Memetic persuasion and WhatsAppification in Indonesia’s 2019 presidential election

Introduction

The influential role of the instant messaging app WhatsApp in mobilising voters around elections is well documented (Cheeseman et al., 2020; Dwyer et al., 2019; Evangelista and Bruno, 2019; Maweu, 2019; Murgia et al., 2019; Tapsell, 2018). This work has highlighted how this communication platform ¹ enables state or corporate players to spread carefully crafted messages that appropriate Internet cultural forms (such as memetic content) through private and public WhatsApp groups in a way that ‘mimics spontaneous grassroots mobilisations but . . . in reality has been organised’, a practice commonly known as ‘astroturfing’ (Monbiot, 2010, see also Bradshaw and Howard, 2017; Kelly et al., 2017). Scholars have also signalled the platform’s end-to-end encryption as limiting possibilities for content moderation and external oversight and hence allowing misinformation and political-sponsored content to circulate freely (Evangelista and Bruno, 2019; Maweu, 2019; Reis et al., 2020).

The use of WhatsApp as a tool for political campaigning is especially important in places that have historically experienced relatively low levels of Internet penetration, such as Malaysia, Indonesia, Brazil, India, Sierra Leone, Kenya and Nigeria. In those countries, the increasing availability of smartphones has afforded mobile Internet access to many users who do not belong to the urban middle-class cohorts that have conventionally been the target of social media campaigns (Baulch et al., 2020; Pereira and Bojczuk, 2018 ; Suwana, 2019). At the same time, since WhatsApp’s rapid popularisation in the late-2000s and early-2010s, urban middle-class social media users have also added this platform to their social media repertoires. Therefore, in many places, the common use of WhatsApp across social classes makes it a useful tool for political campaigners seeking to sway the opinion of large swathes of the population in their favour.

This article contributes to the growing literature on WhatsApp by studying how political parties use this platform in their ‘cybertrooping’ strategies. ‘Cybertrooping’ is the setting up of human teams that create and disseminate content across digital platforms, and by which governments or political parties promote their agenda and/or undermine detractors (Bradshaw and Howard, 2017; Evangelista and Bruno, 2019; Kelly et al., 2017). It does so through a specific case study: the campaign of the incumbent presidential candidate, Joko Widodo, in Indonesia’s 2019 general elections. In Indonesia, by the most conservative estimates, smartphone ownership tripled between 2014 (24%) and 2019 (63.29%; Heimerl et al., 2015), prompting similar increases in monthly active users of WhatsApp in the country. ² Not surprisingly then, as in many places that held polls in 2019 (Brazil, Nigeria and India), WhatsApp emerged as a key platform hosting online campaigners’ attempts to secure swing votes, especially those of millennials who constituted a 42% of the voting population, through dissemination of campaign messages as memetic media, the humorous and visual in character of which were deemed efficacious at engaging young voters (Kartika and Sinatra, 2018). The following research questions inform our study: How did the Jokowi team use ‘cyber trooping’ for their political campaign? How was WhatsApp used during this campaign?

‘Cybertrooping’ (called ‘buzzing’ in Indonesia) often involves ‘political astroturfing’, a concept used to describe the surreptitious manipulation that takes place when state or party actors post campaign messages under the guise of ordinary users (Bradshaw and Howard, 2017; Johns and Cheong, 2019). When engaging in astroturfing, political parties typically pay people who pretend to be ordinary citizens acting independently. But astroturfing can also involve a more subtle practice of making sponsored political messages appear as though they emerged organically from grassroots movements. Below, we show how political elites use meme culture’s association with popular voice to ‘astroturf’ public discourse, and how WhatsApp’s end-to-end encryption, which obstructs external oversight, facilitates astroturfing.

Existing scholarship on the use of WhatsApp in election campaigns shows how memetic media (Milner, 2016) facilitate astroturfing because they are inherently ludic cultural forms that are commonly understood as instances of popular voice (Evangelista and Bruno, 2019; Tapsell, 2018; Tay, 2014). In this article, we consider memetic media as media texts (e.g. images, videos, screenshots, and GIFs) that are created in a way that invites transformation by users who, through their repurposing of these media in new contexts, give them new meanings (Milner, 2016). Memetic media can become a popular Internet meme that triggers endless memetic appropriations, but it can also be ‘a consistently shared, innovatively applied inside joke between two friends [. . .] even if the spread stops there’ (Milner, 2016: 39). Early studies of the political use of memetic media, or what Tay (2014: 47) calls ‘LOLitics’ – the ‘combination of Internet memes and political humour’ – positioned it as an instance of popular voice and vernacular creativity (Jarvis, 2014; Shifman, 2014). Yet, more recent work shows that memetic media can equally serve the interests of elites in their attempts to shape public discourse (Abidin, 2020; Evangelista and Bruno, 2019). However, scholarship on Internet memes has not yet considered how WhatsApp may be implicated in the changing power dynamics of memetic media’s creation and circulation.

