Twitter.mailana.com started by visualizing your circle of friends, but I want to show networks formed around places or ideas too. Now if you pick a keyword you can see how all the people who've talked about that term are connected. For example, the graph above is a screenshot from the social network that's formed around Gluecon.

So what's this good for? My favorite use is discovery, finding new people in an area I'm interested in. For example I've been learning Flash over the last few months, so I searched for the community defined by 'Actionscript'. Immediately I saw an interesting central cluster:

Looking up some of these folks, @mesh is Mike Chambers, a Flash expert at Adobe, @mdowney is Mike Downey a Flex/AIR expert, @ddura and @leebrimelow are Adobe Flash Evangelists, @flashchemist is Vipin Chandran, an expert Flex developer.

What's important about this list is that you know these are well-connected, active people in the Flash community. In a way their position in the graph is like PageRank, with each conversation a vote for their importance. Most other measures like follower count or update totals can be gamed, but someone's place in the overall network is much tougher to fake. Looking at the whole community around a keyword is a great way of discovering the most interesting people, something no other search technique can match.