^{Do you defend free thinking around the world? In democraciaAbierta, we fight against censorship, fear, fake news and manipulation, defending free thinking around the world. Donate and help us to defend free thinking here}

The Republican Party of the United States has an eye on the next elections and is currently investing hundreds of thousands of dollars per month in an enormous data infrastructure, involving state and local committees led by Trump’s 2020 national campaign. All that apparatus contains at least one phone number or email address of every citizen registered in it and is managed by Brad Parscale, the digital media director and strategist of Trump’s 2016 campaign. In that occasion, the nomination of a digital marketing expert and data strategist, who had never before worked in electoral campaigns, was surprising. During an interview for an American television network, Parscale said he expected to have 50 million voters in his database by the day of the next elections.

The Facebook strategy led by Parscale in 2016 played a central role in Donald Trump’s presidential campaign. Estimates indicate the republican team was able to post up to 100 different types of targeted ads on social media in 24 hours. Targeted ads are posts with specific messages for each voter or group of voters. For example, personalized messages may be based on geographic determination – as in the case of regions with more industry employees – or on themes the citizen supposedly prioritizes.

Through that same system, the campaign is able to adjust language to approach the same topic according to specific audiences in gradations that may go from more aggressive conservative contents to a more moderate tone. Moreover, specific groups of voters often do not know what the other ones are receiving. On Facebook, these posts that only appear for specific receivers and not in the official page of the campaign are called dark posts. Once such kind of post is shared, it is very difficult to trace back its origin. Dark posts grant more freedom for the campaign to work with contents for specific audiences without running the risk of losing any public because of more controversial messages, since groups that could feel offended may not come to know about the post or associate it to the campaign.

Following the steps of traditional authors in Political Communication, this article discusses the ways digital media have been affecting electoral campaigns and, in this sense, politics and democracy itself. The discussion is based on the campaigns for Donald Trump in 2016 in the United States and Jair Bolsonaro in 2018, Brazil. By the end of the analysis, I make two suggestions. The first one is the idea of “expanded hypermedia” as a way of understanding contemporary political communication. The second one is the notion of “audience atomization” or the “disruption of the public sphere” as a result of digital political communication in the democratic development of societies.

For now, we must recapitulate some of the things that happened in 2016 in the United States. It is worth noting the analysis of the North American context is especially relevant because developments in communication technologies appear first mostly in the United States. When candidates and parties, as well as campaign and news media professionals, in Brazil get acquainted with new communication technologies, they usually look to the practices already tested by their counterparts in North America as reference. Not by chance, the Academic field of Political Communication presents the traditional hypothesis of the “Americanization” of campaigns, which suggests that electoral campaigns in democracies around the world are ever more Americanized, given that candidates, parties, and journalists follow the practices of their US peers.

In this context, the Trump 2016 Republican platform marked definitively the history of US campaigns, especially, in relation to digital media. It is no small accomplishment, since it happened after Barack Obama’s 2008 and 2012 elections, both won with the intensive use of the internet.

In fact, the history of the internet in American elections is already fairly long. In 1998, the wrestler Jesse Ventura, a political outsider at that point, was elected governor of Minnesota by the anti-establishment Reformist Party with the intense use of his network of emails, which was built around the community of wrestlers and fight events. The network was used to promote campaign gatherings, meetings, and rallies in the state.

In 2000, the Democratic ticket Gore-Liberman developed an online chat to interact with voters. Four years later, in his reelection, the Republican George W. Bush opened a channel for questions and created the famous “Kerry Gas Tax Calculator” website, which calculated, by the gallon, how much more would gasoline cost with the adversary’s plan to increase taxes on fuel.

A simple way to observe the increasing importance of the internet on American presidential campaigns is to “follow the money”. In 2000, the Republican senator John McCain raised US$ 500 thousand in 24 hours on the internet for his campaign. McCain received a total of US$ 6.4 million on the internet during the presidential race, which corresponds to 25% of his expenses with the entire campaign.

In the 2003 Democratic primaries, Howard Dean and his network of blogs (Dean Nation, Change for America, Howard Dean Call to Action, and Blog for America) raised US$ 20 million on the internet or 40% of his campaign costs. John Kerry, who won the party’s nomination, raised US$ 89 million online in 2004 or 33% of all his expenses.

