From Trump Tower to the White House, in 140 characters

Josh Cowls
Apr 12, 2017 · 14 min read

Last week I presented an overview of my thesis research to an audience at MIT. Below are my prepared remarks; I’ll upload a video of my presentation and Q&A in due course.

Thank you all for being here today. I came to CMS in the fall of 2015 with the intention of making the presidential election the focus of my thesis — and in the most basic sense, that’s what I did indeed study, and what I’ll present on today. But few could have predicted in advance just how tumultuous, transformative and — for many of us — disheartening the election would end up being.

Of course, you can only fit so much in to 100 pages and 20 minutes. Thus there are huge and important aspects of the election that won’t get much mention today, including the disturbing rise of fake news, the alleged involvement in the election by foreign actors, and the failure of pollsters to predict the outcome — any of which, by themselves, could certainly have constituted a thesis. But what I present today nonetheless has implications, at least implicitly, for these other questions. In my thesis I offer an overview of the flows of communication between candidates, voters and the media over the course of the campaign. In particular, I explain how this new communication ecosystem was especially beneficial to Donald Trump, a charismatic candidate purporting to be an outsider, who drew on populist and paranoid tropes in his campaign. At the center of his communication strategy, as I will show, is the social network Twitter. And as president, Donald Trump’s use of Twitter continues to dominate the news agenda, providing a powerful personal outlet for the most powerful person in the world.

Background: Paranoid Populism

Before I turn to the 2016 campaign, I’m going to introduce a concept — paranoid populism — and two figures from 20th century American history who embodied it in their media use. Paranoid populism is the shorthand I use for the combination of two related but distinct ideological constructs: populism and (political) paranoia. As we’ll see, paranoid populism as I render it offers a useful lens through which to view the Trump candidacy, and a way to understand and even explain his success.

Although populism in American politics has eluded precise definition, scholars nonetheless agree on several of its characteristics. As its name suggests, populism purports to advocate for “the people” — though in truth, populists specifically revere “ordinary” people — or what Michael Kazin dubs “a noble assemblage”. In fact, it may be more appropriate to refer not to a group of people, but to a place — what Paul Taggart calls the “heartland”.

Populism is defined, moreover, in its opposition to a selfish, coastal elite, unconnected and unconcerned with the plight of the heartland. These appeals to the “pure”, “ordinary” heartland — in contrast to an elite as cosmopolitan as it is corrupt — almost invariably stoke economic resentment as well as cultural antagonism.

Given this objection to coastal elites, populist movements are typically spearheaded by a figure with outsider credentials, especially someone with charismatic communication abilities, who “tells it like it is”.

One final feature typically associated with populism is that it is manifested in the communication strategies of its leaders, as a conscious rhetorical project. I thus argue that Donald Trump’s use of Twitter as an unfiltered means of communication was the essence of his populist appeal.

But populism alone does not fully encapsulate the candidacy of Donald Trump. We also need to consider what I call “paranoia” — a concept which here should be considered in a political rather than a clinical sense. Richard Hofstadter’s influential essay of 1964, “The Paranoid Style in American Politics”, provides an insightful look at the appeal of this strain of thinking and its impact. Hofstadter’s essay — written amidst the rise of far-right Republican Barry Goldwater — pinpoints three characteristics — exaggeration, suspicion and conspiratorial fantasy — which he says constitute paranoia.

Like populism, paranoia is an ideological construct targeting certain sets of people — chief among them, according to Daniel Pipes, both the politically disaffected, and the culturally suspicious. Paranoid thinking shares much with populist ideology: not least an “us versus them” mentality distrustful of elites. Its appeal, like that of populism, can be accentuated by media technology: in 1964 Hoftsadter noted that the quote “villains are much more vivid than those of their paranoid predecessors”.

Yet the two terms are not entirely overlapping: politicians famous for their anti-communist paranoia, including Goldwater and Joseph McCarthy, were not known for populist politics, and McCarthy was up by his own lack of charisma. When combined, however, populist ideology and paranoid thinking can be prove powerful, especially when enabled by the media of the day.

Two figures from twentieth century American history — Charles Coughlin and Pat Robertson — both employed paranoid, populist tropes and tactics in their quest to build dedicated audiences on emerging comms platforms, and their stories are instructive.

