Harsh Taneja (Missouri School of Journalism, University of Missouri) and Angela Xiao Wu (School of Journalism and Communication, Chinese University of Hong Kong) examine the clustering of global internet web usage. To read their full study, please click here.
Scholars of internet governance have traditionally focused on how institutions such as sovereign nation states and multilateral organizations establish public policy. In doing so, experts and policy makers often presume the impact of internet policies on internet usage, but rarely do they examine usage aggregated from the behavior of individual web users.
To address this gap, earlier this year we initiated a research program to link the two disparate areas of internet governance and internet usage with a grant provided by IPO. Unlike most policy research, we took a bottom-up approach–that is, to evaluate user behavior first, and then link the observed patterns to policy initiatives. Our user behavior analysis relies on the web use data from ComScore, a provider that measures internet use from the computers of 2 million people in about 170 countries.
For this analysis, we conceptualized the internet as a network of interconnected websites, linked to one another if they had audiences in common. We constructed this network for the 1,000 most visited websites globally in September of 2009, 2011 and 2013. Analysis of these networks revealed a number of “clusters” of websites, whereby sites within the cluster had more users in common than they did with sites outside the cluster. In each of the three years, we found that the most salient means upon which websites clustered together we
re language and geography (and not content type). We interpreted such clusters as online expressions of place-based cultures, or “regional cultures.” The changes in online regional cultures over the years are particularly worthy of note.
First, we noticed what may be indicative of an ongoing de-Americanization of the WWW. In 2009, the largest cluster was of websites of US based companies, primarily in English. In 2011 and 2013, the cluster consisting of US-focused sites became more isolated as in these years “global” sites (e.g., Facebook and Twitter) become a distinct group, separating from the US cluster to form a cluster of their own. In other words, with the passage of time, just like many other online regional cultures that are distinct from the global cluster, the US-focused sites appear as a regional culture rather than the global center of the WWW in terms of web usage.
Moreover, our data reveal the rise of the Global South on the WWW. The China-based cluster, being the second largest in 2009, became the cluster with the largest number of websites in 2011 and grew even larger in 2013. Similar levels of growths are evident for the India and Brazil-based clusters as well. Simultaneously, we witnessed a gradual decline in the size of many clusters from the developed world, such as Germany, France and Japan as well as the fast shrinking of the Korean cluster in 2013 (The decline in Korea was sharper due to mobile devices displacing computers for web access, more so than in other developed countries). We think that what best explains these reconfigurations in cluster sizes is the spectacular growth of new internet users, as well as of local internet industries, in the Global South, compared to the relative stagnation on both these fronts witnessed in the developed world.
These clusters suggest that internet users are highly fragmented. Our next objective is to look deeper into what appeared to be prima facie—a “cultural fragmentation” of the internet and to locate its shaping forces. We calculated the distance of each cluster from the rest of the WWW, based on a network analytic measure termed “closeness centrality”. The more out-cluster websites users of in-cluster sites also visit, the shorter the distance between the said cluster and the rest of the WWW. Interestingly, we found that the Japan and the China cluster were equally distant from the rest of the WWW, even though one is a completely open internet, and the other is known for notoriously comprehensive access blockage. In general, any association between censorship and distance of a cluster proved hard to establish (This supports our previous study focusing on China’s Great Firewall). We decided to look for other factors, especially national internet-related policies, that could explain the formation of online regional cultures. We will elaborate on this stage of research in our next post.