With the advent of open data and new, powerful methods for analyzing it, we’re learning a lot that could challenge longstanding beliefs on public policy. Politicians, social workers and other civil servants have always had data, of course; they just never had as much and could never do with it what they can today. They should listen to what the computers tell them.

What’s possible

Recent HIV research from Brown University is a great example of what’s possible. Researchers formulated a computer model based on numerous factors relating to drug use, sexual activity and the medical aspects of HIV infection. To ensure it was accurate, they calibrated the model until it could accurately reproduce known HIV infection rates in New York City from 1992 until 2002. They ran the model thousands of times on a supercomputer.

They found that the rate of of HIV infection among New York City injection drug…