Is the best way to lessen incarceration to actually make catching criminals easier to the point that criminals know it? (Or something like that.)

If you want to respect writer Matthew Yglesias, don’t start by reading his Twitter feed.

I still haven’t figured out whether he’s extremely bad at hot takes or if he’s just trolling people.

Either way, his Twitter feed obviously lacks the nuance for which I first started reading him at TAP.

His longer form writing has its regular home at, and he has a thought-provoking piece up about America’s mass incarceration problem and ways we might start alleviating it.

It has a headline that is typically provocative for Yglesias: “The best way to end mass incarceration is to catch more criminals.”

A couple of years ago, Anne Sofie Tegner Anker, Jennifer Doleac, and Rasmus Landersø published a really interesting paper about the impact of a law passed in Denmark that allowed Danish police to add anyone charged with a felony to a DNA database, increasing the share of arrestees added from roughly 4 percent to about 40 percent.

So what was the impact? The authors “find that DNA registration reduces recidivism within the following year by up to 42%.”

That’s a big reduction. Obviously having your DNA sample in some database does not have a lot of rehabilitative power per se. But as with the classic fingerprinting1 of perpetrators, once someone is in the system, it’s easier to catch them if they commit a crime. And indeed, the authors find that databased criminals are less likely to re-offend, but if they do re-offend, they are more likely to be caught. Using some math, they “estimate the elasticity of crime with respect to the detection probability” and conclude that “a 1% higher detection probability reduces crime by more than 2%.” So what do all these registered former offenders do instead of crimes? Well, they “find that DNA registration increases the likelihood that offenders find employment, enroll in education, and live in a more stable family environment.” This is a great paper and a very cool result, and I think it makes a strong case for the expanded use of DNA databases.

But I think it also suggests a better way of thinking about the phenomenon of mass incarceration in the United States than the mode that takes a negative view toward all punitive measures.

In this case, raising the odds that a person will go to prison conditional on the commission of a crime lead to a disproportionately large reduction in the odds of offending, meaning fewer person-hours of prison rather than more. Incarceration is costly, cruel, and has certain properties that tend to encourage crime. We should try to create a society that is dedicating fewer resources to locking people in cages in abusive conditions, and the best way to do that is to reduce the number of crimes being committed.

He covers a lot of ground in this piece, but as is usual for Yglesias when he is at his best, he makes you think a bit more deeply about the issues he presents.

My biggest concern with giving everyday police officers and police officials easier access to expanded DNA databases and enhanced surveillance technology is that I, as a middle-class white guy of a certain age, will not generally have to worry too much about these developments. It’s poor people and people of color who are most abused by law enforcement.

We have a huge, mostly unacknowledged or willfully ignored, racist cop problem in this country. I don’t want to give those bad guys any more tools to arrest innocent people of color.

Facial recognition technology — much of it already shown to be unintentionally, but functionally, racist in its mistakes — scares me the most in this regard.

Leave a Reply