A model called COMPAS which takes a crap ton of inputs to predict recidivism doesn't do substantially better than a simple model you can calculate by hand: http

A model called COMPAS which takes a crap ton of inputs to predict recidivism doesn't do substantially better than a simple model you can calculate by hand: https://www.theatlantic.com/technology/archive/2018/01/equivant-compas-algorithm/550646/ https://www.theatlantic.com/technology/archive/2018/01/equiv... I wonder if the model described in TFA will do any good.