Recommender systems (RS) are machine learning techniques which take user data and predict endorsements of similar output. A couple common examples include content-based filtering and collaborative filtering. Spotify uses these algorithms for playlists and “You Might Like…” audio suggestions. 

What is not so transparent is how these RS work. Research and literature review indicates defects, such as bias and generalization, in the algorithms. Because of this, not every user is provided an authentic recommendation experience. Part of the reason is Spotify’s business model which focuses on advertising and distribution. The larger issue is understanding the “black box” of those processes. 

Public Spotify data was collected and analyzed to comprehend the filtering mechanisms. An attempt to cross-examine the RS methods and determine the accuracies was done. The results were inconclusive though the belief in the flaws of Spotify’s system remains.