In the latest issue of Social Research (wait! come back!...), Marc Mauer has a very smart essay pointing out that the U.S. prison population grew so rapidly during the 1990s not because people were being thrown in prison at a greater rate, but because the length of time served in prison was increasing so dramatically. This bubbled up briefly in Jim Webb's hearing on penal reform last week, but it still doesn't get nearly enough attention.
By Mauer's calculations, if legislatures had left the length of prison sentences alone during the '90s, and the average time served hadn't increased, there would be about 400,000 fewer prisoners in state prisons today (there are currently 1.2 million). And that's without any change in the actual number of people sentenced to prison--before anyone starts talking about treatment programs as an alternative to jail for drug offenses, or graduated sanctions for parole violations, or other wonky measures to reduce recidivism.
The second striking bit comes when Mauer compares U.S. sentences with those abroad. Burglars now serve an average of 16.2 months in prison in the United States, compared with 5.3 months in Canada and 6.8 months in England. A conviction in a U.S. federal court for selling a kilogram of heroin gets you a mandatory 10-year sentence, versus six months in prison in England. Then, of course, there's the prevalence of "three-strikes" laws here in the United States: Around 4,000 prisoners in California are serving life sentences after committing drug or property crimes as their third strike.
Maybe this would be justified if there was any reason to think that longer sentences actually make a difference. But there's not: As Mauer notes, there's no evidence that longer sentences deter crime ("any deterrent effect... is achieved primarily by certainty of punishment, not severity") or reduce recidivism. In theory, of course, you can lock up a prisoner for five or ten years, until he "ages out" of his crime-committing years, and that can reduce recidivism. But doing that for everyone would mean quadrupling our $60 billion-per-year prison system, and I'm hoping no one needs to explain why that would be a horrible idea.
Update: Sorry, one correction to Mauer's piece: It should be 300,000 fewer prisoners, not 400,000.