Based on interviews with eight Jokowi’s online campaigners, the article inquires into how his campaign team was structured to optimise effective creation and circulation, through WhatsApp groups, of campaign-related memetic media. We focus on the following three dimensions of the Jokowi campaign team’s use of ‘cyber trooping’ during the election: (1) the structure of the online campaign team and the oversight of paid workers to create and disseminate content through social media, (2) its modes of producing content, particularly memetic media, and (3) the way it used WhatsApp to circulate such content. The focus on WhatsApp is key, as it highlights the important role the app is playing in the manipulation of public discourse during elections, by virtue of the breadth of its user base (which extends well beyond the urban middle classes), and of its unique architecture (closed, private nature, and groups function). As Evangelista and Bruno (2019) point out in their study of Brazil, ‘due to the nature and architecture of WhatsApp, the visibility of content-spreading strategies is minimal, and this prevents users to realise they are being the target of persuasion strategies’ (p. 15). This observation points to the urgency of bringing to light some of these hidden persuasion strategies.

The article advances scholarship attending to astroturfing within social media election campaigning in several ways. First, our access to ‘buzzers’ (a term used in Indonesia to describe digital campaign specialists who are paid to create, comment, and spread information online) affords a detailed account of the highly orchestrated production, strategically timed release and carefully constructed networks involved in the use of WhatsApp for astroturfing. In this, the study builds on existing scholarship on the WhatsAppification of election campaigns (Cheeseman et al., 2020; Evangelista and Bruno, 2019; Tapsell, 2018). Second, it contributes to work on social media campaigning in Indonesia’s 2019 Presidential election, which has provided an account of the structure of online campaign teams (Saraswati, 2021), but has not examined in detail how both meme factories and WhatsApp are located within this structure, nor their implications for the work of astroturfing. Third, by examining how memetic media were circulated through WhatsApp, we show how this platform is helping to reshape the power dynamics implicated in the creation and sharing of memes.

The use of WhatsApp and memetic media during elections

Recent works on social media election campaigning describe processes for establishing well-oiled professionalised systems for the creation and dissemination through social media networks, including WhatsApp, of campaign messages. Two kinds of political persuasion through social media can be identified. First, data-driven campaigning (such as that implicating the Cambridge Analytica data breach), ³ by which political parties can easily micro-target different users through social media’s advertiser interfaces (Kerr et al., 2018: 11). A second kind of political persuasion through social media involves ‘cybertooping’ (Bradshaw and Howard, 2017; Evangelista and Bruno, 2019; Kelly et al., 2017). In their report, reviewing cybertrooping activities in 28 countries, Bradshaw and Howard (2017) identify five kinds of organisations implicated in attempts to sway public opinion using social media: government-sponsored, political party or politician-sponsored, privately contracted, volunteer groups, and paid citizens (e.g. ‘influencers’). In addition, they identify several kinds of cybertrooping behaviour, which ranges from having a ‘clear hierarchy and reporting structure, integration with campaign and party organisation, with content reviewed by superiors’ to ‘liminal membership but some coordination across teams’ and, at the least tightly structured end of the scale, ‘informal, liminal teams’ (Bradshaw and Howard 2017: 21).

As scholarship is beginning to emerge chronicling the use of WhatsApp to campaign in various countries across the globe, new patterns of social media campaigning are becoming apparent. WhatsApp campaigners rely on the accessibility of micro-publics represented by groups, and they use these public and private WhatsApp groups to spread their political messages, both manually and using automated bulk messaging. In the 2019 Brazilian elections, campaigners relied on marketing agencies experienced in automated bulk messaging to spread messages through WhatsApp groups that were public and web-searchable (Evangelista and Bruno, 2019). But in India, Indonesia, Malaysia, Nigeria and Sierra Leone, WhatsApp campaigners relied on private groups, which they accessed through existing social networks, to spread messages using manual means (Cheeseman et al., 2020; Dwyer et al., 2019; Murgia et al., 2019; Tapsell, 2018).

In different places, WhatsApp-based election campaigns manifest similarities not only in their modes for disseminating content on the platform (through groups), but also in the kinds of content deemed to most effectively shape voter preferences, and also in the structure of campaign teams that will facilitate dissemination of such content. Work on WhatsApp-based campaigns in Brazil, India and Nigeria (and also in Indonesia as this study shows) reveals common pyramid-like campaign team structures, consisting of small numbers of paid workers overseeing content creation and dissemination at the top, and large numbers of unpaid volunteers at the bottom, who sit at a distance from the political party machinery and are therefore freer to engage in misinformation and negative campaigning (Cheeseman et al., 2020; Murgia et al., 2019; Resende et al., 2019).

