Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Tell HN: Uptick in mean-spirited comments

news.ycombinator.com

Post reply on HN
Print view View original

91–100 of 297 posts

Re: Tell HN: Uptick in mean-spirited comments

#91

Post by wnoise »

SandB0x wrote: post #88
maxklein wrote: post #70

Earlier quoted context omitted.

It would be available to everyone and google, since the point is that what you say is forever linked to you on the internet.

> and google > forever linked to you on the internet I'm happy for you as a respected user of this site to know who I am. Google - forever and ever - not so much. As far as comment quality goes, you might cut out some of the crap, but you may also get more sterile discussions from good citizens such as myself. My hunch is that some people just have an attitude problem, and will be a detriment to HN with or without th…

> I don't mind if you as a respected user of this site knows who I am. Google - forever and ever - not so much.

Unfortunately, you can't get the first without the second, as witness the leak of the "journolist" e-mails.

Re: Tell HN: Uptick in mean-spirited comments

#92

Post by rjurney »

pg wrote: post #30

Incidentally, if anyone has any suggestions for technical changes/features/tricks that would help fix the problem, I'm all ears. Fending off the decline everyone thinks is inevitable for forums is the main thing I work on. It has been since practically the beginning.

Flagging an entire thread by clicking 'troll' would help. The up/down voting tends to promote trolling, as the first controversial cliche is the top comment, having spawned a deep threaded conversation with lots of emotions and voting. A different symbol is needed. Something that makes people take pause and say, 'Is this a good subtopic of this link, or is it the exact debate anyone would have predicted?'

Put troll threads/comments at the bottom of the page, or hide them. Or use spam ratings to amplify users' votes who up vote non spam comments.

Short of that, add the ability to minimize entire threads.

Re: Tell HN: Uptick in mean-spirited comments

#93

Post by metamemetics »

maxklein wrote: post #76
metamemetics wrote: post #69

Earlier quoted context omitted.

Seems unnecessary considering some of the most thoughtful discussions come from throw-away accounts doing Tell\Ask HN. It could also exacerbate other issues: http://en.wikipedia.org/wiki/Argument_from_authority [ ie. comment upvotes co-vary with name of commenter instead of content communicated ]

That's why I believe there should be no real disadvantage to using an anonymous account, and no way of telling in the comments if a comment is anonymous or not. All it does is encourage people to link their real life account in a behind-the-scenes manner, which will raise the general discourse level.

> which will raise the general discourse level

On one hand you want to avoid trolls, but on the other hand you want to avoid a circle-jerk and "my boss can google everything I wrote" timidity and un-hackerish conformism. Tending toward either extreme can disincentivize thoughtful discussion and negatively impact a community. So I would be hesitant to declare that a safe assumption.

Re: Tell HN: Uptick in mean-spirited comments

#94

Post by ramanujan »

pg wrote: post #30

Incidentally, if anyone has any suggestions for technical changes/features/tricks that would help fix the problem, I'm all ears. Fending off the decline everyone thinks is inevitable for forums is the main thing I work on. It has been since practically the beginning.

You have the full HN dataset and voting logs. What if you used it to train a simple classifier that would predict the likely karma of a comment as a function of a number of features, including username, username responded to, bag-of-words vector of the comment, time of day, total number of comments in thread, "controversy" of the comment (as reflected by a mix of up & down votes) and similar factors?

The goal could be something like Google's Beer Goggles, but for comments: "it looks like you're writing a flame, are you sure you want to do that?"

Another thought: there's a big difference between a comment at -1 with 10 upvotes and 12 downvotes vs. one that has occasioned 2 downvotes. Might be interesting to see the effect of showing both the number of up & down votes rather than just the sum. The idea is that genuinely trollish comments at -1 will be differentiated from minority points of view.

Re: Tell HN: Uptick in mean-spirited comments

#95

Post by _delirium »

maxklein wrote: post #47
pg wrote: post #30

Incidentally, if anyone has any suggestions for technical changes/features/tricks that would help fix the problem, I'm all ears. Fending off the decline everyone thinks is inevitable for forums is the main thing I work on. It has been since practically the beginning.

I believe that the less anonymous a forum like this is, the more polite it will be. You have said you want people to speak to each other here as they would in the real world: well, I believe the more strongly you tie their real world identity to their account here, the more polite they would be. My suggestion, similar to that experiment in the past, and similar to twitter, add a "verified user" badge in the users pro…

I haven't found anonymity to be very well correlated with politeness on tech discussion forums I'm on. Some of the most abrasive ones are filled with people posting under their real names, and folks like rms, Theo De Raadt, Ulrich Drepper, and a bunch of the comp.lang.lisp crowd have no problem flaming you under their real names.

Re: Tell HN: Uptick in mean-spirited comments

#96

Post by sqrt17 »

maxklein wrote: post #70
SandB0x wrote: post #61

Earlier quoted context omitted.

As an anonymous user I actually kind of like your idea, but can you clarify whether you propose that being "verified" would mean your identity is available to everyone or just to other verified users? ("I was walking home one night and a guy hammering on a roof called me a paranoid little weirdo. In morse code." - Emo Philips)

It would be available to everyone and google, since the point is that what you say is forever linked to you on the internet.

I think there is a distinction between saying stupid things, contributing to sensitive topics, and outright trolling which you're ignoring here. All three of them are things that I wouldn't want to be part of my permanent record, but:

* Saying stupid things (Software X doesn't support Feature Y, which is good if you want to do Activity Z) and then be corrected (because in the meantime, Software X got a Feature Y for the latest version) provides overall value to the discussion because other people are likely to have made the same mistake that I did.

