curious_straight_ca
No bio...
User ID: 1845
curious_straight_ca
8hr ago
I use o1 a bunch for coding, and it still gets things wrong a lot, I'd happily pay for something significantly better.
I dunno, government shutdowns due to ineffective governance and republican infighting sounds exactly like the first Trump administration. Springing this on congress a few days before the deadline when they thought they had everything worked out isn't a great way to introduce your administration, they hate working holidays. And it doesn't work as a show of power if you don't get your desired bill through, it just looks uncoordinated and capricious.
curious_straight_ca
2d ago
Sounds like they shouldn't have a journal controlled by a large corporation as their field's schelling point! They should start their own journal with the old editors. What exactly does Elsevier control that matters, anyway? A name? The only thing would be some amount of prestige you can show to academia as a whole, or the university that employs you. Even then, from wikipedia it "is published by Elsevier and is the official journal of the International Society for Intelligence Research", so maybe the Society can just endorse the new journal.
curious_straight_ca
2d ago
Enforcement changes habits. Right now 'everyone' does it, so everyone does it. If 'everyone' stopped doing it, it'd feel weird to do it, and many fewer people would do it, and enforcement costs go down. It's like smoking, or littering, or drunk driving. https://worksinprogress.co/issue/how-the-war-on-drunk-driving-was-won/
Two underappreciated ideas stick out from this experience. First, deterrence works: incentives matter to offenders much more than many scholars found initially plausible. Second, the long-run impact that successful criminal justice interventions have is not primarily in rehabilitation, incapacitation, or even deterrence, but in altering the social norms around acceptable behavior.
This proves too much! There have been at various points in the past much better ways and much worse ways to fund science than we have today, there are government bureaucracies outside and within the US government that are much more competent than the average in the US, and much less. I'm proposing we go for 'more'.
Academia was once run by people who were good at detecting bullshit. And then people got free money for producing garbage that follows the theme set by the State.
Private institutions that exist today do not produce better work than public ones. The problem is the people and the culture. Free money isn't helping, but the free money isn't just coming from the state, it's coming from all of the people who believe in the system. With zero government funding, superrich people who went to top unis and less-rich people who go to less-top unis would still donate to colleges, and they'd still make money on students. And the academic output would still be as bad as the things those academics post on twitter. There are plenty of people who are the same type of person as those academics, and are immersed in the same culture, but have 'normal jobs', and they say the same things as the academics, because the problem is the people and culture.
curious_straight_ca
3d ago
Disagree. You can't precisely measure quality, but smart generalists can separate bad articles like "How Young People Portrayed Their Experiences in Therapeutic Residential Care in Portugal: A Mixed Methods Study" and "Missandei deserves better": A case study on loving Blackness through critical fan fiction" from the kinds of humanities academic work you might want published. One can tell the difference between 'this might be valuable' and 'this definitely isn't'. The problem is the people funding this stuff aren't doing that.
Social sciences are, in principle, obviously worth funding. Philosophy (Nietzsche was actually a professor of philology), archeology, digging texts out of archives and writing history rank the highest for me, but there's valuable work in a lot of fields. A lot of the best work in economics has directly affected the way we organize the economy and the way businesses do business.
90% of publicly funded 'social science' is not that. It's hundreds of millions of words of repetitive, uninspired analysis of history or literature, like the work of that Ally Louks who blew up on twitter. The thing wrong with her, contra all of RW twitter, isn't that she's too woke or too communist or anything. Michel Foucault was woke for his time, but is obviously worth reading, and thousands of leftist academics have written things worth reading across many different fields. Her work, and 90% of modern humanities academic work, just isn't. And not in the "only 10 experts could appreciate or even understand it" sense, like in research math, but in the sense that there's no interesting content in it at all. There are a hundred thousand academics at various colleges and universities who either aren't smart enough or aren't independent-minded enough to develop good taste about what to research, and are paid (although not paid very much) to write ... really anything, so long as it's topical and isn't too embarassing, and can get published in a junk journal or turned into a book chapter or something.
Now the most valuable work is very valuable, and if you had to choose all or nothing (which you don't!), the best history and economics is still worth funding the garbage. (The money isn't counterfactually going to whatever you think is valuable, it's probably going to more welfare.) Or that's what I'd say in America, but New Zealand probably has a lot less than 5% of the global top 5%, so whatever.
curious_straight_ca
10d ago
Disagree, with an insurance-related CEO one can immediately think "that guy's responsible for denying care to thousands of people, literally killing them", and that goes much farther to intuitively justify it than "the CEO of Ford ... didn't pay workers enough? Polluted?". Maybe from very committed anticapitalists it would, but the average person on reddit or twitter isn't one
IMO tacky can also just mean art that is bad, or art that's attempting to imitate good art while missing the point. I think there are objective (at least with respect to human perception and nature as it actually exists) qualities that this has that this doesn't, without reference to status. If I saw someone express fondness for the latter painting I would think less of their taste and them a small amount, but as a practical assessment of their judgement. That something is often misused doesn't mean the original concept isn't there. The same goes for McMansions - in a well architected house the different pieces fit together nicely, and in a McMansion the individual pieces are exaggerated and they don't fit together at all, and it just doesn't work, in a way that is IMO fairly universal in the way humans perceive architecture.
curious_straight_ca
19d ago
Nobody I know who has children thinks the suffering of pregnancy or childbirth is on the same order of magnitude as the benefits of having children
curious_straight_ca
19d ago
My For You has always been fine, 95% of it is just people i follow already. It's a bit too consistent tbh. I imagine this is because I mostly use Following, so the algorithm's just learned I like that.
curious_straight_ca
19d ago
Courts have certainly gotten worse, due to wait times and case loads. I think technology has helped courts handle some of that (remote sessions). But they are still fundamentally limited in getting people to be physically available at a given time, shortly after a crime, and provide enough time for a judge and some lawyers to talk through the case.
