Eliezer: I can actually think of one case in which the argument "It has a small probability of success, but we should pursue it, because the probability if we don't try is zero".

Say someone is dying of a usually-fatal disease, and there's an experimental treatment available that has only a small probability of working. If the goal is to not have the person die, it makes more sense to try the experimental treatment than not try it, because if you don't try it, the person is going to die anyway.