The scholarship reveals how this pyramid structure enables campaign teams to infiltrate WhatsApp groups in order to spark discussion or intervene in discussion in ways that are favourable to a candidate. In this way, campaign team staff, often on party payrolls, are able to assume the guise of everyday or ‘grassroots’ users to shape public discourse on WhatsApp, in a manner that resembles ‘astroturfing’ (see Johns and Cheong, 2019; Keller et al., 2020). Memetic media play an important role in WhatsApp campaigning, because their appeal to the senses through humour and imagery enables campaigners to commonly address the array of voting publics, with different linguistic, educational and class backgrounds, that assemble as WhatsApp groups, and because their satirical qualities makes them appear to be instances of popular voice, rather than sponsored persuasion (Evangelista and Bruno, 2019; Tapsell, 2018).

The use of memetic media for electoral campaigning on WhatsApp points to new developments in their production and circulation for political influence. Early scholarship on Internet memes positioned them as an index of popular voice. This work focused on the demotic qualities of memetic texts and their participatory diffusion and transformation among networks of Internet users (Jarvis, 2014; Shifman, 2014; Tay, 2014). More recent work on memetic media highlights how meme production and circulation have been ‘institutionalised’, that is, drawn into official election campaigns (Burroughs, 2020), state-run public awareness campaigns and private corporations’ advertising campaigns (Abidin, 2020). Scholarship discussing the use of memetic media in WhatsApp-based election campaigns also implicates political and economic elites in meme production, but more importantly, this work points to a shift in meme circulation (Evangelista and Bruno, 2019; Tapsell, 2018). What we see now is the emergence of centralised modes of producing memetic media, in which elites oversee the fixing of certain kinds of images to official narratives in ‘factories’ – a central point from which the memes fan out. These memes then get circulated through a complex structure of WhatsApp groups which are directly and indirectly supervised by elites.

Social media and elections in Indonesia

In Indonesia, social media election campaigning dates back to the Jakarta Gubernatorial election in 2012 (Lim, 2017; Saraswati, 2021; Tapsell, 2015). The elections saw the ascendance of current President Joko Widodo to the seat of Governor of the capital, after serving as mayor of Central Javanese city Solo for two terms from 2005. In 2012, Jokowi was nominated as the Democratic People’s Party (Partai Demokrasi Indonesia – Pembangunan, PDIP) candidate for Governor of Jakarta, and began to amass a sizable network of active volunteers (JASMEV: Jokowi-Ahok Social Media Volunteers) who used Facebook and Twitter to mobilise young middle-class urban voters’ support for him and his running mate, Basuki Tjahaya Purnama (Ahok) to the Governorship and Vice-governorship, respectively (Tapsell, 2015: 38). A rebranded JASMEV (Jokowi Advance Social Media Volunteers) went on to play an important role in Jokowi’s 2014 Presidential election campaign. Jokowi ascended to the Presidency after winning the 2014 election (Saraswati, 2021; Tapsell, 2015).

The evolution of social media campaigning in Indonesia has proceeded in tandem with the rapid expansion of social media’s user base beyond the urban middle class. The period between Jokowi’s election to Governorship of Jakarta in 2012 and the 2019 Presidential election saw large increases in the number of social media users in the country, as more and more people began to own smartphones and have access to an Internet connection. Between 2011 and 2020, Internet penetration more than tripled, rising from 17% to 64%. The same period saw similar increases in smartphone ownership and the use of mobile devices to access the Internet. In 2012, between 12% and 24% (reports vary) owned a smartphone, and most Internet traffic was generated by desktop access in Internet cafes (Heimerl et al., 2015; Hootsuite & We Are Social, 2011, 2015, 2019). By 2019, 60% of the population had access to the Internet, and 96% of Internet users were accessing the web through their mobile phones. Consequently, both WhatsApp’s and Facebook’s user base increased almost fivefold between 2015 and 2020, which was also favoured by Internet plans offering access to these apps for free (Hootsuite & We Are Social, 2015, 2019; Tapsell, 2015) . Most of those users (66%) fell within the 18–34 age range. This demographic was targeted by the Jokowi campaign because it represents 42% of the voting population (Kartika and Sinatra, 2018)

Assessments of social media campaigning in Indonesia were initially hopeful, but since 2019 scholars have begun to express concerns about its implications for democracy in the country. In 2015, Tapsell (2013) argued that social media were ‘an important battleground’ in the 2014 election that was ‘largely not controlled by [print media and television station-owning] oligarchs but rather dominated by volunteer groups and individual citizens’ (p. 41; see also Holmes and Sulistyanto, 2016). However, Saraswati (2021) sees JASMEV’s establishment in 2012 as a function of an ‘expanding political campaign industry, whose significance increased, and is being used by political elites to maintain their power’ (p. 43). Similarly, Rakhmani and Saraswati (2021: 5) argue that this ‘campaign industry’ coerces voters to participate in populist narratives, by constructing society as constituted by followers of one of two possible strong leaders, whose virtues are defined by their personal traits, rather than their policies (see also Tapsell, 2021). But these works do not consider how the unique features of social media and memetic media help afford the coercion they describe. In their study of the Jakarta 2017 Gubernatorial election, Lim (2017) argues that Facebook and Twitter herded people into ‘algorithmic enclaves’ that polarised public discourse, but this study predates the widespread uptake of WhatsApp, which shapes public discourse in new ways (i.e. not algorithmically). Our article contributes to work on social media election campaigning in Indonesia by attending to the role media technologies, specifically WhatsApp, plays in shaping public discourse.

