What did SCO buy--Unix or the Brooklyn Bridge?

To the outsider, and even to me for some time, the various lawsuits involving the SCO Group follow a relatively simple story line. I'm not a lawyer, but after many interviews with the involved parties and lawyers, the case boils down to divergent interpretations of the subject matter.

The Linux operating system was coming on strong. A group of individuals, some with a history of using litigation to extract wealth from other parties, started to behave like pit bulls. They clamped onto the Linux ecosystem in hopes of exploiting it for riches that for the most part have not accrued to anyone. Recall that Linux and open source are a free "David" in a world of "Goliaths."

The basis of entitlement for those rocking the open source ecosystem is a handful of copyrighted lines of source code. To the untrained eye, it seems like a bit of a stretch considering what percentage of the overall code in Linux the alleged offending code constitutes. Confirming that cynicism are legal experts, vendors, and some of the open source community's squeakiest wheels, who in their writings, speeches, and interviews are making the original claims seem all the more implausible. The common view is that SCO picked the wrong fight and will probably lose.

Based on what I've heard so far from the parties I've interviewed, I'm not nearly as convinced as others are that this will be a landslide against SCO. Nor am I convinced that SCO will win. What I am convinced of is that all sides believe in their hearts that every word of what they're saying is true and that the cases will primarily come down to a judge's interpretation of the terms of Novell's sale of Unix to the SCO Group. What exactly did SCO buy and what is it entitled to as a result of that purchase?

In the case of SCO vs. the World, it isn't difficult to go with the flow and side with the World. After all, when you consider the scope of the outrage that the SCO Group has provoked, the passionate and logically argued treatises from the open source community's greatest thought leaders, such as Eric Raymond, Bruce Perens, and Eben Moglen, and the somewhat singular and universally defiant scoff from the vendor community, the SCO Group must be nothing more than a group of brazen hoodlums with a really bad case of bravado.

On the other hand, if you're even slightly familiar with the role that intellectual property plays in the computer industry, you can't help but be dogged by this question: Why on Earth would any company pay more than $100 million for a technology --as the SCO Group's ancestry did for Unix--if it knew the technology could be opened to the kind of cloning that could irrevocably jeopardize the potential return on that $100 million investment? The SCO Group might as well have bought the Brooklyn Bridge and tried to charge tolls.

Cry foul at SCO all you want for trying to rip the rails out from under the little engine that could (and could bring down Microsoft a notch). Write white papers, blogs, and draw family trees of Unix. Hem and haw about public licenses. At this point, I don't believe that much of the rhetoric and assertions by technology, open source, or self-proclaimed legal experts matter. One "fact" that leads me to this conclusion is that some of the people I've interviewed-- people who were speaking as though they were experts--were themselves confused about some of the cases' simplest facts and, miraculously, even who the litigants are.

Right now, the easiest way to get your head around this very complicated case is to frame it with two primary vectors. Both will probably come down to interpretation. It's about whether SCO owns it, to what it extent it owns it and, if it has some or all of the ownership, just what exactly it is entitled to.

In the first vector, SCO must get a judge to agree with four core beliefs: (1) what is Unix, (2) how much of that intellectual property is owned or controlled by SCO, (3) whether by virtue of the answers to 1 and 2, certain Unix licensees must answer to SCO in terms of fulfilling their contractual obligations, and (4) what those obligations are and, by being party to the development of Linux, were certain licensees in breach of those obligations. SCO's claims on all four fronts are being contested.

In the second vector, SCO must again get a judge to agree on what Unix is and how much of it is owned or controlled by SCO. But, whereas the first vector involves Unix licensees and their obligations, this vector involves any party -- licensee or not -- that may have misappropriated what is owned or controlled by SCO and contributed it to Linux.

For both vectors, after establishing exactly what Unix intellectual property (IP) is owned by SCO and what the nature of that ownership is, SCO must also convince the court that the relevant IP was used during any point in time to establish Linux as an operating system.

SCO interprets the final language behind the purchase of Unix from Novell one way. Novell sees it differently. Now, it's up to a judge to decide. In the end, someone--probably one of the lawyers who helped to draft the agreements between Novell and SCO and who then explained those agreements to either Novell or SCO's top brass--will be proven to have made a critical and stunning mistake.

Ironically, in my on-the-record and off-the-record discussions with the two litigants, both companies referred to the transaction in the same way. Novell talks about "when we sold Unix to SCO," and SCO talks about "when we bought Unix from Novell." But, as it turns out, in the context of the sale, the two have a different definition of Unix. What SCO says it bought is not the same as what Novell says it sold. For the most part, SCO's interpretation is that it literally bought Unix, a purchase that entitled the company to the same control over Unix that the operating system's first owner -- AT&T -- had over it and that has been inextricably connected to the OS each time ownership has changed hands before finally ending up with SCO.

Novell, on the other hand, says that it didn't sell the same "package" that was bought and sold before, but rather just enough of it to allow SCO to build a Unix business. By retaining certain control over pre-existing licensing agreements with companies like IBM and HP, as it says it has done, Novell claims to have complete governance over the performance (the definitions of which are also the subject of legal debate) of those licensees with respect to their obligations. Should this be the case, then it effectively strips SCO of any legal recourse should one or more of those licensees fail to live up to those obligations. Instead, under Novell's interpretation, in the event that one or more of the pre-existing licensees fails to fulfill their obligations, only Novell has the legal leverage to remedy that issue.

As if to compound the interpretation problem, the obligations (as well as the different parties' interpretations of them) are also in the eye of the storm, although not quite in the bull's eye the way the question of what exactly was sold to SCO is. Although this is an oversimplified explanation of SCO's interpretation, the basic idea is that licensees have an obligation to prevent clones of the official Unix (clones that are not subject to the official Unix license) from being developed in such a way that any of their heritage can be traced back to the original Unix.

