My other hypothesis is that people commit crimes for only 2 reasons:
1) They don't think they will get caught. For these types of people, the punishment won't be a deterrent because they believe they will never suffer the consequences. You could literally raise the punishment level to death and it wouldn't affect these people from committing the crime because they don't believe they will get caught.
2) They don't care if they get caught (ie. they don't think or care about the consequences of their actions). This can be split into 2 subgroups: a) They simply don't care about the consequences (ex. acts of passion or they are ignorant to the consequences). b) They think the consequences are "worth" it.
Punishment really only will affect group 2B. The other two groups (1 and 2a) won't be deterred no matter what. So the question really is, how do we optimize deterring group 2B? Solitary confinement, to me, is a more cost-efficient way of increasing the scale exponentially while still not being drastic. Some might say that 90 days in solitary confinement is worse than a few years in jail. If that's true, then it's really cost efficient, and would serve as a cheap, quick deterrent and might prevent recidivism.