Hacker News Reader

Don't worry, the moderator is very sensitive to this issue. There are going to be fairly explicit standards about what should be endorsed and what shouldn't. Roughly things shouldn't be endorsed if they are gratuitously uncivil or content-free, but there will presumably be a fully detailed version of that policy. And there will probably also be a version of showdead for pending comments. (Already anyone who can endorse can see them at /pending.)

We're hoping that this feature will make the site more hospitable to women. That wasn't the only reason for doing it, because users get stupid and/or nasty about lots of topics (or any topic, if they happen to start insulting one another), but certainly some of the most cringe-worthy threads we've cringed over have involved female programmers. So it would be hard to imagine a version of "working" for which pending comments working would not thereby make HN more welcoming to women.

I'm glad this is being considered, but I'm worried that it's easy for non-majority opinions to be interpreted as "gratuitously uncivil", especially on emotion-generating topics.

For example, if a person has had a bad experience being harassed by a coworker and writes a clear and honest comment about this as a problem, that can be easily interpreted (such as by a person unfamiliar with getting harassed) as an uncivil thing to say about your coworker. There is a known pattern of "tone policing" - that people have a pattern of asking less-privileged people to be "not so angry-sounding" when talking about something problematic or upsetting from their lives. I think of this pattern as a defense mechanism to protect yourself from understanding somebody else's pain (especially if it challenges something about your status), and it seems likely to be a bias here, even an unconscious bias.

Instead, I'd love to see explicit standards for comments on a visible community guidelines page, and an efficient flagging system that reinforces these guidelines.

The sort of incivility we're worried about is the more explicit type where someone replies to a comment with "You are an idiot. Don't you realize that x y z?" when they could have simply said "x y z." I'm pretty sure if HN cuts down on that type of comment, people who are earnest but upset about something will be net ahead of people they might be arguing with who are merely thoughtless jerks.

So, a lot of people's first introduction to all of this was the comment Sam made in his article "What I've Learned From Female Founders So Far" about how "we're working on something to improve the quality of Hacker News comments". It is thereby not surprising that people are trying to analyze whether this feature will help that specific issue ;P.

FWIW, I think that this is a different problem than the "you are an idiot" problem: I think a lot of the threads that people are finding issue with when it comes to attitudes towards under-represented groups are not (at least entirely) filled with this "explicit type" of negative comment: it is instead the more insidious, implicit type being stared at.

Regardless, I don't see why the endorsement system will somehow work against either of these problems when the voting mechanism hasn't: the comments that start with "you are an idiot, don't you realize that" are currently getting upvotes, so why would we presume they won't also get endorsements? Are high-karma users voting differently?

I would love to see that kind of comment cut out too, but I am concerned that pending comments won't actually achieve that goal - that biases (unconscious and not) will sneak into the patterns of what gets approved.

The moderator is concerned about that too. But if it did start to happen it would be pretty obvious. So it seems worth trying to see if pending comments can be tuned to cut obvious crap without eliminating stuff that's merely controversial.

That sort of thing needs a flag mechanism, not an a priori vetting of each comment. Enough flags = lock replies, a few more flags = hide comment.

Exactly! And really, we already have a flagging system: the downvote. My understanding is that comments on HN should only really be downvoted for the same reasons as pending comments would not be approved. So why not improve on that existing system? When a comment is downvoted, it should ask the reason for the 'flag', with choices representing the various commenting rules. This should help ensure that people are using downvotes correctly, which should allow them to be acted on more strongly. Sufficient downvotes (or sufficient downvotes with the same reason) could hide the comment entirely.

Mmm. This smells a bit like the League of Legends flagging and Tribunal system, which has worked very well for them. I wonder if HN has enough active readers to make a Tribunal work well.

I've decided to start using a throwaway for political discussions...

I would personally approve any comment that discusses a lived experience in concrete terms.

That being said re: tone policing, there is a segment of the social justice crowd on HN that will make inflammatory generalizations about large groups of people. Generally this will take the form of "X happens because of goddamn white cishet men." Inevitably someone will call these comments out and the response will be that it is justified to speak this way about the "oppressor class" and that anyone who "tone polices" these comments is ignoring the "power dynamics" of the situation.

In my opinion these comments do nothing but add fuel to the fire and I will never approve them. They are wrong for the same reasons gross generalizations about minorities are wrong.

And no, I'm not white, but yes, I am a man.

My take on tone policing is it is a mechanism (whether conscious or unconscious) to tell people they shouldn't be as passionate and concerned about something as they are.

