During Brazil’s 2018 presidential elections, the campaign of the right-wing candidate Jair Bolsonaro invested in social media posts, apocryphal messages sent to WhatsApp groups and direct attacks on other political runners (Oliveira, 2018; Machado et al., 2018). His campaign raised awareness of the supposed ‘menace’ that communism posed to the country (Charleaux and Corsalette, 2018). This anti-communist rhetoric later became viral on social media. In one of these cases, an online montage included Che Guevara with red lips. The right-wing commentator Olavo de Carvalho had used the clip in one of his internet lectures, on the theme of ‘being gay and communist’. During a TV debate, the evangelical candidate Cabo Daciolo had mentioned the possibility of Brazil joining the URSAL, a mythical union of socialist republics, that led to many ‘tweetstorms’ (Chagas et al., 2019). A video denouncing a communist plot to corrupt the elections reached 1.3 million views on YouTube (Mota, Couta and Rocha, 2018).

These series of viral memes and absurdity on social media have obfuscated the broader and continuous use of anti-communist discourse since long before these elections. By proposing a thematic analysis, our aim was to explore the possibility that sub-themes attached to these mentions of communism could reveal more about the intent that helped to promote the topic to internet virality. To that end, we have departed from a well-known anti-communist backdrop that has permeated the politics of the country since the early days of the republic and resurfaced on different occasions. First, we situate our notion of virality by examining several cases in contemporary politics.

Viral content on the internet is a controversial concept, but in which one can see the potential to influence politics in multiple ways. ‘Viral’ is understood as a kind of spreadable media content that has its impact increased as individuals or communities eagerly share it (Postill, 2014; Jenkins et al., 2018) over a short period of time (Nahon et al., 2011). Content virality is a product of new technological innovations, which made unpredictable images possible and the political impact inevitable (Gerbaudo, 2019). The repetition of social media messages, which could suggest support for populist candidates, makes it challenging to tackle and avoid so-called ‘fake news’ (Groshek and Koc-Michalska, 2017). In the face of the limited coverage of journalistic fact-checking, a vast amount of disinformation can go viral at any time without proper investigation and testing for accuracy (Al-Rawi, 2017; Tucker et al., 2018).

The spread of viral content has already stirred contentious issues that have garnered much publicity from all sides of the political spectrum. In the United States, the case of the #occupy movement has echoed over the years to inspire other social justice movements protesting against inequality. The shooting of an unarmed black man triggered the #Ferguson hashtag, which has sparked broader protests for civil rights and exposed the dangers faced by ethnic minorities. In this research, what is meant by the virality of communism or anti-communism does not only depend on the quantification of social media shares – as communism spreads through a variety of expressions and terms – but on its horizontal transmission, either via memes, humour, junk content or fake news, to a range of other users and media platforms (Cebrian, Rahwan and Pentland, 2016).

The design of viral messages in politics, in the shape of memes, can also serve propaganda purposes, campaigns against mainstream candidates and alternative dissent groups (Denisova, 2019). Beyond the use of sensationalist language in politics, nowadays campaigns are more thoughtfully structured to appear highly ranked in search queries – thanks to big data manipulations (Shao et al., 2018; Venturini, 2019). The complexity that stems from the virality of communism lies in the extent to which it sensationalises fictional ‘dangers’, while hiding other agendas. For example, many YouTube videos have seriously debated the ‘implementation of communism’ as an imminent possibility in Brazil, but only to link it to society’s decay, the crisis of morality and ‘patriotic’ conspiracies (Ribeiro, Lasaitis and Gurgel, 2016; Melo, 2019). However, the promotion of viral communism could also unearth superficial and opportunist mentions of communism, as social media networks can still harbour old debates and revive political stereotypes.

