But it doesn't just penalize flamewars, it penalizes all conversations. For example, I have now posted two comments in this discussion and only upvoted the article once. Does that mean this is a worse article than one that I only upvoted and never commented on or a better article than one I upvoted and commented on three times?
There is also the flame war penalty. If the ratio of comments and upvotes is indicating flaming, then the post gets some negative points for the purpose of ranking.
I've always thought that was a ridiculous metric, since it's entirely possible for a thread to have more comments than votes without being a flamewar, and for bad comments to be upvoted. Implicitly, it penalizes users who choose to comment rather than upvote (which is probably the majority,) which is penalizing engagement.
Maybe there's data backing the validity of it up though, I don't know.
> The first two you listed were downranked by the flamewar detector.
Just some feedback that I've found a number of articles fall off the FP due to the flamewar detector that I've felt were good articles/discussions. In fact, I think some of the more valuable discussions tend to have a lot of back and forth discussions relative to the votes.
But I also recognize that flamewars can also look a lot like that.
So I'm wondering if it may be worth revisiting the algorithm for this, and maybe having it factor in a few other things vs. simply the vote:comment ratio (which is what I'm understanding it currently is, but correct me if I'm wrong).
I don't think it necessarily needs to be a lot more complex, maybe simply add to it some standard deviation of upvotes/downvotes (or just a simple ratio), if that's not already part of it.
But I've seen some discussions fall off that I don't remember seeing a particularly toxic discussion happening (e.g. relatively little to no downvoted comments).
Again, happy to see flamewars fall off, but just hoping to see some more interesting/helpful discussions not get caught in the crossfire.
Items with active discussions are often implicitly penalized, because the number of comments are often greater than the number of upvotes, which is interpreted as a sign of low quality discussion or possible flamewar.
>A post with a bad comment to up vote ratio usually means that a flame war is going on.
No it doesn't, not unless most comments typicaly get upvoted, which seems counterintuitive to me.
A bad comment to downvote ratio indicates a flamewar, since there are no flamewars without downvotes, but more comments than upvotes just means high comment velocity (which can go either way) or just that no one is saying anything particularly interesting, which isn't implicitly harmful.
A flamewar detector that hides popular threads to suppress engagement just in case there might be a flamewar is working at cross purposes with the goal of a forum, which is engagement.
I'm seeing this often enough to think it's worth telling people about.
HN has a an automated simple proxy for flame-war detection. One characteristic of flame-wars is that people get into a to'n'fro over the issue, and the number of comments balloons. However, no one else is really interested, so they don't upvote the submisson, and the participants can only upvote a submission once, so the number of points doesn't increase.
Result is that a simply proxy for a flame-war is the number of comments on a submission out-stripping the number of votes.
A side-effect of that is that if there's a mildly interesting submission that lots of people comment on, but very few upvote, then the flame-war penalty will be triggered, and the submission will sink like a stone, never to be seen again.
It's a reasonably interesting collection of suggestions and comments from a fairly wide range of the HN community, but it's now lost, because while people commented, they didn't upvote.
So this is a public service announcement ... if something is interesting enough to comment on, consider upvoting as well, so others can see your comment, and you can then see theirs.
As others have said, there is a heuristic flame war detector. It seems that a penalty is applied if an item has more than 40 comments, and fewer points than comments. It's actually a good idea, but it has many false positives.
That said, if people want to comment on something but don't think it's worth upvoting, maybe it really doesn't belong on the front page.
> Anything with 40 comments and fewer points than comments gets such a penalty, and while it's a good proxy for flame-war detection
Is this really true? There's a strong implication there that the typical HN account is more likely to upvote an article than to comment on one. I upvote comments all the time, and comment frequently, but I almost never upvote articles (though I did find this one to be a great mix of interestingness with accessibility).
Personally I think "generate healthy discussion" should be the goal of all HN threads, which is why I think comment-count penalties are unproductive. How do you tell the difference between "vigorous debate" and a "flame war"? If the algorithm utilized NLP to analyze the tenor of the discussion, that would be one thing, but upvotes vs. comment count is a troubling metric. I want to read threads where people are passionately discussing a topic with long back-and-forth debates. If that's a "flame war" then I'd like more flame wars, please.
Not everyone knows about this, so I thought I'd write a short note[0].
HN has a "Flame War" detector. When it detects a flame war it applies a penalty to the item in question, causing it to drop in the rankings.
The definition of "Flame War" is an interesting one:
* when an item gets to more than 40 comments,
* ... if there are more comments than votes,
* ... ... then it's likely to be a Flame War.
The idea is that HN wants things to be "Intellectually Interesting". TPTB seem to think that if something is interesting then it will get upvoted, and that seems reasonable.
OTOH, if there is a flame war then it's likely to be the same small group of people arguing back and forth. So the number of comments rises, but they can't repeatedly upvote, so the number of votes peaks early. Further, if it genuinely is a flame war then it's unlikely to get upvotes "from outside", and so there we have the conditions.
Occasionally it gets a false positive, so I saw somewhere that whenever the detector gets triggered, the item in question is reviewed by the mods. (Let me add here that the mods make decisions, and while some people will sometimes disagree, I think on the whole they do an amazing job. To them: Thank you.)
I suspect that this is a case in point: http://hnrankings.info/25870571/
I think it's a reasonable idea, the proxy of "more comments than points" seems defensible, and having the backstop of the mods checking things is reassuring. It's always disappointing when what I think is a reasonable discussion suddenly sinks without trace, but I believe HN is the better for it. It would be interesting to know how often it's triggered, and how many are then reversed.
reply