As such, it's relative to the tone of the forum (be it HN, tumblr, irl, wherever). If it's acceptable for me to say "goddamn NSA fucking ruining the internet" then it's acceptable for me to say "goddamn cis people who fucking ruin the internet" [1]

There are plenty of people on HN who speak passionately about many things, and who do so with expletives and curses, blunt and punchy statements. Personally I hope they can continue, but either way I hope it is explicitly allowed or discouraged in the guidelines.

Speaking to over-generalisations; it's a hard one. There is undoubtedly a need to talk about the commulative effect of oppressive groups, and there is definitely a need to talk a about subsections of oppressive groups who are actively doing opressive things (again, whether consciously or unconsciously, knowingly or unknowingly).

Unfortunately we haven't yet figured out/settled on the language to use that differentiates between the two that doesn't involve a whole bunch of clauses and caveats. And given a forum where people are permitted to express their passion, extra clauses and caveats weakens the impact. When you are trying to communicate how fucked up a situation is to someone, it's exhausting and derailing to have to continually validate them and say "of course I don't mean everyone here, implicitly from the context I mean everyone who is engaging in this behaviour".

[1] I had to think a bit then for the phrasing where I could most easily argue "of course I don't mean every cis person" without also weakening the statement with extra clauses.

Excellent points. Are there any examples of guidelines you think would be good starting points?

It's interesting to look at the format of the Django Code of Conduct, although the content would be different since this is for a project instead a forum: https://www.djangoproject.com/conduct/ - including detailed explanations of what it considers important, examples of types of problems, a FAQ explaining the code of conduct, a reporting guide, an enforcement manual, and a changelog.

The Flickr Community Guidelines are also interesting in format: https://www.flickr.com/help/guidelines/ - with many very specific details about acceptable and unacceptable behavior.

Thanks!

I agree with your general sentiment. However, it only takes one member of the underrepresented group (with endorsing capability), or any member of a centrist or neutral party, to approve the a comment. (And as of so far, it doesn't seem like pending comments can be disapproved, e.g. by a hostile party.)

Granted, it's true that by nature of being underrepresented, the probability of that one member showing up to approve the comment may be unacceptably low. But I doubt it. In other words, to use your example, as long as the probability is high enough that at least one woman will show up to approve another woman's comment, there shouldn't be a problem.

Personally, I think that this probability is pretty high, though, as pg mentions, this'll have to be determined empirically. For example, it's entirely possible (as discussed in previous threads), that the population of active endorsers are skewed toward certain groups. Or that even a slightly <100% probability of legit comments being seen can ultimately compound systemic biases in the long run.

I think it is very important to not discount the emotional complexity of forcing someone into a position where their comments must be "endorsed" by a group of people they are perceiving as "hostile" to them.

Even if there are people who would endorse the comment, that is not what the person leaving the comment is likely going to be "feeling" in that moment: they are reading a bunch of comments that they are bothered by--ones which none of the other people around seem to be taking serious issue with (which is a key part of the original problem statement)--and those are the people they are going to perceive as being the ones who must endorse their comment: the ones to whom they are effectively submitting themselves for "endorsement".

(Which, frankly, I think is a separate reason why this endorsement system doesn't make sense as a solution to an endemic problem: most of those issuematic threads are filled with users--many of which have high karma--who are reinforcing the negative comments; it is unclear to me why the negative comments aren't going to have an easy time getting endorsements given that they currently don't have a difficult time getting upvotes.)

I thereby feel like even if this feature "worked" (and again, it isn't really clear to me why this would help, given that it isn't like these users are currently being downvoted or flagged out of the conversation) it still might not be an "appropriate" way to solve these underlying problems (which I would claim are inherently messy and emotional).

I agree with you and your parent. If this system goes into place I'll probably spend far more time on HN than I usually do these days -- at least at first -- specifically trying to approve comments made by other women, in the fear that others with enough karma will dismiss their point of view.

However, that won't stop many of the negative comments you mention, posted by and/or endorsed by high-karma members, also showing up in the thread. It only takes one high-karma approver to allow the negative stuff through and we're back to square one, but possibly excluding some underrepresented groups.

I like the idea in general, that is, I like the drive to improve the quality of HN comments. I'm not sure that flipping 'flag' around into 'approve' is the solution, but sadly I don't have any better ideas. Well, other than post emotion analysis and multiple-tiered karma systems (+1 Insightful, anyone?).

I agree completely. It appears that the problem behind the newly announced commenting changes can be simplified to an "innocent until proven guilty" argument. The new system assumes that the users commenting on HN are immediately guilty of offensive comments.