The spread of anti-communist ideology on mass media is a well-known component of Herman and Chomsky’s propaganda model (Herman and Chomsky, 2011: 29). According to them, Western ideology has mobilised public opinion against a communist enemy as a mechanism of political control. We would like to discuss a similar situation, but in the context of the peculiar course of Brazil’s political history. The country has never managed to establish a communist regime (to the degree of Cuba or China, for example). The first Communist Party emerged in the mid-1920s. It owed much to the many waves of European immigrants moving to Brazil. The first communist groups consisted of small gatherings, in which Trotskyists and Stalinists never convened in full alliance (Dulles, 1985). In fact, fundamental difficulties prevented communist parties from flourishing. The lack of translated Marxist literature, for instance, and the high number of illiterate workers, limited membership of these associations to a few dozen, while the notion of ‘class consciousness’ seemed elusive to some of them (Rosa, 2017: 406).

Contrariwise, existing anti-communist rhetoric was fortified during the first government of Getúlio Vargas (1930–1945). In 1935 a failed communist-inspired uprising to oust him provoked a wave of official repression and censorship of communist ideas (Konder, 1980). Communism then represented the Soviet Union. Vargas’ dictatorship dismantled the Communist Party of Brazil (PC do B), which remained illegal for most of the twentieth century. The ‘Soviets’ embodied enemies who could also be infiltrating agents. They could come to ‘destroy the motherland and its basic values: religion, family, speech, and freedom’ (Motta, 2007: 234). While the origins of this ideological statement remain uncertain, the ‘pragmatic’ alignment with the United States on foreign affairs has come up as one of the hypotheses. Brazil’s automatic buy-in to US policy might have prevented any further advance of communist ideas until the 1960s (Steffens, 2011).

During the military regime that ruled Brazil after 1964, further references to communism in the political discourse sought inspiration in the Cuban Revolution and the growing tensions in Vietnam. The Truman Doctrine stated that the US Army ‘had to act against the communist enemy’. At the same time, France was trying to maintain control over its North African colonies, which rebelled by leaning left. These developments contributed, to some degree, to the strong criticism of the so-called communist ideas in Brazil, many of which had branched into Maoism, Trotskyism, and French-inspired movements (Arruda, 2017). The modern Communist Party of Brazil (PC do B) personified these European-inspired groups of dissidents. Founded in 1962, it has been diverse in its ideas and debates, but has never achieved a high number of affiliates (Giannotti, 2007: 165).

The 1964 military coup made the survival of any leftist opposition even harder. The military dictatorship stirred old fears of an existing communist plot within Brazilian society, accentuating its unhealed divisions. The local communist leadership remained confined to unionist action until the re-democratisation in 1984 (Santana, 1999). Napolitano (2014) points to the pivotal moment that guides much of the modern anti-communism, a discourse that lies in the anti-reformism of the 1964’s military coup’s ‘architects’, to whom communism equalled ‘subversion’. Coming from the top level of government, different types of guidance on how to act toward ‘communists’ have influenced political parties, unions, newspapers and opinion polls (Motta, 2007). During the overthrown government President João Goulart, which preceded the 1964 military coup d’état, conservative values had gained ground based on platforms such as the defence of private property and a national ‘morality’.

In the late 1970s and early 1980s, the anti-communist right invited Catholics to ‘crusade’ against Marxist ideals that supposedly threatened religious freedom (Mendes, 2004, 83). As a result, fear of a ‘red menace’ became established among Brazil’s elites (Motta, 2007). In the face of the growth of the unionist movement in the 1980s (Antunes, 1991), the fragmentation of a single ideal of what communism meant among the leftists might have caused the appropriation of the concept by right-wing groups. Lately, the Communist Party of Brazil (PC do B) has allied itself more closely with the moderate Workers’ Party (PT) than with the old Soviet-style communist ideals. In brief, anti-communism has become a more ventilated concept than even communism itself. More recently, the leader of the country’s biggest left-wing party has showcased close ties to communist parties elsewhere (Gazeta do Povo, 2019). As traditional leftists are discredited amid corruption charges, the so-called antipetismo has carried undertones of the old anti-communism (Davis and Straubhaar, 2019).

It is not possible to stress the whole range of shifts observed in communist and anti-communist discourses in Brazil, but this short background is sufficient to show how mentions to communism intersperses with values of militarism and conservatism. As seen, left-wing groups have, somehow, bought into it, but right-wing supporters were the ones to employ it against their enemies in modern times. The possibility of an elected right-wing candidate has made it more likely that links from different topics related to communism could surge on social media. We aim to detail ‘viral communism’ by dedicating research efforts to identify and analyse the mentions to this motto articulated in brief, but popular, social media posts. Whatever Twitter users were contending about communism, they might also have introduced other themes, as we shall discuss next.

