Social bots (software-controlled profiles on social media) play a disproportionate role in spreading articles from low credibility sources on Twitter, according to a study conducted around the 2016 US presidential campaign. The findings are reported in Nature Communications, and the authors suggest that curbing social bots could limit the spread of online misinformation.
Many assume that social bots contribute heavily to the spread and visibility on social media of articles from low-credibility sources and demand that the platforms crack down on them, but the evidence has so far been anecdotal.
Filippo Menczer and colleagues analysed 14 million tweets that shared 400,000 articles on Twitter during and following the 2016 US presidential campaign (mid-May 2016 to March 2017). The authors found that social bots often amplify articles from low credibility sources (websites that routinely publish various types of misinformation and are identified by reputable third-party news and fact-checking organizations) immediately after they are posted and before they go viral. Bots also target influential users who have many followers by using replies and mentions. The authors suggest such strategies succeed because humans are vulnerable to this manipulation and re-share some of the content posted by social bots. The authors' analysis also shows that disconnecting a small percentage of accounts (about 10%) that are most likely bots could have virtually eliminated the spread of links to low-credibility content during the studied time period.