[My promised concluding thoughts...]
Clearly our disagreements run too deep to do full justice to them in a mere blog post. But I at least hope I've succeeded in indicating where one might reasonably depart from Eliezer's reductionist line. There are also a couple of ad hominem points which struck me as noteworthy. (See my previous post for real arguments; this is mere commentary.)
One is that our beliefs are shaped in reaction to others. Intelligent non-philosophers typically only come across stupid, woolly-headed non-reductionists. The most prominent public intellectuals are typically scientists of a reductionist bent, like Dawkins, whose most prominent opposition is from anti-intellectual rubes and intellectually bankrupt religious apologists. From a purely sociological perspective, it's no surprise that intelligent people might initially be drawn to the former camp. (I know I was.) But that's no substitute for assessing the strongest arguments -- the ones you've probably never even come across unless you've spent a few years doing academic philosophy, or associate with others who have -- on their merits.
Since Eliezer's posts are mostly directed at a general audience - most of whom have not carefully reflected on their beliefs - I agree with 99% of his criticisms. Folks often commit the mind-projection fallacy, are fooled by an empty dispute that 'feels' substantive, and can be irrationally resistant to perfectly legitimate scientific reductions. These are all important insights (though hardly news to philosophers). But he overgeneralizes -- just as to a man with a hammer, everything looks like a nail.
I think the core problem here is methodological. Eliezer assumes that a debunking explanation of a belief is enough to refute it. Rather than doing the hard work of philosophy -- assessing the arguments for and against P -- he shifts to cognitive science, explaining why I might offer such arguments even if P is false. But this is to commit the genetic fallacy. Any reflective non-reductionist will grant that you can explain all the physical facts (incl. their brain states and vocalizations) without reference to any non-physical facts. Of course. But that doesn't imply anything about whether their belief is justified. Explanation and justification are two completely different things.
Reductionists make this error because they assume that all that stands in need of explanation is the third-personal data of science. Hence (they assume), if you can explain all the empirical data - including the vocalizations of your critics - then there's nothing left for said critics to base their arguments on. This type of genetic fallacy is no fallacy, on this view, because a full empirical explanation exhausts all possible justification.
But this is clearly question-begging, or worse. It assumes an indefensible scientism from the start. Non-reductionists take it as given that there is more than just third-personal empirical data that calls out for explanation. There is the manifest fact of first-personal conscious experience, and the normative facts about what we ought to believe and do, etc. A debunking explanation of why we believe in these phenomena is not sufficient unless one has also successfully debunked the phenomena itself. But that requires actually engaging with non-reductionists and the arguments we offer, rather than simply psychologizing us.
Reductionists, when short on real arguments, like to appeal to meta-arguments, e.g. induction on the historical successes of science. 'There have always been nay-sayers, who questioned the ability of science to explain phenomenon X, and every time they've been proven wrong!' It's a familiar sentiment. But it's also pretty weak. If you bother to look more closely, there are principled reasons to think these cases different. All those examples they point to are instances of third-personal empirical phenomena. I grant that science is supreme in that domain. But, to turn the tables, it's never had any success outside of it. So there's no general reason to think that normativity or first-personal subjective experience are susceptible to purely scientific explanation. So, again, these simplistic meta-arguments are no substitute for the real thing.