Opening the black box: exploratory analysis of the Bolsonarista network on WhatsApp

Abstract

The victory of Jair Bolsonaro in the elections of 2018 was marked by several challenges from the point of view of political communication. With little TV time, his campaign was characterized by the strong performance in digital media, especially in WhatsApp. In this paper, we try to understand the communicative dynamics and the electoral strategies that were used by the campaign in the app. We analyzed 194.710 messages from 21 support groups for Bolsonaro between September 5 and November 11. The analysis showed that i) a small group of participants concentrated most of the posts; ii) the most shared links include other social media, such as YouTube and Facebook; iii) there were centralized actions of campaign to neutralize damages to the image of Bolsonaro, as well as to coordinate votes for its parliamentary base. The results show the potentialities of using WhatsApp, but also emphasize that the application should be understood within a broader communicative network.

Keywords
Political Communication; Elections; WhatsApp; Social Media; Jair Bolsonaro

Resumo

A vitória de Jair Bolsonaro nas eleições de 2018 foi marcada por diversos desafios do ponto de vista da comunicação política. Com pouco tempo de televisão, sua campanha se caracterizou pela forte atuação nas mídias digitais, especialmente no WhatsApp. Neste artigo, buscamos entender as dinâmicas comunicativas e as estratégias eleitorais que foram utilizadas pela campanha no aplicativo. Foram analisadas 194.710 mensagens de 21 grupos de apoio a Bolsonaro entre 5 de setembro e 11 de novembro. A análise nos mostrou que i) um pequeno grupo de participantes concentrou a maior parte das postagens; ii) entre os links mais compartilhados, predominam plataformas digitais, como o YouTube e o Facebook; iii) houve ações centralizadas de campanha para neutralizar danos à imagem de Bolsonaro, bem como coordenar votos para sua base parlamentar. Os resultados mostram as potencialidades do uso do WhatsApp, mas também ressaltam que o aplicativo deve ser entendido dentro de uma rede comunicativa em diálogo com outras plataformas.

Palavras-chave
Comunicação política; Eleições; WhatsApp; Redes Sociais; Jair Bolsonaro

Resumen

La victoria de Jair Bolsonaro en las elecciones de 2018 estuvo marcada por varias novedades desde el punto de vista de la comunicación política. Con poco tiempo en televisión, su campaña se caracterizó por el fuerte desempeño en las redes sociales, especialmente en WhatsApp. En este artículo, tratamos de comprender la dinámica comunicativa y las estrategias electorales utilizadas por la campaña en dicha aplicación. Analizamos 194.710 mensajes de 21 grupos de apoyo para Bolsonaro entre el 5 de septiembre y el 11 de noviembre. El análisis mostró que i) un pequeño grupo de participantes concentró la mayoría de los mensajes; ii) los enlaces más compartidos incluyen otras redes sociales, como YouTube y Facebook; iii) hubo acciones de campaña centralizadas para neutralizar los daños a la imagen de Bolsonaro, así como para coordinar los votos de su base parlamentaria Los resultados no solo muestran las potencialidades del uso de WhatsApp, sino también enfatizan que la aplicación debe entenderse dentro de una red comunicativa más amplia.

Palabras clave
Comunicación política; Elecciones; WhatsApp; Redes sociales; Jair Bolsonaro

Introduction¹

The 2018 campaign can be considered a milestone for electoral communication. Particularly for consolidating trends pointed out in previous claims from unofficial campaigns coordination and disinformation architectures that operate on multiple platforms. The victory of Jair Bolsonaro, built on the margins of television electoral advertising, was structured in an underground communication network, which had WhatsApp as one of the main means of disseminating content. During the electoral period, at least 1500 groups² worked as campaign committees, circulating ideas, promoting debates and organizing collective actions, such as caravans and demonstrations. The case emphasizes the recent transformations in the models of adoption and adaptation of technological infrastructures for digital campaign strategies.

Related studies point to the strategies of the candidate for the primaries of the United States Democratic Party, Howard Dean, in 2003, as pioneers of digital mobilization to carry out online and offline campaign activities through the MeetUp website and the political blogosphere (CHADWICK, 2013). In 2008, Barack Obama perfected the techniques by creating a portal called My Barack Obama (MyBO) that coordinated the speech and distributed fundraising and door-to-door convincing tasks to Democratic voters. However, these campaign initiatives begun by the citizen (GIBSON, 2015) were institutionalized, transparent and, to a large extent, controlled by the central committee. It is also important to highlight the role of more clearly concentrated actions, such as the Facebook content segmentation strategy used by Cambridge Analytica in the campaigns of Donald Trump and Brexit, both in 2016. What was observed in 2018 was the consolidation of professional practices in parallel with the efforts of volunteers through extensive network infrastructures that interconnect and feedback various platforms, often beyond the boundary of practices considered legal by the national regulatory framework.

In Brazil, the initiatives to build websites, systems and internal technological infrastructures to mobilize party members and supporters were residual. In addition, as far as it is known, there has also been no significant investment of resources in the micro-targeting of ads on Facebook, as in the USA. In 2018, WhatsApp emerged as the main novelty, especially due to the massive use of the application and its potential for rapid spread of messages. However, the platform holds important technical distinctions with the American systems employed in the previous decade. MyBO and the like are internal technological infrastructures, designed specifically for militants, who sign up and meet local communities who receive tasks and guidance from the committee. WhatsApp, in turn, is an instant messaging application with a limitation of 256 people per group. Although there is a significant effort to guide and control the content, the massive impact requires a capillary branch of tens or hundreds of thousands of groups to viralize the messages. Therefore, there is a component of fluid organization that coexists with the daily networks of family members, co-workers, cultural, religious and sports activities. From an academic point of view, what are continuity and rupture characteristics with the digital campaign strategies studied in previous disputes?

The specialized literature has been devoting more and more attention to the campaign’s operationalization of the application. This work seeks to contribute to these efforts by identifying the communicative patterns of pro-Bolsonaro WhatsApp groups. In order to do this, 194,710 messages were collected from 21 groups between September 5th and November 11th, 2018. The work applies quantitative and qualitative methods of analysis, combining covert observation and data mining.

