Ask HN: Has anyone put thought into the degradation of online communities? | Hacker News Reader

Ask HN: Has anyone put thought into the degradation of online communities?

35 pointsasto38 comments

We've all seen this happen before. You discover this nice new website where there's interesting content and people (let's say slashdot). As more and more people join in, the quality of content/comments gradually starts to decline. Now there's people cracking the same tired jokes/memes over and over again. The people because of whom you signed up there in the first place lose interest and start to leave and that's the end of that. Another ex: Google+ just a few months ago was a place where everyone put an honest effort into making a comment that would be of some value to the reader. The amount of noise there too is quickly catching up to the signal.

I think if you've been on the internet long enough, you've probably seen at least one small online community grow rapidly and then blow up. Has anybody put any thought into how this may be avoided?

HN's is by far the best system[1] I have seen this far but I haven't been here long enough to know if things have changed significantly with time.

[1]write down super specific "guidelines" so everybody's clear as to exactly what is expected here.

View on HN

What you're desribing is the Eternal September[0].

There are strategies to mitigate the degridation of discourse, such as Hellbanning[1]. Are you asking if there is a way to prevent a small community from blowing up? Or are you asking if there is a way to prevent a small community from blowing up and being bogged down by crappy content?

I would say the best strategy would be to have good moderators, and be sure that you do your part to post quality content to the community. Be the change you want to see online. The more quality content, comments, etc. are on a site, the more it will be socially unacceptable to RickRoll[2].

Many sites implement a karma-type system to reward good content and punish bad conent (Reddit, HN, StackOverflow, etc).

However, two online communities come to mind that don't seem to have suffered in quality of content since their inception: 4Chan and C2Wiki[3]. I think this is due to their obscurity from the mainstream and also anarchist nature of their moderation system. On 4Chan, anyone can post, and there are so many posts, anything untasteful to the community is quickly buried in newer or more popular posts. On C2Wiki, anyone can edit, and edits that degrade the quality of the content are replaced just as painlessly with better content. Interestingly, both communities are anonymous.

How can you prevent people from cracking the same tired jokes again and again? Answer coming in 3... 2...[4]

I think first you must create an online community that caters to a specific niche, so users can already relate to one another because they share the same interests. Once people relate to one another, they want to share and converse[5].

Then, if the community grows in popularity, you must figure out how heavily you're going to moderate. Will you agressivley close off-topic posts and potentially turn away quality contributors (like programmers.stackexchange.com)? Or will you let everyone in, and create specific areas for new interests the community develops as it grows (like 4chan.org/sci)?

I see this a spectrum from razor-focused discussion to organic free-for-all. A mistake I belive one mistake that Slashdot made (and is still trying to rectify) is to straddle both ends of this spectrum. They have ended up alienating both the razor-focused users and the free-for-all users and all that is left is the snark.

So, I think, to prevent the degradation of online communities, you must chose which end of this spectrum your community is on, and moderate (or don't) accordingly.

[0]https://en.wikipedia.org/wiki/Eternal_September

[1]http://www.codinghorror.com/blog/2011/06/suspension-ban-or-h...

[2]Was going to be a YouTube link, but resisted the temptation to RickRoll HN because I might get downvoted.

[3]http://c2.com/cgi/wiki

[4]I'm really tired of this joke on HN, FYI

[5]http://www.browncafe.com/forum/ (one of my favorite niche online communities)

Very good. I do a lot of thinking about this too. I wrote this snippet a few days ago:

I think the core tri-state here is

1 users

2 restrictions

3 structure

The users are the target demographic of the sites content. They can be tween girls or seasoned engineers or perhaps people in certain moods (e.g. pornography, games)

The restrictions are the limitations from being say “admin” or “god” on the site. Can someone delete other’s content without any recourse, record or undo? Can they masquerade themselves as someone else? Just what are they permitted to get away with?

The structure is the purpose and intent of the usage of the site. Is there a forum for user contribution? Is there a place, say on think geek for user photos of products?

I think a model with retention and growth must somehow balance the users it's attracting, the restrictions they are given, and the structures of creation they can employ. Together these form a social product.

