From where you said, "Recidivism isn't innocence and this is still a waste of time argument." I took it that you were saying that innocence, not recidivism, is the appropriate metric.



It is ridiculous, I agree, but that is what follows when innocence is the basis for determining whether a prisoner may be released from prison. A man who was imprisoned for stealing a loaf of bread is not innocent, so he cannot be freed if innocence is the standard, so his sentence is a life sentence.




That "recidivism is not innocence" is a weak counter to my earlier point.




It flies as an example of a species of problem to which we may apply the general principle derived from a good-faith answer to the OP's hypothetical situation. (What is good for the set is good for the subset.) Thereby, it flies as an example as to why the question is not silly.




That's how you find out about individuals, and they are getting better at it. But how you find out about groups is you look at how often people released get brought back in for a new offense in light of factors like kind of treatment, completion of treatment, etc. vs. control groups. In large enough numbers the aggregate behavior becomes predictable. You obviously wouldn't use this to the exclusion of more individual methods, but it provides a baseline. My question is not concerned with the individual variations, but with the baseline, which is why I phrased it in terms of "all" prisoners.



Here's why railroad/Sophie's Choice questions aren't very useful in moral philosophy. Using your example, the value of a baby and the value of an adult are necessary inputs to say whether you'd pull the switch. But they are also the true desired output. Nobody really cares whether the switch gets pulled, they only care what that says about the value of babies vs. adults. Asking a railroad question is an underwear-gnome strategy. Step one: frame as railroad question. Step two: ... Step three: profit! My question is not analogous to railroad questions in this sense.

Moreover, while it is hypothetical, it is not fantastical. It may not be practical to test one's hypothetical answer, but the possible answers are nonetheless falsifiable. (Some railroad questions may claim the same; but as I described, they are useless for an entirely different reason.)



I don't think my question disallowed the realistic, good-faith answer, "I don't know."