Loading…
27
Showerthought: I was dead wrong about bail algorithms until I sat through a 2 hour hearing in Cook County
Six months ago I thought risk scores deciding bail was just lazy math dressed up as justice. Then I sat in on a Cook County hearing where a judge overrode the tool, kept a guy locked up on a score of 2, and the guy got caught on camera two weeks later doing the exact thing he was charged with. That one hearing flipped me, because it showed the score wasn't the decision, it was just a starting point the judge could toss. Now I want the tools used more, not less, but only if the override gets written down and checked. Anyone got a link to how other counties track judges who ignore the score?
1 comments
Log in to join the discussion
Log In1 Comment
the_tessa25d ago
The thing is, the score is already just a starting point, that's how these tools are built, judges can always override in either direction. What you saw in that one hearing wasn't the tool working, it was one judge using it and then doing his own thing, which is the whole system really. I get why that flipped you because it does feel good when a human catches something the math missed, but you can't build policy off one case where the override happened to be right. There are way more cases where a judge overrides and the person was no danger at all and just sits in jail for weeks because of a gut feeling. The record keeping you're asking for does exist in some places, a few states log override rates and reasons, and the numbers usually show judges override to lock people up way more than they override to let them go. So if you want scores used more, that part is fine, but the override is the part that worries me, not the math.
3