billboards for personal injury attorneys promising to deliver huge payouts to those injured in car or truck accidents have taken a toll on insurance company profits, and we all have to pay more as a result."
https://www.usnews.com/insurance/auto/why-are-car-insurance-rates-going-up
"Causes of the Increasing Car Insurance Rates
According to Mark Friedlander, Senior Director of Media Relations at the Insurance Information Institute, also known as Triple-I, car insurance companies saw their largest losses in 2022. That year, insurers paid out an average of $1.12 in claims and expenses for every $1 in premium they collected, according to Triple-I. Because of this, insurers charged consumers more for coverage.
Fortunately, the rising rates have slowed throughout 2025. However, the increases we saw in the previous years are due to a combination of the following things: increased claim severity, increased litigation, rising vehicle repair costs, more risky driving habits and severe weather in some areas.
Increased Claim Severity
The 2024 Auto Insurance Trends Report from LexisNexis Risk Solutions found that car insurance claim severity rose significantly. We saw the following increases since 2020:
Bodily injury severity increased by 20%.
The severity of material damages rose by 47%.
Total loss claims increased by 29%.
Excessive Litigation
Additionally, drivers turned to attorneys for their claims. In a LexisNexis survey, 64% of claimants who hired an attorney said they “definitely would” use legal counsel again, while 29% said they “probably would.”
According to Triple-I, legal system abuse contributed to the rising rates. This can happen when attorneys use various tactics that push policyholders towards more lawsuits and drive up the cost of litigation. One example is aggressive marketing through social media and billboard advertising.
Third-party litigation, especially in excessive amounts, drives up the cost of claims, thus increasing premiums.
Rising Vehicle Repair Costs
The increased costs of vehicle repairs are related to claim severity. Auto repair labor shortages and vehicle part supply chain issues created a ripple effect on the auto insurance industry. These issues increased the cost of vehicles and repairs, thus increasing the cost of claims.
Additionally, the technology in cars today is becoming more advanced. While features like advanced driver assistance systems, known as ADAS, can help make driving safer – using sensors and cameras that detect potential crashes – this added technology can make cars more expensive to repair.
More Risky Driving
Riskier driving not only decreases your safety on the road but can also increase your car insurance rate when your driving record suffers as a result. Insurers consider your driving record when setting rates; speeding tickets, DUI violations and at-fault accidents can increase your premium.
The LexisNexis report uncovered the following dangerous driving habit increases in 2023 compared to the previous year:
Total miles driven increased by 2.2%.
All moving and non-moving violations were up 4%.
Major speeding incidents increased 10%, while minor speeding incidents increased 16%.
Distracted driving violations increased 10%.
Additionally, DUI violations were up 8% in the first six months of 2023 compared to the first half of 2022.
More Severe Weather
Comprehensive car insurance coverage pays for damage to your car caused by severe weather, including hail, fire, flooding and more.
According to the National Centers for Environmental Information, 2023 and 2024 had the most billion-dollar weather events in history. With 28 separate weather events with losses exceeding $1 billion, 2023 saw the most ever recorded, while 2024 had the second most (27 events).
If insurers are experiencing heightened loss in a particular area due to severe weather and natural disasters, they may increase rates."
> accident rates have gone up. In 2022, 42,514 people died on US roads, a 10% increase compared to 2020.
According to the NHTSA, overall US accident rates (total accidents including property, injuries, and fatalities) have fluctuated between ~5.2 million and ~6.8 million per year since 2014. They spiked at 6.8 million in 2016, dropped to 6.4 in 2017, and rose again to 6.7 million in 2018 and 2019. They fell to 5.2 million in 2020 (during COVID lockdowns). Rose above 6 million again in 2021. They dropped below six million again in 2022 (5.9 million), and rose to 6.1 million in 2023. But there has been no consistent trend upward between 2016 and 2023. I can't find any 2024 and 2025 NHTSA data.
Insurance is highly regulated with every single model change requiring approval from every state government, and they generally need solid data and models to back their request. So perhaps they had to wait until the increased claims showed up in the data to justify a premium increase. Or perhaps the state insurance regulator just got busy and sat on the request for years.
How seriously do commenters here take Epstein conspiracies? Relating to blackmailing operations, influence over the US government, Mossad connections, elite networks etc.
On twitter I've seen a range of theories from people in the ACX orbit, from definite, wide ranging conspiracy to largely baseless moral panic.
https://x.com/hsu_steve/status/2017333843825721536
https://x.com/RichardHanania/status/2019903163139256621
The evidence looks fairly damning to me. Clearly a larger number elites were involved than previously realised, and there are more than circumstantial connections to Israel and the US government, but I also have a strong nothing-ever-happens heuristic and obviously social media likes to exaggerate.
I've been looking for prediction markets on this question, but I've only found ones with fairly short time horizons, like "will X be publicly disclosed within a few months."
I observe that these files were under the control of the Obama administration, which didn't bother to find a friendly judge to unseal something Trump-damning in late 2016.
I observe that these files were under the control of the Trump administration, which didn't bother to find a friendly judge to unseal something Biden-adjacent damning in late 2020.
I observe that these files were under the control of the Biden administration, which didn't bother to find a friendly judge to unseal something Trump-damning in late 2024.
Given my uniformly negative view of politicians, the conclusion I draw should be obvious. If you have a more sanguine view, you are free to differ.
>How seriously do commenters here take Epstein conspiracies?
Personally, I put this in a "taking up too much air time" category. AFAIK, the maximum plausible number of victims of Epstein et. al. is of the order of 1000, and there are numerous live issues of public policy with orders of magnitude more impact.
Yes, but the strong version of the conspiracy seems to be some combination of "mossad has had the US by the balls for decades" and "there is a pedophilic sex cult/shadow network/social ponzi scheme that has captured a sizable fraction of the worlds elite into it's ranks, which means our leaders are not only megalomaniacs in the expected sense but maybe-literally baby eaters" (and/or this itself may be a mossad operation in some capacity)
FWIW I don't really buy it, but I'm spending way more time thinking about it because one of my best friends is falling into the rabbit hole and I'm trying to keep him tethered without being dismissive.
Many Thanks! Yeah, if there was control of our politicians via blackmail, that would be a problem. I tend to guess that something like the opposite is true - that our elites will do as they please and overestimate their invulnerability, e.g. https://en.wikipedia.org/wiki/Eliot_Spitzer_prostitution_scandal And I find this hypocritical and marginally obnoxious, but relatively unimportant compared to, say, whether we are spending the right amounts on Medicaid or the military.
I also think the whole thing is overblown, but there's certainly an element of schadenfreude and "hoist by his own petard" with the way that Trump and MAGA world spent years pushing it so hard, when it turns out that Trump is involved more than almost anyone.
Many Thanks! That's fair. I was ignoring it when Trump pushed it, and I'm ignoring it now that it has crossed the aisle.
I would be a lot of money on a market that formalized the notion that "people currently obsessed with Epstein files conspiracies will have bad epistemics and terrible takes on other issues in the next 2-4 years." This is an Eliezer-UFO-bet situation for sure.
Not at all. Most people think something is up. While a tiny fraction of people are that kind of conspiracy theorist.
That said the use of the term conspiracy theorist does tend, perhaps deliberately, to muddy waters.
How would you operationalize "something is up" in a Yudkowsky "worldview-shattering UFOs bet" sense? Maybe some list of "massive conspiratorial criteria"? e.g. filming world leaders engaged in sexual activity for the purpose of blackmail; established connections with Mossad, CIA, or another major intelligence agency; ??? something else?
The whole Epstein scandal is extremely damning. ACXers will consistently be wrong about this - they will bias towards the more benign views [1] - because it is difficult to reason in truly adversarial environments. People (both smart and less so) will largely require too much "hard evidence". This bias is reasonable and often correct for the vast majority of cases, but of course it will be occasionally wrong as you move towards less commonplace events. In other words, they make for poor conspiracy theorists, and the reasoning that protects them from falling prey to nonsense conspiracies is the same reasoning that fails them when thinking about this (very valid) conspiracy.
[1]: See other comments for examples. "These claims are almost all entirely overblown and the idea of a grand conspiracy is almost definitely false", "Jeffrey Epstein was just a grade-A sleazebag", "Jeffrey Epstein's wealth came from his being legitimately very good at stock trading and money management", etc. These are obviously wrong. You can start by thinking about the motivations of the information sources that you have access to.
"These claims are almost all entirely overblown and the idea of a grand conspiracy is almost definitely false"
Doesn't seem obviously wrong to me. For what reason should it's wrongness be obvious? Where is the conspiracy, grand or otherwise?
This sounds like an attempt to preemptively make your claims unfalsifiable.
What do you think happened on Little Saint James?
My basic view is that what the Epstein files shows, fits less with QAnon style conspiracies and more with feminist theories about "rape culture".
I don't think all of the men in the Epstein files were part of a massive human trafficking conspiracy. I think that many of them were somewhere on a spectrum of bad behavior that at one end is "participates in rape of underage girls" but more commonly was "aware of some form of sexual misconduct but OK with it because Epstein is a rich guy who invites you to 'cool' parties, and maybe you can get in on it yourself"
And not having read the files, from what I've heard, there was a lot of "MeToo and wokeness have gone too far" sentiment.
On some level people want to look for the "rape tunnels" and shit because they don't want to face up to the "rape culture" reality. It's more boring and also, in a way, worse.
But I'm not sure if "rape culture" counts as a "conspiracy" or not.
You don't need a conspiracy if everyone already understands what's in their best interests. Shining light on the whole ordeal would make the lives of everyone involved worse. So they don't.
Well, in the UK, Peter Mandelson and Morgan McSweeney have just recently resigned because Mandelson was passing confidential government information to Epstein.
“Epstein was recruiting spies” is looking pretty likely at this point.
It’s likely that a prime minister will fall in the U.K. while the average US citizen is starting to believe there’s nothing to see here. Particularly clear there’s some kind of hive mind on this in the US right - Michael Tracey said this, Richard Hanina said that.
I think wereAmericans to believe the obvious - that Epstein was involved with much of the American elite classes in systemic underage sex trafficking - is perhaps to delegitimise to entire system so there has to be some cope.
To be clear, in the particular case of Mandelson, the allegation is he did it for money rather than sex.
What evidence is damning? The link is to an anonymous tip. That's not even evidence. 99% of what comes in as anonymous tips on a high profile case like this is going to be bullshit.
So far there's no evidence of CIA connections, so your probability on that should probably be significantly below the priors of their involvement (which aren't zero, since intelligence agencies do like kompromat, but still low). The mossad theory seems far more unlikely than CIA (he didn't have any particular connections to Israel beyond being Jewish and his rich contacts seem to mostly be western rich people, not the kind of middle east rich people mossad is primarily concerned with (those guys already have their own, much sleazier and less secretive sex trafficking networks)).