Methods

Over the course of 2019 and 2020, we interviewed a total of eight campaigners supporting Joko Widodo’s campaign for re-election in 2019. These individuals were positioned variously in the ‘pyramid’ structure of the campaign team. Five of them were part of Jokowi’s official campaign team: One was in a leadership role in Jokowi’s social media campaign (Maman), one responsible for recruiting creators and overseeing their work (Maya), and three were meme creators who worked out of in a meme factory in Jakarta (Aziz, Pranoto and Sugih). In addition, we interviewed three ‘independent’ content creators loosely (but not officially) linked to Jokowi’s online campaign (Zadrak, Malik and Rudi). ⁴

Our research relied on Fiona’s long standing participation in pro-Jokowi support networks and the trust in them that ensued from that. We sought to assemble a cohort of informants that represented a variety of roles within the overall campaign structure. We adopted a snowball sampling method to build the cohort from pre-existing relationships of trust between Fiona and the campaigners, which is an appropriate way to approach potential research participants in contexts where the gathering of data is contingent upon such pre-existing relationships. We did not have the same access to pro-Prabowo meme creators. Our study is not a critique of Jokowi’s campaign specifically; it treats Jokowi’s social media campaign as a case study that can reveal much about recent developments in the co-option of popular Internet culture by political elites and corporate media institutions, and the role WhatsApp is playing in this process.

We used ‘scroll back interviews’ with our informants (Robards and Lincoln, 2017), which involve ‘co-analyses’ with participants. This involved participants showing Fiona some of the memetic texts that were created and circulated during the political campaign through their WhatsApp groups. We coded the interview data by sorting sections of the transcripts into seven categories, ⁵ relating to the three dimensions of WhatsApp campaigning studied here: the make-up and organisation of the online campaign team, its strategies for creating memetic media deemed to be aesthetically effective in ways useful to the campaign, and for using WhatsApp to circulate these memes. Our analysis of the data focused on these common themes instead of taking each of our interviewees’ statements at face value. We also compared our findings with previous literature that had already reported on similar patterns in the use of WhatsApp during elections. For the presentation of the findings, we did not add as figures in the article; the memetic texts that our participants showed to Fiona since the interviewees did not send these media to us nor did we collect WhatsApp data for this study. However, most of the memetic texts (or similar ones) discussed in this article can easily be found through a quick Google search.

Structure of Jokowi’s campaign team

Jokowi’s 2019 social media campaign team manifested a hybrid organisational form mixing party sponsorship of message production with vast volunteer networks for dissemination of messages, both by campaigners on the party payroll and by volunteers, and both through dedicated Jokowi support groups and through non-political ones such as those emerging from neighbourhood, religious, family, ethnic group and sporting affinities. As mentioned earlier, this was the first time WhatsApp groups had been exploited so intently as part of a political campaign strategy, as the app had seen considerable growth in uptake since the last general election in 2014. This hybrid organisational form has already been documented by previous literature examining WhatsApp political campaigning in Brazil (Evangelista and Bruno, 2019), Malaysia (Tapsell, 2018) and Nigeria (Cheeseman et al., 2020).

Our research affirms the existence of three layers of ‘campaign narrative making’ during the 2019 Indonesia election: positive campaigning, negative campaigning and black campaigning (Saraswati, 2021: 48), which our participants referred to as ‘white’ (positive), ‘red’ (negative) and ‘black’ (disinformation). ⁶ We also found that, similar to the Nigerian case (Cheeseman et al., 2020), those elements charged with ‘black’ campaigning operated at a distance from the official campaign. At the centre of the online campaign team sat the ‘white’ campaigners and their networks, who crafted memetic content from key campaign messages (i.e. amplifying policy announcements, circulating photographs of the candidate on the campaign trail, etc). Here, memetic media production was ‘factoried’: it involved hundreds of carefully recruited and closely supervised young meme creators (three of whom we interviewed – Aziz, Pranoto and Sugih) who worked out of a secret location in Central Jakarta. They were a part of a 200-member team who produced memetic texts, and circulated them on open social media platforms.

