I still haven't gotten through the metaethics sequence yet, so I can't answer that exactly, but if he believed in an "objective" morality (i.e. some definition of "should" that is meaningful from the perspective of fundamental reality, not based on any facts about minds, or an internally-consistent set of universally compelling moral arguments), then he would probably expect a superintelligence to be smart enough (many times over) to discover it and follow it, and that is quite the opposite of his current position. If I recall correctly, that was his pre-2002 position, and he now considers it a huge mistake.
"Fundamental reality" doesn't have a perspective, so it seems weird to draw the lines there. Rather, there's a fact about what's prime, and the pebblesorters care about that, and there's a fact about what's right, and humans care about that. We can be mistaken about what's right, and we can have disagreements about what's right, and we can change our minds. And given time and progress, we will hopefully get closer to understanding what's right. And if the pebblesorters claim that they care about what's right rather than what's prime, they're factually incorrect.
Less Wrong is extremely intimidating to newcomers and as pointed out by Academian something that would help is a document in FAQ form intended for newcomers. Later we can decide how to best deliver that document to new Less Wrongers, but for now we can edit the existing (narrow) FAQ to make the site less scary and the standards more evident.
Go ahead and make bold edits to the FAQ wiki page or use this post to discuss possible FAQs and answers in agonizing detail.