Originally Posted by brokenladder

Well, no. You as an individual will be statistically better off with a system with less Baysesian regret (assuming simply that you are any random individual).

It is possible that there is some particular political faction that would be worse off with the particular distribution of utility stemming from Score Voting. But as there's yet no strong evidence of that, it is rational for all voters to prefer it (until such time as they can demonstrate substantive evidence that it might disadvantage them relative to where they are now).

There is not "something other than Bayesian regret". There are "potential things you did not factor into your Bayesian regret calculation". If voters want to prevent party proliferation, then their utility values for candidates of the major, say two, parties would go up, and they'd have a reason to dislike candidates from additional parties. That's already accounted for in the existing calculations.

That is a valid point, though I doubt there is any method so good at electing centrists that it leaves centrists better off than Score Voting. It would take a specialized series of Bayesian regret calculations. I'm not so interested in that, however, because I already acknowledge there will potentally be a few demographics (perhaps large corporate interests) which will prefer plurality to Score Voting. My major concern is that the vast majority of rational voters will be better off with Score Voting, and that the goal of a benevolent "god" would be to maximize the sum of social utility. (Not "proven", but pretty darned close.)

I want to maximize the utility of white male land owners.

/kidding