Shirky: A Group Is Its Own Worst Enemy

Clay Shirky writes about the Internet, as much as a social and cultural phenomenon as a technical one. His recent article A Group Is Its Own Worst Enemy describes what happens when “social” software (Usenet, mail, instant messaging, weblogs, etc.) is used to support growing, long-lived user groups. In particular, he discusses patterns of failure in these groups and to a lesser extent what can be done to avoid group failures.

It’s far from a recent observation that members of a group can cause that group to fail because their motivations aren’t the same as those of the group as a whole: I think I personally first saw this kind of effect on Usenet in the early 80s. In fact, this is one of Shirky’s main points: people persist in setting up groups mediated by software without learning from the failures (and successes) of previous attempts. Then, of course, they are surprised by the results and write another paper which is in its turn ignored by the next generation of designers…

One of the examples discussed is a well-intentioned but unstructured 1970s Californian bulletin-board system subverted by an inflow of adolescent schoolboys. Some of the most interesting material described comes from pre-Internet psychological work; the point being that failures happen because the motivations of the members of a group can cause problems when they conflict with the purpose of the group, and that this is a human social issue rather than a technological one.

On-line (or partially on-line) social structures are worth trying to protect from the kinds of failure pattern Shirky describes. Unsurprisingly, he can’t give us any instant cures for these ills; it may be that there are none. Instead, he supplies some necessary but not sufficient conditions for a functional long-term community:

• some forms of identity and thus by implication reputation are important
• barriers to participation are required
• a way to spare the group from scale is essential

This is a long article, but it’s well worth the time. One thing it could have done with was a link to the LambdaMOO messages he references, so here is one.

Found through an article by Daniel H. Steinberg at java.net. That site has since closed down.

[Latest updated 2019-12-15 to fix some more disappearing links.]