Testimonial tweeting People ’ s voice ( and eyes ) on anti-impeachment protests in Brazil

Though the idea of horizontal networked, bottom-up communication processes is present in many discussions of the role of digital media in the context of social movements, not many studies point to the processes behind the creation and publication of authorial testimonial content by regular people, that is, content produced in situ at the heat of the moment whose author is neither media nor political actors. Through 10 interviews structured according to Proulx and colleagues’ ICT social appropriation framework (2017), with users that have created and published such type of content on Twitter with the hashtag #ForaTemer in the aftermath of Rousseff ’s impeachment in Brazil in 2016. This paper discusses why and how ordinary people take media matters in their own hands as a form of political action, even though it may apparently be dressed up with different communicative functions, such as emulating the pretense neutrality and objectivity of media discourse. Findings point to a convergent media critique as main motivator, followed by the aim to connect the absent and to scrutinize police abuse. The aggregated result seems to point to a manifestation of collective intelligence, characterized by swarm-like logics and alienation from the complete result of the #ForaTemer publications.


Testimonial tweeting
People's voice (and eyes) on antiimpeachment protests in Brazil Tuiteo testemunhal A voz (e os olhos) do povo em protestos anti-impeachment no Brasil

INTRODUCTION
Dilma Rousseff was Brazil's first female president.Elected in 2010, following two mandates of party colleague Lula and reelected in 2014 by an extremely narrow margin, she would complete four consecutive presidential elections' wins for Brazilian Worker's Party (PT), the strongest left-oriented political party in the country.On August 31th, 2016, after less than half of Rousseff 's second period, the economic struggle that had been affecting Brazil for a few years, along with a series of political scandals, paved the way to what Chomsky called a "soft coup" (Goodman, 2016) and Löwy (2016) called, referencing the military coup in the 1960's, the "farce" that follows the "tragedy": Rousseff was ousted by the national congress in the cusp of the impeachment process that had begun earlier that same year.The "parliamentary coup" (Albuquerque, 2017, p. 1) was not only testimony of the corruption of the legal and parliamentary systems in Brazil, but was also an evidence of the "captured" media system (Guerrero & Márquez-Ramírez, 2004), hostage to economic and/or political elites.For Albuquerque, it is even worse: a "vivid example" of cases "in which the free press seemingly conspires against the democratic order" (2017, p. 1).
Within such context, media coverage of events that push for the opposite direction -the maintenance of the president or, later on, protests against her successor Michel Temer-are expected to be under the influence of such economic and political elites and consequently should be compromised, leading either to biased coverage or no coverage at all -resulting in criminalization of the protests and/or invisibilization, respectively-.Nevertheless, the increasingly mediatized society implies a digital ubiquity (Ganesh & Stohl, 2013) that lead to a sort of omnipresence of mobile digital technologies in the hands of ordinary citizens (Chouliaraki, 2010), which enables the creation of alternative registries of important events.In such context, the value of alternative narratives as those created and/or circulated on social media is magnified, especially when accounting for such an important "out of the ordinary" event (Hall, Critcher, Jefferson, Clarke & Roberts, 1982) with national and international transcendence, since social crises as such are associated with both high uncertainty and high relevance, leading to a higher need for orientation (Weaver, 1980), pushing people to consult with alternative sources of information such as social media.
The debate over media bias on political sensitive matters has long been a theme of interest within the field of political communication and the Latin-American context is by no means an exception (Kucinski, 1998;De Moraes, 2009;Chaui, 2006;Herman & Chomski, 2010;Albuquerque, 2012Albuquerque, , 2017 to name a few).This research, though, goes deeper into empirical work to understand not only how ordinary people perceive media bias or its effects, but specifically how a part of the population deals with it through a process of ICT significant social appropriation (Proulx, Lecomte & Rueff, 2007) that leads to the creation and publication of self-generated visual or audiovisual testimonial content through digital social media, creating evidence of the events as a form of contradicting such biased accounts of mainstream media.
Through interviews with users that have created such type of content on Twitter with the hashtag #ForaTemer in the aftermath of Rousseff 's impeachment, this paper discusses why and how ordinary people take media matters in their own hands as a form of political action, even though it may apparently be dressed up with different communicative functions, such as emulating the pretense neutrality and objectivity of media discourse.

User Generated Content and Political Participation
Though the idea of horizontality of networks, bottom-up communication processes is present in the foundations of many discussions of the role of digital media -and in particular social media-in the context of social movements and street demonstrations (Bastos & Mercea, 2015;Condeza, Santos, Lizama & Vázquez, 2016;Juris, 2004;Santos & Condeza, 2017;Sey & Castells, 2004) not many studies point to the processes behind the creation and publication of authorial testimonial content by regular people -excluded activists, politicians, celebrities and organizations in general.On the other hand, there are some authors that research User-Generated Content (UGC) and political participation, from various perspectives.Dylko and McCluskey (2012) built the concept of Political UGC on top of Eveland's (2003) mix of attributes model and worked it empirically.Östman (2012) researched the causal relationship between UGC creation and political involvement, in this direction.Leung (2009) studied the psychological empowerment on the opposite direction of Östman: how the level of political participation works as a predictor of UGC creation.Building upon Bennet and Segeberg's (2012) connective action -a formulation that suggests that collective action, inspired in its traditional definition by Olson's rational theory of social mobilization, is transformed by connectivity-, Mortensen (2015) developed the concept of connective witnessing, that points to the crossroads between contemporary visual culture and personalized political participation through the creation of testimonial content.Within the field of journalism, there is abundant literature that studies how audiences engage with UGC (Wahl-Jorgensen, Williams & Wardle, 2010), how newsrooms deal with this source of information (Wardle & Williams, 2010;Williams, Wardle & Wahl-Jorgensen, 2011;Kammer, 2013;Wardle, Dubberley and Brown, 2014) or the opportunities professional media outlets create for "citizen journalism" (Thurman, 2008).
Still, none of the cited studies pursue the process of creation of content, as they focus either on the audience, on the content or on the media professional practices.This study sheds light in that direction as I interviewed users that have created content that account for the event they were witnessing to unveil their motives, their communicative practices and the consequences of their acts.

