Alberto Escorcia first noticed the bots in 2010.
Escorcia, a seasoned Mexican internet activist, had been thumbing through his Twitter feed, when he noticed several accounts posting derogatory messages at supporters of a national electricians' union strike. Escorcia realized they didn't stem from real people, as they all repeated the same message and had profile photos of attractive people from other countries. He started documenting a veritable troll boom the following year during Mexico's presidential elections, when tens of thousands of these bots mobilized support for political candidates.
"I realized that it was the start of a new war in Mexico, and that the internet was the new paradigm," Escorcia told Motherboard in an interview at a cafe in Mexico City's historic center. Escorcia has since become an expert on bots and trolls in Mexico, running a blog called LoQueSigue, or "What Follows," to which he publishes data, analyzes hashtags and trends on Twitter, and maps the source of tweets. He also shares information about social media strategies for activists.
Taking on this role in a country such as Mexico, where swarms of automated Twitter accounts exercise a troubling amount of political influence and threaten the safety of activists, has not come without cost to Escorcia's own sense of security. Escorcia explained that his activism has repeatedly been met with violent intimidations, which bridge the virtual and physical worlds. He said that almost every time he publishes one of these analyses or talks about them in the media, he receives death threats.
"It is time that Twitter recognizes their platform has become a weapon in Mexico."
These threats aren't to be taken lightly: Mexico was the third most deadly country in the world for journalists, according to the press freedom group Reporters Without Borders. At least nine journalists were murdered in Mexico in 2016 alone. And while Escorcia is technically covered under a federal protections for journalists and activists, he doesn't trust the government's ability to protect him.
When attacked in the street, Escorcia pressed a panic button the government issues at-risk journalists and activists. Police didn't respond for half an hour. Recently, he began fearing for his life after someone started harassing him at home, repeatedly ringing the doorbell in the middle of the night. He never saw anyone, but the message was clear: Keep publishing about the bots, and something terrible might happen. When Motherboard spoke to Escorcia in February, he was planning on leaving Mexico indefinitely.
"So many people who have been threatened on Twitter stop publishing critical things online, they stop going to protests, and, even worse, sometimes they stop going outside," Escorcia told me. He added that the widespread belief that the Mexican police won't protect activists—even those facing death threats—exacerbates the problem.
"The troll takeover of Twitter is deeply concerning in the context of the wider human rights crisis in Mexico," said Tanya O'Carroll, a technology and human rights adviser for Amnesty International who recently published an investigation of the political impact of bots and trolls in Mexico. "The emerging pattern of well-funded and sophisticated cyberattacks and misinformation campaigns online are fueling this climate of fear and silencing those who speak out."
O'Carroll refers to what many consider bots as trolls or troll farms, since there are actual humans controlling them—though sometimes the trolls are behind hundreds of accounts at the same time.
On March 8, 2016, freelance journalist Andrea Noel, tweeted a surveillance video of a man who had lifted up her skirt and pulled down her underwear while she walking down the street in a hip, upscale neighborhood in Mexico City. Noel tweeted the video with the message "women should be able to walk safely" and requested that people help identify her attacker. Her timeline was immediately flooded with threats from both bots and trolls that included both her location and photos of armed men. Trolls also appeared to tell Noel that she was a "slut" who "deserves to be raped" for wearing the dress.
"I have reported dozens of accounts that have sent me death threats, and nothing has happened."
LoQueSigue published data clouds showing at least 20,000 mentions of Noel's Twitter account in just one day. Many of the accounts had few followers, no photos, and failed to maintain their conversations like a real person would—all prime indicators that the accounts were operated by bots. When a laser shined through Noel's apartment window and landed on her forehead, she decided it was time to leave Mexico City. She'd eventually settle in Tijuana.
A year later, Noel still works as a journalist and continues to use Twitter to share her work. She continues to criticize sexual harassment and the impunity of the harassers. She told me that she still feels on-edge.
"It is easy to say, 'just ignore it,'" Noel admitted. "But when you want to make coffee for breakfast and someone just wrote you saying 'I am going to kill you' or 'I am going to rape you, bitch'? It does affect you."
Mexico is one of the top ten Twitter markets in the world, with close to 10 million registered users. But Twitter invests little money toward monitoring its non-English language users. "You see that Twitter has suspended—for life—the accounts of Milo Yiannopoulos or Azealia Banks," Noel said. "But they don't focus on Latin America. I have reported dozens of accounts that have sent me death threats, and nothing has happened."
