On Monday, two new independent studies commissioned by the Senate Intelligence Committee were made public, providing the most in-depth look at online Russian interference in the 2016 election to date. The studies found that social media networks haven’t been totally forthcoming with lawmakers, and the propaganda efforts were decidedly designed to elect Donald Trump as president of the United States.
The two reports — one by Oxford University’s Computational Propaganda Project and Graphika, and one by cybersecurity company New Knowledge—confirmed much of what we’ve been told about the information warfare perpetrated by Russia’s Internet Research Agency (IRA).
The researchers found that the Russian operation used accounts on platforms that included Facebook, Instagram, Twitter, YouTube, Google+, Reddit, Tumblr, Pinterest, and Vine. According to the reports, the IRA troll factory’s efforts began in earnest during the 2016 Republican primary and promoted Trump’s campaign while disparaging his Republican rivals.
Black Americans were heavily targeted with messages to discourage voting or to vote for one of Hillary Clinton’s more liberal rivals. “Extensive amounts” of the data provided to the Intel Committee by the social networks and analysed by the researchers was previously unknown to the public, and it’s the first analysis of much of the organic content that spread during and after the election.
The authors of both reports were at times critical of the data submission methods used by social media companies and the ways their executives characterised the troll campaigns. The extent of activity on YouTube and Instagram, in particular, was previously unknown or downplayed. New Knowledge broke down the data it received as follows:
Approximately 10.4 million tweets (of which ~6 million were original) across 3841 Twitter accounts, with ~73 million engagements on original content.
Approximately 61,500 unique Facebook posts across 81 Pages, with ~77 million engagements.
Approximately 116,000 Instagram posts across 133 accounts, with ~187 million engagements.
Approximately 1,100 YouTube videos across 17 account channels, engagements/number of views was unlisted.
The reports describe a sophisticated intermingling of accounts that spread influential memes and amplified content created organically by other, non-IRA users. It’s nearly impossible to say where the line could be drawn between Russians imitating Americans and Americans drawing inspiration from Russians.
The Oxford report characterises the evolution of social media as having “gone from being the natural infrastructure for sharing collective grievances and coordinating civic engagement to being a computational tool for social control, manipulated by canny political consultants and available to politicians in democracies and dictatorships alike.”
They often focused on controversial fault lines in U.S. politics like the Black Lives Matter protests and the subsequent Blue Lives Matter counter-movement. Instagram, in particular, was singled out as being an effective platform to build an audience with minorities; the most successful account was identified as @blackstagram, which had 303,663 followers.
The goal of targeting African Americans was to amplify messages that liberal candidates like Senator Bernie Sanders or Jill Stein were better choices than Hillary Clinton — or to encourage them not to vote at all.
On the other side of the partisan divide, Russian operation would focus on issues like Confederate history or veteran’s issues when targeting older groups. When targeting younger conservatives, they would focus on “meme and ‘red pill’ culture.”
Sometimes these efforts were surreal and absurd. One Instagram account, Army of Jesus, morphed from a Muppets Show fan page to focusing on The Simpsons to finally settling on a religious theme. At one point, the account’s counterpart on Facebook offered sexual addiction counseling with an ad that read “‘Struggling with addiction to masturbation? Reach out to me and we will beat it together’ - Jesus.”
It’s easy to dismiss images of Jesus in a MAGA hat or Clinton as Satan but one shouldn’t underestimate the power of constant reinforcement. While white evangelicals may be quite literal in their interpretations and representations of Christianity, they’re also the only religious group that supports Trump.
The researchers did not have direct data from other networks aside from the big four, but they traced some accounts and activities other platforms like Tumblr and Reddit as well as “widespread activity” on Vine, Gab, Meetup, VKontakte, and LiveJournal. Even games, browser extensions, and music apps were weaponised.
Pokémon Go was pulled into the mix when Tumblr was used to encourage players to name their Pokémon after victims of police violence.
The researchers were also critical of tech executives’ handling of the campaigns, their subsequent public reactions, and responses to questions from lawmakers. New Knowledge wrote:
Regrettably, it appears that the platforms may have misrepresented or evaded in some of their statements to Congress; one platform claimed that no specific groups were targeted (this is only true if speaking strictly of ads), while another dissembled about whether or not the Internet Research Agency created content to discourage voting (it did). It is unclear whether these answers were the result of faulty or lacking analysis, or a more deliberate evasion.
In written responses to the Senate in October, Facebook’s representatives said they believed the only people who could answer whether any of the activity was designed to suppress the vote were “investigators with access to classified intelligence and information from all relevant companies and industries.” New Knowledge extensively outlines voter suppression efforts that would be obvious to anyone.
In response to the reports, a spokesperson for Facebook told Gizmodo that it supplied “thousands of ads and pieces of content to the Senate Select Committee on Intelligence for review and shared information with the public about what we found.” They reiterated that they still believe “Congress and the intelligence community are best placed to use the information we and others provide to determine the political motivations of actors like the Internet Research Agency.”
The Oxford report also criticised the social network’s cooperation with the probe, saying that its lack of access to user comments meant the data set was incomplete and that in the future, there should be better coordination with outside researchers. But it called out Google specifical, saying that its “data contribution was by far the most limited in context and least comprehensive” out of Facebook, Twitter, and Google.
It further wrote that networks have a responsibility to protect users’ privacy, but “sharing data about public problems should be more than performative, it should be meaningful and constructive.”
A spokesperson for Twitter told us that its “singular focus is to improve the health of the public conversation on our platform, and protecting the integrity of elections is an important aspect of that mission.” They said that the company has “made significant strides since 2016 to counter manipulation of our service, including our release of additional data in October related to previously disclosed activities to enable further independent academic research and investigation.”