Testimonial Amateur Media
Unlike mediated testimonies, such as an ex-post interview of a witness to the TV News, authentic self-created testimonial media have two main roots that contribute to its relevance as a social phenomenon: its value as a "surrogate sense-organs of the absent" (Peters, 2001, p. 709), enabling those that cannot be present to participate remotely, and its indexical property (Santaella, 1983(Santaella, , 1992;;Peirce, 2003).Which means that the representation -such as the picture or video-is materially connected to the represented object in an unique manner, and this link confers it the perception of authenticity and objectivity.The aesthetics of professional media, though, through its 'proper distance' (Silverstone, 2004) seem to have neutralized the sensibility of the audience.Amateur images, instead, provide an "answer to the growing scriptization of life and the world" (Polydoro, 2016, p. 159;own traslation).
Therefore, the value of testimonial media created by ordinary users goes beyond the previous properties of meta-presence and indexicality.It relies on an "amateur aesthetics" (Polydoro, 2016) that occasionally even include the author's body in what Pantti (2013) calls embodied collectivity.For example, the very precariousness of a "grainy, blurry portrait" (Reading, 2009, p. 69) of the London Bombings (2005) conferred an image first circulated on a blog post, the title of the most iconic image of that tragic event: "It is poor quality and blurred and yet purports to truth and authenticity" (Reading, 2009, p. 70).Discussing the same event, Allan (2007) also underlines the authenticity brought up by citizen video clips made with personal cameras, which "were judged by some to be all the more compelling because they were dim, grainy and shaky, but more importantly, because they were documenting an angle to an event as it was actually happening" (p.13; own emphasis).

ICT Significant Social Appropriation
Thanks to social, technological, economic and institutional drivers (Wunsch-Vincent & Vickery, 2007), that lead to the growing penetration of digital media and digital social networks, the universe of potential content creators and publishers has exponentially expanded.However, not because the possibility exists, it crystalizes, as per Van Dijck's (2009) study with YouTube demonstrates: less than 5 in every 1.000 users are active content creators.The potential to act will depend of access, comprehension and critical adoption and application of the technologies at hand, which is clear in the author's statement: "'participation' does not equal 'active contribution' to UGC sites" (Van Dijck, 2009, p. 44).Besides, content is not spread out just because it is online: there are different types of contents, different motivations (Katz, Blumler & Gurevich, 1974) that attract users to adopt social media (Valenzuela, 2012) to generate and share content (Park, Kee & Valenzuela, 2009).
The process of adoption and use of ICT in daily life to a specific need is object of discussion.Plenty of studies have demonstrated that technological functionalities are appropriated by users in a variety of forms resulting in distinct communicative practices (DeSanctis & Poole, 1994;Isaac, Bresseyre Des Horts & Leclercq, 2006;Livingstone, 2008).Lindtner, Anderson and Dourish define appropriation as "the adaptation and transformation of information systems after they are deployed" (2012, p. 77).Harrison and Barthel (2009) posit that "the history of email has taught us that users may appropriate computer-mediated technologies and fashion them for their own purposes, which sometimes supersede or are at odds with the original purposes of designers" (p.157).
Reyes makes a detailed review of the theoretical models that approach from different perspectives the technology adoption, coming up with three main perspectives: Diffusion, Innovation and Appropriation (2015, p. 74).The Diffusion model, attributed to Everett Rogers (1986), is epistemologically positivist, since the user plays a more passive role (Isaac et al., 2006, p. 9).The model is centered in the idea of dissemination and considers innovation as a definite object, leading, according to Reyes (2015), to a permanently positive bias to innovation.The Innovation model is perceived as a more iterative process, aligned with the idea of appropriation by Lindtner and colleagues (2012) previously mentioned.Within this paradigm, "the concept [of innovation] refers to a whole process, that is, precisely the work of social actors mobilized by the objects conception" (Reyes, 2015, p. 79), supported by Bruno Latour's Network-Actor-Theory (NAT).According to NAT, the subject "is not source but a moving target of a wide set of identities that swarm in his direction" (Latour, 2012, as cited by Di Felice, 2013, p. 64, own translation).Lemos (2015) defends NAT by dissociating it from technological determinism, sustaining that "all technical device must be recognized not as an individuality, closed, finished and autonomously acting upon others, but as a monad, as a network-actor that acts and is acted upon depending on the formed associations (always irreducible ones to the others)" (Lemos, 2015, p. 31).
Though, Reyes defends the adoption of the theoretical framework of appropriation, indicating that "the significance of the use people grant to technologies is the most verisimilar answer" when a researcher tries to understand "why people integrate a technology in their daily life" (2015, p. 81).As we'll see, the ICT Social Appropriation framework by Proulx and colleagues identifies different levels of such appropriation, allowing for the distinction of various communicative practices.Much of the research that adopts Appropriation as a framework, though, is searching for answers in the first two of those levels of social appropriation (technology adoption and technology cognitive skills development).As we'll see, this research focus on the third level, which is when people make technologies useful for their own purposes, in this case, with political intent.Proulx and colleagues (2007) develop their framework dividing the process of ICT social appropriation into six moments: (i) access; (ii) cognitive skills acquisition; (iii) significant adoption, where old practices are replaced by new ones that incorporate technologies; (iv) creative use, that divert from original predicted uses for such technology; (v) collective intelligence phenomenon that rise from the incorporation of the previous steps in a community or a group; and (vi) the complete realization of the appropriation process that culminates in the active engagement of the user or group of users in processes of technology innovation or development of public policies related to such technology.
For example, the hashtag adoption in Twitter as a means for organizing information, aggregating content thanks to the unforeseen effect the # symbol had in the system -being automatically converted into a hypertext that leads to a search with all of the hashtagged content-is an example of a (iv) divergent use of technology at hand, unpredicted by the developers; the programmers of a Linux community or Wikipedia collaborators would be in the level (v) and people who currently advocate, for instance, in favor or against Net Neutrality, free or open source software, WikiLeaks creators etc. could be examples of communities pursuing their own interests or the ideal for public policies regarding their métier, an example of level (vi).However, level (iii) is the one that applies to the heart of this research: significant appropriation of ICT, in which regular users, not necessarily with high levels of technical skills, understand or interpret possibilities and potentials of the technologies and give them an intentional, propositional use, articulated as a communicative practice with, amongst others, political purposes.
Both facets of UGC -creation and diffusion-reflect modes of social and significant ICT appropriation to a higher or lower degree.And within that universe of possibilities, one of the most fascinating, from a political communicational perspective, is its appropriation during political crisis that lead to produce testimonial content, which we'll dig into.

