Hacker News new | ask | show | jobs
by NAHWheatCracker 479 days ago
I think most n/x rating systems are extremely weak and the level of detail is irrelevant. Averaging leads to a soup where nearly everything that isn't objectively awful falls in 60-80% rating. There's no signal to determine if it's good or just mediocre based on the rating alone. Goodreads and IMDB have been utterly useless to me.

Rotten Tomatoes' system has a lot of positives and I see a ton of signal in Rotten Tomatoes scores. It only works with a critical mass of people rating a given piece and I don't think anyone could get that critical mass for books aside from Goodreads.

I had an idea of a rating system where people would have to create a ranked list of the movies/books. Their rating for a given piece would be based on where it is in their list, linearly. Then ratings would be relative to other content. I think this would be much harder to get off the ground than a Rotten Tomatoes system.

1 comments

Criticker does relative rankings, percentile-matched, in kinda the same way. For films (not books).

So if I rate Feature Film I II & III as 30 50 70, and you rate them 70 80 90, we would basically "agree". It's a neat system that I wish other rating systems would use.

[1] https://www.criticker.com/

Very interesting to rate movies and see what it predicts. Sometimes it's fairly spot on and other times it's way off. Very mathematical. Thanks for sharing.