(Read the Executive Summary or download the full report)
Both winners and losers of the 2016 presidential election describe it as a political earthquake. Donald Trump was the most explicitly populist candidate in modern history. He ran an overtly anti-elite and anti-media campaign and embraced positions on trade, immigration, and international alliances, among many other topics, that were outside elite consensus. Trump expressed these positions in starkly aggressive terms. His detractors perceived Trump’s views and the manner in which he communicated them as alarming, and his supporters perceived them as refreshing and candid. He was outraised and outspent by his opponents in both the primary and the general election, and yet he prevailed—contrary to the conventional wisdom of the past several elections that winning, or at least staying close, in the money race is a precondition to winning both the nomination and the election.
In this report we explore the dynamics of the election by analyzing over two million stories related to the election, published online by approximately 70,000 media sources between May 1, 2015, and Election Day in 2016. We measure how often sources were linked to by other online sources and how often they were shared on Facebook or Twitter. Through these sharing patterns and analysis of the content of the stories, we identify both what was highly salient according to these different measures and the relationships among different media, stories, and Twitter users.
Our clearest and most significant observation is that the American political system has seen not a symmetrical polarization of the two sides of the political map, but rather the emergence of a discrete and relatively insular right-wing media ecosystem whose shape and communications practices differ sharply from the rest of the media ecosystem, ranging from the center-right to the left. Right-wing media were centered on Breitbart and Fox News, and they presented partisan-disciplined messaging, which was not the case for the traditional professional media that were the center of attention across the rest of the media sphere. The right-wing media ecosystem partly insulated its readers from nonconforming news reported elsewhere and moderated the effects of bad news for Donald Trump’s candidacy. While we observe highly partisan and clickbait news sites on both sides of the partisan divide, especially on Facebook, on the right these sites received amplification and legitimation through an attention backbone that tied the most extreme conspiracy sites like Truthfeed, Infowars, through the likes of Gateway Pundit and Conservative Treehouse, to bridging sites like Daily Caller and Breitbart that legitimated and normalized the paranoid style that came to typify the right-wing ecosystem in the 2016 election. This attention backbone relied heavily on social media.
For the past 20 years there has been substantial literature decrying the polarization of American politics. The core claim has been that the right and the left are drawing farther apart, becoming more insular, and adopting more extreme versions of their own arguments. It is well established that political elites have become polarized over the past several decades, while other research has shown that the electorate has also grown apart. Other versions of the argument have focused on the internet specifically, arguing that echo chambers or filter bubbles have caused people of like political views to read only one another and to reinforce each other’s views, leading to the adoption of more extreme views. These various arguments have focused on general features of either the communications system or political psychology—homophily, confirmation bias, in-group/out-group dynamics, and so forth. Many commentators and scholars predicted and measured roughly symmetric polarization on the two sides of the political divide.
Our observations of the 2016 election are inconsistent with a symmetric polarization hypothesis. Instead, we see a distinctly asymmetric pattern with an inflection point in the center-right—the least populated and least influential portion of the media spectrum. In effect, we have seen a radicalization of the right wing of American politics: a hollowing out of the center-right and its displacement by a new, more extreme form of right-wing politics. During this election cycle, media sources that attracted attention on the center-right, center, center-left, and left followed a more or less normal distribution of attention from the center-right to the left, when attention is measured by either links or tweets, and a somewhat more left-tilted distribution when measured by Facebook shares. By contrast, the distribution of attention on the right was skewed to the far right. The number of media outlets that appeared in the center-right was relatively small; their influence was generally low, whether measured by inlinks or social media shares; and they tended to link out to the traditional media—such as the New York Times and the Washington Post—to the same extent as did outlets in the center, center-left, and left, and significantly more than did outlets on the right. The number of farther-right media outlets is very large, and the preponderance of attention to these sources, which include Fox News and Breitbart, came from media outlets and readers within the right. This asymmetry between the left and the right appears in the link ecosystem, and is even more pronounced when measured by social media sharing.
