I think I figured out how to restate the intrinsicness principle that Andy and Jim are looking for. (See this post for background to what I’m talking about, and links to the original paper.)
They think the following principle is plausible, and it is.

If A and B are duplicates, and A and B have any phenomenal states, then they have the same phenomenal states.

This gets out of the worry about A or B not being maximal, because then they won’t have any phenomenal states at all.

A similar move can be used to define content internalism, I think. Narrow content is shared by duplicates who have any contentful states at all. Or is that too weak, because it allows for swampmen? Not sure.

What I’d like to be able to do is use this trick to find a general way of defining Ted’s * operator. As stated it’s defined by conceptual subtraction. An F* is something that has all the characteristics necessary to be F except (possibly) being maximal. This makes sense if F is factorisable into maximality and some other stuff. And the probability that this is true for all maximal predicats strikes me as being roughly 0. (+/- about 1.) So we need a more general definition of the * operator. When F* is meant to be intrinsic, then it’s easy – being an F* just means being a duplicate of some (actual or possible) F. But that won’t do for defining uncle*, or, if you are a gung-ho, let it all hang out context externalist, it won’t even do for defining rock*. Hopefully I can figure this one out one of my plane rides tomorrow.