roystgnr
9hr ago
"unless"? Did Bezos leave his first wife before she kicked him out? He was cheating on her for like a year before it went public; I'd assume he would have been happy "getting the best of both worlds" indefinitely if he'd managed to keep getting away with it.
roystgnr
9hr ago
Yeah, "handful" was grossly wrong. 30% for "more welfare", geographically non-uniform, means there's probably going to be somewhere you can get a plurality to vote for seizing the means of production ... grocery retail? really? ...
But my point is just that, if loud vocal support for more welfare seems to be coming disproportionally from college students, that's because of a disproportionality in "loud" and "vocal", not in "support".
roystgnr
9hr ago
Yeah, but if you try denying it to the wrong guy
Yeah, that's what the Schrödinger's Rapist conversation was all about. Men who didn't understand that women inductively match "a stranger is trying to flirt with me" to frighteningly high odds of "and he might get really nasty or dangerous about it if I'm even a bit too gentle or too harsh (or both, because nasty men aren't all big on logic) about shutting it down", arguing with women who didn't understand that making the guys nice enough to listen about this problem more wary just ends up increasing the proportion of public flirtation coming from nasty men who would never have listened in the first place.
You have to be careful with your signalling.
I admit I suddenly developed a lot more sympathy and respect for women with "resting bitch face" (and even for some non-superficial negative personality traits) when I realized how useful that probably is as a defense against accidentally signaling (or being mistakenly perceived as signaling) invitations they didn't want to offer. Even if the person making a pass isn't at all nasty, more explicit rejections are much more emotionally upsetting to give! In the moment it hurts more to be rejected than to reject someone, but (unless there was a reason for rejection more substantial than "I'm not very attracted to you") IMHO the pain of being rejected goes away faster.
roystgnr
9hr ago
Does it? After the last decade I'm very skeptical of "it's just a handful of loud college students" theories, but in the case of support for socialism it may actually be just a handful of loud college students?
Currently the demographic most likely to say US national spending on welfare is "too little" is people with "less than high school" education (vs high school graduates and college graduates), and this is the case far more often than not, although the correlation with education seems surprisingly weak in general - high school graduates' and college graduates' responses are pretty much neck-and-neck.
That's not an apples-to-apples question, I admit. Might there be people who think we should overthrow capitalism altogether and perfect the New Soviet Man or whatever but who also don't want any more money going to today's lumpenproles? It still seems like a good indicator.
roystgnr
10hr ago
States are collectives of people but they aren't people, the people in states should be represented fairly
Taking this consistently, we must conclude that Russia should get to conquer Ukraine, by a vote on the order of 130M to 30M.
Recall that originally state meant, well, "state", not "prefecture".
roystgnr
12hr ago
Where did you read this?
Not OP, but this off topic comment thread was interesting enough that I saved some of the best quotes (from the top reply) and was able to find it again now, a decade later.
This completely tracks, I would like to know more.
You and everybody else still single, right? (...as well as those of us who are married but would like to give our kids useful advice and have grandkids someday, honestly) It feels like the Sexual Revolution burned all the oversimplified restrictive scripts that one might have read in some stuffy old Guide to Mannerly Courting, but then instead of the result being "we're all free! we can just do what feels right!" it turned out to be "you're still screwed if you don't follow the right script, but now everybody disagrees about what the right script should be and it's too contentious to talk about or write down".
I'm not even sure if the above interpretation of proper flirting hasn't also already been obsoleted by cultural changes! (I was already married ten years ago, thankfully, and it's even harder to analyze this stuff from a distance) After ten years of "you find a guy by just swiping right", do women still know or care about the old "you can find a guy by giving plausibly-deniable signals of invitation and escalation" techniques? Ten years after this cute comic prompted the "Schrödinger's Rapist" debate, are men still eager to try carefully interpreting possible plausibly-deniable signals? Granted, the remaining alternatives are awful, but it feels like they're all that's left.
roystgnr
1d ago
75% of blacks voted for FDR in 1936, for example.
Wiki claims "two-thirds of black voters", not much more than his 60.8% of voters as a whole. You have to cut down demographics more finely to get to "76 percent of Blacks in northern cities" specifically.
The charitable view is that blacks migrating to cities etc. supported the dems because they had greater support for unions and worker's rights etc. A less charitable view is that blacks supported dems because they saw it as a way to get handouts from the government.
