Precise Feedback for a Precise Algorithm: Improving the User Experience on YouTube

Amid growing concern over filter bubbles and content diversity, this thesis explores the impact of feedback mechanisms on user experience with YouTube’s recommendation algorithm. The study examines how increased user control can influence their interactions with the algorithm. Based on user interviews, personas were created to understand user behaviors and expectations. A Chrome extension was developed to allow users to report errors in their recommendation feed. Results indicate that this mechanism enhances user satisfaction and a sense of control, though some limitations suggest areas for future improvements. The study also proposes a methodology to evaluate contextual thematic diversity on YouTube, paving the way for further research into recommendation diversity and self-actualization systems.