Last week, President Donald Trump assailed CNN reporter Jim Acosta and suspended his press pass, echoing past comments that the media is the “enemy of the people.” In this context, I sat down with Yochai Benkler, Berkman Professor of Entrepreneurial Legal Studies at Harvard Law School and faculty codirector of Harvard’s Berkman Klein Center for Internet & Society, to discuss the media’s role in the polarization of U.S. politics. Our conversation drew on Benkler’s recent book, Network Propaganda: Manipulation, Disinformation, and Radicalization in American Politicscoauthored with Robert Faris and Hal Roberts.

Citing their study about the creation and sharing of news stories surrounding the 2016 election, Benkler argues that online platforms are not primarily to blame for spreading misinformation and radicalizing the electorate. Our conversation focused instead on the rise of right wing traditional media and the new norms he believes mainstream media should adopt.

—Deborah Chasman

Deborah Chasman: The book focuses on the 2016 election and what made the public sphere so vulnerable to what you call “disinformation, propaganda, and just sheer bullshit.” You resist the idea that technology was the primary driver of that problem, that the manipulation of Facebook’s platform, the Russian intervention, and fake news led to a Trump victory. What did you find in your research that led you to challenge the now common story that extreme polarization has been technologically driven?

The fact is that this twenty-two-year-old dynamic of polarization can’t easily be associated with the Internet.

Yochai Benkler: What really shaped our interpretation was the data. We analyzed just under four million stories online about the election or national politics, published between April of 2015 and the one year anniversary of the Trump presidency, using Media Cloud, a media ecosystem analysis system that we at Harvard’s Berkman Klein Center have been developing with our colleagues at MIT’s Center for Civic Media for a decade. We analyzed how these stories linked to each other—that is, how media producers cited one another as authority or sources, and combined these with text analysis to understand what these stories were talking about and when. These gave us insight into the supply side of political news in the United States. We also analyzed how these stories were tweeted and shared on Facebook, from which we inferred audience attention patterns; and both on the supply and demand sides, we performed network analysis to map the architecture of authority and attention in the U.S. media ecosystem. To these we added detailed case studies of particular controversies, such as how the Clinton emails or Clinton Foundation were covered, or how the Trump Russia investigation was covered during 2017. For these case studies we added text analysis of television coverage to our analysis of online communications.

The data was not what we expected. There were periods during the research when we were just working on identifying—as opposed to assessing—the impact of Russians, and during those times, I thought it might really have been the Russians. But as we analyzed these millions of stories, looking both at producers and consumers, a pattern repeated again and again that had more to do with the traditional media than the Internet.

This finding is complemented by very interesting recent work by other scholars that measures polarizations in attitudes since 1996—the year Fox News was launched—and shows that almost all of the polarization happens demographically in populations that use the Internet and social media less, rather than more. These are people who watch TV and listen to talk radio. So, if the most polarized population uses the Internet and social media the least, to suddenly point a finger at technology says more about our anxieties about the rate of technological change than about what has actually happened to us. The fact is that this twenty-two-year-old dynamic of polarization can’t easily be associated with the Internet.

DC: So the book rejects the older idea that new media symmetrically polarized the electorate.

YB: Yes. If you go back ten or twenty years, remember that some people were relatively optimistic about the Internet and the idea that it would democratize and allow people to really express themselves. I was more in that camp, although I raised a variety of concerns. Others were fundamentally skeptical from the start, particularly with the idea that infinite channels would allow people to tune into only what they would agree with, and we would lose the ability to have a shared agenda. The Internet would essentially just reinforce our own beliefs—and that process would happen symmetrically across the range of views. In the first decade of the twenty-first century, relatively simpler data analysis techniques could be interpreted as supporting that view.

But the new data that we have makes it extremely difficult to continue to hold that position.

The critical transformation happened when media on the right became a multimillion-dollar business that sells outrage—and the real entrepreneur of that transformation was Rush Limbaugh.

DC: Instead you argue that the polarization is asymmetric and not primarily driven by technology. And you make the case that it should not even be called polarization, but radicalization.

YB: Unfortunately, by the end of 2018, to describe what is going on as “radicalization,” as we do, is almost trite. The critical thing that we insist on when we use the word radicalization is that we are not talking just about the crazies. We have an insular right-wing media ecosystem (Fox News, Breitbart, the Washington Times, Daily Caller, and the Gateway Pundit, for example) that has spun out of control and created a propaganda feedback loop, in which what is true or false is entirely beside the point. Its defining characteristic is pushing content that reinforces identity and political in-group membership. To contrast, the left-wing media, which includes outlets such as Daily Kos, Mother Jones, and HuffPost, is part of a single media ecosystem, in which both producers and consumers of news pay attention to a diverse media diet primarily anchored in traditional mainstream media—the New York Times, the Washington Post, CNN, and which stretches all the way to editorially conservative mainstream publications such as the Wall Street Journal and Forbes. In most cases, the left-wing outlets share the reporting and journalistic traditions of mainstream media, and even where they do not, they are constrained in how far they can stray from the truth by the fact that their audiences pay significant attention to these media. So the two wings of the media ecosystem are not operating under the same rules.

