Journalist Patrícia Campos Mello on WhatsApp and the 2018 Brazil Elections

Patrícia Campos Mello. By Lina Ibáñez –, CC BY 2.0

A blog post by Visiting Graduate Student Gabriel Pereira and Research Assistant Iago Bojczuk.

Last year, in our blog post “Zap Zap, Who’s There? WhatsApp and the Spread of Fake News During the 2018 Elections in Brazil,” we discussed how WhatsApp messaging was influencing the Brazilian elections amidst reports of misinformation and ‘bulk messaging.’ We have since continued researching the ways in which WhatsApp has been used in Brazil in recent years for an upcoming research article (co-authored by Gabriel Pereira, Iago Bojczuk, and Prof. Lisa Parks). We interviewed the Brazilian journalist Patrícia Campos Mello on her experiences investigating WhatsApp in the Brazilian elections in order to further understand how WhatsApp’s influence has been perceived and its effects.    

Patrícia Campos Mello is an award-winning Brazilian journalist working at Folha de São Paulo (Brazil’s largest newspaper). In 2018, she became known throughout Brazil for reporting on the illegal bulk messaging by then-presidential candidate Jair Bolsonaro supporters in the Brazilian elections. On 18 October 2018, the scandal first came to light through her news article entitled “Businessmen Fund WhatsApp Campaign Against PT.” Her investigation, which encompassed several articles, became a major topic in the elections, being addressed by presidential candidates, the national and international media, and having direct effects on the electoral process. Several investigations by federal prosecutors on WhatsApp are still ongoing. Since then, she has been harassed by hate messages and fake news on the Internet, as well as by lawsuits filed by the investigated parties.

Gabriel Pereira: Before the 2018 electoral processes even started, there was a growing sense that WhatsApp would play an influential role in the election cycle. Prior to that, had you followed WhatsApp usage in relation to elections?

Patrícia Campos Mello: I cover mostly international affairs, but that interest [on WhatsApp] emerged after I covered the 2014 elections in India. It was one of the first big ‘social networking elections’… with Modi and Twitter. In that country, the WhatsApp usage is larger, given the number of users, and everything. After that, I covered the Trump election in the United States … and that was when the fake news stories emerged in the U.S. context, however, it was not WhatsApp because they do not use that app as much… most of the WhatsApp users are Latinos. Instead, most Americans use text message (SMS). And then the election happened in Brazil and some stories began to appear in BBC Brazil about ‘troll farms.’ It was not exactly that, but they covered some things around that.  On that occasion, we were following the phenomenon involving Bolsonarist groups [groups created and maintained by Bolsonaro’s followers]. This sort of group did not appear overnight. Especially Carlos [Bolsonaro], they’ve been feeding this for two years, a year … Pablo Ortellado, I think has done more research on this.1 So, when they told me to take a look, I had just done another TV story about a generic hepatitis C medication, and they told me, “Take a look at  WhatsApp and see if you can find out more.” And it was then that I started investigating the issue further. I spent three weeks investigating and talking to people… I talked to a lot of people.

GP: Can you tell us a little more about the backstage process as you worked on these stories, how did you get there, and how the stories developed over time?

Chips used by marketing agencies to do WhatsApp bulk messaging. Image reproduced from Folha de São Paulo

PCM: The research process for the story involved talking to dozens of people. Who are the sources with whom we exchanged messages or gave testimonials? We had as off-the-record sources: marketing agency owner, marketing agency client, former marketing agency employee, and current marketing agency staff. So these are people who were dealing with it… and they had all that knowledge, they catered to customers, they were the ones moving the businesses further… And the agencies said, “Look, we cannot discuss this topic at the moment because we’re full of work, because we got a big order of messages against PT [the Worker’s Party] to be sent the week prior to the Election Day.” Some of the messages included phrases such as ‘Let’s clean the country.’ Again, I do not think it was only Bolsonaro’s people who did this. I mean, I know that PT people hired [a similar service], as we know from the statements shared with TSE [the Superior Electoral Court]. However, it was different in the case of Bolsonaro, as it involved businesspeople.

GP: So the story reveals a case that is not a matter as much about fake news or misinformation, but of breach of the campaign finance law using WhatsApp?

PCM: Essentially there are a few things. In Brazil, it is forbidden to use automated messaging software [during Elections]. And they used it. Afterwards, I wrote a piece where I explain how this basically works… it’s a pretty rough thing. The guys go to Santa Ifigênia [a marketplace in São Paulo known for cheap technological devices, often including illegal and smuggled products] and buy a list of CPFs [Brazilian Social Security numbers]. Then there are a lot of people, hired as cheap freelance workers for a marketing agency. They are the ones who send out the messages. These workers do not even know what kind of messages they are sending out.2

Now, for example, WhatsApp has all the conditions [to better understand the situation]. Soon after the first story came out, I interviewed WhatsApp [staff]. Just so you know, WhatsApp had then banned the marketing agencies’ accounts.3 But… First, they are unable to know for sure what the numbers in which these agencies operated are. Because they have a  ‘factory’ of SIM cards… How can you prove that you have banned all the [marketing] agency’s accounts? How will you know which numbers are related to each agency? When I posed that question to them, they were unable to respond. I do not think they can do it.

