Civilian Casualties

The police force was designed to serve and protect its community, but what happens when they begin to serve themselves? Within chapter 5 of “Weapons of Math Destruction”, Cathy O’Neil takes up the topic of justice system models. In order to better “protect” a community, police forces have turned to crime prediction software to optimize police patrolling. One such service, designed by PredPol, takes historical geographic data of recorded crimes and attempts to predict the locations where a crime is likely to be committed. While PredPol stresses “that the model is blind to race and ethnicity”, the applications of such a service brings cause to question this statement. In the United States where cities are vastly segregated “geography is a highly effective proxy for race”. From this, it can be seen immediately the possibility of bias in a model based on geographical data. Driving these models are two kinds of input data: locations of serious crimes and locations of “petty” crimes. Inherently there is a hypothesis that these two forms of criminal activity are related. Due to significant difference in the number of serious crimes and the number of petty crimes reported, the models became flooded with nuisance data points leading to predictions of increased crime in hotspots of a city. Unfortunately, these zones tend to be occupied by the poor and minorities. As is all too common in the analysis of these models, a serious feedback loop ensues from its application. As the police begin to frequent these areas, more individuals are arrested for these petty crimes, overpopulating the prison system with criminals of “victimless crimes”. What O’Neil proposes as a solution, and what I agree with, is to discard the nuisance crime data from the model. While beneficial in some situations, the models are proven to be used in unjust ways, further justifying the racist beliefs of some police officers. Over the course of this reading, it’s hard to not to feel uncomfortable with the use of systems of this sort. Justification of oppressive beliefs is being hidden within “scientific methods”, making it harder for the policed to object to the policing.