[splonk]

There's a fair amount of research on this, although I don't have a link to any public results offhand. But I have sat in on several UX experience sessions with users being interviewed about their motivations for rating, and "fixing" a perceived bad average rating comes up repeatedly.

I'm fairly sure at one point years ago I did see some data showing that rating distribution changes depending on whether the rater sees the average rating first, but it was a long time ago and I don't recall the specific differences now. In practice you're right that this isn't feasible for most use cases.

I do know that the rating distribution is strongly bimodal - on a 5 star scale I think 80+% of ratings will be either 1 or 5 star. Mostly 5 stars - IIRC they were around half of all ratings.