I appreciate the straight talk. I had some questions about the above. Suppose there was a post on the front page that was popular, and "gratifies one's intellectual curiosity", but it happens to conflict with the interests of

A) HN,

B) YC, or

C) Specific individuals within YC, such as sama or pg

Can you guarantee that such stories will not be negatively moderated for that reason alone?

In the interests of transparency, can posts that are negatively moderated be publicly flagged, so that people can form their own opinions on whether the mods are using their powers in a fair and unbiased manner?

Most of us HN readers appreciate the work you're doing in moderating this forum. We're all in favor of mods using their powers in order to further the site's stated goals, but we're also very wary of mods using their powers in order to further their own personal agendas. We're worried that without sufficient transparency, it's only a matter of time before mods start abusing their powers. Hence why posts like this, and accusations of censorship, are such major topics of interest.

An example: the other day someone reposted an old (2011) controversy article about Airbnb. It was a me-too post of the kind users frequently submit in response to something currently on the front page, which tends to be bad for HN because it's repetitive. Also, the article was misleading in several ways (as I understand it) and long out of date in others. Our usual moderation practice is to penalize such stories—i.e. that are derivative of something else or clearly misleading—but we didn't do any of that because it was about a YC-funded startup. We're very clear about this with all YC founders and everyone who works for YC.

Stories about HN itself are a little different because of the evil catnip power of meta, but it's still the same principle: we don't suppress stuff because it's critical of us.

The idea that more data would clear everything up is a fallacy. "Just make everything public" is the sort of argument that sounds great on paper, but I think it would more likely cause just the opposite. People already take things out of context and construct fun, rage-inducing narratives about how awful we are—not because we're special, it's just a fact of life on the internet. Why would it ever stop? Of course it will never stop. What it would do is consume far more of our time and energy than it already does. That would steal resources from the things we should be doing to make HN better, basically giving a small number of users (some well-intentioned, others not so much) the power to do a denial-of-service attack on the good things of HN. Hacker News doesn't magically self-regulate, unfortunately; it takes a lot of work to keep it interesting and stave off entropy, and that's what we need to focus on. Every time one of these things comes up, it takes hours, sometimes a whole day, to answer people's questions about it. There's a burnout risk too.

So I think the balance we gradually arrived at over the years is the right one: be open about the principles we use to moderate HN and answer questions in specific cases. Unless something big happens to change my mind, that's how HN will continue to work; I believe it's what serves the community best as a whole.

HN is quite well moderated, IMO (and I have been a moderator, and I spend a LOT of my time rolling my eyes at how most mods are doing it wrong, damnit). Your comment is a kind of comment I see a lot on HN that sticks in my craw. I blogged about that very recently, so, it isn't you and I don't want to get into a pissing contest, but this is my cranky feeling about remarks of this ilk:

http://micheleincalifornia.blogspot.com/2017/03/hello-did-yo...

"Get a clue, people. Hacker News is an awesome discussion forum because it serves a business purpose. Please stop acting like there is something bizarre or ethically questionable about the mods making sure that this free service actually serves the business that pays all the bills for it. Geez"

I don't agree with the the assertion that HN mainly "serves a business purpose", perhaps in its early days its main purpose was to funnel prospective people into YC (second step from pg's essays, I stumbled on HN this way back in the day).

While you consider the arguments as "inane", I think there is merit in people trying to understand the mechanisms on which HN is being moderated. The "Don't worry your pretty head, it's being done right" argument only drives up more suspicion, because this is exactly the way less than savory practices in government and elsewhere are hidden. I don't think this is the case with HN but trying to argue that the suspicions arise not (just) because people are just jerks or stupid.

If HN serves no business purpose, why does it employ full-time paid moderators?

/sarcasm

There seems to be this assumption that good only occurs when some Christ-like martyr falls on their sword for the benefit of other people. The entire world seems to think that the only good men are dead men who died as martyrs to a cause and if someone is getting something out of the deal, they are evil predators.

I loathe this idea with every fiber of my being and would like to see it wholly stamped out of existence. I believe HN is as high quality a forum as it is because it has a purpose that serves the needs of the people who created it and run it.

