How Deceptive Online Networks Reached Millions in the US 2020 elections

Background

Social media platforms have become central spaces for political communication, especially during elections. Alongside ordinary political expression, however, platforms also host coordinated efforts that mislead users about who is behind certain accounts, Pages, or posts. These deceptive online networks may use fake identities, seemingly unrelated accounts, or artificial engagement to make coordinated activity appear authentic. Even when the content itself is not false, this kind of identity deception can distort public discourse by making political messages seem as though they come from ordinary users rather than organized networks.

Most research on deceptive online activity has focused on politically motivated influence operations, especially foreign campaigns such as Russian activity around the 2016 U.S. elections. This focus leaves out networks that use political content for financial gain, such as clickbait or spam operations that exploit election-related issues to attract attention and traffic. Previous research has also often relied on indirect measures of exposure, such as whether users followed or mentioned network accounts, because direct data on who actually saw network content is typically available only to platforms. This study addresses these gaps by examining networks with both political and financial goals and using platform-level data to measure actual exposure to their content during the 2020 U.S. elections.

Study

The study uses data from the U.S. 2020 Facebook & Instagram Election Study to examine the characteristics, activity, and reach of deceptive online networks during the 2020 election period. The analysis focuses on 49 networks identified by Meta that targeted U.S. users between June 26, 2020 and February 15, 2021, including 13 coordinated inauthentic behavior networks and 36 financially motivated operations. By including both politically motivated and financially motivated networks, the study examines deceptive online networks as a broader category of coordinated activity that uses identity deception while engaging in political discourse.

The authors analyze aggregated platform data from U.S.-based adult active users on Facebook and Instagram, allowing them to measure actual exposure to network content rather than relying on indirect indicators such as follows, mentions, or engagement. They distinguish between direct exposure, when users saw content posted by network-affiliated accounts, and indirect exposure, when users saw network content reshared by accounts unaffiliated with the network. The study also uses individual-level data from consenting survey respondents to examine the characteristics of users exposed to network content and to assess whether exposure or engagement is associated with political knowledge, attitudes, or behavior, while cautioning against causal interpretations of those relationships.

Results

The study finds that deceptive online networks reached a large number of U.S. users during the 2020 election period, but that reach was highly concentrated. Across the 49 networks studied, network content reached at least 36.79 million unique U.S. adult active users on Facebook and 2.98 million on Instagram, representing 14.63% and 1.85% of each platform’s adult active U.S. users, respectively. When users reached solely through network ads are included, total reach rises to 37.33 million Facebook users and 2.98 million Instagram users. However, only a few networks accounted for most exposure: on Facebook, three networks reached nearly 80% of all unique viewers of deceptive network content, and these same three networks accounted for more than 70% of all users exposed across both platforms.

The results also show that unaffiliated accounts played a major role in amplifying network content, especially on Facebook. Users could be exposed directly through posts from network-affiliated accounts or indirectly when non-network accounts reshared network content. For the highest-reach Facebook networks, indirect exposure through unaffiliated accounts far exceeded direct exposure. In the case of Rally Forge, the highest-reach coordinated inauthentic behavior network, 1.3 million users were reached directly, while 13 million were reached indirectly through reshares by non-network accounts. Only a small share of exposed users reshared network content, but their activity substantially increased overall reach.

The study finds that users exposed to network content had similar characteristics across network type and exposure pathway. On Facebook, those exposed were more likely to be older, conservative, highly active on the platform, and among the top 20% of users exposed to content from untrustworthy sources. The authors also examine whether exposure to or engagement with two financially motivated networks was associated with political outcomes, including factual discernment, belief in the legitimacy of the 2020 election, and partisan news clicks. After adjusting for pre-exposure user characteristics, the initially strong associations largely disappeared, and the authors caution that these results should not be interpreted causally. Taken together, the findings suggest that deceptive online networks can reach broad audiences, but that reach in this study was driven heavily by a small number of networks and by resharing from users outside the networks themselves.