> Also, are you really suggesting that crimes against prisoners are not crimes?
No. I interpreted the original assertion and subsequent discussion to be about the effect of prison on crime against the general public.
> The optimal term for minimizing crime is obviously life imprisonment but I'm not sure that information can be put to any productive use
My guess is that the crime rate against the general public as a function of prison time curve [1] is not monotonic. I expect that there is at least one local minimum before the absolute minimum reached at life imprisonment.
[1] it would probably be better to say "curves", not "curve", because different kinds of criminals probably have different shape curves.
> No. I interpreted the original assertion and subsequent discussion to be about the effect of prison on crime against the general public.
Even putting aside the implied exclusion of prisoners as members of the general public, it doesn't actually work that way. Prison inmates can cause physical or psychological damage to other inmates who are about to be released, increasing the subsequent drain they put on public resources and their likelihood of re-offending. Inmates can be threatened with violence if they don't convince their families or associates to commit crimes on the outside. Crime bosses can smuggle in cellphones and bribe prison guards to carry on running their organizations. "What happens in prison stays in prison" is the opposite of what happens in prison.
> My guess is that the crime rate against the general public as a function of prison time curve [1] is not monotonic. I expect that there is at least one local minimum before the absolute minimum reached at life imprisonment.
I agree that it probably isn't monotonic, but it seems like asking the wrong question. If a specific type of punishment causes harm to the convict and causing harm to the convict increases the probability of recidivism then every ounce of that punishment you mete out is counterbalancing itself. Increasing the deterrent also increases the damage. Sure, the two curves may not be perfectly symmetrical and therefore allow you to find a local minimum. But that's like trying to optimize the amount of poison to put in your food as a preservative. Is it not better to just use a non-poisonous preservative?
Their kids will pick up the slack as they grow up without a father.
Also, are you really suggesting that crimes against prisoners are not crimes?
> Offhand, I don't recall seeing data on what the optimal term is for minimizing crime, but I don't recall looking for such, either.
The optimal term for minimizing crime is obviously life imprisonment but I'm not sure that information can be put to any productive use.