Human vs. Machine: Algorithm and Expertise

by Carolyn Heneghan

There’s no denying that music streaming is here to stay and is slowly but surely replacing mp3 downloads (2013 was the first year mp3 sales dropped below streaming sales). Now that streaming has become a popular platform, more companies are looking to capitalize on music listeners’ interest in it. As a result, two models have arisen as different ways to appease customers with personalized playlists and music discovery tactics.

Algorithm-based services, such as Pandora and Spotify, are now contending with expert curation services, such as Beats Music and Songza. Both are slightly similar but differ in ways that may sway users one way or the other. Will one type of playlist creation beat out the other?

How Do Algorithms Work?

Algorithms and expert curation have similarities in that they both involve picking songs that relate to one another in some way. But the difference is who or what decides which songs are related and how they are related to each other in various situations.

Pandora uses specific algorithms that present playlists based on songs’ similar musical traits. The basis of Pandora’s catalog, the Music Genome Project, consists of 400 musical attributes that cover everything from melody, harmony and rhythm to form, composition and lyrics.

Pandora’s catalog constantly grows as it signs on more artists and their songs, each of which take about 20 to 30 minutes to analyze and be added to the catalog. From there, users make a selection based on genres, artists or songs, and Pandora generates an ongoing playlist with songs that have related musical traits based on the Music Genome Project analysis.

Algorithms are sometimes shunned for not creating “satisfying enough” playlists that are too computer-based rather than curated with a real person behind the scenes. Musical trait matching can feel robotic after a time, and it may be a bit harder to discover new music that you might like, versus playlists that run more smoothly than a computer-generated ones.

What About Expert Curation?

Beats Music, on the other hand, has a team of curators, who also analyze songs based on various musical traits. But instead of entering them into a database that chooses the playlists’ songs for them, the curators are the ones who actually create the playlists and decide which songs will go with others.

Users have compared Beats Music to the days when music magazines like Rolling Stone, Pitchfork and alternative publications were a key part of music discovery, where actual humans were behind the selections published for your reading pleasure and for you to either buy or download the album.

Beats has a few different methods of helping users discover music. Just For You offers curated playlists and albums based on any genre you choose when you first click on one of the genre bubbles in the app. Highlights are selections directly from Beats Music staff. Find It is a browsing tool for Genres, Activities and Curators. And finally, a unique addition to Beats, The Sentence allows you to fill in a sentence with a location, mood, the people you’re with and a music genre to generate a customized playlist.

Beats Music has its own naysayers who feel that the service doesn’t have enough of a social component with information about friends’ selections and playlists, which for many users is a primary way that they discover new music. Algorithm-based services, such as Spotify, depend heavily on this type of discovery and devotes entire pages and sidebars to it.

But Is There Actually a Blend of the Two?

Where it gets a little cloudy is that Pandora actually does involve some professional curation at the start, and Beats does use its own tailored algorithm.

The Music Genome Project worked for 10 years to create the base from which Pandora began the algorithm that allows it to play different related songs. With that project, experts analyzed each song before filing it in Pandora’s massive catalog to then be chosen from by the algorithm it uses. So at least at the start, Pandora does employ a bit of expert curation to catalog the songs based on characteristics. But in terms of music discovery, the algorithm takes it from there.

Beats, on the other hand, does actually use an algorithm based on certain genre choices to create the playlists users listen to. But the difference between its algorithm and Pandora’s is that Beats’ is based on a playlist of songs linked together directly by a curator rather than a playlist generated by a catalog of songs organized by similar musical traits.

So Which Is Better?

Ultimately, deciding which is better is entirely subjective and could go either way depending on user experiences. They both have their advantages and drawbacks, and some might find that they enjoy the playlists on either one better than the other. Because both sides have free versions to try before committing to any type of subscription, users can try both to see which better suits their tastes.

So which version will win out in the end? Or will they remain on an even platform and split the user base? Only time will tell as Beats Music is bought by Apple and Songza is potentially bought by Google, which will give the two streaming services leverage and a larger talent and user base on which to grow.

Will yet another streaming model come into play sometime in the future that blows both of these models away? With the constant evolution of the music industry these days, you just never know.

Comments

comments