Trajectories of YouTube consumption based on content exposure over time

The study of content consumption on YouTube and its potential influence on people’s interests is an established research venue. Researchers have frequently considered the political domain as an area of interest. They have oftentimes sought to quantify (I) how users can be mapped in the political landscape and (2) how users take or do not take paths toward radical content. However, these studies have two main fallacies. First, researchers have defaulted to simplified notions of the political landscape. To our knowledge, all studies have considered users on a highly simplified one-dimensional “left-right-scale”. Second, the quantification of their position on that scale is based on higher-level data, such as political leanings of channels, lists, or video titles, rather than the actual content presented in the videos. Unsurprisingly, past studies have recorded varying results and come to opposing positions. We propose to consider the problem by mapping content and users in a multi­dimensional space. We seek to map them based on more granular information about their consumption, in particular, the signals embedded in videos’ scripts. This rigorous approach should aid in clarifying the existing landscape of findings.

MIT Institute for Data, Systems, and Society
Massachusetts Institute of Technology
77 Massachusetts Avenue
Cambridge, MA 02139-4307