“It’s possible”, Michelle Hilmes writes, “that no signal individual had more of an impact on thinking about the radio audience than Charles Coughlin”. Ordained as a Catholic priest in 1916, Coughlin began a regular radio broadcast in 1926, a time when the medium was in its infancy. His “sermons” initially focussed on narrow religious matters, but became over time almost exclusively political in content. Coughlin employed a wide variety of rhetorical techniques and a folksy mode of address to appeal to “the man in the street”.

Coughlin’s audience was drawn to his coverage of America’s deep financial depression and the blame he frequently ascribed to the industrial elite, as well as the Roosevelt administration. In so doing, Coughlin tapped into a then-flexible communication platform around which few norms or laws had been established. He initially enjoyed easy access to his audience through an independent system of radio stations which broadcast his sermons — meaning, in essence, that if one station objected to his content he could simply strike a deal with a rival.

Coughlin’s audience peaked at a remarkable 16 million in 1938, giving rise to the possibly apocryphal notion that “hearing his voice out of every window, you could walk for blocks and never miss a word”. Yet Coughlin’s large audience would not prove eternal. A combination of religious, political and media elites ultimately conspired to take him off the air. As more radio stations eventually turned against him, the outbreak of the second world war — which Coughlin had vehemently opposed joining — saw new limits placed on quote “controversial public issues” — one of the precedents for the Fairness Doctrine, introduced in 1949.

Before his fall from grace, Coughlin had nonetheless demonstrated the appeal of paranoid populism and its ability to thrive on a new communications platform, at least while political and market conditions were favourable, and before regulation had caught up.

Coughlin’s experience brings to mind another paranoid populist firebrand, Pat Robertson, who also utilized an emerging medium to build a similarly loyal if narrower following. Robertson’s Christian Broadcasting Network, launched in 1961, broadcast via satellite –making his religious messaging seemingly descended from the heavens. Robertson’s movement set itself against the perceived liberal excesses of the 1960s, appealing to the more traditional values of ordinary folk. In his broadcasts, Robertson seized the mantle of “evangelical conspiracism”, and built on the tradition of prophesy theology, which predicted an imminent second coming, prefigured by apostasies including drugs and divorce.

This “unmediated Christian perspective” was made possible through CBN’s satellite transmission system, as well as changes in regulation, which made religious broadcasting technically and economically feasible. CBN was eventually available on three thousand cable systems, and the number of Americans viewing religious programming rose from around 5 million in the late 60s to almost 25 million by the mid 80s.

Robertson’s media success bred political ambitions, but he would ultimately struggle to broaden his base, and failed to win a single state when he ran in the Republican presidential primary in 1988. Indeed, his audience actually fell by 52% during his presidential run. Amongst other reasons, it seems that his attempts to “go mainstream” and secure the wider slice of the Republican electorate needed for the nomination forced him into more specific political pronouncements, many at odds with his base.

The experiences of Charles Coughlin and Pat Robertson differ, but share some characteristics. Both were paranoid populists who exploited emerging media technology before either rivals or regulation could catch up. Political factors — for Coughlin the second world war, and for Robertson the realities of majoritarian politics –also eventually hampered their ambitions. But viewed in another way, each had what the other lacked: Coughlin’s platform was broad, but shaky — ripped from under his feet by regulations and nervous networks. Robertson’s audience was deep but narrow, and he was unable to broaden it to encompass a wider swathe of voters when he ran for president. It seems then that any paranoid populist seeking prominence and public office needs a hybrid communications strategy: with both a direct, unfiltered platform, and the ability to get noticed by a much wider audience. Which brings us to Donald Trump.

@ realDonaldTrump

I won’t spend much time here rehashing the story of the 2016 presidential election since it remains fresh in everyone’s minds. But it is worth remembering several of the perceived weaknesses of Trump’s candidacy at its outset. He was up against 16 other Republican candidates — the largest field in history — which meant that before any votes were even cast, endorsements, fundraising and media attention were in high demand. In terms of both endorsements and fundraising, Trump — with minimal support inside the GOP and little organization — was running well behind. [FRAME] Campaign filings in January 2016, on the eve of the first primaries and after some candidates had already dropped out, show Trump running only fourth in the fundraising stakes, [FRAME] while even by the time Trump clinched the endorsement in May he remained disliked in the party, sitting in fourth among elected Republican officials.

As we’ll see, the metric Trump did dominate was media attention. In this respect, Trump had several in-built advantages, including decades of experience dealing with the media — primarily tabloid newspapers — and his years hosting NBC’s Apprentice show. Trump also had a Twitter account, set up in 2009 under the handle @realDonaldTrump, which reached several million followers at the outset of the campaign. And for the past several years, Trump was the informal chief spokesperson of the ‘Birther’ movement — a group which falsely alleged that President Barack Obama had been born outside the United States and was thus ineligible to serve.

Trump, in other words, knew how to get attention, in large part by saying and doing outrageous things. Still, in its early stages his campaign was treated as a joke, with prediction website 538 writing off his chances, and the Huffington Post news website deciding to cover Trump’s campaign in its Entertainment section.

Yet on the trail and on Twitter, Trump’s campaign began to gather steam. And it appears that in both domains, Trump embraced the tropes of a paranoid populist in the manner of those who came before. I ran the text of Trump’s tweets through corpus linguistics software to determine his most common words and phrases. As you might expect, his slogan “make America Great Again” was one of his favourite hashtags during the primary, as this chart shows, coming up 286 times overall, which accounts for around one in every 200 words he tweeted. The phrase perfectly encapsulates the paranoid populism as defined earlier, and is evidence that Trump’s campaign deliberately targeting disaffected voters.

Many people have responded to this slogan by asking when, in Trump’s estimation, was America great? Trump managed to avoid being pinned down on this question directly during the campaign, but his staunch attacks on the legitimacy and record Barack Obama — the country’s first African-American president — are pertinent. By simultaneously alleging that America had seemingly lost its greatness, and by pinpointing the country’s first black president as the chief agent in this process of perceived decline, Trump was making a very particular appeal to America’s “heartland”. As noted earlier, by replacing “the people” per se with “the heartland”, populists can restrict the scope of their appeal to citizens deemed culturally and racially “ordinary”. Through his slogan and other campaign pronouncements, Trump tapped a vein of American “normalcy” that stretched back centuries, seeking to restrict its scope to the country’s heartlands, and implicitly to a particular kind of cultural and racial hegemony.

Turning to the general election against Hillary Clinton, Trump on Twitter defied the traditional migration to the center ground to double down on his paranoid populist rhetoric. As this graph shows, Make America Great Again emerged as Trump’s top hashtag during the general. His second and third most popular phrases were “America First” — a phrase with a long history in isolationism tinged with anti-semitism — and “Drain the Swamp”, another classically populist anti-elite sentiment referring to Washington bureaucrats.

Yet in addition to looking at what Trump said, it’s also instructive to look how he used the platform. Trump began the campaign with almost three million followers, and had earned 13 million by election day. This is an impressive number for any politician not named Barack Obama, and in itself partially explains Trump’s prodigious tweeting throughout the campaign. Twitter though is a famously messy network, riven with bots, trolls and much else, so it is reasonable to assume that Trump was reaching far fewer than three to thirteen million voters each time he tweeted.

One of the strengths of Twitter is that the potential for amplification is built into its architecture through its retweet functionality. As such, it’s possible to analyse the amount of wider reach that Trump’s tweets received during the campaign. So for each month, I calculated the average number of retweets that each tweet received, as shown here: As you can see, the amount of amplification Trump received for his tweets rose considerably over the course of the campaign, from an average of only 200 retweets at the outset to over 12,000 for October 2016. This effect mostly holds even when we control for his rising number of followers across the period:

Trump, of course, started the campaign with a base of followers far in excess of the average presidential candidate. This should disabuse any notion that Trump led a campaign from the virtual “grassroots”: he has enjoyed high levels of name recognition for many years. What he did do was utilise his existing following into a source of much wider exposure for the paranoid populist messaging he deployed — both on Twitter and beyond.

Yet as Trump appears to have intuited, reaching several million followers, or even several million more through retweets, wasn’t enough: he also needed to establish as much exposure as possible in the mainstream media. Another part of Twitter’s architecture is the @reply, which links your own tweet to the account of someone else using the @ key. As this chart shows, in his most common uses of @replies, Trump sought attention from predominantly media sources. In the primary, this mostly meant the at-least-somewhat sympathetic Fox News Channel. In the general election, however, Trump’s media diet appears to have evolved, to include centrist sources like CNN and the New York Times. Yet to a greater extent than the primary, Trump’s references to the media are about more than simply increasing his exposure. Of his 38 references to the Times, for example, fully 25 were to the quote “failing New York Times”. This suggests an evolution in Trump’s strategy. Trump’s tweets during the general election seem to have had less to do with highlighting his media appearances and more to do with objecting to unfavorable coverage by seeking to delegitimize the organizations providing it.

