According to a new study from the University of Southern California’s Information Sciences Institute, more than 15,000 Twitter bots used the #iVoted hashtag on midterm Election day in 2018. This means that out of every five people who posted about the elections on Twitter in the weeks before Election Day, at least one was a bot.
Thousands of these bots originated from Russia, Iran, and Venezuela, with some originating from the United States, as reported by CNBC News. The research also revealed that more than 200,000 bots posted about the Election Day about two weeks after the elections. In comparison, 750,000 humans were found to have posted, translating to about 20% of all tweets as automated.
How were these bots identified? The researchers used a sophisticated tool developed at the University of Indiana called the “Botometer” that uses “machine learning techniques” in order to “determine whether an account is operated by a human or by a software algorithm.” The tool had a 95% accuracy rate, according to one of the study’s four authors and principal investigator Emilio Ferrara.
Twitter also released its own findings last week that “focused mainly on foreign efforts”. Twitter also surprised the public by confirming the removal of 10,000 bots reportedly pretending to be Democrats days before the midterm elections.
Ian Plunkett, spokesman for Twitter, responded to the story, calling the company’s report as the “the single source of truth from our side on the election.”