Donald Trump's Campaign and the Hybrid Media System

The first keynote at AoIR 2017 is by Andrew Chadwick, who explores what the 2016 U.S. presidential campaign means for our understanding of the hybrid media system. Political communication is in the middle of a chaotic transitional period, due in good part to the disruptions brought by newer, digital media; some older media have also been renewed by integrating the logics of newer media. This then represents a systemic perspective that examines forces while they are in flow.

The hybrid media system is built on the interactions of older and newer media logics in the reflexively connected field of media and politics. Actors in this field tap and steer information flows in ways that suit their goals, enable or disable the agency of others, across various older and newer media settings. 'Hybrid' here shifts our conceptualisation from 'either/or' to 'not only, but also'; it foregrounds complexity, interdependence and transition. We pay more attention to boundaries, flux, and liminal spaces, where practices intermeshing and co-evolve.

This also means studying the power relations between social actors, and understanding the systems that exist horizontally and vertically between them and that from time to time undergo long and chaotic periods of change. Within systems, even the most powerful actors often need to cooperate with those who have less power, and power is relational and situationally dependent; in a digital environment, these institutions and their relationships are now comparatively loose, ad hoc, and spontaneous, and highly adaptable to current contexts: they can be understood as assemblages – built around permeable boundaries between different entities engaged in particular endeavours.

Power and systems are also dependent upon time; there is a need for a mastery of technical rhythms, that is, for timeliness, especially in increasingly real-time environments. Temporal power is now enabled or constrained in different ways by different media, and it is enacted and re-enacted differently.

Which brings us to the Presidency of Donald J. Trump. His White House has been in chaos since day one; many early advisers have already been fired or have resigned. Press Secretary Sean Spicer started his job with a need to deal with Trump's blatant lie about the size of his inauguration crowd, for example; rapid work by a Reuters news photographer, his editorial office, and other media made the comparison between Trump's and Obama's crowds echo around the world almost before Trump was finished speaking.

One of the largest demonstrations during the early days of the Trump Presidency, the Women's March, began from a social media post by a Hawai'ian activist, which over night attracted support from a diverse coalition of major activist organisations; it also featured many participants wearing 'pussyhats', responding to candidate Trump's comments about his sexual harassment activities. The Women's March protests occurred on the day after the inauguration, as Trump was giving a public speech at CIA headquarters, and well eclipsed the size of Trump's inauguration crowds.

So, one of Spicer's first acts in office was to defend Trump's lies about the inauguration crowd; but he attempted to follow a pre-digital media approach that presented scripted statements: the arguments he presented were clearly untrue, and quickly debunked by a coalition of mainstream and social media actors who presented copious evidence to the contrary. Such articles, posts, and tweets was further embedded in high-traffic sites including Buzzfeed.

In turn, The New York Times collated much of this material as well as some additional original research in its fact-checking column to demolish every single claim made by Trump and Spicer. Various historical documents and other source data were presented to support the article's conclusions. Trump spokesperson KellyAnne Conway ended up defining Spicer's and Trump's statements as "alternative facts", much to the amusement of professional journalists.

The debunking of the statements, then, involved a variety of actors both professional and amateur, local and global, institutional and individual, who became simultaneously a global, self-mediating force that impacted on the public perception and further media coverage of the inauguration and subsequent debate. Marchers in subsequent protests in Washington, D.C., and around the world continued to refer back to the event, by posting serious or humorous crowd photos from their events, in comparison to the inauguration.

But while such ineptitude may be amusing, Trump's engagement with older and newer media is not always so simplistic. As candidate and President, Trump has efficiently used direct personal attacks and previously unimaginable forms of racist, sexist, and other slurs, which journalists have widely reported but which have not managed to retard Trump's electoral success. Audience engagement with articles about Trump (to the extent that we may believe any such metrics at all) was considerably greater than with material about Hillary Clinton; as an insurgent, non-establishment candidate Trump needed to translate his celebrity capital into any kind of mainstream media coverage. He had to demonstrate to editorial gatekeepers that he would be a serious presidential contender.

Metrics in journalism are now used as an important indicator of audience attention to particular stories and actors; this applies especially also to social media metrics for stories and other news updates being shared by journalists. Trump managed to get a foothold in the race by generating the social media metrics that editors looked for, and that would make them direct more journalistic attention his way. Trump also engaged with mainstream media directly, by essentially livetweeting some Fox News shows; this is anything but random, but instead creates direct connections with the channel and its news staff. This also taps into the informality and deprofessionalisation of cable news.

