So, we're assuming here that there's no problem comparing A and B, which means these valuations are normalized relative to some individual scale. The problem, as you say, is with the scaling factor between individuals. So it seems I end up with something like (AX + BY + FCX + FDY), where F is the value of my partner's preferences relative to mine. Yes?

And as you say, there's an infinite number of Fs and my choice of action depends on which F I pick.

And we're rejecting the idea that F is simply the strength of my preference for my partner's satisfaction. If that were the case, there'd be no problem calculating a result... though of course no guarantee that my partner and I would calculate the same result. Yes?

If so, I agree that that coming up with a correct value for F sure does seem like an intractable, and quite likely incoherent, problem.

Going back to the original statement... "an ethical rationalist's goals in relationship-seeking should be to seek a relationship that creates maximal utility for both parties" seems to be saying F should approximate 1. Which is arbitrary, admittedly.