Bots spread low-credibility content online
This study has shown that fake accounts (bots) on Twitter are responsible for spreading significant misinformation online, especially during the 2016 US Presidential Election. Specifically, the 6% of Twitter accounts responsible for sharing 31% of all low-credibility messages and articles were bots. This is alarmingly disproportional and was found from analysis of 14M messages and 400K articles on Twitter. The bots reportedly spread content at a high frequency within the first 10 seconds after news is published, making it more difficult to determine credibility of this content. Because bots share each others’ content en masse, their messages easily gain traction and appear legitimate. Moreover, these bots cooperate with each other to trick legitimate users in also spreading their false content.
Online bots form dense social networks with many “strong ties” amongst each other. These bots may all be following each other or interacting with each other. If the interactions in the latter case are retweets, this can be represented as a retweet network. A retweet network is a graph of accounts as nodes and edges directed from one node to another is weighted by the number of times that account retweets the other. Here, retweeting activity can be measured for centrality by their in- and out- degrees. The study shows real examples of large-scale retweet networks and how single nodes (like master bots) are central and responsible for information that is spread virally by other bots. The study goes as far as removing the top 10% of most likely bots from a simulated network and observe a large drop in the false content that is spread.
Study: https://www.nature.com/articles/s41467-018-06930-7
https://news.iu.edu/stories/2018/11/iub/releases/20-twitter-bots-election-misinformation.html