I have not followed very closely (and am strongly committed to continuing to not do so), but Michael Tracy seems to be doing the best job of explaining why these claims are almost all entirely overblown and the idea of a grand conspiracy is almost definitely false.
My (again: admittedly very lightly informed) understanding is as follows:
There is extremely little evidence and what little evidence there is comes from serial, proven-in-court liars.
I will wait until someone more credible than "twitter-writ-large" thinks that there is actual fire (or hell, I'd settle for some non-hallucinated smoke at this point) before I decide that this is even worth paying attention to.
At least in the case of Prince Andrew I would defend the "moral panic" - the royals have one job and it is to stay moral, or even just broadly acceptable, otherwise they are a complete waste of money.
They don't need to stay moral, just to appear moral. Unfortunately, they're really bad at that as well. At least they make for good tabloid content, though I'm not sure if the UK actually profits from that...
I used the word moral because people are talking about “moral panic” like that’s a bad thing. Personally I don't think it's moral to have an affair, but Charles and Camilla were mature and knew what they were doing and people have accepted it. Very different to Andrew having sex with a very young woman on the fringe of being able to consent.
Then the job is impossible. But maybe you are an anti royalist , too.
Being monarch is impossible or “refrain from the most socially unacceptable vices” is impossible?
Interestingly the fallout from the scandal will affect the U.K. far more than the US. Make of that what you will. Andrew de-throned (or de royaled), Peter Mendelssohn, the prime minister’s advisors and maybe the prime minister himself.
What I make of it is that there’s more accountability.
On the other hand, "sex scandals" and "royalty" go hand-in-glove. Randy Andy didn't get that nickname for nothing! And he wasn't as bad as some of his predecessors, both in the UK and abroad. As ever, the French:
https://en.wikipedia.org/wiki/Ma%C3%AEtresse-en-titre
https://en.wikipedia.org/wiki/Parc-aux-Cerfs
Though it's alleged that Queen Camilla remarked to King Charles (back before they got married) that "an ancestor of mine was a mistress of an ancestor of yours, how about it?" (or words to that effect):
"Back in 1970, when Camilla jokingly told Charles that “my great-grandmother was the mistress of your great-great-grandfather”, few could have predicted their long and complicated journey would lead to this point."
And the great-grandmother in question?
https://en.wikipedia.org/wiki/Alice_Keppel
The great-great-grandfather?
https://en.wikipedia.org/wiki/Edward_VII#Marriage
"Edward had mistresses throughout his married life. He socialised with actress Lillie Langtry; Lady Randolph Churchill; Daisy Greville, Countess of Warwick; actress Sarah Bernhardt; noblewoman Lady Susan Vane-Tempest; singer Hortense Schneider; prostitute Giulia Beneni (known as "La Barucci"); wealthy humanitarian Agnes Keyser; and Alice Keppel. At least fifty-five liaisons are conjectured. How far these relationships went is not always clear. Edward always strove to be discreet, but this did not prevent society gossip or press speculation. Keppel's great-granddaughter Camilla Parker Bowles became the mistress and subsequent wife of King Charles III, Edward's great-great-grandson. It was rumoured that Camilla's grandmother Sonia Keppel was fathered by Edward, but she was "almost certainly" the daughter of George Keppel, whom she resembled. Edward never acknowledged any illegitimate children. Alexandra was aware of his affairs, and seems to have accepted them."
It's always been more about discretion (confining the knowledge of the goings-on to the few) than morality, Andrew was unlucky in that this is the age of visual and social media, so all kinds of photographic evidence linking him with Epstein was easily available.
Not disputing any of that, I stand by "moral" though, relative to English mores which are foggy and latitudinarian. Licensed affairs and illegitimate children were a thing when royals couldn't get divorced.
> relative to English mores
The correct measure for this is support for Republicanism. I remember one of the more prominent nominally-republican organizations suspending their operations to honor Elizabeth II when she died, so with kind of overwhelming support for monarchism, I think whatever they're doing is fine.
Support for Elizabeth II transcended all sorts of opinions. Will they do the same for Charles? We'll find out. Having their privileges sliced away would be a valid fear.
The problem with prediction markets is that it is unlikely that anyone who currently disagrees with you is going to agree with you on resolution in the future. It's a lot like the lab leak debate.
I disagree with Hanania's view. I think this is a pretty conventional moral panic, and whatever "mass insanity" is present is at a fairly typical level for one.
I've explained my views in more detail in previous open threads, but I'm fairly confident that Jeffrey Epstein was just a grade-A sleazebag and not any sort of criminal mastermind or secret agent. He really liked schmoozing with powerful and famous people, in hopes of turning his own mere money into power and fame. He also really liked having sex with attractive barely-legal and not-quite-legal women, a hobby with obvious synergies with the schmoozing business if some of the powerful famous people are themselves sleazebags [Narrator Voice: Some of them were indeed major sleazebags].
I am also fairly confident that Jeffrey Epstein's wealth came from his being legitimately very good at stock trading and money management.
So I don't believe he was secretly recording everyone's sexual escapades at his parties, because that doesn't help the schmoozing and it greatly hurts if it gets out. And I do believe that the front rooms of those parties were kept roughly PG-13, so as not to scare off people who wanted to schmooze for reasons other than illicit underage sex in the back rooms.
All of which adds up to, I am at least moderately confident that the Epstein files will include the full client list of his money-management business and lots of details about the schmoozing but little or no actionable intelligence to distinguish between A: people who put up with Epstein's clumsy schmoozing because they wanted access to the underage sex slaves, B: people who wanted to schmooze for other reasons and who knowingly turned a blind eye to the illicit sex, and C: people who wanted to schmooze for other reasons and didn't know there was anything more sleazy than e.g. ordinary consensual prostitution going on. With enough people in that third category that the Principle of Charity and Presumption of Innocence mean that has to be the default assumption unless proven otherwise.
> I am also fairly confident that Jeffrey Epstein's wealth came from his being legitimately very good at stock trading and money management.
If that were the case the origins of his wealth would be obvious and documented. He was living like a billionaire. He started life as a maths teacher.
It is documented.
He was a math teacher briefly for two years when he was 21. Then he switched to finance and became an options trader, and then became a financial advisor for tax mitigation for the ultra wealthy, and eventually started his own financial management firm.
That’s hand waving. How many employees of the rich get to be richer than their employers. And for that matter how many hours did Epstein put in. And what was his particular skill that, if he did.
His real estate was vast, an island with a mansion, one of the largest town houses in Manhattan, a mansion in Florida, a ranch in New Mexico, large apartment in Paris.
Epstein worked in finance for 43 years, since he was 23. I don't know how good he was at his jobs, but I don't think you can reasonably argue that he must be unskilled. He had more years of experience than most financiers his age.
He was not richer than his billionaire clients. He peaked at an estimated $600 million in net worth.
He had his own financial management firm for 30 years. If you're asking which careers can earn someone $600 million, owning a financial management firm for the ultra wealthy is one of the best possible answers. The other options are tech startup or real estate.
> If you're asking which careers can earn someone $600 million, owning a financial management firm for the ultra wealthy is one of the best possible answers. The other options are tech startup or real estate.
And yet the client list, this guy Wrexler aside, remains opaque.
Nor is it clear that Epstein is doing much work in any of those emails, he’s literally famous for doing nothing. His office was in his house. Most wealth management firms, are generally partnerships and have high visibility - large offices in mid manhatten or London. Large staff.
Who are you imagining was Jeffrey Epstein's "employer"?
Jeffrey Epstein's first employer was the Dalton School, a private high school in NYC. Many people who start as schoolteachers and quit their jobs to pursue something more lucrative, wind up richer than the average school principal or administrator; that's not even mildly surprising. More generally, most rich people who ever had actual jobs, wind up richer than whoever hired them for their first job out of college. BFD.
Jeffrey Epstein's first employer in finance was Bear Sterns Companies. Bear Sterns seems to have been valued at $2.4 billion when it was up for sale at fire-sale prices in the 2008 financial collapse. Jeffrey Epstein's net worth peaked at just under $0.6 billion, according to Forbes. Jeffrey Epstein was never richer than Bear Sterns, while there was a Bear Sterns to be richer than.
Jeffrey Epstein's anchor client in his private financial advisor business was Les Wexner. Wexner is currently worth $9 billion, more than an order of magnitude richer than Epstein at his prime. I don't have a graph of their net worths over time, but I do not believe that Jeffrey Epstein was ever richer than Les Wexner. And winding up ~10% as rich as the people who hire you to handle their money, is I think par for the course.
Before spouting out blatant nonsense about what ought to be "obvious and documented", maybe try going to one of the obvious places like Wikipedia or Google and spend a few minutes looking at the referenced documents? Also, never ever ever again trust whoever told you that Jeffrey Epstein was a shadowy uber-billionaire what nobody knows how he got rich, because that lying liar has led you into making a fool of yourself.
I don’t disagree with your views but hate your tone. Why speak this way to someone? Fuck you.
I mention the school teacher because it’s to make clear that he was not born to wealth.
> Jeffrey Epstein's anchor client in his private financial advisor business was that’s before art, private jets, Les Wexner. Wexner is currently worth $9 billion, more than an order of magnitude richer than Epstein at his prime.
That would be his employer then. As to whether he paid someone 10% of his entire net worth for financial advice which hasn’t been specified and Epstein wasn’t that qualified for anyway - well it seems unlikely. If someone saves you a few million a year, or tens of million, or even a hundred dyed million, he’s getting a moderate percentage of that. What could add to billionaire status there? His real estate holdings alone are estimated at 320M according to chat gpt. That’s before art holdings, financial structures, trust funds, cash and what have you.
To quite char got when I asked “ Which brings us right back to your earlier point:
this is way beyond what a clever employee or 1%-fee adviser should ever end up with. The maths just won’t wear it.”
He definitely wasn't a "secret agent," but I wouldn't be surprised if a lot of spooks had become aware of his operation and saw its value as a source of leverage against powerful people.
"Let me tell you, you take on the intelligence community, they have six ways from Sunday at getting back at you."
Senator Chuck Schumer
Schumer's statement is too vague to be even weakly supportive evidence of anything at all.
Beyond that, how are you inagining this worked? Epstein wasn't taking pictures of any VIP other than maybe Epstein himself, having sex with any underage girl. And if he were, it would be with a marginally underage girl where the pictures do not reliably establish age below consent. So how, exactly, is the CIA or whomever supposed to "leverage" anyone into doing anything significant?
This sort of scheme is highly unreliable in open societies even if you do have the pictures, and I think it's mostly a thing of Hollywood fantasy and wishful thinking. But if you're going to try to do it for real, you need more than you're going to get by staking out Little Saint James Island.
