The Case for Government Control of Internet Speech Grows Weaker: Filter Bubble Edition

Cass Sunstein has been for some time a capable and influential critic of individual choice and limited government. Over the past decade, he has argued that the Internet is failing liberal democracy. Left to their own preferences, he says, individuals choose to avoid political views that challenge their prior beliefs. They form filter bubbles that exclude contrary views and echo chambers that polarize debates. Both complicate solving national problems.

These alleged filter bubbles and echo chambers comprise expressing and hearing (or reading) speech, both highly protected activities in the United States (or in any polity deserving the name liberal). The harms of filter bubbles and echo chambers should be much more than alleged to justify government actions to “improve” our debates.

Sunstein’s claims about filter bubbles and echo chambers have a certain appeal. We can imagine people choosing to avoid unpleasant people and views. As communications researcher Cristian Vaccari notes:

social media users can make choices as to which sources they follow and engage with. Whether people use these choice affordances solely to flock to content reinforcing their political preferences and prejudices, filtering out or avoiding content that espouses other viewpoints, is, however, an empirical question—not a destiny inscribed in the way social media and their algorithms function.

Both older and more recent studies cast doubt on Sunstein’s claim that the individual choices of Internet users are turning the nation into a polarized dystopia. For example, several studies published in 2016 and earlier indicate that people using the internet and social media are not shielded from news contravening their prior beliefs or attitudes (see the references here). In 2014, experimental evidence led two scholars to state “that social media should be expected to increase users’ exposure to a variety of news and politically diverse information.” They conclude that “the odds of exposure to counterattitudinal information among partisans and political news among the disaffected strike us as substantially higher than interpersonal discussion or traditional media venues.” A 2015 paper based on a panel design found that “most social media users are embedded in ideologically diverse networks, and that exposure to political diversity has a positive effect on political moderation.” Contrary to the received wisdom, this data “provides evidence that social media usage reduces mass political polarization.” A broad literature review in 2016 found “no empirical evidence that warrants any strong worries about filter bubbles.” Just before the 2016 election, a survey of U.S. adults found that social media users perceive more political disagreement than non-users, that they perceive more of it on social media than in other media, and that news use on social media is positively associated with perceived disagreement on social media.

Did the 2016 election change these findings? No doubt all of the studies of that election have not yet appeared. But several suggest doubts about filter bubbles, polarization, and Internet use remain valid. Cato published a summary of a study by three economists who found that polarization has advanced most rapidly among demographic groups least likely to use the Internet for political news. The cause (Internet use) was absent from the effect of interest (increased polarization). Other studies have been more specific. Three communications scholars examined how people used Facebook news during the 2016 U.S. presidential campaign. They had panel data and thus could examine how Internet usage affected the attitudes of the same people over time. The results suggest Sunstein’s concerns are exaggerated. Both Internet use and the attitudes of the panel “remained relatively stable.” A filter bubble did not appear: the people who used Facebook for news were more likely to view news that both affirmed and contravened their prior beliefs. Indeed, over time, people exposed themselves more to contrary views which “was related to a modest...spiral of depolarization.” In contrast, they found no evidence of a filter bubble where exposure to news affirming prior attitudes led to greater polarization.

 Other recent studies have focused on both the United States and other developed nations or just European nations alone. Perhaps data and conclusions from other developed nations do not transfer to the United States. However, cultures and borders notwithstanding, citizens in developed nations are similar in wealth and education. Even if we put less weight on conclusions from Europe, such inform our thinking about supposed failures of Internet speech. 

In 2017, Cristian Vaccari surveyed citizens in France, Germany, and the United Kingdom to test the extent of filter bubbles online. He concluded “social media users are more likely to disagree than agree with the political contents they see on these platforms” and that “citizens are much more likely to encounter disagreeable views on social media than in face-to-face conversations.” His evaluation of Sunstein’s thesis merits quoting at length:

Ideological echo chambers and filter bubbles on social media are the exception, not the norm. Being the exception does not mean being non-existent, of course. Based on these estimates, between one in five and one in eight social media users report being in ideological echo chambers. However, most social media users experience a rather balanced combination of views they agree and disagree with. If anything, the clash of disagreeing opinions is more common on social media than ideological echo chambers.

Another recent study in the United Kingdom found that most people tended to avoid echo chambers. Only about 8 percent of their sample had constructed echo chambers. The authors urge us to look more broadly at media and public opinion: 

Whatever may be happening on any single social media platform, when we look at the entire media environment, there is little apparent echo chamber. People regularly encounter things that they disagree with. People check multiple sources. People try to confirm information using search. Possibly most important, people discover things that change their political opinions. Looking at the entire multi-media environment, we find little evidence of an echo chamber.

Finally, another study of multiple countries found that using social media was related to incidental exposure to news, contrary to Sunstein’s view that older media promoted such unintended exposure while new media do not.

Sunstein’s concerns about filter bubbles and echo chambers appear exaggerated. Accordingly, the case for government action to improve public deliberation fails.