I have done moderating for free. At the scale of HN, unpaid mods would absolutely not work. I wholly approve of them having a personal agenda here. I think the fact that they need to meet a standard of enlightened self interest is part of why they do so much better than most moderating teams.

Is that clearer?

Secondarily, HN has job ads and (more recently) 'Launch HNs' for startups that get special treatment. The job ads are written about in the FAQ and the launches eventually will be, but we're still figuring out the best way to do them. These are concessions HN grants to YC in return for being funded by YC, and our intent is to have them be both secondary and obvious. I agree that it's fair for HN to make these concessions and especially with your reason for saying so: it's the way to have a sustainably high-quality site, which does not come free. What HN doesn't do is make non-transparent concessions, like moderating anti-YC stories differently. But I've covered that elsewhere in the thread.

Any forum with a policy like that will never gain the credibility and respect that HN currently enjoys. If someone wants to create a private forum and moderate it in a way that furthers their personal agenda, as opposed to the community's interests, more power to them, but they should disclose it publicly. As community members, it is entirely reasonable if we ask them whether this is indeed the case.

Thankfully, Dang disagrees with you and isn't moderating HN in a personal-agenda-driven manner (see his post above).

Given "ourselves" is all of us here, a way to illustrate that in a trustworthy way would be to expose the meta data around the story. Which articles were flagged and removed, who flagged, how much it mattered, etc. If not in real time, due to possible exploitation for ranking tweaks, then perhaps in an acceptable offset of time.

I also recommend implementing a cost of suffering flag type for articles which use negative emotional responses to spread the information in the article, especially when it is dissonant and viral in nature. See https://www.youtube.com/watch?v=rE3j_RHkqJc for context.

This article seems reasonable, but here's one indication it's flirting with recursive irrational behavior:

> I have to admit that I found it a bit comforting that I wasn’t the only one who thought this all seemed a bit fishy.

Confirmation bias is still a bias. Making an argument with bias prevents exploring other explanations, such as errors in HN code, activity periods (lunch for example) and sharing of stories within other aggregates which elicit a negative response (a large company monitoring the comments may not agree with post, and consensus there effectively moderates it down).

Publicly pointing out who flagged a story seems like a bad idea. I think the data should always be anonymized. I just think it will lead down a bad path to ostracize people for specific votes or flags.

I do however think it would be interesting to see some data after the fact for some stories say ones that reach 500 points or greater.

Possibly better idea:

Moderators could take a look at unfairly flagged stories and silently adjust flag weights for flag-abusing users.

This is of course based on my unscientific thought that the majority of interesting (IMO) stories flagged of the frobt page are removed by competitors (political or business-wise) rather than moderators.

If, like you say, the moderators never saw the story then who made these changes? I find it hard to believe that foob would put the wrong year in the title of his own post.

Edit: ok, I think I know what happened now. A moderator put 2010 on the title by mistake. I haven't checked the timestamps on the user flags, but it's possible that people flagged the article because they thought it was old, in which case the title change would have acted as a moderation penalty even though obviously we didn't mean it that way. I was wrong to say "moderators never saw it" because I checked the log for moderation penalties, found nothing, and didn't look for title changes. So I made a mistake too.

We're sorry for the mistake. I don't recall this particular thing ever happening before, but plenty of other moderator errors have. We try to admit them and correct them as soon as we know about them.

Look at just the stuff from this case - heavy handed uneven constant moderation (you claim moderators dont see a bunch of stories, but those that you do see like this one, often get adjustments), misreporting, questionable transparency, too much hardly visible power of flags, etc.

I know the rules, but It really seems like at best we are a few moderator changes from having this place turn to something that wouldn't be valuable to people like me.

From the ones I frequent, HN via Dan and team is the only one I've come across which frequently and directly engages the community in a detailed and genuine way about their policies.

There's been times when an older article is posted and adding the year in the title might give better perspective. Or other times where it's clearly a PR post for a company and we just want it off the front page.

Uh... not really?

Left to its own devices, the system will include both extremes and non-extremes. In order to cut extremes out you have to sample what the extremes are, detect them and then minimize them.