Second, what they were able to do is: they banned accounts (the ones they could find) that were related to the four [marketing] agencies I cited in the story. So I asked them the following: For this to happen, you had to detect some kind of abnormal behavior … Do you have some kind of ‘trigger’ [that enables you to detect such abnormal behavior]? Be it someone complaining about spam, be it sending X messages in X seconds… so then you can detect/confirm that a message is probably automated? Then I requested from them: The only thing I want from you –and I’m not asking for any breach of secrecy, no decryption– is the number of messages sent. We know how much it costs per message that gets sent out and, with that information, one can simply show that it’s not in the statement that was declared [to TSE, the Superior Electoral Court]. We know that they banned the agencies, which means they know what the agencies were doing, right? Otherwise they would not ban them … If they banned an agency, it was because they detected an abnormal behavior.6 What triggered such detection? What is the alarm? In this case, then you have to know how many messages have been sent out.

We know exactly how much it costs for each message to be sent, we have copies of the contract. So it would be very easy. But WhatsApp did not [respond]… In fact, the behavior of the media, these big [tech companies]… Facebook was disingenuous.  One of the requests from the STF (the Supreme Federal Court), from [Justice of the Supreme Federal Court of Brazil] Barroso, was to know more about the involvement of WhatsApp, Facebook, and everything else in sponsoring, bulk messaging, etc.4 From WhatsApp, it is obvious you can’t obtain this information because it was not WhatsApp that was contracted to do bulk messaging… instead, it was the marketing agencies that provided such services. But on Facebook, Luciano Hang [a pro-Bolsonaro businessman] had already been fined by the TSE for ‘sponsored content’ in favor of Bolsonaro without properly declaring the spending.5 When Barroso requested that information from WhatsApp, Twitter, etc., Facebook did not put this up. Facebook replied, “No, the Bolsonaro campaign did not pay us.” But of course that’s not what we’re talking about, we’re talking about some people sponsoring pro-Bolsonaro content (maybe they could do that for PT, but it was not the case that time). This is an undeclared campaign donation. In no place Luciano Hang’s sponsoring appeared in Bolsonaro’s declaration for the TSE. Facebook did not report this, although TSE had fined the sponsors. Now, this is a problem with the electoral law because, in theory, you may have several business people paying for sponsoring or bulk messaging for candidates and the candidates may claim that they knew nothing.

In fact, Facebook can control this because they know what is political content and what is not. They know what they sell, right? They know they sold to Luciano Hang to advertise for the Bolsonaro campaign. And this was not properly reported [to the TSE]. In the case of WhatsApp, it’s more complicated. It’s not WhatsApp that gets paid, but rather the marketing agencies that are hired to do this sort of work [of bulk messaging].

GP: For a clearer understanding of how this works from a user’s viewpoint, how is it that a message you receive from a stranger (an unknown contact) may have any effect on your opinions?

PCM: Exactly! I keep wondering about that: it must be very inefficient … it is like when you receive spam emails. But the problem is that this is not the only way they do it. They do this via groups. They spread these messages in groups, and when you are getting a message in the group, there’s a bigger weight, you know? Because then it goes from one group to another. “I received in a group, etc.,” one might claim. These groups are closed. And then you have researchers like Miguel Freitas, from PUC-Rio who did an analysis. He shows how many times each message gets sent out, in a time interval, and concludes there are signs of automated behavior.

GP: I think one of the big questions for me is that nobody really knows how to deal with this as a whole.

Smartphones used by marketing agencies to do WhatsApp bulk messaging during the elections. Image reproduced from Folha de São Paulo

PCM: It’s nobody’s land. When Zuckerberg recently announced that they will do everything to maintain privacy between messages, encrypted… This is the worst thing in the world… and that is WhatsApp. You do not know what goes on inside [the platform and its users]. There are hundreds of groups, private… closed. And then you have people blasting messages of disinformation within the group. With the help of automated systems. Which are not anything too fancy. And then you can spread messages in many groups…. What I think is: some candidates truly have faithful digital followers. And they also have bots and trolls that inflate this [their campaign and/or digital participation]. Even to create narratives.7 And this was always very clear. But in regards to WhatsApp… the groups are, in fact, closed. They are encrypted. (…) We need to have people who understand computer science to track these people. Otherwise this is going to go on completely unpunished. To say such things as: “oh, the group is closed, it is encrypted, and these things are randomly generated.” If you say so, you sit and wait… That can’t be possible! There must be people studying, people with technical expertise which I do not have8 … neither the journalist nor the STF… to kind of see how we can solve this issue.

