Information operations in Bolivia
Several researchers from around the world have published analysis in Spanish language regarding social media manipulation surrounding topics and trends supporting the recent military-civilian coup in Bolivia but their research has not been translated to English or widely reported in the US media. Here are some highlights.
On December 10, the CIDH (Comisión Interamericana de Derechos Humanos — Inter-American Commission on Human Rights) published preliminary findings after a visit to Bolivia between November 22 and 25, 2019. In their report, the CIDH acknowledged the existence of a campaign carried out social media using fake Twitter accounts to support the president of the Civic Committee of Santa Cruz, Luis Fernando Camacho and Jeanine Añez’s interim government.
These brand new accounts were very obvious and many people noted their activity early on (including myself on Nov 9), however claims about fake accounts were rebuked on Twitter with the rationalization that many Bolivians didn’t use Twitter and a large number of people recently joined the platform.
Bolivian social media stats
Facebook is the dominant social media platform in Bolivia. According to Statistica, Facebook accounted for 81.95 % of social media visits in 2019 as compared to Twitter which comes in below Facebook, Instagram, YouTube and Pinterest at just 2.82 %.
“With a continuously increasing number of users, as of January 2019, Bolivia had reached a total of 6.9 million monthly Facebook users, surpassing the audience of 6.5 million users per month met in 2018.” — Statistica
So it is true that not many Bolivian citizens use Twitter and I think it’s entirely possible that some Bolivians made new Twitter accounts in October and November to participate in the public discussion about the events in Bolivia.
Bolivia has a population of approximately 11 million people, so the question we should ask: Were the new accounts proportional to or representative of the total population of Bolivia?
Julián Macías Tovar, social media coordinator for Podemos in Spain analyzed 14 hashtags associated with the coup in Bolivia and found over 92,000 new accounts that were created within one month that supported the campaign against Evo Morales on social media. Julián published 29 pages of analysis in Spanish and also made his dataset available to the public (links google drive) so I downloaded his data and made some visualizations.
Timeline of new account creation activity
General elections were held in Bolivia on 20 October 2019 and were followed by 19 days of protest which culminated in the Bolivian military and police asking President Evo Morales to resign on November 10.
Julián’s dataset goes from October 18 to November 20. New accounts were created every day throughout the entire timespan. Varying totals from between 100 to 1000 accounts were created each day from October 18 to November 5th when account creation ramped up slightly to around 2000 accounts. In the 5 days from November 5 to November 10, between 2000 to 5000 accounts were created each day. November 11, 12, 13 and 14 saw the highest numbers of new accounts. Below is a graph of all 92K accounts, sorted by the day they were created.
There was a spike in account creation on November 11, one day after General Williams Kaliman Romero appeared on television surrounded by the high military command and “suggested” that Morales resign on November 10.
From the 25,000 accounts that were created on November 11, activity started to decrease but was still high. Around 11,000 accounts were created on November 12 followed by about 7,000 accounts on November 13 and just over 5,000 accounts on November 14. So 48,000 accounts — more than half of the total dataset of 92,000 accounts — were created in just four days.
Like I said above, there are probably real people who created new accounts in this dataset. Newsworthy events drive real people to create new accounts on social media platforms and participate in public debates. However I find it highly unlikely that all of the brand new accounts were real humans.
Luciano Galup, Argentine author of the book Big data & Politics: From stories to data. Persuasion in the era of social networks, tweeted on November 12 about several of the suspicious account he found that seemed inauthentic, like the accounts below:
So I don’t believe these are all real people located in or from Bolivia. Especially these accounts that were created simultaneously at the exact same second:
I don’t think all of the 92K accounts are necessarily fake, but to assume these are all real people and representative of Bolivian public opinion is naive.
What we can say with 100% certainty is the result of this account creation activity gives the optics that there is a crowd of Bolivians talking on Twitter with opinions about Bolivian politics; a digital mob that could easily appear like a spontaneous consensus of people when viewed from afar.
The Bolivian expat/US Army vet aka “Cyber Rambo”
Julian found one very interesting account that was programmatically retweeting hashtags supporting the opposition. He found that one single account had contributed more than 13,000 retweets to 14 hashtags supporting the removal of President Morales.
A programmer and US Army veteran named Luis Suarez automated his account to retweet tweets using a custom app named “tfb-suarez.”