In 2008, Barack Obama elevated the numbers to another level. His first presidential campaign raised US$ 750 million, a historical record in the United States and more than twice what his Republican opponent John McCain obtained. From the total, Obama received US$ 500 million on the internet and more than US$ 400 million through what his campaign called “digital effort”, meaning donations that resulted from contact with voters over email, social media, cell phone or on some of the campaign’s websites. In 2008, Barack Obama obtained 3.95 million donors.

In 2012, the Obama campaign model was repeated, but with a lot more people online. On election day in 2008, the Democrat had a little over two million “friends” on Facebook. Four years later, they were 32 million. On Twitter, he had 125 thousand followers on the day of his first election and 22 million in his reelection. Despite the significant amount of money raised by the Republicans in 2012 – Mitt Romney’s platform surpassed US$ 990 million –, the main scene was on the Democratic side. Barack Obama was the first one-billion-dollar American presidential candidate, with US$ 1072.6 million, 4.4 million donors, and more than US$ 500 million obtained through “digital effort”.

Four years later, Donald Trump’s victorious 2016 campaign and his digital communication strategies generated certain apprehension regarding the impact of the internet on electoral contexts and democracy in general. For instance, his campaign acted in a controversial way with respect to the quality of information circulating in the public sphere during the elections. Furthermore, the campaign was marked by the success of questionable forms of speech in the digital environment, the use of illegal data, and the possibility of external influence in Trump’s favor.

As mentioned above, Donald Trump’s campaign was also marked by the choice to place Brad Parscale in the command of social media. Parscale is called the first “purely digital” consultant to lead an important presidential campaign in the United States. He centered his strategy on the importance of Facebook, including over television ads.

The strategist led a Fordist scheme of meme production. The team divided the production of resources (text, image, video, links) to create a great number of variations and distribute them automatically on social media. A traditional US election commentator stated he had never seen a political campaign build a system for Facebook of the same magnitude as Trump’s in 2016.

A controversial information about the developed system, however, was the use of abovementioned “dark posts”, that is, posts that appeared for targeted audiences but not on the campaign’s official Facebook page. Certain messages with negative content on Hillary Clinton sent to specific groups of Democrat voters, such as women, African Americans, and industry workers in key states, were particularly important for Trump’s platform. Messages were sent to discourage these groups from voting for the Democratic candidate.

In that context, over half the costs of Trump’s campaign with media was on digital media. The Republican platform spent and raised, in total, less than Hillary Clinton – the Democrats raised US$ 1,191 million, the Republicans raised 646.8 million – but Trump invested on Facebook more than the Democrat candidate. Estimates show Trump’s campaign messages were three times more retweeted and five times more shared on Facebook than those from Hillary Clinton. One of the highlights of the Republicans’ social media strategy was their greater capacity to set the news agenda. According to a campaign executive, one of Trump’s main strategies was to “keep things moving so fast, to talk so loudly—literally and metaphorically—that the media,

and the people, can’t keep up”.

Cases such as Trump’s imitation of a New York Times reporter with special needs, or his questioning of John McCain’s military past, or even the sexual harassment accusations against him, which could have undermined any campaign in other contexts, did not stick on the agenda driven by Donald Trump’s “steamroller communication”.

Moreover, the platform used bots – automated social media accounts – intensely to boost posts artificially. According to a commentator, bots end up “fabricating consensus” by artificially increasing traffic flow around a candidate or topic. The Oxford Internet Institute found evidence of a much larger quantity of pro-Trump content, and propaganda against Hillary Clinton, disseminated automatically. According to a report by the Institute, the ratio of automatic messages sent by Trump’s campaign in relation to Hillary’s was 5 to 1. This way, the Republican campaign also managed to keep its social media content constantly on the North American journalistic coverage. Estimates indicate Trump received on the whole 15% more coverage on the news than Hillary Clinton during the presidential race.

