Skip to main content

The Chilling Effect Is Real

A new study on the aftermath of the Edward Snowden revelations gives us our best evidence yet.

By Jared Keller


The National Security Agency headquarters in Fort Meade, Maryland. (Photo: Saul Loeb/AFP/Getty Images)

Does the specter of surveillance limit free speech? You bet it does.

According to an upcoming paper in the Berkeley Technology Law Journal, Internet traffic to Wikipedia entries dealing with terrorism-related topics plummeted almost 30 percent in the aftermath of Edward Snowden’s 2013 revelations about widespread surveillance by the National Security Agency. Reuters reports that the analysis provides the “most direct evidence to date” of the “chilling effect,” the inhibition of otherwise legal behavior due to government observation.

Here are the details, per Reuters:

Author Jonathon Penney, a fellow at the University of Toronto’s interdisciplinary Citizen Lab, examined monthly views of Wikipedia articles on 48 topics identified by the Department of Homeland Security as subjects that they track on social media, including Al Qaeda, dirty bombs and jihad.

In the 16 months prior to the first major Snowden stories in June 2013, the articles drew a variable but an increasing audience, with a low point of about 2.2 million per month rising to 3.0 million just before disclosures of the NSA’s Internet spying programs. Views of the sensitive pages rapidly fell back to 2.2 million a month in the next two months and later dipped under 2.0 million before stabilizing below 2.5 million 14 months later, Penney found.

Penney’s results come on the heels of mounting evidence by researchers that the mere threat of government surveillance (consider, say, a security camera that’s not actually recording anything) constrains perfectly legitimate freedoms of expression under the First Amendment. A 2015 study of search activity around 282 specific terms culled from the Department of Homeland Security found that Google searches rated as “more likely to get you in trouble with the U.S. government” saw a significant decline in traffic in the aftermath of the Snowden revelations. While the drop in problematic search results was most pronounced in the United States, there was also a significant, if smaller, decline in non-U.S. searchers. The chilling effect, like the Internet itself, is a global phenomenon.

There’s one potential problem with the chilling effect thesis: If inhibition is a function of the knowledge of surveillance, why is the average American Internet user (as observed through the filter of anonymized Google results) more skittish and inhibited than Muslim Americans who have disproportionately dealt with intense government scrutiny since the September 11, 2001, attacks? A 2007 survey of Muslim Americans on post-9/11 changes in their Internet use found that, while the vast majority of respondents believed the U.S. government was currently monitoring Muslims Internet activities (70.7 percent), a whopping 89.1 percent of respondents said they made zero changes to their online behavior. Even 77.2 percent said they didn’t know any other Muslims who had changed what they did on the Internet to avoid scrutiny by some shadowy intelligence agency.

The chilling effect, like the Internet itself, is a global phenomenon.

What accounts for this potential gulf between the belief in government surveillance and lack of subsequent changes in online behavior among Muslim Americans? There are two potential explanations. The first is methodological: While Internet search data doesn’t lie, people do, especially on surveys administered over the telephone. It may simply be that respondents aren’t willing to telegraph dissent toward measures that are ostensibly meant to keep people safe; consider that a Pew Research Center survey conducted in the aftermath of the Snowden revelations found that a majority of Americans approved of electronic surveillance, despite the empirical drop in search traffic reported in the Berkeley Technology Law Journal. The other option is habituation: Muslim Americans have become so used to discrimination and harassment by law enforcement officials and less tolerant members of the public that the rise of the post-9/11 surveillance apparatus — and the resulting NSA dragnet — is more business as usual. That Muslim Americans didn’t report a chilling effect may itself be a function of the chilling effect.

The experience of Muslim Americans in New York, however, suggests that habituation is rather unlikely — and reinforces the importance of empirical evidence of the chilling effect. In 2011, the Associated Press exposed the New York Police Department’s surveillance of “hot spots” in the city’s Muslim community (read: mosques), including the deployment of informants, undercover officers, and surveillance cameras. A 2012 police deposition concluded that the program didn’t result in any new terror-related leads. But the impact on the community was substantial: A 2013 joint report by the CUNY School of Law and the Muslim American Civil Liberties Coalition found that Muslims of all backgrounds “began to downplay their faith or avoid political discussions and activism” as a result of the sweeps, as the Verge reported at the time. “While some of the comments are clearly in response to single incidents of surveillance,” the Vergewrote, “the knowledge that police might be watching can stifle even innocuous political debate or conversation and sow mistrust.”

That the chilling effect is both measurable and quantifiable in the form of Internet activity is hugely important. For Muslims (and others) who seek recourse in the face of overwhelming surveillance, there’s finally some evidence that the watchful eye of the government may be doing more harm than good.