A second layer involved campaigners (two of whom we interviewed – Zadrak and Malik) who were responsible for creating ‘red’ memes (denoting attack messages). Unlike official ‘white’ meme creators, who only spread memes through open social media platforms, ‘red’ campaigners relied heavily on dedicated Jokowi WhatsApp support groups, as well as WhatsApp groups from their own personal networks (such as sporting, church and high school family groups). Their relationship with the official chain of command was informal, and they were by no means subjected to the same scrutiny applied to factoried meme creators. Two of the volunteer buzzers we spoke to, Malik and Zadrak, reported that volunteers were at liberty to make their own content which may or may not align with the thematic foci of the official campaign. Malik says,

Sometimes the volunteers make videos that the official campaign team doesn’t feel comfortable with, but we don’t care, because we are not part of the official team. (Interview, 7 February 2020)

However, Zadrak and Malik also revealed clear links between the official campaign team and the unofficial work undertaken in volunteer networks. Key volunteer coordinators played important roles in channelling official thematic foci and contents through volunteer networks, and this shows how these networks did not just spread grassroots generated messages, they also helped circulate those carefully planned, produced and supervised by paid campaign operatives. Malik was involved in a WhatsApp coordinating group with official campaign operatives to determine topics for campaigning, and Zadrak said that they received updates on daily themes from social media campaign director Maman and Kresna Abadi (not their real name) of Media Nusantara Citra (MNC), the media company that supported the campaign.

Third, on the fringes of the social media campaign team sat another layer of volunteers who also spread ‘white’ and ‘red’ memetic media through WhatsApp groups and, additionally, engaged in ‘black’ campaigning: perpetuating misinformation and hoaxes. We interviewed one such campaigner, Rudi, who told us that their job included infiltrating pro-Prabowo WhatsApp groups and planting misinformation.

The existence of these outer layers of the campaign structure, responsible for the ‘red’ and ‘black’ campaigning and for also circulating campaign messages through WhatsApp networks, distinguishes the Indonesia 2019 campaign from that of 2014. In 2014, the social media campaign took place on open social media platforms. In 2019, coordinated volunteer groups with loose (but not non-existent) ties to the official campaign worked more closely with WhatsApp networks to create and disseminate positive messages about Jokowi and attack messages and disinformation towards his opponent, Prabowo. Some of these volunteers saw attack messages and hoax messages as essential to influencing swing voters to back Jokowi, and therefore to Jokowi’s victory.

Among those documented, the case of the Jokowi social media campaign team is especially remarkable for the involvement of legacy media institutions in its establishment and capital and resource backing, as reported widely in the Indonesian press. As Souisa (2019) reports, MNC, ‘controls four television stations (RCTI, GTV, MNC TV and iNews), a radio station, Koran Sindo newspaper and online publication Okezone’, and publicly supported Jokowi’s campaign in 2019, and three of our interviewees (Maya, Maman and Zadrak) spoke separately of the important role played by one of MNC’s news directors, Kresna Abadi (not their real name) in both the founding of the online campaign team and the chain of command determining daily topics for campaigning. Maya was a seasoned Jokowi campaigner who had been involved in his successful run for Governor of Jakarta in 2012. In November 2018, they were approached by Kresna Abadi to recruit a team of 200 content creators for the campaign, which Maya recounted how MNC was heavily involved in the establishment of the online campaign team through provision of key resources. It provided the space and technological resources for the meme factory in Central Jakarta – a location deemed convenient as travel to and from it was well serviced by public transport. Maya explains,

The social media team was set up by Kresna Abadi. He contacted me to ask for my help in recruiting 200 content creators in three days. He told me that the creators would be accommodated in the High End Building within the MNC Tower complex . . . I was given a folder containing the contact details of 300 failed applications for positions at MNC, and told to start calling people listed on this file before I started recruiting from other networks. I recruited 60 people from a total of 100 and told them to come in the next day. On the same day Kresna went out and bought computers and hardware. ⁷ (Maya, Interview, 14 August 2019)

This fact highlights the interlinkages between political campaign teams and powerful media institutions – Rakhmani and Saraswati’s (2021: 3) ‘campaign industry’ phenomenon– which demonstrates the ‘inseparable’ links between economic and political interests since the deregulation of the media in the 1990s. What we see happening in the Jokowi campaign team is the big capital backing and direct supervision by legacy media institutions of the surreptitious production of content designed to sway voter opinion by masquerading as the voice of ‘ordinary netizens’. This distinguishes Jokowi’s mode of ‘astroturfing’ from the situation in Brazil and Nigeria where, as far as we know from available scholarship, legacy media was not involved in capital and resource backing for meme campaigns and their dissemination through WhatsApp.

Meme production: planned, supervised, orchestrated

According to our interviewees, memetic media were favoured as a form of political campaign messaging for a number of reasons. First, as a cultural form, memetic texts were seen to align with the cultures of use across a range of platforms targeted by Jokowi’s campaigners. Easy-to-share image memes are as at home on Facebook, Twitter and Instagram as they are on WhatsApp. Second, the humorous and visual character of memetic media were deemed to render them more attractive to millennials from a range of socio-economic categories.

