Jump to ContentJump to Main Navigation
Network PropagandaManipulation, Disinformation, and Radicalization in American Politics$

Yochai Benkler, Robert Faris, and Hal Roberts

Print publication date: 2018

Print ISBN-13: 9780190923624

Published to Oxford Scholarship Online: October 2018

DOI: 10.1093/oso/9780190923624.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use. date: 22 October 2021

The Origins of Asymmetry

The Origins of Asymmetry

(p.311) 11 The Origins of Asymmetry
Network Propaganda

Yochai Benkler

Robert Faris

Hal Roberts

Oxford University Press

Abstract and Keywords

This chapter examines the origins of asymmetry in the American public sphere by charting the rise of second-wave right-wing media. Taking a political economy approach, this chapter investigates how institutions, politics, culture, and technology combine to explain why Rush Limbaugh, televangelism, and Fox News were able to emerge as mass media when they did, rather than remaining, as first-generation right-wing media after World War II had, small niche players. The chapter also considers how the emergence of the online right-wing media ecosystem followed the offline media ecosystem architecture because of the propaganda feedback loop. It shows that asymmetric polarization precedes the emergence of the internet and that even today the internet is highly unlikely to be the main cause of polarization, by comparison to Fox News and talk radio.

Keywords:   public sphere, right-wing media, Rush Limbaugh, televangelism, Fox News, media ecosystem, propaganda feedback loop, asymmetric polarization, internet, talk radio

No one, today, can be simultaneously honest, informed, and successful in the Republican Party.

WILLIAM RUSHER, Publisher, National Review, 1960 (lamenting Barry Goldwater’s loss of the party’s nomination to Richard Nixon)1

FACEBOOK DIDN’T CREATE the asymmetric architecture of the American public sphere. Nor did the internet or the blogosphere. The asymmetry is already clear in patterns of attention to cable news networks and talk radio. As Jeffrey Berry and Sarah Sobieraj document in their book, The Outrage Industry, left-wing partisan commentators like Lawrence O’Donnell on MSNBC can be every bit as vehement and emotional in the outrage business as right-wing partisans like Sean Hannity. Among stories suspect enough to be checked by PolitiFact, Rachel Maddow’s mostly-false or worse ratio (48 percent) is not meaningfully different than Hannity’s (50 percent). But liberal audiences do not pay attention to, or trust, their partisan commentators at rates even approaching those that typify conservative audiences. In the run-up to the 2012 election, Fox News viewership outstripped that of MSNBC and CNN; Maddow was drawing audiences one-third the size of the then-most popular conservative commentator, Bill O’Reilly.2 By November 2016, Adweek reported that “Fox News beat CNN and MSNBC combined across all dayparts.”3 Since Donald Trump’s election, MSNBC has seen a relative increase in viewership, but still, of the top five rated cable news shows, only Rachel Maddow was present on the left, behind Sean Hannity and Tucker Carlson, and ahead of Laura Ingraham and Bret Baier.4 Consistent with the ratings of individual shows, a Pew survey found that only 9 percent of Hillary Clinton voters reported using MSNBC as their primary source of news about the election. The corresponding number for Trump voters and Fox News was 40 percent.5 The asymmetry in talk radio is even starker. There is, effectively, no nationally syndicated liberal talk radio to speak of. From Rush Limbaugh, (p.312) through Sean Hannity, to Mark Levine and a host of others vying for the same market, talk radio captures tens of millions of listeners weekly, and it is almost exclusively right-wing outrage.6 Berry and Sobieraj also document the differences in language and rhetorical devices employed by different outrage outlets. Liberal media tend to have more mockery and direct confrontation between opponents; conservative media have few head-to-head confrontations but more misrepresentative exaggeration, insulting language, and name-calling, along with ideologically extreme activation language.7 Compared to liberal outlets, conservative programs live on outrage; the likelihood that a given program was conservative, as opposed to liberal, increased linearly with the number of instances where one of their categories of “outrage expression” appeared in a program.8 Perhaps unsurprisingly, getting your news from outrage merchants does not help you get a handle on reality. As Joanne Miller and her collaborators and, independently, Adam Berinsky have shown, for Democrats, the more knowledgeable they are about politics, the less likely they are to accept conspiracy theories or unsubstantiated rumors that harm their ideological opponents. But for Republicans more knowledge results in, at best, no change in the rate at which they accept conspiracy theories, and at worst, actually increases their willingness to accept such theories.9

There have been numerous experiments aimed at creating left-wing equivalents of the mainstays of right-wing media, most prominently Air America’s effort to replicate conservative talk radio in 2004 and MSNBC’s strategic shift toward creating a left-oriented mirror to Fox News in 2006. In the early stages of Web 2.0, which coincided with the George W. Bush administration and the Iraq and Afghanistan wars, a flourishing left-wing political blogosphere emerged alongside and in interaction with both mainstream media and the Democratic Party. None of these were able to replicate the industrial organization, market success, or political and cultural significance of the right, and none of these succeeded in creating a symmetrically polarized media ecosystem with balance between the left and right. Instead the left remained firmly anchored, together with the center, in “the reality-based community,” while the right largely shifted to a revival of the nineteenth-century partisan press. As Theda Skocpol and Vanessa Williamson put it in their study of the rise of the Tea Party:

American democracy is, in an important sense, caught betwixt and between in the new media world. The frank, exuberant, all-around partisanship of the nineteenth century is not quite what we now have. True, there are both liberal and conservative bloggers, and on the tube, (p.313) the Fox political slant is weakly countered by liberal-slanted shows on MSNBC. But mostly what America has right now is a thousand-pound-gorilla media juggernaut on the right, operating nineteenth-century style, coexisting with other news outlets trying to keep up while making fitful efforts, twentieth-century style, to check facts and cover “both sides of the story.”10

It would be difficult to offer a more succinct encapsulation of what our data describe. Both the macroscale architecture our data uncover, and the detailed, microlevel data we collect on specific case studies support this basic asymmetry.


Partisan press was as American as apple pie in the nineteenth century. Vitriol, partisanship, and smear campaigns were the order of the day. But as Michael Schudson’s classic work showed, American journalists in the 1920s began to develop norms and institutions that made objectivism and fact-based reporting—removed from partisan opinion—the hallmark of the profession.11 The process was very much an application to journalism of the much broader emergence of managerialism, or Weberian rationality, to all social systems. From Taylorism and Fordism in industry, through the rise of the administrative state and the expert agencies, architecture and urban planning, to the professionalization of science and what we now think of as “the professions,” the era of modernism saw a broad reorganization of social relations across diverse domains into structures oriented around expertise, objectivity, and evidence.12

As norms of objectivity developed for the mainstream, political suppression muted the left and right wings of the American sphere. The revolution in Russia and America’s entry into World War I drove the first Red Scare. The judicial opinions that laid the foundations for what would become, decades later, the Supreme Court’s First Amendment decisions that permitted speech to be suppressed only if it presents a “clear and present danger” were dissenting opinions or later-overturned decisions surrounding suppression of socialist antiwar efforts—The Masses,13 Schenck,14 and Oliver Wendell Holmes’s dissent in Abrams.15 Five times Socialist presidential candidate Eugene Debs was imprisoned for 10 years under the Espionage Act, although later pardoned by President Warren G. Harding. As radio burst onto the scene as the new mass medium in the 1920s, efforts to produce left-wing (p.314) media encountered regulatory resistance. In 1925, the Chicago Federation of Labor launched WCFL. In 1927, the Socialist Party of America launched WEVD (after Eugene V. Debs). Both outlets came under severe pressure from the Federal Radio Commission as propaganda stations and were forced to accept lower-power, constrained licenses that significantly inhibited their growth and adoption.

On the right Father Coughlin became a mass phenomenon and is the direct lineal ancestor of Rush Limbaugh and conservative talk radio. Beginning his career with religious broadcasts in the late 1920s, Coughlin shifted in the 1930s from purely religious broadcasting to politics. Initially he mixed ardent anti-communism with an embrace of the New Deal. After 1934, however, Coughlin became both anti–New Deal, denouncing it as creeping communism, and increasingly pro-Fascist, anti-semitic, and antiwar. Unlike the socialist and labor stations, which the Federal Communications Commission (FCC) had constrained from the start, Father Coughlin may have reached as many as 30 million listeners.16 And his broadcasts became a classic case study in early research on propaganda.17 In 1939, after the war began in Europe, guidelines from the National Association of Broadcasters on selling time to “spokesmen on controversial public issues” and the threat that failure to comply with these would result in FCC nonrenewal of licenses forced Coughlin off the air. It would be five decades before his conspiracy-theory-laden, emotional, propagandist style would return to the airwaves, one year after the FCC finally repealed the fairness doctrine.

During the war, Henry Luce, the head of Time Inc. and publisher of dozens of magazines, responded to concerns about a growing regulatory state and encroaching wartime propaganda by putting in motion a process whose conclusion would become the institutional heart of the post–World War II professional ethos: the 1947 Report of the Commission on Freedom of the Press. The principal concern to which the commission responded was the increasing power of media companies and the prospect of greater censorship by the government under the guise of ensuring that media served American interests. The commission brought together 12 intellectuals to study the role of media and took on the name of its head, Robert Hutchins, the president of the University of Chicago. This coterie of upper-class white men met periodically over a four-year period and issued in 1947 a scathing report on media practices at the time. The legal scholar C. Edwin Baker described the work of the Hutchins Commission “the most important, semiofficial, policy-oriented study of the mass media in U.S. history.”18 (p.315)

The Hutchins Commission defined the key functions of media as follows:

First, a truthful, comprehensive, and intelligent account of the day’s events in a context which gives them meaning; second, a forum for the exchange of comment and criticism; third, a means of projecting the opinions and attitudes of the groups in the society to one another; fourth, a method of presenting and clarifying the goals and values of the society; and, fifth, a way of reaching every member of the society by the currents of information, thought, and feeling which the press supplies.