Luis Suarez tweeted his Bolivian ID to prove he is real and Bolivian and he also posted on Facebook about his excessive tweeting, joking that he’s apparently “Cyber Rambo” now.
Luis also apparently gave an interview to Spanish media El Diario and he posted the entire conversation publicly on Facebook. They asked him what is his current status with the US Army which he declined to answer but he did tell them he was born in Bolivia. Screenshots of the rest of his conversation are in the comments of the below Facebook post.
According to Luis’s LinkedIn, he received a Bachelor’s degree in Civil Engineering from Universidad Católica Boliviana in La Paz which he attended from 2000 to 2004. His profile says he was in the US Army in Fort Hood, TX from October 2010 to October 2016.
So sometime between 2004 and 2010 Luis emigrated to the US from Bolivia and joined the US army. His LinkedIn, Github and Codepen are all linked on his website which is posted below.
Luis’s automated retweets add an unexpected dynamic to the social media activity supporting protests against Evo Morales and his participation in the digital protests raises some interesting questions.
Luis is a Bolivian citizen living abroad with opinions about politics in his home country, which he is entitled to raise on his social media accounts. But is he entitled to automate his account in order to amplify political hashtags? Does his automation activity violate Twitter’s TOS?
If his LinkedIn is correct, then he is no longer enlisted in the US Army, but consider the optics of a man who represents his identity online as associated with the US army and is actively supporting the ouster of a foreign government. I can see how his activity could appear like a member of the US military participating in an influence operation targeting a foreign government.
The appearance of Luis Suarez’s noisy retweet bot in Bolivian political hashtags is a good example of the bizarre and totally random things that exist in Twitter networks.
There were several rumors and Fake News™ that circulated during November so I’ll just address 2 that I think got a lot of attention and were not properly debunked at the time.
Debunk #1: No the CIA did not make a hashtag trend in VA
The below tweet with a screenshot containing the trend #BoliviaNoHayGolpe (translation: there is no coup in Bolivia) claimed it was trending in “VA” and this went semi-viral so some people speculated that it was trending in the state of Virginia because that’s where CIA headquarters are located.
I checked on 3 different devices (mobile, desktop & tablet) and I was unable to view trends at a state level in USA. The only location available in Virginia is Virginia Beach and “VA” is not even an option.
I guess it’s possible “VA” is an option for Twitter trends on some other Twitter application, and if you are able to look up trends at a state level via the state’s abbreviation please contact me so I can update this section, but I do not believe the CIA had anything to do with trending Bolivian hashtags.
Debunk #2: The fake photo of Evo Morales, Chapo Guzman and Pablo Escobar
Henrique Salas Römer, Venezuelan politician, former Governor of Carabobo state and president of the centre-right party Project Venezuela, has a verified Twitter account with 190K followers. On November 6, he tweeted a black and white image that appeared to show Evo Morales sitting at a table in between Chapo Guzman and Pablo Escobar. Even though this was a pretty obvious photoshop, Römer’s tweet is still up and currently has 9,804 Retweets and 9,627 Likes.
I really don’t know what else to say about verified politicians tweeting disinformation and refusing to delete and tweet corrections but it’s a problem that is not going away.
Guatemalan journalist Luis Assardo made a mega-thread of additional analysis that was done by various other researchers. Here are some quick links to more research in Spanish language:
Analysis by Alberto Escorcia of LoQueSigue_from Mexico:
#BoliviaNoHayGolpe: Hay un esfuerzo con bots de crear una narrativa sobre que no hubo golpe de…
En LOQUESIGUE lo detectamos al encontrar decenas de menciones a medios de comunicación usando la frase en común…
Reporting in El Diario (Spain) by Carlos del Castillo
Una campaña coordinada con miles de nuevas cuentas de Twitter y bulos contra Morales lava la imagen…
La salida de Evo Morales de Bolivia tras el golpe de Estado coincide con la creación de decenas de miles de cuentas de…
More info on Luis Suarez’s Twitter activity from data scientist Rubén Rodríguez Casañ from Spain:
Breakdown of Luis Fernando Camacho’s November influx of fake followers by Javier Barriuso from Podemos’s social media team in Castilla-La Mancha, Spain:
More analysis of Camacho’s fake followers by Rodrigo Quiroga, researcher with CONICET from Argentina:
Extensive analysis of fakery in Bolivian social media was published by experts in several countries however it did not seem to break through to the US media. Why not?