GP: This story is incredible, because the impact of the investigation was massive, right? It kind of dominated the discussion in the elections. How did this impact you?

PCM: I think that there are two aspects to it. The first is that everyone was seeing the influence of WhatsApp empirically, qualitatively, and not quantitatively… Then, suddenly, you have a newspaper story that sheds some light on it [the quantitative dimension of the bulk messaging]. The second is that the country was highly polarized. There is Bolsonaro, and his surrounding figures, who are very adept at using social media. Very much so. And they managed to make use of an army of bots. In other words, it is surreal. In these processes, some of these followers attempted to turn my life upside down. For instance, they found a 2013 interview I gave to university students when I said I was a leftist. And they made that viral. And there were many bots sharing it. So I think it was because of these conjunction of factors.

GP: How do you see the relevance of WhatsApp in Brazil today as part of your investigation?

PCM: By 2014, you had people who started doing bulk messaging with SMS (text messages). They did some little thing with WhatsApp, but WhatsApp had a much smaller penetration then. I think, and you must already know, that in Brazil today WhatsApp replaces the telephone. People do not pay to call anymore, you know? You simply use Wi-Fi and call via WhatsApp. You have some interesting surveys by Ibope and Datafolha. The percentage of people who inform themselves through WhatsApp … is gigantic, right?9

GP: After the story came out, there were ‘fake news’ about you being shared on the internet… When talking about Folha de S. Paulo, Bolsonaro even went as far as to say that “Folha itself is over.” What are your thoughts about the role of journalists in the midst of such polarized debates that often involve hate messages and misinformation?

PCM: Journalists in Brazil are experiencing very difficult times, as they are harassed and intimidated via social media. The confrontation with the media is an important part of the current government’s agenda in Brazil. But at the same time, this whole thing makes evident the importance of independent journalism in times of polarization and populism. I have been  threatened many times through social networks and over the phone, and it was a very difficult period for me, but I keep doing my job.10

We want to thank Patricia for taking the time for this interview with us, and for sharing her extensive experience on this subject.

1 GP: Projeto Monitor do debate político no meio digital (Monitor of Political Debate in Digital Media Project) by Prof. Pablo Ortellado and colleagues has a series of reports and publications, in Pportuguese, about the topic. Also worth mentioning is the op-ed written by Cristina Tardáguila, Fabrício Benevenuto and Pablo Ortellado for The New York Times.

2 GP: More details about this investigation can be read in the news piece by Artur Rodrigues and Patrícia Campos Mello, in Portuguese on: “Fraude com CPF viabilizou disparo de mensagens de WhatsApp na eleição”; or in English, summarized: “Identity Theft Enabled WhatsApp Election Blasts”

3 GP: More details about WhatsApp banning the accounts, as well as other documents about the bulk messaging purchase can be read in Portuguese on: “Documento confirma oferta ilegal de mensagens por WhatsApp na eleição”; or in English, summarized “Document Confirms Illegal Sale Of WhatsApp Messages During Election”.

4 GP: More information on the request by Justice Barroso can be read, in Portuguese, on: “Barroso manda gigantes da internet responderem sobre disparos pró-Bolsonaro”.

5 GP: More information about the Luciano Hang being fined by TSE can be read, in Portuguese, on: “Empresário é multado por contratar Facebook para impulsionar conteúdos”

6 GP: More information about Miguel Freitas’ research, in Portuguese: “Relatório enviado à PF sugere caminho para rastrear fake news”. Researchers from ITS-Rio also have shown signs of automation: “Computational Power: Automated Use of WhatsApp in the Elections”.

7 GP: Perhaps the most infamous case of pro-Bolsonaro bots being used in the 2018 elections, were when bots were detected making automated responses to posts by Folha de Sao Paulo that contained the word “bolovo” [a Brazilian delicacy], because of its similarity with Bolsonaro’s name: “Folha publica palavras ‘bolso’ e ‘bolovo’ no Twitter e respostas sugerem ação de robôs pró-Bolsonaro” (in Portuguese)

8 GP: Besides other projects already cited in these footnotes, it is worth mentioning the project by researchers at UERJ, that investigated group dynamics on WhatsApp: “Grupos pró-Bolsonaro no WhatsApp orquestram fake news e ataques pessoais na internet, diz pesquisa” (in Portuguese)

9 GP: According to Datafolha research from 10/27/2018, 66% of Brazilian voters have an account on WhatsApp, 24% shares news about politics and elections on WhatsApp, and 46% reads news about politics and elections on WhatsApp.

10 GP: To read more about threats on journalist, we suggest, for example, the text by Fernanda Canofre: “Brazilian journalists face hacking, doxxing and other threats as election draws near”. In the end of 2018, Folha’s Ombudsman, Paula Cesarino Costa, recognized directly that the “environment is charged for news producers”: “Agonia de grupo tradicional, violência e fraudes marcam ano difícil para jornalistas” (in Portuguese).

Blog in Portuguese / Leia em Português