A recent Union Leader Editorial said predictive policing should be fixed rather than abolished. The problem is that predictive policing can’t be fixed.
Thanks to a FOIA request by the Brennan Center we know that, in NYC specifically, the data falls into three categories: Time and Location; Individuals predisposed to Crime; and An environment that enables Crime.
These categories create a feedback loop.
For example, in NH African-Americans are 2.8 times more likely to be arrested, and 5.2 times more likely to be in jail than a white person, and Hispanics are 2.5 times more likely to be in jail than a white person.
This data is then used in the algorithms that help police “predict” where crime will happen. Police then increase patrols in these predominantly minority neighborhoods, and make more arrests. The cycle continues.
If the models showed more crime in wealthy neighborhoods, the cops might also find illegal poker games and someone taking a bump of coke. However, wealthy people can afford better attorneys and might not be convicted.
Our current justice system relies on getting easy convictions – disproportionately – for victimless offenses; and too many of the crimes with victims are either never closed or lead to a wrongful conviction. The national rate of wrongful convictions is estimated to be between 2% to 10%. Instead of relying primarily on plea deals to get an easy conviction, our justice system should instead put more focus on convicting the actual criminals who harmed one of our neighbors.