At the same time, Trump’s social media strategy was connected to a Big Data strategy, especially, after the campaign hired Cambridge Analytica, which closed operations in 2018. According to the its website, Cambridge Analytica was a “strategic communication company” with the mission to “use data to modify audience behavior”. The company started from a model created by Michal Kosisnki at the University of Cambridge, which analyzes users’ profile through the trail they leave in the digital world, and kept voter profiles in big databases through which it could offer “individual psychological targeting”. Established in 2013, Cambridge Analytica participated in 44 campaigns in the United States before it was hired by Trump’s campaign to take over data operations in June 2016. Besides assisting the campaign in message targeting, C.A. identified the type of message that was resonating with whom and where in order to adjust the candidate’s travel schedule. If, for instance, an article on immigration got many comments, likes, or shares in a certain county in Pennsylvania, the campaign would include that stop in Trump’s travel schedule, and the candidate would give a speech on the theme there.

The importance of that strategy cannot be dismissed in the US electoral model, in which a handful of swing states can define the overall result. As it is known, Cambridge Analytica closed its doors in 2018 after the The Guardian broke the news of the illegal leaking of information about 50 million American citizens stored by Facebook.

But dark posts, bots, and Big Data were not the only noteworthy elements of Donald Trump’s 2016 campaign. Fake news was also an important aspect. It is true that, since the issue became public, a great controversy was set around the expression “fake news”. Many authors consider it wrong and prefer to use “disinformation” to name the phenomenon instead. In this article, I consider these are two types of problem. Fake news is false information presented in a journalistic format and/or originated from websites that fake a journalistic identity. On the other hand, disinformation is false or controversial information that does not have an actual source or is signed by fake sources.

In July 2016, for example, a fake news website called WTOE 5 News published on its homepage and shared on Facebook that Pope Francis had demonstrated “unequivocal support” to Donald Trump’s claim for the US Presidency. The post, which was perhaps the most representative fake news of that moment, got more than a million shares.

Besides being mostly disseminated on social media, fake news seems more connected to conservative voters, at least in the context of the United States in 2016. Estimates indicate there were at least 40 million fake news sharing on Facebook during the elections, with three times more fake news pro-Trump than pro-Clinton. Also, estimates suggest pro-Trump fake news were four times more shared than fake news pro-Clinton.

A researcher on the phenomenon noted the most popular fake news may have exceeded real news in terms of sharing in 2016 and that 75% of adult Americans that had contact with a fake news believed it was true. Donald Trump retweeted a fake story indicating that his support among American industry workers (blue-collar workers) was bigger than that received by any other US Presidential candidates since Franklin Roosevelt. Eric Trump, Kellyanne Conway, and Corey Lewandowski, who were all connected to the candidate, retweeted a story from the fake news website abcnews.com.co stating Hillary Clinton had hired people to disrupt Trump’s rallies.

Digital media use in electoral campaigns is not something recent in Brazil as well, neither the use of bots or the circulation of fake news, which had already appeared in previous elections and political events, like the 2013 demonstrations and the murder of Marielle Franco, a Rio de Janeiro city councilor, in March 2018. In fact, since 2010, a said second phase of online campaigning began in the country, characterized by the withdrawal of legal restrictions and the intensive use of digital resources.

In the 2018 elections, Brazil had more than 110 million citizens with internet access and held the third place among countries with the highest number of Facebook users and the sixth place in terms of Twitter users. Furthermore, in 2018, legal restrictions for online campaigning were reduced due to new rules the National Congress had passed in 2017 (Law 9504/97), which allow “content boost” as a form of paid publicity on the internet.

In that context, one of the digital elements that called attention were the bots. A report published between the first and second rounds of the elections showed that up to 10.4% of all interactions on Twitter and 13.8% among Bolsonaro supporters were initiated by bots. During that period, bots in favor of the candidate mobilized 70.7% of automated interactions on Twitter. An average 1.5 million tweets per day were about the candidates at that point in the elections. Between the 10th and the 16th of October, there were 852.3 thousand bot posts in total, 602.5 thousand in support of Bolsonaro, according to the Department of Public Policy Analysis at Getúlio Vargas Foundation (DAPP/FGV).