Neither of these views is logically inconsistent with the corrected numbers, but Bayes would note that our relative credence in the view with "migrating to cities" in it should increase after we learn that the data shows an effect specifically focused on migrant-targeted cities. (you'd otherwise think "Northern" would push in the other direction; Southerners also went 76% for LBJ)
roystgnr
1d ago
What I don't see is why the solution isn't "just give him the statutory authority". At least until after midterm elections, the Republicans have a 5 seat lead in the House and a 6 seat lead in the Senate. The ballroom idea isn't super popular overall, 28-56, but among Republicans it's got supermajority support, and it's at least generally defensible enough that you'd think someone would be willing to introduce a bill and dare the Democrats to filibuster.
If Trump had difficulty whipping up support among his own party in general, that would be one thing, but the guy can get a room full of people to politely watch (or even smile and nod!) while his health secretary brags about their inability to do grade school arithmetic. I'd think "the design isn't too big or garish, not for state functions" would be an easier sell.
roystgnr
1d ago
we call that jury nullification which happens but I am not sure we consider that allowed.
It varies for different definitions of "we", "that", and "allowed".
Perhaps the strongest case for "not allowed" is that juries are given explicit contrary instructions. Perhaps the strongest case for "allowed" is that there's very few other good reasons to conscript a bunch of random unqualified jackasses off the street to make judgement calls about a trial when there's already a highly trained person, literally titled "judge", right there. The "of their peers" bit isn't added to "jury" because legal types hate concision, it's because that part is critical to nullification protecting against laws that seem good to upper class judges but not to the class of people affected.
(Of course, the strongest case for "shouldn't be allowed" is that often upper class people are just more correct about what's good. E.g. it's much harder to protect unpopular rights if anyone criminally retaliating against their exercise might get let off by a jury nullifying the crime, or at least might be impossible to prosecute in the face of hung juries with some members nullifying the crime.)
roystgnr
2d ago
I'd love to learn of any results that really do make full AC necessary
Okay, there's "every vector space has a Hamel basis", which may have important implications I don't realize since I just use Schauder bases in spaces where the distinction matters.
roystgnr
2d ago
opinions on computer-assisted proofs like the Four Color Theorem
I think the development of formal proof verifiers like Lean mostly quelled the practical concerns about computer-assisted proofs. Nobody's going to trust some two hundred page "proof" just because an LLM spat it out, but formalize it and properly verify the proof steps via a smaller verifier that's been itself closely manually examined, and then the remaining parts of it you have to check manually are more like definitions (when Lean verifies that "All Foos are Bars", does its definition of "Foo" and "Bar" match ours?) and much easier to understand and review. There's a real synergy here in iterating between proof verifiers (which will reliably state whether a proof is correct, but weren't very popular by themselves because they require the proof to be spelled out in tedious precise detail) and large language models (which will translate a colloquial proof into tedious precise detail, but aren't very useful by themselves because they aren't reliable enough to trust without rigorous checking).
The aesthetic concerns are still there, though. There are proofs that you can read through (the highlights of, not the every-trivial-step that you have in something formalized) and they enhance your understanding of the subject, and then there are proofs that just make it from point A to point B via some kind of hideous brute force, and there's a reasonable fear that computer-generated proofs or even just computer-assisted proofs are going to have a lot more of the latter instead of the former. There was quite a lot of excitement recently about a couple newly-AI-proven conjectures (IIRC one on primitive sets, another on Ramsey numbers, both on asymptotic behavior?) because, not only were these about questions that human mathematicians had taken more than a passing interest in, but the proofs were short and insightful. Candidates for proofs "from The Book", to use Erdös' old phrase.
Specific examples would be the Axiom of Choice (either accepting it or not leads to unintuitive results like Banach-Tarski)
Well, everybody agrees that if you accept it then you get certain nice things and certain nasty ones, and that you can have consistent models that accept it and consistent models that don't. There's still a disagreement here, but it's again a disagreement over aesthetics more than over fact.
It's a big disagreement over aesthetics, admittedly. The joke goes: "The Axiom of Choice is obviously true, the Well-ordering principle obviously false, and who knows about Zorn's lemma", and the humor is that emotionally that all feels true even though logically those three things are provably equivalent.
There might even be increasing common ground in the aesthetic question. The Axiom of Dependent Choice is sufficient to prove a ton (I hesitate to say "all", since in my own field we typically just throw up our hands and assume full AC, and I'd love to learn of any results that really do make full AC necessary) of the classic real-analysis and functional-analysis theorems that Zermelo-Fraenkel alone doesn't give you, but it isn't sufficient to force the existence of ugly-seeming things like non-measurable subsets of ℝ, or of insane-seeming things like Banach-Tarski.
roystgnr
3d ago
If an apple fell up once so conveniently that the entire plot happened, I'd consider that bad writing unless, I suppose, the irony of this one unexplained anomaly is the entire premise.
