Tracking and Characterizing the Competition of Fact Checking and Misinformation: Case Studies

Massive amounts of misinformation have been spreading over social media during the 2016 U.S. election season, causing wide public concern about our information ecosystem. How does fact checking compete with misinformation for user attentions? How do misinformation and fact checking differ in their spreading patterns? And what strategies are used by social bots to promote the spread of misinformation? In this paper, we address these research questions by analyzing datasets collected by Hoaxy. We answer the first question by conducting: 1) a survival analysis shows that about 70% of claims will be fact-checked in one week and about 800 tweets with claim links are posted during this period, on average and 2) a cross-correlation analysis shows that the sharing of fact-checking articles typically lags that of misinformation by about one day. Regarding the second question, we answer it from three points of view: 1) we find that false claims can become more popular than the corresponding debunking; 2) when looking at the distribution of types of tweets, we find that fact checking tends to spread in a conversational way; and 3) bot behavior analysis shows that the most active accounts sharing misinformation behave more like social bots. To better understand how social bots spread misinformation, we conduct case studies to answer the third question. By presenting strategies such as the production of a large number of original tweets, the alternating and hijacking of hashtags, and the injection of content into conversations, we demonstrate how social bots take advantages of the recommendation features of Twitter to amplify the spread of misinformation.