(Minor backstory: Blog-buddy Marc 'Armed Liberal' Danziger posted at Winds of Change on the 'social graph' and Google kerfuffle and then asked for comment behind the scenes. Which I provided in a rushed e-mail, and then let the topic simmer through the weekend, distilling several fractions of bogosity from the raw material on offer. Which I now offer, from most trivial to consequential.)
'Social graph' is an unneeded and misleading coinage. Dave Winer lays this out well, in his inimitable style. A graph as referenced is an abstract math and computer science concept, a class of representations for - among other things - networks. Real people schmoozing at parties - real or virtual - are networking, not 'graphing' (that's something you try to beat out of Excel). For the sophisticates, there are well established fields like social network analysis and the virtual business category social network service which the incumbent term invokes. For either, 'social graph' is needless use of obscure terminology, perhaps in service of trumping up a category with no good reason to exist.
The Google story is highly suspect. So what actually happened here? A deliberate workaround on NDAs signed by attendees at a Google technology briefing, with vague mentions of various internal services and external partners that might or not be implicated in an attempt at more 'openness' in the SNS space, and some stringing together of names of folks who might or might not be involved in the project(s). If that's something you'll base a business judgement on, I suggest a job at CBS, not an SNS startup.
Google's immediate impact in SNS is limited. I recommend Danny Sullivan's well-stated comment to the above story. It's not like Orkut and iGoogle have been burning up the world of late. If Google gets too free and loose with their data, most folks won't even have to opt-out, because they've never opted-in. Google's incentive is to come up with something that will attract new community members, not keep them away.
...unless Google's making an evil-stupid move. The Googlers tend to be chaotic but clueful, so I doubt this, but it could happen. The evil-stupid thing would be to reverse engineer social networks out of people's GMail content and use them for marketing purposes. I'm talking address books and message headers, even content. This is known to be possible. It's also so far away from the originally intended use of that information that it's almost certain to create instant and explosive blow back.
Open social network representation is (partly) counterproductive. That includes both SixApart and associated efforts in that direction. If you're concerned about privacy - and SixApart is at least using that rhetoric - then why the heck do you want to make it easier for representations of social networks to be mobilized and shared? Wouldn't you like the shadow of the past to die along with a SNS that might be abusing your confidence? Oh, that's right, we're all (users and SNS providers) going to sign up for OpenID before that happens...
But it doesn't matter, because it's DOA anyway. For the non-technical: Horse, barn. Cat, bag. For the others: FOAF, client-side certs, digital wallets. What do these have in common, other than being dead meat? They all are/were bets that J. Random User wants and will use more secure and capable privacy and portability mechanisms if they are put into their hands. J. User continues to sign up for GMail, Facebook, etc. without any of these mechanisms and complexities. SixApart has one platform, LiveJournal, that is in the SNS space by a stretch and is a fading star. I love you guys (see the URL for this blog), but this is an overreach. Wake me up when someone with a user base and growth curve buys in. (Oh, yeah - SNS' are hit driven businesses, have you noticed?)
There is no singular social network (or graph). SixApart says the right words: "Everyone has many social graphs, and they shouldn't always be connected">. danah boyd nailed this one some time ago (PDF warning). Social networks aren't fragmented only because of technology issues, but deliberately so, as a means of controlling one's own social presentation, hiding and controlling information, even constructing identity. Before working on collapsing all those social affordances into one data structure, you might consider whether that's something the individual will appreciate. And even think about whether then attempting to market based on that unified view is going to be appreciated, or resented.
And you'll never get it out of people anyway.. But let's suppose that in this new transparent society we won't care if our significant others, our minister or rabbi, and some banker all see the same social network. Still won't work. Listen to the Old Fart for a moment here. Back in the day - before there was a real Web - academics sat around talking about the rhetoric of hypertext and how we'd all encode our links. Then we built the real thing - and it didn't happen. Take a look around you - how many typed links or even consistently tagged pages do you see? Real people don't do that.
Now you'll likely admit that marking up Web links is not an emotional topic or something discussed at parties for most of us (unless you're working at Google). Interpersonal relationships are both - see 'gossip'. What the heck makes anyone think we'll get all the nuance of social networks externalized - truthfully and completely and consistently, mind you - if people won't do it for something simple and unloaded like the Web? Doesn't even pass the grin test. So, we're right back into having to reverse-engineer semantics from minimal and partial representations (a la PageRank) but in an incredibly loaded setting. Plenty of step-on-your-crank potential for both incumbent and wannabe SNS'. With the properly perverse attitude, it should be entertaining to watch. Pass the popcorn?