They concluded that the media was failing to meet these goals:

These needs are not being met. The news is twisted by the emphasis on firstness, on the novel and sensational; by the personal interests of owners; and by pressure groups. Too much of the regular output of the press consists of a miscellaneous succession of stories and images which have no relation to the typical lives of real people anywhere. Too often the result is meaninglessness, flatness, distortion, and the perpetuation of misunderstanding among widely scattered groups whose only contract is through these media.

The commission concluded that the failures of media put freedom of expression at risk: “Press practices at times have been so irresponsible that if continued society is bound to take control for its own protection.” They ultimately recommended self-regulation and called on media to commit itself to take its social responsibilities more seriously.

In politics bipartisanship reigned ,and the mainstream dominated in the post–World War II years. The New Deal and Fair Deal saw progressive Democrats and Southern Democrats ally around programs that alleviated poverty and stabilized economic security while largely excluding Black populations from many of their benefits.19 Treaty of Detroit labor relations (agreements beginning in 1950 between the United Auto Workers and the Big Three automobile manufacturers that set labor and wage standards across large industrial sectors and were the foundation of post-war industrial peace and middle-class wage growth) and the sustained economic growth they enabled led to relatively widespread sharing of the returns, and the dominance of Keynesianism took much of the sting out of economic policy politics. The horror of Nazism and the solidarity of wartime experience (p.316) largely discredited the America First wing of the American right. And the Cold War largely united Democrats and Republicans at the expense of the left under McCarthyism.

Against this political background, and in the media and regulatory environment of postwar America, right-wing media remained on the periphery of mass channels. Clarence Manion’s “Forum” was the most successful radio program on the right and was popular from its launch in 1954 to its end in the late 1970s. The weekly program played a significant role in the rise of the postwar conservative movement, supported the Barry Goldwater candidacy, and worked in constant cooperation with the National Review and other conservative outlets. It was nonetheless limited to what a weekly show could do in terms of mass coverage.20 On the left, the Cold War and the Second Red Scare, McCarthyism, and the implementation of Treaty of Detroit labor relations through a more or less centrist accommodation between New Deal Democrats and Eisenhower-Nixon Republicans, meant that left-wing media were marginalized. While efforts like Pacifica Radio, launched in 1946, existed throughout this period, these remained relatively localized and did not reach national influence.

Magazines, with respectable but limited circulation in the tens of thousands up to around 200,000, continued to provide a forum for partisan commentary and news. On the left, The Nation had been published since 1865, and The Progressive and The New Republic (until it was bought in 1974 and moved right) operated since the rise of the Progressive movement before World War I. Mother Jones and The American Prospect joined them in later decades. On the right, Human Events launched in 1944 as a vestige of the America First antiwar movement, and developed over the 1950s and 60s as a major voice in the rise of the conservative movement. National Review, of course, became the sine qua non of conservative media within a few years of its founding in 1955.21 And while these conservative publications and the “Manion Forum” formed an interconnected network of outlets, with some shared publication facilities and with mutual recognition and praise, they never reached the scale and scope of either Coughlin’s audience or what would emerge in the 1990s. The only moment when conservative publications reached a truly mass audience was the publication of three self-published paperbacks in the run-up to the 1964 election, which were published outside of the right-wing media network but sold a total of 16 million copies, containing various conspiracy theories to suggest a Lyndon Johnson presidency would be a disaster (including, in A Texan Looks at Lyndon, insinuating that Johnson had an interest in the John F. Kennedy assassination that preceded his rise to the presidency).22 (p.317) The Goldwater candidacy and anonymous donor purchases of these books in order to distribute them as campaign paraphernalia combined to make these paperbacks instant successes, but subsequent efforts to replicate their success after Johnson won the election faltered.

Ultimately, the first generation of right-wing media was unable to overcome then-prevailing structural barriers to success. Most Americans got their news from the three broadcast networks. Radio still operated under strict group ownership limits, which meant that national syndication required negotiations with many independent station owners. Broadcast operated under the FCC’s fairness doctrine, whose core requirements were that broadcasters cover matters of public importance and that they do so fairly, mostly in the sense that they air competing positions. The doctrine was often associated with a right of reply for politicians who were subject to personal attack and other elements of the broader “public trustee” doctrine that held that private broadcasters holding licenses to public airwaves should act in managing those airwaves as a trustee for the real owners—the American people. While the fairness doctrine did not often result in complete silencing, it made many broadcasters skittish about airing programming that they thought might trigger an obligation to grant free response time to those attacked in these broadcasts. Daily newspapers were local and largely independent. All these conditions made it hard for first-generation postwar right-wing media to reach mass distribution on a national scale. Indeed, the anchor tenants of the first-generation right-wing media system were never, or almost never, economically sustainable as businesses. They relied on donations from listeners, a handful of ideologically committed wealthy individuals, and some corporate sponsorship. On the left these same economic barriers joined the more direct effects of anti-Communist sentiment and political and legal constraints stretching from the first Red Scare into the 1960s.

The one enduring legacy of that post–World War II generation of right-wing media was the basic rhetorical frame of liberal media bias. As Nicole Hemmer traces in great detail, Human Events imparted that sense that mainstream media was biased against conservative beliefs from its earliest days. By the 1960s the insistence that mainstream newspapers were biased and that it was necessary to produce media that were objective, but not impartial, was a basic tenet of conservative media. Hemmer describes a moment when, as part of the years-long battle between the UAW and the Kohler Company about whether Kohler would be a union shop, Manion offered Herbert Kohler an open channel to air his views. The union threatened to sue the station, and, fearing liability, the network of Mutual stations that aired the (p.318) “Manion Forum” canceled that day’s show. Manion responded by setting up his own network, syndicating his show to independent broadcast stations and setting up the first prototype of a conservative network. A basic theme emerged throughout the network of conservative publications in the 1950s and ’60s. Mainstream media were biased against conservative views; the FCC’s fairness doctrine, and the norms of objectivity held by Northeastern publishing, radio, and television elites all excluded conservative views from the media; and the only answer was to create mass-scale media that would balance out that broad media bias by taking a conservative perspective on what to report and how to report on it. That viewpoint of liberal media bias was embraced and repeated vocally by Spiro Agnew, and later by Richard Nixon as the Nixon administration struggled with increasing pressure from traditional journalistic sources—from the Pentagon Papers to Watergate. But that era also coincided with professional journalists receiving the highest levels of trust and positive public opinion, among both Democrats and Republicans, that has been recorded in the General Social Survey.23

Just as there was right-wing criticism of mainstream media, there was a steady flow of left-wing media criticism, most famously Edward Herman and Noam Chomsky’s Manufacturing Consent,24 but no less so Ben Bagdikian’s Media Monopoly25 and work by Robert McChesney, Ed Baker, and others, about the extent to which mainstream media reflect the perspectives and interests of corporate owners, government insiders, and the national security establishment. It was precisely that sense that underlay some of the early enthusiasm associated with the emergence of the internet as a democratizing medium, freeing individual and informal networks of individuals to play a greater role in setting the agenda, reporting the news, and mobilizing in a new networked public sphere.26 But because a left-wing mass media ecosystem never emerged in the same way as it did in the 1990s on the right, this left-wing version of the media-bias argument remained largely within the academic and activist domain and never became a basic narrative shared by large parts of the population. But on the right, televangelism, talk radio, and ultimately Fox News were able to reach mass audiences that eluded the first generation of postwar conservative media, and it was these outlets’ mass appeal that successfully generalized the “liberal media bias” frame to large parts of the population. That imbalance between the left and right’s ability to develop highly partisan mass media outlets, rather than the absence of a left-wing critique of media, accounts for the clear asymmetry in levels of trust in media between right-leaning and left-leaning mass audiences.

(p.319) After Disco: Talk Radio, Televangelism, and Cable News

Over the course of the 1970s, 1980s, and 1990s, a series of technological, institutional, and political changes removed each of the structural barriers that had contained the first generation of right-wing media and created the conditions for the emergence and dramatic success of the second generation right-wing media system that undergirds today’s asymmetric architecture, anchored by Fox News and talk radio.

The 1970s saw substantial moves to deregulate cable and remove many of the byzantine constraints that the FCC had placed on it during the 1960s and early 1970s at the behest of incumbent television broadcasters.27 At the same time, several technological developments increased cable channel capacity and the reach of national networks. Satellite distribution of content to local cable ground stations allowed Ted Turner to launch TBS as the first national cable network. Developments in compression, set-top box, and later hybrid fiber-coaxial systems dramatically increased channel capacity over the course of the 1980s and 1990s, making room for more niche-programming channels to develop. One of the earliest format innovations was CNN. In 1980, when it launched, CNN was a far-out idea—a 24-hour news channel in an era when the major news networks were losing money on creating a 30-minute show at prime time. Its audience grew dramatically over the course of the 1980s. Beginning with a paltry 1.3 million households at a time when network news was viewed by tens of millions across the three networks, CNN quickly grew and within a decade matched the big three networks in news audience. In 1991 the Iraq War marked the coming of age of CNN, as it became the go-to source, and in the early stages the only source, with coverage from inside Baghdad. By 1992 Pew found that about 30 percent of Americans who said they got their presidential election news on TV got it on CNN.28 It was only at that point, with the business model proven and the technological trajectory toward increasing channel capacity settled, that two more 24-hour news channels—Fox and MSNBC—joined the fray in 1996.