Media attention from readers on the left was directed at outlets with orientations ranging from distinctly partisan to relatively centrist. The outlets from the far left were fairly tightly interlinked with sources on the center-left and center and included a mix of stories that were center and left. By contrast, readers and outlets on the right had a more distinct focus and a more insular set of linking and social media sharing practices. These observed patterns are not easily explained by factors that are politically neutral, such as a general idea of homophily or internet algorithms, because such factors should have similar impact on both sides of the political map. Rather, the asymmetry requires a historical, cultural, or social-norms analysis to explain the rightward shift on the right that is not paralleled by a similar leftward shift on the left.
The relative insularity and densely interconnected network of sites in the right-wing media ecosystem may explain why we observe asymmetrically high attention on the right to sites engaged in conspiracy theories, like Gateway Pundit or Infowars, and successful political clickbait sites, like Ending the Fed. This network structure is more conducive to the echo-chamber, information-cascade, and filter bubble effects often discussed as risks of the online environment. Correspondingly, the less insular left wing of the media ecosystem, which is more interconnected with the center-left and center of the media ecosystem, is less susceptible to such problems, which are associated with media spheres that have deliberately untethered from the systems of accountability and recourse built into the broader media ecosystem.
This does not mean that there are no far left sites or left-oriented political clickbait sites; there are and we document them. But readers and media sources on the left paid more attention to traditional media organizations, which in turn moderated the coverage of the partisan left. Media outlets on the right that might have helped to moderate conservative coverage were relegated to second-tier status, losing attention on the right. The Wall Street Journal is a notable example, but our comparison of the network structure of the 2016 election with that of October 2012 shows that even more traditional conservative sites like the National Review or the Weekly Standard lost significant ground to the more radical media outlets on the right between 2012 and 2016. The architectural features of the election media ecosystem facilitated the use of network propaganda and disinformation disproportionately by the right. The last chapter of this report, “Dynamics of Network Propaganda: The Clinton Foundation,” offers a detailed case study of these practices and how they were used both to influence mainstream media coverage and to rally the base on the right.
The abandonment of the Fairness Doctrine in the 1980s created a space on AM radio for extremist views from Rush Limbaugh, Sean Hannity, and, by the mid-1990s, Alex Jones. The expanding market for cable channels in the 1990s made a partisan cable news channel like Fox News, founded in 1996, a thriving business model. In the 2000s, the writable web, or Web 2.0, allowed for the emergence of partisan media outlets in the blogosphere on both sides of the political spectrum, from DailyKos and Talking Points Memo on the left, to Hot Air, PJ Media, and Townhall on the right, although asymmetry in organization, technology adoption, and content were already visible then. Over the past decade, social media have emerged as effective pathways for more extreme views.
In this election cycle, media coverage on both sides of the spectrum was more partisan on social media than on the open web, and more partisan on Facebook than on Twitter. Moreover, these more partisan segments of the media ecosystem appear to have been more vulnerable to disinformation and false reporting. When objectivity and accuracy are at odds with partisanship, fealty to partisan messaging necessitates a loosening of standards regarding truth. The long-standing asymmetry visible in AM radio and cable news reappeared in the 2016 election cycle, both on the web and on social media. Although disinformation driven by political clickbait and partisan media appeared not only on the right, it played a more prominent part in conservative media—at least the portion to which Trump supporters paid the most attention—than on the liberal side of the spectrum. And just as Al Franken’s Air America or Jon Stewart’s Daily Show took the technical affordances of radio and cable, respectively, and mirrored their partisan counterparts with explicit satire, so too in our data we see the Onion and the Borowitz Report occupying parallel spaces to the disinformation-rich sites on the social media right, but, of course, because of their explicitly satirical tone, playing a very different political role. Highly partisan media—the principal incubator and disseminator of disinformation—and Facebook-empowered hyperpartisan political clickbait sites played a much greater role on the right than on the left.