I actually think reforming prosecution and courts to simply take advantage of technology is going to be a huge part of this. Our court system was designed for a century ago. Just adding zoom to the old mountains of process doesn't hurt, but there's a lot of room for efficiency gains without compromising on accuracy or anything else. And just hiring more prosecutors and judges and staff is exactly the kind of thing Scott's suggesting doing in his post instead of spending money on police. (Although I think political energy/will, more than money, is the main constraint)
Technology is a force multiplier for IQ, and criminals are not, actually, that smart. So if you can have some very smart people figure out how to use drones with cameras, or warrants to track phones, or etc etc, in a more systematic way than they currently are (not that that isn't being done, it just isn't being done efficiently because government is slow), that's just good.
curious_straight_ca
19d ago
The thought experiment was intended to be scoped to things like shoplifting and car theft, should've been more specific about that. But I actually think it'd have a moderate effect on murder for a subtle reason - it'd break up cultures of crime by making the moderately profitable activities they sustain themselves not profitable anymore. Like if you couldn't sell drugs, and couldn't steal stuff, and couldn't do welfare fraud, and so on, that culture becomes less attractive
curious_straight_ca
19d ago
The adversary knows the location of X if X is transmitting EM.
At an extreme, make it only transmit for a few seconds every 24h?
For 2 I don't mean enforce it at a software level, just make it physically difficult to disable
curious_straight_ca
20d ago
I mean fixing the laws would allow stores to do the cost/benefit analysis on that themselves, and if they think it isn't then they're probably right.
curious_straight_ca
20d ago
I think those are reasonable points, I'm not expecting that a few minutes of my thinking about this will solve it, just that a few years of smart tech people will. But I think just the location sharing part would be enough - you'd have to make it hard for the thieves to disable without totaling the car, but I think that's doable (just put it on the car's main board? put it in a random location? idk), and then if the cops just reliably physically repossess the car a few days after it's stolen that should make car theft a lot less attractive. A lot of new cars already have GPS and data.
curious_straight_ca
20d ago
In a mathematical sense, I suppose you're right, but people need to get caught in like 99% of cases, and all the benefits will need to be fully reversible in order for it to work
Yes, this was the point of the extreme hypothetical, which I which I should've been more clear about.
I suppose you're right, but people need to get caught in like 99% of cases, and all the benefits will need to be fully reversible in order for it to work
It works for repeated or organized theft, where the criminal's doing it because they're going to resell the goods for money - you can almost entirely stop that by just making costs > benefits.
I think one of the dangers of low punishments is that, even with the punishment, the actions may still worth worth it
Yeah, I'm not proposing low punishments for rape, because the benefit is intrinsic to the act itself and also a very primitive one that's hard to punish, rather than an economic one you can take away.
curious_straight_ca
20d ago
I'm claiming that historically, materially, the reason the divine revelations / religious precepts are what they are are the above - the religious elites actually did think a lot about what the right precepts are, and precepts and revelations were selected across cultures and generations for working and propagating
curious_straight_ca
20d ago
Yeah, what I mean by empower them to is remove all of the obstacles like that that prevent them from doing so. The liability concerns are a consequence of specific laws and precedents, and laws can be changed.
curious_straight_ca
20d ago
I don't think so? Grocery stores have security guards, you can just let them arrest people.
curious_straight_ca
20d ago
The 'minute' part was hyperbole, but I think a day or two should be trivial? If we can modify the design of the car, just have it disable itself in a way that can't be undone without the owner's consent (have many different components do the verification on their chips so the thief can't just swap out one), and broadcast its location. Just location broadcasting (only with the owner's consent once it's stolen) would be trivial to retrofit existing cars with very cheaply. And then you just, like, have the cops go pick up the car whenever it's stolen.
curious_straight_ca
20d ago
It was an unrealistically simple and extreme hypothetical to illustrate the point. Grocery stores have security guards.
The hypothetical was an extreme to illustrate the point. Yes, in practice, enforcement isn't perfect, so a week is too short and you want escalating sentences. The point that effective policing is higher leverage than increasing sentences remains.
Again, realistically, it would be terribly expensive and assumes magically competent cops
... no? Like stores do, currently, have security guards. We could just empower them to arrest shoplifters. That would be fine. (edit: this includes fixing the laws that create extreme liability for doing that)
I presume that stealing will dramatically drop in after the 5 years lag, and possibly wither to nothingness in following decades
Isn't most crime committed by young people? There's a steady supply of fresh young people, and arresting them 5 years after they're young isn't even going to stop them from having kids to form the next (on average) generation of criminals. Like the US, today, does up arresting most violent criminals for long periods of time eventually, and it hasn't fixed the crime problem.
It's basically a slap on the wrist, and only when you're caught
The assumption for the hypothetical is that you're caught every time. It's a slap on the wrist, but you can't actually benefit! So organized small groups stealing over and over wouldn't pop up, because they wouldn't benefit from it. That example was specifically for shoplifting and stealing cars. My argument is they would go down, because you wouldn't actually be able to benefit from doing them anymore. It wasn't intended to apply to rape or fraud. I don't really think there's much you can do about rape on current margins, absent everyone having a camera and audio recorder on them at all times, and fraud's a whole different thing anyway.
O3 can do research math, which is, like, one of the most g-loaded (ie ability to it selects strongly for very high intelligence among humans) activities that exists. I don't think the story that they aren't coming for all human activity holds up anymore.
More options
Context Copy link