How online communities work
This is a collection of writing related to the dynamics of online communities (and a few virtual communities from before home Internet). It tries to explain why online communities end up the way they do, how they form, and how they succeed or fail. I think all of it is interesting, though I don’t agree with everything. I summarize some of the ideas from each link—often not every idea it has. I hope my summaries will encourage you to read them for yourself and serve as a reminder of their content.
Anonymous. Link.
What killed the BBS was leechers, conflict between technical users and those who only wanted to chat, real-name policies, and dilution through a large number of lower-quality boards run by incompetent sysops.
Chip Morningstar, F. Randall Farmer. Link.
Engineers and traditional game designers are tempted to play the role of omniscient central planners when approaching a multi-user virtual environment. Instead, they should observe their (different groups of) players and aid them (in different ways).
See also: the Habitat links collected on Douglas Crockford’s Electric Communities page.
Allucquère Rosanne Stone. Link.
In May 1978 CommuniTree #1, one of first BBSs to run on a microcomputer, went online in the San Francisco Bay Area. Thanks to a 1982 agreement between Apple Computer and the United States government computers with modems found their way to public schools, and shortly teenage boys discovered CommuniTree’s phone number. Within months the unmoderated BBS was destroyed by what a participant called “the consequences of freedom of expression”.
Wendy Grossman. Link.
Part of the reason AOL users came to be despised on Usenet was the poor design of AOL software.
Joel Spolsky. Link.
“Small software implementation details result in big differences in the way the community develops, behaves, and feels.”
- Quoting on USENET makes threads chains of quotes;
- Email notifications mean the user only reads the replies to their own post that arrive in the mailbox;
- Branching discussions are disjointed;
- Sorting threads by the most recent reply leads to certain topics dominating the forum;
- Putting the reply link at the bottom encourages the user to read until the end before replying.
Clay Shirky. Link.
According to Wilfred Bion, all groups talk about sex, find enemies, and venerate a religious icon or set of tenets. Again per Bion, group structure is necessary to defend the group from itself, to keep it from sliding back into these three basic patterns. Communities go through constitutional crises, and the first is the worst, because it needs to establish the rules for making rules. Less is different: small groups can engage in kinds of interaction that large groups can’t.
Shii. Link.
Registration encourages those with nothing better to do to participate and discourages busy experts. It leads to cliques and discussion of people over ideas.
Wikipedia Meta-Wiki. Link.
Polling discourages consensus, encourages false dichotomy and groupthink within factions, isn’t fair, and is misleading. That said, polls can be useful to gauge opinion.
Paul Graham. Link.
Bad comments are mean or stupid; meanness is easier to distinguish and manage than stupidity. Long but mistaken arguments are rare. The most dangerous form of a stupid comment is the short dumb joke.
anaesthetica. Link.
Traditional forum moderation doesn’t work. Scaling can’t be avoided. Instead, a different moderation strategy should be pursued:
- Users must be anonymous;
- Participation must be easy;
- Moderation should focus not on the quality of invidual comments but on the quality of conversations;
- Forums should distingish between origina content, “link-n-blurb” content, and personal content;
- What is shown on the front page should be determined by the quality of conversation.
Eliezer Yudkowsky. Link.
Refusing to ban bad users due to fear of censorship leads to a decline in the quality of a community.
Jeff Atwood. Link.
Meta-discussion crowds out substantial discussion, killing communities.
Alan Crowe. Link.
Blacklisting spam, trolling, and mediocre posts fails because it is an endless game of whack-a-mole. What may work instead is hierarchical whitelists: a user whitelists people, then their client looks at whom those people whitelist, then at whom those whitelist in turn, etc. to select messages.
Robert Scoble. Link.
Blogs become more interesting over time; forums, less.
apenwarr. Link.
Wikipedia changed the idea of wikis: they are written as documentation for outsiders, not as a means of communication within a community.
Xianhang Zhang referencing “Evaporative Cooling of Group Beliefs”, Eliezer Yudkowsky2007. Link.
High-value contributors leave a community when they see it no longer serves their needs. This drives the quality down enough to push the next most valuable layer of contributors to leave.
Jake Seliger. Link.
People who become and remain unpaid moderators are often people with axes to grind, people with no sense of perspective, petty tyrants, etc.
David Chapman. Link.
A different type of person creates subcultures, joins them later on, and profits from them.
allgebrah. Link.
A response to Chapman’s essay. Some subcultures are perennial and have generations, not a one-shot lifecycle.
Siderea. Link.
“If you tell people “the only way to contact me is to break a rule” you will only be contacted by rule-breakers.”
Hotel Concierge. My notes.
Aella. Link.
Communities founded by people who like a specific culture (“possums”) are joined by others who like all cultures (“otters”). Otters may even invite possum friends with opposite values. The community splits: possums notice the culture drift and try to write rules and moderate harder, otters found an otter counterpart.
Gwern Branwen. Link.
Holy wars occur over technologies with network effects.
T. Greer. Link.
You talk to someone who doesn’t share your values differently from someone who does. On Twitter there are no boundaries between communities to enable this. Everyone is mashed together, and in-group messages get hate-retweeted, increasing animosity between groups.
Hacker News. Link.
The top comment by user nostrademons says they splinter off from existing communities.
Freddie deBoer. Link.
When writing film reviews, users of the cinephile social network Letterboxd engage in several types of insincere and netiquette-breaking behavior to stand out and gain “likes”, visibility, and social status. Those include:
- Pretending to go against the crowd;
- Picking easy targets to disagree with and easy political points;
- Writing one-liner reviews that try and fail to be pithy;
- Reviewing movies you probably haven’t seen to be the first;
- Accusing other reviewers of not actually liking more experimental movies.