I don’t think it’s necessarily important to get these right at first blush as much as it is important to know what part you are lacking and what direction you need to turn to based on whats going on.

It has to be reactive (or perhaps, adaptive). When you get the formula unbalanced, it may look good for a while (large growth accompanied by a decrease in quality is the common one), but then it will collapse under it's own disarray.

Metafilter is a nice one. They simply charge $5 to sign up. That's it. Keep (2) open, never have a low barrier with (1) and have it be a, I'm not going to say social aggregator, but a collaborative blogging platform for (3).

Do you know the Myers-Briggs personality indicators? The relevant one here in my mind is Judgemental/Perceptive: http://en.wikipedia.org/wiki/Myers-Briggs_Type_Indicator#Lif...

Good online communities start with a lot of Perceptives, which are often interesting to listen to. At some point the Judgementals start appearing, and the discussion moves, to use a current example, from the structure and potential future implications of Stuxnet to whether Stuxnet is legal or America is hypocrite etc.

The problem is that the arrival of the Judgementals is a positive feedback loop. Judgemental comments breed judgemental counter comments (pro/con) and soon the Perceptives are drowned out in the noise, leaving the discussion and further increasing the share of Judgemental comments.

BTW I think the same happens everywhere where you cannot measure the actual worth of someone's contribution. Politics take over and intelligent people leave.

"The world is made out of two kinds of people..."

Some discussion formats encourage this (Amazon Forums, LinkedIn) and other publishers encourage the battle of the judgementals... The New York Times invented a new section of commentary designed specifically to encourage knock-down drag-out discussions that go nowhere. It's good for the numbers.

Other phenomena contribute to heat death too. Forums on LinkedIn tend to be awful, particularly in the A.I. area. That's because anybody who knows anything has this neural net they're tending that works some miracle, and everything about it is a trade secret. The people who do talk want to rehash, over and over, unproductive questions like "What is intelligence?" ahd "Why has A.I. failed?" while the rest of us are quietly building systems that can do some of the things you do better than you can do it.

Yep, it's always hard to find the place where the builders converge as opposed to those who have opinions. That's what HN used to be, once.

Interesting theory. What do you think of the following test? Someone starts a new online "meeting-place" (yeah, I know, it takes a lot of work for a new one to get traction) and take pains to ensure that all the moderators are Perceptives. Make it an explicit norm of the meeting-place to avoid judgemental comments. Then wait to see whether this new meeting-place degrades less quickly than average.

I tend to think that most people can learn not to express judgements if they want to become accepted by the others at the meeting-place (and that it is possible to exclude those who will not learn).

Seems to me the key is topics - it's easier to stay perceptive when discussing Python micro frameworks than the Middle East.

When HN started featuring contentious topics like that, it brought out the judgmental side of people, and later influenced even technical threads.

I remember first being impressed by HN when seeing a discussion on some PHP framework that didn't contain any of the inane "PHP sucks" slogans I've gotten used to on Proggit. That impressed me not because of my opinion on PHP, but because it was clear these people deal in building things - a language is a tool for them, not a fashion statement. As more recent PHP-related threads show, that's not the case anymore.

It's "Judging", not "Judgemental". It doesn't mean the same thing.

http://users.trytel.com/~jfalt/Ene-med/j-p.html

"Judgers are very organized. They like to "plan their work and work their plan." A Judger likely has a day planner or if she doesn't have one, she carries one around in her head. Work is carried out in an orderly fashion. Even holidays need to follow a plan."

"Perceivers like to go with the flow. If they have a "to do" list, it will likely be just a scrap of paper. They are always looking for new information."

Obligatory references on the subject of the unvalidated Myers-Briggs Type Indicator®:

http://www.skepdic.com/myersb.html

http://www.psychometric-success.com/personality-tests/person...

"Overall, the review committee concluded that the MBTI has not demonstrated adequate validity although its popularity and use has been steadily increasing. The National Academy of Sciences review committee concluded that: ‘at this time, there is not sufficient, well-designed research to justify the use of the MBTI in career counseling programs’, the very thing that it is most often used for."

http://www.indiana.edu/~jobtalk/HRMWebsite/hrm/articles/deve...

