Mining the XFN Network

Rubhub displays XFN relationships (which is not to be confused with FOAF relationships). This is an interesting concept and semantic experiment, although I might disagree with its choice of options. Unfortunately the options are limited. It allows several degrees of romantic relationships (muse crush date sweetheart ) and family relationships (child parent sibling spouse). But it’s unlikely that males/females will be so eager to give their connection/relationship a name when the declaration may actually defeat any possibility of any relationship coming about. So if I indicate that my XFN relationship to Matt is Crush, I doubt Matt will be flattered or even interested 🙂 If an ex-girlfriend knew that I had specified my relationship as being sweetheart/crush via some weblog, she would kill me!

Second, curiously, there is no social relationship to describe a reference to self (because I write/edit 8 different weblogs, that is a real issue). XFN was never really intended to describe weblog relationships, but it was used for WordPress weblogs as a matter of convenience (because developer Matt Mullenweg was working on both projects). But its usefulness wears thin as websites/weblogs start being associated with several people at the same time. Linking to a website is not the same thing as linking to an individual, although for the short term, this synecdoche might be useful for simplicity.

What we need is NOT a language for describing the type of relationship (or intensity of feelings) that two people have. Instead we need to know how these people know each other. Do they meet through an organization? At a conference? At a church? Or school? Is this an email only relationship? Did they meet through an online dating service? A mailing list? A sibling? We need a “is a fan of” relationship designation. Sometimes you know about a person’s only through his website, and yet that is a pretty substantial (albeit assymetrical) relationship. In 2002 at South by Southwest I finally met Cameron Barrett , a distinguished blogger whose blog I had been following for almost a year. I still hardly know him, but I saw his photos of Siberia, know about his getting fired, and even browsed through his DVD library . We chatted for a bit, (and to my knowledge he has never browsed my own weblog–that bastard!), but it was cordial enough–Cam was probably used to being accosted by strangers. The relationship between Cam and me could only be described “am a fan of” “follow the online diary of”.

Also, there needs to be some clarification about geographic contiguity. What do we mean by “neighbors?” Same street? Same subdivision? Same city? On the other hand, I really don’t need to know much about family relationships (if only because none of my family blogs or belongs to friendster).

One interesting part of the XFN Description page is its recognition that assymetrical relationships exist in the real world. Marty might regard Jack as a friend, while Jack might simply regard Marty as an acquaintance. Indeed, judging from RubHub, practical use of XFN has changed the meaning of the term “Muse.” Now it indicates an assymetrical relationship where one person regards the other person as talented/brilliant/creative, but doesn’t yet have contact with.

It goes without saying that the social network described by XFN may not coincide with reality. Judging from Rubhub’s statistics, it seems that Web Designers are the most socially integrated people in society (when it may mean simply that they are using XFN-equipped software like WordPress or they are more likely to tag their relationships). One has only to look at friendster to see how the number of friends doesn’t necessarily indicate strength of relationship ties or even actual contact. My first friendster friend was someone I’d never even met (although interestingly enough, we’ve become good friends and I plan to visit her soon). There are people who have 50, 100 or 500 friends on friendster. It becomes almost an Easter Egg gathering contest. (Admittedly, social network has not gone mainstream yet).

Despite these flaws/idiosyncrasies/limitations, XFN is still an interesting and useful dialect. XML dialects tend to coverge, and bad/incomplete/ tagging still yields useful semantic information. Eventually some genius will figure out how to map XFN and FOAF to whatever social network dialect happens to be the most popular. XML dialects don’t need to be perfect to be useful.







3 responses to “Mining the XFN Network”

  1. Hannah Avatar

    This really is a good post. And very interesting–not just in terms of technology, but…I suppose…in terms of people who are quick to pick up and use terms like “community” and “culture” to talk about the Internet.

    What would an internet kinship chart look like? Not just what kinds of labels we’d use to identify relationships–but what kinds of boundaries/passages would emerge from that? You know. I can’t marry my maternal cross-cousin because of a social taboo. I can’t link to a conservative blog because I’m a liberal. The very process of denoting familial/platonic relationships from romantic/inspirational relationships starts to do that.

    Seriously. This is fascinating. Thanks for the great post.


  2. Robert N Avatar
    Robert N

    what’s interesting about the Internet is how easy it is to map and quantify these kinds of relationships. For example, on, they are able to guess fairly accurately what kinds of books and films would interest me. I’d like to maintain that I’m unpredictable, but maybe I’m easier to predict than I would like to think.

  3. Robert Nagle Avatar
    Robert Nagle

    At the risk of revealing my utter stupidity, I ended up deleting Cameron’s post/response to this topic.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.