Research & Publications

On both sides of the Atlantic, governments, foundations, and companies are looking at how to solve the problem of online dis/misinformation. Some emphasize the demand side of the problem, believing it important to focus on consumer behavior and the use of media literacy and fact-checking. Some focus on legal remedies such as platform-liability and hate-speech laws as well as privacy protections. Meanwhile, others try to raise the quality of journalism and support local news in the hope that creating more reliable content will be a counterweight to the dis/misinformation found online.

In short, there are myriad solutions aimed at addressing the problem of online dis/misinformation. This study looks at one kind of fix: the small companies in the information ecosystem that use natural language processing as well as human intelligence to identify and, in some cases, block false or inflammatory content online. This paper examines thirteen such companies, most of which are building solutions to identify false information online through a combination of people and natural language processing. Nascent and not yet widespread, these businesses are seeking to find new commercial applications for their products and, in some cases, hoping to entice the social media platforms to buy them out.

The gender dynamics of political discussion are important. These dynamics shape who shares their political views and how they share their views and reactions to these views. Using representative survey data from the United States and the UK, we investigate how social media platforms shape the gender dynamics of political posting. We also examine the concept of “mansplaining”—a term used to describe a patronizing form of communication directed at women by men.We argue that the possibility of being mansplained affects who is willing to post their opinions online, and as such, caution should be exercised when using digital trace data to represent public opinion.