meta: add rating/ranking discussion file to workspace

This commit is contained in:
Xymorot 2020-08-06 23:32:00 +02:00
parent 2ab530e95c
commit ae30e504e6
1 changed files with 51 additions and 0 deletions

View File

@ -0,0 +1,51 @@
# Rating ★★★★☆ & Ranking 🥉🥇🥈
**Ratings are a categorization of works into groups of equal merit/worth.**\
The categories can be a binary _good_/_bad_, the classic x-out-of-5 stars, or up to a 0-to-100 rating.
**Ranking is an ordering of works based on their merit/worth.**\
A ranking is like a rating with each work being its own category and the value intervals between the categories don't have to be uniform.
## What's the point?
### Organizing the Library
- search filters
- automatic lists
### Make Public
- [p2p](./p2p-sharing.md) needs to be implemented
- recommendations based on other users' stuff
## Problem: The granularity is too low
For any granularity lower than the number of works to be rated there is a chance of conflict: "I think this one is better than the other, but not as good as to give it the next grade."
### Solutions
**Use unrestricted granularity (64-bit)**\
This also allows the user to defined an arbitrary amount of rating categories.
## Problem: Opinions can change over time
Ratings given years ago reflect the opinions of the user at that time. They may not be accurate any more.
### Solutions
**Motivate the user to always compare works.**\
\*insert good UX here\*
## Problem: Some works are not "fair" to compare
Different things are liked for different factors. Sometimes a one-dimensional comparison in form of a rating/ranking doesn't make sense.
### Solutions
**Make aspects of works rateable.**\
An aspect is a rateable part of a work. In other words: If you can say how good a work does something, that something is an aspect. This also is a way to avoid tags which are not objective, making them an aspect instead.
## Misc
**The under-educated statistician inside me wants the ratings to follow some distribution.**\
This would combine ranking and rating. Ranking with the combination of a static (but customizable) distribution results in a rating for each work.