>Would you rather have 2 family members get killed in a drive-by shooting? Or would you rather have 1 family member get executed by the state? I think your tes

>Would you rather have 2 family members get killed in a drive-by shooting? Or would you rather have 1 family member get executed by the state? I think your test is quite misleading since you are moving the goal posts to "It's better for criminals to kill two people than to wrongfully execute an innocent man" which is an entirely different statement from "It's better to free two criminals than to execute an innocent man". Sentencing all those suspected of a murder will not reduce the murder rate by an equal amount, in fact you will end up with a larger total of wrongfully killed men. From the utilitarian perspective you seem to favor, you would want to minimize the number of dead people, so the correct N ratio depends on the psychological response to risk/reward of the average criminal. For a majority of would be criminals, even a 10% chance of being killed in retaliation is a strong disincentive, so you would get a strong crime reduction response for relatively a large N. As you continue to decrease N, you will encounter hardened criminals, willing to gamble with their life, and eventually those irrational types you mentioned that do not respond to any threat of punishment. But as you go to lower Ns, your own victim count increases, as you kill more and more innocent men in the hopes of dissuading killers from doing the same. So there is a sweet spot, significantly higher than N=1 where the total number of victims is minimized. We can reason the same about the punishment itself, and find a spot (N, punishment) for each crime that maximizes the societal welfare.