Skip to Content

Crime Software: Still Awaiting a Verdict

Scientific data not yet in on how much crime-reduction software predicts, and how much it motivates.
March 6, 2013

Seattle last week became the latest city to install software that tries to predict where crime will strike, giving cops an edge. Indeed, crime reductions have been observed in cities where the tool—called Predpol—is installed and its recommendations followed. But with a dozen cities having joined the bandwagon, one fact is worth noting: no scientific paper has showed to what extent the software itself deserves credit, and whether the power of suggestion, and increased efforts on the officers’ part, might be playing a role.

The whole phenomenon reminds me somewhat of the perpetual war news organizations wage on missed deadlines. Sometimes the organizations decide to buy new software for tracking stories. But you can bet the new product is accompanied by meetings and urgent e-mails: “Starting today, we’re going to get our deadlines on track. Everybody must use the new software.” Presto, stories start moving along more efficiently. But why?

Predpol analyzes the history of where crimes (such as burglaries and car break-ins) occurred—today, yesterday, farther in the past—and places 500-foot by 500-foot red boxes on maps in advance of each patrol shift. Cruisers are meant to visit those areas at any opportunity, standing a higher chance of either catching or scaring off the miscreants. One Los Angeles precinct following the regimen experienced a 25 percent drop in reported burglaries, an anomaly compared to other precincts (see “L.A. Cops Embrace Crime Predicting Algorithm”). Predpol credits the software’s “advanced mathematics and adaptive computer learning” and says the software does better than human analysts.

Where’s the detailed data? This week Jeff Brantingham, a company cofounder and an anthropologist at the University of California, Los Angeles, e-mailed me to say he’s still working getting his research published. He did tell me something new: where the tool was used in Kent, U.K., surveys of the public (who didn’t know the software was in use) showed they perceived more police patrols. And that’s evidence of increased (or wider-ranging) patrols, for sure.

It’s not hard to imagine that software might do better than humans, especially over longer timescales. But let’s face it—police brass want their initiatives to succeed. And if you give an officer a map with red boxes on it—most of them naturally in sketchier areas—and order him to go there, it’s not unreasonable to think he’ll spend more time there than he otherwise might. (Police officers are people, too.) And officers report feeling a heightened state of awareness when in these zones. That psychological effect surely makes them do their jobs better, regardless of prediction accuracy. Finally, if you let software spit out the patrol orders—and abolish any planning meetings at the precinct house—cops will be on the road for that much more time. 

I’m eager to see the published data. Doubtless, it will show a benefit from the software. But even if Predpol’s predictions were nothing but placebos, don’t forget that sometimes a placebo can be just as effective as the real thing. 

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.