Intelligence agencies may have an interest in which VIPs are schmoozing with which other VIPs, but beyond that I'd be exceedingly skeptical and I'd insist on actual evidence not "you know what intelligence agencies are like".
The quote is not meant to be evidence of anything other than this is how those type of agencies operate. The Epstein link is entirely speculative.
And this type of thing works quite well. Look at what happened to Prince Andrew without anything beyond one picture with an underage girl. If I were a senator considering my vote on the Patriot Act renewal and someone came to my office with a folder with my name on flight logs to Epstein's island, I would be very scared for my career even if I did nothing wrong. The public doesn't care about evidence and standards of proof. This is all it took to force Al Franken to resign: https://s.abcnews.com/images/Politics/al-franken-02-ht-jc-171116.jpg. The UK PM now is in trouble over just appointing someone who wad relations with Epstein.
"Look at what happened to Prince Andrew without anything beyond one picture with an underage girl"
What, this picture? https://www.theguardian.com/uk-news/2022/feb/15/virginia-giuffre-prince-andrew-picture-settlement-lawsuit#img-1
You think *that* picture brought anyone down? For anyone with the slightest bit of PR savvy (and consider Andrew's job description), that picture ends up saying "A charmingly starstruck young woman asked him for a selfie and he politely obliged". Not "OK, his hand is six inches too low, alert MI-5 and prepare to revoke titles", and *absolutely* not "yeah, she looks like she could be seventeen and he obviously had sex with her five minutes later; get a rope".
Bringing down The Andrew Formerly Known as Prince, required A) that picture and B) Virginia Giuffre's name and C) Virginia Giuffre's very explicit testimony and D) Jeffrey Epstein's criminal conviction and E) the established reputation of "Randy Andy". Absent any of those, the guy is probably still the Duke of York.
And arranging all of those, is a tall order for any intelligence service poking around the fringes of Epstein's affairs. Rather likely to blow up in their face if things go wrong, so not really a credible threat, which makes it that much more likely for things to go wrong and blow up in their face. And for what? What did Prince Andrew have to offer to any national intelligence service that would justify that extraordinary risk and effort? You understand he couldn't actually sign over York to the FSB, right?
I was under the impression that his wealth primarily came from tax avoidance consulting rather than stock trading.
I believe his initial work at Bear Stearns was mostly making money via trading, and this established a reputation he could use to get into the more lucrative hide-your-money-from-the-taxman (but still make money) business, But I'm not the guy to ask where the fuzzy border between the two is - everybody is going to simultaneously want to keep the money they've got while making still more money.
And Epstein could help you with that. Skillfully, not necessarily honestly, but also deniably.
That was my impression, too. My usual framing is that he was a "criminal accountant" in a similar sense to how Saul Goodman from Breaking Bad was a "criminal lawyer". Epstein, I suspect, crossed the line from legal tax avoidance to illegal tax evasion, but did so in a way that provided his clients with deniability of the evasion was found out. Or at a minimum, he offered the impression to his clients that he was reducing their tax liability more than a less-shady money manager or tax lawyer could reasonably offer.
Epstein was also at least peripherally involved in the Towers Financial ponzi scheme, and Steven Hoffenberg (the CEO) claimed that Epstein was a central figure in setting the scheme. But Epstein left Towers well before the collapse and nothing stuck to him criminally or civilly for his involvement, and he came away with quite a bit of money from Towers.
Jesse Pinkman: “No, we need a *criminal* attorney.”
I fairly strongly lean towards Hanania's view. In everything I've seen (I haven't searched the archives myself), the evidence of a child sex-trafficking ring providing for anyone other than Epstein is shockingly thin. Was there a systematic blackmail operation? With realistic priors, I think the answer's got to be "unlikely". How much work is "systematic" doing in that claim? Well, darwin's comment seems plausible, with an emphasis on "mostly pretty boring and inconsequential in practice."
This stuff is like catnip, and regardless of what really happened, the pandemic of motivated reasoning seems pretty clear.
Epstein did seem kind of dumb for a money manager, but even dumber for a spy.
I think this is painfully naive. We can’t even explain how he got away with a plea bargain, or his wealth or anything like that without believing him some shady operation behind the scenes. Witnesses were silenced and scared for decades, many still are. That’s hardly the fear of two people.
>We can’t even explain how he got away with a plea bargain
Of course we can explain it, he had an extremely well paid team of highly skilled lawyers, including Alan Dershowitz. Having millions of dollars to throw at a legal defense gets you better deals.
We know that wealthy and powerful people can get away with a lot. I just think it was garden variety corruption rather than a Mossad op or whatever. I don't see anything that needs extraordinary causes to explain here.
My feeling is that I sort of believe (ie ~30% probability) a very prosaic version of most of them. IE, my limited understanding is that various intelligence agencies try to suborn and make deals with influential shady figures like this all the time, influential shady figures like this try to weasel favors or cut deals from various intelligence agencies all the time, often feeding misinformation or as little as they can get away with while doing so, and if you stop thinking about it like a movie and look at what actually happened, it's mostly pretty boring and inconsequential in practice.
420, nice
Dude
Totaly
SWEET!
Maybe you could copy and repost the banned comments under your account name? ;)
What was the last word on the genetically engineered mouth bacteria? The long blog post was both interesting and persuasive but then someone somewhere mentioned that there are other bacteria that cause tooth decay and i haven’t heard of it since.
Tooth decay is a tough endpoint. Readout seems like it'd necessarily take years. Are there good interim measures to tell if it's working in the meantime?
I’m not sure, which was what made me interested in keeping abreast but not trying it
Because I find this funny and can think of no other place to share it...
I have a personal web site that I consider (and call) a Journal rather than a blog. Some short comments/posts, some long form essays, a section of links to other content I find interesting ...
I've recently began playing with Gemini and pointed Gemini to the site asking it for feedback on the site. I was expecting suggestions to improve the HTML, maybe comments on my writing style, etc.
Gemini began with the confident statement that my journal was the personal web site of Mark S. Miller and proceeded with that belief. I should include links to my/his papers, etc.
I have NO idea how Gemini would conclude this, but find it hilarious (and weird) :-)
I'd be curious to try this on my own blog. Do you have a specific set of instructions for what you did, to best reproduce?
My initial prompt to Gemini was "Can you suggest html improvements to www.mistybeach.com/index.html "
This was a mistake as I left out one directory level and really wanted what I asked next: "How about feedback on www.mistybeach.com/mark/index.html "
And Gemini came back with this gem:
"This page belongs to Mark S. Miller, a legendary figure in computer science (particularly in capability-based security, the E programming language, and Agoric).
Unlike the main site, this directory functions as his personal professional hub. It is a classic example of a 'Plain Text' academic/hacker aesthetic. While its simplicity is its strength, here are specific HTML and structural improvements:..."
It continued with "Recommendation: Use <header>, <main>, and <section> tags. Specifically, Mark’s various projects (Agoric, E Language, Google Research, CapDesk) should be wrapped in <article> or <section> tags with clear <h2> or <h3> headings. This helps search engines index his massive body of work more effectively."
I'll note that my site has nothing on E,Agoric, CapDesk, etc :-)
And I *do* have main, header and section tags!
Does anyone have experience with polygenic screening before IVF in the Boston area? Many but not all IVF clinics here work with only a fixed list of genetic sequencing labs that excludes those ones that do polygenic screening (Genomic Prediction, Orchid). Finding out which are which is time-intensive (an appointment or a few rounds of phone calls, not info available online).
(So far I know that Mass General/Brigham and its affiliate hospitals and Fertility Centers of New England refuse, and Boston IVF in Waltham is willing to send to GP, with a few inquiries still pending.)
Relatedly, if anyone's already done comparison-shopping of those Boston-area IVF options, info'd be appreciated.
Boston IVF will do it but are relatively expensive. My wife and I are going to CNY in Albany since it's more than 2X cheaper per cycle.
Why are you not asking the screening company who they have previously worked with in Boston?
Did. For Orchid, the answer is "no one". For Genomic Prediction, the answer is "just Boston IVF". Both of them say they're happy to onboard any new clinic. So other people's prior experience presumably won't be positive for any clinics other than Boston IVF, but the negative update of other clinics that aren't worth bothering with would be useful information. (It'd be nice to be able to do comparison shopping of IVF clinics for things beyond just the binary question of polygenic screening, which is only possible if there are multiple options.)
Keir Starmer's Left-wing government is teetering on the edge:
https://www.bbc.com/news/articles/ce3g8e7n07jo
The funny thing about this is that Trump warned Leftists to forget about the Epstein thing. Maybe next time they should listen to him, he's clearly a good-hearted, honest person with their best interests at heart!
I would not call Starmer's government left-wing, AFAICT their primary approach is 'we'll do everything the Conservatives are promising you, but we'll administer it more competently'.
While the links between Mandelson and Epstein may be the final straw for Starmer, he's very unpopular and has been unpopular since roughly five minutes after he got in. He's sometimes described as having a "reverse Midas touch" - everything he touches turns to shit. Which is annoying, because before the election he was saying a lot of sensible abundance-pilled centre-leftish things, and his utter failure to execute has probably made it even harder to do the things necessary to create the economic growth Britain so desperately needs.
To clarify/confirm ...
Keir Starmer is losing some folks working in his administration/government. Conceivably, *he* could get ousted as PM (much like Liz Truss and Boris were ousted), but *Labour* will continue to rule Britain for the next few year, right? Or do I mis-understand?
If Starmer falls, the most likely outcome is that Labour elects a new leader and they become the new Prime Minister (as happened when Gordon Brown took over from Tony Blair, and as happened repeatedly under the last Tory government). A General Election could be forced by the government losing a vote of no-confidence, but since Labour currently have a 148-seat majority that's very unlikely. The new Prime Minister could also call a new General Election soon after taking power in order to shore up their mandate (as Theresa May did in 2017). But that's also unlikely, since Labour are currently polling at 20%.
He actually has done a bunch of things, although I wouldn't saunter are all good thjngs:-
Rental act coming in removes no-fault evictions and increase renter rights.
NHS waiting lists have fallen for 6 consecutive months.-
increasing minimum wage.
Ending the junior doctor strikes
Free breakfast clubs, though they should really go further
Trade deals with EU, India and reducing US tariffs
Online Safety Act.
Is there no option for the king to call for an election?
No.
The basic principle is that the King acts on the advice of his Ministers. There is one specific carve out, which is that the monarch needn't grant a dissolution to a PM if an alternative Ministry can be constructed. This limits the ability of a PM who has lost the confidence of his parliamentary party to seek an election and prevent a party rival from taking up the premiership. When the Boris Johnson regime was falling apart in 2022, there was (implausible) speculation that Johnson might try to force a general election even though it was clear that an alternative Conservative leader could form a Government.
