How many innocent people should we be allowed to arrest and jail in order to prevent a single dangerous person from being free? The Supreme Court has refused to answer this question, but algorithms have, and many courts across the country are going with the algorithm.

At different stages of the criminal justice system, computerized risk-assessment algorithms are slowly replacing bail hearings in determining who goes to jail and who goes free. This is widely seen as progressive reform, but may in fact be leading to more incarceration, not less. While many are warning that these algorithms are biased, racist, or based on bad data, the real problems are in fact much deeper, and even harder to solve.

Guest voices include Megan Stevenson, John Raphling, Renee Bolinger, Georgi Gardiner, and Seth Lazar.

Join Slate Plus to unlock full, ad-free access to Hi-Phi Nation and the rest of your favorite Slate podcasts. Subscribe directly from the Hi-Phi Nation show page on Apple Podcasts or Spotify. Or, visit slate.com/hiphiplus to get access wherever you listen.

Learn more about your ad choices. Visit megaphone.fm/adchoices