Philadelphia is experimenting with risk assessment tools as part of its MacArthur Foundation-backed initiative to reduce the local prison population by 34 percent.

The forthcoming technology will use police data to predict the likelihood of re-arrest among people who have come into conflict with the criminal justice system.

It’s a controversial undertaking: Advocates in the reentry community argue that the tool, pre-loaded with historical data tainted by a legacy of racist policing, will inherently target Black and Brown communities. Richard Berk, the Penn statistician tasked with building the tool, has argued that the tool must forgo fairness for accuracy.

(Side note: Read this 2016 BloombergTechnology profile on Berk.)

Not to mention how much of a boon crime-forecasting technologies can be for strapped police budgets: The predictive technology Berk and his team built for Philadelphia’s Adult Probation and Parole Department between 2006 and 2014 “helped the probation staff handle a 28 percent increase in overall caseload with a staff 15 percent smaller” than before it was introduced.

In a feature on the future of predictive policing for science journal Nature, Penn doctoral student Aaron Shapiro argues new forecasting technologies can only deliver one small sliver of reform.

There need to be “regulatory and institutional changes” made to the criminal justice system, he writes. “We should be wary of relying on commercial products that can have unanticipated and adverse effects on civil rights and social justice.”

Shapiro also gives Callowhill-based mapping firm Azavea‘s predictive policing suite HunchLab a shout for “operating in good faith” and using analytics to “improve policing, public safety and officer accountability.”

Give it a read.