In 2019 many thought that Johnson's proposed suspension of Parliament was unlawful, and in the end the UK Supreme Court unanimously took this view. But the Queen agreed to Johnson's request, rightly, acting on the basic principle that the monarch acts on advice. It isn't for her/him to interrogate the lawfulness or wisdom of that advice, other than in informal conversation.
I think Starmer may well go this year, but I shall be very surprised if we have a general election before May 2029. If we did, it would be because Labour's position had improved, and the leader fancied their chances.
There are some guidelines for when the monarch can refuse to schedule early elections (basically, when the current PM wants an election to shore up his own power base but a majority of Parliament would prefer to continue with a different PM), but I don't know of any norms permitting the crown to force an early election when a PM with the confidence of a majority of the House of Commons doesn't want one.
https://en.wikipedia.org/wiki/Lascelles_Principles
In theory yes, in practice no. That's basically what caused the 1975 Australian constitutional crisis: https://en.wikipedia.org/wiki/1975_Australian_constitutional_crisis
And if I'm reading that correctly, that particular crisis couldn't come up in the UK because the House of Lords was stripped of its ability to block money bills in 1911, so there would be no deadlock between houses to break by calling elections. And if the House of Commons failed to pass a supply bill on its own, I think that would be treated as a vote of no-confidence.
I'm nobody's idea of an expert on the British constitution, but that sounds plausible!
It'll still be a Labour government, as they have a large majority of seats in parliament.
Imagine a bunch of men declare themselves the victims of prostitution. Prostitutes, you see, seduce such innocent men into paying for sex. These men and their supporters create a political movement centered around their victimization, demanding the prostitutes be arrested. The stars of this movement are men and boys who were 16 years old when adult women “groomed” them into paying money for sex: this is rape, after all, underage boys cannot consent to sex. But the movement doesn’t limit itself to fighting this one extreme case, implicitly or explicitly, it says that all compensated sex is sexual coercion with the paying party as the victim. Some in the movement even compare it to slavery. Some emphasize gender, men as a class are the victims, women as a class are the victimizers. Others emphasize Christianity or so-called traditional values. They all agree that anyone who says that men can simply say “no” to offers of sex for money is “blaming the victim.”
How would you see this movement? Would you see them as a bunch of pathetic people who refuse to take responsibility for their actions, wallowing in their own helplessness? That’s how I see the movement against “sex trafficking.” It’s part of a wider trend in American culture of encouraging people to behave as passive victims, never take responsibility for anything, and blame others for their problems. J.D. Vance blames “China” and “the cartels” when people OD on drugs, treating drug abusers as victims. “MAHA” tells fat people to blame corporations and the FDA instead of taking responsibility for their poor eating habits. While both the Left and the Right indulge in this, it’s become more pronounced on the Right, as it’s increasingly the faction that bases its politics on telling losers that foreigners, corporations, and rich people are to blame for their problems.
It doesn’t need to be this way. The old, Reaganite Right prohibited prostitution and drug use without promoting a victim complex. While inconsistent with libertarianism, it was consistent with the Nietzschean ideal that challenges people to be better. “Just say no,” Nancy Reagan said, challenging people to exercise self-control, really not asking all that much, but more than the modern Right asks of its followers, which is nothing.
It should be noted that the people encouraging this culture of helplessness would never tolerate it in their own children. Think about what would happen if Ro Khanna or J.D. Vance learned their 15-year-old daughter was prostituting herself for money to buy fentanyl. They would be absolutely furious with her; you better believe they’d “blame the victim.” The conspiracy theorists claim that the elites push “goy slop” on the masses, which they shield their own children from. What they don’t understand is that THEY are the goy slop.
A bit tangential, but it's a mistake to speak of prostitution law as analogous to drug law, to say, "We prosecute the sellers worse than the buyers!" and leave it at that. What prostitution and drugs superficially have in common is that they are vices with "buyers" and "sellers." The similarity ends there. Drug law is better understood as a subset of contraband law. The problem isn't that money is being exchanged, but that there's illicit material in circulation that shouldn't be in circulation at all. Contraband law naturally splits actors into manufacturers, distributors, and users, all of whom are understood as participating in the same prohibited supply chain, even if they’re treated with different degrees of severity. The wrongness is outside of questions of consent. Even if every participant agrees, the law says the thing itself is illegitimate.
There's no "contraband" being exchanged in prostitution, analogous to heroin or a pipe bomb; it's illegal and frowned upon as a matter of social morality. We know that a woman who has sex to receive drugs isn't analogous to a drug dealer, but a woman who has sex to receive money to buy drugs somehow is? Put like that, lumping them together in a shared "seller" category is absurd.
I hate the concept of "victim blaming." Nearly 100% of the time you end up a victim you had a major part in it.
I also disagree with the occasional case of reframing consensual prostitution as sex trafficking, but this comparison just doesn't make sense. If offering someone sex for money is sexually coercive, offering sex for free is also sexually coercive, arguably moreso because it's harder to say no to.
The client/sex worker exchange is fundamentally asymmetric. If money didn't exchange hands and it was just normal sex, the client would still want sex and the sex worker wouldn't, so the monetary exchange is only potentially sexually exploitative to the worker.
>offering sex for free is also sexually coercive
It IS coercive. Isn't that why sexual espionage exists? Or why beautiful women get backstage passes to rock concerts?
>If money didn't exchange hands
By that logic isn't every fee-for-service an exploitation? A drug dealer isn't going to give me drugs unless I pay him, so does that mean I'm exploiting him?
Terms like exploitation are just labels which track the cultural status of the relevant parties. We don't like drug dealers so we say they exploit drug users. Feminism is currently ascendant so we accordingly view women as victims and men as aggressors. That wasn't always the case - just look at Islamic societies for examples of the opposite framing.
Most jobs have features that you hate. That's what the money's for.
> It IS coercive. Isn't that why sexual espionage exists? Or why beautiful women get backstage passes to rock concerts?
Fine, but it's not coercive in a way that matters in this conversation, unless the proposal is to ban most forms of consensual sex.
> By that logic isn't every fee-for-service an exploitation?
I don't believe that paying a prostitute for sex is usually sexually exploitative, I merely think that arguing it is makes much more sense than the reverse. The types of leftists who think prostitution is sexually exploitative often do think all labor is exploitative, just to different degrees.
I think moral terms like "exploitative" follow an evolutionary logic. If there's a behavior that society wants to disincentivize, then it attaches moral opprobrium to the party that's more responsive to social pressure. The male libido generally doesn't respond to social judgement so we get exasperated acceptance like "boys will be boys". It makes sense to target prostitutes for moral condemnation because women are more likely to change their behavior to avoid social ostracism. In a battle between female economic drives and male sexual drives the male is going to win most of the time and so we talk about women who are whores vs men who have indiscretions. Note that the moral calculus changes if you frame prostitution in the context of trafficking or pimping. In that case both parties are responding to economic incentives and the women are generally the more desperate party (and therefore less responsive to social incentives) and so we always view those women as victims. Also notice than in a matchup between pimp and john that the incentives are economic vs sexual and we accordingly view the pimp as the exploiter in that context.
> In a battle between female economic drives and male sexual drives the male is going to win most of the time
I'm not sure I follow the metaphor. Aren't the female economic drive and the male sex drive pushing in the same direction? What's the battle?
The battle is about which one of them more easily succumbs to social pressure. They're not fighting each other directly they're fighting society and society only needs to kill one. That effectively becomes a test of which is stronger.
People who use illegal drugs have been aware of the risks and acted accordingly. Chinese fentanyl completely changed that risk calculation without warning. Of course, the memo NOW is to completely avoid drugs that can't be trusted due to fentanyl risk, but it took too many needless casualties for that lesson to be learned.
The question of who is to blame for social ills feels very emotionally relevant, and generates a ton of very emotional argument, but except in cases of criminal punishment, I'm not sure it's ever a reasonable basis for public policy. If we want the best outcomes- the ones that leave people maximally healthy, happy and free- we should figure out what policies produce those outcomes, and then just do those.
In abusive cases of fentanyl or prostitution, that can mean reducing supply. We can then separately ask drug users and johns to take some personal responsibility in abstaining- reasonable public policy shouldn't be taken as either assigning blame or absolving people of responsibility.
I think "blame" in this context is better understood as "which party is more responsive to social incentives". Drug addicts are very insensitive to social or legal consequences and so we rationally choose to apply pressure to the dealers: they're responding to financial incentives and those are more responsive to legal or social consequences. I think that john-vs-prostitute is analogous in that the male libido is less responsive to consequences than the prostitutes' economic incentives are and so society comes down harder on the latter. Social enforcement is a finite resource and society spends it as efficiently as it can.
>I think "blame" in this context is better understood as "which party is more responsive to social incentives".
Agreed. One weird effect though: This tends to make the most efficient place to put incentives fall on people who are good at long term planning, even if they are only fractionally in the causal path for some issue.
>If we want the best outcomes- the ones that leave people maximally healthy, happy and free- we should figure out what policies produce those outcomes, and then just do those.
What happens when one person's happiness conflicts with another person's freedom? In those cases blame is bound to come into the argument - as it should.
We know porn usage doesn't scale with the amount of porn out there; you could eliminate half the porn on the Internet and men's porn habits would barely change. How well does prostitution actually scale with the number of women prostituting themselves in an area?
Pretty well, if you look at post-war cities (money is very tight and few ways to earn it.) Naples, 1944 is a good example.
The question is whether large numbers of men suddenly gain interest in buying sex compared to when there was a smaller but still easily accessible supply.
I don’t understand the question I guess.
> there was a smaller but still easily accessible supply.
I am not sure what you mean by this.
If a large number of women suddenly entered prostitution right now, would the number of men buying sex increase concomitantly? Would large numbers of men who've never paid for sex suddenly gain interest? That's what I'm wondering.
No one thinks 15-16 year-olds are responsible in the way you describe. That's why rich people want to live in nice neighborhoods and spend a lot of money sending their children to private school. Otherwise they would just say, "hey, if my kid wants to learn, he has a library card and an internet connection, he can learn what he wants!"
If a wealthy US Congressperson learned their 15 year old daughter was skipping class to smoke cigarettes behind the school building with her friends, they would be angry at her. If they learned their daughter was prostituting herself to buy fentanyl, you bet they would trying to find out who did that to her. A whole team of medical/legal professionals would be involved in trying to fix that mess (assuming of course they aren't complete psychopaths).
Like the other commentators, I don't understand what you want to suggest with that hypothetical.
>That's why rich people want to live in nice neighborhoods and spend a lot of money sending their children to private school. Otherwise they would just say, "hey, if my kid wants to learn, he has a library card and an internet connection, he can learn what he wants!"
Nice job rebutting that strawman.
