Pity the humble pie chart. Not only is it the butt of a joke in a humorous new Apple Computer ad (video), but it looks positively cartoonish next to the fascinating graphs researchers are generating to chart linking and interaction on the Internet.
As the ‘Net continues to grow, social scientists are beginning to understand that they can use blogs, posts, and links in cyberspace to develop and test models of how humans interact with each other.
Assistant professor Lada Adamic and her team at the University of Michigan, sampled over 1,000 political blogs just before the presidential election in 2004, then ran the results through an algorithm to produce a graph of the results. In the graph, the blogs are represented as circles, and the links between them are lines. Blogs which were considered to lean left are blue, while right-leaning sites are red. Finally, the connections are color-coded based on whether they are a conservative site linking to a liberal one (purple), liberal to conservative (orange), liberal to liberal (blue), or conservative to conservative (red).
Unsurprisingly, the right-leaners tended to link to blogs with a similar viewpoint, while the same held true for the left. Interestingly, connections were also occasionally made across the political spectrum, although the polarization of the environment is quite clear. Just before the election, the quantity of activity on both sides was similar—a result mirrored in the outcome of the election, which ran close to dead even as far as the popular vote was concerned.
The graph also displays a slightly large amount of interlinking on the conservative side, which Adamic opines may have been the result of a report on George W. Bush’s military service that was released just before the snapshot sampling was taken.
Similarly, Nielsen BuzzMetrics, a company which studies content and traffic on the Internet, has developed similar software to graph the connections between blogs so that commercial clients can study changing trends. At this time, political commentary and technology are the two most-linked and active types of sites. The Nielson software focuses less on content than the type of software used and numbers of links between sites. The company’s goal is to understand patterns of influence so that it can provide better advice to its customers.
It’s kind of a fascinating concept, and one can easily see that analysis of user habits on the Internet may indeed lead to some breakthroughs in understanding human interaction. Although one researcher points out that they are still seeking the "Mercator map" which will provide a truly powerful representation of online interaction, algorithms that produce graphs such as Adamic’s and Nielson’s are good—if narrowly focused—first steps, and they look pretty cool, too.