The Imprisoner’s Dilemma

I would go a step beyond just "diminishing returns". Higher than optimal incarceration rate actively creates more crime by way of 1-2 million broken homes.

I'm a little worried about this article. It seems to me that it might be emphasizing certain aspects of the data that tend to support a conclusion that is relatively popular in many circles. (Indeed, I'd like to see the U.S. imprison fewer people; but that doesn't mean we should distort the data.)

In particular, the conclusions are largely about the marginal effect of increased incarceration rates in the last couple of decades. So: we should not imprison more people than we are now. Fine. But that is not the same as saying that the huge increase in incarceration since the mid-1970s did not cause a large -- and highly beneficial -- decrease in crime.

According to the graphs, incarceration rates went up by about 60% in the 1990s and have remained more or less flat ever since. But rates more than doubled in the 1980s. This fact, and its possible impact on crime rates, apparently go completely unexamined in the article.

So don't think this article is up to the usual standards of fivethirtyeight.com -- unless I'm missing something.

I thought I have read that with CompStat, precincts also started fudging what crimes they were writing up. Major offenses were getting written up by police as smaller offenses to "cook the books".