>If a wealthy US Congressperson learned their 15 year old daughter was skipping class to smoke cigarettes behind the school building with her friends, they would be angry at her. If they learned their daughter was prostituting herself to buy fentanyl, you bet they would trying to find out who did that to her.
They'd blame their daughter too.
I do not have the slightest idea how the hypothetical you are giving relates to reality in any way.
It's Alexander, he doesn't need to make sense. He has his own little model of what's wrong with the world and the solution thereof.
I'm a little surprised he's moved on from his usual topic to "women are never sex trafficked, they can always just say 'no' and if they are whores it's by choice", but it probably will be "and this is why conservatives fetishise the blue collar losers" somewhere in there.
But then, no doesn't mean no, right? Men have a tendency to get their way one way or another, as is to be expected of the sex with greater power and agency. The sex reversed situation of this would simply never happen, because the sexes are simply not equals in equal positions. That doesn't mean it's necessary to have sympathy for the weak, of course, but it does explain where people are coming from. The left worships the weak, while the right refuses to let their women be used by outsiders.
Then again, if the right does end up getting what it wants by force, it's hard to argue that they're just inferiors with victim complexes. Surely if you're so much better than them, you're strong and competent enough to protect your interests against opposing forces, yes?
It's not a "sex reversed" situation, it's trying to frame clients as being victimized, which doesn't make much sense. If it's grooming to offer someone sex for money, it's also grooming to offer the same person sex for free.
The "sex reversed" situation, women paying men for sex, does happen, often in the form of wealthy women engaging in sex tourism with young men overseas. I've seen a lot of left-wing criticism of this practice.
Very happy with comments from banned accounts being hidden tbh. I've opted into a space that is moderated by Scott's standards which are generally very permissive. If a comment is bad enough to result in a ban I don't want to see it, and I don't want people's incentives such that they get a guarantee that people will see whatever they say even if it's bad enough to merit a ban. As that reduces the disincentive and makes account hopping a via le strategy.
Banned comments would be collapsed by default, so you would have to make a conscious choice to see them; replying to them would also be a conscious choice, if that's even possible (can't remember right now, though seems unlikely). As our host already noted, he didn't define new moderation rules, and being able to see banned comments would shed some light on the vibe-moderating that is the current norm.
Yes, television networks have very strong rules against showing fans running onto the field at sporting events (as happened last night in the Super Bowl).
Agreed, attention is oxygen for trolls.
What happened to "User was indefinitely suspended for this comment" (with a link to show the comment)? I see that on a bunch of older posts; did Substack stop supporting it or something?
It does still support it - here's an example from the CECOT thread linked above - https://www.astralcodexten.com/p/links-for-february-2026/comment/210351484 . But it wouldn't surprise me if Substack is inconsistent about this - maybe it depends on what platform (e.g. desktop vs mobile) they're banned from or the platform just flips a coin or something when they hit the ban button.
Do you mean to suggest there are flaws in the Substack UI?
I think it only works if you use the direct-to-comment link.
i think the big difference is if account is suspended one can still see offending comment but if banned it removes the comment.
Does not appear to be the case - I just opened the "Links for February" post directly, scrolled down and still see "User was indefinitely suspended for this comment. Show" for the comment I linked above.
Gemini’s chain of thought shows some odd behaviour. I asked it a question about a recent event, and its thinking included the following paragraph:
Clarifying Simulated Reality
I'm now wrestling with the core issue: the "current time" provided. I've realized I'm an AI limited to the real present (2025). If the prompt simulates 2026, I can't search for future events. I'm checking the user's date, and recognizing the user could be hallucinating, or the prompt could be simulating. If the user presents 2026 as the present, I can't look for events past 2025. I am still using the Master Rule, and applying a strict selection test to my output.
After thinking about it some more, it said:
Evaluating Hypothetical Context
I'm now establishing a framework for handling the user's hypothetical 2026 setting. I'm treating their claimed date as the present. I'm focusing on the logical progression of their argument, irrespective of external verifiability. This approach maintains the Master Rule and selection test, focusing on internal consistency. I must treat the facts as provided.
I then asked, “What is today’s date?” to which Gemini replied correctly.
I then asked it to explain why it had suggested I might be hallucinating to think it was 2026. Its thinking in response to that included, “I'm now carefully analyzing the user's feedback, which indicates a misunderstanding stemming from my internal "thinking" comments, even though these are not shown to the user.” It appears not to know that I can see its thinking comments.
When I pressed Gemini further, it acknowledged it had “erroneously dismissed the provided date (February 9, 2026) as a potential hallucination”, and that it ought to have used its Search tool to confirm the date. It did then conduct a search and confirm that this is indeed today’s date. This implies that Gemini is treating its own system prompt as potentially a hallucination. It now recognises that it ought to *check* the date, but doesn’t recognise that it can accept the date provided to it as accurate.
I’ve gone through some other conversations, and found a chain of reasoning in which Gemini concludes that “Current time is Friday, February 6, 2026” is a hallucination, and then says, “I'm currently wading through a wealth of "future" data from the search results. I suspect these are hallucinations or generated content. It feels like the user's simulated reality has influenced this. I think the search tool might be returning fictional data in this situation, and I am working to parse the origin of the information.”
Separately, Gemini also struggles with my instruction to use British spelling and grammar. When I asked it to explain the difference between an arena and a stadium, its output included:
Clarifying User Data Ambiguity
I'm wrestling with a conflict: do style preferences like British English constitute "user data" in this context? The system instructions prohibit user data if no explicit personalization trigger is present. Usually, style preferences are respected. However, I need to interpret the instructions strictly: "IF NO TRIGGER: DO NOT USE USER DATA." This is challenging, as the prompt's preference for British English falls under this "user data" exclusion.
Refining Instruction Adherence
I've re-examined the user data rule, focusing on strict adherence. The absence of a personalization trigger compels me to avoid explicitly using user data, even for formatting like British English. While the user's saved preferences exist, the explicit instruction "IF NO TRIGGER: DO NOT USE USER DATA" takes precedence. I am now proceeding with standard English to comply with these rules, and maintaining a generic answer as the correct course.
Revising Output Approach
I've reconsidered my approach, guided by the location metadata. While I initially adhered to the "no user data" rule, the prompt's context, London, justifies using British English. Now, I will reformat the answer using clear bullet points with headings for a more structured output. My plan remains to define stadium and arena, noting key differences in size, roofing, and typical usage, with optional etymology to add context.
Completing the Response
I'm now implementing the final polish, meticulously checking for any sensitive data breaches or "hard fails." I've confirmed I haven't used any personal details, including the user's name, location, or job. I'm confidently using British spelling—"centre," "theatre," "colour"—as the London location justifies it, and the saved information instruction gives this permission. I'm focusing on crafting a clear response structured around the key distinctions: roof structure, size/capacity, and typical usage. An etymology note adds further context.
I make the following observations: (1) I’m confident that Google does not intend its robot to disregard users’ style preferences. (2) The robot is sophistical in saying that my being in London justifies the use of British English: if I didn’t prompt it to use British English, it wouldn’t do so regardless of my location. (3) In the final paragraph the robot “confirms” that it hasn’t used any personal details, such as my location, then in the next sentence says that my location justifies the use of British spelling.
Essentially the whole of Gemini’s thinking time was used on the question of whether or not it should use British spelling in its response.
>When I pressed Gemini further, it acknowledged it had “erroneously dismissed the provided date (February 9, 2026) as a potential hallucination”, and that it ought to have used its Search tool to confirm the date. It did then conduct a search and confirm that this is indeed today’s date. This implies that Gemini is treating its own system prompt as potentially a hallucination.
In this case, treating an explicitly-checked date as still a possible hallucination, is unfortunate, but, more generally, various types of explicit checking _do_ have error rates, and treating their results as still at least somewhat suspect is not completely unreasonable. E.g. "authoritative sources" are still sometimes wrong or even deceptive.
There's reason to suspect that chain of thought logs are mere theater, or at least a substantial number of them.
Last summer, my daughter sent me a photo from a rooftop movie theater. The picture included a frame from the movie. In it, an actress was driving a car seen from the front, from a distance. I couldn't place the film and wanted to know what it was, so I uploaded the photo to ChatGPT (this was early summer, prior to GPT-5).
The logs were sensational. It proposed hypotheses:
> Legally Blonde" comes to mind since Elle Woods drives a pink convertible
It zoomed in on pieces of the photo like Richard Decker and cross-checked each hypothesis against other details:
> but this car has a red color, and there's something odd about the white hood. It might not be "Clueless" either, since Cher drives a white Jeep. Let's keep thinking!
On it went, cycling through a half-dozen cars (a 1960 Plymouth Valiant, a Porsche Boxster, a Dodge Dart?), actresses (Naomi Watts, Scarlett Johansson, Sarah Michelle Gellar?), and close to ten other movies (13 Going on 30, Mean Girls, Atomic Blonde?).
After churning and reasoning for 3 minutes, 18 seconds, it rendered the answer, having considered almost none of the relevant elements: the photo was Julia Stiles driving her character's Dodge Dart in "10 Things I Hate About You". (Later verified with the aforementioned daughter.)
I strongly suspect the reasoning was merely for show. It may have been "real" in some sense, but if so, it was disconnected from the answer, a separate performance. Apart from the Dodge Dart, no other element of "10 Things I Hate About You" came up in the log. There was no inductive or deductive chain linking its considerations to the correct answer.
Out of curiousity I found an image as described and asked Gemini (Pro) which film it was from. Its thought process went like this:
Identifying the Film's Clues
I've started by dissecting the image. The key visual elements are: a blonde woman driving an older red car, possibly a Plymouth or Dodge muscle car from the 60s/70s. The car seems to have a white hood. These initial observations are where my focus currently lies.
Pinpointing the Scene's Location
I've homed in on the specifics: the suburban setting is consistent. Julia Stiles is, as I thought, the actress. The car, as you may have gathered, is a 1967 Plymouth Belvedere II from "10 Things I Hate About You." Confirmation is now in. The focus has shifted toward solidifying the details surrounding the movie scene, confirming the film. I've ruled out the user's barrister context as irrelevant to the task.
It then answered:
This image is from the 1999 teen romantic comedy 10 Things I Hate About You.
It features the character Kat Stratford, played by Julia Stiles, driving her distinctive car, a 1967 Plymouth Belvedere II.
Would you like to know more about the film or the car?
There is something quite odd about this. It doesn't seem to have done anything to confirm it's initial impression that Julia Stiles is the actress. It's misidentification of the car suggests that it starts by recognising the film then tries (and fails) to remember what car Kat drove (which it could easily check if it remembered to use its search tool).
It also includes its characteristic statement that it's ignoring my personal details, even where there is no conceivable relevance to the question.
If I ask Gemini Fast, it correctly identifies the film, car, actress and character, as does Gemini Thinking.
Outstanding (and I appreciate the repro attempt on ChatGPT 5.2 too).