Sure, I don't really refer to Myers-Brigges as such but to the J/P distinction, which for me is a useful way to think about this.

Stack Exchange

Jeff Atwood and Joel Spolsky used to run a podcast where they described exactly the techniques they had discovered.

- Communities are segmented according to specialty so as to concentrate expert-level talent. Compare the kinds of questions listed on an SE site to what's asked on any other Q&A site out there. The math questions are asked by actual mathematicians, the photography questions are asked by actual photographers, etc. (You'll also notice that---beyond the gaming site---most of the silos are dedicated to a profession.)

- Many of the professional sites are dedicated exclusively to people in that profession. The Theoretical Computer Science silo explicitly requires PhD-level material. The Quantitative Finance site opens the FAQ with "if you aren't earning a living at this, it's probably off topic".

- Most SE sites explicitly prohibit subjective questions. Something like, "What's better, Ruby or Python?" will be closed immediately.

- Any post can be edited by another user, just as with Wikipedia. So grammar and spelling tend to be pretty decent.

- Moderators are given broad powers to close questions, delete posts, suspend users, and even delete accounts.

- There is a separate Meta site for questions about the site. That segregates questions about the silo's topic from questions about the silo.

Reddit has some siloing as well, but they don't prohibit memes, which is why the content looks the way it does. They also don't have user editing or strong moderation. It shows.

Reddit has some siloing as well, but they don't prohibit memes, which is why the content looks the way it does.

Reddit hardly prohibits anything. Individual subreddits do filter a lot of content, including memes.

This doesn't really fix everything, though. Problems still arise, and things still get out of hand from time to time. There was some community drama on the front page today...

And it goes the other direction, too. One of the most strictly moderated subreddits, AskScience[1], is about as dry as it gets. I don't know if I'd call that group a "community".

1: http://www.reddit.com/r/askscience

Clay Shirky has studied this extensively and is probably the best resource. "A Group Is Its Own Worst Enemy" [0] is perhaps the best single piece on what you're thinking about.

He offers several design guidelines toward the end of the piece, which I suspect informed the design of HN (and the posting guidelines therein).

Another technique he mentions is that of MetaFilter, which disables their new user page when they feel the effects of scale.

[0] - http://shirky.com/writings/group_enemy.html

I don't know how to solve the problem, but I have thought about it; some thoughts:

The computing industry seems to keep repeating cycles with local and remote processing - thin terminals and mainframes, then the desktop PC, now "The Cloud" is mainframes 2.0, then mobile apps are starting to bring native code back to the user again (although many are tied into the cloud) - similarly, online communities keep alternating between distributed (BBSes + IRC), centralised (myspace, AOL), distributed (forums + IRC again), centralised (facebook + twitter). As such, the churn of users from one system to another is pretty much constant. (There are many reasons for users to move from one system to another, but this is the one that I've seen as an inevitable constant since the internet began, and other reasons are different each time).

How does users moving around relate to quality of community though? Perhaps I'm being subjective and biased, but it seems to me that the highest signal to noise ratio happens at the front of the wave, the masses with the pop culture are the middle, and the ocean of spam trails behind. As such, if we assume that users will move from one system to another no matter what, the degredation of community at a given location is inevitable.

Taking slashdot as an example - the technical side of it, while not perfect, was fairly good and did a fine job back when a handful of technically-inclined people used it; but when it became overwhelmed with average people, there's nothing the system could do to stop it (I don't think HN is immune to this either, it just happens to be at the front of the wave right now). Also in the case of slashdot, the editors started treating the average people as their target audience, which accelerated the decline...

Also, thanks to things like Ubuntu, slashdot's niche of "linux users" stopped being a niche, thus opening the floodgates - I find that communities are generally better when you need to put effort into finding them, rather than them being the default that the lazy people go to for a popular subject (note that this isn't quite hipster sentiment - that would be to say the subject itself becomes worse when it gets popular).

Yes. I flirted with the idea of an online community with a half life. For example, every 6 months the URL would completely change and only the top 1% of members would be notified of the new location. It would be in their hands to repopulate the community.

