The Open Source Community Imperative

Notes from a panel at OSBC

Peter Fenton, Accel Partners, Moderator
Ranga Rangachari, GroundWork Open Source Solutions. Has a hybrid OSS model, 100 customers.
Daniel Frye, IBM, works on their open source strategy team, manages a ton of OSS developers.
Adam Tractenberg, eBay developer evangelist
Clint Oram, SugarCRM, from inception a commercial open source entity
Mike Olsen, sold his company, Sleepycat, to Oracle today

• Peter Fenton fesses up to the irony that a VC is talking about community, but notes that enterprise software model is dead and the big hits lately have been from communities. Communities appropriately applied will be the big breakouts. Open source as an invention, a form of sales and marketing to bring costs from 60% of revenue to 20%.
• Daniel: Communities are quite different, from an analysis we found that some build around a single commercial effort, entity or individual. Looks and acts different from a community that is very broad, with multiple interests. Differences in governance models. Have learned, in some cases the hard way, it is really not easy to start communities. Much easier to leverage, adapt and adopt communities. Most important thing when building one is asking why people would join. Raymond's Rule of Everyone Scratches Their Own Itch.
• Peter: Striking differences in the level of community contributions to the code. Sleepycat does all the contributions, for example.
• Daniel: Most are less that 10%
• Clint: first thing we did was to start forums for $200. From then we ahve continued with tools for people to collaborate, SugarForge. Tools kicked off the community into high gear.
• Adam: we had a community of buyers and sellers before developers. We saw people screen scraping to automate, a latent demand for web services. We had a community that was reaching out to us. Gradually releasing commercial constraints for leveraging our platform. There are still costs for maintaining a web service. Making it more free has enabled greater growth, spreading eBay in areas we haven't anticipated, the best part of the initative.
• Peter: they are often viewed as cost centers, how do you explain that.
• Adam: eBay is metrics driven. Didn't start that way with the project, but now 45% of all listings go through a web service. We track (paraphrasing: everything).
• Mike: It's okay to pay attention to the cost center aspect, but you have to take a longer view. Fundamentally you want a happy community producing customers for you. The things that haven't worked as well are when we have tried to learn lessons that are objectively true in business school. How to do direct marketing. Had to build out infrastruture, cold call for lead gen, that was at best breakeven. Doesn't give us the same scale or engagement. On the developer forum, watching interactions, you get a very good idea for who is interested. The one way programs don't work. It has to be two way.
• Clint: When we focused inward on our internal operations, we neglected the community sites and had a bit of backlash for not participating enough. Had to refactor priorities. I then focused full time on the community.
• Ranga: not having a community person has been successful. Lots of splinter communities, keep them engaged by contributions. 10-15% of an engineer's time is spent contributing back to the community.
• Peter: VCs see it as a radical distirbution advantage, what about the development advantage?
• Mike: we segment the value of distribution and development (what we do). The model we have works for us, the Linux model works for other projects. Different assumptions for business models and strategies for communities.
• Daniel: the design of the community and participation happens after deciding the business model.
• Clint: Don't shift models in mid stream
• Adam: I have venture backed and hobbyist developers.  Need to help them connect, differently.  Uses a developer event that only costs $50, because they want developers to connect with powersellers.  Have to look at it holistically, beyond providing support and forums.
• Ranga: needs are different for different communties, and includes the wider ecosystem
• Peter: some have resisted the profit motive, like Apache
• Daniel: they are against direct monetization, but they are fine as a leverage point.  They don't want or need to give up control.
• Ranga: each views OSS as an end.  No customer says I like it, but wont pay for it.  An area where you get what you pay for, we see no resistance.
• Daniel: a tension would exist when a commercial entity only takes advantage without giveback. 
   
• Adam: to have a commercially successful web service, you have to align your needs with your developers.  Such as when efficiency in a market wins for all parties.  Tim O'Reilly's example of people who use Google are using Linux.  Lots of people leverage inside the firewall to create services outside the firewall.
   
• Clint: set clear goals for the project.  With that you have to be careful about how you mix the two models.  With SugarCRM.com we also have community interaction we don't interject in the forums.  The tension is, I am trying to interact to get a solution, I don't want to be cross sold support.  Need to keep the commercial and community activities seperate, but acknowledged.
• Daniel: some companies have gotten these things confused
• Clint: and some taking firm stands, like Apache
• Ranga: seen this before, where people see the true open source community as a channel -- a big mistake.  Have to treat them as different entities.
• Peter: as the entity extracts profit, RedHat example creating animosity.  How do you handle the tension when it exists.
• Mike: should I take this one?  There are guys who will, might and won't pay.  Your job is to try to move some of them along from might to will, but don't try it with the wont's.  There are some people that rejects the company should make a profit, but you need to be polite and not alienate the broader community.  Our view is rational people do not object to you making a living and the irrational are not going to be important for your business over time.  Need to engage constructively or shut your mouth.
   
• Daniel: the last thing you do is pick a license.  No license is toxic, they have different purposes.
• Peter: SugarCRM picked one that matched their community without being religious
• Clint: became psuedolawyers to figure it out, the success of our endeavor would not happen if we were the only ones to make money.  So we chose the MPL to let others make money, which helped build a vibrant community for OSS and commercial.  Releasing our stuff today under a microsoft shared license because it works for some customers.
• Daniel: the legal folks need to be more tightly integrated into the team than any other I have seen.  Lawyers need 3-4 years in this field.  You want a team that can converse as developers, how code flows and how community works.
• Adam: one of the most suprising things at eBay was how clued in our lawyers were on OSS.  I wanted a fairly popular license, let people make money, keep us from being sued, not give up our patents and it worked out with relatively little back and forth.
• Peter: how does Oracle or others value a community
• Mike: you would have to ask the deal team at Oracle, which was pretty sophisticated.  What impressed me coming in was the deep interest in maintaining the community.  They want to sustain the community, and support us in how we interact with them.  I believe this is true, prove to yourself by just watching.  We won on the technical merits, but Oracle doesn't need to come to my 25 person company.  The reach and breadth of our channel, the differences within our community.  Fairly Darwinian, can't predict all needs, but they will be revealed.
• I asked about the risk when there is an acquisition or change of control that the community is disinfranchised, how would this discount the value of a company built on a community.
   
• Adam: Oracle buying sleepycat injects risk into the community, but he is right that the risk can be managed.
• Daniel: from Sleepycat ass a different form of leadership ranging to more broadbased communities, I can't think of where this has happened.
   
• Peter: VA screwed up Weblogic.
• Susan from Apache: we do care about commercialization.  Isn't it true that the nature of the community changes when there is a financial stake in it?
• Daniel: The community itself can't be acquired.  When we make a big play, we go for a broad-based community, it is less fragile.