Memetic media have been implicated in Indonesian election campaigns since at least the 2014 presidential elections. Wadipalapa (2015) argues that during the 2014 campaign, political discussion took place in open social media platforms (especially Twitter), and involved a meme culture arising from the development of meme-creating applications for Android phones, giving rise to a ‘meme culture that challenged media oligarchy’ (Wadipalapa, 2015: 10, see also Duile, 2020). But our research also shows that behind the façade of humour, spontaneity and lightness that memetic media present lies a production and circulation process that is highly orchestrated through categorisation, top-down planning of themes, close supervision of their production and strategic timing for their release, and sponsored by elements of the media oligarchy.

Our interview data revealed five topics under which the memes associated with Jokowi’s 2019 social media campaign generally fell: bolstering Jokowi’s Islamic credentials, trumpeting his achievements, undermining Prabowo’s Islamic credentials, questioning Prabowo’s masculinity and amplifying Prabowo’s poor human rights record. Memes trumpeting Jokowi’s achievements often depicted him as a heroic developmentalist and as a stable family man, while attack messages questioned Prabowo’s piety and his marital (divorced) status.

We interviewed Maman, director of Jokowi’s social media campaign, who told us that a core team of 12 people was responsible for determining thematic foci, which were updated on a daily basis. They say,

There is a set time frame, so at 12 midnight every day we come up with new hashtags, issues, counterclaims to amplify and the creators’ job is to produce various kinds of content to fit the themes. (Maman, Interview, 12 August 2019)

In addition to receiving instruction on thematic foci on a daily basis, the work of meme creation was closely supervised. Maya told us that five supervisors were based in the meme ‘factory’, and each was responsible for overseeing the work of 40 meme creators, and sometimes, they would help creators settle on suitable wordings for memes. Creators were expected to pump out 50 memes a day, and achieve record numbers of comments, likes and shares in a daily log. High engagement (the sum of likes, shares and comments) was rewarded with bonuses. Maya told us that ‘white’ meme creators’ job was to produce memes with a positive spin according to themes that were determined at the beginning of each shift, of which there were three: morning, noon and night.

Our interviews with meme factory creators show how daily topics were derived from occurrences on the campaign trail. One day, creators were asked to focus on Jokowi’s refusal to take a helicopter to highland town Sukabumi, because he preferred to travel by land so he could mingle and meet with ordinary people along the way. On another, they were to focus on the locally made bomber jacket Jokowi was wearing. The creators’ job was to transform photographs, or create or adopt illustrations, to create memes targeting young social media users. Pranoto and Sugih said about this event:

When Jokowi wore that bomber jacket he was criticised by Prabowo supporters, but there was a backlash from millennials – organic not orchestrated! – and many of them began to wear the same kind of jacket around the palace. (Pranoto, Interview, 16 August 2019)

Yeah many of the youth are apolitical but that bomber jacket debate made them loyal to Jokowi. (Sugih, Interview, 16 August 2019)

Creators also told us that as well as transforming photographs from the campaign trail, they also drew on images from popular culture such as Kpop or anime, and also created their own illustrations using Corel Draw and PixLab. Volunteer creators who circulated memetic texts on WhatsApp groups had a broader repertoire of memetic media to work with, because they were responsible for creating attack messages, as well as circulating ‘white’ campaign messages deriving from the meme factory. Zadrak, founder of the pro-Jokowi online support group, Teman Jokowi (Friends of Jokowi) told us that volunteer campaigners worked closely with the official campaign directors to ensure that the messaging was uniform. Zadrak explained,

On social media, the battle is being fought by the second, not by the day. In the mornings we would focus on Jokowi’s achievements, in the afternoons we would focus on attacking the opponent, and in the evenings we would focus on light themes that pointed towards Jokowi without mentioning him directly. For example, memes playing on the phrase ‘I only have love for one’ (Cintaku cuman satu) allude to Jokowi (number one on the ballot) without being overtly political, and this kind of meme targets young people (Zadrak, Interview, 10 August 2019).

Another source of memetic content was the Presidential debates. Creators were on the spot during debates looking out for moments with memetic potential, for example, gaffes by the opposing candidate that could be extended through image play. In her study of the use of memes in the 2012 US election, Tay shows how perceived gaffes by Mitt Romney were exploited for meme creation by ‘ordinary citizens’ critical of the candidate. The treatment of Romney, she argues, shows how ‘LOLitics employ gaffes as opportunities for ordinary citizens to further communicate political criticism by treating him as an open text for play’ (Tay, 2014, p. 51). In the case of Jokowi’s campaign, Prabowo’s gaffes may well have been exploited by ‘ordinary’ Jokowi supporters, but they also fed a sophisticated astroturfing strategy by the official campaign team. This became particularly apparent in the second televised debate on 17 February 2019, in which Jokowi asked Prabowo to detail any plans for supporting ‘unicorn’ (valued at between USD1 billion and 10 billion) startups, to which Prabowo replied, ‘What’s a unicorn? Those online things?’ (Damar, 2019).

