I've updated the applet to only show 2-degrees by default, so it doesn't go nuts with a gazillion nodes flying around when it first loads up. It's also color-coded (blue=boys, pink=girls).

I'm still trying to get the image thumbnails to show up correctly. Java applets have a security constraint that keeps them from loading any resource (including .jpg files) from a site other than the site the applet was loaded from. Since the images are hosted on photos.friendster.com, and the applet is loaded from banksean.com, I had to hack up a proxy in perl to route the .jpg requests through banksean.com. The applet isn't fooled quite yet though. For some reason it still won't display the images. To see it try anyways, let your mouse pointer hover over a node and wait for a JFC internal window to pop up like a tool tip near by.

On the scraping side of things, I realized I wasn't capturing the entire network for users who had more than 40 friends. Friendster paginates the friends list in groups of 40, so there is some data missing.

On the stats side of things, I've come up with one called Mackness. It's the number of single people of the opposite sex in your immediate network, if you are single. The reports I've run place me way down at the bottom. And if I run the opposite query, the number of single people of the same sex in your immediate network, I show up first :/

When I fix the pagination problem with the scraper, I'll start posting stats maybe on a weekly basis. I'm also considering making this into a bot. Like an automated Friendster user. You can add it to your friends list (it auto-accepts all requests) and it will start gathering stats for you. There are probably a lot more interesting things that could be done with a FriendsterBot, I bet.

Two things intrigue me about this project. First off, I'll bet it will produce more valuable information to its constituents than TIA will. (TIA seems kind of far fetched to me. I think it's more likely to be the feeding trough for IT shops who get government contracts than the ruthlessly efficient totalitarian control structure some are led to believe.) It's far cheaper and solicits contributions from the net at large, meaning it can grow organically.

The second thing that interests me about it is that, like social network mappers*, the GIA looks like a proto-semantic web site. I'm not sure if they're using topic maps or RDF, but they should be. From the Data Model page: "All of the information submitted falls into two categories: An entity or a link." Since they support data versioning in the workflow description, they must be using some kind of parameterized "link" concept.

* Social network mappers as Semantic Web: Notice profiles for inanimate objects or abstract concepts on Friendster? Given the limited semantics of friendster, users post their ideas of relationships between these things. Kind of like a concurrently edited topic map. "RIAA" is friends with "Pure Evil" and so on. All that was needed was a constrained universe of {profile, friendship} to focus its development. And of course, FOAF is intended to be both a social network map and an example of semantic web development. I just think that these accidental developments like spontaneous topic maps on friendster are making more progress towards that end than FOAF is.