The sampling comes by letting extremes pop up. The detection comes by letting users tag stories by content and emotional weight. You can minimize them with moderation tags and meta-moderation flags and automating balancing algorithms.

Currently you support story flagging, comment flagging, up/down voting. You have no way to identify what the significance of these are, though. Without better tools for your community maintain itself, of course you need moderators. But there are simple ways around this that would cut most of the moderation out and allow the system to simply re-weight stories as they become more obviously extreme in one or multiple tag/flag configurations.

Here are the categories of the 30 newest stories at time of writing: social media, privacy, programming, sysadmin, espionage, consumer products, politics, health, games, science. The stories vary in emotional rating (1-5, meh -> i am upset). As tags come in for any given story, trends emerge about how that story is perceived: it is relatively emotional, it is political and science based, and it is upvoted very quickly. Clearly, this is more likely to be controversal than the low-emotion programming topic that takes an hour to reach front page, so the latter should get higher priority, because it is less controversial and more "important to HN".

Notice that no moderator is needed to figure out what is controversial and change the balance, and we didn't even need to do something complicated like analyze speech patterns in comments or story content.

So what's to stop malicious users from tagging a post they don't like in such a way that the system sees it as controversial?

I assume that similar to HN meta stories resulting in upvoting trances, seeing [flagged] leads to ignoring the submission.

Interesting that you make a drug analogy. What's wrong with letting HN users have their crack? Unlike actual crack, there's no danger of side effects, dependency, withdrawal, etc. Maybe the "upvoting trance" is the expression of a sincere communal desire? It's not like it's "trashy" content... meta is simply self-reflection. Why penalize it?

Maybe the fear is that the front page will be overrun with meta content... that makes sense, but I feel like one or two meta posts couldn't hurt. I think self-consciousness is a positive attribute for a community.

> Unlike actual crack, there's no danger of side effects

There's a big danger of side effects to HN itself. When taken excessively—the natural tendency—it makes the site less intellectually interesting, which alienates the best kinds of user (and attracts the less best kinds). Maybe if we could be like David Cross and just do crack once each...

> one or two meta posts couldn't hurt. I think self-consciousness is a positive attribute for a community

This I agree with but HN is extremely well-resourced in this area. The intention behind penalizing such stories, as I tried to explain above, is not to suppress them but to countervail.

How about if there was a bit associated with every post that indicates whether it's meta or not. A submitter can set it when they post if they like; a mod can set or clear it later. Perhaps users could flag posts as being meta as well, i don't know.

Then we just take all the meta posts off the main page, and show them on another top-level page like Ask HN or Show HN.

I i think i'd be happy for meta posts to appear on the 'new' page. There aren't that many of them, so they won't flood it. If they get upvoted, they get promoted to a different place to substantive posts, so they won't clog it.

It's a bit like a GC which allocates large objects in the nursery, so you can still, eg, create a scratch buffer quickly, but promotes them to a large-object heap instead of the normal tenuring generations.

And besides, if a submission is flagged off the front page by that point I would think the damage is done short of resetting the gravity of the post to allow it to be in the same position on the front that it "should" have been. I'm not even sure if this is something they can do.

Talk to them... they're people.

In any case, someone other than HN (e.g. Algolia) could construct a page/queue of flagged posts for review. It would be used by a subset of the audience for the "new" page.

No permission needed from HN moderators. Users already have the power to vouch for stories, they only need a timely way to review.

Technically and opinion aside, given the weighting function, what is supposed to happen when someone vouches for an eight hour old thirty-five point story? Does it move to page the middle of page two? How is justice done?

To quote Dennis Moore blimey, this redistribution of wealth is trickier than I thought.

I don't think flagging is creating problems in terms of HN's quality. As I indicated, there were lots of reasonable criteria under which to flag the previous post. To me, this post has characteristics that meet those criteria as well.

Where vouching might have a role is a quality story with very few points that is flagged dead based on misunderstanding or malintent. A story that hits the front page has had many eyeballs upon it and if it is killed from there by organic means, that's the way the system is supposed to normally work. Even when I think the story does not deserve it.

