In my discussion in comments with Micah Cowen, an important component of ethics has emerged: the ideas of enforceability and enforcement.
It seems to me that enforceability and actual enforcement constitute important demarcation criteria. There are those ideas, beliefs, preferences and desires which can be enforced and those that cannot be enforced. Among the ideas that can be enforced, there are those which you approves of enforcing (in a broad sense, from police and prisons to social opprobrium to perhaps criticism), and those you do not approve of enforcing.
It's an arbitrary semantic decision how to label these demarcation criteria, or the ideas on one side or the other of each criterion. You can label as "ethical" only those preferences which you (or someone) is willing to enforce or you can apply the label more broadly to any sort of preferences established by abstract reasoning. More to the point, though, is that regardless of the label, our discourse about preferences we are willing to enforce seems very different from our discourse about preferences we are not willing to enforce.
And, frankly, I find descriptions of people's preferences — especially their preferences about my own behavior — they're not willing to enforce to be monumentally tedious. If you disapprove of I behave, or how I think, or my specific beliefs or preferences, and you're not willing to actually force (or try to force) me to change in favor of your preferences, I'm just not interested. (If you're my friend, and you're willing to withdraw your friendship, that's a kind of force. (A kind of force I generally approve of.))
As I noted yesterday, Colin McGinn "ethically" disapproves of faith-based belief formation. I'm not (and this will come to as a shock to my regular readers) all that thrilled about faith myself. But, absent an intention to actually force people to comply with that preference, the theist can simply employ the ethical equivalent of the Universal Philosophical Refutation.
"So what?" the theist might say. Just these two words are an adequate refutation, but he might elaborate. "You don't approve of how I form my beliefs. So what? I don't approve of how you form your beliefs. I don't approve of your taste in ice cream either. Neither one of us are going to do anything at all about our disapproval; we will even stay friends so long as you don't discuss your dumb atheism with me (and I'll keep my marvelous theism to myself). So I have absolutely no incentive, no motivation whatsoever to do anything but ignore your preference and live my life as before."
Of course, these demarcation criteria are unimportant if ethical statements in general are truth-apt in the same sense that statements about objective physical reality are truth-apt. (We run into a slight problem of self-reference when it comes to the ethics of believing the truth, but the problem is relatively benign: that it's true that it's right to believe the truth is self-referentially coherent; that its true that it's right to believe falsity is self-referentially contradictory.)
But the problem is that it's really hard, apparently impossible, to consider ethical statements as truth-apt. It's a little like God: If God were to exist and the Christian Bible accurately represents His commands, then a lot of ethical truths would follow. But proving the antecedent is the crux of the biscuit: Lacking confidence in the belief of God and the Bible, everything that follows is just specious bullshit.
It's the same with unenforceable and unenforced ethical beliefs: If ethical beliefs really are truth-apt, then unenforced ethical beliefs are just as important as enforced beliefs. But again, the burden is to show the truth of the antecedent, lacking such proof, unenforced ethical beliefs are at best private preferences and at worst specious bullshit.