Another major media development in the 1970s and 1980s was the emergence of televangelism and evangelical Christian broadcasting. Despite their long presence as part of the American religious experience, in the 1940s and 1950s evangelicals were largely marginalized on the airwaves through the combined efforts of the major mainline religious organizations and the broadcast networks.29 Religious broadcasting during that period was treated as part of the broadcasters’ public interest obligations, which the (p.320) FCC allowed broadcasters to satisfy by airing mainline religious broadcasts on a “sustaining” basis—that is to say, unpaid.30 In 1960–1961, the FCC made two regulatory changes that would ultimately enable televangelism to replace mainline churches on the airwaves. Intense lobbying by evangelical broadcasters led the FCC to change its rules in 1960 to permit broadcasters to treat paid religious programming to count against their public interest obligation. And evangelical broadcasters paid. By the end of the 1970s, paid evangelical programming used this opening to almost entirely displace mainline religious programming on mainstream broadcast stations. In 1961 Congress also passed the All Channel Receiver Act, which required all TV sets to be enabled to receive Ultra High Frequency (UHF), not only the established Very High Frequency (VHF) channels. This change made the largely fallow UHF licenses usable, and these became the basis for a new crop of Christian broadcasters. By 1978 there were 30 new religious stations. More importantly, syndication of the top 10 religious programs, through paid programming, accounted for half the religious programs aired. As cable capacity and adoption expanded, cable reception in turn evened out the quality differential between UHF and VHF stations, and the new possibility of combining cable local distribution with the satellite national syndication that had enabled Turner’s launch of TBS, also enabled Pat Robertson to launch the Christian Broadcasting Network (CBN) a few months later. Robertson had purchased a defunct broadcast license in 1961 and began broadcasting his “700 Club” in 1963. By the mid-1980s CBN’s viewership was third only to CNN and ESPN.31 This rise in audience share for Christian broadcasting coincided with a reorientation of evangelicals toward politics. In support of this reorientation, Robertson changed the format of the “700 Club” in 1980 to include the first investigative journalism and news reporting segments aired by religious broadcast.32 The reach of these broadcasts is a matter of some controversy, but 20 million American households is a reasonable estimate of the number who watched at least some religious programming in the 1980s.33 While televangelism viewership declined in the wake of the scandals that hit televangelists in 1987, most prominently Jim and Tammy Bakker, the new role of evangelicals as a major pillar of the conservative Republican coalition made sure that Christian broadcasting remained an important and distinguishing element of the right-wing media ecosystem—one that came out of a distinctly different moral universe than the framework that underlay the objectivity norms of professional journalism.

Like cable and televangelism, AM talk radio emerged out of an interaction between technological and regulatory changes. The tragic story of how David (p.321) Sarnoff—legendary radio pioneer, head of RCA, and founder of NBC—used litigation and FCC channel allocation proceedings to frustrate and delay adoption of Edwin Armstrong’s superior FM radio technology is oft told and well known.34 It delayed widespread adoption of the technology for decades, well after it drove Armstrong to suicide in 1954. It was only in the 1970s that the number of FM receivers came to equal that of AM receivers, and it was only then that a significant new generation of DJs began experimenting with using the technology, particularly its ability to transmit stereo, to play longer format, non-top-40 music.35 By the early 1980s FM radio far overtook AM, and music in particular was shifting to the higher fidelity, stereo-capable systems. AM needed something new, that did not suffer so much from the technical deficit. That something would turn out to be talk radio. And, as cable did for Turner’s superstation model, satellite distribution to ground stations allowed national syndication on a scale and quality that transmission over copper wire had not, significantly increasing the potential reach of this new format.36

But for political talk radio to emerge, one regulatory piece had to fall in place. Driven by a deep ideological commitment to free markets, FCC Chair Mark Fowler, perhaps known best for his quip that a television is just a “toaster with pictures,”37 led a campaign to repeal the fairness doctrine throughout most of his tenure in office. The doctrine had had a varied life since 1949. Throughout the ’50s and ’60s, first-generation right-wing media outlets saw the fairness doctrine as a direct threat to their ability to use radio. However, Nixon’s use of the FCC and the increasingly sophisticated strategic use of fairness doctrine complaints by both conservatives and liberals alike evened out the partisan effects by the 1970s.38 Fowler built a record of these complaints and worked over years until the FCC repealed the fairness doctrine as inconsistent with the First Amendment. The repeal was completed in 1987.

In 1988 Rush Limbaugh’s daily three-hour shock-jock right-wing talk radio show became nationally syndicated. His visceral, emotional style; his unabashedly partisan exhortations and commentary; and his sheer capacity to sustain three hours a day of programming launched the second-generation right-wing media ecosystem (as well as saving AM radio). By 1990 he was reported to be syndicated on over 300 stations and reaching 5 million listeners a week.39 In 1992 he was reported to have 14 million listeners every week, and launched his syndicated television program Rush Limbaugh Show, which was produced by Roger Ailes.40 That year, Ronald Reagan sent Limbaugh a letter lauding him as “the number one voice for conservatism in our country.”41 In 1993 National Review’s cover anointed Limbaugh “The Leader (p.322) of the Opposition.”42 When Republicans took the House in 1994, Limbaugh was feted as the “majority maker,” an honorary member of the freshman class of the 104th Congress. Newt Gingrich’s former press secretary described Limbaugh as the most important person other than Gingrich in achieving the new Republican majority,43 and Tom DeLay credited him with giving the Republicans their ideological marching orders in the 1993–1994 run-up to their victory.44 By 1996 the Pew report on voter media consumption already treated Limbaugh as one of the major sources of news for voters, noting that Limbaugh “has many more Republicans and twice as many conservatives in his audience than business magazines.” In that survey, 37 percent of respondents said they got news about presidential candidates and the campaign from talk radio, and about 20 percent answered that question with Focus on the Family or the Christian Broadcasting Network.45 The style Limbaugh developed, and has retained since, includes strong emotional appeals to audiences, continuous criticism of mainstream media aimed to undermine trust in media, systematic efforts to undermine trust in government whenever led by Democrats, and policing of Republican candidates and politicians to make sure they toed the conservative line.46 In 1996 Limbaugh abandoned his syndicated television program and focused on radio, where he has been reported (occasionally with skepticism) to attract on the order of 14 to 20 million listeners weekly since the early 2000s.47 Roger Ailes, in the meantime, found a new partner in Rupert Murdoch and launched Fox News.

Talk radio was supercharged beyond Limbaugh by the industry consolidation that followed the loosening of ownership limits in the Telecommunications Act of 1996. That act was the most extensive reform of communications law since 1934. Most of it dealt with enhancing competition in telecommunications. But the act also significantly reduced the ownership limits on radio stations and made mergers easier. Throughout the preceding decades, FCC radio regulation reflected a clear commitment to localism—to the idea that local radio stations should serve the local community—by preferring to license locally owned stations—and to competition, by setting clear caps on how many stations any single company could own, both nationally and locally. As Clinton Democrats came to embrace deregulation alongside Reagan Republicans, both constraints were loosened significantly. The result was a rapid consolidation in the radio station markets. Clear Channel Communications in particular went on a buying spree. By one account the firm owned 43 stations nationwide before the 1996 Act, and eight years later owned over 1,200 stations, reaching over 100 million listeners.48 One of those purchases included Jacor Communications, which, (p.323) like Clear Channel, had started to grow through acquisitions after 1996. But Jacor brought to Clear Channel more than its 230 stations. It also brought Premier Radio, the producer of the radio shows of Rush Limbaugh, Sean Hannity, and Glenn Beck. Hours of right-wing radio propaganda were now seamlessly syndicated throughout the United States; and radio, to this day, is the communications platform that reaches the most U.S. adults daily.49 One politically liberal survey of talk radio in 2007 found that talk radio featured conservative programming over liberal programming by a 10:1 ratio.50 Conservatives gladly acknowledge that talk radio is overwhelmingly their turf.51 Air America, the liberal counterpart to conservative talk radio, largely failed as a commercial enterprise. Democracy Now! succeeded in establishing itself as a listener-supported daily radio broadcast in about 350 cities across the United States, initially on a handful of Pacifica radio stations and now occupying an hour a day, sometimes repeated for another hour. Lacking the benefits of Clear Channel’s distribution network or commercial funding, the program is concentrated in relatively liberal states and regions. In part, the weakness of left-leaning radio may reflect the diversity of the liberal coalition. Conservative talk shows aim at and capture a large and relatively homogenous block—white, Christian, and older. Liberals include Black and Hispanic populations, each seen as its own market segment,52 and liberal radio also had to contend with NPR, an enormously successful radio system that maintains a palpable liberal editorial tone while adhering to mainstream journalistic norms in reporting—in that regard more like a liberal version of the Wall Street Journal, than like Fox News or MSNBC. And in the 2000s liberals tuned into late night comedy, particularly John Stewart and then Stephen Colbert, for their mass media partisan diet. In the news segment of the market, nothing emerged to match Fox News as an alternative partisan news system, dedicated to providing sustained news coverage from a consistently partisan perspective until MSNBC shifted strategy in 2006, a decade after Fox News and nearly 20 years after Limbaugh.

