Predictive algorithm in use to deny convicts parole
Many Maryland residents are familiar with the film “Minority Report,” a science fiction thriller that depicts a futuristic world that has stumbled upon predictive crime fighting. In the movie, three gifted “precogs” are able to see the future and predict violent crimes, allowing law enforcement to stop the act before it even happens. A few twists and turns lead to the program’s inevitable failure, but chief among them is this question: how can you punish someone for a crime they did not commit?
You would think this sort of thing would strictly be left in “fantasyland” — a mere Hollywood idea. But you would be wrong (kind of). A predictive crime formula is in use right now here in Maryland, which can highlight homicide convicts that are likely to repeat an offense. The program, also in use in Pennsylvania and soon to be in use in Washington D.C. (where “Minority Report” takes place), can be used to deny homicide convicts parole.
Using a variety of factors and elements — a couple dozen in total which include age, criminal history, geographic information and types of crimes committed — the formula can “identify eight future murderers out of 100,” according to the formula’s lead inventor.
Now, this formula certainly has good intentions; but the precedent it sets is a scary one. Are we on our way to “precogs” that can see the future? No, of course not.
But does this send us down a path where convicted criminals are basically denied parole simply because of their troubled past? That’s certainly possible. Denying someone a chance to prove they have been rehabilitated — someone who just may have a chance to contribute something positive to society after their ordeal — simply because a few numbers deem them a repeat offender seems like a slippery slope.
Source: Wired, “U.S. Cities Relying on Precog Software to Predict Murder,” Kim Zetter, Jan. 10, 2013
- Our law firm has experience handling violent crime cases; so to learn more, please visit our Baltimore criminal defense page.