3 ms·
There is no quantitative measure for the "net positive" you describe. The point is that through the consequentialist lens, there exists some cost/benefit ratio
by nosefurhairdo 3y ago
There is no quantitative measure for the "net positive" you describe. The point is that through the consequentialist lens, there exists some cost/benefit ratio where family punishment becomes morally correct. This is in contrast to retributivists, whom according to Caplan believe that "...it is morally impermissible intentionally to punish the innocent..."
It's much easier to form coherent moral arguments when you identify clear boundaries for moral behavior versus the "ends justifying means" nature of consequentialism. Some might find that intellectually lazy, others might view it as a requirement for a consistent morality.
And no, the argument that family punishment doesn't work is not compelling when up against Caplan's historical examples of their efficacy. This is just back-rationalizing a morality not actually grounded in consequentialism. A consequentialist who isn't totally historically ignorant should at least be able to entertain the prospect of family punishment satisfying their moral framework, regardless of whether they fully buy in.
- InSteady 3y agoTo me, the way you've characterized this (which rings quite true) is a great example of why a lot of philosophical arguing doesn't feel terribly relevant in the modern world. What practical person is going to constrain themselves to have no clear boundaries whatsoever, simply because they find it useful to frequently assess actions by their downstream effects instead of exclusively by initial conditions? Don't get me wrong, it can be useful to do thought experiments under these kinds of strict constraints, but only up to a point. Under retributivism, family punishment could also be incredibly effective up until the point they declare "no punishing of the innocent." That it is a gotcha that consequentialists are prohibited for doing same is just silly. People make allowances, exceptions, and put hard limits on otherwise open-ended systems in order to make them practicable. Why would consequentialism fall apart if you had to put a few sanity checks into it, such as "don't punish the innocent?" For the sake of thought experiments, sure, see how each side plays out under very strict rules. But the real world is messy, doesn't fit nicely into one particular "ism" all the time. Your entire belief system shouldn't be predicated on thought experiments.. therefore there is no reason to get so dogmatic about building moral justifications from only one direction. Especially when a flexible or hybrid approach offers such obvious advantages.