I'm reading Alasdair MacIntyre's book After Virtue, which is a general attack on modern ethical philosophy and (among other things) a defense of Aristotelian attitudes towards morality. It's a well-written and occasionally provocative book; I'll have more to say about it when I finish it, but I wanted to flag a supposed counterexample to the no-ought-from-is principle that MacIntyre attributes to A.N. Prior:
He is a seaman. Therefore he ought to do whatever it is a seaman ought to do.
Let's replace the noun to sharpen the point:
He is a homosexual. Therefore he ought to do whatever it is a homosexual ought to do.
On the idiomatic reading of the inference -- "he is a homosexual, therefore he should behave like one" -- the ought, if it's a moral ought, doesn't follow from the is. (Ask Ted Haggard.) If, on the other hand, it is a prudential ought -- if he doesn't behave like a homosexual he will cease to be a homosexual -- then the inference goes from a fact to a fact. Alternatively, it goes from a desire to continue to be a homosexual and a set of beliefs about the world and language to a course of action that would fulfill the desire. This isn't an is-ought inference, it's instrumental reasoning.
Let's try the narrow reading, which I take to be:
for all X and q (if all H's ought to do X and q is an H then q ought to do X).
p is an H.
for all X (if all H's ought to do X then p ought to do X).
Assuming ought statements obey first-order logic, this is valid reasoning; however, the conclusion has no normative force. To make the conclusion normative one would have to satisfy the protasis with an ought statement, in which case one would have deduced an ought from an ought.
Of course it's possible to construct an ethical system with very few extralogical principles; utilitarianism is an example; but you can't do it without some extralogical input, and there is no reason to suppose that a small number of universal core principles gives you a better ethics than a laundry list of ought statements. (And good reason to suppose that there's no Occam-like principle -- all known simple systems are simplistic.) So it's difficult and probably impossible to come up with a useful way to adjudicate between moral systems that's independent of whether you like the results -- which limits the power of moral reasoning to guide your actions.
NB I was misquoting from memory; Prior's original example said "sea-captain" rather than "seaman." For obvious reasons I will not correct the post.