METHOD
The procedure included computer-assisted and manual data processing.The former includes (i) data selection and capture, (ii) data screening and (iii) automatic data coding; whilst the latter includes (iv) manual data coding and (v) in-depth interviews with selected individuals.Though it is difficult to dissociate completely one step from another, this paper will highlight the findings ENERO -JUNIO 2018 founded mainly on the last step of the research: ten in-depth semi-structured interviews with users that were identified through the previous steps as creators of testimonial visual content, who were contacted individually and interviewed by the author.

Case
As previously mentioned, the case selected was the aftermath of Rousseff 's impeachment, specifically two street demonstrations that took place, with different characteristics: Table 1 Two street demonstrations that were analyzed to select the users that created testimonial content on Twitter.Source: Author.
The choice of events with different characteristics is justified since the population that attends to a street protest on a Wednesday evening (Day 1), organized impromptu, on the summit of the political crisis is not entirely the same as people that attend to a planned, family, Sunday demonstration with daylight (Day 2).Also, the second protest happened predominantly in São Paulo with some repercussion in Rio, Salvador and a few other capitals.This makes the analysis more contextualized and the conversation with interviewees is supported by information about that specific event in that specific place, helping get in depth with the informants.

Data
The complete dataset 1 -including messages published with the hashtag #ForaTemer in both days-contained 115,701 tweets.Those tweets were screened with data analytics software -Tableau and Excel-using metadata attributes to find originally developed content.This procedure was important for the idea was to find the users who produced themselves the content, so any form ENERO -JUNIO 2018 of retweet was excluded from the final dataset.Also, due to the understanding that the power of testimonial media resides on the indexicality of the imagery (moving or still), that relates immediately to the object represented (Santaella, 1983(Santaella, , 1992;;Peirce, 2003), all text-only messages were also excluded.The final dataset was composed of around 4,000 tweets that were visualized individually on the native platform www.twitter.comto have an appreciation closer to what a regular user would have when accessing such content.The tweets were then coded, amongst other criteria, as testimonial or not testimonial and users coded as ordinary or not ordinary.Testimonial tweets had to comply with three criteria: (i) original content that (ii) referred to the event (iii) in an indexical way (photos, videos, Animated GIF, location apps and others).Ordinary users were considered as such regarding political participation, meaning all those that in the context of the protest that were acting as "ordinary citizens" (Chouliaraki, 2010) -even journalists if they were publishing in their own personal channel.Politicians, activists, Political Associations, Media Twitter users were, therefore, considered not ordinary.
The final result was, 1.223 tweets coded as testimonial and analyzed with more depth.From those, 886 (72%) were created by ordinary users.Within such final sample, users were selected to perform the interviews, according to the criteria to be discussed in the following section.One interview was also conducted with an alternative media outlet (Mídia NINJA) because of the (confirmed) suspicious that it accepted and even encouraged ordinary users to send them their content.

