Predictive Policing Is Racist Garbage

Seeking to dress up the racist repressive apparatus of the capitalist state in modern guise, courts and police departments around the nation have been turning to “predictive policing” tools which claim to foretell mathematically who is likely to commit crime in the future. This marriage of all-American racism with Silicon Valley hype about “machine learning” is straight out of a dystopian science fiction story, recalling Minority Report’s department of “Precrime.”

The leading edge of this has been in sentencing of convicted criminals. Whether under mandatory sentencing guidelines or in the hands of individual judges, sentencing in capitalist law has always maintained and reflected the overall structure of racist American society—victimizing the working class, the poor and minorities in the service of enforcing capitalist “law and order.” Now, so-called risk-assessment tools and methodologies have been adopted by an increasing number of states from Utah to Wisconsin and Florida. These tools, often developed by private companies and contracted to state agencies, “score” those dragged before the courts on their supposed likelihood to commit future crimes, and are taken into account in sentencing and parole decisions.

We do not live in a science fiction novel, and machines do not “think” smarter or differently than the people who have commissioned and built them. The algorithms and formulas deployed—considering factors ranging from gender to supposed “beliefs favorable towards crime” (termed “antisocial cognition”) to education level and employment history—are no less race- and class-biased than the rotten court system that they are built to serve. This pretense to “evidence-based” sentencing in fact amounts to a sort of prejudice-laundering that takes the gross racism of capitalist society and packages it up in “impartial” form by attributing it to a machine rather than those who built the machine.

Though using modern technology, these tools which claim to predict people’s innate criminal tendencies and future acts recall the phrenological pseudoscience of the 19th century. While the racist mad scientists of that era claimed to be able to assess people by the size and shape of their skulls, the idea of calculating people’s criminality by their employment history and upbringing, or their “antisocial cognition” is no less grotesque. In taking into account factors directly related to wealth, class and social position, these algorithms make concrete the visceral contempt of America’s racist rulers for those whom they dominate and exploit.

Ironically, by encoding the moral calculus of the ruling class in mathematical formulas, these tools illustrate how racist the American state is, to the very core. For example, a recent investigation by ProPublica (“Machine Bias: There’s software used across the country to predict future criminals. And it’s biased against blacks.” 23 May) obtained the risk scores of thousands of arrestees in Broward County, Florida (produced by a system developed by the private company Northpointe). Their conclusion: “Only 20 percent of the people predicted to commit violent crimes actually went on to do so.” However, Northpointe’s tool was very good at predicting race. Black defendants were wrongly flagged as future criminals at almost twice the rate of white defendants.

Even when controlling for age, gender and criminal history, ProPublica found that “Black defendants were still 77 percent more likely to be pegged as at higher risk of committing a future violent crime and 45 percent more likely to be predicted to commit a future crime of any kind.” For example, the study contrasted the cases of two petty theft arrests. A 41-year-old white man with two prior convictions for armed robbery was given a score of 3 (a low risk of reoffending) but was later convicted of grand theft. On the other hand, an 18-year-old black woman was given a score of 8 (a high risk), despite having only four juvenile misdemeanors; she has not since reoffended.

These tools are not only being deployed by courts, but, increasingly, by police departments seeking to predict future criminals to monitor. For example, the Chicago Police Department (revealed last year by the Guardian to be operating a “domestic equivalent of a CIA black site” at its Homan Square building) boasts of its “Strategic Subject List” which it claims predicts those likely to be involved in gun violence. Cathy O’Neil, author of the new book Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy, highlighted on her blog (mathbabe.org) a recent study from the Journal of Experimental Criminology. The study, “Predictions Put into Practice,” determined that the list had no impact on homicide levels. Being on the list made no difference in people’s chance of being shot, but did mean that they were more likely to be surveilled by the police and arrested. O’Neil summarized the study’s conclusions: “From now on, I’ll refer to Chicago’s ‘Heat List’ as a way for the police to predict their own future harassment and arrest practices.”

Fundamentally, the racist, brutal nature of the American justice system is not a product of “bias,” individual or systemic. The capitalist state exists to defend the rule of the bourgeoisie against the working class and the oppressed. In America, it is inextricable from the racial oppression and segregation of black people, rooted in the history of slavery, which is used to divide the working class and undermine prospects of integrated proletarian struggle. It will take a socialist revolution to sweep the capitalists out of power, dismantle and break up their racist system of cops and courts, and put in place a society where genuine equality is possible.