Andres Monroy, a professor at University of Washington, has researched the role bots play in influencing political thought in Venezuela, as well as the murders of Twitter users who tweeted about violent crimes in Northern Mexico. Monroy told me that companies like Twitter must walk a very fine line—they "want to increase engagement and have more people use the platform," he said. "So how much they should invest without blocking too many accounts, killing real users while trying to controlling trolls or bots?"
It is also impossible, Monroy added, for Twitter to understand the political context in every country where their users are based.
O'Carroll believes that Twitter not localizing its strategy is at the heart of the problem. "It is time that Twitter recognizes their platform has become a weapon in Mexico and it is being used to silence and punish critical voices," O'Carroll said. "Instead, they are choosing to ignore the presence of orchestrated troll networks on their platform and behave as if it were somebody else's problem. The company can and must do more."
Twitter's official policy reads: "We believe in freedom of expression and in speaking truth to power, but that means little as an underlying philosophy if voices are silenced because people are afraid to speak up." Twitter's rules section also states that the company will not tolerate "behavior that harasses, intimidates, or uses fear to silence another user's voice." Twitter was contacted for this story but did not respond in time for publication.
Twitter has played a major role in large mobilizations around the world, from the Arab Spring to Occupy Wall Street. It has been a key protest tool in Mexico, especially over the past five years, as human rights violations have continued to rise during the presidency of Enrique Peña Nieto. Shortly after the disappearance and murder of the 43 students from the Ayotzinapa Teachers College in 2014, protests erupted in the streets and online when the attorney general refused to answer reporter's questions after discovering the students' bodies. Instead, he told reporters, Ya me canse. "I am tired."
Users began tagging their political critiques with #YaMeCanse, which became the most-used hashtag in Mexican history—and immediately came under attack by bots. As documented by local activists like Escorcia and an international group of scientists, automated accounts started tweeting the hashtag out with links to pornography or violent photos in the hopes that Twitter will flag the hashtag as spam and block it. Twitter activists responded by adding numbers to the end of the hashtag, changing the number every time they came under attack. They eventually got to #YaMeCanse25.
Digital activists in Mexico often refer to the agents that hijacked these hashtags as "Peñabots" since they defend Peña Nieto and the political establishment. In 2016, hacker Andres Sepulveda told Bloomberg Businessweek he was paid $600,000 by the Institutional Revolutionary Party to hack the election, using 300,000 bots to drum up support for Peña Nieto.
While the use of bots in Mexico has greatly expanded and often serves as a marketing tool, it's still common to see armies of bots warring for the interests of politicians and against social movements. On January 1, the price of gas shot up 20 percent in one day, an enormous jump in a country plagued by a deflated peso and a struggling economy. Civil society started to mobilize against the hike, protesting in the streets, blockading border crossings and highways, and tweeting about the price hike, also called the "gasolinazo."
A new hashtag started to emerge in the tweets about the gasolinazo protests, urging people to loot with the phrase "#SaqueUnWalmart" which translates to "Loot a Walmart." Looting did also occur in various cities throughout Mexico, but not on the scale in which it was promoted online. Fearing chaos provoked by possible looting, ordinary citizens refused to leave their homes.
LoQueSigue and Signa Lab, a data research lab based out of the Western Institute of Technology and Higher Education in Guadalajara, noticed that something was fishy with the hashtag. LoQueSigue published an analysis of the tweets showing that close to 500 fake users tweeted the hashtag, with the majority hailing from two northern Mexico City districts.
Signa Lab concluded that "Loot A Walmart," in tandem with the gas-hike hashtag, helped generate a mass panic, leading people to look down on the protesters and fear leaving their homes. The data lab also found that #SaqueUnWalmart had links to #Hail100cia, a hashtag affiliated with the online group Legion Holk. Members of Legion Holk allegedly used social networks to celebrate a recent school shooting in Monterrey, Mexico.
Luis Natera, a researcher with Signa Lab, said that the hashtag led to curfews and generated collective fear. "With this hashtag, they want to influence the conversation, to change trends, insert false information, and mitigate... criticism [of] the government," Natera explained.
Mexico is a year out from its next presidential election cycle. Researchers like Natera fear that the digital battleground will only intensify, with more bots backing candidates and swaying votes. Escorcia, on the other hand, hopes the threats he and other activists have faced will subside by 2018.
"I have hope that this will go away soon," he said. "Mexico is full of brave people who endure a lot and will soon make changes."