Interviews
From the final sample of ordinary users that created testimonial content, then, ten users were selected to proceed with in-depth interviews.The main criterion for the selection of such interviewees was the diversity within the dataset of testimonial tweets.The objective was to approach both heavy and light users (as per their status updates by the time of the data capture); media professionals and non-professionals; users that tweeted a lot during the protests but also someone that just tweeted once; currently active and inactive users; even one non-Twitter user whose content ended up on Twitter was interviewed; users with different tweeting patterns -for example with different narrative functions such as to communicate facts or to mobilize, rallying people to the demonstration-.Though this strategy might allegedly lead to epistemological obstacles to the generalization of the findings, it proved, as we'll see, to be vastly informative, and on the other hand important patterns that emerged despite such a diverse group are strengthened for the same reasons.
All interviewees were asked to read, before the interview, an informed consent with the characteristics of the project, the expectations, their rights and so on; which they signed authorizing the use of the information.The interviews ENERO -JUNIO 2018 were conducted through various means -such as FaceTime or audio calls on WhatsApp for example, as per the availability or preference of the interviewees-and they followed a prebuilt structure based on Proulx and colleagues' (2007) ICT Social Appropriation levels.
The social profiles of the users with their names, professions, age group, and characteristics that are of interest to the research, are as follows, listed by alphabetical Twitter handle: Andrea (pseudonym): Andrea is in her 30's, who makes a living as a selfemployed filmmaker and photographer of events, in special marriages (in association with her life partner).She is licensed but not graduated in sociology and, despite not having many followers, she is very active on Twitter and such activity is predominantly political.This characteristic turned her into an interesting source for she had tweets with up to 22 retweets despite only having a bit more than 100 followers at the time.Thanks to the adoption of the hashtag she was retweeted by Spanish speaking users and that boosted this specific tweet.Also, she is a heavy user as a consumer, for she had a very large amount of likes (close to 12,000 at the time of the research, while the average for the others was 1,600, ú refer to Table 2).These characteristics and her use of Twitter were the main interests behind the interview.
Beatriz Leandro (@biaaleandro): Professional of International Relations, currently fully dedicated to her MSc studies in Sociology, in her late 30's.At the time of the protests, she worked with São Paulo's City Government.Has one son -who attended to a few protests.She was selected for one main reason: though she had left Twitter (her handle now belongs to another person), I could locate and interview her.The relevance was to understand why someone with political activity on Twitter decided to leave the platform and assess if it was somehow related to the protests and specifically with the testimonial content created by her.
Maria Claudia Branco (@CacauCB): Claudia is in her early 30's and though she is a journalist, her use of Twitter is as a very elaborate character, with a particular acid, humorous, political language.Probably due to that Claudia displays a good performance, for her tweets got some reaction from the audience.With only about 1,600 followers at the time of the protest, she had one tweet that received 90 RT and 104 likes and her other tweets also had high numbers with sophisticated and original choice of words and images.She holds a MSc in journalism and works as an intelligence analyst for the presidency of the country in a Strategic Communication and News Agency.Her name is distorted and just her friends know she's the one talking on Twitter.There were multiple criteria that Claudia met, but the most impor- tant for choosing her is that a picture detected as UGC was actually from a friend of hers that had published elsewhere -this was the successful tweet just mentioned.The friend (Hussein) was also interviewed to understand his options and how the process was from the other perspective.
David Gormezano (@dgormezano): David is a French journalist, works for the Spanish version of France 24, a French 24 hours news channel financed mainly by the government.David lived in Brazil because his wife had a job opportunity and though he was not working professionally when the political crisis took place, he followed the protests as an interested politicized citizen.He was chosen not only for being a privileged international observer, but also for being a journalist.It was even more interesting to find out he was not working at that moment, fitting well the definition of ordinary citizen, despite his professional communication background.
Leonardo Silva (@leoborjinha): Leo is a 19-year-old second-year advertising student in Uinpampa (state-funded public University), in small town São Borja in Southern Brazil with little more than 60,000 habitants.His profile was chosen for he seemed to be an ordinary user generating only content about the repression.After the interview, it ended up even more interesting for he was not present, he was retweeting from his hometown, but due to technical knowledge -he studies advertising and works with digital media and social network marketing-and his expressiveness it seemed that he was living what he was commenting.The interview was then redirected as someone that was not present but was performing mediated witnessing -creating his content by capturing media content, such as pictures from TV-with a clear pattern.He is also very active on Twitter: despite being so young and having joined Twitter in November 2013, he had almost 13.000 status updates at the time of the protests, averaging 13 messages a day.
Luiza Geiling Cruz (@LuizaGeiling): Luiza is a 19-year-old literature student, that works at an NGO in São Paulo.She was chosen for her metadata on the Twitter profile: she joined Twitter exactly in September 1th, 2016, (the day after Day 1) right after the impeachment and she was quite active for a very short period, less than a week.Besides, she tweeted she would be monitoring for police brutality before the second protest (more on this on the Findings and Discussion section).Her experience turned out to be very important to the research for she had the most visceral experience, having faced high levels of police brutality from a short distance and being herself hurt during the first protest.The consequences in her process of ICT appropriation will be discussed further on.

