5 ms·

Controlling Positive Feedback Loops in Online Communities

22 points by dsirijus 14y ago

davedx 14y ago
"You need to be a nazi", that's quite a conclusion to make. Any evidence of this actually working?
- dsirijus 14y ago
  Stackoverflow.
- lutusp 14y ago
  Steve Jobs, who built what is now the world's most valuable company, at least in part by being intolerable? Parenthetically, I strongly recommend the recent Iverson biography, it's first rate. I worked with Steve Jobs, I couldn't stand him, and I eventually refused to work with Apple. I am sure there are many others who came to the same conclusion. But this didn't stop Apple as a company.
- tomflack 14y ago
  offtopic.com which had registration closed for a large amount of time in 2002-2003ish? It was still a massive forum but the community kind-of "solidified" and didn't grow larger, for the most part quality was pretty good. It seems to me that once critical mass is reached, closing registration is a good move. It works for other quality-oriented social sites like Oink too.
lutusp 14y ago
It would help if the author actually understood what "positive feedback loop" means in engineering. It's not what he thinks, and social-science terms like "positive feedback" don't translate well in the more rigorous sciences. In strict engineering terms, feedback loops must always have a gain less than unity, and a stable, reliable system normally has negative net feedback. But this doesn't sound "right" to a social scientist, who understands these terms in an entirely different way.
- dsirijus 14y ago
  I've made two examples to help others discern between positive feedback loop (engineering) and positive feedback (social sciences). Where from have you drawn that conclusion? Genuine interest.
  - lutusp 14y ago
    > I've made two examples to help others discern between positive feedback loop (engineering) and positive feedback (social sciences). I posted only because you used the entire term "positive feedback loop", which is so specific that it loses flexibility -- it can only refer to one thing and conflates the engineering and social sciences meanings. "Positive feedback", when not accompanied by "loop", can be stable. But a loop with positive feedback is generally unstable, because the output feeds the input. Here are pictures of loops with positive, unity and negative feedback: http://i.imgur.com/pxANv.png http://i.imgur.com/pxANv.png http://i.imgur.com/od4CS.png http://i.imgur.com/od4CS.png http://i.imgur.com/vC6yV.png http://i.imgur.com/vC6yV.png Very clear, and all solutions to this classic damped oscillator equation: http://i.imgur.com/h5wKA.png http://i.imgur.com/h5wKA.png Where: a = angular frequency b = damping factor Without any reference to the present exchange, this kind of language is often used by those who want to elevate an argument, make it seem more like physics -- more scientific, more deterministic, but usually in a field that doesn't have those properties.
    - dsirijus 14y ago
      > I posted only because you used the entire term "positive feedback loop", which is so specific that it loses flexibility -- it can only refer to one thing and conflates the engineering and social sciences meanings. There's life to positive feedback loops outside engineering without any conflation. If it's more defined in one field, it doesn't mean it has no value in plenty of others. http://en.wikipedia.org/wiki/Positive_feedback_loop http://en.wikipedia.org/wiki/Positive_feedback_loop
- JoeAltmaier 14y ago
  So, is there some way of controlling the gain? Hellbanning is one way but thats bang-bang (gain is 1 or 0). How about, limiting the visibility of comments to a subset of the community. If they comment/answer, then show the thread to more. If they ignore (its a rant etc) then it quietly sinks into oblivion. Just an idea.
buro9 14y ago
I agree with most of this. In fact I agree so strongly that having witnessed the rise and fall of many a community I've run or been a member of that I think that there is a natural lifecycle of a community. I also believe that whilst you can artificially lengthen the life by controlling inputs, you can also kill a community by the very same methods... possible to do but very difficult in the long term. Instead of trying to control the inputs (censor, dictate), I'm embracing the side effects of a lack of censorship... a shorter life span perhaps, but a community that burns brighter during it's life. Effectively what this means is that I'm building community software that has at the very heart of it the notion that a community will die, and that a successful community will schism during the death phase as members attempt to preserve the bits they love. The idea being that communities rip themselves apart, and that at some point HN will do so too. And when it does it won't be replaced by a single "new HN", but instead by many smaller communities each serving a niche that existed within the larger. Even though those niches appear smaller than the thing they emerged from, they are likely to be larger in volume (active users, posts per day, upvotes) than HN itself was before it. As a metaphor for this, think of a kind of community cellular division: the splitting of a community into smaller microcosms that will eventually grow and then split themselves. In some ways you could argue that StackOverflow actually did this preserving the dictatorship through the Stack Exchange network. Except, I don't really accept that a dictatorship can know when a specific community needs to sub-divide itself to survive. I think that comes from within the community. In the software I'm creating, the very tools to gradually manage the creation of new microcosms is given to the users... in much the way that users can create subreddits and that helps reddit to grow.
ChrisNorstrom 14y ago
I agreed with him up until his solution. Because "controlling input" is also unsustainable when a community grows. Who is going to control input? If you're talking about HN, even Paul Graham tried to control input and it's not working. Or maybe he's not doing it enough. Listen to the Mixergy.com interview with Paul Graham. At exactly 49:40 Paul says, "[hacker news] was originally called startup news but after 6 months we changed it to hacker news cause we got sick of reading about nothing but startup stuff." And yet here we are again, HN is nothing but startup news, I just gave up and joined in, no use swimming against the current. The sad part was that HN was originally a much better nursery to startups because all startups begin their life as a project, a hack, a mod, a solution. HN was all about code back then, which put off a lot of people who couldn't code (myself included) and slowly it turned into more startup news to allow for a larger audience. The question is which came first? Was it necessarily a bad thing? How can PG / Should PG force the community back into more of a hacker culture than a startup culture? I came up with an idea to control who the members of a community are by means of crowd-sourcing (http://www.chrisnorstrom.com/2011/02/invention-creating-and-maintaining-exclusive-communities-through-crowdsourcing/ http://www.chrisnorstrom.com/2011/02/invention-creating-and-...) but still, how can one control the hive-mind of the community to prevent it from wandering away from its original intent? We've seen what happens when one doesn't control the community. Maybe its time to play secret-dictator?
- JumpCrisscross 14y ago
  Federating that power of censorship, e.g. by weighting down-votes by karma (for HN), could help the community scale. Then one just needs to keep an eye on the most powerful censors and ensure they (a) aren't abusing their privilege, and, (b) are effectively policing along the right metrics.
  - lmm 14y ago
    You've just suggested solving the problem of positive feedback loops by adding a massive positive feedback loop in the moderation system. That's not going to work.
    - 14y ago
JumpCrisscross 14y ago
Assuming the conclusion is true, this has interesting implications for our highly connected political media environment, which suffers from similar positive feedback loops with a preference shown to cheaply produced vitriol over informed debate. I'm not suggesting we censor, in fact I take this articulation as a reason to be sceptical of the author's conclusion, but any policy for managing a medium-sized discussion-oriented community should be at least in part scalable to the national stage.
webwanderings 14y ago
What this article is identifying as a problem is actually not a problem but a chaos. Chaos in the community is not a problem in itself but rather a next stage of dialog. You have to cultivate and live through the chaos in order to get to the next level, which is understanding. By censoring, you are pulling your community into a wrong and aimless direction. The bells and whistles like flags, up/down votes, are just distraction between the authentic dialog/substantive community and aimless/useless chatter.

