February 10, 2006
CoComment: Automating distributed conversations?
Organising the network of posts, resources, and comments into conversations is a fairly tricky business, especially as the spam-bots have made systems like trackback increasingly fraught, and even commenting via web APIs like Atom is discouraged in the absence of a consistent approach to self-asserted identity. Into this fray has stepped CoComment, a service that tries to link together conversations.
Unfortunately, the service has a number of limitations, not least of which is that there seems to be an element of network segmentation:
"The walled garden of conversation. coComment currently tracks comments left in response to yours, but so far only those left by other coComment users. Who cares what newbies have to say anyway? No, to be fair I'm sure this is a technical matter - but it sure would be nice to see it overcome." (Marshall Kirkpatrick, Social Software Weblog)
Ultimately any such system is less about automating distributed conversation - people have a stake in determining contextual relevance - than about filtering based on identity. What is really needed to bring conversations together is a consistent model for identifying the commenter. If that mechanism existed, then I wouldn't need CoComment to scrape my comments from other people's sites - I can make a response within my own environment, and post it direct to the other person's site while retaining my own copy. My authoring process also is my repository process - isn't that easier than requiring an intermediary to capture my comments in transit?
What's preventing such a simple measure is the need to stop comment spam, which requires better identity assertions. CoComment, while interesting, is a stopgap while we tackle the real issue of who (or rather, what) we let into the conversation, at which point we can tackle the tracking and management issues of distributed conversations at their source.
Unfortunately, identity schemes, both self-asserted and mediated, are so thick on the ground right now its hard to pick a winner. Clearly Shibboleth/SAML is the dominating force in higher education, Liberty is moving the enterprise crowd, InfoCard is going to be big in Windows-land, and in the wider web we now have SXIP, OpenID, mIDm, Identity Commons, LID, YADIS...