Can computers predict urban crime? UChicago Study Says Yes

A group of social scientists at the University of Chicago claims to have invented a computer algorithm that: can predict future crimes up to a week in advance with an accuracy of 90%.

They tested the model’s accuracy in eight major U.S. cities, each measuring about two blocks by two blocks in small squares. Starting with their home base, Chicago, they studied historical records of violent crime and property crime within each square, recording the changes and different patterns over time to make predictions for future events. They say the model worked equally well in seven other cities (Atlanta, Austin, Detroit, Los Angeles, Philadelphia, Portland and San Francisco), giving them an algorithm with “predictive accuracy far greater than in the past.” reaches”. to write.

Their tool differs from previous models, which tended to pin crime to geographic ‘hotspots’, relying instead on what their paper Nature Human behavior calls “spatiotemporal point processes that unfold in a social context.” By analyzing hundreds of thousands of different patterns, they claim they are able to determine the risk of crime at a specific time and space. They say this allows them to see not only how crime changes over time, but also how the police force evolves next to it.

This “crime prevention” style has never found a big following among criminal justice reformers — and may even raise flags with those deterred by the pre-crime screenplay famously depicted in the 2002 film. Minority Report, based on a story by Philip K. Dick.

In fact, models seeking to predict crime before it ever occurs have a history of being inaccurate, exacerbating racial inequalities and also justifying police concentrating on affluent areas of the city. One of the most infamous models was the Chicago Police Department’s surveillance system, which was used between 2012 and 2019 to monitor people it said had a “high propensity for violent, gang-related crime.” A City of Chicago Inspector General report later found that of the 398,684 individuals on this “strategic suspect list” (which included every person arrested and fingerprinted since 1993), only 16% have ever been confirmed to be gang members, and one Chicago Tribune research found 13% had never been charged with a violent crime.

The researchers at the University of Chicago write that they are well aware of these abuses. They claim that their algorithm can even be used to control the police themselves, almost karmically: Their algorithm has found evidence that the police of these cities respond better to crime in predominantly white, higher-income areas than to crime in less affluent neighborhoods.

Nature adds its own quasi-disclaimer to their research, via a side comment entitled “The Promises and Dangers of Crime Predictionwritten by Andrew Papachristos, a Northwestern University sociologist known for his research on gun violence. Papachristos has criticized police for ‘commander'[ing]” its past research and use it to identify “strategic topics” and guide enforcement operations.” He applauds his colleagues’ new model, but leaves this ominous comment: “However, the question of what others will do with these powerful new statistical research tools is perhaps a more fraught task.”

Leave a Comment

Your email address will not be published.