The campaign’s Social Media director, Maman, related how their team exploited the gaffe, generating memetic media that specifically addressed young people as tech-savvy voters by rendering Prabowo as an object of ridicule. They said,

Each debate throws up material that we can use to generate memes. The unicorn gaffe is a good example of something that can be converted into campaign material that becomes viral, because it plays right into the concerns and interests of millennials and can be used to convince them to vote for Jokowi. We spent about a week developing pieces of writing and memes playing on the unicorn gaffe, and people were influenced by that. (Maman, Interview, 12 August 2019)

Indeed, after the debate, memetic texts quickly began to circulate associating Prabowo with horned equine creatures in pastel hues, and media outlets supportive of Jokowi perpetuated the myth of unicorn-related memes as a purely grassroots phenomenon by reporting it as an instance of ‘netizens’ critiques of Prabowo’s debate performance (Damar, 2019).

Dissemination of campaign memes through WhatsApp groups

Dissemination of memetic texts on WhatsApp groups was a task relegated to those on the outer layers of the campaign team, where ties to the official hierarchical structure were looser, and volunteers rather than paid creators played a bigger role in the creation and dissemination of particularly negative (‘red’ or ‘black’) memes. We interviewed three volunteers who were active in this area: Rudi, Malik and Zadrak. Zadrak oversaw campaigning by a network of Jokowi supporters which was responsible for disseminating official ‘white’ campaign messages and also engaging in ‘red’ (but not ‘black’) campaign messaging. Malik oversaw circulation of pro-Jokowi campaign messages on around 270 groups, each with around 100 members. Rudi, president of Cyber Projo, the social media department of pro-Jokowi mass organisation, Projo, oversaw the execution of a ‘black’ campaign that infiltrated public, pro-Prabowo WhatsApp groups and planted hoax messages aimed at turning Prabowo sympathisers away from him.

Zadrak told us that they supervised the circulation of pro-Jokowi campaign messages on around 178 WhatsApp groups, 150 of which were dedicated Teman Jokowi groups, and the rest were personal groups including an array of family, sporting, church and alumni groups. They said,

My priority is to spread the memes. Moderating content is not that important to me. I just spread the memes far and wide. (Zadrak, 2019)

Malik told us that their job was to design attack messages but not to share them (this task was relegated to group admins who decided on the timing of release of particular memes), and Rudi told us that they encouraged supporters to ‘just share’ the memes, and not to add comments.

All three campaigners told us that they did not closely monitor chat in the WhatsApp groups. However, this is not to suggest that they never intervened in the WhatsApp chat groups they used to circulate campaign messages. Zadrak told us that anti-Jokowi comments most commonly emerged in their personal WhatsApp groups (simply because they were not dedicated pro-Jokowi groups) and explained that they made sure to respond to such comments by posting a pro-Jokowi image meme, or an infographic containing statistics that shed positive light on Jokowi’s performance. Similarly, not only Rudi, but also Malik and social media campaign director Maman spoke of joining pro-Prabowo public WhatsApp groups and planting incriminating (mis)information about key Prabowo supporters, or fake information that made Prabowo’s campaign appear un-professional.

The extent to which ‘red’ and ‘black’ messaging played a deciding role in the election result is difficult to determine. What is notable is how the campaigners exploited the technical affordances of WhatsApp – namely bulk messaging, groups and the forward function– to skirt efforts to regulate the circulation of disinformation and hoaxes. Rudi told us that Projo used a bulk messaging service to blast messages through WhatsApp networks similar to what marketing agencies did to spread messages through WhatsApp groups during the 2019 Brazilian elections (Evangelista and Bruno, 2019).

Although WhatsApp limited the number of times a message could be forwarded by any one user to five following the spread of misinformation during the 2019 Indian election (Potkin and Damiana, 2019), our participants told us that limiting forwards did not affect very much their ability to exploit the platform for ‘red’ and ‘black’ campaigning. Zadrak told us they responded to the policy by distributing the task of forwarding campaign messages across different ‘users’ – meaning several WhatsApp accounts linked to distinct numbers but operated by a single person:

Maybe the policy limited our ability to spread messages a little, maybe it meant that we had to work a little harder, but it didn’t make that much of a difference. It just meant that instead of forwarding a message to twenty groups from one number, we distributed the twenty forwards across four different numbers. It didn’t really affect the campaign that much. (Zadrak, 2019)

Zadrak’s response shows how campaigners responded to the platform’s efforts to regulate the circulation of problematic content by exploiting platform affordances – in this case, the way WhatsApp identifies distinct phone numbers as distinct users. In fact, in Indonesia, many users have dual sim-enabled phones and the ‘black’ and ‘red’ campaigners we spoke to owned several such phones, allowing any one campaigner to behave on WhatsApp as if they were a number of different users. ⁸

