> The normalization error, what you are calling subjectivity, is already accounted for in VSE calculations.

I’d love to see an actual paper of the methodology of VSE (the FAQ is less complete than I would like, but helpful in that it shows lots of problems – for instance, while the brief description claims VSE tests elections with “voters who cluster on issues in a realistic way”, none of the descriptions of the different voter models mentions any tie to any empirical research on how voters actually cluster, instead it simply models three different, apparently chosen because of intuitive/aesthetic appeal, empirically ungrounded, abstract ideals); there are several dimensions of it which seems quite subjective/arbitrary rather than objective, making its conclusions also arbitrary, and, worse, it seems to simply ignore known effects like cultural differences in applying rating systems without concrete grounding (which effects both score-based and limited-ranks systems, but not particularly forced-preference or vote-for-one systems.)

> I’d love to see an actual paper of the methodology of VSE

See the references on https://en.wikipedia.org/wiki/Social_utility_efficiency