The "Uber driver rating" problem of star-voting

Thinking about the poll offered earlier today, I was put in mind of a web serial author's complaint about a similar rating system (Found in their author notes at the bottom of the chapter). Pastafarian says:

As a side note, I wanted to mention something about RR's review/rating system. Some people aren't aware, but a story's Rating and Rank on RR have a bunch of effects on the story's growth, particularly in the long term. As a result, you can model a rating or review as being either an endorsement (5 stars) which makes the practical statement "this story should have more readers", or a range of magnitudes of the practical statement "this story should have fewer readers" (anything below 5 stars).

I describe this as being out of sync with the intended statement by many reviewers/raters, just as people who give sub-10/10 ratings to customer service representatives don't intend for their rating to be a condemnation (whereas the CSR's manager will generally consider anything below 9/10 on every metric to be a failure on the CSR's part). This is part of why I would prefer an up/down 2-point scale, which would be in alignment with the intent of the reviewer/rater.

And I was thinking about exactly this issue when I rated (https://civitai.com/models/118756?modelVersionId=135523) 4 stars. I said

Quite effective at responding to even esoteric prompts. My only concern is that some of the women rendered, absent other context, can have faces that are read as a little young.

I also have had quite a lot of trouble with multiple/malformed legs. This model has earned the distinction of making images good enough and correct-to-prompt enough that I want to fix them in img2img instead of just shrugging and moving on.

And I felt guilty. This model is interesting enough to post images of and to explore. Models that aren't interesting, I don't post. But because of the persistent issues of "legs" I was having, I wanted to indicate that it was worthy-of-attention but not quite perfect.

This is not how, in practice, the star rating system seems to work right now. Everything is 5 stars, which means that it's approval rating, where anything less than 5 stars is a punishment.

I hope these thoughts spur some further reflection on rating systems, and I'd be delighted to continue this discussion if people find it useful.