Since the 2010 presidential elections, there have been profound changes in the way electoral campaigns run content online (Jamil and Sampaio, 2011). On the one hand, the diversity of users publishing messages has laid out a fresh universe of political affinities; on the other, it has become more difficult to identify the sources and mechanisms of dissemination. Despite not all Brazilian candidates having active accounts on popular social networks, the 2014 presidential debate saw growing online hostility between right- and left-wing militants on social media. Research conducted on WhatsApp in 2018 attests that audio, video and emojis made up the majority of messages in groups said to be strictly about politics (Caetano et al., 2018). On both Twitter and Facebook, users supporting the right are more likely to post politically polarising content, even though similar analysis regarding the left is still missing in the literature (Chaia and Brugnano, 2015; Penteado and Lerner, 2018).

Twitter is not the most popular network among Brazilians by the number of users (Chaffey, 2019). Nonetheless, this platform has turned out to become a critical forum because of politicians’ intense posting and the news media that it triggers. The use of Twitter had steered candidate endorsement on other platforms during the 2014 presidential elections (Ortiz et al., 2017). A new feature of the social media activities of users identified with the ‘new’ right relates to the assemblage of unrelated concepts and images, the mechanised spread of memes and extensive use of photo montages (Ribeiro et al., 2016). In 2019, President Jair Bolsonaro caused a series of controversies after posting offensive and provocative tweets (O Dia, 2019).

The extent to which the virality of communism on Twitter springs from this context of amalgamation of resources and evolves into other themes is the main interest of this research. We have pursued two main goals: first, as said, to map textual cues that could indicate other themes in posts related to ‘communism’ in the context of the 2018 elections; second, to qualify the source of these themes seen in viral stories. To that goal, we have sought a link between Brazil’s current affairs and its documented history of anti-communist episodes. This study has coded these linkages using a reflexive method, as we shall see.

Many studies have drawn on Twitter to engage with the criticality of user-generated content. This informs how individuals can understand complex events, for example the intricacy of popular interpretations of the Zika virus epidemics (Fu et al., 2016). Regarding political campaigns, Recuero et al. (2015) found that Twitter hashtags were relatively important to candidates’ engagement with users during the 2014 presidential run-up. But since then, the growth in the number of so-called ‘influencers’ has also played a crucial role in the spread of highly politicised content, like the images disseminated during former president Dilma Rousseff’s impeachment proceedings, in 2016 (Sandim et al., 2016). A black-and-white depiction of a young Rousseff showed a woman sitting in front of a military court for being a member of a Marxist organisation. Subsequently, populist candidates such as Bolsonaro, started to build a stronger presence on Twitter (Soares, Recuero and Zago, 2018).

Previous thematic analyses have based many studies on sociological health research (Reutter et al., 2009). Scholars have especially examined the subjectivity of perceptions on social media (Caton and Chapman, 2016). At least on an exploratory basis, this method has helped scholars to understand the linkages promoted by text cues such as slang, political jargon, localisms and connections to other broader subjects, such as mentions of race, religion and extremist propaganda (MacNair and Frank, 2017). The present study followed a similar path, but worked with a relatively small sample of tweets (n = 530), which was suited to cover a crucial period of Brazil’s recent political developments: the six months before the electoral run-up (April to October 2018).

Considering a theme as ‘an attribute, descriptor, element, and concept’ that ‘organises a group of repeating ideas,’ we followed detailed orientations from past studies before setting up the analysis (Vaismoradi et al., 2016: 101). Even if robotised resources or artificial intelligence produced some of the sample posts included here, this fact would not compromise our main aim of classifying the tweets and the linkages they establish. This type of ‘relationship code’ (Vaismoradi et al., 2016: 103) would not be possible via API research, for example, a method that depends on taxonomic repositories but does not capture random associations with a theme, such as rumours and non-objective language (Wilson et al., 2017). The same limitation exists in the Advanced Search resource, which extracts data from a repository that can be modified or curated over time (Driscoll and Walker, 2014). Conversely, this factor has not compromised our quest for qualifying selected tweets, as we have not sought to obtain the total breadth of tweets at the moment of their publication.

