>Why can't we just say some things are "self-evident" and worth fighting for?
Because nothing is actually self-evident in the real world. It's all still made out of meat, with no convenient teletypes up to Plato's Heaven where the Self-Evident Propositions, Ideal Forms, and A Prioris live. "Self-evident and worth fighting for" isn't a factual claim, it's an effort to persuade people to believe (what you label) a delusion.
>And if this is true, then why are moral considerations such potent rhetorical devices? The only possible explanation I can think of is extraordinarily condescending.
Well, there are three possibilities here:
* Typical antirealism. Moral appeals are in fact appeals to ontological delusion. This sounds condescending, but it's the corner you back yourself into by trying to have your moral appeals and your anti-realism at the same time. You're the one saying, "let's appeal to morality, but only ever to my morality, left completely unaccountable to external facts beyond my intuitions." If you tried that in math, physics, sociology, psychology -- you'd be thrown out of the room.
* Moral appeals are used as verbal shortcuts to ultimately nonmoral things. For example, the moral stricture "do not harm the innocent" may ground out in "I feel empathetic pain when I see children get hurt". In this case, the moral appeal has a little bit more actual force, since it grounds out in a real motivation to which we can actually, well, appeal. In fact, if we can manage to talk about the rational content of that empathetic pain (something which antirealists vehemently disavow but which basically the entire rest of humanity for all of history upheld), we can achieve...
* Moral realism: moral appeals work because they appeal to real things that have motivational power. The empathetic pain is a real moral motivation, pain is bad, especially when the damage it accompanies is too permanent for it to be a learning experience. This is the typical view among basically everyone who hasn't been carefully trained to pooh-pooh our innate emotions and motivations as having no normative force, and remains the majority opinion among both professional ethicists and laypeople.
>First, the realists play this game all the time; each branch of moral realism basically corresponds to a different -- usually embarrassingly pseudo-scientific -- way of justifying a priori ethical concerns.
Well, I usually don't see the pseudo-scientific ones. I usually see the pseudo-mathematical ones. But oh well, my ethical concerns are a posteriori, just like my epistemological, uh, knowledge.
Mind, if I want to be dickishly precise, I'm a realist about value and interests at the organismic level, and merely observe that moral norms and imperatives at the social level often have deep anthropological reasons that ground out in the genuine interests of the individuals involved. Where they don't, those norms and imperatives tend to get changed over time.
That might be a form of constructivism at the social level, but constructivism isn't really a form of realism or antirealism (it roughly says, "morality comes from running this algorithm on these data, but you might need some nonmoral interest to get someone running the algorithm at all), and it's actually different from naive social constructionism ("people make up morality, agree tentatively on things, and then use it to persuade each-other about stuff").
>From my perspective, moral realism isn't an escape from the problem; it's just a way of ignoring the actually hard problem and playing games that are completely parametric in the actually hard problems.
Would you care to pose the actual hard problem, then?
>E.g., one plausible (and, per usual, embarrassingly pseudo-scientific) hypothesis bridging this divide is that morality is a social construction that necessarily arises from ethical intuitions that have a physiological basis.
That would still be a typical antirealist view, since it says that morality arises in a way that has nothing to do with moral facts as such.
(The nasty tricky part is what you're willing to count as a moral or normative fact in the first place. Fix that and the questions get a lot clearer. Make that slippery and you get hundreds of years of philosophizing instead of knowing what to do.)