Bramblethorn
Sleep-deprived
- Joined
- Feb 16, 2012
- Posts
- 16,758
So you just submitted a story, maybe a couple. It's getting good ratings but you know it should be doing better, and you've come up with a great new idea for how Literotica should be rating its stories, instead of the current "vote from one to five stars and average".
Sorry to break it to you, but your great new idea is probably worse than the existing system. Here's some likely reasons why...
1. Your system is basically "most views" in disguise. Since high ratings help stories find new readers, this becomes self-perpetuating: the stories with highest ratings get the most viewers which pushes their ratings even higher, and soon it becomes impossible for a new story to catch up. (Examples: "number of likes", "likes minus dislikes".)
2. Your system is very granular: there are only a few possible scores a story can get. As well as making it impossible to establish a ranking between the top stories - which might not be such a bad thing - this means that any individual vote has either no effect on the score, or a massive effect. (Example: median-based scores.)
3. Your system doesn't acknowledge the fact that even if 95% of readers loved a story, some people might legitimately dislike it, and their dislike should be able to affect the score too. (Example: trimming the top/bottom X% of votes.)
4. Your system significantly reduces the number of legitimate votes coming in. A lot of the story movements that people get het up about here are nothing more than random noise in your scores, and the fewer voters you have, the bigger that noise is going to be. If you scare away a large number of legitimate voters in order to prevent a small percentage of bad-faith voters, you're actually making scores less meaningful. Also, sweeps are probably catching most of those votes anyway. (Example: anything that depends on voters registering.)
5. Your system creates perverse incentives because you didn't stop to think "if I was an asshole who wanted to maximise my score, how would I abuse this?" (Example: scores based on "engagement" that encourage authors to troll readers or to post misleading clickbait blurbs). To be fair, the current system also suffers from this, since reader attrition gives long stories a big advantage, and the hotlist systems create some weird scenarios where a 1-vote can actually improve your standing because a bad vote is better than no vote.
6. Your system is basically "let's make 2 the new 1".
7. Your system is basically "what if it went up to 11?"
8. Your system is computationally expensive and doesn't scale to a site which has half a million stories, probably tens of millions of individual votes, and updates several times a day.
9. Your system doesn't have any way to translate the last 20 years of rankings data for all the stories already on this site, so we need to throw out all those votes and start over. (Actually, this one might not be such a bad thing, but it's a significant change and not to be embarked upon lightly...)
10. Your system is too complicated. Most voters and authors don't understand it and will get angry about it, even if it's otherwise perfect. If this is you, don't be sad, you have a promising future in cricket.
Probably some others that I missed, but 10 seems like a nice round number to start with.
Sorry to break it to you, but your great new idea is probably worse than the existing system. Here's some likely reasons why...
1. Your system is basically "most views" in disguise. Since high ratings help stories find new readers, this becomes self-perpetuating: the stories with highest ratings get the most viewers which pushes their ratings even higher, and soon it becomes impossible for a new story to catch up. (Examples: "number of likes", "likes minus dislikes".)
2. Your system is very granular: there are only a few possible scores a story can get. As well as making it impossible to establish a ranking between the top stories - which might not be such a bad thing - this means that any individual vote has either no effect on the score, or a massive effect. (Example: median-based scores.)
3. Your system doesn't acknowledge the fact that even if 95% of readers loved a story, some people might legitimately dislike it, and their dislike should be able to affect the score too. (Example: trimming the top/bottom X% of votes.)
4. Your system significantly reduces the number of legitimate votes coming in. A lot of the story movements that people get het up about here are nothing more than random noise in your scores, and the fewer voters you have, the bigger that noise is going to be. If you scare away a large number of legitimate voters in order to prevent a small percentage of bad-faith voters, you're actually making scores less meaningful. Also, sweeps are probably catching most of those votes anyway. (Example: anything that depends on voters registering.)
5. Your system creates perverse incentives because you didn't stop to think "if I was an asshole who wanted to maximise my score, how would I abuse this?" (Example: scores based on "engagement" that encourage authors to troll readers or to post misleading clickbait blurbs). To be fair, the current system also suffers from this, since reader attrition gives long stories a big advantage, and the hotlist systems create some weird scenarios where a 1-vote can actually improve your standing because a bad vote is better than no vote.
6. Your system is basically "let's make 2 the new 1".
7. Your system is basically "what if it went up to 11?"
8. Your system is computationally expensive and doesn't scale to a site which has half a million stories, probably tens of millions of individual votes, and updates several times a day.
9. Your system doesn't have any way to translate the last 20 years of rankings data for all the stories already on this site, so we need to throw out all those votes and start over. (Actually, this one might not be such a bad thing, but it's a significant change and not to be embarked upon lightly...)
10. Your system is too complicated. Most voters and authors don't understand it and will get angry about it, even if it's otherwise perfect. If this is you, don't be sad, you have a promising future in cricket.
Probably some others that I missed, but 10 seems like a nice round number to start with.
Last edited: