when foreign state actors aren’t involved. New technology is neither
inherently good nor inherently bad, nor is it inherently neutral. Every
technology has certain affordances and certain limitations, and though they
may not be value-laden, they do tend toward certain kinds of social effects.
When a society is already experiencing social conflict or upheaval, those
effects often become more drastic and less predictable. For many countries
experiencing social change and rapid technological adoption at roughly the
same time, that tech has, for better or for worse (or a little of both),
changed those societies forever.
Bots in Brazil
Brazil’s political landscape is fraught, fractured, and unstable. It is a young
democracy, born out of dictatorship only in 1985. As of 2014, there were 28
political parties with representation in Brazil’s two chambers of Congress, and
governance almost exclusively happens by coalition. This can lead to
uncomfortable, even disastrous alliances, like the one that reelected Dilma
Rousseff to the presidency in 2014. As part of her coalition building in 2010,
she named a member of another party, Michel Temer, to be her vice presidential
running mate. However, when calls for Rousseff’s impeachment came almost
immediately after her reelection in 2014, Temer joined the opposition. When
Rousseff was removed from office in 2016, Temer became the new president.
While there were certainly real issues about Rousseff’s presidency raised by
real Brazilian citizens, it appears that social media bots played a significant role
in swaying public sentiment and contributing to the calls for impeachment
that were ultimately successful.
Social media–based disinformation is a staple of Brazilian politics. Political
campaigns paying for the propagation of information on social media are illegal
in Brazil. Brazilian law also dictates that all social media campaign activity must
be controlled by “natural persons”—that is, not run via automation. 13
13Dan Arnaudo, “Brazil: Political Bot Intervention During Pivotal Events,” in Computational
Propaganda, ed. Samuel Woolley and Philip N. Howard (Oxford: Oxford University Press,
2018), p. 136.
96
Chapter 6 | Democracy Hacked, Part 2
However, as these laws are difficult-to-impossible to enforce, automation and
online rumors are part and parcel to many political campaigns in Brazil from
the presidency down to local elections.
In the case of Dilma Rousseff, it appears that both her campaign and that of
her 2014 presidential election opponent, Aécio Neves, used social media
bots to spread campaign messages and/or smear each other. According to
researcher Dan Arnaudo, Neves used bots more extensively in the
presidential campaign than Rousseff.14 Once Rousseff won the election,
though, most of the anti-Rousseff bots stayed online and almost immediately
joined the pro-impeachment campaign. In Arnaudo’s words, “The online
electoral campaign never ended, and these networks became key tools for
generating support for impeachment. ”15 Adding insult to injury, the fact that
Rousseff’s social media apparatus now belonged to the administration meant
tighter restrictions and stricter oversight, giving the pro-impeachment
networks a distinct advantage online.
Bots can have a variety of effects. Sometimes those effects are minimal, even
null. After all, without an audience, even the most active and inflammatory
bots are just “shouting” into the proverbial wind. This is particularly the case
now in 2018, as platforms have taken measures to prevent (if not eliminate)
automation-fueled abuse. For instance, Twitter is now more aggressive about
deleting bots and preventing new accounts from appearing in search results or
driving traffic toward “trending” topics until they demonstrate signs of
“organic” use. Facebook has drastically pulled back on what can be done via
their application programming interface (API), making automation far more
difficult to accomplish. However, in 2014, the doors to the platforms were far
more open, and automated posts, comments, retweets/shares, and likes/
favorites were easy to manage at scale on both platforms, leading to the
“hacking” of algorithmic content selection in Facebook’s news feed and
Twitter’s real-time “trends.” In fact, as Arnaudo states, on Twitter “the most
generated by bots.” 16
It is impossible to quantify exactly how much impact automation and other
inauthentic/fraudulent online activities had on Brazil’s election and
impeachment results. However, it is clear from both the engagement levels of
the most shared content and from the continued financial investment of
campaigns—despite its illegality—that the impact is not trivial. And when
14Ibid.
15Ibid., p. 137.
16Ibid., p. 140, based on Éric Tadeu Camacho de Oliveira, Fabricio Olivetti De França,
Denise Goya, and Claudio Luis de Camargo Penteado, “The Influence of Retweeting
Robots during Brazilian Protests,” paper presented at the 2016 49th Hawaii International
Conference on System Sciences (HICSS), Koloa, DOI: 10.1109/HICSS.2016.260.
Data versus Democracy
97
Twitter trends, Facebook news recommendations, and YouTube’s “Up Next”
autoplay videos significantly impact the content that users encounter, even
small direct impact can be compounded when manipulated output becomes
input for subsequent algorithmic analysis. (Think back to Chapter 3’s discussion
of the algorithmic feedback loop.)
“Weaponizing” Facebook in the Philippines
Brazil isn’t the only country to see a social media disinformation apparatus
continue on, retooled, after the election for which it was first created. In May
2016, after the election of President Rodrigo Duterte, the citizens of the
Philippines saw his campaign’s powerful social media propaganda outfit become
a critical part of the new president’s administration.
Rodrigo Duterte started off at a disadvantage. If campaign spending on
traditional media is any indication, he had less funds to work with than some
of his rivals. However, he had a powerful social media apparatus. Duterte’s
team was driven online because “limited campaign funds forced them to be
creative in using the social media space. ”17 They took advantage of “training
sessions” that Facebook provided to the presidential candidates ahead of the
election. In a country with more smartphones than people and where 97% of
the people who are online are on Facebook,18 that Facebook mastery proved
indispensable.
Headed by Nic Gabunada, Duterte’s social media team consisted of several
hundred connected volunteers. “It was a decentralized campaign: each group
created its own content, but the campaign narrative and key daily messages
were centrally determined and cascaded for execution.” 19 (This is a similar
playbook to the Russian operations in the United States that same year that
we explored in the previous chapter.) Notably, the content being advanced by
Duterte’s supporters were not simply policy positions and innocuous
statements of support. Many Duterte supporters called for violence against
his critics, leading to situations like an online mob in March 2016 threatening
violence, even death, against a small group of students opposed to Duterte. 20
17Chay F. Hofileña, “Fake accounts, manufactured reality on social media,” Rappler, last
updated January 28, 2018, www.rappler.com/newsbreak/investigative/148347-
fake-accounts-manufactured-reality-social-media.
18Lauren Etter, “What Happens When the Government Uses Facebook as a Weapon?,”
Bloomberg Businessweek, published December 7, 2017, www.bloomberg.com/news/features/
2017-12-07/how-rodrigo-duterte-turned-facebook-into-a-weapon-with-a-little-
help-from-facebook.
19Maria A. Ressa, “Propaganda war: Weaponizing the internet,” Rappler, last updated
October 3, 2016, www.rappler.com/nation/148007-propaganda-war-weaponizing-
internet.
20Ibid.
98
Chapter 6 | Democracy Hacked, Part 2
Journalist Maria Ressa (founder of Rappler) calls this a “death by a thousand
cuts” strategy to intimidate or silence critics, where a large number of users
overwhelm a target with online threats. 21
writes: “Since being elected in May 2016, Duterte has turned Facebook into a
weapon. The same Facebook personalities who fought dirty to see Duterte
win were brought inside the Malacañang Palace. ”22 Gabunada told another
Rappler journalist that “there was a need to continue campaigning [after the
election] because they got ‘only 40% of the votes’ and needed more than that
to allow Duterte to effectively govern. ”23
This involved far more than a president tweeting or even a team of bots
amplifying his messages. It was part of a coordinated media strategy aimed at
consolidating Duterte’s power and silencing his opponents, whether politicians
or journalists. Soon after the election, Duterte boycotted private media for
two months, refusing to talk to any independent reporters and only pushing
propaganda through state-controlled media. Simultaneously, pro-Duterte
trolls and sockpuppets (fake accounts controlled by real humans) attacked
independent journalists that criticized the government. When Ressa published
her article, “Propaganda War,” which unveiled many of these activities, she
was immediately attacked by pro-Duterte trolls, including both death and
rape threats and calls for her to leave the country. In Ressa’s words, since
Duterte’s election, “when someone criticizes the police or government on
Facebook, immediate attacks are posted. ”24
According to Etter, this isn’t simply a case of the Duterte administration
“hacking” the system or taking advantage of Facebook’s targeted advertising
features. In addition to providing training sessions to the Philippine presidential
campaigns, after the election Facebook “began deepening its partnerships with
the new administration.” 25 This wasn’t exceptional, but “what [Facebook] does
for governments all over the world … offering white-glove services to help it
maximize the platform’s potential and use best practices.” Etter continues:
Even as Duterte banned the independent press from covering
his inauguration live from inside Rizal Ceremonial Hall, the new
administration arranged for the event to be streamed on Facebook,
giving Filipinos around the world insider access to pre- and post-
ceremonial events as they met their new strongman.
21Ibid.
22Etter, “What Happens When the Government Uses Facebook as a Weapon?”
23Hofileña, “Fake accounts, manufactured reality on social media.”
24Ressa, “Propaganda War.”
25Etter, “What Happens When the Government Uses Facebook as a Weapon?”
Data versus Democracy
99
Though Facebook has since made it more difficult to perform some aspects
of the instant trolling that has plagued Duterte critics, the attacks on
independent journalists continue under Duterte’s rule. In November 2018,
the Philippine government announced its intent to charge Ressa’s media
outlet, Rappler, with tax evasion. Rappler characterized the accusations as
“intimidation and harassment. ”26
It doesn’t look like this problem is going away any time soon.
Consolidating Power in Myanmar
The Duterte administration is not the only government to use Facebook as a
weapon against their own people. Recent investigations have also uncovered
a large-scale information operation being conducted by the Myanmar military
against the minority Rohingya people of Myanmar.
Violence and oppression are nothing new in Myanmar. Myanmar is a young
quasi-democracy. It has long been ruled by dictators and autocrats, and neither
government oppression nor untrustworthy state-controlled media are
strangers in that country. There is also a long history of violence between the
Buddhist majority and the Rohingya people, a Muslim minority population
that lost their rights to citizenship almost overnight in 1982. 27
For years, the Rohingya have been on the receiving end of legal discrimination
from their government and violence from their compatriots who do not share
their religious or ethnic identity. In addition to perpetuating religious
discrimination in education, work, and legal standing, Myanmar’s Buddhist
population, often led by their monks, have routinely held rallies where they
spew hate speech at the Rohingya. Not infrequently these rallies have even
resulted in the death of some Rohingya people.
Both the scale of the violence and the specificity with which the Rohingya are
targeted led global watchdog agencies to characterize the anti-Rohingya
violence as an act of (attempted) genocide. 28
26Alexandra Stevenson, “Philippines Says It Will Charge Veteran Journalist Critical of
Duterte,” The New York Times, published November 9, 2018, www.nytimes.
com/2018/11/09/business/duterte-critic-rappler-charges-in-philippines.html.
27Krishnadev Calamur, “The Misunderstood Roots of Burma’s Rohingya Crisis,” The Atlantic,
published September 25, 2017, www.theatlantic.com/international/archive/
2017/09/rohingyas-burma/540513/.
28Timothy McLaughlin, “How Facebook’s Rise Fueled Chaos and Confusion in Myanmar,”
Backchannel, WIRED, published July 6, 2018, www.wired.com/story/how-facebooks-
rise-fueled-chaos-and-confusion-in-myanmar/.
100
Chapter 6 | Democracy Hacked, Part 2
Myanmar also has a “rich history” of propaganda. Timothy McLaughlin of
Wired magazine writes that “Myanmar had spent decades reliant on state-run
propaganda newspapers. ”29 For just about the entire history of mass media
technology—newspapers, radio, television, the internet—Myanmar’s
government has controlled the primary channels of media distribution. And
so for those living in Myanmar today, they have known two main sources of
information throughout the bulk of their lives: state-controlled media in a
dictatorship an
d local rumors or gossip. According to former U.S. ambassador
to Myanmar, Derek Mitchell, “Myanmar is a country run by rumors,” 30 and this
assessment is shared by a Human Rights Impact Assessment commissioned by
Facebook, which characterized Myanmar as a “rumor-filled society” up to
around 2013. 31
All of this was happening well before smartphones, Facebook, or the world-wide
web entered the scene. Nevertheless, the arrival of digital technology threw fuel
onto the fires already burning in Myanmar.
Two key things have changed in Myanmar in the last ten years: the arrival of
the internet (which for most in Myanmar basically means Facebook) and the
advent of (some aspects of) representative democracy.
While the military junta that had previously governed Myanmar was dissolved
in 2011, it is generally understood that the first truly open elections in
Myanmar took place in 2015. 32 A number of people in Myanmar already had
access to the internet in 2015, but it was after this point when the government’s
censorship of the press and control over online speech began to diminish
significantly, though not completely. As a result of digital technology expanding
at the same time that long-standing limits on free speech were diminishing,
the general populace of Myanmar exhibited a general lack of media and digital
literacies necessary to navigate such an open information environment as the
internet. And since “for the majority of Myanmar’s 20 million internet-
connected citizens, Facebook is the internet,” 33 this lack of media literacy was felt most poignantly on Facebook’s platform.
The negative results were first seen in the general relationship between
Buddhists and Muslims in Myanmar. The long-standing animosity between
these groups was exacerbated by the newfound freedom of expression and
access to a platform with significant reach. And with some aspects of anti-
Rohingya discrimination baked into the law, the oppression of Muslims at the
29Ibid.
30Ibid.
31“Human Rights Impact Assessment: Facebook in Myanmar,” BSR, published October,
2018, https://fbnewsroomus.files.wordpress.com/2018/11/bsr-facebook-myan-
mar-hria_final.pdf, p. 12.
32Ibid., p. 11.
33Ibid., p. 12
Data versus Democracy