Basically, they use the entire dynamic range. 3 stars for "as expected" is eminently sensible, but the west and especially the US suffers so much from review inflation that you could really remove 2 if not 3 stars of a 5-star scale: either a reddit/hn-style up/down, or possibly have a middle-ground "at expectations".
Yeah, a lot of review inflation in the US. I see loads of 4-star reviews on Goodreads as well as the restaurant review apps. I wish people would use the entire range as it makes the review more valuable.
But even if you encourage that it's still pointless, because you don't know how people have used it, or with what distribution.
This annoys me with ratings used to train recommendations too, e.g. Netflix, where I don't know how they want me to use it. What if I know it is something I would absolutely not even try watching, should I rate it? Or am I telling it I tried it, and that's a higher rating (because I might give something similar a go too) than none at all? Not to mention the old stars not being mapped into thumbs, or the later addition of double thumbs up meaning only more recently encountered favourites get that accolade.
Yes this is practical in theory, but it's bad if their rating is mixed to rest of the world. Worldwide products like games review gets fewer stars by Japanese compared to others, some people think it's a problem.
I remember when I first started using rideshare, the drivers instructed people to leave 4 star ratings if bad, 5 star if good. That’s why everyone has a 4+ star ratingz
