RT on Facebook during the 2020 US Election

RT on Facebook during the 2020 US Election

Client
Completed
Project title
RT on Facebook during the 2020 US Election
Disciplines
Social MediaMisinformation
image

Gap
Gap
There is an ambiguous relationship between government-sponsored media, like RT (formerly, Russia Today), and Facebook algorithmic curation, which may decrease exposure of users to propaganda and at the same time increase confusion and misinformation.

The challenge

Following the Cambridge Analytica scandal and the Mueller report, Russian digital interference in the affairs of democratic states has been driving the debate on digital propaganda. While the existing studies tend to focus just on some forms of such interference, namely deliberate disinformation and fake news, several studies have been raising concerns about the possibility of algorithmic curation systems increasing exposure to Russian propaganda and proposing ways to address it. However, there is still limited understanding of the complex relationship between platform-specific information curation models and the activities of government-sponsored media, in particular the extent to which pro-Kremlin outlets are able to manipulate the Western public on social media.

Approach

In this research, in collaboration with Mykola Makhortykh from Bern University, we seek to move a step closer towards understanding this relationship by asking two research questions: (1) What content did pro-Kremlin media promote on Facebook in the run-up to the 2020 US Presidential Election? (2) How did the Facebook algorithms affect the distribution of this content?

To answer these questions, we explore activities of what is known as the key actor of Russian propaganda abroad—its 24-hour news broadcaster RT (formerly, Russia Today) —on Facebook.

We use a mixed-method approach and combine algorithmic auditing methodology with frame analysis. We use virtual agents to understand how users of Facebook that follow RT are exposed to its information in Facebook’s news feed and search results.

Outcome

RT's election-relevant strategic framing on Facebook included promotion of content that was anti-Biden leaning with an emphasis on anti-establishment narratives. However, due to algorithmic factors, such as randomisation and localisation of content curation, individual agents were exposed to eclectic RT content without an overarching framing.

We found that only a fraction of the content published by RT throughout the period of our study appeared in our agents’ feeds.
We found that only a fraction of the content published by RT throughout the period of our study appeared in our agents’ feeds.

We ran a Jaccard index and found that our agents had been exposed to very different content with almost no overlap.
We ran a Jaccard index and found that our agents had been exposed to very different content with almost no overlap.
Despite extensive engagement with RT pages, agents were also exposed to  the content of other information sources (e.g. legacy media) as part of Facebook’s algorithmic curation that prioritizes reputable news sources. However, this exposure differed substantially between agents located in different states.
Despite extensive engagement with RT pages, agents were also exposed to the content of other information sources (e.g. legacy media) as part of Facebook’s algorithmic curation that prioritizes reputable news sources. However, this exposure differed substantially between agents located in different states.

Affiliation

image