This was not always the case. There was a period from about 1940 (when Father Coughlin gets pushed off the air on the eve of World War II for spouting the America First Committee’s pro-fascist views) until 1988 (when Rush Limbaugh becomes nationally syndicated) when both left and right media existed, but they existed on an almost noncommercial basis. There was the National Review and the Manion Forum on the right, and The Nation and the New Republic on the left. But all of these were relatively low circulation, living mostly off of some combination of subscriptions and donations from major supporters. Both sides were relatively peripheral to the mainstream debate, and the mainstream media, both TV and print, operated under what were then still relatively new standards of professional journalism. The critical transformation happened when media on the right became a multimillion-dollar business that sells outrage and stokes hatred. The innovator and the real entrepreneur of that transformation was Rush Limbaugh.

In the book, we look at the history of what made Limbaugh, and then Fox News, possible. Technology (advances in FM radio, satellite, and cable technology) is part of the story, but not all. Rather, cable deregulation in the 1970s and ’80s created the conditions for the emergence of a much wider array of channels, and deregulation of broadcasters allowed for new formats to emerge. As Christian fundamentalists became politicized in the 1970s in response to the women’s movement, for example, televangelists emerged using paid programming that had become legal as the FCC loosened broadcasters’ public interest obligations. Combined with cable and satellite syndication, Christian Broadcasting Network was the third most subscribed and most watched cable TV station in the 1980s. Limbaugh’s syndication, in turn, was only possible with the repeal of the fairness doctrine in 1987.

For me, the single most important audience for this book is the professional newsroom.

In an environment where you have lots more channels, media companies moved away from trying to serve the middle of the audience to thinking about segments. Executives might say, “If I can grab 20 to 25 percent of the population, and give them something unique, I’m better off than trying to fight with eight other channels for a share of the average viewer/listener.” That is what Rush Limbaugh found, and that is what Fox News copied, and that is why, to this day, in the multichannel environment, Fox is the most successful cable channel even though its ideology limits its audience. Everybody else has to compete for audiences across a broader range.

So the critical change was the emergence of a business model that stokes anger and outrage and reinforces identity-based politics just when the Republican coalition began to lean heavily on two deeply identity-based communities, namelyt, angry whites in the wake of the civil rights movement and the religious right pushing back against the women’s movement. These two audiences form the core of the outrage industry. By the time Internet sites such as Breitbart came around in 2007, you couldn’t compete on the right if you did not follow this model.

DC: So, there was a new business model on the right, but there was a different journalistic standard too.

YB: It is important to recognize that, from the start, journalism had nothing to do with the format that emerged in the right-wing media. Limbaugh himself is all commentary. And Fox News very early on has this pretend news component, but with a very strong emphasis on opinion personalities. If you look at the programs today, it is clear that the straight news programs are not popular on the right. Instead it is Sean Hannity, Tucker Carlson, Laura Ingraham, which blend opinion and editorial. It is all identity confirmation all the time.

We call this core dynamic the propaganda feedback loop: you cannot afford to do anything else in that media ecosystem—you lose audience if you push back with facts that do not conform to the identity. That is what brings people to you. What differentiates your product is that you tell people it is okay to be angry, it is okay to hate, it is OK to not believe anyone who tells you something that doesn’t fit your bias. And politicians on the right are stuck in the same feedback loop. Those among them who want to stay tethered to reality get vilified or simply ignored.

DC: The book argues that mainstream media gets implicated in spreading the bias. Could you explain how that happens?

YB: That was something we tried to understand after we saw it in the data, rather than the other way around. We analyzed what mainstream media covered over the course of the election, for example, how many stories and how many sentences associated Clinton with scandals—primarily email and to some extent the Clinton Foundation and Benghazi—and what the stories said about Trump as well. Many more of the Trump stories were associated with his agenda, with immigration, with jobs, et cetera. So we tried to understand why the media covered the election that way. What we find, very clearly, implicates journalistic practice.

