February 09, 2004

SCO, Linux and Freedom

As a long-time user of Linux (I've done almost all of my work on Linux since 1995) and a member of the free (as in "free speech", not "free beer" - click here for a discussion of the distinction.) software movement, I'd like to expand on Geoff Pullum's comments on SCO's attack on Linux. First, a bit of background. Linux is a Unix-like operating system kernel, originally written by Linus Torvalds, subsequently developed under his direction by a large group of volunteers. Unix was originally written at AT&T Bell Laboratories. It is usually used together with software produced by the GNU Project, the founding and leading light of the free software movement, a combination referred to by some as GNU/Linux. The Unix trademark currently belongs to The Open Group. The copyright to a version of the Unix code that originally belonged to AT&T was transferred via a chain of transactions to the Novell Corporation. SCO purchased certain rights from Novell, including, according to SCO, the Unix copyright.

There are actually two things at issue. First, SCO has sued IBM, claiming that IBM contributed to Linux intellectual property belonging to SCO. Originally, this took the form of a trade secret claim, in which SCO claimed that IBM had put SCO trade secrets into Linux. Judging from SCO's filings with the court, they appear to have abandoned this claim and to have shifted to a copyright claim. This claim does not involve the old Unix code. Rather, SCO's claim is that under their contract with IBM, IBM's own work on its version of Unix constitutes a derivative work of the original Unix code and that therefore IBM has infringed SCO's copyrights by contributing its own code to Linux. One of IBM's defenses is that SCO is just plain wrong about the contract and that SCO has no claim to IBM's code.

Second, SCO has been claiming that large amounts of Unix code to which it owns the copyright have been copied into Linux and that it is therefore entitled to demand license fees from Linux users. SCO hasn't yet tried to sue any Linux user; in fact, it hasn't sent out any invoices. As a result of SCO's public claims, Red Hat, a major Linux distributor, has sued SCO for damaging its business.

One major issue here is of course whether there is any infringing code in Linux. For the most part, SCO has refused to identify any. They originally claimed that the infringing code had been identified by a team of MIT mathematicians. When pressed to identify them, SCO first retreated to the claim that its team had formerly been associated with MIT, then just became silent. It looks like they just made them up. Anyhow, SCO eventually presented a couple of snippets of allegedly infringing code at a trade show. Even then, they obscured the code by using Greek letters. This pathetic attempt at encryption was broken without delay, and the code shown not to be infringing. You can read Bruce Peren's analysis of the allegedly infringing code here.

More recently, SCO identified some more allegedly infringing code. This turns out to consist entirely of header files that define what is called the Application Binary Interface. Everybody else believes that these files are in the public domain and not copyrightable.

The other major issue is whether SCO in fact owns the copyrights on the UNIX source code that it claims has been incorporated into Linux. Novell claims that it did not transfer these copyrights to SCO. SCO has sued Novell for "slander of title" but the case has yet to come to trial. On my reading of the contract, Novell is right.

SCO also has a problem because until very recently SCO programmers made contributions to Linux and SCO itself distributed Linux. That puts them in a very poor position to complain. Furthermore, by demanding a license fee for Linux, SCO has violated the GNU General Public License (GPL) under which Linux is distributed, which automatically terminates SCO's right to distribute Linux. In response, SCO has alleged that the GPL is invalid. All of the legal opinions that I have seen are that SCO is spouting nonsense. See, for example, this article, entitled SCO: GPL Unconstitutional. Lawyers Scratch Heads. If the GPL is invalid, then SCO has no right to distribute Linux at all.

SCO would like everyone to think that it is angry Linux advocates who are responsible for the MyDoom worm, but this is almost certainly not the case. (Here's a good summary of the evidence.) To begin with, analysis of the worm's code shows that only 1/4 of the machines infected by MyDoom will launch distributed denial of service (DDOS) attacks on SCO. After attacking SCO, MyDoom will attack users of the Kazaa peer-to-peer file sharing network, which has nothing whatever to do with either SCO or Linux. MyDoom creates "backdoors" in all of the machines it infects, allowing them to be controlled from the outside.It also installs a keylogger, allowing crackers to see everything the user types, such as credit card numbers and passwords. Another piece of evidence is the fact that the volume of email containing MyDoom was so large from the outset that only spamming techniques could have created it. The evidence is that MyDoom was created by professional criminals for the purposes of spamming and identity theft. The DDOS attack on SCO is simply a diversion.

[Update 2004/02/10: According to this Eweek story, the latest version of MyDoom is now circulating. This version attacks Microsoft's main web site. It does not attack SCO.]

There are some excellant sources of information on the SCO affair. The Free Software Foundation has issued a statement and the Open Source Initiative has issued a position paper. The single most important source is Groklaw, which includes frequent news updates, extensive commentary and analysis, and an archive of documents. Groklaw is an amazing and inspiring phenomenon in its own right. It started out as a blog by Pamela Jones, a paralegal and Linux user. Her thorough research and penetrating analysis attracted a large readership, and the site turned into a collective, volunteer research effort. Groklaw members report leads, scan and format legal documents, and send in eyewitness reports from sessions of the court. Groklaw is an outstanding example of the role that the network can play in the functioning of a free and open society.

What does this have to do with language and linguistics? Well, in part I'm going on about it just because it is important, even if it doesn't have to do with language. SCO's attack on Linux and more generally on the free software movement, is an attack on freedom. But there is a more specific connection to language, namely the role that free software is beginning to play in the maintenance of minority languages.

The companies that produce proprietary software are out to make money, and they generally aren't terribly interested in small markets. At best, they get around to it when they feel like it. The result is that proprietary software tends to be available only for major languages. It is much easier to adapt free software to smaller languages, where by "small" I mean both languages with small numbers of speakers and languages whose speakers, though comparatively numerous, are too poor or otherwise disadvantaged to be well served by proprietary software. In some cases, proprietary software vendors have moved to adapt their software to small languages only after free software did so. This article describes how Microsoft released Welsh versions of its software only in response to the creation of Welsh versions of Linux and other free software. Here are web sites where you can read about how Linux is being adapted to the languages of South Africa and India.

Only free software was used in the generation of this post.

Posted by Bill Poser at February 9, 2004 12:52 AM