I may have read your post wrong but I thought the centre channel comprised of stripping out the differences and only allowing through what's common to both channels.

Okay, I think that's what I meant.

If you summed the two channels, you'd get everything as if it were a mono signal (including things only on the left channel for instance), so that's not what you want.

If you subtracted the two channels, you'd get only the stuff that was unique to the outside channels, which is also not what you want.

I don't know the exact mathematical operation that would result in "only the center channel sounds appear in the center", but simply summing the two channels would suck. Does the DSP simply sum the two channels to get the center channel, or does it Do The Right Thing?