Friday, April 2, 2010

Sponsored Communities: Letting Go is Hard to Do

While preparing to teach my Thursday night technology strategy class, I saw a tweet by Santa Clara University professor Terri Griffith (and then by others) reporting that Business Week had posted my column on the difficulty firms have in managing sponsored open source communities.

Entitled “Open Innovation's Challenge: Letting Go Is Hard To Do,” the column was requested by Michael Arndt of Business Week as part of a special report on open source, open innovation and related topics. I used the column to link the research on open source and open innovation to the specific issues that big companies are having in surrendering control in sponsored open source communities.

In my open innovation blog, I enumerated the academic research behind these observations. Here I want to look at the second half of the story: the difficulty firms have in letting go, specifically Google, Intel and Nokia.

I thought it was interesting that there are at least five (now four) firm-sponsored open source communities in mobile platforms: Nokia’s Maemo, Intel’s Moblin, Google’s Android, Nokia’s Symbian and the LiMo Foundation; all except Symbian are variations of embeded Linux. In four cases, there is clearly a single firm driving the process to meet its respective strategic goals.

In this blog, I’ve often commented on this topic, such as Nokia’s tight control over the Maemo community and Google’s over Android. That this keeps coming up over and over again suggests that it’s an inherent problem.

As Siobhán O’Mahony and I found in our research, firms that don’t let go will have trouble convincing outsiders to participate, because they don’t know whether or not they will be able to benefit from their contributions.

Of course, maybe this is just a temporary expedient, of holding the reins tight until the community is up and running, but eventually intending to share governance. An encouraging sign is the decision of Nokia and Intel to pool their efforts (forming Meego), and turning over community management to the Linux Foundation, which is much better situated to build a cooperative open source community.

Still, the inescapable truth is that we really have only one example of a big firm creating a sponsored open source community and then really letting go. In the Business Week column I wrote:

But the best role model is Eclipse, formed through IBM's 2001 donation of its Java development software. IBM executives decided to share control when they realized "they needed Eclipse to become independent to achieve their strategic goal to have the broader Java ecosystem adopt Eclipse," says Mike Milinkovich, executive director of the Eclipse Foundation. Since then, the foundation has been able to attract outside participation not only through its formal processes, but also through new bottom-up initiatives created and led by outsiders.
Due to length limits, other insights from my interview with Milinkovich that ended up on the cutting room floor. Here is a missing paragraph:
Concerned about potential for IBM domination, “a lot of companies watched the operations of the Eclipse Foundation with eye towards “is this organization truly independent?’” said Milinkovich. The foundation was able to attract outside participation not only through its formal processes, but by encouraging new bottom-up initiatives that could be created and led by outside members, independent of IBM.
Another factor that was encouraging was that Milinkovich said that other firms and sponsored communities have come to it for advice. As a consultant to Symbian Ltd. during its final two years, I know that the Symbian (and Nokia) folks spent many months studying and talking to Eclipse in setting up the Symbian Foundation, which earlier this year released 40 million lines of source code as open source. Symbian has done a good job of getting the process right, but still the bulk of the resources are being supplied by Nokia — which then, as now, ships the overwhelming majority of Symbian phones.

Milinkovich said that LiMo also came to Eclipse for advice. However, he did not hear from Google (or Intel) in setting up their communities — whether because they consulted someone else or because they thought they knew everything, neither of us can say.

The fact that Eclipse stands alone is an existence proof that letting go is hard to do. But perhaps two or three years from now, we’ll have other examples of firms dispersing control to build a truly open sponsored community.


Mike Demler said...

Hi Joel,

Good articles here & in BW.

In the case of Android, do you regard allowing companies to take the source OS, branch off and modify for new purposes (i.e. beyond smartphones) as one way of letting go or just fragmentation?

HP/Compaq has modified Android for the Airlife netbook. MIPS & Sigma Design did the same for a set-top box. There are other examples as well.

Is there an example of this type of branching in other sponsored "open source" communities? Does your definition of an open source community require that all code converge back to the origin, so that that the "branchers" can then share in governance? I'm just wondering if this is how any other open source communities evolved.

Joel West said...

This process of “forking” is inevitable in open source, as noted in my earlier posting on Android forking.

Forking is not openness, forking is just a property of an open source license.

True openness requires sharing governance and participation, as Siobhan and I found in our research on sponsored communities. It’s hard to say whether the current lack of openness is temporary or genetic.