Results indicate that, despite being underground, WhatsApp composes a network of circulation of electoral content in interface with other platforms, since YouTube, Facebook, Instagram and Twitter represent 65.2% of the links posted in the groups. In addition, there is a considerable concentration of messages in super participants, since the 100 most active users were responsible for 41.14% of the total messages. The most engaged activists acted as inter-group articulators: 88.3% of users post in only one group, while 11.7% sent messages to two or more. Findings from the qualitative study point out that these groups were used by the campaign to neutralize damage to Bolsonaro’s image, to coordinate vote seeking tactics and to build a climate of popular adherence to the candidacy.

This paper is divided into four parts. In the first, we analyze the specialized literature on the transformations of strategies and technological infrastructures used for the digital campaign, contextualizing the emergence of WhatsApp in Brazil and the scientific evidence available. In the second, we explain the methodological procedures which employed a combination of a quantitative and qualitative approach and the ethical dilemmas of research in public groups. In the third part, we detail the results of the empirical study, focusing on the communicative patterns of the bolsonarista WhatsApp and on the ethnographic-inspired qualitative immersion. At the end of the paper, we review the findings in the light of contemporary literature on digital campaign and point out research paths to be explored in future research.

Transformation of digital campaign strategies and infrastructures

The victory of a candidate from the political fringes of intolerant discourse who adopted a rental party months before the election, without an organizational structure in the states and without a party alliance that guaranteed television time for electoral propaganda, was unlikely. However, Jair Messias Bolsonaro was the first elected president with a media strategy predominantly supported by digital channels in Brazil. Despite the race that year having several atypical elements and the deterioration of democratic institutions, resulting from the effects of a political crisis that had dragged on for five years, the knife attack on Bolsonaro and the arrest of the main competitor, former President Luiz Inácio Lula da Silva, then placed first in public opinion polls, it is undeniable that the election raises questions for the research area of political communication.

Thus, in 2018, a pattern that was already being noticed in 2010 (Twitter) and 2014 (Facebook) became even more evident, that is, the use of social media platforms for the purposes of electoral propaganda, not only through the efforts of the candidates themselves and their committees, as well as hybrid initiatives by actors who gravitate around broader party networks (KREISS, 2016); and apocryphal actions of digital guerrilla operated by often anonymous nuclei and not always linked to parties (ALVES, 2016). What drew attention in 2018 were the coordination and articulation strategies between professionalized efforts - financed by the campaign committee or by other actors - for the production of messages, combined with the decentralization of dissemination by sympathetic groups.

The conjuncture of contemporary political-electoral communication can be analyzed by the key of the hybrid media system, which discredits a vision of linear evolution or replacement of the media, in favor of an intricate technological and social media that intersects and cross schedules between the mass media and digital platforms (CHADWICK, 2013). This pattern was evident in Donald Trump’s strategy of scheduling the mass media for his provocative tweets (WELLS, 2016); and in Barack Obama’s strategy to raise donations in digital crowd funding and investing resources in television advertising. A challenge of great relevance in this scenario is to apprehend the communicative traces of a multifaceted and hybrid infrastructure of digital technologies used for the production of electoral propaganda.

The concept of campaign initiated by citizens investigates the practices of systematizing the engagement of supporters in electoral actions through the creation of centralized digital spaces³ in which the central committee distributes tasks and unifies the discourse (GIBSON, 2015, ALVES, 2017). To this end, US parties have historically invested in the use and creation of various resources such as electronic systems, databases with voters lists, social media monitoring methodologies, among others. Kreiss (2016) calls prototype politics the development of human and technological resources in the Democratic and Republican parties to influence the field of technological performance, creating apparatuses and internal knowledge to the parties, such as massive lists about voters’ interests, behaviors and ideas.

Although the international cases and concepts being developed by the authors shed light on some practices that have been observed in Brazilian digital campaigns in recent years, it is essential to be cautious and prudent in the application of these approaches, observing the local institutional, legal and cultural peculiarities. Specifically, the election of former retired captain Jair Messias Bolsonaro, although it is a mostly digital example, has no parallels with the elaborate digital system infrastructures developed by the US Democratic or Republican parties, which were linked clearly and institutionally to the party organization.

In practice, Bolsonaro’s strategies are close to campaigns for Brexit in 2015 and Donald Trump’s in 2016, considering the triggering of underground unofficial communication strategies. Still, there are peculiarities to be pointed out. Kreiss and Mcgregor (2018) elaborate a pioneering analysis on how technology companies such as Facebook, Twitter, Google and Microsoft have allocated digital consultants dedicated to campaign committees in the United States, providing a series of boosting, monitoring and analytics services directly linked to the staff of candidates. They noticed the growth in spending on digital political ads by Trump and how his campaign did not directly use the robust digital infrastructure of the Republican National Committee (RNC), instead composing a lean team of professionals directly linked to specialists from social media companies.

Again, this scenario has some similarities, but it stands out for its distinctions. InternetLab report points out that Jair Bolsonaro did not declare spending on boosting content to the TSE (Superior Electoral Court), as well as investing insignificant amounts in digital strategy. In practice, the report suggests that Bolsonaro’s campaign was consolidated into two main pillars: (1) the strong personalized appeal around his image, capable of attracting millions of followers to his official channels; and (2) the active and intensive involvement of voters to amplify the reach of their messages and convert votes. On one hand, the declaration of campaign expenses to the TSE indicates reduced investment in consultancies, agencies and paid content on the internet; on the other hand, a survey by Patrícia Campos Mello published in Folha de S.Paulo⁴ revealed millionaire contracts by businessmen to buy services for shooting WhatsApp messages in large scale for the benefit of Bolsonaro; illegal practice that would constitute campaign donation by companies. How did this happen and what is known about the use of WhatsApp in 2018 so far?

WhatsApp is a private messaging application that allows the creation of limited groups of up to 256 people. The company reported that it has a base of 120 million users in Brazil⁵, reaching 56% of the population; and 98% of people who own a smart phone access the application daily according to Statista data⁶. A research by Agência Aos Fatos on the informational habits of Brazilians on the Internet support that social media or instant messaging applications are the main channel for accessing news, reaching 33.8% against only 10% of direct visits to news channel websites⁷.