ENERO -JUNIO 2018
Mídia NINJA (@MidiaNINJA): Interview conducted with Social Media manager Driade Aguiar for this alternative politicized (even partisan, according to some sources) media.She personally has no formal education and has been in charge of social media for them for a few years, since the group started (circa 2011).Driade was 27 years old by the time of the interview.Besides what became quickly obvious after analyzing the data (that Mídia NINJA is the most relevant content creator within the datasets from a quantitative perspective, both as a broadcaster and as an influencer), this alternative media organization is also known to be decentralized, so I assumed it could be an attractor of UGC -as was confirmed by the interview-.So, looking from the perspective of an alternative, politicized, decentralized, young media organization promised to be very interesting, especially since there is so much work on big media's use of UGC (as per literature review on previous section), and not so much dedicated to this kind of organization.
Roberta Prescott (@RobertaPrescott): Journalist, specialized in IT and Telecom, at the moment of this research she was at her late 30's working as freelance to national and international media.One of the main reasons for interviewing her is that she was the only interviewee that produced testimonial content on both protests; others like Claudia also tweeted but not her own content, for she did not attend to the first protest; or participated on both events but did not create content for Twitter on the first such as Luiza.Besides, she is the ordinary user with more testimonial tweets (30) the analyzed patterns of her content were very enlightening for she moved from a more distant, journalistic-like content to more engaged as the repression took place.One of her tweets had very good engagement with 28RT and 28 likes though she had just 1,001 followers at the time.
Marcelo Costa (@screamyell): Publicist, works as a cultural journalist.Self-defines as a "conteudist" (as in "content-creator").Super active on Twitter with almost 180.000status updates by the time of the protests (while the others' average is below 7,000, refer to Table 2), he is about 40 years old and runs a rock fanzine over the internet that has been active for over 20 years.Demonstrated excellent Twitter and Social Media skills and knowledge.His profile mixes personal (such as personal and political opinions and even his hobby beer tasting) and professional (his fanzine information that names the handle).
Hussein (no Twitter handle): Law student, Hussein Cavalcante was (and is) not active on Twitter, he is much more active on Facebook (for political matters) and Instagram (personal).He is on his early 20's and lives nearby ENERO -JUNIO 2018 the typical place of gathering for protests in Sao Paulo (Paulista Avenue), therefore he participated in plenty of them.He was chosen specifically as someone from outside Twitter that was included in such network by one of those unmapped organic flows of social media: he published on Facebook2 and his friend Claudia (other interviewee) liked the picture and 'curated' it, publishing it on Twitter.This is not uncommon, and another interviewee, Beatriz, also published a video received from a friend, as per her memory, via WhatsApp.But as the source, the process also varies, as Beatriz gave complete credit ("Video: Tania Bustamante") but did not ask for authorization; while Claudia did ask, but wrote "I just received this emblematic image" with no attribution.
Table 2: Quantitative profile of the users as to the number of testimonials (per each day of the protests), number of followers, followees (Friends), Favorites and Status updates, alphabetically ordered by Twitter handle.Source: Author.

FINDINGS AND DISCUSSION
The interviews point to a few convergences that reinforce the literature and others that add interesting points to be investigated further.The analysis will be presented following the same first five levels of Proulx and colleagues (2007) ICT's Social Appropriation model3 that structured the interviews, which are: 1) Access; 2) Cognitive and Technical Mastery; 3) Significant Use of Technology; 4) Creative Use; and 5) Collective Adoption.Additionally, there is a section assessing the perception of personal impact users had due to their actions.

Level 1: Access
Though ten interviews are not by any means representative, it is remarkable that all of the interviewees that were ordinary citizens (this excludes Driade Aguiar from Mídia NINJA) had a degree or were studying for it (sociology, law, journalism, advertising, international relations, literature).All of them had internet mobile access, though with different levels of quality and quantity: while literature student Luiza waited to get home to publish her content, Roberta published a lot of content, including videos, in real-time, as they were created.Twitter heavy-user Marcelo had knowledge of hot-spots on the route, which he used to save some data from his personal plan.
Other variations occurred, as it would be expected.The quality of the imagery, for example, which depends directly of the device, varied a lot, especially with the videos.As to the liveness of the imagery, some published live and others when got back home, for different reasons, but only Luiza said explicitly she was worried about her data plan.Beatriz said she preferred to enjoy the event and publish media afterwards.Coincidently they were both occasional users on Twitter, and currently neither is active in that social network.This suggests that, for regular users of Twitter, liveness matters.Roberta (Personal Communication, October 17th, 2017) and Leo (Personal Communication, February 16th, 2018) defined Twitter as "real time" and David as "live" media (Personal Communication, October 31st, 2017), something that converts it into a live storytelling tool for this kind of event.

