Lessons from an Open Source Legend: Brian Behlendorf visits the GovLab

(co-authored with Luc Lewitanski)

The GovLab welcomed Brian Behlendorf, one of the most prominent and enduring figures in the open source software movement, co-creator of the Apache Webserver and Foundation, board member at the Mozilla Foundation and CTO of the World Economic Forum, to give us a two-day teach-in on a variety of topics. On July 9th, he addressed the entire organization with a lunchtime talk about his career and some broad observations about what the GovLab and our community can learn from the open source movement.

A self-described mediocre student in the Berkeley Computer Science department, Brian Behlendorf was fascinated by the pre-Web Internet and was an active member of various communities online. He was interested in learning where the Internet came from: he compulsively read the Internet Engineering Task Force’s Request For Comments and found himself inspired by the fact that geeks shared their research very widely. He refers to this as a pronoiac environment (as in, the antonym of paranoiac) where trust is ubiquitous and openness is obvious. At 21, Brian found himself reading the HTTP working group boards where Sir Tim Berners-Lee and a young student named Marc Andressen were arguing over standards. He quickly realized that he was learning more about computer science from his online communities than at school. Brian felt entitled to express his opinion. He began to make his own suggestions and learned how to ask smart questions in order to be heard. He had ingrained himself into a highly meritocratic community built around the power of peer review (or flaming) as opposed to credentials.

Brian regaled us with stories from two decades as a founding member and active participant of the open source community. As a Unix Sherpa for Wired Magazine, he found himself spending time writing patches for the web server software that the NCSA had created. This lead to what he calls his role in “co-birthing” Apache, a seminal open software product. Apache first resided on a Unix box at Wired that Brian owned. He named the product after the Native American Tribe, which was the last to survive to the onslaught of the Western forces – drawing an analogy with the hegemony of Microsoft. Others later pointed out that the name created an unintentional pun given the community’s emphasis on trading patches. Since its creation 17 years ago, Apache has consistently been used for 50% to 60% of the entire web.

Brian shared a couple of insightful mantras by way of Matt Mullenweg (founder of WordPress) that have guided him: “Usage is like oxygen to applications,” “If you’re not embarrassed by your 1.0, you waited too long to get it out,” and an unsourced one:  “The last defect is fixed when the last user is dead.” This iterative, empirical mindset provides a useful framework for GovLab’s upcoming crowdsourcing projects. He also warned us: when maintaing a platform, it’s very easy to pick up technical debt, which is the accumulation of short-term thinking, bad architectural decisions, and promises that you can’t throw away. To avoid these pitfalls, Brian recommended that the GovLab read Karl Fogel’s Producing Open Source Software and the Agile Manifesto. With regard to crowdsourcing expertise, Brian advocated engaging experts virtually through asynchronous methods such as mailing lists and wikis in order to keep a perfectly transparent record of both formal and ad hoc discussion threads. This record allows people to easily monitor what approaches have failed, or as he calls it, “where the bodies are buried”.

After 20 years and countless projects, Brian has developed a cogent theory about how Open Source communities begin, grow, and achieve what he calls “steady state” — a maturity in which enough users of a piece of software contribute back the improvements required by its broader community of users. Sensitive to the political dynamics of these communities, Brian articulated the community manager’s role as engineer, traffic controller, facilitator, curator, and, very occasionally, judge and jury. In particular, we reviewed the case of the Direct Project, a medical record messaging standard that he championed during his tenure at the Department of Health and Human Services. In this case, consensus was achieved by allowing the different proposed standards to compete in a bake-off. He estimated that perhaps as few as 40 contributors on the wiki and 15 core programmers wrote code for this project. However, 1500 people closely followed the email discussions on the subject. We shouldn’t be surprised to note that successful online communities often demonstrate this power/law distribution. Today, this standard has become the default for health care providers to easily share semantically meaningful medical records.

Those of us working on evolving institutional models to be more open have much to learn from the open source movement. At the GovLab, we want to research how the cultural, procedural, and community processes that form organically to support these efforts can be transported into non-technical modes of collaboration. Brian and his colleagues have an important role in distilling what has made open source so successful and sustainable. Sometimes, the engineering community’s broad ability to create and scale technological solutions distracts us from studying the underlying new organizational dynamics that helped drive innovation. Engaging experts of Brian’s caliber allows the GovLab to glean insights and adopt methodologies that better address the challenges we face across the public sector.

Brian was a pioneer in realizing that the ideals behind what made open software work effectively could be applied to our institutions and social systems: collaboration and community can work concretely to help us make better decisions and solve public problems in new ways.


No comments yet.

Leave a Reply