The platform is seen as an ideal technological tool for viral marketing, that is, actions that are spontaneously and quickly passed on by people themselves (FOUAD, 2017). Interviews with participants in the Gezi Park protests in 2013 in the city of Istanbul, Turkey, indicated that the application was used for sociability among members and exchange of information for allowing faster download than Facebook groups, in addition to building environments for private and controlled access, more conducive to the development of bonds of affinity and trust within the group (HACIYAKUPOGLU; ZHANG, 2015). Respondents also informed that the information exchanged by the application was considered more reliable because, in general, they knew personally who recommended them, usually a family member, close friend or a colleague from social activism.

There is a growing volume of Brazilian research aimed at understanding problems related to the use of WhatsApp in the 2018 campaign. Baptista et al. (2019) analyzed Ibope opinion poll data and indicated that 57% of people consider the application to be an important source of political information. Furthermore, the data suggested that users perceive Facebook as a space more polluted by fake news than WhatsApp. The research also showed that 14.4% of respondents said they shared false information on the application in a conscious and purposeful way.

Some investigations were made using a technique of infiltrating groups in the application and daily monitoring of message exchanges⁸. A study about 100 political groups on WhatsApp point out the coordination of the flow of messages between bolsonarista groups through common administrators’ members who act as connectors and informal organizers of the networks, as the four moderators who managed 17 candidate groups (MACHADO; KONOPACKI, 2018). Another empirical analysis mapped the systematization of the spread of false news, fundamentally carried out by an interconnected organization of Bolsonaro supporters and showed how managers incite collective attacks and disqualifications, distribute tasks, align tactics and exclude users who contest or question the discourse of the discourse of the group (DOS SANTOS et al., 2019)⁹. Cesarino (2019) classified the bolsonarista network on WhatsApp in layers: (1) Official: they are the most exclusive circles that have the presence of Bolsonaro, his children, candidates and closest campaign managers; (2) Unofficial: content disseminated vertically and unilaterally by managers; (3) Militancy: these are more populous groups with horizontal exchanges between members; and (4) Personal networks: created by the users’ personal bonds; and (5) Media ecosystem: referring to the circulation of content and links from other web platforms. Thus, an interconnected infrastructure of groups is composed with internal and external message flows that generate the viralization processes on Whatsapp.

Still, there is a lot to empirically investigate about the use of this tool in the digital campaign, especially to understand if there are different standards and strategies by the candidates and what the distinctions are; in addition to understanding how private groups connect with other digital platforms and social media websites. This paper aims to contribute to these studies by offering initial findings derived from the analysis of data from the exchange of messages in pro-Bolsonaro groups on WhatsApp during the 2018 election. Thus, the investigation has an exploratory character in order to map and inform possible future questions about the political-electoral use of this tool. We focus on three points of analysis: (1) what the most disseminated information sources in the groups are; (2) what the pattern of concentration in the engagement of users in the production of messages is; and (3) what the traces of possible coordination in the intergroup messages flow are.

Methodology

WhatsApp is a great challenge for empirical researches that are interested in the data of the tool, namely, characteristics of groups, administrators, users and messages. During the election, the application was referred to in the press as a “black box” for not providing access to the information circulating in that space for researchers and society. WhatsApp, being a private channel for exchanging messages and with end-to-end encryption, does not have an interface for academic access to data. Not even aggregated and anonymous statistics on groups for political mobilization purposes are provided by the company.

In this way, research designs have been testing other forms of sample composition. The most common is a combination of methods, such as non-participatory observation and data mining. Thus, the process begins by identifying and entering public groups, which are discovered from lists that are available on the candidates’ websites. Once accepted into the group, WhatsApp allows the user to save messages to a file on the smartphone and export to a computer. All messages exchanged in the group are recorded, as well as multimedia content, such as photos and videos.

The exported file is a notepad that contains messages posted since the user joined each group. It is a text that records information in an unstructured way, that is, without an accessible format for analysis. For this reason, we developed a script in R statistical language to transform the notepad into a structured model, tabulated in rows and columns, with variables such as area code, anonymized user code, date, time, message, link, hashtags and whether it has multimedia content. To avoid questions about the exposure of data from users who did not agree to participate in the research, we replaced cell phone numbers in the tab by unique codes generated randomly.

Ethical issues on WhatsApp research

Studies on digital media always instigate reflections on the ethical limits of scientific research. Challenges include the growing blurring of the public-private boundary, as well as privacy dilemmas in times of intense data and metadata production. On top of that, the fast pace of creating new applications and technological devices, which change the circulation and information sharing networks, has to be considered (TOWNSEND; WALLACE, 2016, WILLIAMS; BURNAP; SLOAN, 2017). This last topic, in particular, applies to this work, since there are few documents on research ethics on social media that include specific guidelines on data collection on WhatsApp (VERHEIJEN; STOOP, 2016). This situation forces us to argue and justify the ethical guidelines that drove the research from the more general guidelines brought by the documents. There are three main issues that need to be addressed.

The first refers to the consent of the observed, which ideally should be requested from each individual, but which can be restricted to communicating to the group administrator, who is responsible for announcing it to all users (TOWNSEND; WALLACE, 2016, p. 6). In practice, however, these procedures would make research impracticable in radicalized groups and in polarized environments, such as neo-Nazi groups (TOWNSEND; WALLACE, 2016, p. 11). With proper proportions, this context resembles that of our research, which took place during a very polarized political campaign and in which university and scientific research were under systematic and virulent attack on the bolsonaristas groups. The researchers, consequently, would be enemies to be excluded, since they would be seen as infiltrated militants by individuals. Therefore, the covert research technique, provided by the resolution no. 5 10 from the National Health Council¹⁰ and used in related research, was applied to this work (CHAGAS; MODESTO; MAGALHÃES, 2019).