22 points by dsirijus 14y ago

davedx 14y ago

"You need to be a nazi", that's quite a conclusion to make. Any evidence of this actually working?

dsirijus 14y ago
Stackoverflow.
lutusp 14y ago
Steve Jobs, who built what is now the world's most valuable company, at least in part by being intolerable? Parenthetically, I strongly recommend the recent Iverson biography, it's first rate. I worked with Steve Jobs, I couldn't stand him, and I eventually refused to work with Apple. I am sure there are many others who came to the same conclusion. But this didn't stop Apple as a company.
tomflack 14y ago
offtopic.com which had registration closed for a large amount of time in 2002-2003ish? It was still a massive forum but the community kind-of "solidified" and didn't grow larger, for the most part quality was pretty good. It seems to me that once critical mass is reached, closing registration is a good move. It works for other quality-oriented social sites like Oink too.

lutusp 14y ago

It would help if the author actually understood what "positive feedback loop" means in engineering. It's not what he thinks, and social-science terms like "positive feedback" don't translate well in the more rigorous sciences. In strict engineering terms, feedback loops must always have a gain less than unity, and a stable, reliable system normally has negative net feedback. But this doesn't sound "right" to a social scientist, who understands these terms in an entirely different way.

dsirijus 14y ago
I've made two examples to help others discern between positive feedback loop (engineering) and positive feedback (social sciences). Where from have you drawn that conclusion? Genuine interest.
- lutusp 14y ago
  > I've made two examples to help others discern between positive feedback loop (engineering) and positive feedback (social sciences). I posted only because you used the entire term "positive feedback loop", which is so specific that it loses flexibility -- it can only refer to one thing and conflates the engineering and social sciences meanings. "Positive feedback", when not accompanied by "loop", can be stable. But a loop with positive feedback is generally unstable, because the output feeds the input. Here are pictures of loops with positive, unity and negative feedback: http://i.imgur.com/pxANv.png http://i.imgur.com/pxANv.png http://i.imgur.com/od4CS.png http://i.imgur.com/od4CS.png http://i.imgur.com/vC6yV.png http://i.imgur.com/vC6yV.png Very clear, and all solutions to this classic damped oscillator equation: http://i.imgur.com/h5wKA.png http://i.imgur.com/h5wKA.png Where: a = angular frequency b = damping factor Without any reference to the present exchange, this kind of language is often used by those who want to elevate an argument, make it seem more like physics -- more scientific, more deterministic, but usually in a field that doesn't have those properties.
  - dsirijus 14y ago
    > I posted only because you used the entire term "positive feedback loop", which is so specific that it loses flexibility -- it can only refer to one thing and conflates the engineering and social sciences meanings. There's life to positive feedback loops outside engineering without any conflation. If it's more defined in one field, it doesn't mean it has no value in plenty of others. http://en.wikipedia.org/wiki/Positive_feedback_loop http://en.wikipedia.org/wiki/Positive_feedback_loop
JoeAltmaier 14y ago
So, is there some way of controlling the gain? Hellbanning is one way but thats bang-bang (gain is 1 or 0). How about, limiting the visibility of comments to a subset of the community. If they comment/answer, then show the thread to more. If they ignore (its a rant etc) then it quietly sinks into oblivion. Just an idea.

