On the Friday before the presidential election in France, documents purported to contain unverified information that was damaging to the Macron campaign were leaked online, just hours before a nation-wide moratorium on election coverage by the media was set to begin.
While the discussion was tweeted nearly 350 thousand times on social media, the disinformation campaign had limited effect on the voting public in France.
But why is this the case? In a new research paper “Disinformation and Social Bot Operations in the Run Up to the 2017 French Presidential Election,” Emilio Ferrara, Research Leader and Principal Investigator in the Machine Intelligence and Data Science (MINDS) group at the University of Southern California‘s Information Sciences Institute and Assistant Research Professor in the Department of Computer Science, finds new evidence that this disinformation campaign was limited in its scope and impact as the campaign itself was mostly shared by foreigners outside of France. In his study, Ferrara looks at the timeline of the election and leverages his previous experience in studying Twitter bots during elections to identify anomalies, patterns and profiles of those who shared and consumed, and revived links to content that was intended to undermine the Macron campaign in favor of Marie Le Pen. The paper will be appearing in August 2017 in First Monday.
Ferrara, who has a long history of distinguishing bots versus humans on social media, studied a dataset of almost 17 million posts from April 27, 2017 through May 7, 2017, France’s election day. Through a combination of machine learning and other modeling techniques, Ferrara was able to discern bots from humans with over 80 percent accuracy based on the characteristics of the Twitter accounts, including the frequency of account postings and the customization of the account profile. Ferrara then developed a profile of those who engaged in sharing the leaked and possibly false information about the Macron campaign. Eighteen percent of the MacronLeaks promoters, says Ferrara, were bots (compared to 15% of bots that were deployed in the US Presidential Campaign).
Ferrara says that accounts engaging in sharing and promoting “#MacronLeaks” and the affiliated files were mostly foreigners which, Ferrara notes, favored “alt-right” topics. MacronLeaks, he says, was tweeted about more often in English (177,695 tweets out of 350,000) than in French (135, 397 tweets).
Ferrara hypothesizes the existence of an underground market of reusable political disinformation bots.
Ferrara believes the fact that MacronLeaks was mostly shared among English speaking audiences could be a significant reason this disinformation campaign was not more detrimental to the Macron’s election.