Fighting Fake News Is Not the Solution

By Masha Gessen, The New Yorker

06 January 18

espair pervaded American newsrooms following the 2016 election. It wasn’t just that most of the media had failed to foresee the outcome of the vote; it was the realization, as one colleague put it, that “nothing we do matters.” Out of this despair, a new resolve and many a subscription drive were born, based on the widely shared belief that if more people consume more-accurate stories, American democracy may be saved.

In the popular imagination, the public is divided into two segments of roughly equal size: the “liberal bubble” and the “right-wing bubble.” In fact, there has never been much evidence that this picture was true, and two recent data points contribute to disproving it. One is a large study of the reach and impact of fake news; the other is opinion-poll data on the tax-reform bill that Congress passed and President Trump signed into law in December. Together, they burst the two-bubble theory by showing that most Americans are better informed and less gullible than you might think. That, in turn, suggests that fighting “fake news” is not the solution, or perhaps even a solution, to our current political problems.

For the fake-news study, three political scientists from three different universities—Andrew Guess, from Princeton, Brendan Nyhan, from Dartmouth, and Jason Reifler, from the University of Exeter—combined data on Web traffic in the month before and one week after the election with responses to an online public-opinion survey by 2,525 Americans to determine who consumed fake news, and how much. For their definition of “fake news,” the authors relied on an earlier study by the economists Hunt Alcott and Matthew Gentzkow, who looked at stories that are “intentionally and verifiably false and could mislead readers.” The economists’ study suggested that every American adult had been exposed to at least one fake news story in the leadup to the 2016 election, but relatively few people—roughly eight per cent—actually believed them.

The economists’ study was based on what people recalled seeing; the new study by the political scientists uses more and harder data. The conclusions, however, are fairly similar. First, they found that Trump supporters read fake pro-Trump stories while Clinton supporters read fake pro-Clinton stories, but the latter group consumed a lot less fake news than did the former. The study did not directly address the question of how far in advance of reading the fake stories these voting preferences had been cemented; the authors concluded that “the ‘echo chamber’ is deep . . . but narrow.” The biggest news, in other words, is that while many people were exposed to fake news stories, few were taken in by them as measured by how many similar articles they went on to read. About ten per cent of news consumers sought out more fake news, and they read an average of 33.16 fake stories, according to the political scientists.

An earlier study that used a different approach to data collection—combining social-media sharing, hyperlinking patterns, and language use—yielded similar results. In an article published in the Columbia Journalism Review last March, scholars and researchers from Harvard, Ritsumeikan, and the Massachusetts Institute of Technology argued that, in media consumption, “polarization was asymmetric.” In other words, rather than two bubbles, there was one, positioned far to the right of the political spectrum. A majority of Americans, the study showed, get their news from a variety of different media. They are routinely exposed to opinions they don’t share; they do not live in an echo chamber.

A second important observation in the study by Guess, Nyhan, and Reifler concerns the effectiveness, or, rather, the ineffectiveness, of fact-checking as a genre of article or Web site unto itself, such as the Washington Post’s Fact Checker or PolitiFact. “Not only was consumption of fact-checks concentrated among non-fake news consumers,” the authors wrote, “but we almost never observe respondents reading a fact-check of a specific claim in a fake news article that they read.” The study published in CJR, on the other hand, observed that while audiences of right-wing-bubble media and the rest of the media hardly overlapped, language had a way of migrating from Breitbart into the mainstream media. The authors identified the two topics that dominated false conspiratorial narratives—Hillary Clinton’s e-mails and the threat posed by immigration—and traced the mainstream media’s disproportionate focus on these topics to the fake-news sites’ obsession with them. Together, these observations suggest that those ineffectual fact-checking pieces might have been a primary vehicle of that migration—such as, for example, when the Post fact-checked Trump’s claim, made in an interview with the conspiracy-theory purveyor Sean Hannity, that Clinton’s e-mails caused the death of an Iranian defector.

Still, the most salient, consistent, and counterintuitive result of these studies is that the image of the American public as divided into two equal partisan bubbles is wrong. Opinion data on Trumpian tax reform is real-life proof that most Americans share a fact-based view of reality. Poll after poll showed that voters opposed the tax bill, and that they did so on the basis of accurate information: they believed that it would benefit the rich. If there were indeed two equal-sized information bubbles in this country, one might reasonably expect half the population to buy Trump’s incessantly repeated line that the bill constituted a tax cut for the middle class. One would also expect roughly half the voters to support repealing the Affordable Care Act. That a majority of Americans support Obamacare and do not support the tax law is proof that accurate reporting still matters—sort of.

Members of Congress who voted for the tax bill, which will disproportionately benefit the very wealthy and will gut Obamacare, may be justified in assuming that they can afford to make their donors happy at the expense of their voters: partisanship and gerrymandering, they reckon, will keep their seats safe. In other words, an informed public is a necessary condition of democracy, but not a sufficient one. Democracy may indeed die in darkness, but light is no guarantee that it will survive.