But the whole idea of attaching a one dimensional number to a preference seems flawed. How do you even compare movies, books, songs you enjoy? There’s a broad sense of liking or not something in general, but attaching a 4.1 stars to this and 4.3 stars to that seems delusional. And it’s not context-free either: start taking into account the time of day, state of mind, the phase of the moon or what have you, and you might as well throw the whole concept out.
>But the whole idea of attaching a one dimensional number to a preference seems flawed.
I think the idea of attaching a one dimensional number to almost anything in computer mediated communications is wrong.
I believe we should rate things with a vector, not a scalar.
Example: Your post Is 4.5 insightful, 0.2 snark, 5.0 truthful, 0.0 political
Re:The Platform Is the Enemy
by BlueTemplar on Sunday January 10, 2021 @11:47AM
(#25716522)
The law of large numbers deals with the last issue.
A much worse issue is that the aggregated rating is going to depend on the expectations that people had of the product. And bad marketing can screw that up, especially if it attracts the kind of people that are a bad fit for the product. And the smaller you are, the more the 'marketing' is out of your hands.
I think there's a distinction that's being glossed over, here: people can be desirous of eating at some restaurants more than others because of the results, but no longer pay attention to what those restaurants have in common -- what the underlying reasons are for the outcome at those restaurants that they prefer. Something which helps decide where to eat would ideally highly rate restaurants you'd enjoy, while not requiring you to learn specific details about the preparation of really excellent food.
I disagree with the assertion that letting a rating service choose the restaurant (or a list of restaurants of the highest rating) will result in not having desires about the experience at that restaurant, but I think "nuanced preferences" in the initial sentence must mean "knowing a lot about the kind of restaurants I like", instead.