The absence of consent is not a problem when it comes to public environments, such as ethnographic observations on a street, for example. Groups on social media, however, are not so clear about these definitions, as they mobilize diffuse elements from each sphere. The recommendation of the main ethics guidelines in research on social media is to take the user’s expectation to be a possible target of observation. The question to be answered is, therefore, whether pro-Bolsonaro groups brought this expectation. We understand that our object is located in the interface between the public and the private, since they are less public than Facebook groups (which can have thousands of members) and, certainly, less closed than family groups. Data collection is justified for the following reasons: the groups i) had open invitations; ii) had many participants; iii) had some internal area code diversity; and iv) were created for the specific purpose of discussing public issues and mobilizing campaign actions. That said, although, eventually, some users may have the expectation of not being observed, there are sufficient reasons for us to understand the group as a space of interest to the academic community and society in general.

Finally, how the data will be presented in the analysis is what matters. In the case of our work, the absence of consent imposes the need to present the data in an anonymous and aggregated way. In this regard, there is no exposure of the observed and no damage to their images. The identity and individual uses were duly erased, bringing only collective behaviors and the content of messages to the focus of the analysis. In this regard, there is no exposure of the observed and no damage to their public images. No interactions between users were analyzed, nor were any experiments or interactions with members performed. The work was based on silent observation and presentation of aggregated statistics.

Data

The database of this research was composed of 194,710 messages published in 21 declared support groups for the candidate Jair Messias Bolsonaro, from September 5th to November 11th, 2018. The groups were chosen manually, from lists on bolsonaristas websites and their choice aimed to represent geographic distinctions in Brazil. However, the manual clipping is not intended to contain a representative sample of the support groups’ population, nor to account for other patterns of message exchange, such as family, work or leisure activities.

Analysis of the communicative patterns of the groups

The first axis of analysis was dedicated to understanding which information sources were most circulated on WhatsApp groups in the sample. For this, we recorded¹¹ the links to the websites that were published by users. Out of the 194,710 messages, 19,435 contained URLs, representing 9.81% of the total. In general, links are less published than images (19.2%) and videos (16.2%) and more common than audios (3.87%).

The analysis of the websites shows that 65.2% are from social media, such as Facebook, Twitter, Instagram and Youtube. Graph 1 shows the total number of links to each of the websites:

Graph 1
Shared link sources

The result raises the discussion of how instant messaging groups create spaces with greater privacy for their users, but are also interconnected to a networked media ecology that is visible on social media platforms. In this regard, WhatsApp works as an interconnection channel that promotes communication flows between other platforms, producing complex information networks that do not depend only on a website. In addition, as the reports published by ITS and the Public Agency have already suggested, managers disclose links and addresses to guide collective attacks or to expand the channel subscription list. In this first case, many indications to dislike videos of the singer Anitta were found, a result of her position in defense of #EleNão (#NotHim) and a general positioning of the groups against the artist’s musical style.

In addition, it is very interesting the role of YouTube as the website most recommended by bolsonaristas networks, with 8,956 messages, against 1,896 from Facebook. This suggests the strong adoption of an audiovisual grammar, above all, through the popularization of personalities that gain visibility and recognition from peers to talk about politics on the platform, the so-called youtubers.

In order to deepen this discussion, we raised which channels received the most recommendations, finding: 1) Jair Messias Bolsonaro (107); 2) Bernardo Küster, who introduces himself as a Catholic journalist who supports Bolsonaro; 3) PSDB (32); and 4) PT (31). The first two refer to the construction of a political content consumption base directly connected to the candidates or to journalists / youtubers who openly support the causes; the PSDB and PT are, again, guidance for negative comments. It is interesting to point out how opponents’ channels circulate in these groups, suggesting that there is a practice of monitoring the opponents’ discourse and an organization to direct militants to dislike or criticize the videos.

The only information website that appears on the top five is O Antagonista with 737 publications. This is an indication of a behavior against consuming news from the mass communication and an indication of informational habits oriented to digital native press, with more advocacy editorial lines and partisan content. There is the disclosure of addresses to Folha de S. Paulo, O Globo, Estadão, Revista Veja and R7, suggesting that these people have some contact with news from these organizations. However, more accurate qualitative research must be carried out to understand how these groups receive and interpret these texts.

The second issue to be explored is the concentration of user engagement in the production of messages. The concept of super-participants analyzes the behavior of the most active users in the production of messages in an online environment, measured by Graham and Scott (2014) as a minority that produces at least 2% of the messages in forums with less than 100 thousand posts or that creates more than 2 thousand in those with a total of more than 100 thousand.

In the sample data, there is a considerable concentration of activity. 5,288 different numbers that posted a message were found, of which 3.2% (173 users) were responsible for 50.12% of all publications. Graph 2 shows the cumulative sum of posts and the number of users. It is remarkable how there is a small group of 640 users who are responsible for 75% of the engagement registered in the sample and a long tail of people who have a lurker behavior, that is, they are very rarely involved in the discussions. According to the authors’ criteria for identifying super-participants, only seven users made more than two thousand posts.

Graph 2
Concentration of message production by users

Finally, this paper aims to identify the coordination features in the flow of intergroup messages. For this purpose, we recorded which users published in more than one group. The result shows that 88.3% of the people (4,670) participated in only one of the sample groups. On the other hand, there is a minority that, in fact, works systematically to circulate messages based on intergroup sharing. So much that there is a specific case that posted in 16 different groups. An article from Revista Época¹² shows how there were citizens hired by the campaign to create and manage spaces on WhatsApp, as well as an industrial production of memes and texts that fed the discussion in the period.

Graph 3
Concentration of message production by users

Graph 3 demonstrates users who have posted on more than one group. The y-axis contains a selection of profiles that have posted more than 200 messages and the x-axis indicates which group they have posted on. The color has a gradation from gray to black, indicating the total number of messages the user has posted on that group. It is noticeable that several profiles have aligned features, that is, they have posted on more than one group. As several groups are from different regions of Brazil, it is very likely that these managers are the curators of the content, also organizing the activities of the members and guiding the propagation dynamics of the flows.

Furthermore, we found a much higher average of posts when crossing the identification of the most engaged participants with those who post on more than one group. Users who posted on more than one group had an average of 129 posts compared to 23.8 of those who posted on just one group. Anstead (2015) argues that the current conjuncture of political communication reconfigures the concept of electoral coordination, as it used to be structured in bureaucratic strategies aligned with party leaders. In the digital campaign, coordination appears in loose and fleeting alliances, formed to amplify the dissemination of messages.

