> But is that empirical evidence talking about peoples' overall position, or their position with respect to one particular exchange?
It's about overall position, but the outcome of every particular exchange is a change to overall position; now, if the asymmetry in exchanges were essentially random, this wouldn't matter, but when you have a system with distinguishable classes, and there is a clear relationship between class role and which side of the asymmetry of exchange you tend to fall on across exchanges—say “capital” having the greater gain and “labor” the lesser—then you can end up with a situation where the pattern of exchange, each seemingly mutually beneficial but asymmetrically so, ends up producing more disutility from increasing inequality on one side than utility from absolute gain.
It's about overall position, but the outcome of every particular exchange is a change to overall position; now, if the asymmetry in exchanges were essentially random, this wouldn't matter, but when you have a system with distinguishable classes, and there is a clear relationship between class role and which side of the asymmetry of exchange you tend to fall on across exchanges—say “capital” having the greater gain and “labor” the lesser—then you can end up with a situation where the pattern of exchange, each seemingly mutually beneficial but asymmetrically so, ends up producing more disutility from increasing inequality on one side than utility from absolute gain.