Editor's note: This article came from detailed notes, though not a full transcription of the event. As a result, some of the names and dates may be incorrect. On 03 August 2004, Jonathan Zittrain clarified some of his comments further, as you can read below.
On Monday night of OSCON, several open-source-friendly lawyers held a mock court to debate SCO's recent Linux-related lawsuit with IBM. While the lawyers and audience itself were definitely most sympathetic to the opposite position, the idea that SCO has no copyright claims over the Linux kernel, as host Pam Samuelson pointed out, is a position that any good lawyer can reasonably argue.
Jon Zittrain, from Harvard Law School, who worked on Eldred v. Ashcroft, argued the pro-SCO side. Pam was careful to point out that Jon much preferred the IBM position, but since neither IBM nor SCO laywers could comment on the case without facing censure from the presiding magistrate, and she could not find anyone else willing to argue SCO's position, Jon took the job. Mark Radcliffe, an IP laywer from Palo Alto, California, argued the pro-IBM position.
Jon started by explaining where SCO, at least the current iteration, came from. He traced it to Caldera, mentioning that the initial public offering in March 2000 produced a market value of $1.1 billion dollars. Today it's worth $61 million, thanks to the inappropriate, wrongful, and legally actionable IP theft from many parties, including IBM.
Jon next explained a brief history of Unix, starting at AT&T's Bell Labs in 1969. AT&T failed to make much of Unix until 1984, when the court-ordered breakup allowed them to pursue other markets it had left alone due to anti-trust concerns. Unix had a very collaborative and loose development, especially with universities. With the anti-trust concerns out of the way, AT&T tried commercial development.
You might remember that Richard M. Stallman started the GNU project also in 1984, intending to recreate a Unix-like system without using any Unix code.
AT&T owned almost every aspect of Unix and entered into some 80,000 licensing agreements with around 6,000 institutions. Jon claimed that, oddly, there were really no licensing templates -- these were very unique licenses. Thus, any licensing dispute with any group depends on the specific language of the license. He gave the impression that this was a mess, since decisions made on any one dispute would translate to other licenses only with difficulty.
Most importantly, IBM and AT&T made a deal where IBM could see and modify AT&T's SysV Unix code. IBM could redistribute any derived software it made but not the source code itself. IBM had to pay royalties and keep the source and aspects of operation of Unix confidential.
SCO's chain of title to Unix (at least as the company claims) is somewhat Byzantine, Jon said. In September 1995, Santa Cruz Operation (SCO Zero, not the current SCO One), bought almost all of AT&T's rights to Unix, apparently from Novell. Novell, retained 95 percent of the royalties from existing licences, but also retained veto power over licensing.
SCO maintains that Novell's power applies only to continuing licenses, not new ones. If this is true, Novell has no power over new licenses that SCO negotiated after this agreement.
Part of the confusion comes from the amendments that were eventually attached to the SCO-Novell agreement. Things started to fail with amendment one, which listed exceptions to the agreement. Purportedly, it left out all of the copyrights. According to Jon, SCO claims that this was a typographical error, corrected a year later in another amendment, itself confusing and mostly unreadable.
Interestingly, Darl McBride, now head of SCO, negotiated that contract--for Novell. He's prepared to testify on behalf of SCO now though.
SCO took no action until after the head of IBM software gave a keynote at LinuxWorld in January 2003, committing IBM's Unix and AIX expertise to improve Linux. This surprised SCO, who decided to play David to IBM's Goliath. (With a valuation of $143 billion, IBM is definitely larger than the $61 million SCO.) Jon explained that SCO merely wants to protect its rights as negotiated in IBM's agreement with AT&T.
According to Jon, SCO contends that:
SCO's ongoing discovery process is attempting to figure out exactly which code is infringing. SCO's interpretation of copyright law suggests that copyright protects not just literal code, but the sequence, structure, and operation of a program. So Linux doesn't have to copy Unix code line by line to infringe.
Even if they haven't shown specific code, Jon continued, SCO can still claim that there is infringement of certain ideas, methods, and concepts of SysV, with the reasoning that this is why IBM was so certain it could help improve Linux.
Even if copyright law doesn't cover this, IBM's contract with AT&T does. This contract prevents IBM from divulging this code.
Jon then provided musical and publishing examples of copyright precedents, citing the case where a jury decided that George Harrison subconsciously infringed on the song "He's So Fine" while writing "My Sweet Lord" and a case where a magazine published only a few words from Gerald Ford's upcoming autobiography but still infringed because it published the only words anyone would want to read from an otherwise turgid memoir.
