The three things are:

• scalability;
• opportunities for participant initiative;
• feedback (responsiveness).

These things interact, sometimes in strange ways, and there are tradeoffs. I'll start with scalability. The key here is removing the human component of administration, and that's every type of administration, whether it be setting up accounts, installing upgrades, providing help to users who can't figure out how to do things. The usual solution is to outsource administration by using a web service such as blogger, myst-technology, or typepad to host individual blogs.

Opportunities for participant initiative means that people can raise issues on their own and start their own threads. This winds up being hard for many people in leadership roles to accept because it implies ceding control, sometimes to people in extreme followership roles. Allowing participant initiative can also work against scalability because there is inevitably some clean-up that an administrator has to do when people make mistakes, someone posts something grossly innapropriate, or SPAM creeps in.

However, removing initiative makes the channel one-way and destroys a kind of social scalability that occurs when more than one person is trying their best to contribute good ideas. When this type of initiative emerges, a leader can effectively cede some of the less value-adding aspects of their jobs (e.g., answering basic questions) and focus more on figuring out newer, better ways to manage the group.

Third is feedback. People need to feel that when they contribute, it is having an effect. Again, this can work against scalability because people are usually looking for feedback from someone in a position of authority. That person can spread their attention only so far. Here again, allowing individual initiative can help by allowing alternative sources of authority to rise up and take some of the burden. An advantage to this approach is that these people may be seen as more credible, particularly in times of crisis.

I'll have more to say on these topics in the future. In particular, I left some of the specifics vague. That's because many of them depend on the context: type of software (blog, wiki), type of organization, purpose of community, etc.