There is a trend towards "semantic web" and distributed repository of content in the web. It is a major change from "website-centric" model. In not so remote times users became members of a community around one website which provided them with ability to publish and ability to form a community: that is to comment each-other's content, to share information in a number of ways. Today this model changes. Ability to publish content is now a commodity. Everyone has access to one or another blogging tool and does not need much technical knowledge. How about sharing information and forming a community? Well, it seems web itself is becoming a community with the help of XML (and xml-based protocols).
A concept of "track-back" pioneered by Movabletype is a "glue" that ties together *content* of separate websites. After a user publishes an article on his website his software "pings" (or notifies) a number of websites about the fact that the content has changed. Not only that. It provides link to a chink of XML in the article body that can describe the article: topic, url, author etc. So, based on this notice, the remote website can do the following thing:
- the remote site owner can make a comment on the article and publish it on "his" website, not the original one. His program will inform the website of originating article about the comment. A link is formed on both of the websites, connecting them semantically (on common topic).
- remote website may be a centralised website directory listing all newly updated websites (http://www.weblogs.com/.
- remote website may be a "distributed content repository". It may maintain a listing of topics and with each ping it receives this listing grows. For example, if every member of the Geeklog community would notify "geeklog.net" on each its article about "geeklog themes", then visitors to "geeklog.net" could access a listing containing not only links to websites, but also date of publication and short contents summary.
- Finally there is an important use for Geeklog. That remote server to which an article athor sends notification about his article may actually be not remote, but *his own server*. But the "track-back" script would maintain a collection of links to the website articles that are classified and categorised. This would overcome limitation of Geeklog software that does not allow assigning several topics to one article.
Now from open content to open functionality. Weblogs are opening up allowding all sorts of applications to communicate with them through XML-RPC (xoops for example), through a number of standard APIs. This allows posting from other programs and again many other things with content syndication.
I would think this becomes a very important feature of a blog software, and Geeklog should pay attention. So far we are stuck to "website-centric" model. There is huge potential given Geeklog advanced community features.
Finally, some ideas about implementing this. As an intermediate step anyone can use "stand along trackback script (perl)" from Movabletype. It seems to work with a modified Geeklog template, with standard Geeklog code. I'll try to make a demo in a few days. As to implementing XML-RPC on top of Geeklog, I've came accross an interesting script with which one can call any method of a PHP class through XML-RPC. Maybe we can use it on top of existing Geeklog? GL2 is a great promise and will have many features but I try make the best of what we have today.