That recidivism technology Philadelphia is getting might not account for racist policing

It turns out Philadelphia’s potentially racist recidivism algorithm will “lose some accuracy” if it isn’t designed to be, well, a racist recidivism algorithm.

Funded by the $3.5 million MacArthur grant and developed by Penn statistician Richard Berk, the fairness and accuracy of the pending computer algorithm — being built to predict the probability of recidivism after arrest — was the subject of a recent City Council hearing.

There, REDEEMED founder and criminal justice reform advocate Bill Cobb asked Berk if the technology will account for what Technical.ly Philly described as a “legacy of racist policing and policies” (a legacy that includes significantly higher rates of arrest, incarceration and unemployment among Black and Brown men than any other population).

“If we strive for fairness, unfortunately, we’re going to lose some accuracy,”said Berk. “That’s a price. It’s a price you may choose to pay.”

According to Technical.ly, Berk removed himself from the ethics and morality of the algorithm despite questioning from reform advocates like Cobb and Defender Association chief Keir Bradford-Gray, the latter of whom reminded everyone how criminal justice is fundamentally supposed to be fair and unbiased.

“We tend to deal with things on the backend” when it comes to reforming the criminal justice system, Bradford-Gray told us recently, rather than work to “limit who’s coming into the criminal justice system” in the first place.

In other words, even if the algorithm is built to consider things like the lasting trauma of childhood poverty, it will not truly break the cycle of recidivism until it can prevent that traumatized child’s potential trajectory into the criminal justice system.