The results indicate that WhatsApp has interfaces to the campaign on the Internet in a general way, often being scheduled and scheduling communicative and performance actions on other websites, such as YouTube. In addition, the Brazilian case is distinct from European and American experiences, due to the political use of an instant messaging application rather than the parties’ electronic systems and historical databases on voters lists.

The findings contribute to understanding the coordination of the activities of Bolsonaro’s digital militants. On one hand, it is evident the formation of digital elites, minorities with high digital literacy, some paid other volunteers, to produce content, manage communities, approve members, orchestrate the flow of information, define tactics and distribute tasks, much like the concept of digital vanguards of Gerbaudo (2017). On the other hand, it cannot be ignored that, in a second moment, there is an involvement of people who share the messages from the campaign groups with family, work or leisure activities networks, as pointed out by Cesarino (2019). O Globo¹³ disclosed that Bolsonaro campaign’s communication team distributes content to 1,500 groups, which would reach a maximum of 384,000 people in a first wave of propagation. It is bound to affirm that for a viral electoral marketing strategy on WhatsApp to have national reach, it would depend, fundamentally, on the role of these people to share messages with their networks of personal contacts, creating wider propagation cascades. Otherwise, the tactic would have reduced effectiveness. That is, it is necessary to convince supporters and the electorate in general to share the content in the campaign groups in order to achieve the planned impact.

Qualitative immersion analysis

In addition to promoting Jair Bolsonaro and attacking his opponents, WhatsApp groups also had great strategic importance in conducting the campaign, directing themes, neutralizing attacks and guiding actions by the bolsonarista electorate. The widespread use of WhatsApp in a presidential campaign in Brazil was a novelty and provided several strategies for engagement and persuasion. In this topic, we analyze two strategies that stood out among the analyzed ones: the preemptive framing and the coordination of the proportional vote in the majority electoral system.

The debate on media frame is widely known in communication studies and refers to how themes are presented, by different actors and institutions, in the public debate (SCHEUFELE; YVENGAR, 2017). Some close concepts, such as agenda-setting and priming, related to the content, become themes in the public debate and, therefore, have a more universal character once they are guided. The framing, on the other hand, is not only in constant dispute between the interested actors, but also has its effectiveness related to the interaction established with the recipients of the content (SCHEUFELE; YVENGAR, 2017). In this regard, researches on framing usually consider factors such as the performance of journalistic vehicles, actors and political organizations, the influence of the broader cultural context, and also reflections on how framings are constructed, such as the amount, the order and emphasis of the information presented (TEWKSBURY; SCHEUFELE, 2009, MARQUES; MONT’ALVERNE, 2019).

The preemptive framing refers to a centralized strategy that seeks to provide interpretive frameworks about a situation that may or may not happen in the future (LAKOFF, 2017). The proposal, therefore, aims to anticipate interpretations about facts and events, providing an early framework, which will act as a shortcut for individuals to quickly organize events that can destabilize beliefs. The preemptive framing can be more direct, placing themselves as a previous counter narrative, or subtler, delegitimizing the future emitter without referring specifically to the theme. Analyzes of Donald Trump’s tweets show how the President of the United States used the strategy, for instance, to reframe anti-racist protests that would occur in football matches, framing them as anti-patriotic and disrespectful acts to the American flag (COOLEY, 2018).

In Jair Bolsonaro’s campaign on WhatsApp, three main uses of this strategy were highlighted: a) rumors and complaints about fraud at electronic voting machines, which provided the militancy with several arguments to contest a possible defeat; b) the circulation of a video of the then candidate Joice Hasselmann accusing Veja magazine of receiving R$600 million to slander Jair Bolsonaro; c) screenshots of an alleged WhatsApp conversation in which left-wing groups agreed to dress up as bolsonaristas and attack the # EleNão (#NotHim) protests.

The material on the alleged fraud at the electronic voting machines was abundant and diverse, and included, among others, the circulation of videos of digital influencers, an interview by the former Federal Police Chief Protógenes Queiroz and considerations by a professor, Diego Aranha, from the State University of Campinas (UNICAMP), on the alleged electronic voting machine violation. It is worth mentioning that this discussion was not restricted to WhatsApp groups, but was also raised in the public debate by figures such as Danilo Gentili and Carlos Bolsonaro himself. Bolsonaro’s victory removed the need to use this framework, but the narrative is sustained even after the electoral period¹⁴.

The video published on YouTube by the then candidate Joice Hasselmann on the alleged payment of R$600 million to some press vehicle was essential to neutralize damage to Bolsonaro’s image. A few days after the video, widely circulated on WhatsApp, Veja magazine published an article that promised to wreck Bolsonaro’s candidacy, but which, in practice, did not have much repercussion. In the groups, the magazine article was received without surprises and its true motivation (the alleged payment) invalidated the statements and evidence raised against the candidate. In fact, the “wrecking article” ended up being complementary to the accusation made by Joice Hasselmann, “confirming” its authenticity and working to build an increasingly suspicious atmosphere about the performance of the press.

The third emblematic case was the circulation of screenshots of a WhatsApp conversation in which alleged leftists agreed to attack the #EleNão (#NotHim) protests wearing shirts representing Bolsonaro. These images circulated through different groups and anticipated the interpretation of any possible aggression that could occur during the protests. There was no record of attacks and the matter was forgotten without further clarification.

The three cases allow us to observe the varieties of this political strategy and, above all, its excellent cost-benefit ratio. The cost for errors was very low and the gain from anticipating a case that actually occurred – for example, the case of Veja magazine - was extremely relevant. The case of electronic voting machines worked as a long-term framework, being constant throughout the campaign and involving the circulation of content from different sources with reference to the theme. In this regard, complaints from authorities, explanations from alleged experts and citizen registries constituted a plural framework that reinforced each other and collaborated to the construction of the framework before the vote. The other two cases have shorter temporality, focusing on specific events, but are distinguished in relation to anonymity, which occurred only in the anticipation of the #EleNão (#NotHim) protests.

