On Wednesday, Twitter released a collection of more than 10 million tweets related to thousands of accounts affiliated with Russia’s Internet Research Agency propaganda organization, as well as hundreds more troll accounts, including many based in Iran.
The data, analyzed and released in a report by The Atlantic Council’s Digital Forensic Research Lab, are made up of 3,841 accounts affiliated with the Russia-based Internet Research Agency, 770 other accounts potentially based in Iran as well as 10 million tweets and more than 2 million images, videos and other media.
Russian trolls targeting U.S. politics took on personas from both the left and the right. Their primary goal appears to have been to sow discord, rather than promote any particular side, presumably with a goal of weakening the United States, the report said.
DFRlab says the Russian trolls were often effective, drawing tens of thousands of retweets on certain posts including from celebrity commentators like conservative Ann Coulter.
Some of the tweets posted:
“Judgement Day is here. Please vote #TrumpPence16 to save our great nation from destruction! #draintheswamp #TrumpForPresident,” said a fake Election Day tweet in 2016.
“Daily reminder: Trump still hasn’t imposed sanctions on Russia that were passed 4,193 in the House and 982 in the Senate. Shouldn’t that be grounds for impeachment?” said another tweet in March of this year.
Multiple goals
The Russian operation had multiple goals, including interfering in the U.S. presidential election, polarizing online communities, and weakening trust in American institutions, according to the DFRLab.
“The thing to understand is that the Russians were equal opportunity partisans,” Graham Brookie, one of the researchers behind the analysis, told VOA News. “There was a very specific focus on specific ideological communities and specific demographics.”
Following an initial push to prevent Hillary Clinton from being elected in 2016, the analysis identified a “second wave” of fake accounts, many of which were focused on infiltrating anti-Trump groups, especially those identified with the “Resistance” movement, exploiting sensitive issues such as race relations and gun violence. These often achieved greater impact than their conservative counterparts.
“Don’t ever tell me kneeling for the flag is disrespectful to our troops when Trump calls a sitting Senator “Pocahontas” in front of Native American war heroes,” tweeted an account posing as an African-American woman named “Luisa Haynes” under the handle @wokeluisa in November 2017. The tweet garnered more than 32,000 retweets and over 89,000 likes.
“They tried to inflame everybody, regardless of race, creed, politics or sexual orientation,” the Lab noted in its analysis. “On many occasions, they pushed both sides of divisive issues.”
Iran trolling
Iran’s trolling was primarily focused on promoting its own interests, including attacking regional rivals like Israel and Saudi Arabia.
However, Iran’s trolling was less effective than the Russian posts, with most tweets getting limited responses.
This was partially because of posting styles that were less inflammatory, according to the report.
“Few of the accounts showed distinctive personalities: They largely shared online articles,” according to the report. “As such, they were a poor fit for Twitter, where personal comment tends to resonate more strongly than website shares.” Generally, many troll posts were ineffective, and “their operations were washed away in the firehose of Twitter.”
All of the accounts linked to the massive trove of tweets released by Twitter have been suspended or deleted, and the analysis notes that overall activity from suspected Russian trolls fell this year after Twitter clampdowns in September and June 2017.
But, that does not mean political trolls do not still pose a threat.
“Identifying future foreign influence operations, and reducing their impact, will demand awareness and resilience from the activist communities targeted, not just the platforms and the open source community,” according to the report.
…