Applying network analysis to the “network of networks”
The case of Usenet discussion groups

Itai Himelboim
School of Journalism and Mass communication
University of Minnesota
Shawn Lavelle
Department of Computer Science
University of Minnesota

Nicholas Jankowski, at the 2004 New Research for New Media symposium, emphasized two future research agenda topics for new media: network analysis and the access and analysis of log data. The Internet, which has long been addressed as “the network of networks,” had rarely analyzed as a network. Because the Internet is computer based it holds detailed logs for its own networks for researchers to use.

New media has been traditionally studied on the individual level. The individual user has been studied using variety of tools such as interviews, questioners, eye-tracking and bio-feedback, in a range of contexts such as human-computer interactivity, internet access, and internet participation. The WWW was studied mainly on the websites level, using methodologies such as content analysis. However, the Internet is more than the sum of its parts. The connectivity of users in discussion groups and web-pages in the WWW contains more information about social implications of this new medium.

From a political science perspective, revealing the structures of users’ and websites’ connectivity leads to a better understanding of the accessibility of information and the distribution of power on this allegedly free and open to all democratic sphere. The study of the WWW as a network shows a highly skewed distribution of power in the WWW, where few websites attract a disproportional amount of hyperlinks and control the distribution of economic and political power on the Web.

This study takes a network approach to analyze Usenet discussion groups. A network analysis of more than a hundred Usenet groups from a variety of topics – politics, health, entertainment and technical support, shows that while topics of discussions can predict some of the variance in the structure of the networks, overall Usenet discussion groups are very hierarchical, and are characterized by a highly skewed distribution of responses. In other words, while Usenet groups are open to all to participate and exchange opinions, ideas and information, they are typically controlled by few, highly connected, participants.

This paper will discuss both the theoretical and the methodological novelty of studying Usenet discussion groups on the network level.

HOME