[forgive the lack of charts, my blogging service is in the middle of an upgrade] I've been thinking about another driver of features in Open Source software given the recent noise surrounding the harmony project.
I'll get back to harmony in a minute. First let's examine another Open Source java project, lucene and talk about what makes it successful.
Lucene is an open source source engine that is very popular. You might not have heard of it but you have probably used it. Wikipedia uses it, as well as a whole boatload of other sites that are listed on the lucene website, and I'm pretty sure a whole barge full of commercial sites are using it but not telling you (which is perfectly in line with its license). Why is lucene the de facto open source search engine (and possible the de facto commercial behind the scenes search engine)?
Being open source doesn't hurt, but when it is an important component like search, having features is also very important. And who better to implement a search engine than Doug Cutting?
What is it about Doug that makes him such a good open source coder for search engines? Well he has done it before. He did it at Parc, Apple and Excite. (He's also leading development on an open source replacement of Google called Nutch. See this interview for more details about Doug)
Lucene is a very nice piece of software that does what people want (well barring those that have ported it to their favorite language). It doesn't make sense for someone else to re- implement the wheel, when Doug already knows about benchmarking search engines and can design one from the ground up that will be fast because he's done it a few times before. So that coupled with a pretty strong community around lucene leaves it as a prime example of an open source gaining features due to expertise in an area.
Whats another example? Well I guess I can mention Safari, since I haven't for a while :). Sure Safari is also a "fork" but let's ignore that for a while. The main developer of Safari, David Hyatt, worked on the Netscape, Camino and Firefox browsers prior to working on Safari (and KHTML). The man obviously knows his stuff, and hence MacOS users get a browser that is compliant with the latest web standards (plus it's open source as well).
Whoever hired Hyatt was pretty smart. I'd bet than even if Apple didn't get the head start they did from KHTML and built a closed source browser it would still be very good. Makes you wonder if MS would try to pluck him off to work on their latest incarnation of a broswer.
Let's tie this all up. Will Harmony succeed? I think that lies in the hands of IBM (why IBM? well they just acquired a company called GlueCode who is pretty involved in the Harmony initiative). Will they allow their team to work on the JVM (apply their expertise with such fun things as JIT (which you probably don't/can't code in your spare time)) or donate code to it (ala eclipse)? If so, Harmony could be done relatively fast and would be a boon to Linux, non x86 platforms and IBM as well. If not it could be years before harmony gives us a good implementation.