Murdoch and Ailes’s innovation was to capture market share in the 24-hour news channel not by competing on the same model as CNN, but by offering a similar 24-hour news format oriented to leverage the market that Limbaugh had proven—giving right-wing audiences an outlet they did not have within media constrained by professional journalistic norms.53 Fox News combined the format of CNN’s 24-hour news network model with the audience-segmentation strategy that had been developed by the Christian broadcasters and Limbaugh in the 1980s and early 1990s. Rather than serving the median viewer in hopes of capturing a share of it, which (p.324) was the prevailing theory of how to compete in the three-network world of over-the-air broadcast, channel abundance made room for stark audience segmentation as a robust and successful strategy. Within five years of its launch, and particularly after September 11, 2001, Fox News caught up to and then passed its major rival, CNN. By the 2016 election cycle American right-leaning audiences had been exposed for two decades on television (and nearly three on radio) to a propagandist mass media outlet built on feeding its viewers with news that fit and reinforced their world view while constantly pointing fingers at all other media sources as biased. The strategy paid off for Fox in producing an immensely loyal viewership, and for the Republican Party with a core of support highly resilient to the vicissitudes of real-world failure or transient political winds.

Studies by Pew of the news habits of Republicans and Democrats demonstrate the significant difference between the two audiences. A 2014 study found that 47 percent of “consistently conservative” respondents identified Fox News as their “main source of news about government and politics.”54 Another 11 percent mentioned “local radio” as their main source, and in a breakdown of sources of news that meant that Hannity on radio, Limbaugh, Glenn Beck on radio, and Beck’s “The Blaze” all ran ahead of ABC, CBS, or NBC, much less CNN.55 By contrast, “consistently liberal” respondents were spread out much more evenly across various media, including primarily CNN (15 percent), NPR (13 percent), MSNBC (12 percent), and the New York Times (10 percent). Conservatives not only watch and listen to Fox News and talk radio, they also express high trust in these sources. A full 88 percent of consistently conservative viewers trusted Fox News, 62 percent trusted Hannity, and 58 percent trusted Limbaugh. Among “consistently liberal” respondents, NPR, PBS, and the BBC were the most trusted sources, hovering around 70 percent. MSNBC, which since 2006 has mounted the most explicit effort to mirror the Fox News strategy for the left, received only a 52 percent trust score from consistently liberal respondents and was trusted by fewer than half of the “mostly liberal” respondents, all of whom trust CNN and the major television networks to a degree largely consistent with the trust patterns of respondents who were “mixed” liberal and conservative (Figure 11.1).

The Origins of Asymmetry

Figure 11.1 Trust in media by ideological identity.

Source: Pew Research Center.

This pre-election snapshot reflects one moment in a long trajectory of declining and divergent levels of trust in journalism between conservatives and liberals, Republicans and Democrats. Gallup has been measuring distrust in media since the mid-1990s.56 In 1998 and 2000 party identifiers of both parties, and independents, were closest to each other in the level of trust in (p.325) media: 52-53-59 (R-I-D) in 1998, and 47-53-53 (R-I-D) in 2000. Election coverage separated democrats from Republicans in 2001, but 9/11 brought the two groups together in 2002: 39-52-65 in 2001, 49-52-59 in 2002. Republican trust in media dropped over the remainder of George W. Bush’s first term in office, from 49 percent in 2002 to 31 percent in 2004, where it stayed fairly stable until 2015 and the beginning of the 2016 election cycle. In 2016, just before the election, it dropped precipitously to 14 percent and remained at that level in 2017. These patterns of extremely low trust in media are highly consistent with the pattern of attention that we observe in our data—with conservative audiences tweeting and sharing on Facebook stories from within an insular right-wing media ecosystem and largely ignoring most other (p.326) sites. By contrast, Democrats’ trust in media rose over the course of George W. Bush’s first term, from 53 percent in 2000 up to between 60 percent and 65 percent until the first year of the Barack Obama presidency, after which it saw a gradual steady decline, down to 51 percent in 2016, followed by a sharp upswing in 2017 back to 72 percent, its highest value in the Gallup set and closely matched only by the 2005 Gallup poll (which likely coincided with coverage of Hurricane Katrina). Again, this trust is consistent with our behavioral observations about which sites users who are not on the right tend to share and amplify. The category of “independents” that show up in these polls should not in fact be treated as a completely distinct third group. Surveys suggest that this group is made up mostly of Republican and Democratic leaners (slightly more Republicans than Democrats), with only 15 percent of all independents truly not leaning toward one or the other party. Both sets of “lean-party” respondents are very similar to party identifiers of the party toward which they lean.57 Unsurprisingly, independent trends “split the difference” between the responses of party identifiers. The pattern suggests (a) systematically less trust among Republicans throughout the period, and (b) a correlation between trust in media and the party of the president in power. If mainstream media are consistently somewhat critical of whoever happens to be in power, then partisans of that president will see media as less trustworthy whenever that party is in control and more trustworthy when the opposite party is in control.

The longest relevant time series on the question of trust in media is the General Social Survey’s question on confidence in the press. Most revealing is the partisan breakdown of the question that is the inverse of Gallup’s question. Where Gallup reports all those who have “a great deal” or a “fair amount” of trust in media, the GSS also collects data on everyone else—in particular, those who have “hardly any” trust in media (Figure 11.2).58

The Origins of Asymmetry

Figure 11.2 General social survey responses: percent of population responding that they have “hardly any” confidence in the press (1973–2016).58

Several patterns emerge. First, extreme distrust of the press media was very low in the 1970s and the immediate post-Watergate era, although it briefly bounced upward during that era among Republican Party identifiers. The Reagan-Bush years in the 1980s to 1992 saw Democrats jump in their high distrust from 14 percent to about 20 percent, while Republicans saw their distrust rate hover around 30 percent during the 1980s. Republicans saw a clear inflection point from 1990, when their “hardly any” category made up 29 percent of Republican respondents, to 47 percent in 1993, as Bill Clinton became president and on the eve of the Gingrich revolution. Republicans response remained stable until 1998, when it again inflected upward from 44 percent in 1998 to the upper-mid 50s (56–57 percent, up to 60–61 percent) (p.327) from 2002 until the 2016 election. Among Democrats, by contrast, there was an inflection point from 1990 to 1993 (22 to 35 percent), and distrust stayed at that level except for a brief peak (41%) during the Lewinsky affair in 1998 percent (which was also the Republican’s lowest “hardly ever” answer, at 44 percent), from where it gradually declined until 2006. It again shifted upward from a low of 31 percent in 2006 to a high just under 40 percent over the course of the Obama years.

Given the great number of factors involved in broad, population-level trends of trust and distrust, it is impossible to identify a clear causal line from media coverage to levels of trust in media. It is hard, for example, to lay the 1990 dramatic increase of distrust in media at the feet of Rush Limbaugh, given that “hardly ever” responses among Democrats increased by 50 percent (from 22 percent to 35 percent) between 1990 and 1993, a very similar degree of change as that among Republicans during the same time period (29 percent to 47 percent). By contrast, the Republican inflection point in 1998 certainly could reflect a Fox News effect similar to the effect we describe at the end of this chapter. It is difficult to imagine that the sustained attacks on the credibility of the media by political leaders and the partisan press for over two decades have been entirely inert and have not contributed meaningfully to undermining their audience’s confidence in the independent press. And it is difficult to imagine that attacks such as these by Fox News, the news source (p.328) conservatives trust above all other news sources in the country, have no effect on the fact that precisely those conservatives who most trust Fox News also have the lowest trust in all other media. And the result is quite an outlier by international standards. The Reuters Institute found that U.S. media were more starkly polarized than in other countries, primarily in the sense that right-wing viewers did not look at most mainstream publications and that self-described right-leaning survey respondents showed much lower trust in news media than self-described left-leaning respondents.59 The most powerful finding in the Reuters Institute report was that “[i]n the United States the headline rate (38%) is up, although there is a 15-point gap between this and trust in the sources you use (53%). Only Hungary, another deeply polarized country, has a bigger gap between general trust (31%) and the sources you use (54%).”60

Existing in a media ecosystem dominated by media whose role is to confirm your preconceptions and lead you to distrust any sources that might challenge your beliefs is a recipe for misinformation and susceptibility to disinformation. At the end of the day, if one side most trusts Fox News, Hannity, Limbaugh, and Beck, and the other side most trusts NPR, the BBC, PBS, and the New York Times, one cannot expect both sides to be equally informed or equally capable of telling truth from identity-confirming fiction.

As we discussed in more detail in Chapter 3, the propaganda feedback loop that creates the patterns of distrust and media attention and the dynamic way in which media production and consumption patterns feed into each other and into the behavior of political elites, seem sufficient to explain the sustained differences between Democrats and Republicans in their susceptibility to conspiracy theory and rumor.61 The audiences of talk radio, Fox News, and other conservative media outlets are subject to a sustained flow of identity-confirming news and attacks on potential sources of error correction, and are informed by their political elites that certain facts and attitudes are identity consistent and that challenges to these facts reflect bias of the other side, rather than being the products of professional, objective norms. Right-wing audiences are systematically disconnected from potential sources of disconfirmation. As the dissonance between what they receive from their own media and what they receive from outside increases, a deeply asymmetric trust structure develops which associates identity-confirming news as trustworthy and identity-disconfirming news as suspect. This does not require any special psychological profile; it is merely a direct consequence of the architecture of the media system. And it means that a population with high trust in bias-confirming news and high distrust in bias-disconfirming, (p.329) professional-norms-driven media will be more vulnerable to disinformation campaigns than a population that has generally higher trust in professional journalism on average, but lower trust in any given media outlet. The latter population will, on average, check and cross-reference rumors, conspiracy theories, and other modes of disinformation more than the former. And it is onto that baseline asymmetric structure—of media outlets, political elite practices, and media consumption and trust patterns—that the internet and social media were grafted. And, unsurprisingly, the different architectures of the two parts of the media ecosystem resulted in quite different susceptibility to the new techniques of network propaganda. As our case studies in Chapters 3 to 9 showed, Russian propaganda, commercial clickbait entrepreneurs, sockpuppets and botnets, and straight-up partisan disinformation campaigns all operated differently in the right-wing media ecosystem than in the center and left. And whatever solutions we will embrace, if all we do is treat the discrete, social-media-focused manifestations of the underlying structural difference, our answers will be partial and unstable.

