After attending SGFooCamp (photos), I've been meaning to playwith Google'snew Social Graph API a little more.

The SGAPI is a fairly simple and powerful tool:

• it looks for data captured on the web by the Google crawler, namely:
  • XFN links (normal webpage links with rel="something" markers),
  • FOAF information, which I won't go into
• this data defines a graph, with nodes being webpages (which are URLs, and URLs are people) and edges being relationships
• it's not a proprietary Google data store
• there's one RESTful GET call in the API that returns the graph given a starting URL

That sounds kind of abstract, but it's really quite simple. For example, consider all the web services you use:your blogger.com account, your flickr account, your del.icio.us account, your twitter account, and your home page. They're all"you"and should have additional information to indicate that they belong to the same person. XFN defines this asthe rel="me" relationship.

Let's look at a specific example: Mark Kuznicki'saccount. Usingthe magic of thedemo application, we can see that Mark has a number of accounts that are claiming that they're him:

Why? Because Mark entered the URL of his home page and these applications added a link marked with rel="me" on the appropriate A tag. (If you don't believe, follow thelinks and look at the source).

Now, why does it (currently -- 2008-02-20) call these "possible" connections? Simple: because Mark hasn't linked back to these sites on his home page with A linksmarked rel="me".

Why should Mark do this? Simple:

• because if he does this he can have identity consolidation on all his public social networks, that is, by giving only his home page URL we can definitely know all the social networks that Mark belongs to (and not just claim that Mark is a member of)
• and by knowing this, we can start exploring other (web) relationships that mark is involved in, such as rel="friend"
• and if he starts doing this, pretty soon we have a solution to the YANS problem: when Mark joins a site, he need merely specify his home page and his friends are automatically found -- and without having to screw around with the password anti-pattern or manually re-entering all friends.

We're just putting the finishing touches on a new (but critical!) feature for the the BlogMatrix Platform -- "Networks". A Network is somewhat analogous to the Facebook concept of a Network, that is, it's something that you join to identify yourself as belong to something. The key feature of a Network is that it aggregates all posts from all it's members.

• unlike Facebook, you can join as many networks as you'd like
• the ideal use of a Network in the BlogMatrix Platform is to represent a team, either as an official organizational unit or something ad-hoc created for a particular purpose or project
• networks can be created in three different modes:
• • private; that is, only the creator can add accounts to the network. This is useful if you'd like to follow a number of people across an organization. They do not have to consent to membership
  • join with automatic membership. This is pretty obvious.
  • join, but with approval needed. In this mode, you may request to join a network, but it's up to the owner whether you're added or not.