This seems obviously correct to me, except that empirically it's just wrong. Off the top of my head I can't actually think of any other examples in which it's wrong, though; is there some meta-irony here about how there's this one unexplained anomaly in the category of narrative quality of anomalies?
roystgnr
3d ago
Isn't "fanny" the classic? A silly term for "butt" or adjective for "belt pouch slung over your butt" in the US; a vulgar term for female genitalia in the UK.
roystgnr
6d ago
Gemini with memory on seems to make reasonable guesses about the reasons why I ask a question, which so far is only a little useful for me, but which could possibly make it a better source of answers than I often am for some of the sorts of XY-Problem questions I sometimes get from others.
roystgnr
6d ago
Lowest non-COVID workforce participation rate in my lifetime, but that just means 62% down from a peak of 67%. Optimistically I'll guess there's just more and more jobs with flexible hours. If the nature of the work allows for it, a smart boss will realize that if you like to eat lunch at 2 that just means you'll be able to get back with less delay and there'll be someone around to handle any emergencies at noon.
roystgnr
6d ago
My second YouTube search came up with the same top result, and far be it from me to refuse a polite request.
But I'm not making a third search. My kids occasionally use my YouTube account on our living room computer, and at some point I'd worry about the algorithm deciding that my current math/science/art balance needs more "art" and that my "artists" need fewer clothes.
roystgnr
6d ago
I'd never call myself a "never"-Trumper, because any reading of history will tell you that things can get much worse than him, but if I lived in a swing state I'd probably have held my nose and voted for the other shameless crook in 2016, the other mentally declining old man in 2020, and the other dishonest auto-coup fan in 2024. Maybe that's close enough, practically, even if I'm still not certain that any of those votes would really have been the lesser evil?
I still loved Ron Paul. If we'd made him Emperor he could have been a sufficiently radical (reactionary?) libertarian to throw the country into anarchy, but as a mere President I bet he'd have been great. Remember back when we had the chance to work on fixing a federal debt that was "only" ten trillion dollars? Good times.
roystgnr
6d ago
I watched one video, and I'm still not sure how much of that is her voice. I know I should have gotten inured to modern over-AutoTuned production over the decades, but 𝅘𝅥𝅮𝅘𝅥𝅮 I really don't think I'm strong enough! 𝅘𝅥𝅮𝅘𝅥𝅮
She's got great taste in underwear, though.
roystgnr
6d ago
Just look at the original image
Which follow-up joke would you say is even harder to miss: "Okay: 100100101011010101011...", or "I'd love to, but my spacesuit is still in the shop"?
roystgnr
6d ago
But my small town has multiple comparable parks in easy distance
I assume that what you mean is "comparable quality-per-area", but it's amusing to imagine a little town with like 5000 people, 10% of whom are employed maintaining the 5 square miles of neighboring park/zoo/lake/forest/museum/hiking/garden/ice-skating/boating complexes.
roystgnr
6d ago
Ooh, I should add this to my list of Goodhart's Law examples.
No, that one is fraud.
There are already around 60 categories on that list. We may need to build a lot more jails.
roystgnr
6d ago
Google Earth is a thing.
It is! It uses a near-sided perspective projection.
You may have missed the joke.
roystgnr
6d ago
The sharpest discontinuous decline in show quality in my mind was obviously between seasons 8 and 9; the internet seems to agree. IMDB ratings also seem to agree that "The Principal and the Pauper" (season 9 episode 2) was the most blatant turning point: second-worst-rated episode up to that point, and the only thing that beat it was a clip show. On the other hand, it's still higher rated than the average of basically every season from 17 to 37, so there's something to be said for the power of dull continuous decline too.
If we were to say that the Simpsons was all good in seasons 1-8 except for a few clip shows, then in average ratings it doesn't decline past "mostly good" (where the average episode is at least as good as all those early episodes) until around season 12; if we set our sights higher (yeah, some of those season 1 episodes were meh) then season 10 was the dividing line, and the last good individual episode was probably 2024.
Now I'm curious; I'm going to watch that one.
... wait, it opens by claiming to be the Simpsons Series Finale? Are those ratings just people's way of saying to let the show finally die with dignity?
AI is still at the state where it will sometimes confidently tell you something completely wrong, and yet for simple blocker issues this is still immensely helpful, because you can just try out what it told you, and 90% of the time you get to declare victory, and the other 10% you're just back where you started.
In fact, there's a bit of grey area in between - twice recently I've seen an AI come up with a solution which was definitely wrong or incomplete in some way, but which was much easier to fix than solving the problem from scratch would have been.
For complex issues and design issues, AI can easily paint you into a corner by generating reams of redundant/spaghetti/inflexible code that solves your immediate problem but is unmaintainable in the long run, but in general it's getting better so fast that I'm not sure how long this warning will be necessary.
More options
Context Copy link