Re: That's silly though - Edit 1
Before modification by Burr at 05/03/2010 11:13:57 PM
There is some percentage at which it becomes either more practical or more ethical (or both) to release everyone.
Not really. The only way your 'example' could ever come up is if the government never jailed any REAL criminals and let them run loose. Not even the worst governments have that poor a record. It will never be anything like that.
It is an important question because of its implications for determining what counts as having been rehabilitated. If you have 100,000 thieves with an average 5% probability of reoffending, then you know about 99,500 of them are rehabilitated and 500 of them aren't, but you can't necessarily identify which ones are which. Is it practical or ethical to continue the imprisonment of 99,500 rehabilitated prisoners just because you can't identify the 500 non-rehabilitated prisoners? If so, then what if the probability of reoffending were even less, so that there were only 1 non-rehabilitated prisoner among two million rehabilitated prisoners? At what percentage does the balance lie?
Recidivism isn't innocence and this is still a waste of time argument. In actuality, recidivism isn't that hard to predict if you don't have a political agenda. Of course, no system is foolproof so you will invariably release some folks who will re-offend and keep some who never would.
If innocence were the appropriate metric, then all prison sentences would be life sentences. If we'd rather concern ourselves with reasonable measures, then it is far more relevant that a rehabilitated prisoner set free poses no more risk or harm to society than any other person.
However you will quite simply never reach some kind of extreme 'we have to let everyone go now'.
Hundreds of juvenile convictions by one corrupt judge were tossed out about a year ago, and the juveniles' records were expunged. (http://www.cbsnews.com/stories/2009/03/26/national/main4895883.shtml) Likely, due process was violated for all of them. But it was, say, within the realm of plausibility for investigators to find a e-mail where the corrupt judge states something like, "65% of your wards you only got because of me, Harry." Therefore, it was within the realm of plausibility for the situation to match my hypothetical one. Thus, it is not some kind of extreme even in the particular.
Besides, if you couldn't identify WHO was going to re-offend, how could you ever have the stats you describe above? The entire thing is nonsensical.
The stats are modeled on past observations. That's why you don't know who; you only know the aggregate behavior.