I don’t see what difference the stochastic dominance point makes to the conclusion. Under severe imprecision (P3), it seems at least as hard to compare actions with the (imprecise) stochastic dominance relation, as with (imprecise) EV-broadly-construed.
Your argument seems to be pretty strong insofar as you point that many times we don’t really know what the expected value of an action is.
It is defused if we can retain some notion of “a is better than b” without needing to defend that we can compute the expected values to any degree of granularity.
Stochastic dominance and similar concepts provide a way to do this in a way that may be apparent to many readers. We can be confident that a is stochastically dominant over b while being extremely uncertain about their respective expected values. The space of problems over which we are able to make decisions expands.
The stochastic dominance point is helpful, ty
I don’t see what difference the stochastic dominance point makes to the conclusion. Under severe imprecision (P3), it seems at least as hard to compare actions with the (imprecise) stochastic dominance relation, as with (imprecise) EV-broadly-construed.
The difference is:
Your argument seems to be pretty strong insofar as you point that many times we don’t really know what the expected value of an action is.
It is defused if we can retain some notion of “a is better than b” without needing to defend that we can compute the expected values to any degree of granularity.
Stochastic dominance and similar concepts provide a way to do this in a way that may be apparent to many readers. We can be confident that a is stochastically dominant over b while being extremely uncertain about their respective expected values. The space of problems over which we are able to make decisions expands.
Cheers