New Oxford study shows evidence of post-Snowden “chilling effects” in what people read on Wikipedia.
A new Oxford University study has published empirical evidence showing that government mass surveillance programs like those exposed by Edward Snowden make us significantly less likely to read about surveillance and other national security-related topics online.
The study looks at Wikipedia traffic before and after Snowden's surveillance revelations to offer some new insight into the phenomenon of "chilling effects," which privacy advocates frequently cite as a damaging consequence of unchecked government surveillance. What it found is that traffic on "privacy-sensitive" articles dropped significantly following what author Jon Penney describes as an "exogenous shock" caused by revelations of the NSA's mass surveillance programs and the resulting media coverage.
The articles were chosen based on keywords from a list of terms flagged by the Department of Homeland Security, used for monitoring social media for terrorism and "suspicious" activity. For example, Wikipedia articles containing the 48 terrorism-related terms the DHS identified—including "al-Qaeda," "carbomb" and "Taliban"—saw their traffic drop by 20 percent.
The results also mirror a similar MIT study from last year which found that users were less likely to run Google searches containing privacy and national security-related terms that might make them suspicious in the eyes of the government.
Perhaps even more alarmingly, the study seems to show a long-term drop in article views on these topics that lasts well past the initial shock of Snowden's revelations, suggesting that people's' calculations about what to read on Wikipedia may have been permanently affected.
"It means that the NSA/PRISM surveillance revelations, covered by media in June 2013, are associated in the findings not only with a sudden chilling effect but also a longer term, possibly even permanent, decrease in web traffic to the Wikipedia pages studied. This indicates a possible chill," writes Penney.
While the study makes a convincing argument, identifying chilling effects empirically has always been tricky. There are plenty of other factors that can influence behavior online that may not necessarily have to do with being afraid of the government, for example. But given the specific timeframe and the huge amount of other evidence showing that surveillance tends makes citizens more prone to self-censorship and conformity, Penney's study seems to at least be an anchor point for future arguments that these programs have caused harm.