Mudblood open source
I talked with a prominent open source developer - Gianugo Rabellino of Apache/Cocoon fame - today, getting his feedback on the business models of Red Hat, Alfresco, MySQL, SugarCRM, JBoss, and others. He made it clear that the open source development community (if it, in fact, exists) strongly prefers pureplay open source projects, rather than "mudblood" open source companies.
No surprises there.
I then spoke to a group of architects (quite senior) from Fortune 500 companies at a recent OSDL advisory meeting. I asked them the same question: do you care if your software is open and, if so, how open is open? Almost universally, I was told that everything should be 100% open (and they then gave some good reasons as to how this would influence their decisions). They even said they'd pay for it, but I didn't believe them on that one, human nature being what it is. :-)
What wasn't clear to me in either conversation is how to quantify community involvement. None of the companies above made the pureblood cut, and yet all are making healthy revenues. Is this because each is a sell-out? Profitable sell-outs, but sell-outs nonetheless?
His suggestions?
- Hire a prominent open source developer. Google did this with Greg Stein (and Chris DiBona). Novell did it by acquiring Ximian (with Nat and Miguel as part of the package). I'm not convinced that these or other developer-hires did much for the top lines of these companies, but certainly Nat's and Miguel's involvement with Novell gave it a revived freshness (which now only real revenues can sustain).
- Actively participate in relevant open source communities. Alfresco, for one, has been doing this, but I think we could be doing better. And maybe MySQL could have benefited from more active InnoDB involvement...? :-) I know that RightNow actively contributes code to the HtDig project, as well as bug fixes to MySQL, too. Is this enough to make these companies "pure?"
- Open up all of one's source code. No "baitware," he said. So, no RHEL vs. Fedora, Community vs. Enterprise, etc. This sounds great in principle, and I admit to being biased toward openness in software, but I'm not convinced that the developer community that Red Hat lost by moving to the RHEL/Fedora split was not more than made up by the committed developers and customers it gained as a result.
And so, the question: does community matter? Or, rather, does the pureblood development community matter?
As I've written before, of the top 50 contributors to SugarCRM's development community, 95% are SugarCRM partners. Are they tapping into the pureblood open source developers? Maybe not. Are they tapping into an audience that matters and actively contributes? Yes.
Despite my apparent assurance on the matter, I'm anything but convinced by my arguments. If you have any ideas on how open "open source" should be, please ping me.
2 comments:
IMO, the "openness" is not about the availability of source code, but about the openness of the community.
An opensource product with a closed community controlled by a single company does have the same kind of lock-in consequences as traditional software vendors. Also some software vendors allow their customer to buy a source licence, thus effectively giving them access to the source code (with some strong restrictions about what they can do though).
Now community diversity can perfectly be achieved through commercial companies diversity, as it gives users the choice when they need support or need some new features in the product.
This is what happens in communities like Apache, where contributors, although obtaining committership as individuals, often contribute as part of their paid work. So many companies effectively have the ability to provide expert support and services, thus giving the choice to users.
On the other hand, there are also many GPL'ed products developped by a single company. That company "owns" the contributors that are its employees and thus the intellectual property of the product, which is what permits the dual-licensing model ("comply with the contraints of GPL or buy a commercial license"). And this doesn't motivate third-party contributors as they have no real benefits when contributing.
So the license has a big influence of the community that gathers around some software. And I'm happy to see Alfresco using MPL!
Back to customers, I think they currently don't really make the difference between the different kinds of opensource communities. I regularly give some "what is opensource?" seminars and people are often incredibly amazed when they see how these communities work and are organized.
So we're in a transition phase, and customers are slowly being educated. In a few years from now, they'll probably understand what openness really is, and how it ensures choice and long-term vitality.
In Response to:
An open source product with a closed community controlled by a single company does have the same kind of lock-in consequences as traditional software vendors.
--
Lock in is not a product of open verses closed, controlled verses community driven or even STANDARDS!
Lock in is due to cohesion of code.
You can be locked in to the JCR, Portlet standard, LDAP whatever.
There is a myth going around that open source can help prevent lock in. Only you can prevent lock in.
You choose the level of dependence when you make your interface decisions and that's all there is to it.
Licensing is a really problem space. Licensing is an issue because of the sheer number and variety which exist. I think Microsoft seems to understand this best. The are creating the "shared source" model with what seems to be a very few well defined licenses.
I was also glad to see Alfresco choose MPL, it will make a big difference in their community building efforts. Its hard to argue with a license which is essentially to my understanding (rape and pillage my code, but don't take credit for my work.)
Post a Comment