I haven't figured out how to determine the top 1% but this was more a pipe dream a than serious potential project. Anyone is free to take the idea and run with it.

I think this is a fun idea. People who run sites like reddit or HN or fark are attempting to make themselves middlemen to advance various of their interests. As a user I don't lose very much by finding one cool link on reddit and another on HN. To me it is just another url. But if a prolific respected user of HN migrates fo a new place, that probably causes the proprietor of HN more pain than the users of HN. Because unless you are for some reason confining yourself to the ghetto that is HN, you most likely are looking in a few different places for your enlightenment. A community with a fuse, as it were, would be quite healthy.

The NewsWelcome page is excellent.

(http://ycombinator.com/newswelcome.html)

Meatball Wiki has many people talking about creating and controlling online communities. It's focused on wikis, but is applicable to other types of communities.

(http://meatballwiki.org/wiki/)

The Japanese origin of anonymous posting (leading into the chan culture) is interesting and is very different to the western use (where "anonymous == arsehole".)

(http://www.ojr.org/japan/internet/1061505583.php)

> Q: Why did you decide to use perfect anonymity, not even requiring a user name?

> A: Because delivering news without taking any risk is very important to us. There is a lot of information disclosure or secret news gathered on Channel 2. Few people would post that kind of information by taking a risk. Moreover, people can only truly discuss something when they don't know each other.

> If there is a user ID attached to a user, a discussion tends to become a criticizing game. On the other hand, under the anonymous system, even though your opinion/information is criticized, you don't know with whom to be upset. Also with a user ID, those who participate in the site for a long time tend to have authority, and it becomes difficult for a user to disagree with them. Under a perfectly anonymous system, you can say, "it's boring," if it is actually boring. All information is treated equally; only an accurate argument will work.

Clay Shirky's "A Group Is Its Own Worst Enemy" from 2003 is a bit of a classic.

http://www.shirky.com/writings/group_enemy.html

I think a community's approach to moderation can also accelerate its degradation.

Consider a site where rules are overly broad, moderators are chosen by their ability to assert rules blindly and are actively dissuaded from using their own judgement, and meta discussion is not allowed (i.e. about the site and its rules). Furthermore, people who disagree with each other often enough are silently (shadow) blocked from each other.

We've read a lot of stories here about the importance of seeing things you disagree with and not creating an echo chamber for yourself; consider why a community would actually impose that on users.

These are symptoms of a community, IMO, chasing the lowest common denominator: sanitize the site as much as possible to sell ads, because disagreement debate drive eyeballs away. A good CPM can certainly be established in a place where everything is awesome and people are never wrong. Dictatorships-with-secret-police generate ad revenue, but not any real discussion.

I find many of the posts on this thread to be interesting, but I personally do not think that this entropic process is a problem. I like decay and I am attached to the plasticity of the internet.

Xianhang Zhang has written a lot about social experience design (what you're talking about he touched on a little in the post about the Evaporative Cooling Effect): http://blog.bumblebeelabs.com/social-software-sundays-2-the-...

It's tough.

The site I feel worst about is Tvtropes. A few years ago they had the great idea of basing a universal theory of culture derived from Japanese animation. Brilliant!

Then they get in trouble with AdSense because of drivel about sexual fetishes in tv and comics written by virgins, and it's just overrun by so many people that are completely incoherent that anybody who knows anything about writing or directing or acting just can't stand it.

Wikipedia avoids this fate by being aggressive about deleting stuff. Wikipedia is one of the least friendly web communities out there. But, it really has managed to go mainstream without falling apart.

pg has written at length on the dilution of online communities: http://paulgraham.com/hackernews.html

Seems obvious, but surely there are some who haven't come across it yet.

I just want to chime in and say that I've been thinking about and researching solutions to this for years. Only recently have I been pushed towards creating a real product that does solve this problem (due to enough people liking the idea at a Lean Startup Machine competition for us to win). The idea is relatively simple...we prioritize content by segregating people into dynamic communities consisting of users who are highly similar behaviorally. The goal is to create a million little hacker newses behind the scenes.

