The thing that annoys me about it is that it tries to sidestep the problem of how to get from matter to consciousness. The way Eliezer frames it is that other theories invoke magic to explain consciousness but this one doesn't. That makes it sound so good. But you do in fact need to get to consciousness somehow!
Put differently, how is it different from epiphenomenalism (= consciousness appears when atoms do specific things)? Eliezer has called epiphenomenalism "the most deranged idea in all of philosophy", so it better be different.
When in doubt, draw causal diagrams! I think it comes down to this: both agree that the following two things are important...
... but they disagree on how they are related. Epiphenomenalism says that the relationship is causal; do one and the other appears, i.e.,
whereas functoinalist reductionism says that they are inherently the same, i.e.,
but I'm still annoyed by framing this as less magical than panpsychism, which after all just says that
I don't think there is any sense in which functionalist reductionism happens "within physics" but panpsychism doesn't. If you get to say "consciousness is what an algorithm feels like from the inside", then I get to say "consciousness is what atom feels like from the inside". And then we can talk about how this fits the data, at which point
Moat