The Heisenberg uncertainty of social networks

One of my New Year's resolutions is to open up this weblog to a wider audience. So on first mention of something obscurely technical, I'll try to define it. Today's obscurely technical topic: FOAF.


The FOAF (friend-of-a-friend) project, according to its homepage, "is about creating a Web of machine-readable homepages describing people, the links between them and the things they create and do."

Here is a trivial but valid FOAF file:

<?xml version="1.0" encoding="utf-8"?>
<rdf:RDF xmlns:rdf=""
<foaf:weblog rdf:resource="" />
<foaf:name>Jon Udell</foaf:name>

My FOAF <Person> record could say more about me: geographic location, interests, projects, and so on. But its main purpose is to list a bunch of other <Person> records -- those of my friends and associates -- and thereby create a web that can be traversed by software.

I don't currently maintain a FOAF file, for reasons that David Weinberger solidly nails in this posting:

It's like thinking that the invitation list for your wedding actually reflects your circle of friends and relatives. No, you had to invite Barry-the-Boozer because he's your cousin and you couldn't invite Marsha because then you'd have to invite her husband Larry-the-Ass-Grabber and her daughter Erin-the-Snot-Flinger.

If you want to get at the real social networks, you're going to have to figure them out from the paths that actual feet have worn into the actual social carpet. [David Weinberger: Corante: Many-to-Many]
Exactly right. In a connected world we have all sorts of ways to measure relationships, but that doesn't mean we can or should try to declare them. Bill de hÓra says why not:
Hand crafted logical ontologies are not sufficient precisely because they want to be certain. They don't drift with your interests over time, they're rigid, they're deterministic, they can only see around so many corners. In short they age badly, and they evolve badly. [Bill de hÓra]

I realized long ago, for example, that maintaining a blogroll by hand was going to be a losing proposition, and switched to a system that simply echoes the list of feeds to which I'm currently subscribed. Commenting on David Weinberger's posting, Julian Bond echoes that idea:

I think that systems that exploit the other activities of people to build maps of social networks will always be more accurate and have less inherent bullshit than system where the participants are consciously building networks. This may mean that systems like Spoke and Plaxo that derive metadata from email use have more relevant data than systems like Tribes, Friendster and Linkedin. [Julian Bond]

The impetus for all this recent discussion was, in part, the recent announcement by Six Apart that its TypePad blogging service has been automatically generating FOAF files from users' blogrolls. The announcement also mentions that a new service, Plink, is available for browsing the newly-enlarged web of <Person> records. As Matt Jones found out, you don't need to maintain a FOAF file to be Plink'd, you just need to be mentioned in somebody else's FOAF file.

Given the Web and the many agents dedicated to exploring its interconnectedness -- Google, blog searchers and mappers -- this approach seems to me at best redundant. So when LinkedIn asked me to explicitly define my relationship with someone, by choosing from a list of options, I declined. A Google query mentioning our two names would have been the best way to define our past, current, and future relationship.

I'm sympathetic to the FOAF cause, but I hope for a more general approach. The reality is that every document published to the Web can help to define a relationship -- by linking to, quoting from, or more subtly supporting or refuting another document. Of these actions, linking is the only one that's always unambiguously machine-readable. We can do much better. Six Apart has the right idea in leveraging TypePad's blogroll editor to build FOAF files under the covers. We need to extend that to all of our content production: email, blogs, everything.

To that end, I'd like to see Don Box's Word-to-XHTML project (which may or may not supplant an earlier but not very useful translator posted to MSDN) turn into more than a solo skunkworks effort.

Former URL: