On the Virtues of Explicit Relationship Ontology


Clay Shirky is an extremely smart man. So it’s with trepidation that I say that I think he’s passed over something important about Friendster/Orkut-style sites that try to make relationship explicit.

The relevant Shirky-text is this one, a generally smart deconstruction of new features at Friendster and Orkut. When talking about Orkut’s “please confirm that this person is your friend” dialog box, he lays down the rhetorical hurt:

Please make sure? I have 30 million years of primate social experience wired between my ears — I know instantly whether someone is a friend or not. There are times when I forget people’s names and I still remember whether they are friends or not. What Orkut means, of course, is “Please do the socially awkward thing of explicitly denying a social overture, to give us more accurate information about you.”

But then, he goes on to describe the shortcomings of Orkut’s new 1-5 scale for rating how close a particular friend is:

That wasn�t enough, however, so now they�ve added this linear scale of friendship that would be laughed out of a freshman sociology course, and then they say tell me the data is private. Of course it�s not private � that data isn�t for me, it�s for Orkut. I don�t need it in the first place, because I am a monkey, descended from a long line of such monkeys, whose main talent consists of keeping track of relationships. Measured on the time scale of our social capacity, fire is a recent invention and agriculture is still a novelty.

This is mostly right, I think, but it omits a possibility of great importance. I may be a monkey when it comes to my own relationships, but reading other people’s relationships is not always trivial. If they’re willing to codify these things to some extent—“have met” vs. “friend” vs. “long-time friend”—it helps me greatly, if I’m trying to skim the shape of someone else’s local social net.

One of the virtues of Friendster is that it lets me get a sense of how far away someone is from someone else, socially. Knowing that someone is one link from someone I know—even if the link measure is a crude one—is very different from seeing that there’s no such two-hop connection in the system. Knowing that A and B are friends changes my perception of both of them, especially if I don’t know much else about them.

Of course, this possibility doesn’t come up with Orkut’s friend-ranking: they keep this closeness rating entirely private to you. That makes it useless to me, and useless to Clay. And if it weren’t private, there’d be lots of social tensions of the form Shirky notes for the “is this person your friend?” question: am I forced to overstate my friendships out of kindness, pity, or fear?

But still. I’ve been saying since I got into this social-network thing that I would love to be able to modify my social horizon based on closeness. A close friend of a close friend I’d like to see displayed; an acquaintance of an acquaintance I’d rather not. Orkut kind of understands this in its relationship to groups: if the only thing I have in common with someone is membership in a shared interest group, I don’t care much about their friends. The people I’m linked to directly, though: their friends “matter” to me in the sense that anyone on a social networking site matters.

So yeah. It’s a little unfair to ding Shirky for not pointing out a positive possibility that Orkut doesn’t understand when he points out something else Orkut doesn’t understand. So let’s not think of this as a dinging, so much as pointing out a thesis that he might have embraced, but didn’t.