WhatsApp not only afforded buzzers with protection from the regulatory impulses of the platform itself and the state, but it also enabled them to spread campaign messages (‘red’, ‘black’ and ‘white’) to new cohorts outside young middle-class voters residing in cities. As mentioned, smartphone penetration increased significantly between 2014 and 2019, promoting considerable expansion of both WhatsApp and Facebook’s user bases in the country. Importantly, the Jokowi campaign’s targeted demographic, millennials, dominate these user bases, making both platforms crucial sites for communicating with key sections of the voting populace. Our interviewees identified WhatsApp as especially important for communicating with a wide range of voters across rural–urban, lower class-middle class divides. Malik said,

Even low income people use WhatsApp now so using WhatsApp groups to campaign is a good way to expand the network (Interview, 7 February 2020).

Yet campaigners also identified other dimensions of WhatsApp that were conducive to the efficacy of campaign messages and their capacity to sway voters’ opinions of the candidates. Some of our interviewees pointed to the intimate qualities of private WhatsApp groups (which the ‘red’ campaigners largely relied on) resulting from the way admission to such groups rests on admins manually adding a number. Malik said: It adds a human element to it. You have to manually let someone in, and then you greet them. (Interview, 7 February 2020).

Malik also saw WhatsApp as a useful tool for monitoring the virality (and therefore, success) of memetic texts they created. They said: ‘If a meme I created returns to me through five of my WhatsApp groups as a result of forwards, I consider it viral’ (Interview, 7 February 2020).

Conclusion

In this article, we have examined the use of WhatsApp for circulating memetic media in Jokowi’s 2019 social media campaign to highlight the platform’s role in facilitating the elite manipulation of public discourse – a phenomenon commonly referred to as astroturfing. Our research reveals how paid operatives and volunteers with looser ties to Jokowi’s official campaign were responsible for circulating on WhatsApp both ‘white’ (positive) official campaign messages and ‘red’ (negative or attack), and in some cases, ‘black’ (false) messages that were deemed unofficial and yet remained part of an overall layered structure. These campaigners were responsible for establishing, building and maintaining dedicated WhatsApp Jokowi support groups and also for ensuring that political campaign messages were spread through non-political groups, such as family groups, ethnic groups, religious groups, school parent groups and sporting clubs that were part of campaigners’ everyday networks. Jokowi’s official campaign team leveraged these networks to spread campaign messages at a convenient distance from the official campaign, and conveniently hidden from regulatory authorities as a result of encryption.

Our study contributes to scholarship in several respects. The use of WhatsApp in election campaigning is growing, and we now have a picture showing patterns and variation in how campaign teams are structured to accommodate WhatsApp for political messaging, especially through the dissemination of messages through groups. The fact that communication on WhatsApp is end-to-end encrypted makes it impossible to observe it from the outside, hence, interviews and ethnographic methods become crucial to further elucidate how state or corporate players use WhatsApp to sway public opinion. However, it is often difficult for researchers to access campaigners and gain the trust necessary for them to be willing to speak on the record about the processes and practices implicated in WhatsApp-facilitated astroturfing. In this respect, our study stands out. We build on the emerging body of work attending to WhatsApp and election campaigning (Cheeseman et al., 2020; Dwyer et al., 2019; Evangelista and Bruno, 2019; Maweu, 2019; Murgia et al., 2019; Tapsell, 2018) by providing a detailed account of how WhatsApp is positioned within a broader structure in which campaigning relies heavily on the orchestrated production and strategically timed release of memetic media, and of how WhatsApp campaigning affords these messages efficacy and broad distribution.

Second, this article contributes to existing literature chronicling new developments in the production and circulation of memetic texts, leading to their co-option by political and corporate elites in their attempts to shape public discourse (Abidin, 2020; Evangelista and Bruno, 2019). Jokowi’s campaigners relied on memetic media to extend partisan messages to millennials who constitute a large proportion of the voting population, and image memes were deemed efficacious for their humorous and visual qualities, enabling them to appeal to urban and rural, middle and lower class voters alike. The irreverent humour inherent to meme culture makes these texts appear to be unofficial, hailing from the grassroots, but in fact, their production and circulation in Jokowi’s 2019 campaign was part of a highly orchestrated, well-funded, centralised and carefully planned process of persuasion.

Third, the study also contributes to existing scholarship on the evolution of social media election campaigning in Indonesia. This work has shown how social media campaigning has been co-opted by political and economic elites since around 2012, who have now developed well-financed systems for surreptitiously manipulating public discourse through strategic production and dissemination of content. What is missing from this work (with the exception of Lim, 2017) is considered attention to the technological infrastructures that make such manipulation possible. Lim’s study is the only work so far that discusses how platform affordances work to help shape the qualities of public discourse, and the introduction of WhatsApp in recent years as an indispensable campaign tool flags the urgent need to return to this important question.