* Contributing to sensitive topics (be it gender equality, the question of "race", or any political question) will always send the goon squads your way whenever you voice a non-majority opinion. Admittedly, HN has a policy that these contentious topics are not great material for posting, but you get one of these once in a while and it's hard to make an informative contribution to such an issue without drawing ill will from anyone (which will result in downvotes to an anymous HN poster, but could be potentially harmful if those trying-to-be-well-informed-but-not-necessarily-popular opinions ended up on your permanent record). As an example, consider the question of founders being able to raise kids at the same time. While there are people who do a good enough job at it to be able to step up and say, I can do it just fine, and people who can claim anything because people will listen to them anyways, anyone in between would run the danger of drawing the ill will from people of one side or the other for claiming that it's possible or not so possible.

* Outright trolling is something you would not want in the comments, but there's a very fine line between a pointed remark that someone made, and trolling. And as with blog posts, successful trolls are still successful when they are not anonymous.

I think one of the problems here is that, when the audience grows large enough, a pointed post will draw a larger number of upvotes than a boring-but-informative one. So it may be useful to partially decouple the comment-ranking function of up-downvotes with the karma calculation to disincentivize karma whoring.

For example, one could reduce the weight of comment up-/downvotes whenever the total number of (comments+upvotes+downvotes) of an article crosses a certain threshold - say 100. In this way, having a +20 voted comment on an article that barely made it to the frontpage would count more than a +60 voted comment on an incendiary topic that draws a total of several hundred up/downvotes.

Re: Tell HN: Uptick in mean-spirited comments

#97

Post by invertedlambda »

wnoise wrote: post #91
SandB0x wrote: post #88

Earlier quoted context omitted.

> and google > forever linked to you on the internet I'm happy for you as a respected user of this site to know who I am. Google - forever and ever - not so much. As far as comment quality goes, you might cut out some of the crap, but you may also get more sterile discussions from good citizens such as myself. My hunch is that some people just have an attitude problem, and will be a detriment to HN with or without th…

> I don't mind if you as a respected user of this site knows who I am. Google - forever and ever - not so much. Unfortunately, you can't get the first without the second, as witness the leak of the "journolist" e-mails.

Agreed. If you don't want your presence to be known on the Internet, there's a very simple way to accomplish that.

Don't post, don't publish, don't comment. Simple as that.

Some time ago another person made a comment about pearls and some sort of livestock. Throwing...things...in places?

Re: Tell HN: Uptick in mean-spirited comments

#98

Post by joelmichael »

ramanujan wrote: post #94
pg wrote: post #30

Incidentally, if anyone has any suggestions for technical changes/features/tricks that would help fix the problem, I'm all ears. Fending off the decline everyone thinks is inevitable for forums is the main thing I work on. It has been since practically the beginning.

You have the full HN dataset and voting logs. What if you used it to train a simple classifier that would predict the likely karma of a comment as a function of a number of features, including username, username responded to, bag-of-words vector of the comment, time of day, total number of comments in thread, "controversy" of the comment (as reflected by a mix of up & down votes) and similar factors? The goal could b…

I hope you would reward controversy rather than punish it? I don't come to HN to read happythoughts.

Re: Tell HN: Uptick in mean-spirited comments

#99

Post by JesseAldridge »

pg wrote: post #30

Incidentally, if anyone has any suggestions for technical changes/features/tricks that would help fix the problem, I'm all ears. Fending off the decline everyone thinks is inevitable for forums is the main thing I work on. It has been since practically the beginning.

Possible solution 1: Smaller communities; Dunbar's number. I know it's been suggested before, but here it is again... When everybody knows everybody, people are nicer. Just create one unique, isolated instance of the site for every X number of users. Distribute current users randomly (or whatever) among the instances. Assign new users to the least populous instance. If an instance gets over max_number of users, split it in two.

Don't bother trying to do fancy stuff like connecting votes between communities; if something is really good people from each community will submit it separately. Also, it might be nice to be able to move between communities... but that sounds like something for version 2.

You could have these smaller communities run in parallel with the main site if this seems like too big of a shift.

Possible solution 2: Not sure if this really addresses the issue squarely -- but what if, when ranking stories and/or comments, you factor in the number of votes in the intersection between the current user and the submitter/commenter. Something like:

define votes(username): return ids of last n upvoted items for username

item_score *= (length(intersection(votes(user_x), votes(user_y))) / (n / 2))

Just use the original algorithm if the user hasn't yet cast n votes.

One man's garbage is another man's gold, as they say.

You could do something similar by checking downvotes as well. If two people are polar opposites (one person upvotes stuff the other downvotes), then comments by one could be greyed out or hidden for the other.

If there were simply a box I could check to make my votes public, then maybe somebody could write a Chrome or Firefox extension to pull this data and handle the re-ranking.

Neither of these two ideas would be perfect, but I think they would improve the user experience significantly. If nothing else they would be interesting experiments.

Re: Tell HN: Uptick in mean-spirited comments

#100

Post by ramanujan »

joelmichael wrote: post #98
ramanujan wrote: post #94

Earlier quoted context omitted.

You have the full HN dataset and voting logs. What if you used it to train a simple classifier that would predict the likely karma of a comment as a function of a number of features, including username, username responded to, bag-of-words vector of the comment, time of day, total number of comments in thread, "controversy" of the comment (as reflected by a mix of up & down votes) and similar factors? The goal could b…

I hope you would reward controversy rather than punish it? I don't come to HN to read happythoughts.

Indeed, the idea is that a "controversial" comment can evoke both disagreement and support, while a trollish comment would just result in downvoting.

OTOH, some particularly clever trolls are of course intended to provoke division of an intellectually unproductive sort. Not sure if machine learning could identify that...though features computed from the commenter's past history could be informative.

Post reply on HN