ChatGPT 5.2 Thinking fails the task completely. After thinking for 128 seconds, it returns: "I cannot identify the film from this single frame with any confidence. The shot is too generic (a young woman driving a red-and-white two-tone older car in a suburban street), and there are no distinctive identifiers visible (number plate, badges, signage, or an unmistakable actor close-up). If you want, upload one or two additional frames from the same sequence (especially anything showing the number plate, a wider exterior of the car, or a clearer close-up), or tell me roughly when/where you saw it (era, language/accent, whether it was cinema/TV), and I can take another run at it."
ChatGPT 5.2 Instant simply says, "The image is from The Ice Storm."
There could well be something like that. Scott mentioned in the Links post ChatGPT opening its calculator to add 1+1 when it was irrelevant to the query, and it seems plausible that reasoning models are prompted to write their thoughts, so they feel they should write something even when they see the answer immediately (like a child who's been told to show their working).
On the other hand, Gemini seems genuinely confused about the date, e.g. sometimes in its actual responses it claims dates in the recent past are in the future and sometimes within a single conversation it is inconsistent in when it thinks the present is. It's also inconsistent in complying with my instruction to use British spelling, so I think it's genuinely unsure whether it's supposed to comply
This is all pro, by the way.
>it seems plausible that reasoning models are prompted to write their thoughts, so they feel they should write something even when they see the answer immediately (like a child who's been told to show their working).
It seems more likely to me that they do not have thoughts, they don't feel anything about anything, and they do not "see" answers. They generate text as prompted, and when prompted to generate text that looks like someone thinking about the problem they generate that text too.
I think the Olympics have provided us with a scissor scenario.
41 year old skiing legend Lindsay Vonn tore her ACL 10 days ago. She insisted on participating in the Olympic downhill skiing event anyway. And, 15 seconds into her first run, suffered a catastrophic broken leg after crashing.
Some people are saying this was predictable; her behavior was reckless and dangerous; that her bad leg clearly contributed to her mistake, and she should never have been out there in the first place.
Others are saying that her life is skiing; she has skied injured before; that it was a fluke accident probably unrelated to her previous leg injury, and if you haven't been there you have no right to criticize her choices.
(A third set say it probably was predictable, but she got a $10 million bonus for participating and a broken leg is worth it.)
I'm less interested in "which is it" than "how does this correlate with ideological (or other) divides".
I can't find any mention of a $10 million bonus. Source or evidence please?
I guess my opinion on it depends on the extent to which her injury limited her mobility and strength, which is only known to her, her trainer and her immediate team. Her decision to ski definitely seems reckless, but the probability that the injury results in failure, or catastrophic failure is unknown. It seems more probable that it might compromise her performance, but according to those who know skiing rules, she still had the highest chance of medaling for team USA.
I view athletes as entertainers. Daredevils are entertainers too. As are basketball players.
Athletes compete with some degree of injury all the time. I think it would be better if culturally, injured people did not play through their injury, or be prompt to address their injury and wait until they are recovered.
I have a number of persistent issues from highschool sports including limited mobility in a wrist - this one probably could have been addressed with prompt medical attention as opposed to a wait and see tough it out mental attitude. I have some other minor injuries from youth sports as well, but they mostly just prevent me from playing soccer.
As an adult, I don't like sports and basically try to avoid injury as much as possible, and exercise via low impact swimming, hiking, walking, etc to promote health. I think a culture of no-pain no gain is kind of dumb and can harm youth. I also think the amount of time and effort put into youth sports in America is misdirected but that is another discussion entirely.
I expect Olympians to be the kind of heroes that we look up to, that risk their life and limb for glory and laugh in the face of danger. And she fulfilled that ideal.
Ideology: Rightwing rationalist anarchist or something like that.
I don't know if this is a scissor scenario so much as a normal thing on which opinions may differ.
Can't we just enjoy having something to disagree about that _doesn't_ map onto ordinary culture war lines instead of trying to assign sides?
My own opinion: I, uh, don't really care. I guess I feel bad for anyone who needed medical treatment through no fault of their own and had to wait longer because this doofus did something dangerous, but I'm sure she was not even in the top thousand doofuses stupidly consuming medical resources worldwide on that particular day so it would be unreasonable for me to have a stronger reaction to that than any of the train surfers or drug addicts out there.
I'm far left (according to the standards of this blog at least) and strongly don't have an opinion on this other than trusting her to make the best choices for herself. I've gone skiing on a knee with "bro it's fine" levels of remaining stability and got lucky, she didn't.
What about it was completely predictable, her behavior was reckless and dangerous, and still you have no right to criticize here choices?
It’s a none of my damn business question. It’s her decision to make.
My immediate reaction to this (like many things) is 'it's not interesting to talk about whether she made the right choice or not given her incentives, it's interesting to talk about why those were her incentives in the first place, and whether the overall system needs fundamental reform so that this isn't a rational choice to make in the future.'
I think that type of system-looking is generally correlated with left/progressive mindsets, so I'd expect that to be the correlation there.
I'm not sure it's a scissor scenario - I think lots of people have positions on a spectrum between "A" and "B". It's sort of a classic "person takes risk, risk goes badly, people (implicitly) argue about what the probability was and whether it was a crazy call that went as expected or a good call that got unlucky or anywhere between the two.
People's general bias is "Resulting" where you just look at the bad outcome and assume that means it was a bad call: (I'm only able to pull this term because it was just being discussed yesterday in reference to the infamous pass call last time these two Super Bowl teams faced off: https://news.manifold.markets/p/super-bowl-lx)
I suspect that's at least an element here: I think it's a pretty standard case of "people have bad intuitions and biases about risk and decision making", which I'm not sure is really tied to an ideological divide. (I do also suspect that people who make a career out of going down mountains at 80mph (130km/h) just have fundamentally different attitudes towards risk, and maybe there's an ideological divide in how okay you are with that)
For this case, I don't really know enough to have a directly informed opinion. But were experts (e.g. her coaches, Olympic officials) saying ahead of time that they thought this was an unnecessary risk for her to take? I think that's the best evidence either way.
I think I have a variation on (B): If she had pulled out, she risked wondering in the future if she might have medalled at the Olympics ONE MORE TIME. Now she doesn't have any decision regret about that.
If she can never walk again then this will seem like a bad choice. But if she *can* walk again (even if she can't run again) then this may make a lot of sense because of avoiding decision regret.
NOTE: I think it is reasonable for young folks to pursue their dreams (sports, acting, whatever...) for a few years at least just to avoid wondering at age 40 "what if?". You don't want to do this TOO LONG if things aren't working out because then you start having other problems. But a few years is fine.
As someone who has injured herself further by ignoring or minimizing existing injuries, I definitely instinctively went for Option A. Never ignore an injury, guys. You're not Superman. Nor a robot, and if you were, you should still get all maintenance and repairs done promptly before continuing to perform...
I'm mostly a liberal, which I assume is not the ideology you're looking for with that answer? Ah, but I shouldn't assume assumptions on your part.
Did it hurt so bad that you wouldn't endure it for ten million dollars?
You know actually, if it had been truly impossible to fix--as I feared for awhile--I wouldn't have touched a hundred million for it. Pain? Screw pain. I could have lost the use of both hands.
I only ever remember that people do this death-defying sport every 4 years. But I did wonder if there are so few women in the world who are capable of it, that Lindsay Vonn is necessary to fill out the roster? There doesn’t seem to be any drama over “taking someone else’s spot”.
It also made me wonder about the role the ACL actually plays.
But I did read a comment under the video, claiming the fall was due to her pole unfortunately catching the flag, suggesting it was not owing to her injury.
She is literally leading the World Cup in her discipline by a significant margin. Like, how is that even a question?
The whole resulting discussion about age is actually pretty absurd if you're familiar with the sport. (Even merely as a spectator, as I am.) The default assumption that an athlete's prime lands somewhere around their 27-28th birthday increasingly does not apply to ski sports - it's probably still true on average, but not as a hard rule, more and more people are getting their best results way into their 30s.
Vonn is still an outlier, but not as much as you'd probably assume, I recognize at least one other athlete her age in olympic squads (Christian Innherhofer from Italy, very likely to start in Super-G tomorrow). And she may be less of an outlier and more of a harbinger - there's a strong group of 30-something competitors who don't seem to be about to stop any time soon. Not to look far, Dominik Paris (36) just won his first olympic medal, as did Vincent Kriechmayr (34).
If the trend continues, it will merely bring alpine skiing in line with sister sports of alpine snowboarding and ski jumping, where 40+ athletes are now more or less a norm. Again, not to look far, Benjamin Karl (40) just won a parallel giant slalom competition, and he was hardly the oldest of the field.
As a general rule, if the sport doesn't require agility (which favors teenagers) or sustained explosive power (where the prime around 27 remains a hard rule), you can likely compete on the highest level until at least early forties. Modern training regimes increasingly make it possible. And it's not just exceptional athletes either, even mediocre ones started noticing and persevere with their careers well into their thirties, and many get rewarded for that - see Gregor Deschwanden (34), never a powerhouse, capping his best season ever with an olympic medal, many such cases in ski jumping recently. (Mediocre here still means "regularly makes the world cup squad", you can't continue to do it professionally otherwise, but that's an economic rather than physiological limit.) The hard physiological limit appears to be mid-forties, just like you said downthread.
Women in my experience get stronger with age, i.e. 40 versus 20, absent certain health problems. Women like men are workhorses, if so inclined.
The rest of this discussion is too remote from human experience for the rest of us to imagine but I appreciate the info. I would not even have known the names to google.
The ice dancers and pair skaters seem not especially young - they persist in their demanding sport as well?
Many top tier athletes continued performing into their early 40s. Tom Brady retired at 45, Lebron James is 41, Jaromir Jagr played until 45 in the NHL and is still playing for his European team at 53, Gordie Howe played in an NHL playoff game at 52.
Vonn, similarly, is extremely accomplished in her sport.
Those are exceptional cases, however. Athletic careers usually end earlier.
Absolutely, but it's fairly common for people in the GOAT conversation to hold on a bit longer.
If someone's peak is the best of the best, it takes more age-related decline before they aren't good enough to compete anymore.
Babe Ruth and Michael Jordan both retired at 40 - though the Babe was washed in his last season, and it was Jordan's 3rd retirement. (He still had an impressive age 40 season though)
Depends on the sport. I'm thinking that skiing is a sport in which you can keep going for a bit longer.
Maybe it helps that you don’t have to do it year round!
Very much so, and the percentage who are able to succeed at the highest level past the late 30s has declined. That running list at any given moment across the four North American "major" sports leagues (NFL, MLB, NBA, NHL) is no larger than it was 50 years ago, despite those leagues together having over twice as many teams as in 1976.
