The Minority Report, in the real life
The classic Philip K. Dick novella (and Tom Cruise action flick) just took a step closer to reality. This time, instead of using “precogs” to track down crime before it happens, we’re using computer algorithms.
This isn’t the first algorithm used by police to identify criminal hotspots, but it uses a new approach that’s intended to eliminate (or at least identify) biases in policing.
* The report was published in the Nature journal under the title “Event-level prediction of urban crime reveals a signature of enforcement bias in US cities” (via Bloomberg).
* Using Chicago as a test, it was able to predict criminal activity with 90% accuracy, with similar results in other major US cities.
* It breaks urban areas into 1000 square-foot-tiles, then uses historical crime data to predict where crime is most likely to occur.
Different approach:
* This differs from previous models, which worked from the assumption that criminal activity originated from fixed hotspots and spread from there.
* Social scientists at the University of Chicago argue that this approach ignores the “complex social environment of cities, as well as the nuanced relationship between crime and the effects of police enforcement.”
* Instead of directly policing these hotspots, the new algorithm pulls from “hundreds of thousands of sociological patterns” to predict increases in crime in certain areas a week in advance.
Limits of predictive algorithms:
* In the past, they’ve had a notoriously low hit rate (paywall), and often lead to even worse biases in policing low-income areas and individuals.
* Minority Report’s mutant precogs make mistakes, and these types of algorithms are no different.
* Will this one be any better, or is predictive policing best left to the realm of science fiction ?