Algorithmic culture and filter bubble: the case of YouTube’s recommendation system

Abstract

Day by day, algorithms of all kinds are becoming part of our daily routines and help us to improve our daily lives. The recommendation systems of many platforms we daily use, are using different algorithms in order to function properly. A strand of the relevant research is currently exploring the impact of algorithms on the identity of users and culture more generally, which leads to the notion of “algorithmic culture”. Pariser (2011) first wrote about the phenomenon of the “filter bubble” and how they are being created, both by users and algorithms. Yet, it still remains controversial today if the phenomenon is real, and has split the academic community, as many tried to prove that such a phenomenon is not created by the algorithms. The YouTube platform has one of the most widely used recommendation systems and will be the focus of analysis for this thesis. The thesis examines the platform of YouTube for the existence of a commercial filter bubble in the case of music culture and ponders its impact on identity, or in this case, the music taste of users. Following the method of algorithm auditing, two fake accounts were created and loaded with two different types of music content in order to impersonate two different types of user with different music taste. By analyzing the recommended videos of the two accounts we show how different kinds of bubbles emerged through the recommendation of the platform, and how the platform of YouTube can faction as technology of the self.