Open Source Talent Drain - How Success Can Hurt Open Source Projects

I am attending the 451 Group Conference in Boston this week. Raven Zachary, the Open Source analyst for the 451 Group, led a session on the future of open source.

One issue Raven raised was the difficulty of scaling a successful open source project once it has become successful. The hallmark of a successful open source project is that the lead contributors hit the trade show circuit while big players like IBM cherry pick key project contributors to bolster their own offerings.

Raven gave the example of the Tomcat team, where a number of key contributors have been hired away, leaving the team roughly the same size it was 5 years ago, despite vastly higher usage and community needs. The solution would seem to be a model where the contributors could make enough money to keep working on the project as it becomes more successful (aka the JBoss model).

For long-term success, it may be that an open source project must be financially successful enough to cherry pick its own best contributors. Ruby on Rails may be the anti-example to this strategy, so it will be interesting to see how they handle their own success long term.

A related problem is that once an open source project is successful, its contributors can command significant consulting fees for doing work that others could do if the product had better documentation or usability. There is an odd negative incentive here, where making an open source product more user-friendly could cost the core contributors money. In the worst case, the incentive is to add cool but half-baked features so you have more stuff to talk about at conferences.

Some other interesting points from Raven's talk:

• Simplest definition: open source is community supported software that is freely available on the internet and which has minimal restrictions on redistribution. A more complete definition is here. Many popular open source products don't exactly fit this definition, so open source is at some level a business positioning as well as a development and distribution model.
• Support and services model is most accepted: customers are loathe to pay license fees to "upgrade" a supposedly free thing. I'm not so sure this is the whole story. I had a subsequent discussion with David Skok at the conference, and he felt the key to JBoss' success had been their ability to upsell management software to JBoss users on a subscription basis, taking their ASP from $10K to $50K.
• Vendor opportunity to reduce stack complexity: customers are struggling with how to manage 15 different open source vendors, each with its own licensing and support community. A related issue is how to handle integration issues between these products, an area being addressed by companies like Red Hat and SpikeSource. Our own WaveMaker product integrates over two dozen open source applications, so this trend is both real and accelerating.