A Rejoinder to Dressel and Farid: New Study Finds Computer Algorithm Is More Accurate Than Humans at Predicting Arrest and as Good as a Group of 20 Lay Experts 1

[...]specifying relevant risk factors and providing feedback on the degree to which they predict recidivism "loaded the deck" for laypeople to predict more accurately than experts would in the much more complicated context of a real criminal case-with relevant and irrelevant information provided by the defense and prosecution and with no feedback about recidivism after an individual leaves the courtroom. [...]several studies show that if probation officers correctly identify the existence of dynamic criminogenic factors through the application of risk assessment and then attempt to ameliorate them through appropriate interventions, they can reduce an offender's likelihood of recidivating. [...]it is critical to note that when empirically constructed risk instruments capable of identifying dynamic criminogenic factors are not being used by community corrections staff, officers will often engage in supervision practices that focus on addressing issues uncorrelated with crime. [...]this study overlooks an entire body of literature where community corrections professionals working in the criminal justice field discard risk assessment recommendations for their own "seat of the pants" judgments.