Level 2: Cognitive and Technical Mastery
The preparation of the equipment is a reflection of the level of spontaneity of the creation of testimonial content, which is important in the context of this research that has as its object exactly testimonial content that was produced as effect of the opportunity, not of the voluntary pursue of such content.Of course there are different levels of preparation specially as a function of the type of user (an activist, a politician, a journalist and a user with no formal communication skills are expected to be different in that sense), but all of the interviewees shared a lower common denominator: part of the ritual of attending to a protest is getting the mobile phone fully charged and usually also assuring there is enough disk space, regardless of to what measure the user anticipates any sort of documentation of the protest.From that point on, there are differences: Roberta, journalist, took two devices in order to preserve her personal device, while using an older one to document the event, due to fear of being robbed amid the crowd; Marcelo, who works as a cultural journalist -though has no formal studies in journalism-included in his preparation a spare battery to charge the phone during the event, so that he could make more registries; Andrea, ordinary user, said though she wasn't planning to make images, it is part of a digital culture and that it is "a thing of our time" (Personal Communication, November 14th, 2017), as she regularly registers her life moments with pictures or videos.
Regarding digital skills, while the three journalists and the cultural producer had a clear understanding of social media and Twitter functionalities, ordinary users made a more layman use of the platform, exploiting it as a regular user.This reinforces the perception of Twitter as a biased tool that attracts niche users, such as journalists, politicians, opinion leaders and so on.Still, all interviewees made a conscious use of the hashtag as a strategy to join the larger conversation.The adoption of the hashtag was in no case a fortuity or a "sheep-shepherd" situation, as all users declared they wanted something in the lines of "reach more people" (Claudia, Personal Communication, October 19th, 2017) than in their followee network and to enhance the visibility of the hashtag as a political statement.Many were familiar even with the trending topics dynamics on Twitter.Marcelo mentioned a sense of "community" formed around the hashtag and stated that #ForaTemer was simultaneously a symbol of revolt with the political coup and an "aggregator" (Personal Communication, October 25th, 2017).David, who also works as journalist for French media (off duty at that time), had a very clear picture of this multifunctionality of the hashtag, pointing three functions: (i) a statement, (ii) a visibility enhancer and (iii) a bridge to the European audience (Personal Communication, October 31st, 2017).Beatriz, a sociology Masters' student, admits having little digital skills, but her adoption of the hashtag was equally conscious: "to be found", she said (Personal Communication, November 13th, 2017).
The use of the hashtag as a source of information was the other face of the same phenomenon, since the interviewees admitted using it to follow other protests that they were not able to participate: "It's a research filter", defined Andrea (Personal Communication, November 14th, 2017).Leo said in order to "have access to views different than mine (…) I just go ahead and put #Fo-raTemer and I can see the many interpretations [of an issue] on this medium [Twitter]" (personal communication, February 16th, 2018).
In terms of audiovisual background, the levels varied a lot, so it does not seem that users deem necessary such background to create their audiovisual content on a platform such as Twitter on an event such as a protest, they look for other values they perceive themselves when consuming the same kind of content: proximity (friend, friend of friend as author), real time, lack of commercial interest or political bias are the ones that showed up during the interviews.

Level 3: Significant Use of Technology
There are many different ways that users may exert their power as content creators in significant ways in the context of a street demonstration.This research identified three main manifestations of it: (i) mediactivism, as in providing alternative narratives for the event as a reaction against media bias; (ii) ENERO -JUNIO 2018 political action, as different types of content created by the interviewees held some degree of political intent; and (iii) networking, which means intentionally connecting or being connected to the protest via the adoption or consultation of the main aggregator related to the events (#ForaTemer).
Luiza went to the first protest (Day 1), and uploaded no pictures.But what she lived left her enough scars to change that attitude: a few meters from her a sound bomb ("moral effect bomb" is the official euphemism in Brazil) was launched and a piece of metal left a young 19 year-old girl blind of the left eye.In the heat of the repression she got hurt trying to escape to safety."I was really mad with all that had happened the 31 [August 31st protest] (…) I got hurt, they've hurt people close to me, then on the fourth [September 4] I went thinking about it" (Luiza, Personal Communication, December 11th, 2017).Luiza is mentioning the following tweet: Luiza's story may not be the general case, but is indeed a great motivational factor for people to be ready to document cases of police brutality during social protests in Brazil.This personal/witnessing experience is reinforced by the perception of a very biased media that tries to silence or criminalize the social movement and in particular the protests against Dilma.In Luiza's words, the "totally dishonest" traditional media "was part of the coup" and "without it, the coup might not had happened" (Luiza, Personal Communication, December 11th, 2017).She says that they clearly took sides and there they stood all along the process.This perception is shared by all the interviewees, no exceptions here.The difference lies in the way each one uses alternative sources of information and with what degree of confidence they treat it, such as the reference of French journalist David, regarding alternative media such as Mídia NINJA or Jornalistas Livres: he stated that this kind of independent content, as well as UGC in general, "should be consumed with extreme caution" (David, Personal Communication, October 31st, 2017).He states that Globo's4 coverage of the protests was "Unbelievable", as in a negative perception of an intentional invisibilization tactic by the largest media conglomerate in South America.To illustrate, he recalls they were talking about "stuff like Fashion Week" instead of the massive protests he had attended to.Leo, despite being more conservative on his political views as per the interview, agrees with David: "I don't give any credit to what TV says.Specially Globo" (personal conversation, Personal Communication, October 31st, 2017).Hussein (Personal Communication, November 14th, 2017) defends testimonial content precisely as a reaction to biased media and insists that who had been on the protests "knew it was not what the media reported".He had also been testimony to police violence: "I knew that [testimonial UGC] was true because I had lived"."I don't trust media", Hussein blatantly asserts.
In line with mainstream media distrust, emerge not only the professionalized "mediactivism" (Driade, Personal Communication, September 22nd, 2017) that characterizes collective groups of alternative media such as Mídia NINJA and Jornalistas Livres, but also the "sort of activism [where] you are media yourself (…) helping connect people who are not there" (Andrea, personal communication, November 14th, 2017).This connective function, related to the idea of a networked ecosystem of friends in social media, put users in a double role: informers and informed.As Marcelo explained, he was out of the country for another round of protests of 2013 -that became known as Jornadas de Junho (or "June Rounds" in a free translation)-and he was very pleased to be able to follow closely thanks to testimonial content available on social media (particularly on Twitter via #ForaTemer), as if participating from distance.Due to that experience, he felt the "duty" to do the same for others.
Hussein (Personal Communication, November 14 th , 2017) said he published images during the gathering stages in the Facebook pages that created the "event", as a form of incentive for people that are sitting back home to encourage them to get up and join the protest.This was not a practice described by the others, but confronting the media was, as previously discussed, a form of activism.Beatriz (Personal Communication, November 13th, 2017) used social media (Twitter at the time, Facebook currently) to meticulously persuade acquaintances or even friends who had a different perception as hers, or at least, in her words, "sensitize", as in raise awareness.
Hussein's case of content that call for action might be the most explicitly political use of testimonial content on Twitter to mobilize constituents to the street demonstration, but other forms of content, despite not being as explicitly political, had political implications and, according to the users' statements, had political motivations as well.Journalists such as David and Roberta did adopt much more neutral language in their tweets, as usual in the news making business, but their intent was to counter-inform, in an attempt to disrupt media bias, both national and internationally.This observation is intrinsically related to the generalized perception that mainstream media was either concealing the protests or misinforming about them, such as through emphasizing violence or downsizing the attendance.Even selfies, a relatively frequent way of self-ex-ENERO -JUNIO 2018 pression during the protests studied, may be interpreted as an expression of the collective identity that help users construct a desirable self (Polletta & Jasper, 2001), beyond narcissistic interpretations of personal exhibitionism on social media.
Networked logics seem to have worked in two ways for testimonial content: connecting the absent and building trust.The former means that, as previously mentioned, through testimonial content users connected people that were unable to participate in the protest: Marcelo, Andrea, David, Claudia and Beatriz gave similar accounts that lead to the idea of network, connecting those that are not in person, many times reciprocally as they themselves were connected through the media produced by others.The latter means that users trusted testimonial content, not only due to aspects related to the amateur aesthetics (Polydoro, 2016;Pantti, 2013), but also because of the direct connection with events, if not by personal experience, through the connection with the author of a testimonial content, such as what happened with Hussein and Claudia: while the former published on Facebook an image, Claudia, her friend, asked for permission and republished on her Twitter.Evidently, the direct connection with the author of the pictures assures her its authenticity.Along the same lines, Leo stated that "if the person that took the picture and published in her own [social media] user, then I'm cool with it" (Personal conversation, February 16th, 2018), meaning the content under such procedure, is trustable.
It is important to note that, though all interviewees considered themselves to some measure to the left of the political spectrum, the interviews corroborate data from previous studies of attendance to the demonstrations (Ortellado, Solano, Moretto, 2016) that defend a diversity of political orientations: "I'm not petist5 , more like PSOL6 " (Beatriz, personal conversation, November 13th, 2017); "I think both Lula7 and Aécio8 should go to jail" (Marcelo, personal conversation, October 25th, 2017); Andrea voted for PT, would vote again, but clears out that she's "not necessarily petist nor anti-petist" (Personal conversation, November 14th, 2017).Mídia NINJA stands out, for it seems to be more aligned with PT lately, as per media information and interview with David and Leo, but not necessarily at the time of the events.Claudia also stated that she became not only a petist, but a dilmist9 , as a result of the polarization of the political crisis in Brazil last years.

