Kevin Munger & Joseph Phillips, A Supply and Demand Framework for YouTube Politics, Preprint, October 1, 2019.
Abstract: Youtube is the most used social network in the United States. However, for a combination of sociological and technical reasons, there exist little quantitative social science research on the political content on Youtube, in spite of widespread concern about the growth of extremist YouTube content. An emerging journal- istic consensus theorizes the central role played by the video “recommendation engine,” but we believe that this is premature. Instead, we propose the “Supply and Demand” framework for analyzing politics on YouTube. We discuss a number of novel technological affordances of YouTube as a platform and as a collection of videos, and how each might drive supply of or demand for extreme content. We then provide large-scale longitudinal descriptive information about the supply of and demand for alternative political content on YouTube. We demonstrate that viewership of far-right videos peaked in 2017.
Tyler Cowen's take (I've not yet read it myself):
From a new and very important paper by Kevin Munger and Joseph Phillips from Penn State:The most extreme branches of the AIN (the Alt-Right and Alt-Lite) have been in decline since mid-2017.However, the Alt-Right’s remaining audience is more engaged than any other audience, in terms of likes and comments per view on their videos.The bulk of the growth in terms of both video production and viewership over the past two years has come from the entry of mainstream conservatives into the YouTube marketplace.…despite considerable energy, Ribeiro et al. (2019) fail to demonstrate that the algorithm has a noteworthy effect on the audience for Alt-Right content. A random walk algorithm beginning at an Alt-Lite video and taking 5 steps randomly selecting one of the ten recommended videos will only be recommended a video from the Alt-Right approximately one out every 1,700 trips. For a random walker beginning at a “control” video from the mainstream media, the probability is so small that it is difficult to see on the graph, but it is certainly no more common than one out of every 10,000 trips.That authors suggest (p.24) that if anything the data suggest deradicalization as a more plausible baseline hypothesis.Of course this is not the final word, but in the meantime so much of what you are reading about YouTube would appear to be wrong or at least off-base.