Lawrence Sherman: A lot of what I’m saying depends on the accuracy of forecasting, and simply because of weather forecasts and other experience people have had, they tend to be a bit skeptical because there is a range of error around any forecast. When the RAND Corporation developed the idea of selective incapacitation with NIJ grants 30 years ago and recommended that we try to use prison much more selectively to keep the prison population down — to keep the cost down — the National Academy of Sciences review panel was very skeptical because they thought the false-positive rates, the errors would be way too high. And the tragedy is that by keeping our hands clean, criminology stayed away from decisions that judges and prosecutors and police were going to make anyway. And when they made those decisions without guidance from criminology, what they decided to do was to play it safe and to put as many people in prison as possible. And so 500 percent growth in the prison population later, I think it’s time for criminology to ask, “Do we want to get our hands dirty? Do we want to make the best predictions that we can make even if they’re going to have error?” Because those predictions will be much more accurate than people looking at a rap sheet and saying, “This guy’s dangerous; lock him up,” when in fact his first offense was not until age 21 or it’s all been property offenses. The cost-effectiveness of locking that person up as opposed to keeping close tabs on him out on the street would be very low. We now, I think, have to accept that the prediction tools have gotten a lot better — we have supercomputers, we have vast databases that are comparable to the weather forecasting databases. Weather forecasting has gotten twice as accurate in the last 20 or 30 years than it was beforehand. And most of us don’t pay any attention to that sort of thing. But just compare it to volcano prediction, and you’ll see how much better weather forecasting is than other kinds of forecasting. And crime is getting very close to at least a weather forecast for the next day, or at least for the next 12 hours. If you’re going to look in the short term like, in the next two years, is this person going to commit a horrible crime of the kind that we absolutely have to protect the public against? We can make those kinds of forecasts with much better accuracy than judges and prosecutors are making by the seat of their pants. And I think the prosecutors, when given the opportunity to base those decisions on those models, will be grateful to have something to hang their hat on as opposed to having to take all of the blame and get none of the help that criminology can offer.