Skip to main content

The Ear vs. The Machine

How algorithms are defining our music tastes

By: Deedee Barrientos


Trends have this generation in a chokehold. What someone says, does, listens to or how they dance is spread in a matter of seconds and the next day, a million other people are dancing or dressing the same way. We see this everywhere: in videos, shows, movies, social platforms and music streaming services. 


But we don’t decide what trends on our own. Most media platforms use algorithms to determine what is trending. 


A Google insider, who requested anonymity on the grounds of job security, explained that music algorithms consist of recommendation and ranking systems. These systems create the trends page. 


In a technical manner, this problem has many solutions and is eventually broken down into obtaining information. If you consider it a two-sided network – the tracks produced by artists and the listening audience – this network sustains itself by meeting both sides’ needs. Artists meet their creative needs by sharing their music and the audience meets their consumer needs by accessing a music library.  



From this connected network, information like past engagement data, location and age are considered to decide what to recommend to listeners. By looking at what other listeners with similar tastes are streaming, the algorithm picks what has the most engagement and recommends it to different listeners. The algorithm also learns more about listeners based on how they react to those recommendations.  


Most people think when a song is trending everyone else is listening to it. While that may be partly true, trending pages are not as unitary as most believe. If something is trending, it doesn’t mean it’s universally trending, it just means it’s trending within a community of listeners with similar interests. This is why trending songs work on people: they’re personally catered. 


“In order to figure out which item to match to which user, we have essentially a retrieval aspect where we try to grab relevant items for this user,” the insider said. “Then in ranking, we essentially order the items in such a way that it would be in likelihood that this item would fulfill some objective for the user.” 


The worldwide trending section is more closely related to the public consensus of trends. Worldwide trend pages are made up of the most popular tracks or videos that people want to see. This is calculated by what has the most engagement. 


Because of how the algorithm works, it might seem impossible to escape trends. However, the algorithm will only recommend listener-specific trends by looking at their past listening history. So it’s really up to the listener to control the type of song recommendations they’ll see.  


By listening to various playlists and artists outside of your comfort zone, you can get a broader scope of recommendations. This will feed the algorithm more information about you as a listener and give you music recommendations that aren’t kept in a box of the same genre.

Leave a Reply