What About the Internet?

In March1995 the Drudge Report already had 1,000 email subscribers.62 For perspective, the yahoo.com domain had only been registered in January of that year, and Microsoft would release its first version of Internet Explorer in August. At the time the internet enjoyed wide acclaim as a democratizing technology. In 1997 the Supreme Court hailed it as a platform on which “[t]hrough the use of chat rooms, any person with a phone line can become a town crier with a voice that resonates farther than it could from any soapbox. Through the use of web pages, mail exploders, and newsgroups, the same individual can become a pamphleteer.”63 That image captured what many of us writing about the internet at the time thought. Reduced costs of producing and sharing information, news, and perspectives would, for the first time since the emergence of mass circulation print and broadcast, enable citizens to participate in setting the public agenda and mobilize for action around our intense political concerns, rather than following the agenda set by media owners, the advertisers who paid, and political elites to whom they paid attention.64 In a 1999 article, for example, one of us described a copyright lawsuit by the Washington Post and Los Angeles Times against the conservative Free Republic website, and argued that the amateur, citizen-produced commentary and speech produced on Free Republic offered an important new degree of diversity and free speech in the public sphere that would (p.330) be lost if courts continued to prefer commercial organizations as the core speakers in the networked public sphere. Within a couple of years, however, a counternarrative emerged that blamed the internet for causing fragmentation and polarization,65 phenomena that we, in Chapters 10 and 11, describe as a consequence of longer-term political processes and changes in the radio and television media ecosystems.

The first dozen or so years of the twenty-first century continued and deepened the pattern. Building on new empirical observations that online attention tended to follow power law,66 rather than normal distributions—that is, a very small number of sites garnered a very large portion of the attention—several authors argued that in practice the internet replicates the broadcast model of attention, with a very small number of sites accounting for most of what people read or heard.67 Some empirical studies, most prominently pioneering work by Nancy Glance and Lada Adamic, tended to support Cass Sunstein’s arguments about polarization and fragmentation of political discourse, rather than a reconcentration on a small number of supernodes.68 Importantly, those early efforts to map the blogosphere tended to observe a symmetric pattern of polarization. A different line of argument against the democratizing effects of the internet was that they simply were not that important. Markus Prior emphasized that most people were simply distracted—and enjoyed broad access to entertainment and other formats that left them simply uninformed—leaving only the more partisan and polarized to participate.69 Matthew Hindman used extensive data analysis of blogs and their use to argue that the relative size of the political blogosphere was negligible in the overall scheme of internet usage, and that those who wrote the top blogs were as much a part of the elite as the op-ed writers.70 Nonetheless, as we discuss in more detail in Chapter 12, throughout those years more of the academic work combined qualitative assessment, survey data, and quantitative analysis to argue that the internet had a more democratizing effect, whether emphasizing mobilization, diversity of viewpoints, or organizational transformation as the elements that made for a networked public sphere in which power to set the agenda and influence political discourse was more widely distributed.

Although Drudge and the Free Republic were early adopters, the “Web 2.0” period in the first decade of the twenty-first century saw more or less equal growth on the two sides of the partisan divide online. In one of the first studies that looked systematically at the differences between the left and the right in this period, one of us collaborating with Aaron Shaw showed that the primary differences during this period revolved around organization (p.331) and action orientation.71 Bloggers on the right tended to write many more sole-authored sites that primarily shared links to stories found online with brief comments. The Free Republic was, in this regard, unusual on the right. On the left there were more group blogs and blogs that enabled users to write their own diaries. The left also had much more of a focus on action orientation—raising funds in particular—than the right blogosphere did.72 Daily Kos was the standard-bearer of this format. Howard Dean’s insurgent presidential campaign in 2004 is generally celebrated as the first political campaign supported by the “netroots,” young, mobilized, liberal, internet-based activists, which continued to play a significant role in the 2006 and 2008 campaigns,73 but the decentralized, self-organizing approach was as central to Ron Paul’s supporters on the libertarian right as it was for those of the netroots on the left. Of the highly visible sites during that early Web 2.0 period, several on both sides of the political divide continued to be influential in the 2016 cycle, such as the Daily Kos and Talking Points Memo on the left, or Michelle Malkin, Townhall, and RedState on the right. However, the most influential were either founded in that Web 2.0 period as media ventures, such as Huffington Post (2005), BuzzFeed (2006), or Breitbart (2007), or founded more recently, after 2010. Because the techniques and research questions that have been brought to bear in the last few years were not available at the time, we do not have a systematic study to which we can compare our current results to those that described the blogosphere in its first few years. Studies did not include mainstream media; they described much shorter snapshots of time, and often included blogrolls, as opposed to specifically story links, which means that they reflected to some extent affinity networks rather than authority or attention networks, as ours do. Because of the differences in what was possible and done at the time, we do not know whether the actual linking patterns were similarly asymmetric during the first few years of Web 2.0, or whether the networked public sphere was more symmetric then, unlike on TV and radio. The earliest analysis we have, from our data, goes back to the month before the 2012 election (Figure 11.3). And while the details are slightly different, and the overall architecture is less clearly segregated asymmetrically, it is structurally similar to what we found in our data from 2015 to 2018. We see a well-separated right-wing sphere, though it is more integrated and offers a more prominent place to media outlets we characterized as center-right in 2016, particularly the older conservative publications like the National Review or Weekly Standard. The rest of the media ecosystem was anchored around mainstream media in 2012 as it was in our present study. Because we only captured data for one month (p.332) prior to 2012, this architecture is likely noisy and only indicative of the trend toward the patterns we now observe.

The Origins of Asymmetry

Figure 11.3 Network map of political media in the United States during the 2012 election. Nodes sized by media inlinks.

As social media, in particular Facebook, replaced Web 2.0 technologies central to the blogosphere, the debate took on the familiar structure of the prior decade and a half, triggered most powerfully by the Arab Spring (and before that the Iranian Green Wave moment) and the brief exhilaration associated with seeing a region long-plagued by repressive and corrupt regimes rising up to demand democracy.74 Terms like “liberation technology”75 or “emancipatory politics”76 set the frame for Hillary Clinton’s State Department to adopt its Internet Freedom Agenda as one of its core programs for expanding democracy throughout the world. The Arab Spring, and after it Occupy Wall Street and the Indignados in Spain, were the high-water mark for the argument that the internet democratizes and liberates. The apparent short-term media and attention victories were followed in most places by political losses, while the digital trails, particularly on Twitter, enabled a new generation of data-collection techniques to evaluate the structure of organization in online mobilization. Sandra González-Bailón (p.333) and her collaborators’ research on the Indignados movement showed that the structure of online mobilization was more hierarchical and centered on major influencers as central nodes.77 Mayo Fuster Morell emphasized the ways in which the anti-austerity 15-M Movement in Spain was built on earlier mobilization efforts in the Free-Culture movement.78 Zeynep Tufekci’s analysis of these protest movements, from Tahrir to Gezi Park and Zucotti Park, led to a new ambivalence about the double-edged nature of social media for mobilization and the possibilities it opened for repression.79 Nagla Rizk’s analysis of the post-revolution networked public sphere in Egypt similarly underscored the rising role of both the Egyptian Army and the Muslim Brotherhood on Facebook, and their displacement of the earlier, more liberal Egyptian blogosphere, as central dynamics in the unfolding of the story of the Egyptian revolution,80 with only Tunisia offering a major counterexample.81 In all, as the dust settled from the Occupy movement and the Arab Spring, the liberating potential of decentralized social mobilization began to give way to the consideration of the benefits of more organizationally structured politics.

As networked discourse shifted from the open web to social media, large parts of the media ecosystem became much more opaque. While Twitter opened its data and gave rise to a generation of studies that have taught us most of what we know about discourse and mobilization on social media, Facebook created an enormously powerful platform that was in large parts unobservable by anyone who wanted to understand how the internet was affecting democracy. Anyone, that is, except Facebook scientists. The most significant study of polarization on Facebook was done by a team inside Facebook and published in Science in 2015.82 Using data from the last six months of 2014, the team found that Facebook users who shared news did so in a polarized way, but that they did so primarily because they shared what their friends shared, and they were segregated into politically homogeneous communities. It was users’ sharing patterns, not the design of Facebook’s algorithms that led to the polarization. Setting aside the troubling self-serving nature of a team in Facebook publishing a study that exonerates Facebook from responsibility for polarization in American politics, using data the company would not share with outsiders except under a nondisclosure agreement, the pattern of polarization they describe dovetails quite well with what we observed. We showed that based on Facebook sharing, the two sides of the media ecosystem look more symmetrically polarized than when measured by tweets or inlinks, but nonetheless there was significant asymmetry, with the right being much more skewed. The 2015 Facebook paper similarly showed a broad symmetry, (p.334) but with the more extreme right-wing content accounting for a larger share of right-oriented sharing than the most extreme left-wing sites accounting for sharing on the left, and sharing of sites on the left extending much further toward the middle of the ideological spectrum. The comparison is imperfect, because the Facebook team were not looking specifically to compare, but the published materials offer support for the proposition that the asymmetry we observe clearly since 2015, and was also visible in our 2012 data, was also there in this very large study conducted between our two observation periods.

