5 Comments
User's avatar
Hyolobrika's avatar

I'm not seriously suggesting this but it would be ideal if people who serve in or regulate courts have been a) the victim of a crime and b) the victim of a false conviction, just so they know what it's like and can make decisions accordingly.

Expand full comment
Peter Hollingsworth's avatar

The risk in both cases is that you start with totally wonky priors. If the confidence is unreasonably high you end up way off the evidence driving your outcomes.

Expand full comment
Hyolobrika's avatar

Considering the question of how many false convictions we should tolerate, I think it heavily depends on the punishment.

If you're getting the death penalty for it, the courts had better be damn sure that you're guilty. If you are getting a small fine, or maybe going to a Scandinavian prison, maybe not so much.

Expand full comment
Tom Chivers's avatar

that's true, it's an expected value calculation: probability of outcome × impact of outcome

Expand full comment
eric's avatar

Be interesting to revisit the subject after the obvious trail verdict. Maybe worth saying, though, that along with the US freedom to publish more about ongoing trials comes a highly politicised legal system, along with prosecutors mounting big PR operations to sway jurors in advance of selection and actually during the trail, often using 'evidence' inadmissible in court. Judges in the UK take careful note of the likelihood jurors might have meaningfully have had access to a foreign publication before considering reporting restrictions.

Expand full comment