Chicago PD's predictive policing tool has been shut down after 8 years of catastrophically bad results


In 2012, Chicago PD collaborated with the RAND Corporation and the Illinois Institute of Technology to automatically generate "risk scores" for people they arrested, which were supposed to predict the likelihood that the person would be a "party to violence" in the future (this program was called "TRAP" — Targeted Repeat-Offender Apprehension Program" — seemingly without a shred of irony). Now, that program has been shut down, and the City of Chicago's Office of the Inspector General has published a damning report on its eight-year reign, revealing the ways in which the program discriminated against the people ensnared in it, without reducing violent crime.


Jail abolition lawyer Shakeer Rahman has published an excellent Twitter thread going through the report's highlights. A few of the points he makes:

* People who were assigned high risk scores by TRAP were subjected to "enhanced prosecutions"

* Routine traffic stops and other activities that pose a high risk for racial profiling were pretenses for generating TRAP scores

* Your TRAP score went up even if were acquitted or had your case dismissed (Rahman: "In other words, police run arrest people on racist/arbitrary whims, and then those arrests become 'data' showing you should be targeted again")

* Cops could access TRAP scores at will, and routinely violated the rules about sharing these scores outside of policing/criminal justice contexts

(via Naked Capitalism)