As we look at the internet and its history, it seems that we have some anecdotal evidence of the early success of right-wing-oriented sites, but little evidence from the first dozen or so years of political speech on the internet of the asymmetries we observe today. We certainly have nothing like the sustained evidence of asymmetry in cable television or talk radio. By the second decade of the twenty-first century, however, we do in fact see such a differentiation, and our data from 2012 and early 2015 suggests that it is already well established by then, rather than that it first emerged during the 2016 presidential campaign. If we understand that online media and offline media interact in an integrated single large media ecosystem, we can also understand that online media on the right and the left faced different incentive structures and different constraints. As mainstream media came online, online left media had to contend with audiences that had a moderate trust in mainstream media and an architecture of use that exposed them extensively to mainstream professional media offline. Selling wild bias-confirming conspiracies risked harming a publication’s reputation, or consigning it to amusement rather than news, because of negative feedback effects from that sphere’s most trusted sources. Online left publications faced the reality-check dynamic. On the right, presented with an audience already highly attuned to talk radio and Fox News and deeply distrusting of all media outside their ecosystem, right-wing sites that came online—Breitbart, Daily Caller, and beyond—faced very different competitive pressures, and positive feedback reinforcement for bias-confirming news, irrespective of its veracity. Heightened partisanship was a positive differentiator, and manifest falsehood (as long as it was only pointed out by mainstream outlets that could be painted as corrupt and lying) was not a particularly costly practice. Indeed, when Hannity amplifies an Infowars story, the falsehood becomes a virtue. Right-wing online media had to adapt to a media ecosystem that was already a decade or two into the propaganda feedback loop. When Breitbart was launched in 2007, it had to compete for an audience that had already listened to Rush Limbaugh for 20 years, and to Fox News for 11. When the Huffington Post launched in 2005, Air America (p.335) was barely a year old, and MSNBC had not yet shifted from trying to copy CNN to trying to differentiate as the mirror image of Fox.

It is not hard to explain why a population exposed to news outlets whose core professional commitment is to provide partisan bias-confirming news will be more susceptible to partisan identity-consistent rumor, conspiracy theory, and disinformation than a population whose media consumption habits are anchored in media committed to fact checking and objective reporting. Those who exist in a media ecosystem committed to objectivity (and the commitment need not be perfectly followed; it merely needs to restrict the practices in the main) will usually encounter no confirmation, or refutation, of the false information in the media they trust the most. By contrast, those who pay attention to the partisan media will not generally receive disconfirmation and will instead often receive confirmation and amplification of the false facts. When the Petra News Agency in Jordan was hacked and had a story planted supposedly exposing an interview with the Saudi now–crown prince boasting that the Saudis had funded 20 percent of Hillary Clinton’s campaign, the story was reported on the Zero Hedge and then republished on the Fox News website and Infowars.83 In addition to reinforcing the conspiracy theory fringe claims, the partisan media persistently reiterate for their readers, viewers, and listeners that the mainstream, professional media are the ones who are biased, partisan, and “fake,” and that only they, the partisan media, are in fact those who offer “fair and balanced” reporting. This makes the corrective role of professional institutions asymmetrically available to the two populations.

We recognize that our findings and the media history we review could be explained by the work in political psychology that follows from the rigidity model developed by John Jost and collaborators.84 Synthesizing decades of work stretching back to the 1950s, Jost and his coauthors proposed a theory that suggested that people adopted conservative political positions as a way of satisfying a particular combination of cognitive style, personality traits, and epistemic and existential needs. Many of these could certainly provide a simple answer to why Limbaugh or Fox News succeed on the right to a degree that Al Franken or MSNBC never did on the left. If dogmatism, intolerance of ambiguity, and a need for closure, order, and structure are strong predictors of politically conservative attitudes, these certainly would also predict a rapid conversion on a single authoritative source of news that offers only confirmation of belief-consistent news and reassures its viewers and listeners that other sources that raise doubt are liars. The observed patterns of media consumption on the right and the left are also consistent (p.336) with this model. In particular, the significantly narrower menu of media sources used by consistently conservative audiences, and the high trust these respondents report in their own media sources combined with their high distrust of other sources, even when those other sources are highly trusted by everyone else, including nonpartisans of mixed views, are all consistent with the rigidity model.

Our own research does not, however, incorporate any measures that could deny or refute this effect. The rigidity model has long been contested by psychologists who view it as too partisan in its research orientation, and recent work in particular raises concerns about several core methodological concerns, in particular whether it captures the range of positions that count as conservative, particularly the overlap between economic and identity issues. More directly relevant to our findings, the critique questions whether the rigidity model can account for exposure of study subjects to discourse that would shape the alignment of their political answers more than their personality or cognitive style.85

Neither the propaganda feedback loop model we describe in Chapter 3 nor the political economy story we offer here for how that feedback loop emerged on the right in the United States between the 1980s and 2000s, depend on a psychological theory. They merely require that one side of the political ecosystem begin to adopt such a strategy before the other. The positive feedbacks between the benefits to elites, who gain a reliable audience, the benefits to the broadcasters, who gain a loyal market segment, and the beliefs and attitudes of the population begin to reinforce each other within the feedback loop. As we explained in Chapter 3, the mainstream media then become a source of confirmation for the opposite side of the political spectrum because they, on average, will find more lying in the bias-confirming media (because there will be more of it there, by design) and provide its opponents with truth-based reasons to reject their opponents claims. The side that adopts the hyperpartisan media strategy enjoys, in this regard, a first-mover advantage in a positive feedback dynamic that then both reinforces and disciplines those who are in it. Certainly, endowing the population of the side that chooses the bias-confirming media with a baseline personality that increases the prevalence of preferences for such media would make the tipping easier and faster. But the model does not require a differential psychological makeup for the propaganda feedback loop to emerge alongside, and in opposition to, media that continue to operate under the reality-check dynamic. Once the population being studied is pervasively served by a highly asymmetric media system, studies that depend on responses from that (p.337) population will have to control for the asymmetric effects of media exposure and discourse on shaping the political beliefs of respondents.

Fox News vs. Internet: Political Impact Assessments

Moving from identifying patterns of media practice and consumption to outcomes is hard. We are most persuaded by three studies by economists that look at the effect of Fox News, and at the effects of the internet relative to Fox News, over the past 20 years. These studies all support the proposition that Fox was the more important component of the right-wing media ecosystem, even in the 2016 election.

Over the past decade, two major studies have sought to quantify the “Fox News Effect” on America political polarization. The work was pioneered by Stefano DellaVigna and Ethan Kaplan in an article published in the Quarterly Journal of Economics.86 Because the cable industry and cable channel offerings were growing and changing fast in the late 1990s, Fox News was introduced at different times in different cities. In particular, Fox News’s successful completion of carriage deals with major multiple-system operators like TCI (later bought by AT&T) sped up its adoption, while its failure to negotiate carriage on other cable carriers like Adelphia (taken over by Time Warner after its bankruptcy) delayed its adoption. This variation between towns in the timing of Fox News entering the market provided a natural experiment for the effect of Fox News on voter turnout and partisan voting. DellaVigna and Kaplan were able to show that introducing Fox News to a town increased the Republican vote share by 0.4 percent to 0.7 percent, and estimated that because it likely was responsible for about 200,000 more Republican votes nationwide in the 2000 election. In particular, they estimated that Fox News was likely responsible for slightly over 10,000 votes in Florida in an election won by George W. Bush by 537 votes. Because the effect was present even for Senate seat elections that were too local to be covered by a national cable channel like Fox News, DellaVigna and Kaplan argued that the Fox News effect acted at the general ideological activation level, rather than specific candidate support, and was particularly significant in increasing Republican turnout in otherwise Democratic districts.

DellaVigna and Kaplan’s work covered the first election after Fox News began to play a significant role. By 2017, two other economists, Gregory Martin and Ali Yurukoglu, were able to replicate these results for 2000, improve their precision, and extend the analysis to cover the 2004 and 2008 results as well.87 Martin and Yurukoglu combine several of the most sophisticated techniques (p.338) available to analyze the partisan bias and effects of media on voting patterns. They used a novel instrumental variable (the placement of each of the 24-hour news channels on the dial) combined with text analysis to measure the degree of partisan slant of the program transcripts from CNN, Fox News, and MSNBC over the course of the 2000, 2004, and 2008 elections, applying a technique developed by Matthew Gentzkow and Jesse Shapiro to measure media bias in newspapers.88 To these they add voter data, surveys, and Nielsen data to provide the most detailed and robust evidence of the central role Fox News has played in the partisan polarization of American media and voting patterns. In this study, Martin and Yurukoglu replicated and strengthened DellaVigna and Kaplan’s original finding that Fox News made a critical difference in 2000. They further showed that because it increased both its viewership and its partisan slant, the effect of Fox News became more pronounced over time. While their data picked up the strategic shift MSNBC made in 2006 to try to mirror the Fox News strategy on the left, they did not find that MSNBC had a meaningful effect on voting behavior. Most importantly, Martin and Yurukoglu found that the primary effect of exposure to Fox News was on voters who started the election cycle as centrists or Democrats, and over the course of the election cycle were converted to Republicans. While this effect declined from 2000 to 2004 and to 2008, it was substantially higher than the impact MSNBC had on converting viewers who were originally Republicans.

