Music Discovery: Interpolation vs. Extrapolation
In the realm of music discovery, the big question is: “How do I find music I like”.
At a time where more music is available to us than ever before, this experience should be a luxury, not an inhibitor.
Today, there are two main ways in which streaming services tackle music discovery. On the one hand you have companies leaning more towards human curation (like Pandora and Apple Music) while the others are more focused on intelligent algorithms (Spotify, Google Play). Both approaches have advantages and disadvantages when you think of the sheer size of music available today.
Regardless of their approach, all streaming services today have one thing in common - interpolation.
In mathematical terms interpolation is defined as a method of constructing new data points within the range of a discrete set of known data points.
If we translate that into music terms we get: “a method of finding new songs within the range of a set of known songs”
When thinking of Discovery, this might seem counterintuitive because this process in itself imposes a limitation on what is discoverable.
There is a very simple and logical reason for why streaming services take this approach: there is too much music and not one service has been able to accurately process and analyse all of it.
If we rely on humans, we would need a very, very, very, very (you get my point) large team of curators to go through it all — which is why Pandora do not have a huge catalogue of music compared to other services — The upside is that their Genome Project has made their database clean and a lot more accurate than any other service (less is more?). On the flip side, you can find whatever you want on Spotify, but they only curate music from a very small percentage of that database themselves.
Today the approach of interpolation is connected to “related artists”. All streaming services have their own set of rules for how their artists are “connected”.
An artist can be related to another because; they collaborated on a song, they are signed to the same label, fall under the same genre, they have a similar style (singing/rap flow), they have the same fan demographic and social reach or they share the same location (in rap these comparisons are more apparent — Kanye West and Common are connected because they are both from Chicago, they have collaborated on several albums etc.)
The combination of related artists and the use of interpolation don’t really allow for great discovery as you very quickly find yourself getting stuck in a loop.
Below is a Spotify list of related artists to Drake.
I then created a radio station with Drake as the seed artist.
A few things to notice: As I continue to listen further through the playlist, a lot of the artists you see here, are repeated — including Drake every 5th track.
Also, all of the ten related artists (see above) appear in the first 2 hours of this particular station.
Now, none of this is surprising or in fact wrong, but when you feel like discovering new music, you are very hard pressed to find it this way.
This is true for all streaming services, because with interpolation you eventually get stuck in a loop.
So what if we were to use Extrapolation?
“In mathematics, extrapolation is the process of estimating, beyond the original observation range, the value of a variable on the basis of its relationship with another variable.”
When we think of that in terms of music discovery, it actually means we can try to look beyond the set of songs we already know and incorporate new songs into our “safe” list.
Here you can see an image that visually explains the difference between interpolation and extrapolation. You can see that with interpolation you are very quickly limited, but with extrapolation you can actually continuously find new connections. The size of our song list or database, being the only limitation.
Every professional DJ I know, uses a combination of interpolation and extrapolation — this is their formula for how a sequence of songs fit together in their set. It is this formula that allows them to take their audience on a musical journey.
We do a very similar thing with Muru.
The truth is, all of us enjoy discovering new music. However, it needs to happen at the right time and in the right context. DJ’s know and understand this better than anyone else.
Extrapolation in music discovery, gives us the opportunity to be surprised, delighted and excited about music. And dammit, we need to get more excited about music again!