Cohesion
By Sam Ruby, January 17, 2003.
This document describes some of the currently popular techniques used to promote collaboration and group forming via weblogs. If you are not familiar with the concepts of weblogs, RSS, HTML, and referers, it might be helpful to review these concepts as described in Manufactured Serendipity.
co·he·sion (k
-h
zh
n)
- The act, process, or condition of cohering: exhibited strong cohesion in the family unit.
- Physics. The intermolecular attraction by which the elements of a body are held together.
- Botany. The congenital union of parts of the same kind, such as a calyx of five united sepals.
I see three basic views of trackback / pingback / automatic linkbacks
- Remote comments
- Two directional hypertext links
- Ridiculously easy group forming
These are each explored in turn below.
This is the concept I was able to grasp first. I have a comment enabled weblog. Comments are entered using a small textarea and formatted using a restrictive set of rules. There's no up front authentication. An alternative would be that you enter your comments on your weblog, with whatever tool makes you comfortable and using as rich of formatting as you desire. Some small excerpt is then placed on the remote weblog containing the essence of your comment, and a pointer to where more can be found. This has additional benefit in that it provides some authentication that you, in fact, did author the comment. Because such remote comments promotes cross linkage, they also tend to promote exploration and group forming.
Two directional hypertext links 
Hypertext links on the web are one directional. To achieve two way communication effectively, you need to be to traverse them backwards. There are two basic approaches taken to date: push and pull. There are two popular push techniques today: trackback and pingback. Both require the content authoring software to actively notify the remote weblog of the link. Typically, the receiving end does some validation by actually fetching the page and ensuring that a link is present.
There are two basic pull approaches that I am aware of. Some use JavaScript on the client and invisible images to transmit referer information via a URL. Others scan the referer logs on the server. Some simply present this information as is, some validate it, and some attempt to correlate it to specific days or even specific weblog posts. A few go a step further and actually try to extract the comment, either from the html page or by finding the associated RSS and parsing the content:encoded and/or description found there.
Some attempts to visualize this have tried to impose a hierarchical organization on what essentially is bidirectional, quite possibly cyclic, graph. To date, it appears that this data is too dynamic and distributed to be displayed as such in a static manner except possibly when limited to a single level of depth. A single level of depth also helps the dynamic implementations as it increases responsiveness and doesn't attempt to impose a hierarchy.
Ridiculously Easy Group Forming
This appears to have been one of the primary motivations of TrackBack, and the one most difficult to explain. The best way I have found is to explain it by example: you are at an event (say, a conference) with a bunch of people you don't know. Some of you will capture your thoughts in a weblog, and it may interest for people (whether they attended or not) to be able to find all of these thoughts. Google often helps (after a few day time delay). Some people often collect these links and manually update and publish them.
This can be automated using the techniques described above. Somebody posts a weblog entry. This is either orchestrated in advance or simply emerges. Others link to it. The list of people who link to the post is published, forming something qualitatively different than a simple set of comments. What is achieved is essentially a persistent but time sensitive directory of a given topic.
To me, Sébastien Paquet captured it by modifying Reed's law thus:
The value of a group-forming network increases exponentially with the number of people in the network, and in inverse proportion to the effort required to start a group.
Wiki, InstantOutliner, MultiAuthorWeblog, Groove.
Hypertext links within the context of a weblog adds a bit of "clumpiness" or texture to the underlying fabric which is the world wide web. Some techniques are available today to accelerate this process. This is an area ripe for further research and experimentation.