Mathew Aslett from 451 Group and Tim Bowden are making the case that selecting an open source license influences the commercial success of an open source proejct. From Tim’s blog:
‘When it comes to takeovers and valuations, I think the role of GPL as a strategic weapon is often under appreciated.’
I would like to suggest an alternative, and more important, strategic weapn: trademarks. MySQL was purchased for $1 billion because it is a very well known and successful trademark; how can you miss by being the ‘M’ in LAMP. JBoss consistently protected their trademark and was purchased for $350 million by RedHat. Sun knows the value of good trademarks and their protection of OpenSolaris is causing problems in that community. However, there are also examples of non-gpl licensed project and companies: Sleepycat was purchased by Oracle, Zimbra by Yahoo, although still independent SpringSource is doing very well with their Apache licensed Spring project.
GPL is a fine license but is not the secret for success. Having a well known, protected trademark is what really drives success. Owning the trademark to a project allows that company to leverage it for profit; to the exclusion of others.
How do you develop a strong trademark? Start with great technology and a development/support team that is passionate about building a community. The choice of license is secondary.
Ian Skerrett 
Ian, Without a doubt building a strong trademark is an absolutely essential requirement to building a strong business. I’m firmly of the opinion though, if your code is BSD licensed rather than GPL’d, you’re making it much easier for your competitors. I guess without hard numbers (difficult as they would be to collect) we’ll have nothing but (presumably) informed opinion and gut feel.
I still believe license is more important than trademark. A license clarify and governs the relationship between the provider and recipient. To software developers, it is the “bottom line” and the basis of any decision to adopt a software or not.
Microsoft had a weak trademark called “Windows” (They actually managed to get a jury to rule that the “Windows” is too generic in the English language to qualify as trademark and had to pay Lindows a lot of money to get it to change name to Linspire), that still did not stop Windows API to be an allergy to a substantial number of people in the Free Software camp.
“Free software” is not a trademark, and I think FSF will have problem trademarking it. As a concept it is probably bigger than GPL, but it will certainly fail your successful trademark definition.
However, I can see the importance of “Trademark” if we look at the business side of software development and distribution. From a developer’s viewpoint, I take license over trademark any day.
Tim,
Thanks for the comment. I am not really sure why it is easier for competitors? If they fork the project, they will need to give it a new name. If they try to offer services, they won’t be able to use the trademark. I might be missing something but I just don’t see how the license selection impacts your competitors.
ctrambler,
My statements were more in the context of business side, ie. how do you increase a company valuation; sorry I wasn’t clearer. I agree some developers are more focused on licenses. However, I do think a strong developer community leads to a strong trademark and this brings in more developers.
Ian, when the underlying project is GPL’d, every vendor is operating on more or less the same code base. No significant differentiation can be done between vendors products.
With a BSD licensed project, any vendor can release a version with exclusive proprietary features, that other vendors will take time and expense to duplicate, only to see the original vendor of that feature then release their code into the project code base (if they’re smart) thereby cementing their dominance (in the short to medium term at least) over that part of the project turf.
If you want to know what competition in a BSD based project looks like, you could do worse than study the UNIX wars. Lots of almost-but-not-quite compatible versions from many vendors, non of whom could cement their dominance of the UNIX market. None with any great incentive to contribute back to the shared UNIX core (ok, so it wasn’t organized in the same way most open source projects are today, but I still believe it’s a valid comparison). In the end, the inability to dominate the market hurt them all.
Now look at Linux. Want to take on Red Hat in the enterprise space? Forget it. Second place Novell has no chance of catching up. Whatever feature Novell may wish to add to Linux, Red Hat automatically gets it too. Certainly Novell can offer the same product as Red Hat (with some delay as their engineers get on top of the latest features), but why bother changing vendors then? There’s nothing much they can do to effectively differentiate themselves. Sure, they can offer better service (on substantially the same product) but even there Red Hat’s got an out. Don’t like RH’s service? Buy RHEL from HP, or IBM, or Sun (or …). Either way, RH gets some dough.
Has a “Red Hat” emerged in the post UNIX wars *BSD market? Not a chance. No BSD vendor can prevent another offering a better BSD, and no vendor is able to prevent a re-run of the UNIX wars in the BSD project space. Result: no vendor is able to rise and dominate BSD. I believe that’s broadly a pattern for any BSD licensed project.
One exception to my argument I think, is where the product is being applied in different markets (by type, and /possibly/ by geography where there are significant differences in market culture). While Red Hat rules the enterprise Linux Market, it looks increasingly like Canonical (with Ubuntu) will carve out for itself the top dog spot in the Desktop Linux space. Mandriva (formerly Mandrake Linux) tried snaring the desktop linux spot some years ago, but the market wasn’t ready for it. Everyone’s held back since then, but it looks like Canonical may now have picked it right. If they manage to make something of it, how do you challenge them for it? You can’t offer anything they can’t offer because of the GPL. I believe it will become a re-run of the enterprise linux play, only this time it’s going to be Canonical with Ubuntu on the desktop. I just can’t see any way an alternative vendor can effectively compete and win in those circumstances.