I am playing with Google's new Social Graph API. For those unfamiliar with the term, a social graph is a map of the connections between people—it's basically what one gets by combining two or more individuals' contact lists, and drawing lines between who knows who.

Social networking sites, like Facebook or LinkedIn, are designed to let individuals create and explore contact lists and social graphs. But, these sites have specific and commercial motives and, one way or another, attempt to "own" these social connections.

As an alternative, many people have been working towards ways that one can create and explore their social network on their own websites—under their own independent control and ownership. So, as freely as one can have a website or a blog, one might have their own mini-Facebook.

Google's new service fits into this more distributed approach to social networks by indexing social information on any website, and providing a convenient way to aggregate that info into a consolidated view (e.g., my friends).

In order to work with this new service, one need only post a public contact list formatted in either the XFN and/or FOAF formats. These are (semantic) web data formats that have been around for many years.

I've done a good deal of work with XFN and FOAF, and I even have an old contact list of sorts published in both formats. You can see how Google sees My Connections according to the contact list I posted several years ago on my old iCite.net blog.

It's interesting to think about these specific formats becoming more widely used. FOAF is really generic in that it, in itself, only really represents who knows who. But, FOAF can be extended through something called the Relationship Vocabulary, that makes it more on par with the terms in XFN.

XFN represents a much more robust, but ad hoc, set of relationship types, e.g., friend, spouse, neighbor, colleague, etc. And, it's interesting, in looking at my own contact list, seeing how XFN isn't (in my opinion) consistently designed in relationship to time.

For example, in XFN, you can have met someone. But, someone can also be a neighbor. Once you meet someone, they are always someone you've met. But, thinking of someone as a neighbor might mean that they are currently one's neighbor or (possibly) that they have, at one point, been one's neighbor.

So, given a lot of info of this type, suddenly useful in a public context, the lack of clarity around these concepts (in XFN, but in FOAF as well) might be the basis of some strange assumptions made by applications and people. But, I guess we'll see what comes of this—hopefully stuff that's interesting in a good way!