Abstract
When online protests erupt, fake social media accounts can swoop in to hush them up, or amplify their own agenda.
Politically active Twitter-using Mexicans have become deeply suspicious of fake accounts that leap into action when an awkward hashtag emerges. The bots disrupt it by jumping into the discussion, burying the protesters’ messages in a sea of spam. The Mexican word for this practice of silencing voices online is tecnocensura (technocensorship). It first became prominent during the 2012 presidential election, when it was widely reported that #YoSoy132 (a protest against media bias towards Peña Nieto’s campaign) had been leapt on by fake accounts. According to the Anonymous hackers’ collective, these so-called “Peñabots” also appeared to help boost positive messages for Nieto during the elections, such as #PeñaGanaDebate (Peña wins debate).
“We are at a point where you do not know what is real,” said blogger Alberto Escorcia, who through his investigative project Lo Que Sigue has made map-like diagrams showing how the Mexican bots swoop in. “This is not just seen on Twitter, but also on the internet at large, and it is affecting our ability to make free decisions.”
Bot armies are by no means unique to Mexico. Lawrence Alexander, an Open University student who conducts data research in his spare time, carried out a study of 20,500 pro-Kremlin accounts on Twitter, finding the tell-tale signs for fake accounts: lack of location or time-zone information on their profiles, no interactions with other users. Along with bots that had been operating for two or more years, gaining more credibility through their longevity, he also found spikes in the number of new bots coinciding with the start of the Euromaiden protests in Ukraine (late 2013) and the pro-Russian unrest in eastern Ukraine (early 2014).
Bots waded in to discussions after Russian opposition politician Boris Nemtsov was murdered in February, saying Ukrainians killed him
“What I noticed in all of these cases was a group of accounts tweeting the same identical phrase. Repeating a keyword-laden sentence would tend to amplify a story or subject,” Alexander told Index. “What the bots are doing is against Twitter’s rules, but it’s clear many of these fake accounts are passing undetected and slipping though the net. In some cases, they do get caught out; the network of bots tweeting about Nemtsov, for example, was suspended shortly after I identified it.” These bots had waded into discussions after Russian opposition politician Boris Nemtsov was murdered in February, repeating messages saying Ukrainians killed him.
A protester takes part in the #EPNnotwelcome demonstration, during Mexican President Enrique Peña Nieto’s state visit to London in March
Credit: Guy Corbishley / Demotix / Press Association Images
Social-network manipulation is now a business. In March, St Petersburg blogger Marat Burkhard told Radio Free Europe about the two months he was employed by a mysterious agency to push pro-Putin messages in online comments, totalling up to 30,000 comments a day between 300 staff. Turkish President Recep Tayyip Erdogan has also been accused of having a mighty bot army, and, after the 2013 Gezi Park protests, announced he was taking on 6,000 “social media experts”. In April, the New Republic magazine in the US reported on a “click farm” in the Philippines, where it found the work was generally more commercial than political, but techniques were getting increasingly savvy. The “click farm” company employed real people to manage fake accounts, with each account registered to a different mobile sim card, filed away in case Facebook requested a text message to prove posters weren’t machines. From tens of thousands of cards, the employee digs out the right one to respond to the verification text, then continues to pose under multiple fake names, boosting the “likes” of different companies, products and people.
On Twitter, the holy grail is to become one of the top 10 “trending” topics of the moment. “Getting on this list means many, many more eyeballs,” said Gilad Lotan, chief data scientist at US start-up studio Betaworks. “Whether it’s a conference or a movement, people are coming up with strategies to try and get on that list. One strategy is to use a unique term that wouldn’t have been used before. Then, the key is getting lots of people to use it at the same time.”
Speed and timing are crucial. Although the exact workings of Twitter’s algorithms are not made public, it is understood that trending topics are not picked solely for their popularity, but for a spike in activity. This is why the protest movement OccupyNewYork never trended, because it was sustained over a long period. But, around the same time, Steve Job’s death and Kim Kardashian’s wedding did, because there was sudden peak in interest. This could also be why Mexico’s long-running YaMeCansé protest about the disappearance of the 43 students also dropped off the trending list.
As a research experiment, Lotan bought 5,000 followers for £5. He watched how paid-for bots operate, seeing that those that followed him also followed businesses, musicians and politicians, who appeared to have paid for the privilege. Extra followers get users more visibility and credibility, and, in turn, more genuine followers. In their 2012 US presidential election campaigns, both Barack Obama and Mitt Romney were accused of buying fake followers. Their teams denied it, and, indeed, anyone can buy followers for someone else’s account, so directions needn’t have come from the top.
Social media networks are under pressure to delete fake accounts, but at the same time people need anonymous accounts to share information, especially whistleblowers and people who live under oppressive regimes, where they might be jailed for a tweet.
Twitter was asked for a comment on their bots policy, but as this article went to press had not responded.
This is not just seen on Twitter, but also on the internet at large, and it is affecting our ability to make free decisions
“We all need to think more deeply about media literacy, and understand where the content coming to us is coming from,” said Lotan. “Is it an algorithmic system or curated by people we follow? Do we actually follow them, or it something else pulling the strings? We need to question the content we are seeing and not take it for granted.”
© Vicky Baker
