Clay Shirky: A Group Is Its Own Worst Enemy - Social Software Design(clifford.webservepro.com)
clifford.webservepro.com
Clay Shirky: A Group Is Its Own Worst Enemy - Social Software Design
http://clifford.webservepro.com/pipermail/nec/2003-July/000022.html
13 comments
Heh. Meta discussion that is also on topic.
Note that the article is from 2003. This problem has been around for a long time. As far as I know, no one has ever really solved it.
One thing I've noticed on Reddit and Digg is that discussion stops being a conversation when the number of posts gets too high.
It's still a conversation when you have forty or fifty posts: you can read the whole thing and understand what is being said before you contribute. When every conversation has over 100 posts, you don't have one conversation any more. You have a bunch of parallel conversations, often repetitive. Worse, you start seeing more posts where the author is not trying to talk to another person, but to stand on a soapbox and make points to the crowd.
I don't think this is (necessarily) a failure of etiquette or good manners, just a limitation on how much information people can digest in a casual setting like a web forum.
Of course, as you get more users, you will also get a higher absolute number of people who are intentional troublemakers or who are incapable of understanding the community standards. Unfortunately, these guys have a tendency to completely take over a thread.
One possible solution that I'd like to see someone try is intentional sharding of the community. Costly, but worth it IMO if it kept some enclaves of high quality discussion. You could give everyone read only access to any shard, which would be a decent option for the lurking majority.
Though I don't know of anyone who has solved the problem of community deterioration, my gut feeling is that it is solvable, someday. At least to the extent that a good conversation is about good manners and paying attention, and not about showing up at the place that is currently in fashion.
Note that the article is from 2003. This problem has been around for a long time. As far as I know, no one has ever really solved it.
One thing I've noticed on Reddit and Digg is that discussion stops being a conversation when the number of posts gets too high.
It's still a conversation when you have forty or fifty posts: you can read the whole thing and understand what is being said before you contribute. When every conversation has over 100 posts, you don't have one conversation any more. You have a bunch of parallel conversations, often repetitive. Worse, you start seeing more posts where the author is not trying to talk to another person, but to stand on a soapbox and make points to the crowd.
I don't think this is (necessarily) a failure of etiquette or good manners, just a limitation on how much information people can digest in a casual setting like a web forum.
Of course, as you get more users, you will also get a higher absolute number of people who are intentional troublemakers or who are incapable of understanding the community standards. Unfortunately, these guys have a tendency to completely take over a thread.
One possible solution that I'd like to see someone try is intentional sharding of the community. Costly, but worth it IMO if it kept some enclaves of high quality discussion. You could give everyone read only access to any shard, which would be a decent option for the lurking majority.
Though I don't know of anyone who has solved the problem of community deterioration, my gut feeling is that it is solvable, someday. At least to the extent that a good conversation is about good manners and paying attention, and not about showing up at the place that is currently in fashion.
I recall PG saying something a while back about a "rollback" feature that would give you the option to only display those comments and submissions created by users above a certain account-age/karma threshold. So in effect you could turn a knob and the site would roll back to, say, the user base from 300 days ago.
I thought it sounded like a pretty elegant (if potentially divisive) solution, maybe worth serious consideration.
I thought it sounded like a pretty elegant (if potentially divisive) solution, maybe worth serious consideration.
easy in principle, pretty hard in practice.
It's a recursive problem you'd basically be talking to nobody (and you wouldn't be reading this comment on your writing either).
After all what's to stop those 'above' you from setting the clock back even further.
Eventually only those people not using that feature will be able to communicate, which puts you back at square one.
It's a recursive problem you'd basically be talking to nobody (and you wouldn't be reading this comment on your writing either).
After all what's to stop those 'above' you from setting the clock back even further.
Eventually only those people not using that feature will be able to communicate, which puts you back at square one.
>> It's a recursive problem you'd basically be talking to nobody...
Presuming this was done where each person had fine control over the date, you may be right. However, suppose each person only had a single binary control, essentially "hide everyone newer than me."
Furthermore, rather than "newer than me" being defined by a join date specific to each user, instead it would be defined by the start date specific to a particular generation of users (the news.yc folks do seem to track mass user migrations). Visually (bars are new.yc join floods, time moves to the right):
This would still provide each user with the ability to view the site more-or-less as it was upon first arrival, without actually allowing each user fine granularity in locking everyone out.