buro9 14y ago

I agree with most of this. In fact I agree so strongly that having witnessed the rise and fall of many a community I've run or been a member of that I think that there is a natural lifecycle of a community. I also believe that whilst you can artificially lengthen the life by controlling inputs, you can also kill a community by the very same methods... possible to do but very difficult in the long term. Instead of trying to control the inputs (censor, dictate), I'm embracing the side effects of a lack of censorship... a shorter life span perhaps, but a community that burns brighter during it's life. Effectively what this means is that I'm building community software that has at the very heart of it the notion that a community will die, and that a successful community will schism during the death phase as members attempt to preserve the bits they love. The idea being that communities rip themselves apart, and that at some point HN will do so too. And when it does it won't be replaced by a single "new HN", but instead by many smaller communities each serving a niche that existed within the larger. Even though those niches appear smaller than the thing they emerged from, they are likely to be larger in volume (active users, posts per day, upvotes) than HN itself was before it. As a metaphor for this, think of a kind of community cellular division: the splitting of a community into smaller microcosms that will eventually grow and then split themselves. In some ways you could argue that StackOverflow actually did this preserving the dictatorship through the Stack Exchange network. Except, I don't really accept that a dictatorship can know when a specific community needs to sub-divide itself to survive. I think that comes from within the community. In the software I'm creating, the very tools to gradually manage the creation of new microcosms is given to the users... in much the way that users can create subreddits and that helps reddit to grow.

ChrisNorstrom 14y ago

I agreed with him up until his solution. Because "controlling input" is also unsustainable when a community grows. Who is going to control input? If you're talking about HN, even Paul Graham tried to control input and it's not working. Or maybe he's not doing it enough. Listen to the Mixergy.com interview with Paul Graham. At exactly 49:40 Paul says, "[hacker news] was originally called startup news but after 6 months we changed it to hacker news cause we got sick of reading about nothing but startup stuff." And yet here we are again, HN is nothing but startup news, I just gave up and joined in, no use swimming against the current. The sad part was that HN was originally a much better nursery to startups because all startups begin their life as a project, a hack, a mod, a solution. HN was all about code back then, which put off a lot of people who couldn't code (myself included) and slowly it turned into more startup news to allow for a larger audience. The question is which came first? Was it necessarily a bad thing? How can PG / Should PG force the community back into more of a hacker culture than a startup culture? I came up with an idea to control who the members of a community are by means of crowd-sourcing (http://www.chrisnorstrom.com/2011/02/invention-creating-and-maintaining-exclusive-communities-through-crowdsourcing/ http://www.chrisnorstrom.com/2011/02/invention-creating-and-...) but still, how can one control the hive-mind of the community to prevent it from wandering away from its original intent? We've seen what happens when one doesn't control the community. Maybe its time to play secret-dictator?

JumpCrisscross 14y ago
Federating that power of censorship, e.g. by weighting down-votes by karma (for HN), could help the community scale. Then one just needs to keep an eye on the most powerful censors and ensure they (a) aren't abusing their privilege, and, (b) are effectively policing along the right metrics.
- lmm 14y ago
  You've just suggested solving the problem of positive feedback loops by adding a massive positive feedback loop in the moderation system. That's not going to work.
  - 14y ago

JumpCrisscross 14y ago

Assuming the conclusion is true, this has interesting implications for our highly connected political media environment, which suffers from similar positive feedback loops with a preference shown to cheaply produced vitriol over informed debate. I'm not suggesting we censor, in fact I take this articulation as a reason to be sceptical of the author's conclusion, but any policy for managing a medium-sized discussion-oriented community should be at least in part scalable to the national stage.

webwanderings 14y ago

What this article is identifying as a problem is actually not a problem but a chaos. Chaos in the community is not a problem in itself but rather a next stage of dialog. You have to cultivate and live through the chaos in order to get to the next level, which is understanding. By censoring, you are pulling your community into a wrong and aimless direction. The bells and whistles like flags, up/down votes, are just distraction between the authentic dialog/substantive community and aimless/useless chatter.