3 ms·
> Paternity tests for example seem to have ~1.1% false positive rates Realistically I think the existing criminal justice system would be hard put to claim a w
by ajross 3y ago
> Paternity tests for example seem to have ~1.1% false positive rates
Realistically I think the existing criminal justice system would be hard put to claim a wrongful conviction rate lower than that, so... I'm not sure this argument cuts the way you want it to. If you have a 99% accurate test of guilt, that seems like more than enough probable cause to continue investigating, and probably good enough to convict with just a little extra corroboration.
While there's a privacy argument to be made, I really don't think you're going to get very far arguing that this is going to cause quality problems. Frankly I think it's likely to help with reducing false imprisonment (i.e. by fingering the right suspect earlier, before the wrong one ends up in court).
- Cpoll 3y agoThere's an old Bayesian thought exercise: Imagine that the test for Condition X has a 1/100 false positive rate (specificity), but X occurs in only 1/1000 people. If you select a random person and they test positive for X, it's more likely that they hit the 1% false positive than that they actually have X. (This comes up all the time in practice). So if you have a database of 10,000 people and 1.1% false positive rate, you're likely to get several false positives. Unless you stop searching as soon as you find a positive, of course...
- breather 3y ago> If you have a 99% accurate test of guilt, that seems like more than enough probable cause to continue investigating, and probably good enough to convict with just a little extra corroboration. If you're facing a jury (as you probably are in a criminal case) people have been convicted on far less than this with no other corroboration. It's all in the narrative put before the jury and their understanding of the evidential burden.