You know how, for the past decade, you’ve been able to track everything you’ve played on your iPod, iTunes, Winamp, Windows Media Player, Spotify, and more — and have all of that stuff create a Last.fm profile, which is useful for all kinds of reasons?
No? Well then, do you use Facebook? It’s basically doing the same thing.
Richard Jones released Audioscrobbler ten years ago today [this article originally posted on November 20, 2012], inventing the concept of tracking the music people play in order to inform music recommendations.
He merged it with Last.fm just a year later, in 2003, in return for 15 percent of the company. When CBS acquired Last.fm for $280 million in 2007, Jones earned a healthy chunk of change for his invention — we estimate around $38 million.
Ten years later, scrobbling has changed the world, even if the people using it might not be familiar with the word. The scrobbling concept has become interwoven throughout the internet. People use it every time they sign in to a music service with Facebook.
We caught up with Jones for an email interview, which we present here on the tenth anniversary of scrobbling. (All interviews are edited for length and clarity.)
Eliot Van Buskirk, Evolver.fm: How did the idea of scrobbling occur to you?
Richard Jones, inventor of Audioscrobbler: I had been reading about collaborative filtering algorithms, and it occurred to me it would be fun to try and use them to discover new music. ‘People who like A,B & C also like D…’ but for music.
There wasn’t really a suitable source of music data to base recommendations on, though. I considered crawling the P2P systems of the time (Soulseek was popular back then) — but in 2002, when the world was still reeling from Napster melting down, people who listened to MP3s tended to hoard massive collections of music they had no intention of listening to, just because they could.
This made the list of music you were sharing via P2P a poor data set to base recommendations on — yes, I downloaded the full Metallica back catalog just because I could, but I didn’t really listen to all of it, and I wouldn’t want recommendations based on it.
Always-on internet was rapidly becoming the norm, and more and more people were listening to MP3s on their computers with software such as Winamp. So scrobbling was born: I wrote a plug-in for Winamp that reported what was being played in realtime — the perfect data source for recommending and discovering music.
Evolver.fm: When did you realize that it was a big deal?
Jones: I remember a few occasions where I got a ‘This is going to be bigger/cooler than I expected’ vibe. Early versions of the Audioscrobbler website did little more than present all the data and cross-link everything. You could click an artist name from your profile page, see which tracks by that artist were most popular, click on one of the ‘top listeners’ of that artist, and end up looking at someone else’s music profile who listened to the same artist as you — then start the process again, clicking on a different artist.
That mechanic was core to the whole experience, and still is today. I had no idea that would be such a fun and interesting thing to explore until I’d built it. I remember thinking I had discovered something pretty cool at that point. That was the ‘Audioscrobbler minimum viable product.’
The next occasion I remember was when I spent some time digging in the logs and running the numbers, a few months after I started. The handful of friends I had initially convinced to install my Winamp plugin ‘so I can collect enough data for my dissertation’ had told all their friends, and before long, I was seeing new people signing up every day from all over the world.
People were starting to write about Audioscrobbler in early 2003 (waxy.org possibly being the first blog) and there was coverage in The Guardian (the first mainstream press article, if I remember correctly), BBC, etc. The surge in interest and signups was highly motivating.
As silly as it sounds, the other occasion that stands out for me is a bit later on, when we got our first Google Adsense payment in the mail. We were struggling to afford rent, servers, etc. The first month we ran ads on the website, we made a few thousand pounds. I think that was when it first dawned on me that what we were doing could turn out to be a viable business, as well as a labour of love.
Evolver.fm: How did Audioscrobbler get acquired by Last.fm? What was/is your reaction to how widespread its use became after that?
Jones: I’m not sure ‘acquired’ is the right way to look at it. We teamed up and pooled our resources. Last.fm and Audioscrobbler had both started completely independently at around the same time. There were many similarities in the long-term plans for both projects, the first meeting went really well, so we decided to merge the two projects.
Once we presented a single, unified website, the resulting whole was definitely greater than the sum of the parts, and the growth sped up dramatically. When that happened, we shut down the Audioscrobbler legal entity, and I ended up with the same percent ownership of Last.fm as the other Last.fm founders.
Evolver.fm: What was your reaction when you saw Facebook implement, essentially, your scrobbling idea on a mass-market level? Did that take you by surprise, or did you always know that scrobbling was for everyone?
Jones: I think scrobbling and the idea of a music profile is important enough to warrant a product in its own right. To do it justice, you need a dedicated team and a dedicated product. Facebook have done a fine job of bolting it on to their existing product, but it’s definitely one of those features that deserves to be more than an afterthought.
Evolver.fm: Does scrobbling need to be an open standard, do you think? Or is it fine as part of Last.fm, Facebook, and other proprietary networks?
Jones: I’ve entertained this idea a few times in the past, but practical and logistical concerns tended to stop any efforts in their tracks. You need a lot of data in one place to be able to do interesting things with it. Hosting and crunching lots of data requires lots of hardware, which isn’t cheap. Decentralising or otherwise federating the scrobbling data makes it less useful overall, unless you have a central way of browsing it, which is costly.
One possible model for the future is something like Musicbrainz.org — a foundation or non-profit entity that holds the data. Open source and non-commercial entities can use the data for free. Commercial use of the data is subject to a fee, which in turn pays for the foundation to run their infrastructure. There are a lot of details to consider, but something like that should be possible.
Evolver.fm: Can you believe it’s been ten years since you released Audioscrobbler? I can’t; it seems like yesterday. What have you been up to that you’d like people to know about?
Jones: It certainly makes me feel old. I turned 30 this year — most of my 20s were spent building Audioscrobbler and Last.fm. Absolutely no regrets though. There is no substitute for working with a world-class team on a product you are passionate about that is used by millions of people.
When I left Last.fm in 2009, I spent a while working on something called Playdar, which is relevant to your next question…
Evolver.fm: What’s the next problem in music of that scale that needs solving? Or is it a secret?
Jones: The next problem is one of access to music and interoperability. ‘Access’ is solved, in the sense that you can download or listen to millions of songs from YouTube, Amazon, iTunes, or Spotify. Stuff that isn’t there can be found via P2P, music blogs, and a host of other sources.
So given that access to music is easier than ever, I challenge you to send me a playlist that includes stuff from an unsigned artist, some artists that aren’t on iTunes or Spotify because of digital rights issues, and some major label artists.
The main sources of music don’t really interoperate. If all the tracks you want are on Spotify, it will work fine — just make a playlist on there. If they are all on iTunes, I could download the entire playlist. What if I need to access three different services to get all the tracks in a playlist? What if I dont have a Spotify account because I use a competing service like rd.io? Did I mention I’m actually in Europe, and half the YouTube videos you sent me just say ‘This video is unavailable in your country?’
After leaving Last.fm, I spent some time working on Playdar.org — a ‘music content resolver’ designed to provide an abstraction layer that hides the boring details about where music comes from, so people who build apps and products that play music don’t have to worry about that. It should come from whichever sources make the most sense to the individual listener, according to their country, their preferences, and whichever subscription service they are signed up with.
If this access problem is solved in a satisfactory way, I fully expect to see an explosion in the number of amazing new music services. Building a music app that plays you music today is a legal minefield, and royal pain in the ass. If that problem were to effectively go away for most developers, I think we would see a lot of innovation.
(Photo courtesy of metabrew)