Tuesday, May 1, 2012

Visualizing the liberal-conservative gap

Network visualization and analysis is a growing research area. Popular accounts of its rise that I've enjoyed quite a lot are Manuel Lima's gorgeously illustrated Visual Complexity: Mapping Patterns of Information, and Christakis' and Fowler's Connected: The Surprising Power of Our Social Networks and How They Shape Our Lives. The most recent example of this kind of work that I've seen is Adrien Friggeri's Agreement groups in the United States Senate (click on the link to see an interactive version).

Friggeri is a PhD student at the Ecole Normale SupĂ©rieure de Lyon. I am not able to comment on the method he used to create that graphic, an algorithm devised to identify cohesion within groups of people; he links to a paper about it, titled 'Triangles to Capture Social Cohesion', which includes some interesting charts as well. The conceptual framework and programming behind this project are beyond my knowledge. However, the results are captivating.

In the visualization, each ribbon represents a Senator. The right-left flow of lines is a measure of how much each particular person voted with his or her herd or moved away from it. What I find more worrying —although not surprising— is the lack of more cross-party collaboration. With a few noticeable exceptions, the divide between the political parties in the U.S has been consistent in the past 15 years. This is why Friggeri's graphic reminded me of the classic network plot designed by Lada Adamic and Natalie Glance for their paper 'The Political Blogosphere and the 2004 U.S. Election: Divided They Blog' (see below), which revealed the weak interaction between liberal and conservative bloggers.

Another comment I'd make is that I would like to see this timeline extended toward the past. Would it show a more integrated and collaborative Senate?