People who commit crimes rarely believe they will be caught, and as a result weight that conditional probability differently than an ideal reasoner might expect

People who commit crimes rarely believe they will be caught, and as a result weight that conditional probability differently than an ideal reasoner might expect them to. If you believe your chance of being caught is 0% then the punishment is not a deterrent. Alternatively, if you are desperate, even if that chance is not 0% and you know it, it might still be worth it (say you're desperate because you have a previous felony conviction and now can't support yourself). Now, one of the hard things to measure, is whether the sentences does deter everybody who doesn't commit crimes. I assume that there are crimes you and I have had the opportunity to commit, but weighed the odds of being caught and decided that the outcomes from committing the crimes were not worth it. It's difficult to measure that, but it's also not very useful, because it is hard to say whether the deterrent came from the punishment, or from the fact that it was lawbreaking in the first place, or that the label of "felon" is risky enough given the known outcomes. In the place I'd argue we want the most deterrence - violent crime - our harshest punishment seems to have no real deterrent value. (I would be interested to see a review of this using more modern statistical / ML techniques to examine the data, though). https://www.dartmouth.edu/~chance/teaching_aids/books_articles/JLpaper.pdf https://www.dartmouth.edu/~chance/teaching_aids/books_articl...