Raw

Earlier, when I saw the news about Yahoo! doing a "social job network" - Kickstart - I was going to post saying: This highlights a few things, like how online social network management systems are only really useful when tied to specific domains (possibly including the domain of " hangin' out, dude"). To work with data that crosses domain boundaries (e.g. describing people and their relationships plus jobs) life gets easier when you have a data description language that can be extended into whatever domain you want - the design is just about reflecting what you want to model:

[ a foaf:Person job:experiencedIn [a job:Skill; foaf:name "bricklaying"]; job:isLookingFor [a job:Vacancy; *cut*

That's why I didn't bother posting before - I knew I'd have trouble making the point without first writing a description of e.g. a Person and their capabilities and a Job and its requirements; demonstrating how RDF allowed such information to be merged by simple addition; how SPARQL could be used to pattern-match across the domains to pull out vacancies matching a person.

Are Yahoo! using RDF for Kickstart? Can't remember Mr. Beckett mentioning it, but I wouldn't be at all surprised (either way there's likely an NDA on it right now).

But then I saw another post from Dare on the topic : The Difference between a Social Network Site, a Social Graph Application and a Social OS, in which he provides a few working definitions. Good stuff, though I'd probably go for a slightly different emphasis.

Starting with the fact social graphs are something that can be described without necessarily being technology dependent - social networks are a few orders of magnitude older than the Internet, after all. Then there's how you talk of the network for a particular community. Rather than using the confusingly-overloaded term context, I suspect it might be more useful to talk in terms of views of a single (huge) conceptual graph of all social networks. Two reasons I suspect this - it seems closer to the reality and is a better fit for describing things on the web. Entities and their relationships all identified by URIs.

I'll slap on a caveat - in the current arbitrarily fragmented social graph representation on the Web,  subsets of the data are appearing disconnected by design, in the extreme case in what Dare calls Social Operating Systems: " All your social interactions whether they be hanging out, chatting, playing games, watching movies, listening to music, engaging in private gossip or public conversations occurs within this context.". A technical analogy there would be having to use Microsoft Internet Explorer to explore the Microsoft Internet. In the real-world this would be like saying Catholics can only meet other Catholics in church, and there'd certainly be no after-work beers with colleagues...

Another suspicion of mine is that the only way to get around this is to flip the views inside out, rather than thinking in terms of collections of people and their relationships, make it much more individual-centric. More like the way we, as individuals, carry on our social relationships more generally. This would mean the primary representation (and ownership) of a person's online social graph isn't divided up between the silos of different sites and/or applications, it remains squarely with the person concerned or an agent acting on directly their behalf. The online services (for whatever purpose) just get views of that. This seems like a logical progression from Dare's remark:

Where Facebook has hit the jack pot is that they have built a platform where applications that are compelling once they have a critical mass of users can feed off of  Facebook’s social graph instead of trying to build a user base from scratch. Contrast the struggling iLike website with the hugely successful iLike Facebook application.

There are a huge number of different factors which lead people to use Facebook's window on the Web, but I can't think of any that would mean the Web itself can't act in the role of platform here - and if it can, then it will. It's not unreasonable to see the windows on to the social Web being more determined by what the end user wants rather than by the arbitrary demands of current systems.

Judging by the range of suggestions on the social graph mailing list, we're not short of technology to make this so.

It would be nice to see danbri & libby's early intuition on the significance of social networks on the Web come to fruition. 'Course people have suggested the Facebook kind of thing is just a fad, and that fads come and go. Ok, predictions for next season: data silos and stovepipe applications will be out, distributed data and Semantic Web applications will be in.

@en