Been fighting most of the day with lib problems, compile failures, and segfaults. Blechie. In interesting tech, Stewart writes an interesting pieces about social software and outlines his views of five characteristics:
- Identity
- Presence
- Relationships
- Conversations
- Groups
One characteristic also worth bringing up is History. This of course is a (perhaps the only?) characteristic of web communities that few other systems have. Specifically, a shared, referenced, indexable/searchable history. I’ve spoken before about the temporal mapping that takes place (this is just another way of describing it).
This is as good of a time for me to dig back and make some H20 comments. The Rotisserie approach certainly is interesting, and is ideal for the learning environment and encouraging active participation. In fact, this would probably be a better format for most online ‘debates’, but I’m not sure of it’s effectiveness in general conversation.
As far as filtering messages (finding worthwhile discussion) in general purpose forums/communities, I think that the best approach may be a combination of collaborative/predictive filtering via trust metric, prioritization via activity/popularity algorithms, and community moderation, preferably along more than one attribute axis. The secret to making it all work of course is to make it minimally intrusive upon workflow, ideally something that you’d do anyway (ie, how long you hover over a message contributes to a weighted view rank; remote scripting rating systems, key events).
Also see: W3C Annotea Project, Crit, XLink, [add more about typed links for semantic systems]