amalcon

So, you're postulating that there is something of a Simpson's paradox going on in NYC between two largely separate cultures -- that of people with college degrees, and that without? This is theoretically possible, but conveniently not directly testable due to secret ballots. It is however indirectly testable in two different ways -- and the numbers still don't back you up.

1) Exit polling (https://www.timesunion.com/capitol/article/voted-trump-new-y...) does not back you up. More college educated New Yorkers reported voting for Trump than total fraction of San Franciscans that voted for Trump.

2) Middlesex County, Massachusetts (roughly, the north half of greater Boston) is an alternative example to NYC that has a very similar proportion of college degrees compared to SFO (about 1% less). Middlesex County is in the northeast. About 42% of those voters voted for Trump, which is more than New York and more than double SFO. Suffolk County (with Boston proper) has very similar numbers to NYC for both college education and voting for Trump. This can't be explained as a cultural difference between NYC and Boston because the Suffolk ratios are so similar. It can't be explained as an educational difference between SFO and Boston because the Middlesex educational numbers are so similar but there are proportionally more Trump voters than NYC. It can't be explained as a cultural difference between Suffolk and Middlesex because first, they are direct neighbors sharing a public transit system, and second, Middlesex is considerably more traditionally leftist than Suffolk (leaning on that cultural difference would undermine your original argument).

Most of the growth is. If we removed AI, the US would be in a severe recession right now. Most of the absolute market cap, employment, capital, and other metrics that are not directly coupled to growth paint a better distributed picture.

Of course, growth like this can only continue for so long without changing that.

So, "commoditize your complements". A big part of Microsoft's old playbook (and a smaller part of its current one), so we know it can work.

You are simply incorrect about that. More people voted for Trump in NYC (the typical exemplar of "the northeast") than in SFO (the exemplar of the part of California that is more pro-AI than NYC).

If you were in NYC, some of this may be a Boston/NYC difference. NYC is much more anti-Waymo (which I interpret mostly as an extension of being much more anti-car in general). There are also substantially different cultures between the two re: art and finance, which may have reflected on the AI questions.

I don't think you can decouple Trump from himself in this way without a concrete explanation. If some factor causes one to dislike something just because Trump likes it, then that same factor would influence voting. We have numbers for voting. When there are lots of other plausible explanations than Trump, the onus would be on you to demonstrate why more people are taking reverse-psychological cues from Trump in a place where more people also voted for him. Gesturing at "culture" doesn't do that.

You did claim that it was political:

Its just the thing that upsets me the most about your everyday liberal professional, how much they hate AI just because Trump is pro AI

That's what I'm refuting here -- that people here dislike AI just to be contrary to Trump. This is implausible.

And trying to lecture a life long New Englander who currently lives within the MBTA zone on what the culture is like in the northeast is maybe not the best look. We aren't afraid of Waymos here; the only people I've met who weren't jealous of them were taxi or Uber drivers. That supports my hypothesis, not yours.

That observation[0] isn't particularly useful for arguing that it's a political allegiance thing. A lower fraction of voters in the pro-tech part of California (SFO, roughly) voted for Trump than, say, in NYC. Your observation cuts against the argument that political allegiance is the main driver, not in favor of it.

[0]- Assuming it's true, which in my experience it isn't especially, but this isn't the best argument.

I think "your everyday liberal professional" is more likely to be anti-AI because it is being advertised as a direct threat to their livelihoods, rather than for any more abstract reason. The latter includes both political allegiance stuff and the reasons those professionals would likely cite, like its energy requirements or the economic damage if the bubble bursts.

Whether or not you think it likely that this threat will materialize is beside the point. Workers never like it when someone is advertising that they will put you out of work.

I don't think a bifurcation of cognition is necessary to explain this. You get both people who claim to always think with words (to the point where they don't understand how it's possible to think without one), and people who claim never to outside of linguistic tasks like writing (to the point where they need "internal monologue" explained to them). In my experience, if you get down to it -- most people will have experienced cases of the thought preceding the word (you can skip the word, or not, but either way it demonstrates a thought without a word) and of the word preceding the thought (people who claim not to have an internal monologue will usually still do this sometimes when writing).

It's just that subvocalization is one powerful technique for organizing thoughts, but not the only one. If you had a swimmer who only ever learned the crawl stroke (freestyle), they might wonder how someone can swim without it. The difference, of course, is that the breaststroke is a demonstrable physical act whereas cognitive techniques are not. The monologuers are just more practiced at subvocalization-based reasoning, and the non-monologuers are more practiced at other types.

It's not that hard to exert limited control over which technique you're using -- and I think there are multiple nonverbal techniques. I'd encourage people to explore the space a bit. Subvocalization-first approaches are very good for linear progressions, whereas subvocalization-last (or never) approaches are usually better for big-picture reasoning.

The obvious solution to the problem of abuse of the asylum system (which is in fact a real problem) is to hire more immigration judges so that claims are processed in a timely manner. Apart from getting rid of the asylum process entirely, that's the only long term solution. It's something Joe Biden made progress on (though he did overstate his progress), and Donald Trump reversed.

It doesn't show up in the short term numbers because it is a long-term solution. But Americans aren't interested in long-term solutions.

I am not interested in being dragged into an argument about this, and of course the supreme court can try to calvinball whatever they want. That being said, your interpretation is at a minimum disputed by one of the largest law firms in the country (https://www.thompsoncoburn.com/insights/pro-forma-session-10...).

Congress has remained in technical session (by having one member open and close the House each day) constantly for decades. Actually ending the session would be a rather significant departure, and I've heard of no plans to do so.

