10 Years of SourceForge.net
It's often difficult to notice when you're in the midst of making history. Some fine people in San Francisco went about their unremarkable lives in 1967, only to discover years later that they were at ground zero of the "Summer of Love." In the summer and fall of 1999, I spent some time working next door to four noisy, Mountain Dew-swilling misfits working on a renegade project within VA Linux Systems. Little did I know that their efforts would become known as the world's largest open source development site.
I refer, of course, to SourceForge.net, which launched on November 17, 1999. Most people think of SourceForge.net these days as another huge web site with lots of ads, but very few understand its humble beginnings or how challenging it was to even launch the darn thing without the powers-that-be at VA killing it off in a fit of well-intentioned hari kiri. The history and beginnings of SourceForge.net can teach executives and managers today the value of trying crazy things that might (and probably will) fail; of letting your young guns run wild with imagination; and not squashing innovation within your company. Unfortunately, there are also quite a few lessons on how to squandor success, but that's for another day. Today is about SourceForge.net, the site that was before its time and how it came to be.
SourceForge.net began as a project called "Cold Storage" and was headed by Tony "fusion94" Guntharp. His team included Uriah "precision" Welcome, Drew "dtype" Streib, and Tim "bigdisk" Perdue. The first thing that might surprise you is that Cold Storage was never intended to be a developer collaboration site at all. In fact, Tony's team was trying to solve a vastly different problem - finding and locating different versions of open source software. At the time, there were a smattering of sites called SUNSites, with the most popular housed at UNC, before it mutated into ibiblio.org. But each SunSITE repository was at the time FTP-only, and there was no canonical index: each SunSITE would have different projects and of the projects that were shared, some were not synced. It was possible to download different versions depending on what SunSITE you used. There were security ramifications to this, as one could never be sure which version included a bug fix or patched a security hole. The grand plan of Cold Storage was that all open source software in the world would be indexed and mirrored and made available in an expansive archive. This way, developers and users alike could always be sure they were getting the latest patched releases, or in some cases, that they could downgrade to a version they knew was reliable. On the opposite end of the spectrum was a site called Freshmeat.net. Freshmeat kept an index of the latest versions of all open source software, and linked to the latest downloads. In this case, there was a master index, but users were at the mercy of the latest and greatest, and there was no interface for finding a specific older version.
While they were busily creating Cold Storage, a happy accident occurred. VA housed a number of open source projects at openprojects.net as a sort of open source community service for goodwill purposes - until the day that a massive hardware failure prevented angry sysadmins and developers from accessing their projects. Tony and team were tasked with fixing those issues. While they were working on that, they realized that they could combine the project hosting services of openprojects.net with their Cold Storage plans, and voila! Cold Storage quickly morphed into SourceForge. 67 days of hard PHP/MySQL coding later, SourceForge.net was launched.
Actually, it should have been that easy, but it wasn't. This is the part of the story where what had been an interesting engineering project morphed into a major political undertaking. Enter the bean counters: those who didn't want to launch the service because they couldn't see the business reasons behind it. There were numerous incidents where the project was on the chopping block, only to be saved by a friendly executive or two, including then-CEO Larry Augustin. The typical, cookie cutter MBA-bred CFO takes a look at the P&L sheet, says "WTF?", and quickly reaches for the "Funding Denied" stamp. Lost in that strict P&L approach is the vision behind a project like SourceForge.net, and the idea that building the center of gravity for open source development and attracting a critical mass of developers is actually valuable.
After finally gaining the blessing of the company, the SourceForge guys went on to build out the site, before launching it officially on November 17. By that time, it was recognized as a valid company project. I'll never forget when, a few weeks before launch, a marketing person at VA issued an all-hands-on-deck call to action for employees to help add projects to the site. I'll also never forget working next door to those four crazy guys who spent long hours locked in that room banging out SourceForge code. I often wonder how many companies in the world had the kind of chutzpah, grand vision, capacity for innovation and near-insanity to bring about something like SourceForge.net. Not
These days, *forges come a dime a dozen, and finding a place to host a project is not nearly as daunting as it was in 1999. At that time, offering shell accounts, version control, bug trackers, mailing lists, a unified web front end, and web subdomains *for free* was unheard of. SourceForge.net was truly before its time.
Hats off to Tony, Uriah, Tim and Drew. You really changed the world.
(Many thanks to Tony who helped me recall much of the chronology)