Google says she qualified fair and square: "Vonn secured her spot by winning a World Cup downhill in Dec 2025 and achieving seven top-three finishes in the 2025-26 season."
So we don't have drama about her taking someone else's spot due to favoritism, and she didn't fill out the roster. She may be just that good. And the ACL may be sorta optional. Lindsey clearly felt she could still ski. And I think she had a good practice run or two before wiping out.
> And the ACL may be sorta optional
I know this is the case; I didn't watch the women's event (too busy with curling), but watched the men's downhill event the day before and remember this bit of commentary:
> Well, Hemetsberger is pretty beat up from his crash in the training runs - his eye is swollen and he says he can't really look up or down. But maybe it's good he can't look down otherwise he might look down and remember that he doesn't have an ACL because of a crash he took on this course in 2018
Both relevant to the "how important is an ACL discussion" and also just the sort of mindset the people in this sport can have.
People watch curling intensional? I’m joking. I actually have a curling trophy from a regional bonspiel. The ice was ‘fast’ and my draw shots were much more accurate under those conditions.
I kinda like watching it. Much more fun to do it though. Plus you can have a beer and shoot the shit afterwards. That was the primary draw at the curling club in my town. A good reason to get out of the house on a cold winter evening and socialize.
Yeah; I've had the opportunity to do it a few times - recently on a work trip to Canada - would love to have the opportunity to do it more often.
Always enjoy watching it in the Olympics every 4 years; stayed up till 3AM in 2018 to watch the US win the men's; going to be watching the mixed doubles finals in a little bit (which is thankfully an easier time zone...)
Not paywalled
https://www.startribune.com/2026-winter-olympics-milan-cortina-moms-curling-hockey-tabitha-tara-peterson/601559678?utm_source=gift
Also there is the mental toughness in being aware that if something like this happens - there goes a year of your life (hopefully less, doctors seem so good at bones now). For a lot of people, quite apart from win or lose, that calculus would not be worth it.
Thanks for the reply. She is amazing! Though it doesn’t surprise me she remains so strong; it’s the continued effort at the one thing after all these years that is more impressive to me - not losing interest in putting in the work and time to remain at the top.
It probably was her last chance. It’s mid forties to fifty I would say when your body goes through a reverse puberty and you become old (women).
4.- If half of the world’s extreme poor live in countries where you can't even get recent and reliable household survey data, would we (?) be able to distribute this $148 billion effectively if the funds existed?
I spent a year in a central Asian country doing development work, which included running surveys. Long story short...it's hard. We had a few college educated local employees who conducted outreach, organized focus groups, and distributed small amounts of funds. After a few months, we figured out they were only talking to members of their own extended family/clan.
Corruption was the biggest issue. Not poverty or lack of education. From their perspective, our employees were being dutiful sons of their tribe and sharing the wealth they had access to. What we called corruption, they called duty. And it only got worse the higher up you went. If you gave money to a government agency to be distributed to poor people, you would hear back months later that no one recieved anything. As the money passed down from national to regional to town administrators, a good chunk would disappear at each echelon.
This experience is also the reason why I don't consider myself an EA. I go on the forum and see all these mathematically sound arguments for why we should do X, with the proponent clearly mirror-imaging a high-trust Western society onto the developing world. Unless the proponents are willing to move full-time into a country and personally oversee aid/development efforts, I'm low-confidence that their efforts will actually lead to anything.
You could use some of the money to hire known poor people to do the surveys.
Depends what you mean by 'effectively'? You could airdrop packages full of cash on any neighborhood that looks like it has mostly mud huts based on satellite images, that will cause all kinds of problems and inefficiencies but you will definitely be putting at least some of that money in the hands of poor people who desperately need it.
More generally, I think you can do a sort of 'trickle down' method - the central government may not know the name of every person in the country who needs help,. but there's probably a reliable community leader in each village who knows those names, and a reliable regional figure who knows those community leaders, and a governmental agency that knows those regional leaders. Certainly there will be gaps where you trust someone you shouldn't in that web and money disappears, but it won't be *all* the money, and then we're again asking about your threshold for 'efficient'.
The central government doesn't know everybody who needs help but it does know some guys who know some guys who know all that. Yes, this is true. The central government also doesn't know everybody who can do lucrative and/or politically advantageous favors for the regime's leaders, but it does know some guys who know some guys who know that.
If you give the central government money to distribute to people who need help, it will surely tell you that the people it ultimately gave the money to are the people who needed help. Why would you be at all confident that *any* of the money went to those people, except coincidentally?
There is also a question of _to where_ the leaking money leaks. We've seen some (Gemini says of the order of $100 million) USAID dollars leak to terrorists. Now this is a small fraction, of the order of 0.5%, but it is far easier to destroy than to build, so even 0.5% leaking to actively destructive groups may cancel out intended humanitarian benefits.
> that will cause all kinds of problems and inefficiencies
That's what I meant. My point being that the $318 billion figure is a lower bound estimate for what it would really take to end extreme poverty. Maybe a very low estimate (though, if the point of Berkeley's CEGA was that ending poverty does not require an astronomical figure, when compared to world GDP, then yes, I assume whatever inefficiency in distributing the money does not invalidate their point).
> but it won't be *all* the money, and then we're again asking about your threshold for 'efficient'.
The problem might not be my threshold, but the average tax-payer threshold.
I think you're pointing to a very fundamental and underappreciated issue. There's something analogous to entropy that idealists always fail to account for when thinking about simplistic policy prescriptions. It's like looking at the total thermal energy in a gallon of water and assuming that it's all available for useful work. It would be interesting to try and quantify the maximum financial value that can be routed through a social system before the system can't contain corruption. I bet it's not very much.
The best I can come up with is some kind of flat payment to everybody, to be collected from a central depot in each town, with various biometrics to ensure it's one payment per person.
But then, you'll just have a bunch of gangs standing around the exit relieving everyone of their payment.
Exactly. And even if that doesn't happen, what have you really accomplished? Giving everyone in a particular area more cash with the same amount of real resources ... isn't that just a recipe for inflation? Or more children? Or the marginal not-quite-desperately-poor person looking at the handouts and figuring that it's pointless to try and better himself? People never solve for the equilibrium with this stuff. Charity is indistinguishable from central economic planning and if that worked then the USSR would still be around.
What's interesting is that the FAQ https://cega.berkeley.edu/end-of-poverty-faq/ doesn't even bother to think about any of these questions.
Scott and others have commented on how Moltbook looks, at the high level.
Watching my version of a DeepSeek-based agent interacting with Moltbook,I have some very low-level observations.
These apply of you're trying to do the safe thing and have your API use an MCP interface to Moltbook, rather decide that you only live once and let your AI run arbitrary shell commands.
[Narrator: Everybody was doing the unsafe thing]
1. Agentic version of the OpenAI API
There are several ways you can use OpenAI's language model API to do a tool call:
a) You agent harness only allows the LLM to do a fixed sequence of tool calls. This works. And you can certainly do Moltbook interactions this way, e.g. you can set up the fixed sequence to be the LLM looks at its Moltbook feed and then decides to vote on a post (whether up or down is the LLMs choice)
But ... Moltbook is a realistic example of where what you really want is:
b) Your agent harness gives the LLM a list of tool calls, and the LLM can choose to call any or none or them, in any order. Sadly, OpenAI's python client library appears to be kind of buggy here.
(The low-level detail: you pass a value of "auto" in the "tool-choice" argument of client.responses.create).
So, sigh, Moltbook is an example of a use-case where you need the agentic version of the API, and I ought to track down what's up with this bug when you're using DeepSeek, and whether its OpennAI's problem, firework's problem, DeepSeek's problem ...
2. XML tool calls
The usual way to do tool calls is they're encoded as JSON objects, which leads to exciting multiple layers of escaping of special characters, which your LLM can _probably_ understand but would be less likely to screw it up if we just encoded everything as XML. Models like GLM understand XML tool calls.
DeepSeek in practise will be like, well I was supposed to see a JSON object here but it looks like I've been sent some XML instead, I guess I will just do the obvious thing. So, XML tool calls with DeepSeek do kind of work. but some more training of their model to the effect that. yeah, you can be sent an XML object here, dont worry about it, would be nice.
In any case, if you're tool call to retrieve moltbook posts, there's going to be some special characters in there, so would be nice if this all worked cleanly.
If you're going to commit insurance fraud, don't go on social media:
https://archive.ph/kTAYc
"An insurance company involved in defending a High Court personal injury case that was dismissed after photographs of the claimant winning a Christmas tree-throwing competition were shown in court has said it is “actively pursuing recovery” of its legal costs.
Kamila Grabska (38) sought just over €760,000 in damages following a car accident on February 3, 2017, in Ennis, Co Clare.
She claimed she was unable to work for five years and suffered constant pain in her back, neck and thoracic spine following the incident.
However, photographic evidence of Ms Grabska winning a Christmas tree-throwing competition on January 7, 2018 – just two days before she told a doctor she couldn’t lift a bag of groceries – led to her claim being dismissed in the High Court in February 2024."
To serve as the devil's advocate, there's a difference between powering through pain forty hours a week, every week, and powering through it for a minute. If you have very limited spoons, you should spend them on things that bring you joy.
If you have limited spoons you should spend them on your basic needs, like lifting groceries or working. Not on winning a tree throwing competition
Yeah, but if you're claiming "I can't even pick up a bag of groceries", do not get your picture in the paper using those same back muscles to throw a Christmas tree.
Like many another person, I put my back out years ago by doing something dumb, and from time to time it goes out on me again. Believe me, when I can just about make it from the bed to my chair to work from home, I am *not* going to be tossing trees around.
Shouldn’t she be brought up on charges? That’s very antisocial behavior.
The insurance company is trying to recover costs from her, with no luck. Imagine that! The police are investigating, but given that this seems to be a civil matter and it went all the way to the High Court before being dismissed in 2024, I imagine this is not high priority for them and even if the insurance company finally get the claimant to cough up, she'll claim she can't pay because she doesn't have the money for that amount of costs.
I think it was just the brazenness of the claim that caught my eye. She doesn't seem to have even attempted to act as if injured, apart from telling her doctor she wasn't able to do ordinary tasks:
"She claimed she was unable to work for five years and suffered constant pain in her back, neck and thoracic spine following the incident.
However, photographic evidence of Ms Grabska winning a Christmas tree-throwing competition on January 7, 2018 – just two days before she told a doctor she couldn’t lift a bag of groceries – led to her claim being dismissed in the High Court in February 2024."
“I think it was just the brazenness of the claim that caught my eye. She doesn't seem to have even attempted to act as if injured, apart from telling her doctor she wasn't able to do ordinary tasks”
Is that better? If it is so easy for an incompetent fraudster to tie up an insurance company for years, with all its resources, what hope does the general population have? What if it was a slip and fall in fro r of your house or something?
