The Geopolitical Secrets Hidden on Wikipedia
WHY YOU SHOULD CARE
Because high schoolers’ favorite homework source might be more useful than you think.
By James Watkins
Many people say they want a revolution, but we’re pretty sure that not many say they want a revolution in geopolitics. But there are at least a couple. Once upon a time, a group of scientists decided to apply their expertise in molecular and genetic interactions in your DNA to the field of global political stability. To achieve their goal, they’d use every high schooler’s (and college student’s) favorite homework resource — Wikipedia.
Sounds like an episode of Scooby-Doo, Science Edition. The results are arguably as cool.
A country’s geopolitical stability can be measured on Wikipedia.
The Wikipedia Dispute Index correlates with a number of established measures of geopolitical stability, about as well as they correlate with one other. Rob Russell, professor of protein evolution and cell networks and a specialist in computational biology, worked with two colleagues at the University of Heidelberg — his wife, biotechnology entrepreneur Gordana Apic, and bioinformatics research scientist Matthew Betts — to create the index by running a relatively simple calculation: On how many pages linked to each country’s homepage does Wikipedia helpfully notify the reader “The neutrality of this article is disputed”? (A notice all too familiar to those of us who research civil wars in obscure countries on weekends.)
The index relies on the “guilt by association” principle, a concept commonly used in biology, says Russell. If a country’s homepage is linked to many disputed Wikipedia pages, it is likely that there is unsettled public interpretation of, say, certain historical events, which can be used as a proxy for … well, something. “We don’t really know what it is measuring,” says Russell. “Something about how angry the contributors” to Wikipedia are in a certain country, he speculates. Nevertheless, a 2011 study published by the researchers showed that the measure remains stable over time, seems to cancel out “false positives” on aggregate and responds to real world events. The values for Georgia, for instance, jumped around the time of the 2008 conflict with Russia.
There remains scope to refine the index, such as by combining different-language editions of Wikipedia, if only Russell had the resources. “I suppose I had this naive dream that I would be approached by the CIA to try to beef this up,” says Russell. Well, that might not be so naive: In 1994, the CIA commissioned the Political Instability Task Force (PITF) to develop more accurate ways to chart and predict political instability in different countries.
Understanding political instability is extremely important, but also resource-intensive — to the tune of “tens of millions of dollars” of CIA money, says Dr. Monty G. Marshall, director of the Center for Systemic Peace, who has worked with the PITF for 18 years. Indices tend to aggregate measures of economic development, levels of discrimination, governance and many others, none of which are perfectly objectively comparable between countries or over time. Meanwhile, the Wikipedia Dispute Index is easy to compute and its methodology is wholly objective. An index by the Economist Intelligence Unit based off the work of the PITF correlates strongly with Russell’s Wikipedia Dispute Index.
To be sure, the initial study was exploratory, and while the findings are promising — “things don’t usually work this well first time through,” Russell says — they are not conclusive. While it may be a “curiosity,” says Marshall, the Wikipedia Dispute Index “has no direct analytic utility or explanatory value.” He tells a story of how, in the PITF’s work regressing many factors against the indices it created, one time, the strongest association was simply with the “country code,” demonstrating how “correlations between apparently unrelated variables can occur randomly or spuriously.”
Nevertheless, applying the science of networks to the internet is an established technique, with Amazon book recommendations another example of how “guilt by association” can result in surprisingly accurate results, especially if you provide Amazon with a lot of information by buying a lot of books. And there’s a hell of a lot of information on Wikipedia (40,323,058 pages and counting). Finding ways to mine this information could yield insights we never knew existed. It might not be the best source to reference in your high school history papers, but perhaps Wikipedia could teach the CIA a thing or two. Chances are they’re already watching.