Looking at the 2016 election, anyone assessing the impact of cable versus that of the internet has to contend with the fact that Trump supporters were, overall, primarily to be found in demographic groups that were the least attentive to online news sites and social media. Trump voters overwhelmingly reported tuning in to television: Fox News (40 percent), the networks (12 percent), and local TV (5 percent). Facebook was the primary source for only 7 percent.89 Moreover, the relationship between social media and internet use for election news and support for Trump and Clinton, respectively, across age groups and gender raises doubt about the role of online media in Trump’s rise. If anything, it seems use of social media is inversely related to support for Trump. Of white voters in the 18–29 age range, whose social media and internet use for election news was 53 percent,90 fewer than 40 percent voted for Trump.91 Trump’s share of the vote increased with age. Trump’s share of the white vote was over 54 percent in the 45–64 age range, and over 58 percent in the over 65 group, while social media use in these demographics was between 5 percent and 1 percent, respectively. Similarly, women reported using Facebook for news much more often than men do, by (p.339) a ratio of 62 percent to 38 percent,92 and supported Clinton over Trump by a healthy 12-point difference.93 In another recent paper, Levi Boxell, Matthew Gentzkow, and Jesse Shapiro exploited these significant age-based differences in partisanship and internet use in a formal analysis of the contribution of the internet to polarization. Looking at eight different measures used across political science to measure partisan polarization, and at data ranging back to 1972, they found that most of the polarization that has happened since 1996, the first year for which there is regularly collected data on internet use, occurred in the populations least likely to be internet connected (those aged 65 and over), while polarization of those most likely to be online and use social media has been fairly stable over this period.94 It is simply unreasonable to pin the blame for patterns of trust and distrust in media, or the rise of Trump, on a medium that is consistently used less by demographic groups that express that distrust or support the president and is used more by populations that hold the opposite position on both questions. (p.340)


(1.) Quoted in: Nicole Hemmer, Messengers of the Right: Conservative Media and the Transformation of American Politics, Politics and Culture in Modern America (Philadelphia: University of Pennsylvania Press, 2016), 157.

(2.) Jeffrey M. Berry and Sarah Sobieraj, The Outrage Industry: Political Opinion Media and the New Incivility, Studies in Postwar American Political Development (Oxford; New York: Oxford University Press, 2014), 12–13; “November 2012 Ratings: Fox News Daytime Shows Have Best Months Ever,” accessed February 14, 2018, http://www.adweek.com/tvnewser/november-2012-ratings-fox-news-daytime-shows-have-best-months-ever/157704.

(3.) A.J. Katz, “November 2016 Ratings: Fox News Has Most-Watched Month Since 2012,” Adweek, November 30, 2016, http://adweek.it/2gWcPfF.

(4.) A.J. Katz, “November 2017 Ratings: Fox News Is Basic Cable’s Most-Watched Network For 17 Months,” November 29, 2016, http://adweek.it/2i1MT0L.

(5.) Jeffrey Gottfried, Michael B1arthel, and Amy Mitchell, “Trump, Clinton Voters Divided in Their Main Source for Election News,” Pew Research Center’s Journalism Project (blog), January 18, 2017, http://www.journalism.org/2017/01/18/trump-clinton-voters-divided-in-their-main-source-for-election-news/.

(7.) Id. Berry and Sobieraj, at 39–41.

(8.) Id. Berry and Sobieraj, at 42–45.

(9.) Joanne M. Miller, Kyle L. Saunders, and Christina E. Farhart, “Conspiracy Endorsement as Motivated Reasoning: The Moderating Roles of Political Knowledge and Trust,” American Journal of Political Science 60, no. 4 (2016): 824–844, https://doi.org/10.1111/ajps.12234. Adam J. Berinsky, “The Roots of False Beliefs: Political Rumors in America from 2010–2017” draft chapter.

(10.) Theda Skocpol and Vanessa Williamson, The Tea Party and the Remaking of Republican Conservatism (New York: Oxford University Press, 2016), 125–126.

(11.) Michael Schudson, Discovering the News: A Social History of American Newspapers, (New York, NY: Basic Books, 1995).

(12.) David Harvey, The Condition of Postmodernity: An Enquiry into the Origins of Cultural Change (Oxford, [England]; Cambridge, MA, USA: Blackwell, 1989); James C. Scott, Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed, Nachdr., Yale Agrarian Studies (New Haven, CT: Yale University Press, 2008); Charles S. Maier, “Between Taylorism and Technocracy: European Ideologies and the Vision of Industrial Productivity in the 1920s,” Journal of Contemporary History 5, no. 2 (1970): 27–61; Charles S. Maier, “The Politics of Productivity: Foundations of American International Economic Policy after World War II,” International Organization 31, no. 04 (1977): 607–633.

(13.) Masses Publishing Co. v. Patten, 244 F. 535 (S.D.N.Y. 1917).

(14.) Schenck v. United States, 249 U.S. 47 (1919).

(15.) Abrams v. United States, 250 U.S. 616 (1919).

(16.) Sheldon Marcus, Father Coughlin: The Tumultuous Life of the Priest of the Little Flower, 1st ed. (Boston: Little, Brown, 1973).

(17.) Institute for Propaganda Analysis, Alfred McClung Lee, and Elizabeth (Briant) Lee, eds., The Fine Art of Propaganda (New York: Octagon Books, 1972).

(18.) Victor Pickard “‘Whether the Giants Should Be Slain or Persuaded to Be Good’ ”: Revisiting the Hutchins Commission and the Role of Media in a Democratic Society,” Critical Studies in Media Communication Vol. 27, No. 4, (October 2010): pp. 391–411.

(19.) Ira Katznelson, When Affirmative Action Was White: An Untold History of Racial Inequality in Twentieth-Century America (New York; London: W.W. Norton, 2006).

(21.) Id. Hemmer, at 3–48.

(22.) Id. Hemmer, at 167–170.

(23.) Tom W. Smith, Peter Marsden, Michael Hout, and Jibum Kim, General Social Surveys, 1972–2016 [machine-readable data file]. Data accessed from the GSS Data Explorer website at http://gssdataexplorer.norc.org.

(24.) Edward S. Herman and Noam Chomsky, Manufacturing Consent: The Political Economy of the Mass Media (New York: Pantheon Books, 2002).

(25.) Ben H. Bagdikian, The Media Monopoly, 5th ed. (Boston: Beacon Press, 1997).

(26.) Yochai Benkler, The Wealth of Networks: How Social Production Transforms Markets and Freedom (New Haven, CT: Yale University Press, 2006).

(27.) Stanley M. Besen and Robert W. Crandall, “The Deregulation of Cable Television,” Law and Contemporary Problems 44, no. 1 (1981): 77–124, https://doi.org/10.2307/1191387.

(28.) “Campaign ’92: Survey VIII” (Pew Research Center, July 8, 1992), http://www.people-press.org/1992/07/08/campaign-92-survey-viii/.

(29.) Peter G. Horsfield, Religious Television: The American Experience (New York: Longman Press, 1984), 8.

(30.) Jeffrey K. Hadden and Anson D. Shupe, Televangelism: Power and Politics on God’s Frontier (New York: H. Holt, 1988), 8–10.

(31.) Id. Jeffrey K. Hadden and Anson D. Shupe, at 138.

(32.) Razelle Frankl, Televangelism: The Marketing of Popular Religion (Carbondale: Southern Illinois University Press, 1986), Table 2, 92. Razelle Frankl, 118.

(33.) Jeffrey K. Hadden and Anson D. Shupe, Televangelism: Power and Politics on God’s Frontier, 155–159.

(34.) Paul Starr, The Creation of the Media: Political Origins of Modern Communications (New York: Basic Books, 2004), 382–383; Tim Wu, The Master Switch: The Rise and Fall of Information Empires, 1st Vintage Books ed. (New York: Vintage Books, 2011), chap. 9.

(35.) Brian Regal, Radio: The Life Story of a Technology, Greenwood Technographies (Westport, CT: Greenwood Press, 2005).

(36.) Kathleen Hall Jamieson and Joseph N. Cappella, Echo Chamber: Rush Limbaugh and the Conservative Media Establishment (Oxford; New York: Oxford University Press, 2008), 45–46.

(37.) Brent Staples, “Just a Toaster with Pictures,” The New York Times, February 8, 1987, https://www.nytimes.com/1987/02/08/books/just-a-toaster-with-pictures.html.

(39.) Lewis Grossberger, “The Rush Hours,” The New York Times Magazine, December 16, 1990, https://www.nytimes.com/1990/12/16/magazine/the-rush-hours.html.

(41.) Id. Hemmer, at 270–271.

(42.) Id. Hemmer, at 271.

(43.) Jamieson and Cappella, Echo Chamber, xiv; Hemmer, Messengers of the Right, 171.

(44.) Jamieson and Cappella, Echo Chamber, 46.

(45.) “TV News Viewership Declines,” TV News Viewship Declines (Pew Research Center, May 13, 1996), http://www.people-press.org/1996/05/13/other-important-findings-12/.

(46.) Jamieson and Cappella, Echo Chamber.

(47.) Id. Jamieson and Cappella, at 46; “Is Rush Limbaugh’s Business Model in Trouble?,” Politico Magazine, accessed February 9, 2018, https://www.politico.com/magazine/story/2016/05/is-rush-limbaugh-in-trouble-talk-radio-213914; “Rush Limbaugh’s Audience May Be So Much Smaller Than You Think”—Business Insider, accessed February 9, 2018, http://www.businessinsider.com/rush-limbaughs-audience-may-be-so-much-smaller-than-you-think-2012-3.

(48.) Maria Figueroa, Damone Richardson, and Pam Whitefield, The Clear Picture on Clear Channel Communications, Inc.: A Corporate Profile (Ithaca: ILR School, Cornell University, 2004).

(49.) “State of the Media—Audio Today 2017” (Nielsen, June 22, 2017), 3, http://www.nielsen.com/us/en/insights/reports/2017/state-of-the-media-audio-today-2017.