Following what the literature indicates about the importance in the presentation of the framework, it is important to highlight that the preemptive framings that circulated on WhatsApp were inserted in an exclusive narrative and were shared by ordinary users, on the margins of official electoral communication. Among the three cases presented, the most “underground” was that of the #EleNão (#NotHim) protests, since that of Veja was published by the journalist Joice Hasselmann and that of the electronic voting machines was, in fact, an organization of different contents, mixing from public interviews to personal reports. All of them, however, were presented as exclusive information, which would guarantee the recipients a different perspective on future political events.

The second strategy was initiated between September 19th and 22nd and sought to coordinate the majority and proportional votes, through “digital flyers”. The main post listed Jair Bolsonaro’s allies for all positions in dispute in all states of the country, in addition to the Federal District.¹⁵ Entitled “Bolsonaro’s Brazil Team”, it had 112 nominations and it was widely spread among the groups analyzed, being posted 135 times, by 76 different cell phone numbers.

This type of post sought to achieve the so-called “coattail effect”, which consists of the relationship between the votes for majority candidacies and the number of seats achieved in proportional positions. There is a debate about which majority position really matters to leverage proportionals (SAMUELS, 2000, BRAMBOR; CLARCK; GOLDER, 2004) and also about the reverse coattail effect, in which candidates for congressmen and councilors organize their territorial bases to aggregate votes for majority positions (AVELINO; BIDERMAN; BARONE, 2012, MACIEL; VENTURA, 2017). Despite more specific discussions on the topic, it is a strategy widely used by political parties and candidates, who struggle to demonstrate proximity and obtain explicit support from candidates for majority positions.

The election of Jair Bolsonaro is an interesting case to think about this issue, as PSL federal bench went from an elected candidate in 2014 to 52 in 2018, 38 of whom were neophytes. The circulation of “flyers”, working as an informational shortcut to guide the vote, shows how the proportional candidates sought to approach the figure of Bolsonaro. The identification of this strategy allows to complement the frame pointed out by the accusation published in Folha de S.Paulo about mass shootings by WhatsApp on the eve of the elections¹⁶, showing how the application was an important piece not only for the election of Bolsonaro, but for the election of his allies in positions of different levels of government.

The publication of the flyer “Bolsonaro’s Brazil Team” was eventually accompanied by a text that emphasized the importance of electing congressmen from PSL or aligned parties, due to the need to have a strong bench in the Congress for the approval of projects promised by the president. There is nothing new in the strategy of associating proportional and majority candidates, but in this case, the change in the circulation format, which allows great capillarization through user shares, in addition to easy storage and access by cell phone, is noteworthy. It is noted, therefore, how WhatsApp expands the horizon of campaign actions, establishing means for high-speed and intense communication.

Discussion and conclusions

This paper aimed to identify some communicative patterns and campaign strategies that were used on WhatsApp groups to support Jair Bolsonaro’s candidacy. The results of the analysis oppose both narratives about the spontaneity of popular adherence and the views that isolate WhatsApp as a defining element of the presidential campaign.

Bolsonaro’s campaign was marked by the short time spent on television and the supposed economy of resources. Anchored in the narrative of a popular uprising, it was built as an exemplary initiative of the “new politics”, detached from the usual electoral practices and ties. The analysis of the WhatsApp groups, however, brings several elements that indicate that, despite the new media, there was articulation and direction of communication by WhatsApp. The identification of super-participants (with a high concentration of posts and inter-group activity), as well as the coordination of votes and the anticipation of framings, indicate the strategic use of the application in the election campaign.

Regarding the campaign strategies, the observation shows how WhatsApp brought agility and capillarity to Bolsonaro’s communication. The ability to create certain frameworks for events that occurred in the heat of the campaign is an example of how WhatsApp allows communication that is more closely linked to the communicative daily lives of citizens, escaping the rigidity of HGPE or press coverage.

This centrality of WhatsApp in the redefinition of strategies, however, should not be considered in isolation. The massive presence of links to videos on YouTube or Facebook posts corroborates the view of Cesarino (2019) and Dos Santos et al. (2019) about the existence of a broader network of information circulation and content production in which WhatsApp appears as a centerpiece within a larger system. Understanding the connections and sharing dynamics between the content produced for each platform is one of the challenges that the new communicative environment imposes.

Finally, the importance of further studies on dissemination, reception and engagement strategies applied to WhatsApp specific dynamics is highlighted, as well as the uses and appropriations of the content in circulation to form opinions and perceptions of citizens with varying degrees of political engagement.

• 1
  This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001.
• 2
  Available at: https://oglobo.globo.com/brasil/time-digital-de-bolsonaro-distribui-conteudo-para-1500-grupos-de-whatsapp-23134588. Accessed on: May 30th, 2020.
• 3
  The most well-known case was My Barack Obama (MyBO), a digital portal for registering voters sympathetic to the Democrat’s candidacy for the post of President of the United States in 2008 in which strategists organized and distributed volunteers in activities such as canvassing - door to door visits to present guidelines, recruit new volunteers or collect donations.
• 4
  Available at: https://www1.folha.uol.com.br/poder/2018/10/empresarios-bancam-campanha-contra-o-pt-pelo-whatsapp.shtml. Accessed on: May 30th, 2020.
• 5
  Available at: https://olhardigital.com.br/noticia/whatsapp-revela-numero-de-usuarios-no-brasil/68604. Accessed on: May 30th, 2020.
• 6
  Available at: https://www.messengerpeople.com/pt-br/whatsapp-no-brasil/. Accessed on: May 30th, 2020.
• 7
  Available at: https://aosfatos.org/noticias/11-graficos-que-mostram-como-as-pessoas-consomem-noticia-na-internet/. Accessed on: May 30th, 2020.
• 8
  Available at https://www.bbc.com/portuguese/brasil-45666742. Accessed on: May 30th, 2020.
• 9
  Available at: https://apublica.org/2018/10/grupos-pro-bolsonaro-no-whatsapp-orquestram-fake-news-e-ataques-pessoais-na-internet-diz-pesquisa/. Accessed on: May 30th, 2020.
• 10
  Available at: https://bvsms.saude.gov.br/bvs/saudelegis/cns/2016/res0510_07_04_2016.html Accesses on: May 30th, 2020.
• 11
  The registration of addresses was performed during the notepad tabulation process using the mining technique and organization of text called regular expressions. The code identifies patterns, sequences and variations in the textual structures of website addresses.
• 12
  Available at: https://epoca.globo.com/como-funciona-maquina-de-whatsapp-que-pode-eleger-bolsonaro-23180627. Accessed on: May 30th, 2020.
• 13
  Available at: https://oglobo.globo.com/brasil/time-digital-de-bolsonaro-distribui-conteudo-para-1500-grupos-de-whatsapp-23134588. Accessed on: May 30th, 2020.
• 14
  Available at: https://www1.folha.uol.com.br/poder/2020/03/sem-apresentar-provas-bolsonaro-diz-que-houve-fraude-eleitoral-e-que-foi-eleito-no-1o-turno.shtml. Accessed on: May 30th, 2020.
• 15
  A few positions were not filled, especially in Northeastern states.
• 16
  Available at: https://www1.folha.uol.com.br/poder/2018/10/empresarios-bancam-campanha-contra-o-pt-pelo-whatsapp.shtml. Accessed on: May 30th, 2020.

