Sentencing / Post-Conviction

Software Is Predicting Hard Crime

  • Print.

Cue Tom Cruise from Minority Report, where he played a “pre-crime chief” in the future who solved murders and made arrests before they happened. Researchers are pushing reality’s envelope nearer that fiction.

For one, software developed by Richard Belk, who teaches criminology and statistics at the University of Pennsylvania, predicts which probationers and parolees are probably going to murder again, or be murdered, reports.

It’s been helping probation officers in Baltimore and Philadelphia determine how much supervision is needed for parolees. Now Washington, D.C. is trying a tweaked version of the software that looks at other crimes, too.

Berk’s algorithm works with about two dozen variables, including the crime, a person’s criminal record, where they live and—the greatest predictor—their age at the time of the crime. The younger you were, the more likely you are to reoffend, Berk says.

Berk says the D.C. experiment with lesser crimes could lead to helping determine bail and sentences.

Then there’s predicting where crime will occur.

The Memphis, Tenn., police department, with help from the University of Memphis, began using software and technology in 2006 that helped reduce serious crime by more than 30 percent and serious crime by 15 percent.

Known as BLUE CRUSH (Criminal Reduction Utilizing Statistical History), the initiative uses IBM software for predictive analytics to work over data ranging from crime patterns, crime reports, profiles of known criminals, weather and other variables and identify likely hot spots for crime at various times, IBM said in a press release last month. The data allows police to better target patrols and checkpoints.

Two police departments in the U.K. also recently began using the CRUSH system, the Telegraph reports.

Give us feedback, share a story tip or update, or report an error.