(50.) John Halpin et al., “The Structural Imbalance of Political Talk Radio,” June 22, 2007, https://cdn.americanprogress.org/wp-content/uploads/issues/2007/06/pdf/talk_radio.pdf.

(51.) Skocpol and Williamson, The Tea Party and the Remaking of Republican Conservatism, 125–126.

(52.) “Audio Today: Radio 2016—Appealing Far and Wide,” (Nielsen, February 25, 2016), http://www.nielsen.com/us/en/insights/reports/2016/audio-today-radio-2016-appealing-far-and-wide.

(53.) Bruce Bartlett, “How Fox News Changed American Media and Political Dynamics,” SSRN Electronic Journal, June 3, 2015, http://dx.doi.org/10.2139/ssrn.2604679.

(54.) Amy Mitchell et al., “Political Polarization & Media Habits,” Pew Research Center’s Journalism Project (blog), October 21, 2014, http://www.journalism.org/2014/10/21/political-polarization-media-habits/.

(56.) Gallup Inc., “Democrats’ Confidence in Mass Media Rises Sharply From 2016,” Gallup.com, accessed March 1, 2018, http://news.gallup.com/poll/219824/democrats-confidence-mass-media-rises-sharply-2016.aspx.

(57.) Samantha Smith, “5 Facts about America’s Political Independents,” Pew Research Center (blog), July 5, 2016, http://www.pewresearch.org/fact-tank/2016/07/05/5-facts-about-americas-political-independents/.

(58.) Tom W. Smith, Peter Marsden, Michael Hout, and Jibum Kim. General Social Surveys, 1972–2014 [machine-readable data file]/Principal Investigator, Tom W. Smith; Co-Principal Investigator, Peter V. Marsden; Co-Principal Investigator, Michael Hout; Sponsored by National Science Foundation. NORC ed. (Chicago: NORC at the University of Chicago) [producer and distributor].

(59.) Nic Newman with Richard Fletcher, Antonis Kalogeropoulos, David A.L. Levy, and Rasmus Kleis Nielsen, Reuters Institute Digital News Report 2017, http://www.digitalnewsreport.org/.

(60.) Id. at 19.

(61.) Miller, Saunders, and Farhart, “Conspiracy Endorsement as Motivated Reasoning: The Moderating Roles of Political Knowledge and Trust.”; Adam J. Berinsky, “The Roots of False Beliefs: Political Rumors in America from 2010–2017” draft chapter.

(62.) Blumenthal v. Drudge and America Online, Inc., April 22 1998. https://epic.org/free_speech/blumenthal_v_drudge.html.

(63.) Reno v. American Civil Liberties Union, 521 U.S. 844, 870 (1997).

(64.) Yochai Benkler, “Free as the Air to Common Use: First Amendment Constraints on Enclosure of the Public Domain,” New York University Law Review 74, no. 2 (p.435) (May 1999): 354–446; Yochai Benkler, “From Consumers to Users: Shifting the Deeper Structures of Regulation Toward Sustainable Commons and User Access” Federal Communications Law Journal 52 (2000): 561–579.

(65.) Cass R. Sunstein, Republic.Com (Princeton, NJ: Princeton University Press, 2001).

(66.) Albert-László Barabási and Réka Albert, “Emergence of Scaling in Random Networks,” Science 286, no. 5439 (October 15, 1999): 509–512, https://doi.org/10.1126/science.286.5439.509.; Lada A. Adamic, Lada A., and Bernardo A. Huberman., “Power-Law Distribution of the World Wide Web.,” Science 287, no. 5461 (2000): 2115–2115.

(67.) Erzsébet Ravasz and Albert-László Barabási, “Hierarchical Organization in Complex Networks,” Physical Review E 67, no. 2 (February 14, 2003), https://doi.org/10.1103/PhysRevE.67.026112.; Matthew Scott Hindman, The Myth of Digital Democracy (Princeton, NJ; Oxford: Princeton University Press, 2009).

(68.) Lada A. Adamic and Natalie Glance, “The Political Blogosphere and the 2004 U.S. Election: Divided They Blog,” in Proceedings of the 3rd International Workshop on Link Discovery, 2005, https://doi.org/10.1145/1134271.1134277.

(69.) Markus Prior, Post-Broadcast Democracy: How Media Choice Increases Inequality in Political Involvement and Polarizes Elections, Cambridge Studies in Public Opinion and Political Psychology (New York: Cambridge University Press, 2007).

(70.) Hindman, The Myth of Digital Democracy.

(71.) Others who had looked at this question on a smaller scale or more based on experience were Bowers, C., & Stoller, M. (2005)., Emergence of the Progressive Blogosphere: A New Force in American Politics, New Politics Institute (2005). Available at: http://www.newpolitics.net/node/87?full_report=1; David Karpf, (2008b). “Understanding blogspace.,” Journal of Information Technology and Politics 5, (no. 4) (2008b): 369–385. M.R. Kerbel, M. R. (2009). Netroots: Online Progressives and the Transformation of American Politics. (Boulder, Colo.CO: Paradigm Publishers, 2009).

(72.) Aaron Shaw and Yochai Benkler, “A Tale of Two Blogospheres: Discursive Practices on the Left and Right,” American Behavioral Scientist 56, no. 4 (April 2012): 459–487, https://doi.org/10.1177/0002764211433793.

(73.) Micah L. Sifry, The Big Disconnect Why the Internet Hasn’t Transformed Politics (Yet) (New York: OR Books, 2014).

(74.) Zeynep Tufekci and& Christopher Wilson, “Social media and the decision to participate in political protest: Observations from Tahrir Square,” Journal of Communication 62, no. (2) (2012): 363–379 (2012).

(75.) Larry Diamond, “Liberation Technology,” Journal of Democracy 21 (July 2010): 69–83.

(76.) Philip N. Howard, The Digital Origins of Dictatorship and Democracy: Information Technology and Political Islam, Oxford Studies in Digital Politics (Oxford; New York: Oxford Univesity Press, 2010).

(77.) Sandra González-Bailón, Javier Borge-Holthoefer, and Yamir Moreno, “Broadcasters and Hidden Influentials in Online Protest Diffusion,” American Behavioral Scientist 57 (July 2013): 943–965.

(78.) Mayo Fuster Morell, “The Free Culture and 15M Movements in Spain: Composition, Social Networks and Synergies,” Social Movement Studies 11, no. 3–4 (August 1, 2012): 386–392, https://doi.org/10.1080/14742837.2012.710323.

(79.) Zeynep Tufekci, Twitter & Teargas (2016).

(80.) Nagla Rizk, Lina Attalah, and Nadine Weheba, The Networked Public Sphere and Civic Engagement in Post-2011 Egypt: A Local Perspective., http://www.arabnps.org/egypt/.

(81.) Escander Nagazi, Jazem Halioui, and Fares Mabrouk, An Accelerated Story of the Emergence and Transformation of the Networked Public Sphere: The Case of Tunisia, http://www.arabnps.org/tunisia/.

(82.) E. Bakshy, S. Messing, and L.A. Adamic, “Exposure to Ideologically Diverse News and Opinion on Facebook,” Science 348, no. 6239 (June 5, 2015): 1130–1132, https://doi.org/10.1126/science.aaa1160.

(84.) John T. Jost et al., “Political Conservatism as Motivated Social Cognition.,” Psychological Bulletin 129, no. 3 (2003): 339–375, https://doi.org/10.1037/0033-2909.129.3.339.

(85.) Ariel Malka, Yphtach Lelkes, and Christopher J. Soto, “Are Cultural and Economic Conservatism Positively Correlated? A Large-Scale Cross-National Test,” British Journal of Political Science, May 30, 2017, 1–25, https://doi.org/10.1017/S0007123417000072.; Ariel Malka, Yphtach Lelkes, and Nissan Holzer, “Rethinking the Rigidity of the Right Model,” inJarret T. Crawford and Lee J. Jussim, eds., The Politics of Social Psychology, Frontiers of Social Psychology (NewYork: Routledge, Taylor & Francis Group, 2018).

(86.) Stefano DellaVigna and Ethan Kaplan, “The Fox News Effect: Media Bias and Voting,” 122(3) QJE 122 no. 3 (August 2007): 1187–1234.

(87.) Gregory J. Martin and Ali Yurukoglu, Bias in Cable News: Persuasion and Polarization., NBER Working Paper, April 5, 2017.

(88.) Matthew Gentzkow and Jesse M. Shapiro, “What drives media slant? Evidence from US daily newspapers,” Econometrica, 2010, 78 no. (1), (2010): 35–71.

(91.) “How the 2016 Vote Broke Down by Race, Gender, and Age—Decision Desk HQ,” accessed April 6, 2018, https://decisiondeskhq.com/data-dives/how-the-2016-vote-broke-down-by-race-gender-and-age/. (based on CCES Common Content 2016).

(92.) Elisa Shearer and Jeffrey Gottfried, “News Use Across Social Media Platforms 2017,” Pew Research Center’s Journalism Project (blog), September 7, 2017, http://www.journalism.org/2017/09/07/news-use-across-social-media-platforms-2017/.

(93.) Alec Tyson and Shiva Maniam, “Behind Trump’s Victory: Divisions by Race, Gender, Education,” Pew Research Center (blog), November 9, 2016, http://www.pewresearch.org/fact-tank/2016/11/09/behind-trumps-victory-divisions-by-race-gender-education/.

(94.) Levi Boxell, Matthew Gentzkow, and Jesse M. Shapiro, PNAS 114, no.(40) (October 3, 2017): 10612–10617 (Oct. 3, 2017), http://www.pnas.org/cgi/doi/10.1073/pnas.1706588114.