What would you think of W3, much bigger, sadder, and blander?
Anyway the point of the repugnant conclusion is that any world W1, no matter how ideal, has a corresponding W2.
This is a central argument of Phil Torres’ paper against space colonisation: there will be space wars!
Probably, we should include in the calculation not only averaged wellbeing of the individuals, which is “goodharted” measure of social wellbeing, but also the properties of the whole world, to which any individual may have access.
Why is average wellbeing a goodharted measure?
Offing those with low wellbeing increases average wellbeing.
That in itself can be solved (if you break the symmetry between killing / not allowing to live), but it still remains that a tiny super-happy population (of one person in the limit) is what’s aimed at.
It ignores many important aspect of human wellbeing:
1) preference to stay alive, even if it is unpleasant + some other preferences
2) time relation between observer-moments: e.g. short intense pain is not very important
3) non linear preferences about intensity of pleasure and pain. people could work a lot of time for short pleasure
4) social values (family), intellectual values (knowledge, diversity of experiences)
Declining marginal moral weight answers this with a built-in preference for diversity. More reference classes are good, more quantity isn’t worth very much trade-off in intensity.