Fake news appears more connected to WhatsApp than to Facebook in the Brazilian context and its characteristics are closer to “disinformation” than to “fake news”. The newspaper El País monitored three public WhatsApp groups in favor of Jair Bolsonaro during September 2018 and stated that two of them distributed disinformation conspicuously. At the time, there were at least 100 public WhatsApp groups in favor of the candidate.

Among the known false information that circulated during the electoral period are the alleged fraud in electronic ballot boxes in the first round; the connection of Manuela D’Ávila (PCdoB) – who was running for vice-president with the Worker’s Party (PT) candidate Fernando Haddad – with Adélio Bispo de Oliveira, the criminal who stabbed Bolsonaro during the campaign; the claim that candidate Fernando Haddad would have praised incest in one of his books; the accusation of Joice Hasselmann, who was then running for Congress, that an important news organization would have received R$ 600 thousand to oppose Jair Bolsonaro’s candidacy; a text publicized by Eduardo Bolsonaro that claimed the Facebook group “Women United Against Bolsonaro” would have bought its more than a million members from a previously existing group; the controversial “gay kit” that Jair Bolsonaro himself presented during an interview on the prime time news show Jornal Nacional. On that occasion, Bolsonaro referred to the book “Aparelho Sexual e Cia - Um guia inusitado para crianças descoladas” (“Sexual Organs and Cia – An unusual guide for cool kids”), written by Helène Bruller and illustrated by the swiss cartoonist Phillipe Chappuis, indicating that the work would have been distributed to Brazilian public schools by the Ministry of Education.

The Facebook advertising and propaganda tactics of Donald Trump’s campaign were not used in the Brazilian context. There is no evidence that Bolsonaro’s candidacy paid for a big advertising campaign, neither for significant microtargeting actions on that social media. In fact, through years of work, the politician had secured a much more solid presence on Facebook than his opponents by the 2018 elections.

For instance, in 2014, our research group at PUC-Rio monitored the number of posts and average interactions on Facebook of eight federal deputies of the state of Rio de Janeiro that were then campaigning for reelection, including Jair Bolsonaro. According to our analysis, Bolsonaro presented at the time an average number of interactions (likes, comments, and shares) 29 times higher than the second place on the list with almost 30 thousand interactions per post. A study by DAPP/FGV shows Jair Bolsonaro’s 2018 campaign reached a peak 1.3 million interactions and 6.6 million views in a single live transmission.

Hence, while Facebook was the center of Donald Trump’s campaign in the US context of 2016, WhatsApp had a greater relevance in Brazil in 2018. The encrypted messaging app does not offer advertising services. In this platform, the message is only disseminated if users spontaneously forward it, at least theoretically. In addition, electoral legislation prohibits candidates from sending messages in mass and to phone numbers that were not consensually offered to their campaigns. This networked propaganda infrastructure is really efficient, from a communication perspective, when group administrators and the voluntary militancy are coordinated.

Regarding Jair Bolsonaro’s campaign, the creation of giant militant groups in parallel with a system of sophisticated tools (supposedly) funded by Brazilian business executives have raised academic and legal discussions. On that issue, the newspaper Folha de S. Paulo published on October 18, 2018, in the period between the first and second rounds of the election, that a number of business executives were (illegally) funding WhatsApp campaigns in favor of Bolsonaro with contracts worth up to R$ 12 million. A year later, the newspaper published an official statement from WhatsApp confirming “the activities of companies that offered the massive sending of messages, in violation of our terms of use, to reach a large number of people” during the Brazilian election. The case is currently in the Brazilian Superior Electoral Court (TSE).

I do not mean to say that Jair Bolsonaro was elected because of WhatsApp or Donald Trump because of Facebook and fake news. The election of public figures like Bolsonaro and Trump cannot be explained with few variables. Countless elements related to politics and democracy, the economy, social, family, and individual values, and so on, are behind these victories, along with a fragmented and complex communication system composed of diverse actors and media. For example, Bolsonaro not only had a strong digital presence (on WhatsApp, Facebook, and Twitter) but also obtained extensive journalistic coverage on television and news media in general because of the stabbing episode. Also, he benefited from the anti-Worker’s Party sentiment (“antipetismo”) brought on by corruption scandals and operation Car Wash.