Level 4: Creative Use
Again, Mídia NINJA stands out for a very creative solution to a complicated problem: how to, on a tight budget, spread its wings all over the country to cover mainstream and marginal stories with citizen help and still maintain the aesthetic quality and the political commitment predicated by a mediactivist collective such as they are, and that defines and identifies them on social media?Driade explains what she calls a "political-aesthetical didactics" as a set of adhoc real-time orientations to eventual collaborators that send her content via Social Media, so that they are able to capture better content that follow Mídia NINJA's aesthetical-political standards: "Don't you have a frontal picture?", or instructions such as the inclusion of a monument that would serve as an index to identify the territory (city, square etc.); wider or narrower frame according to the sent picture, holding up the cell phone horizontally instead of vertically and so on (Driade, personal communication, September 22, 2017).

Level 5: Collective Adoption
Considered as a whole, the group of users, each one in her role as ordinary individual that adopted Twitter as their platform to create and publish content documenting the protests, could be considered a loose, weak-tied collective, linked by a few factors: use of Twitter; critique of the mainstream media coverage; lack of professional image-making skills; and, above all, usage of personal resources and channels to create an "alternative narrative" (Marcelo, Personal Communication, October 25th, 2017).In a social appropriation approach, this may be interpreted as a manifest form of collective intelligence.It is not the same as a completely conscious community of software developers around Linux, for, in the case of an "ad-hoc publics" (Bruns & Burgess, 2011) formed around a hashtag on Twitter, each individual user has less control and understanding of the overall 'product' or result of their efforts, which can be interpreted as a sort of alienation.Nevertheless, the commonalities amongst users, their motivations, their political-mediatic aspirations, point to a swarm-like perception of contributing with a grain of sand to the alternative narrative of the event, contrasting with mainstream media.Marcelo states that protests were "solemnly ignored by mainstream media" (Personal Communication, October 25th, 2017); French journalist David said that even international media were giving "too much space to the opposition " (Personal Communication, October 31st, 2017); Roberta wanted to "echo" with alternative sources of information about the events that informed beyond the violent or repressive more graphic episodes (Personal Communication, October 17th, 2017); Claudia wanted to contradict mainstream media's discourse that relied solely on dualities such as violent-pacific protest or unpopular-massive attendance and at the same time pursued evidences of support for Rousseff, more than opponents to the impeachment.

