The Musician Network

While at Jazz Fest last spring, I was in total music overload. Every musician and their mother seemed to be in New Orleans that week. Our crew’s itinerary was packed with shows spread about the city. While I didn’t always know who I was seeing, my friends were knowledgable about Jazz/Funk musicians and were able to give me context. We had a lot of conversations that followed a formula:

Me: “Who are we seeing tonight?”

Friends: “Oh it’s a supergroup with X, Y, and Z who play in bands A, B, and C normally”

After having the same conversation about 20 times, an idea dawned on me: there are tons of collaborations in the music world and those collaborations likely form a graph. Wouldn’t it be cool if you had a way of searching for a musician and seeing their network of collaborations? Thus the idea for the musician network was born.

The Data

Now it was time to prove my hypothesis. So, I asked myself, where can I find these collaborative relationships between musicians? Where else but ole’ reliable Wikipedia. A musician’s Wikipedia page has great metadata on it. Among that metadata is a musician’s collaborators and associated acts. Take a look at David Bowie’s page.

Here we see all of the bands and musicians that he worked with throughout his career. I was able to extract the collaborations between artists by using these links. I wrote a screen scraper web crawler that grabbed these links, navigated to each of those links, and record the artist that was navigated from. After running the algorithm for about 1 day, I was able to capture a large portion of the musician network from Wikipedia and store it in a database. You can see the technical details of that process here.

The Visualization

Brian Eno

Now that I had the data, I wanted to display the information that I captured. I built a web visualization that represents data in a graph. It enables you to type any artist’s name and see their corresponding network. After experimenting with the tool I built, I made some observations.

  1. The degree of interconnectivity between many different musicians is astounding. Seeing a visual representation of the connectivity illuminated this in a way that I never could have fathomed otherwise.
  2. Bands tend to form clusters in a graph. Whenever you see a tightly grouped cluster of nodes, it most likely a band.
  3. Artists within a similar genre, unsurprisingly, are very tightly grouped together. If a musician has done work outside of their primary genre, it is made fairly obvious by a disjoint node outside of their primary network. For example, Jay Z normally collaborates with other Rap and Hip Hop artists. But in the early 2000’s he created a collaboration with Linkin Park, a punk/rock band, which was outside of his primary genre. This is illustrated in the graph, with the Linkin Park nodes being disconnected from his primary network. This pattern is replicated across many other artists.

Go play!

The tool is now published and ready to be experimented with. Check it out at I’d love to hear what other people discover with it. If you find anything cool or have any suggestions for enhancements, shoot me a line at !