Nevertheless, digital media are, unquestionably, a new element that radically affects electoral campaigns and, indeed, politics and democracy themselves. Online campaigns have become an important object of study for researchers in Political Communication in Brazil and around the world. Pipa Norris at Harvard University suggests the idea of “postmodern campaigns” or “postmodern political communication”. According to the author, in the modern period, television was at the center of political communication, along with the professionalization of political marketing, the increasing importance of opinion polls, and strategic planning commanded from a headquarters. Differently, the postmodern moment presents a permanent campaigning system, a multivariate and fragmented environment of media and actors, and a more volatile behavior on the part of the electorate.

Philip Howard, who is a professor at the Oxford Internet Institute, indicates the use of data is a central characteristic of current “hypermedia” electoral campaigns. “Hypermedia” campaigns develop in digital environments with targeted ads distributed online. In that context, citizens, who traditionally are consumers of political content, become potential producers and disseminators of messages as well. Hypermedia campaign distributes selected content to selected audiences, having computational propaganda and advertising, with the intense use of voters’ data, as one of its key elements.

Hence, “expanded hypermedia” would constitute a permanent campaign with the use of data and social media in the midst of a complex media environment that demands various types of political communication (pre-modern, modern, and postmodern) managed independently, but coordinated as a whole and complementarily. Researchers of the National Institute of Science and Technology in Digital Democracy (INCT.DD) gathered data from the Superior Electoral Court (TSE) showing that Brazilian federal deputy candidates in 2018 spent almost R$ 15 million boosting digital media content, but R$ 140 million publishing print material, including the famous “santinhos” (“holy cards”), and another R$ 80 million on militancy and street mobilization.

How can this type of political communication affect democracy? What consequences can it bring to democratic processes in Brazil and around the world? In 1927, a series of lectures the American philosopher John Dewey had proffered in the previous year in Kenyon College, Ohio, were published under the title “The Public and Its Problems”. In that work, Dewey presents the important notion of “the eclipse of the public”.

According to the philosopher, the complexity of the modern world and its countless forms of affecting societies had eclipsed the “public”, an element that, for Dewey, is central to democratic self-governance. Lost in the multifaceted dynamics of modern life, the “public” is unable to emerge from the deliberation of “common” interests, failing to constitute an actual “political community”. That phenomenon would damage democracy through the detachment between politics and the social development of modernity.

Almost a hundred years later, the constitution of a “public” to propose and debate common problems still seems to be a problem. However, that problem today may not have so much to do with the complexity of modern life as it has with the fragmentation and clustering of societies in multiple and varied mini-publics. By that I do not mean echo chambers or bubbles, which relate to the citizen’s informational aspect. The point is that disruptive communication builds political communities that have little connection to each other and often construct parallel political realities. Because of such weak interdependence, these realities tend to conflict when meeting in the public space, which at times can lead to violence. Here, we may call this phenomenon the “disruption of the public sphere”.

In fact, during the modern era of political communication, one of the greatest fears of political philosophy was the famous “tyranny of the majority”, when a larger portion of citizens – the majority – can impose democratic damage on a minority. In the context of mass democracy centered on television and opinion polls, the clearer the majority, the more politicians could adapt their positions and discourse to adjust to the largest number of voters. In that model, the thoughts that matter most are those of the majority, which are not necessarily correct, as noted by the American philosopher Walter Lippmann.

Today, the danger comes less from the politics of following majorities than from the capacity of certain political agents to create their own minorities through digital and targeted political communication. Basically, it matters little if most of the population is against the deregulation of guns, for instance. The important thing is to gather the largest number of minorities by the time of the elections. By doing this way, politicians can bring those interested in the preservation of family values together with those in favor of the deregulation of guns, the ultra-liberal in economics with those against political correctness, progressive gender politics, and the preservation of the environment. The threat for democracy today is no longer the tyranny of the majority. Currently, the real danger comes from the tyranny of these "fabricated minorities", constituted through digital political communication.

**

This work received the support of the Coordination for the Improvement of Higher Education Personnel of Brazil (CAPES) – Grant code 001.