Panel with Brian Behlendorf from Apache/Collabnet, Josh from PostgreSQL, Chris Hoffman from Mozilla, Larry Wall from Perl and David Wheeler from Bricolage. Moderated by Christian Einfeldt, “How these communities result in 1 + 1 = 5…”
Blogger’s Note: It’s interesting to think how lucky these guys are, and I mean this in the best way. They have been able to manage large scale projects and arose through some of the purest merit. Now they have significant respect from a broad community and the business world themselves. Non-technical managers do not have these opportunities, to learn and ascend among large groups. Closest thing to it may be in the non-profit world, but the respect of business peers is not the same. Being a startup entrepreneur has levels of meritocracy, however relationships and capital structures provide boundary conditions. Yet picture that young marketing manager with a crappy day job. Could she play a role and ascend in these communities to just as much prominence.
Larry Wall: small group of core developers, different groups and specializations. Have some people still working on Perl 5 and another end of the candle working on 6.
David Wheeler: can speak for those at Kineticode, who knows how many are using Bricolage. Companies and organizations like Portugal Telecom and WHO look to sponsor development. Some like RAND want to part of the community.
Chris Hoffman: Sun and IBM were there at the beginning as corporate sponsors. A number of companies looking to roll Firefox out to even 100k employees. There are also curiosity companies, who see the development model and want to tap into the collaborative process.
Brian Behlendorf: Strange to start this conversation by asking what companies have been involved. Never recognized them explicitly as special entities. It’s a membership organization based on individuals, who happen to move jobs, and may have first allegiance to Apache over their jobs. However, have 1k developers who have commit levels, 5k contributing patches and on the list. Some are living the dream of writing software you then give away. Seimens wrote us a $5k check once, and IBM and Sun gave us machines. Tried to keep licenses clean for the benefit of corporations. Some legal questions were beyond us, so we created a forum to talk about legal issues and invite lawyers to participate in the conversation — has worked out really well, helping us understand nuances. Building bridges to corporations is important but we have been successful without significant initiative.
Josh Berkus: individual developer membership, and have talked about formal corporate participants for issues like J2EE compliance. They say to get an engineer to participate on a list.
Chris Hoffman: Most enterprises don’t know how many of their engineers already participate in open source projects.
How should a company who wants to contribute get started? (the panel seemed to resist this question when it was asked before)
David Wheeler: Haven’t had companies come to contribute, only to fund the development of a feature. They tell them to develop it within the mix, but not retain copyright, although they have had some say they have a script and look to contribute. Wouldn’t want a company to dump code on them, would want a dialogue on licensing and how they can make the project successful.
Larry Wall: In the Open Source Innovation panel there were metaphors that is software more like an encyclopedia or a poem? A language design is more like poetry. Our project is not terribly typical with others. Have to interact with the design team, open to feature suggests and patch contributions that are fits — then it’s in. Most of the pressure went away when Perl 5 went out. Perl 4 had feature pressure. If you make your core extensible, you end up with CPAN instead. Most contribution has gone into CPAN.
Josh Berkus: International contributors that would love to go full time that need funding. Some projects are forum or IM meetings where decisions get made, so you need to participate if you want to have a voice in decisions.
Brian Behlendorf: Just join the use and developer mailing lists. 30 top level projects, a couple of releases within it have diverse problems that companies could contribute to. Adding new functionality can be done within the community. Have a process called The Incubator where new projects start from anyone’s proposal, get resources of a project, can exit when you meet a criteria, including packaging and shipping but also having more than a single constituency within the community that participates in the project. Don’t want a project to disappear if a company abandons it. Contributor license agreements for individual and corporate participants to make redistribution clear. After The Incubator it can go into one of or a new top level project. XMLBeans as an example that went through the process. Feels like bureaucracy, layers of indirection, but good for quality products without dependence upon companies.
Chris Hoffman: Most of the help we provide is developing a strategy for people to be involved. The harder thing with Firefox and Thunderbird is you can incubate and prototype and promote, but people want assurances that it will appear in the main release. Technical hurdles, licensing, community around it have to happen not just because of the requirements itself but the quality burden for 40M+ mainstream users.
Brian: gut instrinct and intuition has been my guide. If someone else has a good effort going, look to join it. Find people that are trying to solve the same problems. Choose the simplest licenses, to lower the transaction costs to joining your party as low as possible.
Chris Hoffman: when IE launched, Barksdale said we needed to do something significant — open source the browser and fast.
Unless you ahve a legal reson to open, you dont want to incorporate too soon. You don’t want to add structure too early.
Brian: Humility. Say I have a answer, not the answer
I asked about if they see a rationale for when property should be opened (similar to how transaction cost analysis helps inform outsourcing)
David Wheeler: some contribute because it is not thei core competency, dont want to take care of something outside their expertise. Is this improtant to our business. Do we have anything to contribute t o a project outside.
Brian: If it is in the bucket of context, look to partner with others, like Geoffrey Moore said. Sometimes may want to put something core out. Started Subversion at Collabnet, an improvement of CVS, started a community and did the work out in public. Put 4 FTEs on it, large for a startup, and built a community of 20 people, now according to Necraft there are 8k public installs and did 1.0 a year ago. Needed it to be competitive with other software engineering tools out there, a business decision to open, although it was core and it amplified effort and commoditized the space.
Larry Wall: My low level implementation guy in Austria, compiler guy in Taiwan — its outsourcing!
Brian Behlendorf: Most businesses donate and consider it a promotional expense and don’t track it. Open question is if we will have to start tracking value of donations of IP as non-profit organizations.
Josh Berkus: For an established project with a community around it, it is tough to conceive of how to have venture support without new risks.
UPDATE: Niall Kennedy posts his reflections on the session
Ross Mayfield is CEO and co-founder of Socialtext, an emerging provider of Enterprise Social Software that dramatically increases group productivity and develops a group memory.
He also writes Ross Mayfield’s Weblog which focuses on markets, technology and musings.
