The idea of prisons, frankly, disturb me, but I’m not sure whether it disturbs me more or less than the idea of hardened criminals being free. I think the general thinking in Western-Northern Europe is that prisons are not great, and I’ve mostly just defaulted to this belief.

Prison And Crime: Much More Than You Wanted To Know was my first read on the topic, and as usual, Scott Alexander does a great job of trying to consider incentives, hidden variables, and other potential problems with different analysis. His summary:

  • Long prison sentences can theoretically decrease crime through deterrence, incapacitation, or rehabilitory aftereffects.
  • Deterrence effects are so weak that we might as well round them off to zero.
  • Incapacitation effects are strong. The exact strength depends on how many people are already in prison, but a reasonable estimate at the current margin is that each prisoner-year prevents one violent crime and six property crimes.
  • The magnitude of aftereffects are unclear, and probably range from slightly beneficial to detrimental depending on the population and length of prison term being studied. …

Disappointingly, it appears that whether prisons are a net-positive or net-negative depend on a number of factors, and is not possible to answer in general. However, I think the research on the main negative effects of prisons needs more focus, as if we could eliminate that, then we could make the whole thing a lot more cost-effective.

Specifically, I’m talking about the imprisonment after-effects, that is, how likely is somebody who was released from prison to commit a crime again. Given the incredible disruption that prison is to your life and social connections, unsurprisingly, prison sentences of about 5 years or longer have a huge negative effect here: somebody whose life was completely ruined by imprisonment is highly unlikely to be able to reintegrate. At the same time, imprisonment of a few months is a much more manageable disruption, and suffers from much milder negative after-effects. It seems fairly obvious to me that the logical choice would be to try and eliminate this negative after-effect by seriously focusing on rehabilitation and reintegration. In this sense, I would argue that the system should not have any memory in that whether you’ve been ever imprisoned or not should not be allowed to impact your ability to get a job. I don’t have any good ideas on how to achieve this though, other than getting rid of the obviously regressive punitive laws in the US.

Scott Alexander does add a conclusion:

Prison is less cost-effective than other methods of decreasing crime at most current margins. If people weren’t attracted by the emotional punch of how “tough-on-crime” it feels, they would probably want to divert justice system resources away from prisons into other things like police and courts.

This broadly makes sense to me, but I would like to highlight the “at most current margins” part. As discussed, whether prisons are a net-positive or net-negative depend on a lot of factors, and one of these is current crime and incarceration rate. As extreme examples, not having any prisons at all does not make sense, in the same way as imprisoning 25% of the population doesn’t make sense either. This implies that there is a point between these two where more prisons and more imprisonment ceases to be a net-positive. Scott Alexander’s conclusion says that most places are at a point in this curve where more imprisonment is clearly a net-negative.

This is important, because it does not mean that from a cost-analysis standpoint prisons are always ineffective, just that the US in particular is (very much) overdoing them, and should be focusing much more on other measures. In this way, Europe is currently doing significantly better.