I hope everybody is having a great end of the year! I just wanted to remind everybody that voting is closing at ~05:00 UTC on December 30th for the Anime Awards. In case you missed it, there are also other mediums you can vote for:
It does. There was a bit of a discussion about this in the voting announcement thread here. Basically, I wanted to do ranked choice voting, which isn't as punishing to shows that don't get a vote, but that type of question didn't work in some browsers, so we fell back to the STAR voting. It should still result in more accurate results than a single, non-transferable vote though. It's something I want to fix for the next time we do this (if we do).
If you implemented the voting algorithm (STAR in this case) in the code instead of just using a library, consider implementing this more generalized variant of STAR; not sure if there is a name for it:
Instead of scoring candidates from 0 to 5, allow voters to give scores between m and n inclusive, where m and n are integers. If the voter choose not to give a score to a candidate, assign a default score of k where k is an integer and m <= k <= n
For example,
m = -2 # minimum score
k = 0 # default score
n = 3 # maximum score
What I liked about scoring methods - that I'm unsure if ranked-choice voting methods can achieve - is how expressive a voter can be.
For example, if I like anime A way more than anime B and hence want to give A a much higher chance of winning, I can express it in my scores (e.g. by giving A a score of 3 and B a score of 0)
Another case is if I absolute detest an anime C, I would be able to give it a score of -2.
Note that allowing negative scores likely come with its own problems, especially with strategic voting.
Also note that the typical scoring method - that's being used for Lemmy ani.social awards right now - simply uses the following parameters:
m = 0 # minimum score
k = 0 # default score
n = 5 # maximum score
I don't think other commenters' complaints are really about STAR itself, but its parameters of m, n and k during the scoring phase.
If these parameters can be properly tuned such that unvoted candidates get a middle score instead of the lowest score, that should solve the problem where these unvoted candidates appear to be voted against.
P.S. I'm still of the opinion that these awards are just another form of popularity contest, so if a candidate hardly get voted for because it was barely watched by most of the voters, that means that majority of the voters won't really want it to win anyway.