Freed Fedora Foundation Might Get Participation Boost

by Timothy Prickett Morgan

As part of its Red Hat Summit two weeks ago, the commercial Linux distributor announced that it would be creating the Fedora Foundation, an organization separate from Red Hat that would control the contributions to various open source development projects that are currently under the sole control of the company. This move by Red Hat mirrors that of Sun Microsystems with its OpenSolaris development community, which goes live today.

The ironic bit is that to better compete with Red Hat, the dominant seller of tech support and maintenance services for the Linux platform, Sun had to take its Solaris Unix platform open source, create a license that affords some sort of intellectual protection to developers, and create an organization free from Sun that could eventually--and the word "eventually" is important--control the destiny of the future development of the Solaris operating system. Sun is largely moving to a free distribution, fee services pricing model, too, just like Red Hat established many years ago. And now Red Hat has to emulate the OpenSolaris community by breaking the contributors free of the company. Red Hat might even do a better job of this than Sun. (The devil is always in the details, and given Sun's deep hold on Java, which should have long since been a truly open source, GPL-licensed technology, we'll have to wait and see.)

Red Hat was founded as a means of getting commercial support for an existing Linux operating system and the other elements of the Linux stack, and Red Hat gradually became big enough and influential enough to be a respectable contributor to the Linux platform. But with its implementations of Red Hat Linux 2.1 and 3, Red Hat essentially broke away from the Linux pack and decided what elements of future Linux code it would commercialize, and to a large extent took on the job of creating and supporting that code on its own shoulders. For instance, Red Hat brought big chunks of the Linux 2.5 development strain back to Linux 2.4, getting the jump on many of the features that were set to be available in Linux 2.6. By doing so, Red Hat got something of a jump on SUSE and other distributors who were sticking with the normal Linux roadmap and had to sit idly by and watch as Red Hat sold Linux 2.6 features on its Linux 2.4 implementation while they waited 18 months or so to get the real Linux 2.6 out the door. But, when Linux 2.6 was available, Red Hat ended up lagging as it had to wait about six months between when Novell and others got Linux 2.6 out and when it could get Linux 2.6 to market. Now that Red Hat is more or less in lockstep with the both the development and production Linuxes offered by its commercial competitors, it will be interesting to see what the newly formed Fedora Foundation will do in terms of roadmaps and feature adoption.

Almost two years ago, Red Hat set up the Fedora Project as a means of giving its engineers and other Red Hat enthusiasts a means of testing out alpha and beta code and of shaping the future development of Red Hat's desktop, workstation, and server products. But the Fedora Project is sponsored by Red Hat, and the contributors are for the most part Red Hat software engineers. So it is not exactly free from encumbrances, and the meritocracy of the Fedora Project is necessarily dominated by the Red Hat engineers who contribute. (The same will hold true of the OpenSolaris community, which will be dominated by the 1,000 software engineers that Sun has and who will gradually move from being software engineers paid by Sun to create closed source Solaris to being software engineers being paid by Sun to create OpenSolaris and test and certify the commercialized version of OpenSolaris, which will be called Solaris, as it is today.) Through no fault of its own, the Fedora Project is not something most coders are going to feel empowered to join. Software developers may be joiners, but they join things to contribute, and if they don't think they will have any sway in the organization, then they probably won't join it.

So having the Fedora Project go one step further away from Red Hat to become the independently run Fedora Foundation is, in terms of boosting contributions from outside of Red Hat, a pretty good idea. But then again, so is the idea of having the smartest coders in the world working on Linux development itself with Linus Torvalds and all of the other thousands of open source projects that eventually end up contributing the thousands of applications that run on Red Hat, Novell, Mandriva, Turbolinux, Debian, and other Linuxes. While I am convinced that a free Fedora Foundation--one might even shorten that to simply the Fedoration--is an improvement over the Fedora Project, the simple fact remains that what Red Hat does in terms of development and integration beyond what the open source projects offer themselves isolates Red Hat as much as it links it to the open source community. In a proper world, Red Hat would not have to be necessary and open source projects would be developed with the necessary integration and testing principles built in from the get-go; all certification would be done by the individual developers of each project, or that responsibility would be delegated to a neutral, communally funded body. A kind of Switzerland. If Red Hat, Novell, and all the others have their own arms-length development communities, the logical thing to do next is to get economies of scope and scale and merge them all into one giant Linux development and open source certification organization--which is what Open Source Development Labs should really be, in my opinion.

But, you can't go public on such a business model, and you cannot control customers with it either. Which is why it never happened. And just so I am not misunderstood, the way that Red Hat, Novell, Mandriva, and others create their Linux implementations is orders of magnitude more transparent than the way Unix platforms were developed, and is a monumental improvement that has, in fact, forced Sun to change its tune with Solaris. And if Hewlett-Packard and IBM had any sense--and the proper licensing to allow it to do so--they would take their Unixes open source and give them away while there is still a chance that they might survive and create a community as Linux has done (with perhaps less stratification, though).

In any event, Red Hat has promised to create the Fedora Foundation and will eventually move all Fedora Project development work--which includes the core Linux platform plus the desktop, SELinux, and stateless Linux (for appliances) variants created by Red Hat, documentation for the products, internationalization and language translations, new support for hardware virtualization features, and the new Fedora Directory Server I told you about last week--to the Fedora Foundation. Importantly, Red Hat says that code contributed through the Fedora Foundation will be secured by copyrights owned by the foundation--not by Red Hat. This is Red Hat's way of proving that it is really taking a hands-off attitude for the Fedora Foundation.

Red Hat used the launching of the Fedora Foundation as a springboard to talk about what it really wants, which is a reform in existing software patent policies in the United States and the evolving ones in Europe, which are heading in the same direction as software patents here in the States. Red Hat wants software patents awarded under a much higher scrutiny than they currently are--think about how stupid it is to be able to patent one-click buying--and to allow for more legal questioning to get bad software patents reversed. Mark Webbink, Red Hat's deputy general counsel, also said that Red Hat wants to create a software patent commons, where software that was patented could be donated to a foundation and kept out of the control of any commercial entity, which would be apt to sue to protect its patents. Such a software patent commons would be like lawyer repellent--if you have a large portfolio of dubious software patents (and they are dubious by their very nature, not because you did anything wrong, because software patents are in many ways silly), then another company will think twice about suing you for infringing its portfolio of probably equally dubious software patents. What happens today is the lawyers make a little money as two parties rattle their sabers, and then they end up cross licensing all the patents any way. We could just cut to the chase scene and create a software patent commons and let anyone use them.

"Patents are not equal to innovation," said Webbink in a statement proposing the idea of the software patent commons. "More often, innovation occurs despite patents. What we observe today in the software industry is the use of patents to maintain market share, even where that market share has been obtained by anticompetitive means. We need to move away from a system of software patents compromised by trivial, incremental enhancements that block innovation, to a system that is aimed at rewarding substantial innovation." He said that Red Hat has put up its own portfolio of software patents to protect coders against infringement lawsuits, but did not detail what those patents are. He did, however, provide a link to Red Hat's position on patents, which you can read here.

IBM, Novell, and Red Hat have promised not to sue any open source projects for infringing their patents. This is obviously not the same thing as setting up a software patent commons, which means giving away the rights to hard-won software patents to the open source community. It is hard to let go, but vendors might eventually have sense to do it. But patents have been such a strategic advantage that it will be very tough for them to do it.