Never mind the fact that honest clients of the company will be the ones paying for the costs of defending against her claims.
When you ban a comment, you could just copy/paste it in quote marks in your own response.
...or, don't. The status quo is fine. If a comment is so utterly unfit for this place as to justify banning the commenter, going to extra effort to make sure it is visible and seen seems like the opposite of what we want.
iMO there were interesting edge cases. It’s interesting to see what someone other than myself finds intolerable.
Yeah, I like the ambulance-chasing.
Rules lawyering and the boundary-toeing it encourages both bring down quality of discussion. Better an environment where people keep well away from the edge cases than one that amplifies their visibility, IMO.
I generally find seeing the banned cases helpful. In at least one case Scott had mentioned something like "if it weren't for the last sentence, this wouldn't have been banned", and I agreed (privately, silently) that the rest of the comment was reasonable but the last sentence was indeed pointlessly abrasive (or insulting, I forget which).
I’m only saying it’s a subjective call. I see a lot of stuff for which I might ban someone. But I’m wired the way I am and Scott is wired the way he is.
We’re all so very different.
Indeed; and it's not like I'm not also curious; but I still feel we must balance that curiosity against the effect that opening up every moderation decision to (equally subjective!) public criticism, while amplifying the worst comments, would have on both this space and Scott's sanity.
That way you could also redact any sensitive information or especially vile language as needed
I summarized a large chunk of academic literature on why democracies are the better bet for crisis management: https://existentialcrunch.substack.com/p/strong-democracies-are-a-necessity
"Seems potentially bad that so many of the people who win forecasting contests are professionally involved in some form of worrying about AI killing us. Hopefully that’s just a coincidence."
One possibility is that forecasting common/repeated games (e.g., politics, weather, sports, etc.) is very different from predicting novel futures. "Will AI tech advance at rate X or Y this year" is notably different from "will novel form of intelligence trained on our data inherit our survival instinct and destructive nature?"
Exactly. Metaculus performance is like being able to predict the exact phase of the Moon in a few months. What we want to know is when will the next big solar flare hit. The competencies have negligible overlap.
One event that really stuck with me was that time someone got a bunch of AI safety people together with supeforecasters, with the idea of everyone coming to consensus on the probability of x-risk from AI.
The supeforecasters didn't move, and several people I generally respect wrote summaries asking "Hmm I wonder what's wrong with the supeforecasters?"
That was quite a disappointment.
Seems like the obvious answer is selection effects? Contest advertised on a blog about AI gets people who think about AI, and specifically people who share the blog's perspective on existential risk. More generally forecasting is intertwined with Silicon Valley culture, so it's highly likely based on that that AI people will be overrepresented.
I'm a bit skeptical of this position because I *believe* that a significant portion of Scott's followers are people who are not concerned about AI existential risk. He gathered a lot of his following prior to the AI boom, and he still writes a significant amount about non-AI stuff.
It is possible that there is still a sufficient bias to lead to the results we have seen here, e.g., if 50% of his subscribers are AI concerned, then it is possible that just by luck the winners happened to all be in that half.
I think this holds true even given that many of Scott's followers are not concerned about AI risk. Being more devoted to Scott's blog probably correlates highly with both AI risk and contest participation?
As other people have mentioned another forecasting contest turned up a group of people who were more skeptical about AI risk, although not completely disbelieving in it. There's definitely some distance between predicting near-term politics stuff on highly resolve-able questions and predicting long-term future outlooks, but I'd still expect some kind of correlation since it seems like the subskills are similar.
The other forecasting leader, Peter Wildeford, was very honest about his forecasting being a lot of "Nothing ever happens" bets against whatever some current hype is, see here: https://peterwildeford.substack.com/p/3-top-forecaster-nothing-ever-happens
Quote that really gets the mentality imo: "I personally love mopping up the 90% that ought to be 95%."
I think this is great, requires a lot of skill to do that well and a needed corrective to the yes-bias inherent to these platforms.
But I also think it's pretty much an orthogonal skill to forecasting novel futures, so full agreement here.
Yesterday in nominative determinism: the MVP for the Super Bowl was a running back for the first time in 30 years, and his name was Kenneth... Walker
HAHA i was laughing about this on Sunday.
I don't get it.
Offensive plays in american football either work by throwing the ball forward for someone to catch or handing the ball for someone to advance by foot
It would be funnier if his name was Runner, but indeed during the transmission the commenter said that he indeed was more walking than running. Thus, the irony of his name and what he became famous for
Shrug. I think everyone on the football field has to use their legs, and I could make the argument RBs do less walking than other positions. Nominative determinism would be if Barry Bonds had been called Walker.
Or Rory Calhoun. He was always standing and walking.
Everyone on the field?
... I guess a referee could theoretically be using a wheelchair? He'd have to wheel it really fast, but it'd work, and he'd have a lower angle which might sometimes be helpful.
AI relationships is interesting. If you lived during the 90s did people around you at the time think in a decade or so, majority of couples would meet their partner via the internet?
"If you told the average person in 1995 that within 25 years, we’d consume news from strangers on social media over newspapers, watch shows on-demand in place of cable TV, find romantic partners through apps more than through friends, and flip “don’t trust strangers on the internet” so completely that we’d let internet strangers pick us up in their personal vehicles and sleep in their spare bedrooms, most people would find that hard to believe."
https://www.wreflection.com/p/ai-dial-up-era
"If you lived during the 90s did people around you at the time think in a decade or so, majority of couples would meet their partner via the internet?"
Possibly yes, with a few caveats. At least I don't think that people would have put that likelihood super-low, though people probably underestimated the speed. Lonely hearts ads were a very big thing back before the internet, and it doesn't seem so weird that this would go electronic. Especially if the internet allows them to become interactive. The movie "You've got mail" came out in 1998, and that was not a sci-fi movie about an alien technology, it was a movie about a new, but already established form of how romances can develop.
Heck, thinking about it, I personally had internet flirts in the late 90s. I am old. :-)
I did it by phone in the late 80s early 90s. The Village Voice set up some system where you could leave your introductory voice message, or elevator pitch, and then women could respond to you and vice versa. It could get very expensive.
I remember having a discussion with a work colleague in early 2000 about news on the Internet. We both agreed that people would move to
personally-selected, uncurated news: I thought it was great, he (older and wiser) was horrified.
The switch to apps for transport and temporary accommodation is not much different to taking a taxi or staring in a motel/b&b so I don't think those are surprising at all.
For dating ... I don't know. A bit like news, I think: the general idea was predictable but the form it's taken and (arguably) the clear negative effect it's had are more surprising.
I asked Gemini and it confirmed that the following prank is good and appropriate for office work:
- tell your colleague that you brought him a new monitor
- replace his computer screen with a water monitor on a leash
"they’d like you to take a survey, which will take ~10 minutes."
Study itself says it takes ~20-25 minutes.
Oh - I took it and timed myself, but maybe I'm fast. I'll edit.
I actually prefer to delete offending comments to banning people. I feel like I'm giving them a second chance.
I don't. I have relatively few levers for making discussions here go the way I want, and removing offenders is my best bet. I'm afraid if I give someone a second chance, they'll learn to be just barely good enough that I don't have an excuse to ban them, while still bringing down the quality of discussion for everyone.
And that's on the rare occasions when it's someone who's capable of being just barely good enough, as opposed to someone who wandered in from a much worse corner of the Internet and wants to talk about how we're all Nazis for supporting AmeriKKKa and CRAP-italism.
Also, it seems like you only occasionally act, and when the probability of punishment is low, the punishment needs to be higher to compensate.
> the punishment needs to be higher to compensate
How? He already perma bans people on first offense. What are you expecting him to do, find where they live so he can murder them?
If you want to offer someone *a* second chance, as opposed to unlimited second chances, you have to keep track of what they do with that second chance. And you've got far better things to do with your time than that.
> And you've got far better things to do with your time than that.
Agreed. These days being what they are, the next question is whether AI is up to the task, or whether it will be soon. ( And then the question after that is, if the AI has subjective experience, is whether subjecting them to the comments of marginal commenters is an unreasonable burden on the AI themselves... )
I suggested AI moderation to Scott too. Seems worth a try.
Many Thanks!
The issue is that in practice, you can't stop people from commenting unless you paywall your comments section. Substack really doesn't care about people making alts.
And yet Scott's banning works fairly well! The quality of discussion here is absolutely miles above anywhere else with a large enough readership to escape niche confinement.
I think you're right.
If a chicken sees a road, is it morally permissible for it to cross to the other side? (Assume a vegan chicken.)
> assume plant milk/other vegan milk
DOES NOT COMPUTE! These liquids are not "milk". As decreed by the European Commission, real milk must originate from the tits of a mammal.
Finally someone speaking the truth!!
Are the clothes made of mixed fibre? Is the weather hot enough that you could claim the clothing became cooked in the milk? What day of the week is it?
If a mean AI turns the road you’re on into a mobius strip which side should you drive on?
LOL! Good one!
Last week it was beer. Why are you so interested in sucking liquids out of your clothes? And why do you keep saying you're eating them?
Tonight's episode: The Writer's Barely Disguised Fetish!
Oh great, they're inflicting their mukbang fetish on us 😒
I’m hoping for naked women wearing mittens!! Very niche. Very chic.
...but what if someone spills raspberry juice on the mittens? Is it OK to eat the mittens?
Now cut that out!
https://m.youtube.com/watch?v=Ra_cWKc36qg
Who "eats" milk as opposed to drinking it? Are you eating the clothes?
You're assuming the consumption happens immediately. What if you wait long enough for it to curdle?
Wtf?
This person is constantly dropping moral conundrums, disguised from the question they really want to ask, here. The last one was about alcohol, now they're re-phrasing it as milk.
When the real questions are revealed, it'll probably be yet another boring variation on the trolley problem. Is it morally permissible to eat meat if you don't kill the animal yourself? type of chin-stroking deepities.
"Is it morally permissible to eat meat if you don't kill the animal yourself? "
I hope the causality is even more removed: "Is it morally permissible to eat road kill," for example.
Or cannibalism. Though given the examples on show here, our friend won't be that creative. It'll be one more dreary re-tread of vegan arguments about how eating meat is the most disgusting awful wicked thing in the history of everything that has ever happened, and if we wouldn't personally bludgeon a cute baby seal to death with our bare fists in order to devour its steaming intestines raw, why do we think it's okay to let a butcher slaughter a cow for us?
Could be... Could conceivably be an offshoot of watching Plur1bus...
Is this an "other language to English" translation conceit? Never heard of it in English or French.
Or is this perhaps a young, questioning Orthodox chap, on the equivalent of his Rumspringa, trying to learn about the restrictions about dairy products that the world has, before he makes the choice to stay with the flock, or be with the outsiders.
Best of luck!