HIST 390

This article about the complexity of Pandora really fascinates me. Firstly, because I have never once used Pandora once in my life, I am more accustomed to Google Play, Soundcloud, and using Youtube Red to listen to my music, that or just downloading from links I find on subreddits of bands that I follow. I always forget that since I use other forms of media to listen to my music of how big Pandora still is, the same goes for my view on Spotify, another program which I have never used. Secondly, because of how this article is describing Pandora’s algorithm. This article somewhat describes Pandora’s algorithm as out of date and one that needs improving, which I both agree and disagree. For example, which taking about the rating system is done my actual people and not an AI algorithmic system that would automatically filter songs, I agree that Pandora should stick with actual people. I say this in comparison to how YouTube as been handling its rating system. YouTube uses an AI system to rate videos, especially all the VEVO music accounts to show in a viewers recommendation bar. This system backfired a little when popularity = good for the AI system because it then flooded viewers recommendation tabs, which in turn sparked out cry from some of the viewers. By having a human rating things, taking out their own biases into the ratings, popularity doesn’t always mean that the song would be good and flooded on a front page of the app. The part of their algorithm I do agree with is the idea of having a global chart for musicians using the app to see where the majority of their fans are that way in turn the artist could pander to that audience for more revenue which Pandora could charge the artist for using the map which could generate revenue for the app (even thought the artist could probably just use another app to see where his/her majority following would be at).

No Comments »
Filed under: Uncategorized