This is not evidence that media consumers on one side are savvier than the other. Research suggests that liberals and conservatives alike are saddled with behavioral biases that support a willingness to believe and disseminate misleading information that aligns with our group identities and worldviews, and to gravitate toward circles of like-minded individuals that reinforce shared values and group identity. Our observations suggest that, given this baseline similarity, the network structure and practices of media on the right reinforced these basic dynamics, while network structure and media practices throughout the rest of the media ecosystem, particularly the greater interconnection of sources across the center, center-left, and left, moderated them.
The asymmetry in the degree of partisanship and message discipline between the right and the rest of the media ecosystem influenced the coverage and agenda of the campaign across the media spectrum. The center and left of the media sphere paid attention almost exclusively to negative articles focused on personal scandals or failings of both Donald Trump and Hillary Clinton, although the level of attention to (and criticism of) Trump was several times higher. The Breitbart/Fox-centered media system focused on attacking Clinton’s character and highlighting her negatives while touting Trump’s positives or deflecting attention from his negatives. It also produced a steady flow of stories focused on Trump’s substantive agenda, particularly his hardline stance on immigration with a strong Islamophobic inflection. As a result of right-wing media’s successful framing of the election discourse around immigration and its more uniform and disciplined (or partisan) messaging, overall coverage of the two candidates was very different. Coverage of Clinton was dominated by her emails, Benghazi (early in the cycle), and the Clinton Foundation (after the convention). Over the course of the campaign, Trump’s views on immigration received more coverage than any of his scandals, although the coverage was often critical. Specific scandals, most importantly the Hollywood Access tape, dominated his coverage at discrete moments in the campaign. Additional coverage of Trump focused on other negative stories, but even stories such as the Trump University scandal received no more attention than the candidate’s views on trade. In short, attention to reports of Clinton scandals exceeded attention to her stance on issues, whereas attention to reports of Trump’s scandals was balanced by attention to his stance on the issues and reinforced his focus on immigration, his campaign’s primary substantive issue. Despite the clear separation between the two media spheres, the right-wing media succeeded in shaping the agenda across the political spectrum in a way that the Clinton campaign did not.
Breitbart’s influence is clearest on the topic of immigration. A clear finding in our data is that the right-wing discussion of immigration was much more focused on Islam and terrorism, and the narratives gravitated more often to issues of identity threat (significantly, white identity) than on economic insecurity or the loss of jobs. On this topic of immigration, Breitbart played a particularly influential role—one substantially greater than its significant role in the overall election discourse. In discussions on Twitter related to immigration, Breitbart was the most prominent node by a wide margin. By contrast, as measured by linking behavior and social media shares, alt-right and white nationalist sites such as VDARE, WND, and Daily Stormer played a small role in the immigration debate and do not appear to have been as central to the Trump sphere as some reports, including some of their own triumphal celebrations of the Trump victory, would suggest. As described in further detail in the immigration case study, less extreme right-wing media outlets, including Fox News and Daily Caller, alongside Breitbart, employed anti-immigrant narratives that echoed sentiments from the alt-right and white nationalists but without the explicitly racist and pro-segregation language.
