Have they ever publicly discussed what their algorithm is? I'd be quite interested to hear from a place with reasonably high traffic how they go about it.
> Btw, none of this is new—it happens every time there's a major ongoing topic with divisive qualities. The principles we use are: (1) downweight the follow-ups so there isn't too much repetition; (2) upweight (not a word - I just mean turn off user flags and software penalties) the ones that have significant new information; and (3) downweight the hopeless flamewars, where the community is incapable of curious conversation and people are just bashing things they hate (or rather, bashing each other in the name of things they hate).
---
So its things like "the ratio of downvotes to post comments is looked at" to help detect flame wars. That then makes it down weighted and not show up on the front page as much.
Likewise, common things in titles (ChatGPT) gets down weighted so that they don't have a "here is a whole bunch of them that dominate the front page". If you browse https://news.ycombinator.com/newest much, you'll occasionally see lots of things on active topics.
But once it gets enough positive engagement in a post, it becomes up weighted.
