A corollary of the “innocent until proven guilty” standard is that there’s just no way to compute such statistics. It’s impossible by design to distinguish between a clearly innocent person and a very suspicious person whose guilt couldn’t quite be proven without looking into the details of their case.
I've never understood that. There's a claim you are guilty of robbery. You get charged and go to court. You plead not guilty. The da insists and you go to trial. During trial, you manage to provide cctv showing that you were 1000 miles away. The jury says "not guilty".
Is there any reasonable way to say "we don't know, he might be guilty, we just didn't reach 'beyond reasonable doubt '"?
"It’s impossible by design to distinguish between a clearly innocent person and a very suspicious person whose guilt couldn’t quite be proven"
Oh, you mean another innocent person? (Emphais mine)