2. Flags _soft_ downvote people unlock easiers

3. Mods have a ""super"" flag which is the ranking equivalent of aging a post pre-maturely. So the weighting algorithm calculates the post as being much older then it actually is allowing it to slide off the front page faster.

There's no reason why a few people clicking 'flag' should ruin the experience for everyone else.

It's true that the effect of flags isn't perfect in every way, but we have to evaluate this on a whole-system level. I'm not aware of any way to make the whole system work significantly better (otherwise we'd have done it already) and certainly getting rid of flags wouldn't be one.

There's no reason to have a 'flag' link once a story hits the main page, for example.

Anything is preferable, at this point, to the methods used. Perfectly good articles are getting dropped and it comes across as censorship.

Is the secret handshake that important to the success of HN? Why throw newbies for a loop or make them have to think about this?

Aka, I've noticed that most of my stories I submit never reach the front page, but a nearly identical, or sometimes the exact link will make it to the front-page. I can only speculate that there is some high karma aspect to the ranking algorithm that works against high karma HN users.

Kind of seems opposite of how it should be but maybe there is a good reason for this approach that the mods can share.

I take a little bit of offense to this statement. The article identifies stories that had sudden extreme drops in position and then speculates that this could be caused by a hidden flagging threshold and/or moderator intervention. I strongly suspected that moderation was responsible in some of the cases but I also emphasized that this was because I didn't expect those stories to get heavily flagged and that we couldn't know for sure. All of these stories were effectively removed from the front page whether it was due to a hidden threshold or a moderator. I thought it was interesting to look at which stories were removed even if we could only speculate on the mechanism in each case.

You've confirmed that there are hidden thresholds in place that severely penalize stories and that the moderators make manual adjustments. I don't think that that makes the post completely wrong even if the moderator adjustments tend to be smaller and all of the stories I looked at were affected by automated means.

Do you have any thoughts on whether any of the listed stories are appropriate for the front page? As I said before, a few of these stories seem like "bad" posts but most of them appear fairly typical to me. Flagging is obviously necessary but maybe there is an issue with the user-specific weighting or how one of these thresholds kicks in if there seems to be a high percentage of collateral damage? I'm sure that you've spent a lot more time thinking about this than I have so I'm genuinely very curious to hear your opinion.

Edit: I took a look at your list and don't see any pattern in those stories. Some belong on HN and indeed got a lot of time on the front page. Others clearly don't belong on HN. In a few cases moderators intervened to penalize the story or turn penalties off, but in a solid majority moderators didn't intervene at all. In a few cases a bunch of points were dropped by the anti-voting ring software so the 'real' score wasn't as high as it seemed.

To me all this shows, and the reason I described it as 'completely wrong', is that you can't reconstruct what's going on from timestamps, point totals, and story rank. There's simply too much data that isn't public. I know that's frustrating to the large section of HN's userbase who love to know what's going on and figure things out. But I don't believe making it all public would be helpful, so here's the balance we've arrived at over the years: we make public the the principles of how the system works, including moderator intervention, and we answer specific questions about specific cases.

Make it public?

The upside is that it is easier to see who is gaming the system. Or that could be another downside if the answer is "the moderators".

> Or that could be another downside if the answer is "the moderators".

To me it seems like an upside if we know when moderators are gaming the system.

But you're right, there are some forms of security that don't require obscurity. For voting systems though, I would categorize them as "cat and mouse" systems, which unfortunately fall into the obscurity category.

Someone finally has enough data to confront your obvious bias and enough people saw it that you had to back up and apologize but this is a rare exception.

Mods removed submissions and individual posts they find distasteful and it has nothing to do with HN guidelines, they are applied unequally just as some members get special treatment. hN is just not that honest of a place to have an open discussion on a variety of topics.

no administration leads to 4chan like boards and newsfeeds

/b/ is the best known and least moderated board of 4chan; only child pornography and brony stuff is removed.

Then maybe the system needs to be changed?

> None of this is new information, btw. I've posted about it plenty over the years and am always happy to answer questions.

Is there anything in particular that should make us trust the information you provide? You could be lying for all we know.

Without sounding too snide, it sounds like what's happening is that we do all of the work by posting articles and comments, and hacker news controls the narrative.