Into the Mainstream

This concludes my analysis of how Trump used Twitter to both deploy paranoid populist tropes and target media sources, seeking both attention and retribution for perceived criticism. It seems to evoke something of a hybrid strategy, stoking the small but committed base of followers through a direct, unmediated slew of messages, and reaching out, however angrily, to try to gain a foothold in the mainstream media.

For those whom Trump was reaching directly through his Twitter account, media exposure presumably did not much affect their support one way or the other. But for the larger bulk of prospective voters whose support he sought, Trump required if not a supportive, then at least an interested mainstream media.

Of course, Trump was not the first candidate seeking media exposure. For decades, campaigns have tried to find cheap and efficient ways to get through the media’s “gatekeeping” process, which restricts the flow of information to the public. In the past, everything from press releases to photo ops have been used by campaigns to try to seize the media spotlight. If it works, this can result in “free media” — as opposed to expensive TV ads.

Another theoretical concept is useful here for thinking about how the media covers news. In a book about the Vietnam War, Daniel Hallin introduced the three spheres of political discourse in America. First, there is the sphere of consensus — “Apple Pie” issues about which most everyone agrees and for which an opposing viewpoint does not need to be presented. The second sphere is that of legitimate controversy — this contains issues which are reasonably disagreed about, such as whether marijuana should be legal. The third sphere is the sphere of deviance — issues so outside the mainstream that journalists condemn them, such as heinous crime. (Points for anyone who gets the chianti reference here.)

Yet there are two counterintuitive things to consider about Hallin’s spheres. First, they are permeable: issues can move between them over time — consider, for example attitudes towards women’s suffrage between the 19th century and today. Second, studies show that journalists disproportionately cover deviance, even as they condemn it.

Over the course of the campaign, Donald Trump made myriad deviant statements. In many cases, there were no shortage of people decrying them as abnormal or wrong. But it is also true that the sheer interest in Trump’s campaign often outweighed concerns about condemnation. In an interview at Harvard on the eve of the election, CNN President Jeff Zucker expressed regret for the hours upon hours of live, unfiltered footage of Trump’s rallies that his network broadcast. Of course, this coverage was an enormous ratings hit, which undoubtedly contributed to the networks’ decisions — but without a filter of condemnation, this coverage served to normalize and even partly legitimize Trump’s candidacy.

My analysis suggests that Trump’s deviance on Twitter meant his account was also over-covered with respect to other candidates. For my research I queried Media Cloud, a database developed by MIT’s Center for Civic Media, for the Twitter handles of each of the Republican candidates over the course of the primary. As the graph shows, even when we measure it alongside poll ratings, mainstream media references to @realDonaldTrump were far in excess to the Twitter handles of his rivals, who barely get a look in.

Of course, correlation between Trump’s deviance and the disproportionate coverage his tweets got doesn’t prove a causal relationship. And many of these references to Trump’s Twitter were, presumably, condemning him. But it nevertheless says a lot that Trump’s paranoid populist tweeting received so much mainstream coverage during the primary election — a time when media attention contributed considerably to a candidate’s viability.

So what have I shown today? First, that Trump’s campaign contained many of the paranoid populist tropes of his ideological forbears, and that much of this ideology manifested itself in his communication strategy, which was unusually focused on Twitter. Second, and related, that Twitter offered Trump a direct, unfiltered line to his base of committed supporters — sending messages which were amplified by them through the network. And finally, that Trump, who begged for media attention even as he besmirched the outlets which provided it, succeeded in getting through the gatekeepers in large part because of the deviance of his paranoid populist stances.

Thus Trump melded the loyal following of Pat Robertson with the broad appeal of Father Coughlin to become America’s first paranoid populist president. Every election is decided by a myriad of factors — but in the case of 2016, my research suggests that without @realDonaldTrump, we wouldn’t have President Trump.

Thanks for listening — I’d welcome your questions and if you want to discuss this further … I’m on Twitter.

Josh Cowls

Writing about the ethics of data and AI; political…