The hard problem is whether there's a difference between awake people with qualia and perfect sleepwalkers (aka "p-zombies") without qualia, and, if so, what causes that difference.
Is the answer even relevant ?
As far as I understand, there currently exists no "qualia-detector", and building one may be impossible in principle. Thus, in the absence of any ability to detect qualia, and given the way you'd set up your thought experiment about the sleepwalker, there's absolutely no way to tell a perfect sleepwalker from an awake person. As far as everyone -- including the potential sleepwalker -- is concerned, the two cases are completely functionally equivalent. Thus, it doesn't matter who has qualia and who doesn't, since these qualia do not affect anything that we can detect. They are kind of like souls or Saganesque teapots that way.
It certainly matters to the subject! I sure wouldn't want to lose my ability to experience regardless of whether others can ever notice (or whether it's possible).
Your objection here strikes me a little bit like behaviorism. Yes, there are valuable things to be gotten most of the time from such an approach, but behaviorism suffered from an unwillingness to acknowledge that people had thoughts. After all, thoughts didn't demonstrate themselves in behavior beyond being talked about, in which case it was the talk that was part of the scientific domain, not...
I encounter many intelligent people (not usually LWers, though) who say that despite our recent scientific advances, human consciousness remains a mystery and currently intractable to science. This is wrong. Empirically distinguishable theories of consciousness have been around for at least 15 years, and the data are beginning to favor some theories over others. For a recent example, see this August 2011 article from Lau & Rosenthal in Trends in Cognitive Sciences, one of my favorite journals. (Review articles, yay!)
Abstract: