5 ms·

This thread rose to the top group of the front page last night (you can see I posted here then, I happened to see it). Then it sunk quickly and disappeared. I w

This thread rose to the top group of the front page last night (you can see I posted here then, I happened to see it). Then it sunk quickly and disappeared. I was a little dismayed because the cynic in me was thinking along the lines of it being removed for being antithetical to YC company success. I went to bed. To my surprise, it was back up near the top this morning with almost a thousand votes and hundreds of comments. TripleByte may have chosen to burn their reputation irreparably, but I have gained a lot of faith in YC and the mods here.

dang 6y ago
It fell because of a software penalty called the flamewar detector. We review posts that get that penalty because there are often false positives. I saw it on the list last night and restored it (https://news.ycombinator.com/item?id=23280488 https://news.ycombinator.com/item?id=23280488). That was the only action any moderator took on the post. I'm glad I saw it quickly enough, because there would have been a nightmare of a flamewar about us 'suppressing' the post if we had missed this, when in reality it would just have been an accident of timing. That raises the obvious question of why we have such software if it causes such problems, but the answer is simply that it helps more than it hurts, overall.
- ganstyles 6y ago
  Yeah, thanks for the response. I saw you mention this downthread after you made the comment. Thanks for the work you do moderating this place.
- alexpetralia 6y ago
  Hi dang, sent you an email about this, but perhaps it would be useful to include a page on HN recording "recent moderator actions". This could make the process more transparent for users and help them understand your actions (rather than producing conspiracy theories every week).
  - dang 6y ago
    The question is whether that would raise more objections and protests than it would answer. Almost everything we do is defensible to the community, because if it weren't, we wouldn't do it in the first place. I say 'almost' because we make wrong guesses, but then we're happy to admit mistakes and fix them. That doesn't mean it's all self-explanatory, though. On the contrary, it can take a long time to explain because there are many complexities, tradeoffs, and non-obvious aspects. Meta threads and discussions tend to invite objections from the litigious type of user. Such users are rarely satisfied, but have a ton of energy for meta argument, so it's easy to get into a situation where any answer you give leads to two or three fresh objections. Such objections have to be answered with great care, because if you slip up and say the wrong thing, people will use it to drum up a scandal (edit: and will quote it against you for years to come!). This consumes a lot of mental and emotional energy. (Edit: btw, this is asymmetrical: the people raising objections and making accusations are under no such restriction. They can say anything without downside, no matter how false it is or what they accuse you of. They can make things up with impunity and people will believe them by default, because on the internet you are guilty until proven innocent, plus everyone loves the underdog. These are additional reasons why it's easy to end up in a situation where every comment you spend an hour painstakingly composing earns you a bunch more counterarguments and demands.) These arguments tend to be repetitive, so you find yourself having to say the same things and defend against the same attacks and false accusations over and over. This is discouraging, and there's a high risk of burnout. Disgruntled users are a tiny minority, but there are more than enough of them to overwhelm our limited resources. I fear this outcome, so we've always shied away from adding such a system. We do want to be transparent, and we answer whatever questions people ask, but it feels safer to do it ad hoc as questions come up. There's no specific question you can't get an answer to, other than a few special cases like how HN's anti-abuse software works. There's an opportunity cost issue too. The vast majority of the community is pretty happy with how we do things—I know that because if they weren't, we'd never hear the end of it, and then we'd say sorry and readjust until they were. I think it makes more sense to do things to keep the bulk of the community happy, or make them happier, than to pour potentially all our resources into placating a small minority—especially since, once you've done this job for a while (say, a week) you know that nothing you do will ever be completely right or please everyone. On the other hand, if I could ever be persuaded that a full moderation log would satisfy everyone's curiosity and reduce the overhead of misinterpretation, complaints, imagined malfeasance, etc., then we'd be happy to do it. This question has come up repeatedly, so if you're curious to read previous answers, see https://hn.algolia.com/?dateRange=all&page=0&prefix=true&query=by%3Adang%20meta%20litig&sort=byDate&type=comment https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que....
- cheez 6y ago
  Your salary should be at least 250K
- ravenide 6y ago
  Thank you for not incorrectly saying "that begs the question of..."
  - dang 6y ago
    Bullet dodged.

