LessWrong 2.0 Reader

"0.12% of the population (the most persistent offenders) accounted for 20% of violent crime convictions" https://inquisitivebird.xyz/p/when-few-do-great-harm

Thanks for the response! I must protest that I think I'm being misinterpreted a bit. Compare my quote:

the point of RCTs is to avoid resorting to regression coefficients on non-randomized sample

To the:

The point of RCTs is not to avoid resorting to regression coefficients.

The "non-randomized sample" part of that quote is important! If semaglutide had no impact on the decision to participate, then we can argue about about the theory of regressions. Yes, the fraction that participated happened to be close, but with small numbers that could easily happen by chance. The hypothesis of this research is that semaglutide would reduce the urge to drink! If the decision to participate was random, and I believed the conclusion of the experiment, then that conclusion would seem to imply that the decision to participate wasn't random after all. It just seems incredibly strange to assume that there's no impact of semaglutide on the probability of agreeing to the experiment, and very unlikely the other variables in the regression fix this, which is why I'm dubious that the regression coefficients reflect any causal relationship.

That said, I think the participation bias could go in either direction. I said (and maintain) that the lab experiment does provide some evidence in favor of semaglutide's effectiveness. I just think that given the non-random selection, small sample, and general weirdness of having people drink in a room in a hospital as a measurement, it's quite weak evidence. Given the dismal results from the drinking records (which have less of all of these issues) I think that makes the overall takeaway from this paper pretty negative.

There are the predictable lobbies for increasing the price taxpayers pay for prisoners, but not much advocacy for decreasing it.

Similar disclaimer: don't assume these are my opinions. I'm merely advocating for a devil.

If we're going for efficiency, I feel like we can get most of the safety gains with tamer measures. For example, you could cut off a petty thief's hand, or castrate a rapist. The actual procedure would be about as expensive as execution, but if a mistake was made there is still a living person to pay reparations to. I think you could also make the argument that this is less cruel than imprisoning someone for years—after all, people have a "right to life, liberty, and the pursuit of happiness", not a right to all their limbs and genitals.

Another thing we can do is punish not only the criminal, but their friends and family too. We can model people as having the policy to take certain actions in a given environment. The ultimate goal of the justice system is to decrease the weight of certain defective policies in the general populace, either through threat, force, or elimination. When we get good enough mindreaders, we can just directly compare each person's policy to the defective ones, and change the environment to mitigate defection. Until then, we have to make do with approximations, and one's culture, especially the shared culture among friends and family, is a very good measure for how similar two people's policies will be. So, if we find someone defecting, it makes sense to punish not only them, but their friends and family for a couple generations too.

(For people reading this thread who want an intro to finemapping this lecture is a great place to start for a high level overview https://www.youtube.com/watch?v=pglYf7wocSI)

A fenced-off city that will inevitably be compared to a Holocaust ghetto.

My paraphrase of Gandalf: "Many that die deserve life. Can you give it to them? Then do the next best thing, and deal out death in judgement to the many that live who deserve it." 

But why's that a bad thing?

Because otherwise everyone will gleefully discriminate against them in every way they possibly can.

I am sure there are some interesting uses of agented AIs in can configure for automated OSINT but this feels quite large a task given I am bottlenecking more in who to hand the data to rather than it being insufficiency rich.

Know any preconfigured agency menageries for something like this?