April 27, 2016 at 9:13 AM
In June 2013, reporters at The Washington Post and the Guardian ran a series of stories about the U.S. government's surveillance programs. According to documents leaked by Edward Snowden, the National Security Agency was harvesting huge swaths of online traffic — far beyond what had been disclosed — and was working directly with top Internet companies to spy on certain people.
Glenn Greenwald, one of the Guardian journalists who reported the disclosures and a surveillance skeptic, argued in a 2014 TED talk that privacy is a critical feature of open society. People act differently when they know they're being watched. "Essential to what it means to be a free and fulfilled human being is to have a place that we can go and be free of the judgmental eyes of other people," he said.
Privacy advocates have argued that widespread government surveillance has had a "chilling effect" — it encourages meekness and conformity. If we think that authorities are watching our online actions, we might stop visiting certain websites or not say certain things just to avoid seeming suspicious.
The problem, though, is that it's difficult to judge the effect of government-spying programs. How do you collect all the utterances that people stopped themselves from saying? How do you count all the conversations that weren't had?
A new study provides some insight into the repercussions of the Snowden revelations, arguing that they happened so swiftly and were so high-profile that they triggered a measurable shift in the way people used the Internet.
Jonathon Penney, a PhD candidate at Oxford, analyzed Wikipedia traffic in the months before and after the NSA's spying became big news in 2013. Penney found a 20 percent decline in page views on Wikipedia articles related to terrorism, including those that mentioned "al-Qaeda," "car bomb" or "Taliban."
"You want to have informed citizens," Penney said. "If people are spooked or deterred from learning about important policy matters like terrorism and national security, this is a real threat to proper democratic debate."
Even though the NSA was supposed to target only foreigners, the immense scale of its operations caused many to worry that innocent Americans were getting caught in the dragnet. A Pew survey in 2015 showed that about 40 percent of Americans were "very" or "somewhat" concerned that the government was spying on their online activities.
The same survey showed that about 87 percent of American adults were aware of the Snowden news stories. Of those people, about a third said they had changed their Internet or phone habits as a result. For instance, 13 percent said they "avoided using certain terms" online; and 14 percent said they were having more conversations face to face instead of over the phone. The sudden, new knowledge about the surveillance programs had increased their concerns about their privacy.
Penney's research, which is forthcoming in the Berkeley Technology Law Journal, echoes the results of a similar study conducted last year on Google Search data. Alex Marthews, a privacy activist, and Catherine Tucker, a professor at MIT's business school, found that Google activity for certain keywords fell after the Snowden stories were splashed on every front page. Both in the United States and in other countries, people became reluctant to search for terrorism-related words such as "dirty bomb" or "pandemic."
Penney focused on Wikipedia pages related to sensitive topics specifically flagged by the Department of Homeland Security. In a document provided to its analysts in 2011, the DHS listed 48 terrorism terms that they should use when "monitoring social media sites." Penney collected traffic data on the English Wikipedia pages most closely related to those terms.
This chart from the paper shows how the number of views dropped after the June 2013 news articles. The amount of traffic immediately dropped and stayed low for the subsequent 14 months.
To be clear, this traffic drop could have come from anywhere — Penney did not track the country of origin — but the United States accounts for 43 percent of English Wikipedia traffic, more than any other country.
Penney narrowed the list to the most suspicious-sounding articles, as judged by an online survey he administered. The results became even more dramatic.
Here, in black, are the combined monthly traffic totals for the Wikipedia pages related to the 31 top words on the DHS list. In the year and a half before the Snowden revelations, traffic to these pages was rising. After June 2013, traffic not only fell immediately, but continued to decline over the next dozen months.
For comparison, the chart also shows the combined page views for 25 Wikipedia pages that are security-related but not terrorism-related. These are less provocative articles containing the words "Border patrol" or "Central Intelligence Agency." There was a slight but statistically insignificant dip in traffic for these pages, which makes sense because people may not be as worried about visiting these kinds of pages.
The Wikipedia data suggest that the Snowden revelations had a noticeable impact on people's Wikipedia behaviors, says Penney. "I expected to find an immediate drop-off in June, and then people would slowly realize that nobody is going to jail for viewing Wikipedia articles, and the traffic would go back up," he said. "I was surprised to see what looks to be a longer-term impact from the revelations."
Penney has provided evidence that spying programs, once the public knows about them, cause collateral damage. It's unlikely, of course, that the patterns here were caused by actual terrorists changing their Internet habits.
Instead, the study suggests that the shift in Wikipedia traffic was the result of people who stifled their curious impulses because they didn't want to seem like they were doing anything wrong. "This is measuring regular people who are being spooked by the idea of government surveillance online," Penney said.
That's one plausible conclusion we could draw from the data. There is, however, an alternative explanation for these results.
The Snowden revelations ignited a huge debate about the NSA. Stories about government surveillance dominated the news cycle for months. Perhaps people stopped looking at terrorism-related Wikipedia articles not because the Snowden leaks made them paranoid, but because the news distracted them from their previous curiosity about terrorism.
In other words, maybe it wasn't a "chilling effect" that caused the dip in terrorism-related Wikipedia traffic — but rather the short attention spans of online audiences.
It will require more research to fully understand what happened. But even if the evidence is still being examined, chilling effects continue to occupy a prominent position in privacy debates.
In March 2015, the American Civil Liberties United filed a lawsuit in federal district court challenging the NSA's surveillance practices, with Wikipedia's parent organization as one of the eight plaintiffs. Writing with his colleague Lila Tretikov in the New York Times, Wikipedia founder Jimmy Wales accused the NSA of tracking Wikipedia users:
So imagine, now, a Wikipedia user in Egypt who wants to edit a page about government opposition or discuss it with fellow editors. If that user knows the N.S.A. is routinely combing through her contributions to Wikipedia, and possibly sharing information with her government, she will surely be less likely to add her knowledge or have that conversation, for fear of reprisal.
The result, argued Wales and Tretikov, "represents a loss for everyone who uses Wikipedia and the Internet — not just fellow editors, but hundreds of millions of readers in the United States and around the world."
In October, a judge threw out the case, declaring that the plaintiffs "have not alleged facts that plausibly establish an injury attributable to the NSA's Upstream surveillance." In other words, there wasn't enough evidence that anyone was harmed. The decision has been appealed.