How London Cops Used Low-Tech Version of Predictive Policing
While reporting on how big data could be used to help law enforcement identify the next felon, or conversely, target innocent people, I heard one anecdote about predictive policing that was decidedly low tech, but no less intriguing.
It came from Peter Ship, a former detective with London's Metropolitan Police Service, who described an effort that he and his colleagues used in the 1990s to try to reduce the number of rapes occurring in the city.
Using police files, they wanted to see whether they could find common patterns of behavior of convicted rapists that might help investigators understand the criminals' progression to violence -- and perhaps prevent it. At the time, it was a new way to approach their work.
Since this was before all the hype around "big data" and what algorithms could do, detectives studied the files by hand. No doubt it was a slower process, but amid the page turning, one surprising theme jumped out: Long before they were arrested for assaulting women, many of the rapists had been arrested for the same crime - stealing women's underwear from clotheslines.
With this discovery, investigators started flagging certain men as being at high risk for committing rape. They were placed on watch lists and paid preemptive visits by social workers, Ship said. It helped speed investigators' work and allow them to more quickly identify suspects, he said, although he didn't specify to what degree it reduced the number of rapes.
"It's about getting ahead of the game," Ship said.
The Metropolitan Police Service declined to comment.
The anecdote highlights law enforcement's longstanding interest in predictive policing, an approach that is gaining traction with the growth in analytics software and the accumulation of vast amounts of personal data through e-mail, social media and cellphones.
Of course, as highlighted in my story, using computers to try to identify suspects before they commit the crime, like the precogs in "Minority Report," is both promising and problematic. Jim Adler, the former chief of privacy at Intelius, a major seller of online background checks, did a research project on this using tens of thousands of criminal records. By examining only a few details on each individual in the reports, he said he was able to identify most of the felons, without knowing ahead of time that they were arrested for serious crimes. However, a number of non-felons were also wrongly pinpointed.
Which just goes to show you that there may be the perfect crime, but as for the perfect crime-fighting tool? They're still working on that.
Bloomberg reserves the right to edit or remove comments but is under no obligation to do so, or to explain individual moderation decisions.