Larry Ellison tomorrow will eventually unveil Oracle + Sun strategy, while waiting I took the chance to ping my friend Carlo Piana and ask also his opinion on the European Commission’s clearance of the Oracle/Sun merger.
I have talked a lot about the subject and I hesitate to speak more. I am obviously cheerful for the outcome of the case, but some recent remarks leave me with a bitter aftertaste and risk to spread some more fear, uncertainty and doubt.
I don’t want to dwell much on the market issues, whether Oracle did in fact compete with Sun in the database market or other similar arguments. I formed a clear opinion about it well ahead of any merger, and people like Josh Berkush, who knows a great deal of the business, convinced me that this was a non-issue. The overwhelming evidence I have seen just confirms that. Whatever.
The part that interests me most is the role of business models versus licensing and competition. As a lawyer and Free Software advocate, I concentrate more on the games that the license permits, rather than on how people exploit them to do business. It is however a central issue.
It is a fact that the GNU GPL allows a great control over the relicensing of derivatives, including software that extends and incorporates the original one. It’s an effect and a feature of copyleft licenses, and it just works as intended. Suggesting anything contrary is simply stupid.
Now we have companies that decide to keep the entire copyright over a project and do what is called “dual licensing”. This is perfectly fine and consistent to the aims of Free Software. GNU is a perfect example, all those who say “Free Software, or open source, equals bazaar” are misguided. MySQL AB decided to do it. They were not the first, they were not the only. Sun bought MySQL AB and decided to retain that model. What is the price of this? You can steer the project, sure, but you must ask assignment of copyright, or only accept permissively licensed patches.
You cannot accept a copylefted patch whose author does not assign you the copyright, otherwise you must ask his permission to distribute the resulting combination (and to sell exceptions to it), or from then must be content with the highest common denominator, in this case distributing under the GPL.
So in a dual licensing setting you trade off third parties’ contributions to gain control. Is this still Free Software, is still
this open source? My friend Simon Phipps would say “it depends”, and he has a good “open source scorecard” to decide. I’d say it is a temporarily diminished or dormant form of Free Software. There are two exits: the sole copyright holder ceases to be such because it starts accepting copylefted patches, or a fork eventually kicks in. As I said elsewhere, forks are quite unusual because the games are against them on both sides, favouring a meeting of interests somewhere. But not impossible and not even unfeasible. Everybody acknowledges it. The GPL in particular is /designed/ to embed the solution to an hostile takeover of a Free Software project, including GNU/FSF — let alone a publicly traded company.
Here comes the FUD. The opponents to the merger (some of them, actually), speculate that a fork to MySQL would miss an essential asset, that of selling exceptions to the GPL in order to embed proprietary extensions to the application. All true, but the fact that this is under no definition an “essential asset”. It is what MySQL has lived of in its early stages, it is not a mandatory feature of all Free Software projects, it is not even what Sun is making money from (this is common knowledge, I have seen the actual numbers, but I am bound to obvious confidentiality). Well, Sun is not making a lot of money overall, and this is the problem.
I have read what you have written a few weeks ago, that the dual licensing is not the only business model for a Free Software project, and that is plain truth. If a particular entrepreneur is a one-trick-poney and relies on a single business model, an essential element of which is outside her control because she has disposed of it, that does not mean that others could not do better in innovative ways, business-wise. The question is “does a strong copyleft license allow a pure Free Software project to live”? The answer is pretty straightforward, as there are plenty of examples, from Drupal to Samba, from KDE to GNOME, that confirms so. Then also a fork can live, especially if the entrance level is lowered by the poor performance of the current project leader.
A dual licensed GPL Free Software project is software that just awaits to be sole GPL. If Free Software depends on a perpetual dual licensing strategy to live, then it is not Free Software, it is free as in “free ticket to the trashbin”, and frankly I don’t give a damn if it is directed there. All who say that the GNU GPL does not allow to do business in every area of the market, and suggest that a project like MySQL should become a weak-copyleft to make its codebase survive and prosper, are just talking rubbish, in my very humble opinion.
A number of lawyers took part in the debate, ranging from GPL’s original author Eben Moglen to people with a strong legal background like Groaklaw’s editor. Despite listening to lawyers talking about open source business maybe awkward at the beginning, but it is likely the “natural revenge” we all deserved for being so comfortable to discuss licensing agreements for years!
The benefits and issues with all this will be for another blog post, soon.