I know, right? It used to be easy: just look for writing in a very long-winded style, almost as if the author is being paid per word, in a place where that sort of writing didn't belong. I think it was because that type of writing represented a disproportionate fraction of the tokens in the training data due to the long-winded-ness. Somewhere around a year ago, they figured out some way to deal with that problem.

Making election day a (normal) national holiday would not help. The type of service sector or emergency service job that makes it a problem to vote needs to work more on a national holiday, not less.

Making it a kind of super holiday where we mandate closure of everything from gas stations to non-emergency hospital services might help, but you're never getting the votes to make that change.

The goal is obviously to make US taxpayers pay for it, but give room for the admin to lie and say we aren't. Similar to how Mexico was going to pay for the useless border wall.

Of course, point 1 (which includes immediate stoppage of the fighting in Lebanon) isn't happening -- which sort of renders the rest of the thing moot anyway.

I need to give them credit for one thing: the sheer chutzpah to claim that America will get Gulf allies to pay for it, thereby reminding everyone about the promise that Mexico will pay for the border wall.

The US should decouple the Hezbollah-Israel war from the Iranian one.

Inclusion of Lebanon in the cease-fire is an Iranian condition, not an American one. Since Iran clearly has the leverage here, it's unclear what you expect the US to do to cause this decoupling.

The most lucrative uses of deep neural networks today are ad targeting and recommendation algorithms. Those are other places.

Yeah, they have gone up a lot since I bought mine too. I did get Qwen3.5-122b running on all-GPU (on a 128GB machine) under a minimal Arch Linux setup (I do my GUI work on a much cheaper box). It worked, but Qwen3.6-35b is performing almost as well and a lot faster.

Still cheaper than a new Mac. Maybe not cheaper than a used one.

A Strix Halo with similar RAM is considerably cheaper. Still not cheap, mind, but performance is OK (not great) and it will run more or less the same models.

I do find it hilarious that Asimov wrote many stories about how simple bright-line rule-based systems are ineffective for restricting agency. Those stories were first published in the 1940s.

80 years later, we have something approximating AI, and we're trying to restrict it with simple bright-line rules. Not because we never learned that lesson, but because we simply haven't come up with a better way to do it. Probably because a better way to do it just doesn't exist.

The hilarious part, though, is that it's not the AI that's working around the rules. That's the scenario that's been in science fiction, but it's not what's happening. It's the human users making use of our agency to get the AI agents to work around the rules. Despite calling them "agents", current AI agents don't seem to be able to that particular something. Yet, at least.

Haha. Reminds me of how volt-amperes are technically the same unit as watts, but if you see VA in an electrical specification you know it means a different thing than it would if you saw W.

The issue is that this isn't even in Trump's top two cryptocurrency scandals since he became president for the second time. #1 is the access selling, #2 is the pardon of the Binance guy.

Nobody seems to care about either of those, either.

Not just recently. It's been that way for decades -- since at least the Reagan era, but arguably going back to the New Deal.

So, there would now need to be new negotiations as to the future of the Iranian nuclear program? Except with Iran in a weaker position than they actually are today, due to a decade of no progress on enrichment.

How is that worse than the current situation? Iran is closer to nuclear capability now than in your counterfactual, yet those who disagree with you are the ones who want a nuclear-capable Iran?

Rules of thumb:

The more your toolchain (compilers, linters, etc) can statically verify, the better agents will do.

The terser the code, the better agents will do.

The more often similar problems have been solved in open source, the better agents will do. Agents seem particularly good at plumbing together different pieces of software.

Anything that requires a judgement call, as opposed to having one obvious way to do it, will get worse results from an agent.

As the scope of the request grows, agents get worse at it. This can be mitigated somewhat using various techniques ("write a plan", "do step 1 of the plan", etc) but never fully resolved. At some point the task is so big that it's necessary to do large parts by hand.

His base is still with him. Frankly, I can't find a material distinction between the beliefs of his base and the belief that Trump is the physical embodiment of America (and therefore anyone who opposes him in any way is an enemy of America).

However, his base is smaller than it seems like it is. It definitely does not include everyone who voted for him. A substantial fraction of his voters just prioritized other things over the rule of law.

That prioritization was a critical mistake - both because he was never going to deliver on those other things, and because giving up rule of law is catastrophic. It is understandable, though. It's these people who may be starting to break fron Trump.

Wrong. Republicans have been nasty forever (e.g. calling Democratic voters baby-killers who hate America since at least Reagan, redoubling that under Bush, redoubling again under Trump). They still win elections, and Democrats responded for decades by ignoring it.

That is what has led to Trump. What little evidence we have indicates that bullying just works. It's not the only thing that works -- see Obama -- but it does work. Ignoring the bullying doesn't.

What exactly do they have deep regrets about, though? Do they regret voting for someone with his style? Do they regret empowering an obvious corrupt liar? Do they regret supporting someone who focuses on who to blame and hurt, rather than on things that might actually help (albeit in regrettably marginal ways in all three cases)?

Or do they just regret that they were fooled by this guy, specifically? That he's not accomplishing his stated goals, whether or not he is taking his promised actions? If it's this one, then it's only a matter of time before another charlatan does the same thing better.

In legal terms, one would call this a "statement against interest". If a politician says he does not care one bit about the welfare his constituents, it is reasonable for those constituents to take him at his word no matter how untrustworthy he is. It only hurts him if we believe it.

Iranian negotiators probably shouldn't believe him, because it's not against interest in that context. There probably just aren't any on Hacker News.