How to design better online conversations

Bad discourse isn't a behavior problem, it's a design problem...Amazon is the granddaddy of pooled user feedback—the reviews its customers produce, individually and collectively, provide a huge competitive advantage. Get Satisfaction is designed to help people complain, yet they do it in a way so useful that businesses pay for the service. Stack Overflow helps programmers pool their knowledge; HowardForums does the same for mobile phone users, English Companion for teachers, Ravelry for knitters. So if the conversation on HowardForums is good and the conversation among commenters on CNN.com isn't, does that suggest a design principle beyond "Avoid politics"?

Yes, it does, if you recognize that a rhetorical tragedy of the commons is occurring in many forums. All the participants have an incentive to have good conversations, but each participant also has an incentive to get the most attention. This tension suggests that increases in individual anonymity or in group size also increase the likelihood that someone will start acting like a jerk. Both anonymity and scale reduce what Robert Axelrod calls "the shadow of the future"—the sense that our current actions will have consequences down the road.

That provides some options for turning the jerk dial down. One is to make identity valuable: Stack Overflow won't let new users post until they have exhibited enough other behaviors—visiting the site, responding in helpful ways to other posts—to earn the karma for full participation. Another approach is to partition public platforms, thus reducing the incentive to publicly act out. Twitter does this by segmenting its audience: I can rant all I like, but only to the users I can persuade to follow me. Yet another approach is to enlist users in defensive filtering. Amazon sometimes refuses to publish a post, but most of its policing is done by customers who flag offensive reviews and elevate those they find helpful.

via