Journalistic core practices have never been perfect but, broadly speaking, they have worked reasonably well. That is largely because, until recently, both political parties in the United States and the major actors—corporations, unions, nonprofits—more or less complied with a set of elite norms about how much you could attack basic foundational facts, how much you could fabricate. This meant that the model of journalistic objectivity and balance—being neutral and reporting on both sides—was not systematically biased in favor of one major party or the other. It reflected, more or less, the elite consensus range of views. Trust in media largely oscillated with the party in power: critical coverage meant that if your party was in power, your trust in journalism declined, and then rebounded when the other party took power.

The flows of falsehood are so unequal that when mainstream newsrooms try to be neutral they are complicit in creating a false impression that there are two legitimate sides to the story.

In the highly asymmetric system we have today—when one side systematically produces propaganda and has no internal checks, and the other side is much more constrained by fact-checking because its readers pay attention to a broader range of media—the flows of falsehood are so unequal that, if you try to maintain neutrality, you are essentially complicit in creating a false impression that there are two legitimate sides to the story. To say this side says “x” and this side says “not x” systematically reinforces and amplifies the lies.

This happened with reporting on climate science. When one side is propagating false science and the other is propagating science, neutrality in the form of giving credibility to scientists far outside the scientific consensus created misimpressions in the public. We see the same dynamic now in essentially everything in the political sphere.

DC: The book suggests that the right-wing media alone—speaking to their core audience—would not have resulted in a Trump victory. You go as far as suggesting that the mainstream media threw the election to Trump.

YB: There is no definitive measurement of the total number of people exposed to Fox News and talk radio, but we can infer from responses to surveys that we are looking at somewhere between 25 and 35 percent of the population. They make up the 51 percent of Republicans who in August of 2018 told the Quinnipiac Poll that they think the press is the enemy of the people. You have to live inside that universe—and only in it—to hold such an extremist and radicalized view of the media and of democracy. Then there is this chunk of between 15 and 25 percent of the population with political views that are more fluid; these people maybe lean Republican and maybe sometimes watch Fox News and sometimes they don’t. That to me is the audience that ends up being tricked into misunderstanding the actual choices because of a failure of mainstream media to adjust strategies to asymmetric propaganda.

For me, the single most important audience for this book is the professional newsroom. I keep trying to talk to journalists and to editors. People look at the data and recognize that it fits their experience. Implementing a fundamental change in editorial policy, giving up on opportunities to frame an eye-catching headline, risking the scoop in more skeptical vetting of email dumps—all these are very difficult, but necessary, changes to make in the professional culture.

DC: Yet you are optimistic about convincing the mainstream media to employ these new strategies. What would change if they did?

YB: The stark finding in our data is the disappearance of the center right. It has disappeared because the right-wing media does not pay attention to anything that tries to straddle the center and the right. Traditional Republicans have to choose between moving ever farther right or being ignored. For example, we document how Fox News reasserted its dominance over Breitbart after the 2016 primaries and election by becoming more extreme and more exclusively focused on reinforcing right-wing narratives.

We argue that journalists need to shift their emphasis from acting in demonstratively neutral ways—such as criticizing both sides—to insisting on verifiability.

The critical question, then, is how the 15 to 25 percent of the population who systemically or at least occasionally vote Republican, but whose media diet is not exclusively Fox News and its satellites, will get better information. Much of this population still watches the TV networks. They still read major newspapers. Those networks and papers have a real role to play in stopping disinformation by emphasizing clear truth-telling, rather than balance, adopting strategies that underscore the falsehood of statements that are false, rather than leading with their newsworthiness. We argue that journalists and editors need to shift their emphasis on acting in demonstratively neutral ways—such as criticizing both sides, or pretending that there are parallels between quite different phenomena on the right and the left—to insisting on verifiability. Stating the truth, explaining the methods and sources, and treating assertions of neutrality as facts that need to be fact-checked, not as editorial choices about shading center.

We have not looked at the 2018 data yet, but impressionistically I think that we are beginning to see a little bit of that happening—more effort to identify false statements as false before they are reported. This is the sandwich strategy of reporting: what I am about to tell you is false, here is what was said, let me remind you this was false and what the political interest is behind the falsehood. Sometimes the mainstream media may simply need to refuse to cover things. The very racist immigrant caravan ad that Trump ran just before the 2018 election was a recent example. The question is: do you cover it? Or do you say, “I’m not going to give you a link. At most I’m going to tell you there was a racist appeal that was worse than anything since the Willie Horton ad, even worse because the president created it.”

