Let's move this out of the Top Writers Section thread since this really has nothing to do with it.
Usoki's idea for clipping people's votes who are outside the average isn't really the best way to go at all... and can be gamed easier than most other modes of dealing with vote gaming (which is the main reason to even have an algorithm in the first place).
Personally, I always liked places that have complex weighted algorithm to do theirs. The main thing is that they can't be an algorithm that they publicly tell anyone what it measure or how it works. The more details you give out about it, the easier it is for those who want to game it. I think the IMDB does a fairly good job with this. I like giving more weight to people who have been around longer, voted more often, voted a wider range of stories (both in quality and authors), and possibly other things (written stories, donated, or whatever you choose). How much weight each of these things is where you need to tweak things to get the right balance.
Yes this will tend to create an "elite" class of members, whose votes are worth more than others, but that's how these systems work. And anyone can make theirs worth more by being more involved in making the site better (reading and voting more over time). It's not going to be an uncrackable elite class, or a glass ceiling of any kind, just something that you have to actually be involved to become part of. You have to care. That way people who jump on, vote their friend's stories up, and leave, will never be worth as much as a vote as the person who reads lots of stories over the years and votes every one fairly. It will also weed out the people who only ever vote thing either 10 or 1, or always just vote stories they like 9 or 10, and make their ratings become slightly watered down.
But, the main thing that's really needed to make any good weighted algorithm work well is having a LOT of people voting... we don't really get those kinds of numbers on most of the stories. With the low numbers we get on voting, no method is really going to work well and be mostly free from gaming. If we go with a weighted system it's going to either greatly skew things because the low numbers are affected by highly weighted system, or it's not going to change anything because of a lightly weighted system. We just don't have the numbers to make it work really well.
Usoki's idea for clipping people's votes who are outside the average isn't really the best way to go at all... and can be gamed easier than most other modes of dealing with vote gaming (which is the main reason to even have an algorithm in the first place).
Personally, I always liked places that have complex weighted algorithm to do theirs. The main thing is that they can't be an algorithm that they publicly tell anyone what it measure or how it works. The more details you give out about it, the easier it is for those who want to game it. I think the IMDB does a fairly good job with this. I like giving more weight to people who have been around longer, voted more often, voted a wider range of stories (both in quality and authors), and possibly other things (written stories, donated, or whatever you choose). How much weight each of these things is where you need to tweak things to get the right balance.
Yes this will tend to create an "elite" class of members, whose votes are worth more than others, but that's how these systems work. And anyone can make theirs worth more by being more involved in making the site better (reading and voting more over time). It's not going to be an uncrackable elite class, or a glass ceiling of any kind, just something that you have to actually be involved to become part of. You have to care. That way people who jump on, vote their friend's stories up, and leave, will never be worth as much as a vote as the person who reads lots of stories over the years and votes every one fairly. It will also weed out the people who only ever vote thing either 10 or 1, or always just vote stories they like 9 or 10, and make their ratings become slightly watered down.
But, the main thing that's really needed to make any good weighted algorithm work well is having a LOT of people voting... we don't really get those kinds of numbers on most of the stories. With the low numbers we get on voting, no method is really going to work well and be mostly free from gaming. If we go with a weighted system it's going to either greatly skew things because the low numbers are affected by highly weighted system, or it's not going to change anything because of a lightly weighted system. We just don't have the numbers to make it work really well.
Comment