Live data from Hacker News

Ask HN: Are HN comments automatically downvoted by algorithms?

news.ycombinator.com

21–28 of 28 posts

Re: Ask HN: Are HN comments automatically downvoted by algorithms?

#21
post #19
post #18

Earlier quoted context omitted.

This comment is shown as fine to me, whereas it really is dead: https://news.ycombinator.com/item?id=10245424 I'd love to hear a non-malicious explanation for that.

That comment was killed by a spam filter. Those generate false positives sometimes. We've fixed that and also unkilled the comment. In the future, please do what the HN guidelines ask and email such questions to us at hn@ycombinator.com. I only saw this thread by accident.

I'm sure you can see that the post linking to the (simplifying) anti-YC blog being accidentally killed by spam filter can raise questions about whether it was really an accident. I'm not saying it was not - I'm just saying the bayesian probability of this being an accident doesn't look that great to me.

Re: Ask HN: Are HN comments automatically downvoted by algorithms?

#22
post #6

There are a lot of people on HN, and a lot of people with downvote privilege, and there are a lot of people who just generally downvote because they disagree with something. Even if your opinion is completely positive, if people disagree, and a lot of people see it, you will get downvoted just because they disagree (and no one will reply to your comment to disagree, it's a drive-by-downvote). There's also verificatio…

I find that really annoying when people downvote a comment because they don't agree with it. IMO, the downvoting privilege should be used only to indicate a flawed argumentation or negative content.

Re: Ask HN: Are HN comments automatically downvoted by algorithms?

#23
post #20
post #17

Earlier quoted context omitted.

HN not only shadow bans commenters, they will also poison placement of comments (algorythmically) in a sub thread based on who writes the comment.

We nearly always inform people when we ban their accounts: https://hn.algolia.com/?sort=byDate&prefix&page=0&dateRange=... . Lots of software affects both story and comment rankings. "Poison" is a rather tendentious way to describe that. HN's goal is high quality in all things, to the extent it's possible. That doesn't just happen by default; systems are necessary, and it gets harder as numbers grow.

"Poison" is a rather tendentious way to describe that

NB...The words in my signature didn't come from a random sampling of the dictionary...they were from a screen grab of HN moderation tools...

On the topic of changing subjects, no not really.

Those words are outrageously inflamatory.

Re: Ask HN: Are HN comments automatically downvoted by algorithms?

#24
post #6

There are a lot of people on HN, and a lot of people with downvote privilege, and there are a lot of people who just generally downvote because they disagree with something. Even if your opinion is completely positive, if people disagree, and a lot of people see it, you will get downvoted just because they disagree (and no one will reply to your comment to disagree, it's a drive-by-downvote). There's also verificatio…

Perhaps some random percentage of down-votes should require a response (every third, or something).

Re: Ask HN: Are HN comments automatically downvoted by algorithms?

#25
post #23
post #20

Earlier quoted context omitted.

We nearly always inform people when we ban their accounts: https://hn.algolia.com/?sort=byDate&prefix&page=0&dateRange=... . Lots of software affects both story and comment rankings. "Poison" is a rather tendentious way to describe that. HN's goal is high quality in all things, to the extent it's possible. That doesn't just happen by default; systems are necessary, and it gets harder as numbers grow.

"Poison" is a rather tendentious way to describe that NB...The words in my signature didn't come from a random sampling of the dictionary...they were from a screen grab of HN moderation tools... On the topic of changing subjects, no not really. Those words are outrageously inflamatory.

Nobody found them inflammatory when they sat in the public HN source code for years, but once they appeared in an accidental screenshot, they became outrageous!

We're happy to answer good-faith questions about HN moderation, but this seems like trying to stir up drama.

Re: Ask HN: Are HN comments automatically downvoted by algorithms?

#26
post #25
post #23

Earlier quoted context omitted.

"Poison" is a rather tendentious way to describe that NB...The words in my signature didn't come from a random sampling of the dictionary...they were from a screen grab of HN moderation tools... On the topic of changing subjects, no not really. Those words are outrageously inflamatory.

Nobody found them inflammatory when they sat in the public HN source code for years, but once they appeared in an accidental screenshot, they became outrageous! We're happy to answer good-faith questions about HN moderation, but this seems like trying to stir up drama.

The verbiage of 'poisoning the well' is a well understood figure of speech.

Re: Ask HN: Are HN comments automatically downvoted by algorithms?

#27
post #4

Personally, I downvote destructive comments and flag harmful ones. My impression is that others do the same. I've never seen anything to suggest that there is algorithm beyond those that affect community members under shadow bans. My understanding is that people wind up in that state by hand, but I have no direct knowledge.

Thank you for this. I hadn't realized there is a difference between downvoting and flagging comments until I read your post. I hope other HN users know of this difference.

Re: Ask HN: Are HN comments automatically downvoted by algorithms?

#28
post #21
post #19

Earlier quoted context omitted.

That comment was killed by a spam filter. Those generate false positives sometimes. We've fixed that and also unkilled the comment. In the future, please do what the HN guidelines ask and email such questions to us at hn@ycombinator.com. I only saw this thread by accident.

I'm sure you can see that the post linking to the (simplifying) anti-YC blog being accidentally killed by spam filter can raise questions about whether it was really an accident. I'm not saying it was not - I'm just saying the bayesian probability of this being an accident doesn't look that great to me.

Imagine making a scoring system for domains that that takes into account flags and votes on comments that include said domain. Also, your post had a word written in ALL CAPS, arguably two. You can see other comments linking to the same domain at around the same time [1] that aren't marked as spam.

[1] https://hn.algolia.com/

Post reply on HN