The Story
I am one of the Co-Founders of Napster. I was there in the beginning, hacking on the code, fixing bugs, adding features, before the investors, before the strung out rock bands like Metallica started complaining, before we ever came to California. It was an amazing, crazy, epic experience.
I resigned Napster on October 31st, 2000 and left 2 weeks later, conspiciously the same day the Napster-Bertelsmann agreement was announced, and roughly the same day that BMG CEO Strauss Zelnick was sent packing.
All The Rave: The Rise and Fall of Shawn Fanning’s Napster
Ever wonder what really happened to Napster? Heard rumors but never the real story? The definitive source cataloging the saga from start to finish, the internal strife, the chaos and insanity, and the rise and fall of an amazing technology that defined an era is now available (Amazon, B&N).
Chock full of things you didn’t know from the mainstream media but would be surprised to learn, this book is an informative read.
Origins
So you want to know about Napster, huh? Stuff in the magazines no good? Newspaper stories got boring? Get the feeling something isn’t quite right? Onward, gentle reader! Here’s some rad facts about reality and my involvement with it:
Napster was essentially Shawn Fanning’s first Windows program.
Napster v1.x was the first prototype, which used UDP as its primary network transport. Shawn Fanning wrote both the client and the server by himself. The last version was 1.2 (I believe) before the design was dumped, due to inadequate to support the amount of network traffic Napster generates.
Napster v2.x was the second prototype, which instead used TCP as its primary network transport (much more reliable).
The concept was founded in September of 1998 by Shawn Fanning, solely. Not by a car salesman uncle, not by a coked out wannabe kid, all opportunistically claiming credit for other good people’s ideas and work. Just one enlightened kid with the resolve to follow through.
The company was incorporated by Shawn’s uncle, John Fanning, in May of 1999, and I am told that this was done without Shawn’s immediate knowledge or involvement. Newspapers have accurately reported the morally deplorable distribution of equity (Uncle 70%, Shawn Fanning 30%). Gotta love family!
Before Napster had taken its death grip on the Internet (before it was even known by the general public, April-May of 1999), Shawn would solicit his friends for help (Napster was never a simple piece of software). I was one of those friends. Sometime in June, he gave me full control of the backend and I became the sole developer of the server. Before June, Evan Brewer (another friend of ours) had been acting in a systems administrator role for Shawn. When I took full control over the server, I also took over systems administration duties as well.
Napster relocated from Hull, MA to San Mateo, CA in August of 1999, christening its first moments in time as a “real company” (beforehand, it was just Shawn and myself doing the development).
During my tenure at Napster, the backend engineering group had grown by a significant two: Ali Aydar, a close personal friend of mine and someone who helped teach Shawn software development, and Jordan Mendelson, a superbly bright software architect.
Ahh well. So long, old friend.
Some Technical Misconceptions
I’ve seen a lot of misconceptions over the last decade about what Napster was capable of, usually in comparison to other alternatives. So here, have a nice pile o’ facts:
By spring 2000, Napster servers were linked together.
Many, many people looked at the online user count and assumed that was the total number of users available to them – this was not correct. The user count was a farce, limited only to users on the local server – a stark contrast to all other file-sharing alternatives, who [intuitively] reported the total number of users participating in the system.
Because of the scale problems that result from linking servers in a full-mesh model, combined with the sheer amount of traffic we received, the servers were split into 2 clusters. This gave users a 50/50 chance of being able to reach their friends. This was unfortunate, but the primary function of the network was to locate and distribute content, and that took priority.
Why, then, show only local users if searches and downloads extended to so many more? As custodians of the largest and most content-rich file-sharing network in the world, management was concerned that we would attract too much attention before the company was prepared for it, and so we were ordered by executive management to configure the servers to report local users only.
Of course, anyone could have tested the hypothesis by simply searching for something hard to find, initiating the download and then looking up the source’s user info – one info field would say (remote), indicating that user was not local. And, of course, you could message anyone anywhere on the Napster network, not just on your local server. Unfortunately, no one reporting on this subject thought to try this.
By November 2000, Napster regularly peaked over 1M concurrent users daily. Rumours were that at the time this was very close to the average concurrency of AOL.
Despite a large number of (bogus) resumes on the ‘net from ex-employees at Napster, during my tenure only three people were primarily responsible for the development, deployment, and well-being of the service.