SCO's argument, as Jon put it, is that "the infringing code is the heart and soul of Linux and Unix."
The rest of SCO's position seems somewhat weaker.
For example, as proof that the case has merit, Jon suggested that SCO's lawyers believe in the case. Of course, they charge an hourly fee for the case much lower than the industry average--in exchange for 20 percent of the judgments, acquisitions, and financing that result from the case. In October 2003, when SCO raised $50 million, the laywers netted around $8.9 million. Clearly, they're willing to put their money where their mouths are.
As to the claim that Linux developers could rewrite the infringing code, SCO claims that this does nothing to satisfy the infringement. There are smaller penalties for unwitting infringement, but there are still penalties. Even though SCO failed to register its copyright (as the AT&T code came about before the U. S. signed the Berne Convention, giving authors automatic copyright protection without manually attaching a license to every copyrighted item distributed) before filing the suit, it still deserves statutory damages and can claim those in figure suits.
Jonathan Zittrain writes:
I think what I meant so say was: Thanks to the Berne Convention (and the Copyright Act of 1976), one no longer needs to formally register a copyright in order to enjoy protection. However, one must register before suing in order to have access to "statutory damages" -- damages in the absence of actual harm. SCO didn't register before suing IBM, so cannot claim statutory damages for copyright in its case. Of course, now that it's registered, any new cases could have such damages contemplated if all the other elements of infringement were proved.
Jon also mentioned some of the justification for the other suits and SCO's Linux licensing program. Most importantly, SCO believes that the open process of development used for the Linux kernel does not protect against improperly copied code. On the other hand, its amnesty program shows that SCO has a sense of mercy and compassion.
Of course, some commentators (as well as IBM) have pointed out that SCO has actually distributed Linux and related software under the GNU GPL. Jon explained that SCO's position that the GPL violates the U. S. Constitution. By the way, they didn't know then that they owned parts of the IP in Linux. They were angry to discover that.
Jon finished by describing the open source crossroads. It's a victim of its own success, with increasing mainstream acceptance by large corporate behemoths. Lawsuits are a fact of life in a pond this big. SCO's advice to developers is to get used to this.
There are many developers building groups, such as the Open Source Risk Management group, to offer protection from and education about copyright and ownership issues. Alternatively, users have the option of paying the $700 license fee. Otherwise, you can take your chances.
Jonathan Zittrain explains further:
The $700 license gets a promise from SCO not to sue for simply running Linux, but it requires the licensee to agree not to make any changes to the Linux source. So it's not much of an option for someone who actually wants to code!
Jon stepped slightly out of character (or put his tongue more firmly in his cheek) to explain that the nature of the failed dot-com company is to sue. If you've contributed to code without expecting remuneration, you'll have to wonder when they'll sue you too.
Of course, SCO hasn't yet sued anyone without a larger market capitalization than their own.
Mark Radcliffe then took the stage, to argue IBM's position. As you might expect, he started by saying that SCO hasn't given the full story in some respects.
Mark framed his talk rhetorically around the phrase "strong magic". SCO needs strong magic to make the claims that it's making. Remember, this is a business that has failed twice, first with Unix and then with Linux. He suggested that SCO's management has asked "How can we turn this smoking crater into a business?" How can we turn an antique Unix codebase into Unix code?
At its base, Mark explained that SCO's claim is a contract dispute. The meaning of the contracts is vitally important. Though the contracts are messy in some cases, they're very clear in other cases.
Unfortunately, the contracts don't fully define what SCO bought. They did buy the UnixWare business, but the contracts define it unusually. They did not buy the copyrights, as they appear in the original list of excluded assets.
Amendment 2, issued a year later, attempted to change this, but being an exception to an exclusion, it's not clear. IBM believes that the exception to the exclusion of copyrights does not apply in this case. (This was difficult to follow, perhaps a sign that the contracts are especially strange.)
One of SCO's biggest claims against IBM is that IBM violated its confidentiality agreement with AT&T. However, SCO voluntarily dropped its trade secret claim earlier this year.
Mark put forth the argument that claiming trade secret protection for Unix is impossible. To keep a trade secret, you must take reasonable steps to protect the secret and the secret must not be widely known in the industry.
Of course, Unix is widely known. AT&T distributed Unix widely before the anti-trust agreement 20 years ago. After 1984, when AT&T tried to make a business out of selling Unix, the USL-BSD suit failed. Now that this confidentiality has gone, SCO cannot retrieve it.
Mark continued with a series of explanations of why the infringements do not apply. Even if they did, however, IBM supports Novell's conjecture that Novell has the ability to waive violations. Novell sent IBM a letter formally waiving any restrictions on Unix on SCO's behalf. That's very important.
