3 ms·
I think you're talking about number of rating points and how the algorithm does not sort them by points. This is correct, the algorithm for HN for example, rel
by sova 9y ago
I think you're talking about number of rating points and how the algorithm does not sort them by points. This is correct, the algorithm for HN for example, relies on vote-ups, gravity "time for decay", and some other factors to calculate sequence of top listings. Total vote ups is just one factor whose strength decreases over time.
Addendum:
(= gravity* 1.8 timebase* 120 front-threshold* 1
nourl-factor* .4 lightweight-factor* .17 gag-factor* .1)
https://news.ycombinator.com/item?id=1781417 https://news.ycombinator.com/item?id=1781417
== Thread where Paul Graham chimed in
- gus_massa 9y agoI mostly agree. A simplified formula is: upvotes / time^2 * penalties where penalties are a lot of small and big penalties that are added to the story automatically or by the mods. The most relevant penalties that are not added by the mods are: * automatic penalties for some sites that usually have off topic stories * penalty for flamewars and controversial stories, when they have much more comments than upvotes * penalty for getting too many flags by users In this case, I guess the story got enough flags to get some penalization, but not enough flags to get killed. For a nice black box analysis off the penalties, you can read the article wrote by kens: http://www.righto.com/2013/11/how-hacker-news-ranking-really-works.html http://www.righto.com/2013/11/how-hacker-news-ranking-really... (original submission: https://news.ycombinator.com/item?id=6799854 https://news.ycombinator.com/item?id=6799854 (920 points, 1460 days ago, 190 comments)) (It was written 4 years ago, so many details may have changed, but I think that the general idea is still correct.)