Child of a Metal God

Can you identify behaviours or other externally-measurable/-visible things that fit your thesis? What about things that don't? Can you control for self-interest at all, i.e. whether society is getting better or worse for various small groups, various large groups, most people, and you? It seems like over-fitting is the biggest pitfall here, so using a methodology which works against that, or at least which makes you very aware that you probably are over-fitting, is probably just about the best you can do. Ideally, the Veil o' Ignorance™ can force you to consider factors which have nothing to do with your own interest, and actually quite apart from your own interest, but I would guess you're either already used to doing that, or perhaps have tried to align your own interests with universal good as you perceive it to a degree where that doesn't add anything.

Can you identify people for whom society is getting better? Can you identify ways in which it is getting better for most or all people? I feel like I can, in both cases, and so it devolves into weighing various goods, which I am not very good at, since I don't find ends-oriented ethical reasoning very resonant or conclusive. (And my tendency towards virtue ethics means, at societal scale, I tend to be one of those jerks who places value on principles and morality and shit, which I suspect don't add much to any reasonable attempt to ascertain the trend line of society's overall goodness.)