(for the record, such a feature isn't as useful as simply applying social & administrative pressure to keep quality up; I just wanted to mention an alternative implementation that doesn't turn into an arms race)
Presuming this was done where each person had fine control over the date, you may be right. However, suppose each person only had a single binary control, essentially "hide everyone newer than me."
Furthermore, rather than "newer than me" being defined by a join date specific to each user, instead it would be defined by the start date specific to a particular generation of users (the news.yc folks do seem to track mass user migrations). Visually (bars are new.yc join floods, time moves to the right):
time: |--------|---|------|-----|---...
group: 1 2 3 4 5
This would still provide each user with the ability to view the site more-or-less as it was upon first arrival, without actually allowing each user fine granularity in locking everyone out.
(for the record, such a feature isn't as useful as simply applying social & administrative pressure to keep quality up; I just wanted to mention an alternative implementation that doesn't turn into an arms race)
I've been pondering how to solve this problem for years, and I'm no closer to solving the problem than anyone else.
Your approach involves freezing the user base so that nobody new can join. Granted, it's an innovative take on it I've never seen before, but that's still what it is. The problem with all such approaches is that a certain percentage of the user base wanders off over time, so your frozen userbase constantly shrinks. Eventually it gets below the point you care. (Or, in most cases, below the critical mass of the community.) It only works correctly in the static case where nobody ever leaves.
In your case you could raise the threshold, but all that does is expose you to the very things you were trying to get away from. If there was a correlation between "the people who stuck around" and "quality posters" that might work, but I doubt such correlations exist. (Then again, I don't know, as my words imply.)
About two years ago, I scraped a single article with over a thousand comments on Slashdot and looked at all the UIDs. I theorized that all the old folk would be gone. Instead what I found is that the UIDs were fairly uniformly distributed across the entire base, which, if you think about the temporal distribution of people joining, is actually a moderately interesting result. I just eyeballed the list of unique UIDs, I didn't actually plot it out, but it wasn't what I was expecting. Some small core stick around and keep posting, but over time the vast bulk of the early group leave.
Your approach involves freezing the user base so that nobody new can join. Granted, it's an innovative take on it I've never seen before, but that's still what it is. The problem with all such approaches is that a certain percentage of the user base wanders off over time, so your frozen userbase constantly shrinks. Eventually it gets below the point you care. (Or, in most cases, below the critical mass of the community.) It only works correctly in the static case where nobody ever leaves.
In your case you could raise the threshold, but all that does is expose you to the very things you were trying to get away from. If there was a correlation between "the people who stuck around" and "quality posters" that might work, but I doubt such correlations exist. (Then again, I don't know, as my words imply.)
About two years ago, I scraped a single article with over a thousand comments on Slashdot and looked at all the UIDs. I theorized that all the old folk would be gone. Instead what I found is that the UIDs were fairly uniformly distributed across the entire base, which, if you think about the temporal distribution of people joining, is actually a moderately interesting result. I just eyeballed the list of unique UIDs, I didn't actually plot it out, but it wasn't what I was expecting. Some small core stick around and keep posting, but over time the vast bulk of the early group leave.
Over time the vast bulk of any group leave, not just the early adopters.
They'd have to, there are just too many reasons to leave and only a very limited few to stick around indefinitely.
Eventually everybody dies, gets a life or lands in some sort of crisis in their lives that changes it to the point where spending time on social sites is just too expensive.
They'd have to, there are just too many reasons to leave and only a very limited few to stick around indefinitely.
Eventually everybody dies, gets a life or lands in some sort of crisis in their lives that changes it to the point where spending time on social sites is just too expensive.
I liked the bit in the article about MetaFilter shutting off the new-user page whenever they got too much press.
Maybe turning off new signups whenever some popular blog mentions Hacker News might raise the bar somewhat. That would get only those in who find HN interesting enough to bookmark it until the flood drops off.
Maybe turning off new signups whenever some popular blog mentions Hacker News might raise the bar somewhat. That would get only those in who find HN interesting enough to bookmark it until the flood drops off.
This is the article that inspired me to do my PhD thesis.
nice, what are you researching?
Design guidelines for social software. How do we best think and reason about the issues around trying to achieve social goals in software design.
Everyone involved in community projects should read this (doesn't matter whether online or offline)
audio available?
http://news.ycombinator.com/item?id=7354
http://news.ycombinator.com/item?id=24992
http://news.ycombinator.com/item?id=122189
I find it to be relevant to the recent discussions on HN.
Here's a summary of the essay by Jon Stahl http://blogs.onenw.org/jon/archives/2003/10/21/a-group-is-it...