Steve Howe had one of the most unfortunately named sessions of OSBC London - "Openadaptor - Efficient System Integration and Migration" - and one of the most interesting. Steve is Global Head of Open Source Initiatives for Dresdner Kleinwort Wasserstein, a Europe-based investment bank.
The company started its Openadaptor project to "provide an easy and standard way to allow DrKW to interact with a standard message bus." Innocuous enough. The purpose was to "stop the needless duplication of effort that would otherwise have ensued." Again, vanilla. The company wanted to coordinate and leverage a community - including its competitors - to offload support costs and improve development.
Did the dream become reality?
Not really, Steve said. He noted that DrKW received a lot of advice and contributions from outside the company, but not necessarily a community. As Steve said, "we made some good friends," but that's it. Why? Steve suggested it's because the "itch" they were scratching was probably fairly well-localized to DrKW - it wasn't a huge need that lots of people felt.
[UPDATE: Steve corrects me in the comments below: "One slight correction though - I didn't feel the itch was specific to DrKW, but rather to complex organisations with many systems and the need for middleware. ie the itch was smaller than 'I need an operating system that doesn't crash every 5 minutes'."]
Was it a mistake, in Steve's mind? No. But it required a lot more work than expected, because it's very difficult to get outside contributions. Marc Fleury addressed this in his opening remarks, as well. Those who take a "If I built it, they will come" attitude are going to fail. Completely. They won't come. Not without a lot of work.
Some of Steve's conclusions, based on the Openadaptor experience:
- Community development of software is possible at the lower end of the stack as there are fewer design choices to be made.
- At an even slightly more complex level such as openadaptor it becomes much more difficult to develop communally.
- Highly complex applications will never be able to be developed communually as there is too much scope for religious differences of opinion. [Asay: This is a different riff on a theme r0ml raised at OSCON 2003 and reprised at OSBC 2004 ("(More) Missing Open Source Projects").]
- There is scope for open source business software, but the development model will need to be different to that of traditional open source.
- Finally, remember the key differentiator between open source and proprietary software is quality, not cost. It is important not to lose support for open source as a whole by overselling what is possible today.
Posted by Matt Asay on June 28, 2006 01:52 AM