Even if there were a violation, Novell has waived it for IBM.
For example, in Amendment X to the Novell-SCO agreement, IBM acquired "perpetual and irrevocable rights" to Unix. Mark claimed that IBM in fact has a statement from SCO explaining that restrictions on the use and confidentiality of Unix do not apply to IBM's code.
The biggest question so far is what kind of code could and did IBM share with Linux. Is this actually Unix, or is it derivative code, to which SCO can claim some rights?
SCO has made no statement as to which this code actually is. IBM believes that Linux is not a derivative work of anything SCO can claim copyright over. Even if it were, it wouldn't matter, as Novell has waived any violations on SCO's behalf.
SCO has claimed the authority to terminate IBM's ability to distribute AIX. Unfortunately, that flies in the face of IBM's "perpetual and irrevocable" rights to distribute Unix. There's a long history of contracts in which "irrevocable" has never meant "revocable"--that's why IBM paid millions of dollars for the license.
In any case, if there were a violation, Novell's letter has waived it anyway.
Even if SCO didn't realize that it wanted to claim ownership over Linux, it distributed Linux under the GPL voluntarily--the very product that it claims violates its rights. Since SCO did it voluntarily, it cannot rescind the permission it gave.
Besides that, consider that anyone who can demonstrate a legitimate claim of overship to code distributed by SCO under the GPL--which SCO now considers infringing--may have a right to damages for SCO violating the GPL by putting additional conditions on the code that the GPL does not allow.
As did Jon, Mark explained that the contracts at the heart of this contractual dispute are somewhat odd. Again, Amendment 2 is an exception to an exclusion that allowed SCO to assert its rights if necessary. If they were truly important, why did SCO wait so long?
Mark proposed that SCO didn't care until it entered the lawsuit business, pointing out that SCO's lawyers had bet heavily on the first version of Napster as well as Aimster. He left the question of SCO's judgment on copyright issues to the audience.
More importantly, the traditional language a lawyer would expect to find in a contract for transferring copyrights actually does not appear in the SCO-Novell contract. Since lawyers try not to transfer contracts accidentally, this appears to be deliberate. Perhaps the parties intended to transfer some copyrights in the future, but IBM does not believe that the current language of the copyright performs this.
As mentioned many times before, there's a very strong possibility that there are no existing copyrights to Unix as owned by AT&T after the 1984 breakup. Again, the USL v. Berkeley lawsuit was an embarassing trainwreck for USL. AT&T had paid no attention to its confidentiality or the formality of copyright laws. The company tried to enforce this copyright, but it had distributed many copies of Unix without the license attached. Think of it as a disregard for their copyright--if they have done that enough, Unix would be known widely and USL would have little room to start enforcing the copyright.
Mark then attacked the similarities to music copyright cases. The ratio of a few seconds of melody found in a three-minute song is immensely greater than the ratio of a few hundred lines of code to a ten-million-line operating system. Even if this code does infringe, it's hardly substantial--besides, Novell has waived the violation.
IBM interprets Amendment X to mean that it can use the methods and information it learns from Unix to make other operating systems. IBM believes that it has done that. If so, the company's position is very strong.
Mark then repeated that SCO has neglected its duties under the GPL by trying to place additional restrictions (its licensing fee, for one) to code already licenced under the GPL. This is actually a copyright violation.
Worse, SCO has neglected its duty to its shareholders by wasting millions of dollars on stupid cases instead of building a viable business. In the end, SCO brought the case to a court of law, not a court of magic. SCO needs to prove its case on facts and the law but have failed to do so.
The end result was enlightening. While SCO's claims had always seemed somewhat dodgy, I had never really understood how Byzantine its claims were--especially in relation to its contract with Novell.
On the other hand, as the question and answer session at the end reinforced, it's uncommon for open source project leaders to make sure that contributors have the right to make unencumbered contributions to their projects. The initial paperwork to vet the copyright of patches is daunting, but it may be legally necessary for now.
Of course, as one questioner pointed out, it's very easy for open source and free software to make its way into proprietary products, never to be found. One large benefit of publicly developing software--the transparency of the process--makes it easier for companies to claim potential (or real) infringement.
The real solution will likely include continuing to fight bad copyright, patent, and trade secret laws and judgments, working with intellectual property lawyers who want to do the right thing, and being very careful that copyrights on distributed code are very clear.
chromatic manages Onyx Neon Press, an independent publisher.
Return to the OSCON 2004 Coverage Page
Copyright © 2009 O'Reilly Media, Inc.