Our procedure was first to harvest results from a Twitter search query using the following keywords: ‘Brasil, comunismo and eleições’ in different orders. We limited our search to tweets published between the 7th of April – six months before the first round (when legal campaigning started) – and the 28th of October 2018 – the date of the second-round voting. Faced with limited methodological choices for starting the theme identification, we resorted to text mining with Nvivo software, from which we obtained the one hundred most repeated terms in the sample of tweets. After filtering the list to exclude repetition and clutter words, we observed how the fifty most repeated words appeared in fifty-three tweets (10% of the sample), before examining these terms’ relationship with the word communism. We relied on the repetition of these terms to determine meaning unities through which we could classify, label and translate into broader themes. After testing each other’s descriptions, we compared them to ensure accuracy among coders. Then we nominated these assumptions as working themes and developed our storyline.

We are aware of the critique that has warned against the potential subjectivity of this method (Vaismoradi et al., 2016). Hence, the coding followed a strict three-stage process of (i) origination, (ii) verification, and (iii) nomination (Table 1). In the face of a lack of paradigms that could inform us regarding acceptable sample sizes, we decided to use 50% of the total sample (245 tweets), which was a number that could sufficiently represent the total amount of tweets published in the period (530 posts). This amount also made for a realistic workload. Each coder considered the same set of 35 tweets per month from the last to the first tweet published.

As part of the procedure, each coder transcribed the content of each analysed tweet and recorded their decisions on a spreadsheet, adding reflexive notes when necessary (see Table 2). The analysis was performed twice to ensure consistency. A few categories were subjected to an intercoder reliability check (Cohen’s kappa). This revealed a 0.741 measure for the ‘democracy vs communism’ category, which appears as a reliable index in content analysis and for such a subjective analysis (Riffe et al., 2019). Below we discuss the main themes identified and their significance for this research. For privacy reasons, we have limited our mention of user information to their handles. By doing so, we have followed other research (e.g. Cavazos-Rehg et al., 2014) that has seen no problem in publishing users’ handles, as this mention – alone – cannot lead to the identification or harm of individuals whose content is already available to the public. Below we discuss the main themes identified and their significance for this research.

We describe how these themes have manifested, from the most recurrent to the least recurrent. We dwell on five main themes, while giving a brief theoretical grounding of these themes presented alongside respective examples from the sample. We have chosen to translate the content of each tweet from the Portuguese, as the English version better illustrates our explanation and eases context comprehension.

This thematic analysis has revealed that Twitter posts mentioning communism were mostly associated with suspicion of electronic ballot boxes, distrust of public institutions, corruption, and perceived threats to religion and morality. Regarding the anti-communist background, Communism remains a useful topic to boost other controversial statements, catch-phrases, montages and personal insults. ‘Viral communism’ emerges mainly from the replicability of well-known images, hashtags, images and humour, which are well-known affective devices for viral content (Miller, 2008). The principal source of our information was the textual element of tweets, but the relevance of visual and interactive material was also acknowledged (Stalcup, 2016). Given the enormous political crisis that loomed during this period in Brazil, the virality of communism, even when rooted in memes and entertainment, should not be misinterpreted as accidental or harmless.

In fact, the extent that hopelessness and disinformation have fed ‘viral communism’ have leveraged real everyday experiences of insecurity and poverty; isolation and hopelessness. These are far from being unfounded concerns – even though they are here associated with unreal factors. The mechanisms which the right has exploited to transfer these concrete issues to the elusiveness of communism in a time of voters’ insecurity deserves a better development in future research. In the case of the Brazilian election examined in this study, users might have acted under the influence of the impeachment of Dilma Rousseff, an ex-communist militant, or simply succumbed to political polarisation brokered by the internet. Whatever the cause, the way that the word communism was used shows that it evidently retains many of its negative moral, economic and cultural stereotypes. This research also reveals that social media design and algorithmic power can accommodate intangible and outdated arguments so that these topics can recur at the expense of relevant political debate.