Yesterday, Tom Coates posted a piece entitled " Trackback is dead. Are Comments dead too?" His argument is that trackback spam has put an end to an interesting attempt to knit together posts between different blogs, that we should allow time for mourning, but we should also begin looking for alternatives. The Six Apart Pronet list has carried a number of posts from people agreeing with his analysis.
Trackback never really took off outside of techie circles. The lack of support for it in blogger and the lack of education of new bloggers as to its advantages ensured that. For those of us who are interested in the technical aspects of blogging, and in the potential it offers to change the way we have conversations, it was a great starting point, but it never hit the primetime.
I wonder whether the lack of visual integration based on trackbacks may also have held it back. Reading comments on blogs is a linear process that requires little mental adjustment, but since trackback content rarely extends beyond excerpts, truly following a conversation requires a deep adoption of hypertext reading skills that may not have yet reached critical mass.
The enthusiasm with which people greet technorati’s cosmos when they see it, and the number of bloggers who don’t use trackbacks but who display a technorati profile on their site may be evidence that simpler tools encourage participation, but that could equally just be the ego-stroking possibilities that blog rankings allow. Some people have suggested making use of tags to follow conversations, but I fear that the imprecision would not make for a smooth flow, and the namespace could quickly become very cluttered.
On the technical side, with the Atom Syndication Format approaching stability I wonder whether this may be a good time to consider the possibilities the use of Atom as both Syndication Format and Publishing Protocol offer for a trackback replacement. A couple of years ago, Tim Appnel posted some thoughts on the next generation of trackback and a number of them could be combined with an extended Atom API.
A new form trackback could consist of an atom entry, posted to a given URI, with the recipient following the conventions of the Atom Protocol (and HTTP) to respond. The receiving interface, having access to the full entry, could apply logic to the entry to ensure it wasn’t spam.
The receiving URI could then build up a representation of the conversation as a feed, perhaps with extenstions to represent the flow of the conversation, and tools could display that linearly, threaded, or however seemed most appropriate. Smart clients could spider out from that representation to see if there were other elements that had been missed, aided by the universal uniqueness of entry IDs.
Offering clear ‘conversation views’ it may be easier to evangelise on behalf of this new tool, and by building on top of an existing API the overheads for tool support would be reduced.
What is most clearly missing from this approach is a way to reduce the spam load. More data allows for more sophisticated filtering, but as noted above it increases processing overheads. It may be that the WSSE authentication used within Atom could be used, along with a key included in the entry being commented on, to require the client sending the ’trackback’ to identify itself, but I suspect this one is going to require further thought.