As I look informally at media coverage of the midterms, I see some cause for both optimism and pessimism. I certainly encounter more instances of clear “the president is lying” coverage in the mainstream press. But I also repeatedly encounter stories about, say, disinformation on Facebook that frames it as bipartisan, and then proceeds to describe an almost purely right-leaning phenomenon. And as for amplifying the extremist agenda, there has been some depressing initial work showing just how fully the New York Times and the Washington Post fell for Trump’s effort to boost attention to “the caravan.” They published a vastly disproportionate number of stories relative to the actual likelihood that several thousand migrants walking to the U.S. border would present a significant challenge to U.S. sovereignty or society. A naive first look at top mainstream outlets in our general dataset confirms that “caravan” and “voter fraud” occur in many more mainstream media stories than “health care.”

Spending a lot of political and intellectual energy right now on figuring out how to improve online platforms is a mistake.

Journalists and editors need some new norms: do not amplify; emphasize falsehood before the falsehood is explained; and be much, much more careful about documentary evidence such as emails. As we learned from the 2016 election, an email dump is like catnip to a journalist. Here is supposedly occult or private data that somehow reveals a hidden truth. Propagandists know that if they put the catnip out, the journalists will dive in, spend days going through it, and then come up with a story that can get a blaring front-page headline—even if it is not warranted.

A couple of case studies in the book show this dynamic. We describe how the New York Times got had by Steve Bannon on the Clinton Foundation and Uranium One story, which was one of the most linked New York Times stories about Clinton in the entire campaign. And yet, there was a real mismatch between the headline, the bulk of the story, and the acknowledgement buried deep in the body that there was no evidence of corruption or wrongdoing. Similarly, the Washington Post did a deep analysis of documents that were released by Judicial Watch and reported that all sorts of donors tried to get access to Clinton. Then again, buried deep in the body of the story was an acknowledgement that donors sought access and Clinton’s aides did not give it to them. But that does not make for front-page news.

DC: So we need to resist the scoop culture.

YB: Which is extremely difficult.

DC: And focus on truth. I saw some complaints on Twitter last week after Trump’s announcement that he was going to use an executive order to end birthright citizenship. People said, why is the New York Times repeating that when it is not possible?

YB: It is even worse than that: the Associated Press actually tweeted it out as a quote without any frame.

For decades media and journalism studies have understood that there is a tension between trying to make money by grabbing people’s attention and selling advertising on it and being professional and honest. This is not new. Work on this tension goes back at least to the 1940s and the Hutchins Commission, which was convened at the end of World War II to assess the proper role of the media in a democracy. But the number of opportunities to get it wrong, fed by right-wing propaganda, is much higher today, and the anxiety about competition from online sources is also much higher. So constraints are harder to comply with, or harder to agree to comply with.

The good news is that we found that the major mainstream media do not have to be so worried about online competition. They continue to define the agenda. They continue to draw attention. They can afford to be more careful. Editors and journalists need to understand that they are under an asymmetric attack, but that they have the breathing space to be the professionals they ought to be, and that they—more than all of the technocratic and technological solutions that are constantly being floated out there—can make a bigger impact. The commitment to professional journalism at its highest level is going to be the single most important thing that we can do to deal with this moment of epistemic crisis.

DC: So you don’t think censorship of online platforms is a good strategy even though so much disinformation lives on those platforms?

YB: I think that spending a lot of political and intellectual energy right now on figuring out how to improve online platforms is a mistake. There is good reason to think that in five, ten, or fifteen years, as the online population becomes the main body of the electorate, we are going to have to solve that problem. But in the near term we need to solve the problem of advertising online; we already have proposals on the table such as the Honest Ads Act. The primary issue is transparency and a prohibition on active behavioral manipulation, which is different from censorship.

DC: Should we have a new Hutchins Commission about the role of the media in a democracy?

The commitment to professional journalism at its highest level is going to be the single most important thing that we can do to deal with this moment of epistemic crisis.

YB: The Hutchins Commission was a creature of its time: a small group of high-status men, asserting a set of norms. We could fix the composition problem, but I don’t think we can recreate the shared ideology of authority figures handing down a set of norms. We need broad debate about this problem across the profession. We need open, public acknowledgement of where the failures are. And we need to leverage the critical forces of the open, networked public sphere that makes up “the rest” of the U.S. media ecosystem to keep journalists and editors honest.

It will be hard. But I know that there is this middle segment of the population that can be moved. And the question of reorienting this audience has to do with changing the incentives that favor radicalizing politics and radicalizing rhetoric.

When you actually look at the electorate, there is still a center-right in the United States. It has just gotten its party stolen out from under it. The Trump party was born of racial and religious anxiety that the Republican Party has long stoked and ridden to power. Now, it has overwhelmed the Republican Party. The question is, what dynamic can allow the Republican Party to become a conservative party, rather than a radical right-wing party. If anyone is able to move the needle today, it is more likely the mainstream media than any other potential actor.