dang 6y ago

It fell because of a software penalty called the flamewar detector. We review posts that get that penalty because there are often false positives. I saw it on the list last night and restored it (https://news.ycombinator.com/item?id=23280488 https://news.ycombinator.com/item?id=23280488). That was the only action any moderator took on the post. I'm glad I saw it quickly enough, because there would have been a nightmare of a flamewar about us 'suppressing' the post if we had missed this, when in reality it would just have been an accident of timing. That raises the obvious question of why we have such software if it causes such problems, but the answer is simply that it helps more than it hurts, overall.

ganstyles 6y ago
Yeah, thanks for the response. I saw you mention this downthread after you made the comment. Thanks for the work you do moderating this place.
alexpetralia 6y ago
Hi dang, sent you an email about this, but perhaps it would be useful to include a page on HN recording "recent moderator actions". This could make the process more transparent for users and help them understand your actions (rather than producing conspiracy theories every week).
- dang 6y ago
  The question is whether that would raise more objections and protests than it would answer. Almost everything we do is defensible to the community, because if it weren't, we wouldn't do it in the first place. I say 'almost' because we make wrong guesses, but then we're happy to admit mistakes and fix them. That doesn't mean it's all self-explanatory, though. On the contrary, it can take a long time to explain because there are many complexities, tradeoffs, and non-obvious aspects. Meta threads and discussions tend to invite objections from the litigious type of user. Such users are rarely satisfied, but have a ton of energy for meta argument, so it's easy to get into a situation where any answer you give leads to two or three fresh objections. Such objections have to be answered with great care, because if you slip up and say the wrong thing, people will use it to drum up a scandal (edit: and will quote it against you for years to come!). This consumes a lot of mental and emotional energy. (Edit: btw, this is asymmetrical: the people raising objections and making accusations are under no such restriction. They can say anything without downside, no matter how false it is or what they accuse you of. They can make things up with impunity and people will believe them by default, because on the internet you are guilty until proven innocent, plus everyone loves the underdog. These are additional reasons why it's easy to end up in a situation where every comment you spend an hour painstakingly composing earns you a bunch more counterarguments and demands.) These arguments tend to be repetitive, so you find yourself having to say the same things and defend against the same attacks and false accusations over and over. This is discouraging, and there's a high risk of burnout. Disgruntled users are a tiny minority, but there are more than enough of them to overwhelm our limited resources. I fear this outcome, so we've always shied away from adding such a system. We do want to be transparent, and we answer whatever questions people ask, but it feels safer to do it ad hoc as questions come up. There's no specific question you can't get an answer to, other than a few special cases like how HN's anti-abuse software works. There's an opportunity cost issue too. The vast majority of the community is pretty happy with how we do things—I know that because if they weren't, we'd never hear the end of it, and then we'd say sorry and readjust until they were. I think it makes more sense to do things to keep the bulk of the community happy, or make them happier, than to pour potentially all our resources into placating a small minority—especially since, once you've done this job for a while (say, a week) you know that nothing you do will ever be completely right or please everyone. On the other hand, if I could ever be persuaded that a full moderation log would satisfy everyone's curiosity and reduce the overhead of misinterpretation, complaints, imagined malfeasance, etc., then we'd be happy to do it. This question has come up repeatedly, so if you're curious to read previous answers, see https://hn.algolia.com/?dateRange=all&page=0&prefix=true&query=by%3Adang%20meta%20litig&sort=byDate&type=comment https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que....
cheez 6y ago
Your salary should be at least 250K
ravenide 6y ago
Thank you for not incorrectly saying "that begs the question of..."
- dang 6y ago
  Bullet dodged.