ENERO -JUNIO 2018
Another phenomenon, adjacent to the concerns of this research and linked to the previous level of Social Appropriation, is the sophisticated forms that Mídia NINJA employs to manage their collaborators so that they can cover as much as they do, with a little budget and a very committed team.Driade (Personal communication, September 22nd, 2017) mentions three types of collaborators to Mídia NINJA during events and also on daily activities that they manage through 84 Telegram10 groups divided by geography, topic, activity and others.The groups were described as: "Livers": those who live in the organization's housing facilities in different cities in Brazil.These are very committed collaborators that gather to plan ahead the coverage of an event, distribute assignments and so on.They also develop the agenda to be covered and participate in domestic duties in the houses.
Collaborators or Mediactivists: people who frequently collaborate but don't live in the houses, they come and go, but may eventually participate on planning activities and have a strong background with the politicalaesthetical principles of the collective.
Eventual or "Floaters": those that show up ad hoc, for instance sending content on Facebook's inbox or offering some other kind of assistance, such as translations, subtitles etc. Mídia NINJA welcomes their content, though it is a process of trial and error with live instructions, as previously described as "politic-aesthetic didactics".This kind of agent may be very relevant as it works both as a connector -due to their large networks, of followers in the case of Twitter-and as an intensifier of the homophilic networks.Alternative politicized media such as Mídia NIN-JA and Jornalistas Livres, Fake politicized characters such as Dilma Bolada, Muda Mais and so on, turned out to be extremely relevant in the dataset, not only in the particular testimonial content sub-datasets, but also in the whole data.They attract testimonial content that otherwise would probably turn out in ordinary users' personal channels or not turn out at all, for mainstream media were proven, at least as per the interviews, to be very biased in favor of the impeachment and against the protests.

PERSONAL IMPACT
Interviewees seem to have a general perception of the creation of testimonial content as a form of "civic duty" (Beatriz, personal communication, November 13th, 2017) that brings them satisfaction as it was clearly perceived as a form of political participation.There were no accounts of important negative ENERO -JUNIO 2018 effects such as conflicts with peers, relatives or at work; the audience's reaction was positive and the act of creating testimonial content has had no clear effect on the interviewees' lives.
The notorious dissatisfaction with the role mainstream media has played during the political crisis that resulted in Rousseff 's impeachment has an important impact on this particular communicative practice.Motivations depart from feelings of injustice and conviction of the right thing to do, such as: when Claudia states she's "defending something she believes" (Personal communication, October 19th, 2017); when Roberta talks about the sensation of "job done" or "fulfilled duty" (Personal Communication, October 17th, 2017); similarly to Marcelo, who mentions that the creation of testimonial content felt "almost like my duty" (Personal Communication, October 25th, 2017).
As to the effect, all interviewees concur that, when not null, the effect was very limited, and things followed its course despite their effort.French journalist David reported that his accounts of the political crisis though social media, amid the political crisis in Brazil, in his condition of foreign journalist and specialist in Latin America politics, was enough not only to open the eyes of his "close circles".He also intended (successfully, according to his perception) to challenge superficial international reports over the crises that were giving too much attention to the political opposition (those against Rousseff, or pro-impeachment) during the crisis.Such bias towards opposition views, according to him, imprinted international media those days, not only in Europe, as he mentions New York Times for example (Personal communication, October 31st, 2017).
Predominantly, though, the reports were of frustration with the developments, since in concrete, the demonstrations did not revert the political crisis and not only the impeachment followed through, but demonstrations were not capable of pushing for an alternative course, such as direct elections.
Finally, the perception of 'preaching for the converted' appears in Andrea's and Marcelo's metaphor of a "Bubble" or a "Big Bubble" to describe their network of followers on Twitter, where all people think alike.Somewhat like Hussein's perception translated as "Internet bubble" where "who is against get out or quite down" in reference to political discussions on the digital social networks.Such perception led users to even disengage from Twitter at all such as Luiza and Beatriz which can be interpreted as a concrete manifestation of the frustration with it as a tool for political deliberation and debate.

CONCLUSION
This research adds evidence to a political significant appropriation of ICT as a means to fight back media bias against social movements in Brazil, not only by organized groups -such as notorious mediactivists of Mídia NINJA-but by ordinary citizenry as well.Though the phenomenon is circumscribed, in this research, to a topic restricted semibounded population (Rafail, 2017) on Twitter -therefore raising questions of representativeness from both from the platform selection as from the data filtering criterion-and ten interviews is hardly an attempt to generalize, there is definitely a communicative practice associated with digital media that works as a bottom-up content generation.Within the context of Rousseff 's impeachment, as per the interviews conducted for this research, it was mainly triggered by media bias, but also by intent to collaborate, participate and invigilate the actions of public forces.
Media and technology are what society makes of it.The relationship between media and democracy is far from an intrinsic one.In capitalist societies, media in general is subjected to its logic and in less professionalized media industries, media may be subject to other types of forces, such as political and economic interest groups (Guerrero & Márquez-Ramírez, 2004;Albuquerque, 2012).In that context, it is important that people, constituents, partisans make use of their own channels as a means to provide alternative discourses.The challenges, of course, are many, as behind the discourse of neutrality of so-called platforms lie "the tensions inherent in their service" (Gillespie, 2010, p. 348) for the need for monetization of these commercial platforms lead them to suppress their (alleged) participatory spirit (Van Dijck, 2013), urging users themselves to reclaim it.
shoot and post any form of police repression on today's protest", citing the state police (Source: Luiza Gelling Cruz, 4 septiembre, 2016; own translation).