Referências

• ALVES, M. Campanha não oficial–A Rede Antipetista na eleição de 2014. Fronteiras-estudos midiáticos, v. 19, n. 1, p. 102-119, 2016. Disponível em: http://revistas.unisinos.br/index.php/fronteiras/article/view/fem.2017.191.10 Acesso em: 31 maio 2019.
  » http://revistas.unisinos.br/index.php/fronteiras/article/view/fem.2017.191.10
• ALVES, M. Redes de campanha na eleição do Rio de Janeiro em 2016. Compolítica, v. 7, n. 2, p. 87-120, 2017. Disponível em: http://compolitica.org/revista/index.php/revista/article/view/122 Acesso em: 31 maio 2019.
  » http://compolitica.org/revista/index.php/revista/article/view/122
• ANSTEAD, N. Data-driven campaigning in the 2015 United Kingdom general election. The International Journal of Press/Politics, v. 22, n. 3, p. 294-313, 2017. Disponível em: https://journals.sagepub.com/doi/10.1177/1940161217706163 Acesso em: 31 maio 2019.
  » https://journals.sagepub.com/doi/10.1177/1940161217706163
• AVELINO, G.; BIDERMAN, C.; BARONE, L. Articulações intrapartidárias e desempenho eleitoral no Brasil. Dados – Revista de Ciências Sociais, v. 55, n. 4, p. 987-1.013, 2012. Disponível em: http://www.scielo.br/scielo.php?pid=S0011-52582012000400005&script=sci_abstract&tlng=pt. Acesso em: 31 maio 2019.
  » http://www.scielo.br/scielo.php?pid=S0011-52582012000400005&script=sci_abstract&tlng=pt
• BAPTISTA, E. A.; ROSSINI, P.; DE OLIVEIRA, V. V.; STROMER-GALLEY, J. A circulação da (des) informação política no WhatsApp e no Facebook. Lumina, v. 13, n. 3, p. 29-46, 2019. Disponível em: https://periodicos.ufjf.br/index.php/lumina/article/view/28667 Acesso em 6 de jun. 2020.
  » https://periodicos.ufjf.br/index.php/lumina/article/view/28667
• BRAMBOR, T.; CLARCK, W.; GOLDER, M., Replication data for: Gubernatorial and Presidential Coattails in Brazil, Harvard Dataverse, v. 1. 2004. Disponível em: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/82CWKQ Acesso em: 31 maio 2019.
  » https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/82CWKQ
• CESARINO, L. Populismo digital: roteiro inicial para um conceito, a partir de um estudo de caso da campanha eleitoral de 2018. Manuscrito em desenvolvimento Disponível em: https://www.academia.edu/38061666/Populismo_digital_roteiro_inicial_para_um_conceito_a_partir_de_um_estudo_de_caso_da_campanha_eleitoral_de_2018 Acesso em: 1 mar. 2019.
  » https://www.academia.edu/38061666/Populismo_digital_roteiro_inicial_para_um_conceito_a_partir_de_um_estudo_de_caso_da_campanha_eleitoral_de_2018
• CHADWICK, A. The hybrid media system: Politics and power. Oxford University Press, 2017.
• CHAGAS, V.; MODESTO, M.; MAGALHÃES, D. O Brasil vai virar Venezuela: medo, memes e enquadramentos emocionais no WhatsApp pró-Bolsonaro. ESFERAS, p. 1, 2019.
• COOLEY, D. Of twit-storms and demagogues: Trump, illusory truths of patriotism and the language of twittersphere. In: LOCKHART, M. (ed.) President Donald Trump and His Political Discourse New York: Routledge, 2018.
• DOS SANTOS, J. G. B.; FREITAS, M.; ALDÉ, A.; SANTOS, K.; CUNHA, V. C. C. WhatsApp, política mobile e desinformação: a hidra nas eleições presidenciais de 2018. Comunicação & Sociedade, v. 41, n. 2, p. 307-334, 2019. Disponível em: https://www.metodista.br/revistas/revistas-metodista/index.php/CSO/article/view/9410 Acesso em 6 de jun. 2020.
  » https://www.metodista.br/revistas/revistas-metodista/index.php/CSO/article/view/9410
• FOUAD, N. Viral marketing effect on digital knowledge acquisition: WhatsApp as a model. Alexandria, v. 27, n. 1, p. 10-29, 2017. Disponível em: https://journals.sagepub.com/doi/pdf/10.1177/0955749017718705 Acesso em: 31 maio 2019.
  » https://journals.sagepub.com/doi/pdf/10.1177/0955749017718705
• GERBAUDO, P. Social media teams as digital vanguards: the question of leadership in the management of key Facebook and Twitter accounts of Occupy Wall Street, Indignados and UK Uncut. Information, Communication & Society, v. 20, n. 2, p. 185-202, 2017. Disponível em: https://www.tandfonline.com/doi/abs/10.1080/1369118X.2016.1161817?journalCode=rics20 Acesso em: 31 maio 2019.
  » https://www.tandfonline.com/doi/abs/10.1080/1369118X.2016.1161817?journalCode=rics20
• GIBSON, R. K. Party change, social media and the rise of ‘citizen-initiated’campaigning. Party politics, v. 21, n. 2, p. 183-197, 2015. Disponível em: https://journals.sagepub.com/doi/full/10.1177/1354068812472575 Acesso em: 31 maio 2019.
  » https://journals.sagepub.com/doi/full/10.1177/1354068812472575
