The Three Camps of Open Source by Mark Stone
October 2009. Matt Asay has a typically thought-provoking post up titled Free software is dead. Long live open source. He extolls the virtues of open source pragmatism, and adds his voice to other recent critics of the more ideological Free Software movement. While I like much of what Matt has to say, and I like all of what he forces us to think about, I want to take issue with some of his characterization and remind folks of the full historical context at work here.
Matt poses a typical contrast between pragmatic open source and ideological free software, as if these are the end points of the spectrum, and as if open source emerged out of free software. Specifically, he says:
“There have long been two camps within what we typically refer to as ‘open-source software.’ The first is led by free-software advocates like Richard Stallman (who, importantly, largely eschew the term ‘open source’ as not being sufficiently concerned with freedom), while the latter is led by no one, but was formally organized in 1998 by Tim O’Reilly, Eric Raymond, and others in Silicon Valley.”
Actually, while the term “open source” was not coined until 1998, the roots of the movement go back at least as far, and evolved independent of Free Software. The real contrast is between Unix and GNU, as emphasized by GNU’s recursive meaning as “GNU’s Not Unix”. And to some extent the contrast is between Berkeley and MIT.
Kirk McKusick nicely laid out the history of Berkeley Unix in his essay, Twenty Years of Berkeley Unix for the original Open Sources volume. (I’ll note that volume was published ten years ago, making it now thirty years of Berkeley Unix.) Part of Kirk’s point is that the idea of “freeing” Unix goes back to the late 70s, and that the Berkeley Unix approach to “freedom” always meant a fairly unrestricted way to use the software, whether in an academic context (UC Berkeley), or a commercial context (Sun Microsystems).
While the term “open source” was actually coined by Christine Peterson of the Foresight Institute, the three people most involved in first pointing out the need for such a term and then evangelizing its importance were Eric Raymond, Larry Augustin (then CEO of VA Linux Systems) and Tim O’Reilly. What’s interesting is that Tim came at this issue very much from a Unix perspective, not a Free Software perspective and not — I’ll point out the distinction shortly — a Linux perspective. Let me put this in context from my vantage point as an editor at O’Reilly at the time. By the time the term “open source” was coined, O’Reilly’s “Unix in a Nutshell” had already sold about 10 million copies. Unix was very big business for O’Reilly, and the importance of Berkeley Unix’s (by this time the three branches of Free BSD, Net BSD, and Open BSD) pragmatic approach to relatively unrestricted use seemed obvious.
What followed from the coining of the term was the very first Open Source Summit, held in Palo Alto. One thing that was fascinating to observe was that there really were two distinct camps there. Larry Augustin made sure that the Linux camp was well represented, and Tim O’Reilly made sure that the Berkeley Unix camp was well represented. And in fact these were two groups that previously had had very little interaction. The Linux guys were younger, more brash, and full of dot-com enthusiasm. The Berkeley crowd was older, more subdued, and saw the dot-com boom as just another cycle. That these two camps came out of that meeting seeing common cause and being willing to work together is a real testament to the perseverance and insight that Larry and Tim had. That they are two distinct camps has been largely obscured by the passage of time, and glossed over not just by Matt Asay but by many others.
Because who was not at that first Open Source Summit? Richard Stallman was not there, nor anyone else of stature from a purely Free Software perspective.
It’s human nature to think in simple dichotomies: Free Software – GNU – GPL vs. Open Source – BSD License. But I would argue that there are three corners to this triangle: Berkeley Unix (and Apache, Sendmail, Perl, and all the other products of the UC Berkeley camp) vs. Linux (and PHP, MySQL, and all the other outgrowths of the dot-com era) vs. GNU.
Linux, in many ways, is the paradox: the most commercially successful piece of open source software, and yet licensed under the GPL. The FreeBSD guys, for example, may well scratch their heads and wonder why their equally good operating system under a more permissive license has enjoyed less commercial adoption. Meanwhile the Free Software guys rail against companies like Google that use Linux commercially without having to release their own software because technically a website isn’t distributing software. The press likes to talk up GPL v3 as a shot across the bow of Microsoft. My own suspicion is that it is more a shot across the bow of Linux. Stallman can call it “GNU/Linux” all he wants, but that doesn’t make it so. Linux is the third camp, not part of the Free Software camp. Not surprisingly, Linus has not taken the bait and has opted not to adopt GPL v3.
Maybe the real lesson is that the magic isn’t in the license, or in the software. It’s in the willingness of camps with contrasting interests to collaborate. When academics, entrepeneurs, and grass roots developers collaborated in the 90s remarkable things began to happen. Sadly, that kind of collaboration is still too much a “manual” exercise: dependent on the personalities involved, and the circumstances that bring them together. Ten years later, we need to be moving to more of a process, codified by a set of best practices.
One way to approach the problem may be under the auspices of an open source foundation, one that is dedicated to understanding and capturing those best practices and processes across projects, and across commercial open source endeavors, instead of trying to advance just one area of endeavor. Certainly that’s part of what we aim to accomplish with the CodePlex Foundation.