Another potential problem is that people will begin to purposefully tone-police themselves in an effort to appease the gatekeepers and ensure their comments are approved. This will likely result an atmosphere in which moderators and those with high karma create - one in which many unpopular opinions are silenced via unapproved comments.

As someone who is an inactive user, with absolutely no karma, I'm not looking forward to having to jump through hoops in order to have comments approved. Why not wait to see if I'm an actual troll or jerk before deciding that I need to be policed?

I'm concerned that giving comment-approval power to people with x amount of karma will reinforce many of the existing problematic aspects of HN instead of opening it up to better discussion

Given that endorsement is a mechanism which can only reduce the quantity of discussion on HN, it is difficult to imagine that it is designed to open up a discussion to more viewpoints.

I think your choice of "female work experience in technology fields" is interesting because it is one of the few legitimate topics where there isn't widespread agreement on basic foundations around the topic.

Others topics that fall into that category include "The Economics of Bitcoin", "Is Google Evil", "Startup X was bought by large company Y".

There's also a wide range of political & economic topics which are always controversial, but sometimes I question their legitimacy as a topic for HN.

I wonder if these controversial topics should have special handling so far as endorsement (and maybe even voting) on comments goes?

Given that endorsement is a mechanism which can only reduce the quantity of discussion on HN

That seems a mistaken assumption. The presence of existing comments often causes me to refrain from posting comments. Often I've been about to reply to something, then notice someone else has already said substantially the same thing, and as a result don't.

And this is not a contrived example. As I said when I first launched pending comments, if someone says something important in a nasty way, it may well turn out to be safe not to endorse it, because someone else will probably show up and make the same point without nastiness.

Plus if HN gets more civil, it may encourage people to comment who might not have before. I know from my own experience that incivility has decreased the number of comments I make. I often find myself about to say something, then decide I have work to do and don't have time to spend the afternoon fighting, and so don't say it.

Only experience can tell us whether one or both of these factors will increase the number of comments, but we can't assume that pending comments can only reduce discussion.

Often I've been about to reply to something, then notice someone else has already said substantially the same thing, and as a result don't.

Hm. Interesting...

I guess unique insights aren't, generally.

Amusingly, this already happens with dead users as well. I remember one thread where several European users were chiming in with agreement, all dead and invisible to anyone who hadn't set the flag. There was nothing nasty about their posts either, they were just the brutally honest type.

> It means that people with currently-underrepresented voices will have to "play nice" with the current mainstream of HN in order to be represented on the site.

Surely there's more to talk about on HN than one aspect of someone's life that puts them out of the HN mainstream?

We should be well rounded enough to be able to participate in some conversation on the site without swimming against the tide. That's not "playing nice", that's just not going against the grain in every single thread.

I love the idea of these guidelines and hope that if they happen, they'll be public! Thanks for outlining the idea so clearly.

I don't think it will be a problem in practice, both because women will be able to endorse comments and because during the test period people endorsed most comments. The only comments that weren't endorsed were troll comments.

The most recent test period was brief (just a couple hours) so it isn't representative of how it will work out in practice, but that's why I think we should turn it on for 24 hours and see what happens.

EDIT: On reflection, I apologize for how dismissive my comment must've sounded. It wasn't my intention.

> The only comments that weren't endorsed were troll comments.

I have always browsed HN with showdead on; I occasionally see dead "troll comments" and I hardly ever see not-dead "troll comments" that are not at least slightly downvoted. It would seem to me that the downvote and flag mechanisms are already sufficient, therefore, against this class of "troll comments".

The real question is whether serious comments--maybe even ones some users would consider "highly informative"--that also happen to have "negative affect" (or, even more difficultly, which are subtly dismissive) will actually have a difficult time getting the endorsements required to be posted on the site.

(Separately, there is then the issue as to whether one's own biases alter what they even consider "negative affect". There are some other comments on this thread that talk about "tone policing" which point out some subtle issues in how people perceive the tone of others when they are in disagreements.)

I also will note that even if getting endorsements is not a problem, it seems important to not discount the emotional complexity of forcing people to submit themselves for "endorsement" to the group of people they may already feel are being dismissive of their viewpoint. I wrote more on this here:

https://news.ycombinator.com/item?id=7484727

I just wanted to thank you for the comment. You make some good points.

I don't see that a brief test period is much of a model for sustained use. I think people will get tired of endorsing any strangers and mostly just look for friends' postings to let in.