Slashdot

These are some elements that I hypothesize create good conversation spaces. I don't have evidence for these yet, and there's not much evidence in the literature, but here goes.

• Access to a home directory, especially when there is no F2F component of the group. This presumes something other than totally anonymous interactions. Http://www.blackplanet.com is a good example.
• Group size: larger groups offer different challenges and benefits. Social loafing, trolling, flaming, and other undesirable behaviors might increase, but so does the ability to use tools like reputation systems, collaborative filtering and so forth.
• Reputation systems allow for some signals of trust, especially in those social spaces where there are no competing social systems (cf collaboratories) or where there are large enough numbers of interactions to provide decent feedback. In general, the ability to provide direct feedback between users is essential.
• Collaborative filtering can help when there is a large amount of traffic to reduce the load on the user.
• Barriers to entry: Various schemes for barriers to entry have been attempted. Disclosure of the offline identity is often used in professionally oriented spaces. Other spaces make the user contribute a certain amount of content before becoming full participants.
• Barriers to exit: Even though the conventional wisdom is that it is impossible to create barriers to exit in online conversation spaces, there are certain methods that do encourage such barriers. For instance, digests of content allow for peripheral participation, allowing users to vascillate between different levels of participation and reducing the cost of participation. Setting higher barriers to entry will also engender to some extent a barrier to exit.
• Degrees of freedom: allowing users to repurpose the technology of the space for them to adapt it to their social context. An example is allowing people to form groups in different ways on EverQuest.
• Mission statements: short, easily repeatable catch phrases that define the context of participation helps create a sense of culture, and puts boundaries around the goals of the conversation. "Node for the ages" in http://everything2.com is a good example.
• Persistent Identity: tie rewards to specific pseudonyms, so that the cost of misbehaving with those pseudonyms gets higher. As Weick says "When people feel their actions are chosen, visible and irrevocable, they feel responsible for them".
• Ability to silence: Sometimes participants need to be censured. On E2, the gods use the Everything Death Borg to deny access to the chat area. On another site, there was a mechanism whereby any registered member could "silence" another once per day. This is often a way to moderate and still distribute the responsibility of moderating.

Several of these mechanisms seem to tied to interaction over time. I think time and duration are parameters that few people have dealt with reasonably while looking at either online communities or online conversation spaces. We'll have to do some more work in that area.