Practical implementation is more of a nightmare, but at this point we're in limited beta and you can sign up to receive an invite at http://foldr.co (just go to launch in the upper right and put in your email for the invite).

This project is my personal crusade, primarily because I have watched so many great communities die. I can't say for sure whether or not the solution will ultimately work, but it won't do any good if nobody tries.

Paul Graham thought about this as HN grew. http://paulgraham.com/hackernews.html

Also, the phenomenon where the people who are most invested in a community are the first to bail when it changes is often called "evaporative cooling".

I run one of the largest current internet forums, there are multiple problems (which I could talk about for hours) but ultimately the only solution is exclusivity. People are affected by their surroundings; as soon as you make a site public it will start to change.

I've been working on a Reddit / HN like site that uses a different ranking system meant to avoid this problem, see this if you're interested:

http://www.thoughtocean.com/what/dilution

It looks interesting, but why is a sequence of users that vouch for the final user detrimental? User(1) would vouch for user(2), which should make user(2) more trustworthy, and therefor have greater voting power. Wouldn't this phenomenon simply stack as user(3) becomes more trustworthy because of the trustworthiness of user2? So that the final user, user(n) would be more trustworthy then a similar user who only had a sequence of two users, with the final one vouching for him? Besides that, however, it looks like a good idea, and i personally am interested in the final product.

Spolsky has an old article on how the UI of the software used affects the community: http://www.joelonsoftware.com/articles/BuildingCommunitieswi...

It's because of Dunbar's number http://en.wikipedia.org/wiki/Dunbars_number

Communities are meant to be 150 people or less, where everyone knows everyone else.

I think the potential relationship between Dunbar's number and online communities is interesting - I had never thought about it that way before. But certainly online communities don't break down after a few hundred people. Why must everyone know everyone else for a community to remain high quality? Or closer to the original question, why do you think communities degrade after everyone doesn't know everyone else?

The only way to avoid it is to have an invite-only site where new invites are interviewed and voted on by a committee/small group of members.

The problem is that as a site gets bigger, more and more people join that have views that are opposed to the original intentions of the site (and owners) and the site degrades over time.

The other problem is that you don't know the age of the person posting. Reddit is now filled with high school/younger kids posting.

It parallels a true democracy. If we had a true democracy, eventually the extremes would rule.

One thing I've seen done that it interesting is a virtual basement. My favorite online forum is advrider.com. It a vbulletin for adventure motorcycle enthusiasts with over 200k users, 500k threads, and 17m posts. It works surprisingly well because it has a "basement" where anything goes. The understanding is upstairs is for constructive conversation, and the basement, especially "Jo mamma" is for time wasting, vitriolic, political/religious/Etc, posts. If you start a pointless thread, it gets banished to jo mamma.

I know of a few online communities that have blown up and survived. The key was a set of sardonic admins and a site owner that paid for the hosting himself.

I think that an ad supported online community is incongruous with a quality online community.

This happens everywhere and all the time. As the community gets more popular and attention, more and more people will sign up and pollute everything. From what I've seen, most of the time, redditors are blamed for that here on HN :P

Tragedy of the Commons. What we really need is more incentive for people to maintain the quality of discussion, like if they were economically invested in the site.

What, like SomethingAwful's "tenbux" policy? I guess it could work, if it wasn't spoiled by cliquey moderation.

See, for example, griefers on Medal of Warfare on Xbox Live.

They've paid for the game, and they're paying for the xbox live account, and there is reputation and reporting.

That does nothing to stop people like "General Minus" from trolling the heck out of other players.

Dedicated trolls / griefers will have many alternate accounts. Team AVO (Minecraft griefers) apparently had several thousand alt accounts. Or trolls will use other means. Grawp (wikipedia troll) posts disguised links that perform a wikipedia action. Lots of people click the links.

The other problem is that customers feel entitled. SuicideGirls, a NSFW site with a set of forums, had severe problems with awful posts. They implemented "spring cleaning" to sort it out, many customers expressed unhappiness, a few people were banned and a new board was introduced and low quality posting was tolerated in the other boards. A lot of drama for a tiny increase in quality on most boards.

Reply on news.ycombinator.com