"Preferences" are also facts about minds.

Not directly, there is no fixed "preference mapping" from minds to preferences that works in general. We can only hope for one that works for humans, constructed for searching preference of humans, because it won't need to work for any almost-humans or not-humans-at-all. I look at a mind and see that its preference is X, you look at the same mind and say it's Y. There is no factual disagreement, the sense of "preference" was different; and if it was the same, the purpose was lost.

I interpret him to be saying something fairly non-dualistic - namely, that morality is not an ontologically basic thing separate from physics.

I did not like the particular way he was trying to make morality relate to physics. I thought it asserted a confused relationship between 'is' and 'ought'.

He also may be saying that moral claims reduce to fact claims in some sense, which is almost true (you need to throw some values in as well).

I think that was a point that he was at least trying to make and it is something I agree with.

Are you coming at this from the perspective of a moral nihilist?

No. That's for people who realise that God doesn't tell them what morality is and get all emo about it. I more take a 'subjectively objective' position (probably similar to what you expressed in the previous paragraph).