This is the intro for chapter 7 of this excellent book:
The emphasis current reform efforts place on reducing punishments for people convicted of low-level nonviolent crimes is understandable, but it should be clear by now that the impact will be limited. Any significant reduction in the US prison population is going to require states and counties to rethink how they punish people convicted of violent crimes, where “rethink” means “think about how to punish less.”
A simple example makes this clear. Assume that in 2013 we released half of all people convicted of property and public order crimes, 100 percent of those in for drug possession, and 75 percent of those in for drug trafficking. Our prison population would have dropped from 1.3 million to 950,000. That’s no minor decline, but this sort of politically ambitious approach only gets us back to where we were in about 1994, and 950,000 prisoners is still more than three times the prison population we had when the boom began. Or consider that there are almost as many people in prison today just for murder and manslaughter as the total state prison population in 1974: about 188,000 for murder or manslaughter today, versus a total of 196,000 prisoners overall in 1974. If we are serious about wanting to scale back incarceration, we need to start cutting back on locking up people for violent crimes.
I think there would be knock-on effects from legalizing drugs (in particular, legalizing the drug markets) that would affect other categories of crime. But we'd still have a lot of violent offenders who are violent in more conventional ways, and we'd have to make an honest decision about how to punish them.