Newswise — ITHACA, N.Y. – In a finding that has implications for the 2022 midterm elections, Cornell University researchers found Russia tried to distract liberal voters during the 2016 presidential campaign with a seemingly innocent weapon – tweets about music and videos.
The strategy resembles techniques used by autocratic governments that control their national media, such as Russia and China, which “flood” social media with entertainment content to distract their citizens from domestic events like protests that they don’t want covered.
“We’ve seen flooding as a social media strategy in both China and Russia; for example, in the past decade Russia has frequently manipulated social media with respect to Ukraine,” said Alexandra Cirone, an assistant professor of government. “But an autocratic country, trying to do it to a democratic country in the midst of an election – we’re in new territory. It’s surprising how well these strategies transplant.”
Cirone and Will Hobbs, assistant professor of psychology and government, co-authored the paper, “Asymmetric Flooding as a Tool for Foreign Influence on Social Media”.
Previous research has shown that the Kremlin-affiliated Internet Research Agency (IRA) used thousands of troll accounts on Twitter and other social media platforms to overwhelmingly support former President Donald Trump’s campaign over the campaign of Hillary Clinton, in an attempt to amplify social divisions between conservatives and liberals. That work has focused on partisan messaging.
But in 2018, Twitter released an important dataset of more than 10 million tweets sent by 3,841 IRA accounts. These accounts represent human-controlled Russian operators, or “trolls,” as opposed to computer-controlled accounts, or “bots.” The dataset included apolitical content, which hasn’t been studied as much as the partisan messaging.
“In China, government-affiliated users might flood social media feeds with Chinese history or inspirational quotes,” Hobbs said. “So, what might Russian trolls use in the U.S.? Entertainment content popped out of the automated analysis method.”
Key words in those tweets included “hiphop,” “remix,” “rapstationradio,” “nowplaying” and “indieradioplay.”
By constructing a timeline of the IRA messages, they found left-leaning IRA trolls posted large volumes of entertainment content in their artificial liberal community and shifted away from political content late in the campaign. Simultaneously, conservative trolls were targeting their community with increases in political content. The effort would have encouraged the right to vote and the left to ignore politics.
Hobbs notes that the mechanism doesn’t mean a Twitter user sees a music tweet and automatically stops thinking about politics. Rather, if a troll suddenly posts a significant amount of music tweets to a feed, it displaces other content.
“If someone posts a ton of entertainment content, and your feed only shows 10 posts at a time, then having eight posts that are now unrelated to politics pushes everything else down in the feed,” Hobbs said.
The findings underscore that entertainment content is not as innocuous as we might think.
“You might think you’re clicking on a cat meme” Cirone said, “but really you’re potentially putting a troll network into your feed that could later start posting divisive content or monetize their following.”
For additional information, see this Cornell Chronicle story.