• GRAHAM, T.; WRIGHT, S. Discursive equality and everyday talk online: The impact of “superparticipants”. Journal of Computer-Mediated Communication, v. 19, n. 3, p. 625-642, 2014. Disponível em: https://onlinelibrary.wiley.com/doi/full/10.1111/jcc4.12016 Acesso em: 31 maio 2019.
  » https://onlinelibrary.wiley.com/doi/full/10.1111/jcc4.12016
• HACIYAKUPOGLU, G.; ZHANG, W. Social media and trust during the Gezi protests in Turkey. Journal of Computer-Mediated Communication, v. 20, n. 4, p. 450-466, 2015. Disponível em: https://onlinelibrary.wiley.com/doi/full/10.1111/jcc4.12121 Acesso em: 31 maio 2019.
  » https://onlinelibrary.wiley.com/doi/full/10.1111/jcc4.12121
• KREISS, D. Prototype politics: Technology-intensive campaigning and the data of democracy. Oxford University Press, 2016.
• KREISS, D.; MCGREGOR, S. C. Technology firms shape political communication: The work of Microsoft, Facebook, Twitter, and Google with campaigns during the 2016 US presidential cycle. Political Communication, v. 35, n. 2, p. 155-177, 2018. Disponível em: https://www.tandfonline.com/doi/abs/10.1080/10584609.2017.1364814?journalCode=upcp20 Acesso em: 31 maio 2019.
  » https://www.tandfonline.com/doi/abs/10.1080/10584609.2017.1364814?journalCode=upcp20
• LAKOFF, G. Trump’s twitter distraction. George Lakoff’s Blog, 27 mar. 2017. Disponível em: https://georgelakoff.com/2017/03/07/trumps-twitter-distraction/ Acesso em 25 maio 2020.
  » https://georgelakoff.com/2017/03/07/trumps-twitter-distraction/
• MACHADO, C.; KONOPACKI, M. Poder Computacional: Automação no uso do WhatsApp nas Eleições. Disponível em: https://feed.itsrio.org/poder-computacional-automação-no-uso-do-WhatsApp-nas-eleições-e969746d231f Acesso em: 1 mar. 2019.
  » https://feed.itsrio.org/poder-computacional-automação-no-uso-do-WhatsApp-nas-eleições-e969746d231f
• MACIEL, N.; VENTURA, T. O Partido dos Trabalhadores na Câmara dos Deputados: a evolução das bases socioeconômicas e territoriais (1994-2014). Opinião Pública (UNICAMP), v. 23, p. 96-125, 2017. Disponível em: http://www.scielo.br/scielo.php?pid=S0104-62762017000100096&script=sci_abstract&tlng=pt. Acesso em: 31 maio 2019.
  » http://www.scielo.br/scielo.php?pid=S0104-62762017000100096&script=sci_abstract&tlng=pt
• MARQUES, F. P. J. A.; MONT’ALVERNE, C. Conflitos, barganhas e desprezo pelo interesse público: enquadramentos do congresso nacional em editoriais da Folha de S.Paulo. REVISTA FAMECOS (ONLINE), v. 26, p. 30158, 2019.
• SAMUELS, D. The Gubernatorial Coattails Effect: Federalism and Congressional Elections in Brazil. The Journal of Politics, v. 62, n. 1, p. 240-253, 2000. Disponível em: https://www.jstor.org/stable/2647606 Acesso em: 31 maio 2019.
  » https://www.jstor.org/stable/2647606
• SCHEUFELE, D. A.; IYENGAR, S. The State of Framing Research: A Call for New Directions. In: KENSKI, K.; JAMIESON, K. The Oxford Handbook of Political Communication Oxford University Press, 2017.
• TEWKSBURY, D.; SCHEUFELE, D. A. News framing theory and research. In BRYANT, J.; OLIVER, M. B. (eds.), Media effects: Advances in theory and research (p. 17–33). Hillsdale, NJ: Erlbaum, 2009.
• TOWNSEND, L.; WALLACE, C. Social Media Research: a guide to ethics. Aberdeen: University of Aberdeen, 2016. Disponível em: https://www.gla.ac.uk/media/media_487729_en.pdf Acesso em: 31 maio 2019.
  » https://www.gla.ac.uk/media/media_487729_en.pdf
• VERHEIJEN, L.; STOOP, W. Collecting Facebook Posts and WhatsApp Chats. In: SOJKA P.; HORÁK A.; KOPEČEK I.; PALA K. (eds.) Text, Speech, and Dialogue TSD 2016. Lecture Notes in Computer Science, vol 9924. Springer, Cham. 2016. Disponível em: https://link.springer.com/chapter/10.1007/978-3-319-45510-5_29 Acesso em: 31 maio 2019.
  » https://link.springer.com/chapter/10.1007/978-3-319-45510-5_29
• WELLS, C. et al How Trump drove coverage to the nomination: Hybrid media campaigning. Political Communication, v. 33, n. 4, p. 669-676, 2016. Disponível em: https://www.tandfonline.com/doi/abs/10.1080/10584609.2016.1224416?journalCode=upcp20 Acesso em: 31 maio 2019.
  » https://www.tandfonline.com/doi/abs/10.1080/10584609.2016.1224416?journalCode=upcp20
• WILLIAMS, M.; BURNAP, P., SLOAN, L. Towards an Ethical Framework for Publishing Twitter Data in Social Research: Taking into Account Users’ Views, Online Context and Algorithmic Estimation. Sociology, v. 51, n. 6, p. 1149–1168. 2017. Disponível em: https://journals.sagepub.com/doi/10.1177/0038038517708140 Acesso em: 31 maio 2019.
  » https://journals.sagepub.com/doi/10.1177/0038038517708140

Publication Dates

• Publication in this collection
  04 Dec 2020
• Date of issue
  Sep-Dec 2020

History

• Received
  16 June 2019
• Accepted
  19 June 2020