This, to me, sounds like a circular argument at worst, and a circular analogy (if there is such a thing) at best.

Mmm. Yes, I think you're right. As I've chewed on this, I've come to wonder if that's part of where I've been getting the impression that there's a hard problem in the first place. As I've tried to reduce the question enough to notice where reduction seems to fail or at least get a bit lost, my confusion confuses me. I don't know if that's progress, but at least it's different!

I guess the categorical difference is that when asking about blood flow, there's someone who experiences the question and the data and the subsequent answer; but when asking about consciousness, it's the very process of being able to understand the question in the first place that we're asking about.

I think that, if you go this route, you arrive at a kind of solipsism.

I'm afraid I'm a bit slow on the uptake here. Why does this require solipsism? I agree that you can go there with a discussion of consciousness, but I'm not sure how it's necessarily tied into the fact that consciousness is how you know there's a question in the first place. Could you explain that a bit more?

Anyway, let's assume that your main criterion for judging whether anyone else besides yourself is conscious is their behavior (if that's not the case, I can offer some arguments for why it should be), and that you reject the solipsistic proposition that you are the only conscious being around (ditto).

Well... Yes, I think I agree in spirit. The term "behavior" is a bit fuzzy in an important way, because a lot of the impression I have that others are conscious comes from a perception that, as far as I can tell, is every bit as basic as my ability to identify a chair by sight. I don't see a crying person and consciously deduce sadness; the sadness seems self-evident to me. Similarly, I sometimes just get a "feel" for what someone's emotional state is without really being able to pinpoint why I get that impression. But as long as we're talking about a generalized sense of "behavior" that includes cues that go unnoticed by the conscious mind, then sure!

In this case, a perfect sleepwalker or a qualia-less computer that perfectly simulates having qualia, etc., is actually less parsimonious than the alternative, and therefore the concept of qualia buys you nothing

It's not a matter of what qualia buy you. The oddity is that they're there at all, in anything. I think you're pointing out that it'd be very odd to have a quale-free but otherwise perfect simulation of a human mind. I agree, that would be odd. But what's even more odd is that even though we can be extremely confident that there's some mechanism that goes from firing neurons to qualia, we have no clue what it could be. Not just that we don't yet know what it is, but as far as I know we don't know what could possibly play the role of such a mechanism.

It's almost as though we're in the position of early 19th century natural philosophers who are trying to make sense of magnetism: "Surely, objects can't act at a distance without a medium, so there must be some kind of stuff going on between the magnets to pull them toward one another." Sure, that's close enough, but if you focus on building more and more powerful microscopes to try to find that medium, you'll be SOL. The problem in this context is that there are some hidden assumptions that are being brought to bear on the question of what magnetism is that keep us from asking the right questions.

Mind you, I don't know if understanding consciousness will actually turn out to yield that much of a shift in our understanding of the human mind. But it does seem to be slippery in much the same way that magnetism from a billiard-balls-colliding perspective was, as I understand it. I suspect in the end consciousness will turn out to be no more mysterious than magnetism, and we'll be quite capable of building conscious machines someday.

In case this adds some clarity: My personal best proto-guess is that consciousness is a fuzzy term that applies to both (a) the coordination of various parts of the mind, including sensory input and our sense of social relationships; and (b) the internal narrative that accompanies (a). If this fuzzily stated guess is in the right ballpark, then the reason consciousness seems like such a hard problem is that we can't ever pin down a part of the brain that is the "seat of consciousness", nor can we ever say exactly when a signal from the optic nerve turns into vision. Similarly, we can't just "remove consciousness", although we can remove parts of it (e.g., cutting out the narrator or messing with the coordination, as in meditation or alcohol).

I wouldn't be at all surprised if this guess were totally bollocks. But hopefully that gives you some idea of what I'm guessing the end result of solving the consciousness riddle might look like.