Our data suggest that the “fake news” framing of what happened in the 2016 campaign, which received much post-election attention, is a distraction. Moreover, it appears to reinforce and buy into a major theme of the Trump campaign: that news cannot be trusted. The wave of attention to fake news is grounded in a real phenomenon, but at least in the 2016 election it seems to have played a relatively small role in the overall scheme of things. We do indeed find stories in our data set that come from sites, like Ending the Fed, intended as political clickbait to make a profit from Facebook, often with no real interest in the political outcome. But while individual stories may have succeeded in getting attention, these stories are usually of tertiary significance. In a scan of the 100 most shared stories in our Twitter and Facebook sets, the most widely shared fake news stories (in this sense of profit-driven Facebook clickbait) were ranked 66th and 55th by Twitter and Facebook shares, respectively, and on both Twitter and Facebook only two of the top 100 stories were from such sites. Out of two million stories, that may seem significant, but in the scheme of an election, it seems more likely to have yielded returns to its propagators than to have actually swayed opinions in significant measure. When we look at our data week by week, prominent fake news stories of this “Macedonian” type are rare and were almost never among the most significant 10 or 20 stories of the week, much less the election as a whole. Disinformation and propaganda from dedicated partisan sites on both sides of the political divide played a much greater role in the election. It was more rampant, though, on the right than on the left, as it took root in the dominant partisan media on the right, including Breitbart, Daily Caller, and Fox News. Moreover, the most successful examples of these political clickbait stories are enmeshed in a network of sites that have already created, circulated, and validated a set of narrative lines and tropes familiar within their network. The clickbait sites merely repackage and retransmit these already widely shared stories. We document this dynamic for one of the most successful such political clickbait stories, published by Ending the Fed, in the last chapter of this report, and we put it in the context of the much more important role played by Breitbart, Fox News, and the Daily Caller in reorienting the public conversation after the Democratic convention around the asserted improprieties associated with the Clinton Foundation.
Our observations suggest that fixing the American public sphere may be much harder than we would like. One feature of the more widely circulated explanations of our “post-truth” moment—fake news sites seeking Facebook advertising, Russia engaging in a propaganda war, or information overload leading confused voters to fail to distinguish facts from false or misleading reporting—is that these are clearly inconsistent with democratic values, and the need for interventions to respond to them is more or less indisputable. If profit-driven fake news is the problem, solutions like urging Facebook or Google to use technical mechanisms to identify fake news sites and silence them by denying them advertising revenue or downgrading the visibility of their sites seem, on their face, not to conflict with any democratic values. Similarly, if a foreign power is seeking to influence our democratic process by propagandistic means, then having the intelligence community determine how this is being done and stop it is normatively unproblematic. If readers are simply confused, then developing tools that will feed them fact-checking metrics while they select and read stories might help. These approaches may contribute to solving the disorientation in the public sphere, but our observations suggest that they will be working on the margins of the core challenge.
The media structures we observe in this study have taken shape in conjunction with political dynamics that are informed by and intertwined with deep-seeded social and cultural value systems. The asymmetries in media structures and function are likewise grounded in society-wide processes that have evolved over several decades. The production, dissemination, consumption, and processing of political information are inherently political acts tied to group identity. If the real challenges come from inside the political system and consist of intentional political communication within a major wing of the American political system, willingly received by a large segment of the American citizenry, then the solution is far from obvious and interventions must confront the political origins of the problem.
One point of optimism from this study is the continued central role of traditional media organizations across the political spectrum from center-right to left. Editorial boards and professional journalists must grapple with the pitfalls of operating in a disinformation-rich environment. Their ability to respond and adjust to this more challenging environment will benefit the majority of the American public that does not occupy the right-wing media ecosystem.
But the efforts to find a technological fix—through changes to the Facebook algorithm or a fact-checking app—are much less likely to be either effective or normatively justifiable if they mean intentional disruption of a class of political communication desired by its recipients and intended to forge a powerful political connection within a substantial wing of the American public. Where is the Archimedean point from which to look at the right-wing media ecosystem and say, “What is going on is so inimical to American democracy that we know that it is an attack on democracy, rather than its expression?” If readers on the right shun fact-checking sites because they read their media to reinforce their in-group identity and share in partisan folklore, then new fact-checking apps will simply be ignored, and imagining that installing them will solve the rise of disinformation is merely whistling past the graveyard. Perhaps the most important antidote is in the hands of American conservatives who do not recognize themselves in the racist, populist, anti-science, anti-rule-of-law, and anti-journalism worldviews that became so prominent in the right-wing media ecosystem. Such conservatives have the legitimacy on the edges of the new right wing to speak from inside the tent, rather than throwing stones at it from outside. The marginalization of never-Trump voices within conservative circles in this election cycle suggests that such conservatives will face a steep uphill battle. Fundamentally, our observations suggest that the answers will not likely come from the technological or legal domains, but from the realm of political culture, norms, and power.