The newly appointed head of Radio 1 said the station will stop using big data to pick their playlists and trust presenters’ instincts instead. But what has big data done for music?
Last month, the new head of music at Radio 1 and 1Xtra, Christopher Price, announced in this interview that the station would no longer use ‘big data’ (YouTube views, Soundcloud streams, social media ‘likes’, etc.) to determine which songs should or should not make their playlists. Instead, Price wants to return to music curation based on the tastes and instincts of presenters. For Price, curation should be about ‘using your ears’, not reading numbers. But with music lovers living in a big data matrix, this may be more difficult than it sounds.
How big data rules the airwaves
The instinct-based approach is contrary to Radio 1’s current practice, as revealed in The Observer two years ago. Before Price’s appointment, the station’s playlist was selected by a committee of 12 producers sitting around a table listening to songs and, more importantly for them, reading out artist YouTube subscribers, Twitter followers, Spotify streams, Shazam ratings, and pretty much any other metric available. It was not quite as simple as picking the tracks with the most views, but the facts and figures were far more influential than the songs themselves.
In a way, resigning playlist curation to the figures that be lessened Radio 1’s influence as a taste-making institution. Presenters could stray from the playlists and promote the new music they liked, but at its core, Radio 1 was following trends, not setting them. Through basing its playlists on whatever was ‘trending’, Radio 1 became just another follower. Yes, the station may have increased the exposure of some new artists (Radio 1 is to thank for Jess Glynne’s ubiquitousness) but they chose those artists because people already liked them, and indeed ‘liked’ them.
How big data sets the trends
With Radio 1, the UK’s self-styled chief tastemaker, taking trends instead of making them, where do listeners turn to discover their new favourite song? There are two possible answers to this question. In the world of electronic music the DJ is king, and the right DJ playing the right track in the right club can create and overnight underground sensation. You will not find DJs checking YouTube hits mid-set. Even the more mainstream, crowd-pleasing DJs like David Puentez stress the importance of choosing interesting songs that fall outside the Top 40.
But what about when you’re not in the club, or if you’re listening to a genre oft avoided by DJs? How do you find new music then? The answer, whether you like it or not, is most likely big data. Load up your streaming service of choice and you will be presented with ‘Featured’ tracks and ‘Trending’ playlists. The ‘New Releases’ are not ordered by date, but by the platform’s curation staff, who have access to some of the largest and most specific listener activity databases in the music world. There is a reason Drake and Kyla’s ‘One Dance’ was at the top of Spotify’s ‘New Releases’ tab for weeks and it has nothing to do with it being a great tune. Streaming services encourage listeners to start with the charts or trends, and each of these is determined by mass accumulation of, yes, big data.
How big data rules discovery
Every Monday, millions of Spotify users awake to their ‘Discover Weekly’ playlist. These playlists consist of thirty songs that Spotify thinks the user will like that they probably have not heard before. Discover Weekly playlists are known for being uncannily good, and have been compared to mixtapes made by close friends. The playlists, of course, are curated by an algorithm that tracks your ‘taste profile’ against listeners with similar interests, based on Spotify’s database of 75 millions users. In simple terms, if most Marshall Jefferson fans listen to DJ Deeon but you don’t, you’ll be getting some Deeon in your playlist next Monday. This is another application of big data to music curation, even if it is more personalised and even if it does promote smaller artists.
Alongside ‘Discover Weekly’ there are countless other online discovery systems, and all of them use big data. Soundcloud’s Recommended tracks function, whereby similar tracks follow the track the user selected, is explained by its creators as such:
“Recommended tracks are selected by an algorithm that returns recommendations through a network of relations and interactions on SoundCloud (for example a user liked a track, a user followed another user, a track is reposted etc.)”
In other words: big data.
Similar algorithms are used for everything from Spotify Radio to Pandora and Last.fm.
Is big data a bad thing?
This is the ultimate question. Like most developments in music technology over recent years, there is no way to definitely state if big data is positive or negative; it just simply is. The way people hear new music has been utterly transformed by big data. Whether it helps an independent artist by recommending their little-played album to someone with the perfect ‘taste profile’ or whether it dilutes the radio into a homogeneous-but-’like’-able mush, big data and music are indisputably intertwined.
Even if Christopher Price succeeds in transforming Radio 1’s curation process, Radio 1 has to play the Top 40, and the Top 40 influences Youtube views and Youtube views influence Facebook likes and Facebook likes influence streams and streams influence ‘Discover Weekly’ which influences streams again, and streams count towards the Top 40 and the Top 40 ends up on Radio 1. It’s all connected. For some listeners, this is fine. As long as they get fresh music in their playlist every week or hear their favourite tune on the radio, they are happy. For others, big data driving music goes against everything they think music should be about.
Want to escape the big data penetration? Find a club with a crate-digging, vinyl-loving, remixing, cross-fading, honest-to-God DJ. And don’t turn back.
Header image by Camelia.boban (Own work) [CC BY-SA 3.0], via Wikimedia Commons
