I took a poll the other day: “Do you feel like you have a music discovery problem?” 1/3 of respondents said yes, the other 2/3 said no. Honestly, I was expecting a stronger skew towards no, but I suppose we’ve been trained to expect a constant stream of new excitement, regardless of the medium.
Today, though, I want to make the case that content discovery is not a problem with a single solution. No matter how good the algorithm, there’s no way that Rdio, Spotify, HypeMachine, exfm, or any blog could possibly single-handedly “solve” discovery. No matter how perfectly they can match content to my tastes, they are inherently devoid of the emotion that comes with disaggregated discovery.
After I graduated college, I spent a few weeks in Scandinavia. I was fascinated by the similarities and differences of cultures and countries, and as a music fan I was on the hunt for great artists and songs that hadn’t crossed the ocean. In a record store in Stockholm, I heard Oh Laura playing over the speakers. While not particularly awesome, and certainly not in line with most of what I listen to, I became obsessed for the next few months (I had to hunt like crazy to get the MP3s in the US). It was real-life context and happenstance that caused me to like the music.
A few days ago, I opened the daily email from The Listserve. A guy I don’t know from Brooklyn who apparently likes baseball and Game Of Thrones recommended everyone check out Lorde, a 16 year old New Zealand female pop artist. On any other day, I would’ve passed over his recommendation (I don’t know him, why would I trust his taste?), but I was feeling adventurous. I checked it out, and after a couple days of listening, I shared it on Twitter.
Discovery comes from all over, and it should. When you’re surprised in a good way, you ascribe more value to the thing that did the surprising. Algorithms can be designed to make better choices for you than your brain is often capable of (“What should I listen to after this Foxygen record?”), but they’re rarely designed to surprise and delight; they’re designed to build trust, which means consistency. We take more risks with our choices than algorithms do, and with greater risk comes greater reward.
Does all this mean companies shouldn’t try to help you discover content? Absolutely not. It simply means those efforts should be designed a bit differently than we see today. Ideally a system would include:
- - A variety of choice, with some close to the user’s comfort zone and some towards the various edges; some suggestions from social data, and some from “likeness” algorithmic data.
- - Additional context to power further discovery. This could be as simple as user-defined tags (Bandcamp does this well for music, Instagram does it well for photos, etc).
- - Easy ties to other points of discovery. Eg - if I discovery something elsewhere, I should easily be able to bring it into this system.
The final point is clearly the toughest. Content platforms want to believe that they are your only source for that type of content — if you have access to “every song in the world” through a streaming service, why would you ever go to blogs and SoundCloud (or even crazier yet, a live show)? But opening up to those inputs will add value to the platform’s ability to better understand your habits.
Discovery doesn’t happen in silos. It never will. It’s as capricious as we are, by design — we appreciate things more when they come from surprising places.
2 months ago - Permalink