Jun 30, 2018
#ElectionWatch: Bot Battlefield at #AMLOFest
Bot-like accounts use hashtag #AMLOFest to spread pro and anti-AMLO messages three days ahead of the election
On June 28, presidential candidate Andres Manuel Lopez Obrador, also known as AMLO, organized a festival to commemorate the closing of his electoral campaign, ahead of election day on July 1.
The festival was accompanied by a hashtag campaign under #AMLOFest, which quickly attracted pro-AMLO and anti-AMLO accounts, who used the hashtag to respectively promote and smear presidential candidate Lopez Obrador.
The hashtag started trending late on June 27 and quickly generated more than 234,000 mentions. According to a machine scan those mentions were generated by a total of 65,900 users. On average, each user generated 3.5 hashtag mentions, which is a high figure. In scans of organic traffic which @DFRLab conducted previously, a typical average rate of posting ranges from 1.1 to 2.2 posts per user. The traffic on #AMLOFest generated 30 percent more, suggesting artificial amplification.
Moreover, 80 percent of all hashtag mentions were retweets, indicating that only a small number of accounts generated unique content, while the remaining users simply amplified it. The high level of amplification indicates automation or — at the very least — a small group of hyperactive users purposefully boosting the campaign.
@DFRLab found that automation was used by both groups — accounts supporting AMLO as well as accounts using the hashtag to spread anti-AMLO messages.
Anti-AMLO campaign
Six accounts, whose locations were set as Puebla state, posted anti-AMLO memes using the #AMLOFest hashtag.
There was reason to believe that these accounts were managed by the same individual or group because they were all created in the same time span of December 24 and 27, 2011, and posted between 21 and 36 tweets since. The fact that they had similar activity patterns suggested that they were part of a very small network that attempted to hijack the #AMLOFest hashtag.
The six users also shared links to a site www.elobscurosecretodeamlo.com, which translated from Spanish means, “the dark secret of AMLO.”
The site consisted of a landing page and an embedded YouTube video called “AMLO’s dark secret”. The YouTube video was uploaded on June 26 and was viewed 13,000 times at the time of this report.
The site as well as the YouTube page was run by an anonymous group or individual and accused AMLO of murdering his own brother and best friend, which was inaccurate and inconsistent.
The hashtag #ObscuroSecretoDeAMLO appeared to have been spread by inauthentic accounts. A machine scan showed that 331 users used the hashtag 1,969 times, averaging at 5.9 hashtag mentions per user, which indicated automation or inorganic amplification.
The hashtag campaign was led by the following ten users, who altogether generated 1,233 hashtag mentions, 62 percent of the total traffic, suggesting the campaign was neither authentic nor organic.
The top ten most active accounts appeared to be a part of an influence network for hire, as they had promoted various political candidates over the past couple of months.
For example, two most active accounts @ElPinche_Cholo and @ivickyvicky2012 promoted #yovotovila (a pro-Mauricio Vila Dosal campaign, who was running on behalf of the National Action Party), #yoconmarianaboy (a Mariana Boy-supporting hashtag, who’s representing the Green Party), and #todosconlomeli (which promoted the PRI candidate, Jesus Lomeli Rosas).
Automated pro-AMLO accounts
Among the most active accounts, which used the #AMLOFest hashtag to share pro-AMLO messages were several bot-like accounts.
The most notable bot-like account was also the most active — @RobotinaChairi1. Although in its bio it explicitly says it is not a bot, there’s overwhelming evidence that suggests otherwise.
Since its creation on June 21, the account has tweeted 2548 times, averaging at 30 tweets an hour without any breaks. Moreover, the account has used the hashtag #NoSoyBot (I’m not a bot) 279 times in less than seven days. Again, that was three ‘I’m not a bot’ hashtag mentions per hour since the account’s creation on June 21.
Apart from that, the account used the URL shortener iftt.com, which allows users to automate their posts according to a number of criteria — for example, retweeting any post with a given hashtag. The use of this software, in fact, is one of the 12 bot indicators the @DFRLab has recently identified.
Another hyper-active account that used the hashtag — @Amelia20870330 also resembles a bot. The account was created on June 26 and since posted 375 tweets, averaging at 187 tweets a day. Notably, all of the account’s 375 tweets are retweets.
Conclusion
In the closing minutes of the official campaign season in Mexico, #AMLOFest showed how a trending hashtag can serve as a battleground for bot-like accounts that are promoting two sides of the argument and as a result skewing the organic online debate. The anti-AMLO accounts clearly failed to hijack the #AMLOFest hashtag to seed their own campaign, accompanied by the unattributed video accusing Lopez Obrador of killing his brother and best friend. At the same time, hyper-active pro-AMLO accounts appeared to have out-tweeted real users involved in the hashtag campaign, further distilling the organic discussion.
@DFRLab will continue to monitor the information space ahead of the vote in Mexico on July 1 with a particular eye on the media blackout 72 hours ahead of ballots being cast. During the intended time of reflection during the blackout ahead of the vote, campaigns are not allowed to hold official events and traditional media cannot cover the campaigns with content like polling data. This regulation does not apply to social media, and thus creates an unintended vulnerability of an unchecked social media space.
Donara Barojan is a Digital Forensic Research Associate at the Atlantic Council’s Digital Forensic Research Lab (@DFRLab).
#ElectionWatch in Latin America is a collaboration between @DFRLab and the Adrienne Arsht Latin America Center at the Atlantic Council.
Follow along for more in-depth analysis from our #DigitalSherlocks.