Additionally, of course, Trump also connected effectively with the right-wing news sphere in the United States, led by Breitbart. This tapped into the (partly) algorithmic optimisation of social sharing: a shift from news access via search to news access via social media, where timing and engagement matter especially much and where specific stories can be made to trend. Trump mastered some of these rhythms of social media sharing, which elevated the right-wing news sources supporting his nomination bid to much higher visibility and ultimately led to his nomination as candidate. In the presidential campaign, he outstripped the attention paid to Clinton by some margin, and even received more attention than incumbent President Obama had received in his 2012 re-election campaign.

But what is the causality here? Did Trump's tweets lead to greater news coverage, or was it the other way around? What other factors may also be relevant here? Retweets of Trump's posts preceded increases in news articles, but the opposite relation did not hold; he posted more when he received less news coverage. This is not disintermediation at all, then; Trump did not bypass the mainstream media with his social media use, but managed to substantially affect the mainstream media agenda by using social media. There may not have been a particular major digital media operation in the Trump campaign at first, but what he did do was enormously effective.

After the primaries, of course, all of this grew substantially. Trump received more organisational support from the Republican Party; by August 2016 some 100 digital staff were in place, and Trump hired both Cambridge Analytica and Breitbart's Steve Bannon. CA built detailed voter models from email lists, voter lists, Facebook posts and tweets, but there remain questions over how successfully these were in fact used. At any rate, Trump certainly used Facebook advertising to great effect, and in doing so benefitted from Facebook's opening of its API to consumer data broker companies, who improved their success rate of matching Facebook accounts to email lists from around 30 to up to 85 percent. This also enabled marketing not only to these users themselves, but also to "lookalike audiences" with similar interests to those who are already signed up to email lists. Such users may not be engaged enough to attend rallies, but may still be susceptible to the same messages as delivered via Facebook.

The campaign also utilised 'dark posts', drawing on Facebook staff who were embedded in the Trump campaign. These do not appear in the campaign's page, but can be used to A/B test multiple, slightly different versions of the same ad to compare audience engagement – and they do not appear as spam to targetted users. Such posts were not just used to engage with Trump's voters, but also to suppress the enthusiasm of prospective Clinton voters, described as "disengagement targets".

It is uncertain how well such voter suppression did in fact work, but any small drop in turnout for Clinton would have had substantial impact in states with thin margins between the two candidates. There is now an urgent moral imperative to start examining this kind of activity, so that both the platforms and the data science firms that operate on them can be held to account – as well as the campaigns themselves, of course.

Additionally, we have now also seen the emergence of moral panics around 'fake news'. These are associated particularly with online and social media, and with the decline in the recognition of major mastheads. There are now various 'fake news' factories that generate a substantial volume of such content which individually may not have any considerable impact, but in aggregate result in substantial visibility; Facebook and other social media platforms are especially crucial for their distribution, while Google's ad platform can also be used to distribute such stories to susceptible audiences. Left-wing audiences attempting to show up the trust of their opponents in such 'fake news' stories by sharing them through these networks also inadvertently help to disseminate them. Further, such stories are also made visible by algorithmically curated clickbait story lists that are even employed by a variety of mainstream media outlets. How any of this works, in these operators as well as in the Internet giants, is an arcane and intransparent science.

And such content, as well as broader debate, is also disseminated and engaged with by social media bots. Some 33% of tweets with pro-Trump hashtags during the 26 September 2016 Trump/Clinton debate came from bots and highly automated accounts; for Clinton, the percentage was 22%. We can no longer rely on the face value of Twitter and other social media platforms, because of the presence of such a substantial number of bots.

But then there are other, less automated rogue operators as well: WikiLeaks was deeply involved in the campaign, following its transformation from an information transparency site collaborating with mainstream media to a site providing unredacted information dumps that appear designed to attack specific politica actors. Leaks about Clinton's private email server and the email traffic of the Democrat party organisation severely damaged Clinton's campaign, and Trump exploited WikiLeaks effectively in his own campaigning. The extent to which such leaks were also driven by foreign state actors – principally, Russia – is still under investigation. These operations, too, represent a global assemblage.

But, we should remember, so was the Women's March: there are global, complex, diverse, heterogeneous assemblages between mainstream, alternative, and social media actors of various stripes on all sides of current political struggles.