D&S affiliate Angèle Christin writes a response piece to Cathy O’Neil’s Weapons of Math Destruction.
One of the most striking findings of my research so far is that there is often a major gap between what the top administrations of criminal courts say about risk scores and the ways in which judges, prosecutors, and court officers actually use them. When asked about risk scores, higher-ups often praise them unequivocally. For them, algorithmic techniques bear the promise of more objective sentencing decisions. They count on the instruments to help them empty their jails, reduce racial discrimination, and reduce expenses. They can’t get enough of them: most courts now rely on as many as four, five, or six different risk-assessment tools.
Yet it is unclear whether these risk scores always have the meaningful effect on criminal proceedings that their designers intended. During my observations, I realized that risk scores were often ignored. The scores were printed out and added to the heavy paper files about defendants, but prosecutors, attorneys, and judges never discussed them. The scores were not part of the plea bargaining and negotiation process. In fact, most of judges and prosecutors told me that they did not trust the risk scores at all. Why should they follow the recommendations of a model built by a for-profit company that they knew nothing about, using data they didn’t control? They didn’t see the point. For better or worse, they trusted their own expertise and experience instead.