Aug 26, 2017
Toxic Hashtags vs. #GobiernoEspía
Bots and trolls in Mexico used a new tactic to manipulate a trending topic on Twitter during two days in June 2017. This new technique affected #GobiernoEspía, an important hashtag which corresponded to bombshell revelations published on June 19 that the Mexican government was targeting journalists, activists and human rights defenders with Pegasus malware.
An investigation conducted by Mesura and Signa_Lab demonstrated how a small number of accounts displaced a trending topic and attempted to derail online discussion surrounding #GobiernoEspía. The investigation was presented by Carlos Páez, director of Mesura, live on-air during Aristegui Noticias’s weekly segment #BigData on June 23.
The following is a translation of their findings.
The size of the nodes (accounts) depend on the interactions each received. Color is determined by the community each node belongs to.
From June 19–20, Signa_Lab captured 500,000 tweets from #GobiernoEspía tweeted by 27,340 accounts. This was a huge, organic conversation taking place while news was breaking about the government spying scandal. The main hashtag #GobiernoEspía trended #1 in Mexico all day on June 19 and also trended worldwide for awhile, then it was interrupted.
#GobieroEspía is a Trending Topic (TT). Signa_Lab is watching and already capturing data for analysis. Here is a graph of more than 10,000 tweets from the last hour and a wordcloud.
Here is the main network that was formed by people tweeting about #GobieroEspía on June 19:
Around 5pm on June 19, three hashtags started trending: #MeChocaQue, #TúTanFresaY and #ANalgadasPuedo. These three “toxic hashtags” were driven by 768 accounts and began moving up the trends bar.
The panel above shows a few sample tweets of the accounts that tweeted multiple hashtags with #GobieroEspía. This technique of spamming hashtags has been documented previously in Mexican Twitter where “undesirable” hashtags are targeted and knocked off the trends bar and then replaced with spammy garbage hashtags.
Now #GobieroEspía is attacked by bots to bring down the Trending Topic, and we show the powerful interaction of users to not dilute the conversation.
This hashtag attack was different, because not only did bots and trolls ride all four hashtags, but they also mentioned 7,857 accounts who were actively tweeting #GobieroEspía. This is new technique that effectively inserted the three garbage hashtags — along with the accounts that were tweeting them — directly into the center of the main network.
Below you can see how the network of 768 accounts, shown in red, wove itself into the gray, organic network:
768 users mentioned 7,857 accounts out of a total 27,340 users
Out of 27,340 users, only 768 accounts or 2.89% of the network, hijacked the #GobiernoEspía hashtag causing it to descend on the trends bar, eventually kicking #GobiernoEspía out of the trending topics entirely.
Operation #Overthrow #GobiernoEspía was successfull; with 3 hashtags (HT) from 5 to 9pm, they managed to displace the hashtag expressing indignation and protest.
The following is a translation of the hashtag analysis by Dr. Rossana Reguillo and the team from Signa_Lab and presented by Carlos Páez:
Over several hours on June 19 the hashtag #GobieroEspía remained a trending topic reaching first place on the trends bar around 1 pm, when a press conference was convened during which the report that proved the use of Pegasus spyware against journalists, activists and human rights defenders was presented. At 5 pm, accounts that usually call for a hashtag to deflate one or more hashtags which are critical, contrary to the system or discuss sensitive and interesting topics for citizens, called for pushing 3 hashtags: #MeChocaQue, #TúTanFresaY and #ANalgadasPuedo.
Until 7 pm, hashtag #GobieroEspía was ranked first in the trending topics at the national level and also gained traction worldwide. The three mentioned hashtags continued climbing position on the trends bar until #TúTanFresaY placed second, just below #GobieroEspía, #ANalgadasPuedo reached the third position and #MeChocaQue was in the fourth position on the trends bar.
The Signa_Lab team followed in real time, the speed with which just a few accounts pushed the three hashtags in such a way that by 8 pm, #GobieroEspía had descended to the seventh place in the trending topics while the other three hashtags were the top three trends on Twitter. By 9 pm, #GobieroEspía had been diluted and even though Twitter users continued to use the hashtag #GobieroEspía, they were no longer visible on the trends bar.
Signa_Lab isolated the red network that invaded #GobieroEspía; again this cluster of toxic tweeters was very small in comparison to the full network. But even though the accounts that tweeted #MeChocaQue, #TúTanFresaY and #ANalgadasPuedo accounted for only 2.89% of the total network, they still had a huge impact.
The network of hashtags associated with #GobieroEspía shows how the three toxic hashtags connected to #GobieroEspía.
Carmen Aristegui points out that besides disrupting and polluting an organic conversation, this tactic of displacing hashtags from the trends bar has very damaging effects:
That matters because when we identify issues that society presumably cares about, it displaces these important issues from ranking on Twitter and it has an important effect because being in “X” position then shifting to position “Y” or disappearing from the trends — that value of an issue of public interest being — or not being in the trends is no small thing.
Carlos and Carmen close out the segment with the same questions Mexican investigators have been asking for years: Who is responsible for this manipulation of social media and are public resources being used to pay for it? They also wonder why can’t Twitter detect this is happening?
To some people, keeping an important hashtag trending and visible on the trends bar might seem like a trivial thing. However in a country where 10 journalists have been murdered in the past 8 months and self-censorship is a survival tactic, bots and trolls manipulating a key platform that citizens use to read and discuss news has far-reaching consequences.
Here is the full 15 minute #BigData segment on Aristegui Noticias. You can select auto-translate in the settings for subtitles in other languages but the machine translation isn’t perfect.
Finally I just want to point out that people in Mexico are so familiar with these tactics of social media manipulation that the subject warrants a 15 minute segment on the news to discuss what happens to important hashtags like #GobieroEspía. There is a level of public awareness and an effort to educate social media users in Mexico that I don’t see happening yet in USA.
Related:
