Linux: The licensing puzzle

The history

This year the Free Software Foundation published a draft for GPL v3. It aimed to preserve better the freedom of the software and its users. One of the new elements was forbidding the usage of free software with forms of DRM that will deny the users their freedom to run, examine, modify and/or distribute the software. (An example of such a DRM is the TiVO device. It works under Linux, but is designed to allow running only versions that are digitally signed by the TiVO manufacturer. Thus, it denies the users the freedom to modify and then run this Linux.)

Many free software luminaries supported the changes. Some, however, did not. For example, the best-known free software project, the Linux kernel, is licensed under GPL v2, without the provision of “or any further version”. And its creator and leader, Linux Torwalds, has spoken against relicensing the kernel under GPL v3.

The official reason was that some of the developers cannot be contacted anymore, and their permission cannot be obtained. However, few will doubt that in practice Linus has the authority to make this decision. In addition, only a very small amount of the current kernel code is written by developers that are out of reach, and can be easily rewritten on need.

Linus has other considerations, too. “I am software engineer”, he says, “it’s not my job to tell the hardware producers what they should (not) do.” According to him, these new elements are politically motivated, a part of an ideological “quest foor freedom, for separating the world on ‘us’ and ‘them'”. “There are a lot of legitimate uses of software where encryption and signing keys should not be given to everyone”, he points.

This position sparkled a huge debate on Groklaw. Here I will prefer to focus on the consequences.

The second release of the GPL v3 draft softened a bit its wording, but in essence its position against the DRM has not been changed. Given that this position is the largest change from GPL v2, and the main reason for the existence of GPL v3, it’s not likely that FSF will bulge. And Linus Torwalds is still unhappy with it. The tone of Eben Moglen, the top FSF lawyer, is concilliary and soft, and negotiations are planned to continue. Maybe a compromise will be reached, and the Linux kernel will be relicensed under GPL v3. But what will happen if it doesn’t?

The development

“Linux” is actually the name not of the entire OS, but only of the kernel – a part that is very important, but is relatively small, compared to the rest. Nearly all the rest of the GNU/Linux (that is its proper name) OS is controlled by the FSF. And FSF already declared its intention to relicense all the free software it controls under GPL v3. As soon as v3 is ready, the relicensing will happen.

As a result, the kernel and the rest of the OS will still be usable together, but only under conditions that do not contradict to GPL v3. Older versions of the rest of the OS will still be usable under GPL v2, but soon will become obsolete. Anyone who wants to use an up-to-date version of the entire GNU/Linux, will have to comply with GPL v3. Those who don’t like it will have the bare kernel only.

Having only the kernel is about as good as having nothing. The perspective of writing from scratch all the rest of the OS is beyond the abilities of almost any IT company, and clearly beyond the financial interest of any. There still could be joint efforts to develop further GPLv2-based versions of the rest of GNU/Linux, but I don’t believe these will be able to hold for long.

On the other hand, developing the kernel only is an enormous, but still easier task. FSF has such a project – the Hurd. Until recently, its development was almost stalled. Technical reasons were cited, but the single most important reason was the presence of another, more developed free kernel, which attracted almost all developers.

Keeping Linux under GPL v2 may change this reason. Most kernel developers are a freedom-loving folk. A lot of them would prefer to see Linux under GPL v3. Some will directly switch to the Hurd, giving it the long-awaited boost. Others will stay, but will license their code under both GPL versions – that is, they will in practice work for the Hurd, too.

If this happens, I’d expect within about an year the Hurd to be able to rival most Linux abilities, and to be OK for most tasks. As a result, more kernel developers will be tempted to switch entirely to it. This process is self-speeding: the free kernels “market” is narrow, and the commoditization on it is very strong.

Also, all free projects have a developers turnaround. It will speed the brains leak to the Hurd: what attracts developers to the free kernels is the magic of the freedom, and the Hurd will have more of it. As it becomes as well-known and established as Linux among the programmers, it will attract almost all of the newcomers.

This is valid for the individual programmers. A lot of the Linux code, however, is novadays contributed by companies. Since GPL v2 is considered more “proprietarity-friendly”, some may decide to support Linux against the Hurd. Others will be generally indifferent. So, we may soon observe a competition for the hearts of the companies. The result of this competition may be what will determine the outcome of the entire Linux vs. Hurd thing.

The possibilities

If the Hurd wins, Linus Torwalds may be faced with the perspective to either eventually relicense the Linux kernel under GPL v3, losing a lot of face, or to see his brainchild dead. Regardless of his decision, the free community will still have the Hurd as a strong kernel. A lot of time and efforts, however, will be lost in pointless internal wars.

Another possibility for Linux will be to become more and more friendly to the proprietary demands. In theory, its relative immunity to them is due to the GPL; in practice, it is maintained by the balance between companies and free developers. If most free developers defect, ways may be found to make Linux less and less free project.

The other possibility, of Linux winning, appears to me less probable. Most of the folks that would go to the Hurd will develop because of the freedom, and will continue the work no matter what. So, in this case we will have a long-standing competition between the two kernels. Or, to be exact, standing until the first larger Linux loss.

In all cases, a lot of efforts will be lost. So, the best outcome is if Linus relicenses Linux under GPL v3, despite all of his considerations.

Don’t get me wrong. I strongly support the right of Linus on his own opinion and choice. However, Linux is not anymore his very personal, homegrown project. A huge lot of huge things depend on it. Hope he lives to the responsibility that lies with him now.

The kings learned this lesson long ago. At least the successful ones.