Allowing that cloning to happen, which SCO contends is the case with Linux, theoretically harms both the owner of Unix and the authorized licensees of Unix. First, to the extent that others are making money on that unauthorized Unix clone, the owner of Unix is being denied certain proceeds to which it is entitled. Second, to the extent that the Unix clone is purchased or used as an alternative to one of the officially licensed versions of Unix (including any version offered by the owner of the Unix intellectual property), the exclusive right of the legitimate licensees to serve those customers' needs has been compromised. Financial harm is involved in both cases.

This doesn't mean that a competitive Unix clone that isn't subject to the terms of the Unix license or to the control of the owner of Unix can't exist. Like other cloned technologies (BIOSes, virtual machines, etc.), SCO's Chris Sontag contends that the only way for a Unix clone to escape such control is for that clone to be a 100 percent clean-room developed clone--a soup-to-nuts version that was developed by programmers who had absolutely no exposure or influence from the source code of Unix and its derivatives (the latter of which couldn't have been developed without exposure to the original Unix source code).

In a number of ways, according to SCO officials, Linux is not a clean room implementation of Unix. SCO claims that Linux contains source code that, at the very least, can be traced back to several licensed derivatives, if not the original Unix source code. The reason SCO sued IBM is because it believes that IBM is party to this non-clean-room-developed clone along both of the aforementioned vectors.

First, IBM was directly responsible for the contribution of code to Linux that was technically off-limits in SCO's view. As a side note, IBM isn't the only party that could end up being sued for such a alleged role in creating Linux. Some speculate that Linus Torvalds, credited with creating Linux, and Andrew Morton, the overseer of the Linux kernel (both of whom are employees of the Open Source Development Labs), will be next.

The second reason SCO is suing IBM is because the company contends that IBM failed in its obligation (which is also now a matter of legal debate) as an official Unix-licensee to protect the integrity of Unix's IP. As a party to the alleged leaks that resulted in the current version of Linux, SCO contends that IBM is responsible for the financial harm that has been caused to the rightful owner of Unix by the illegitimately derived Unix clone.

This much we know: Whereas Novell paid approximately $800 million for Unix, SCO paid approximately $100 million. According to Novell officials, it sold Unix because it wanted to be "out of the Unix business," but the company still wanted a return on its substantial investment.

This much is also known: When Novell sold Unix to SCO, part of the agreement was that Novell would retain the rights to the royalties from certain pre-existing licensees. SCO would become the business administrator of those accounts on Novell's behalf in exchange for a 5 percent fee. As far as I can tell from interviewing both companies, there is no argument over these facts. At the very least, given this framework, the transaction between SCO and Novell constitutes the "sale of Unix" from Novell's point of view because it sold SCO the right to be in the Unix business. In other words, the right to license Unix to new licensees and to benefit from any royalties or payments that come as a result.

But this is where the mutual understanding ends. Novell's interpretation of the "Asset Purchase Agreement" (including an amendment to it) is that, in addition to retaining the rights to any proceeds from pre-existing licensees, it also maintained complete control over the those licenses and therefore, the supervision over the obligations of those licensees.

In other words, regardless of how the legal issue over how the obligations should be interpreted is settled, if a pre-existing licensee's satisfaction of those obligations is in question, it's up to Novell and only Novell to decide what happens next. This is why, when SCO told IBM that its license to Unix had been revoked because it failed to satisfy its obligations, Novell essentially told IBM to disregard the notice. The assumption was that the license wasn't SCO's to revoke. This Novell interpretation, SCO claims, theoretically paves the way for pre-existing licensees like IBM to play a role in the unrestricted creation of Unix clones, thereby undermining SCO's ability to seek a return on its $100 million investment in the Unix IPR, while Novell (which is now partly owned by IBM) can turn a blind eye to that behavior, thus leaving SCO with no legal recourse. SCO argues that Novell's interpretation is ridiculous because SCO would never have agreed to terms that allowed Novell to control the IPR in such a way that a licensee could do whatever it wants without being having to answer to SCO first.

This brings us back to the original question: Why would any one spend $100 million to be in the Unix business if they didn't have the sort of control over the IPR that guaranteed them legal recourse (namely, the right to revoke a license) in the event that they felt that a pre-existing licensee was improperly interfering with their ability to seek a return on that investment? That seems to defies logic. It would be like the proverbial purchase of the Brooklyn Bridge. SCO might as well have just become a licensee of Unix for far less money.

Though we'll never know for sure, it's possible that SCO's interpretation of what it was getting at the time it purchased Unix from Novell is consistent with its interpretation today. SCO's explanation for why it took so long to come forward (another controversial topic) is that it took some time before it discovered what it believes to be misappropriated Unix IP in Linux. The company expected to have the legal leverage necessary to make sure that new or old licensees couldn't undermine their business and, based on that expectation, is now attempting to use that leverage. Novell on the other hand, claims that when it comes to pre-existing licensees, SCO has no such leverage and that Novell intended it that way when the sale was made, and when it signed both the original contract and the amendment to it.

Much of the ensuing public debate has vetted the intentions and the actions of the involved parties. As I mentioned earlier, even the interpretation of the obligations is different. Ultimately these issues are all secondary to how the court interprets the final terms of the sale. It is totally a matter of interpretation.

If, for example, the court rules in SCO's favor, saying that SCO has full control over the IPR and that Novell is only entitled to control over the royalty agreements with pre-existing licensees, the argument then moves onto to the issue of obligations--whether SCO has interpreted them correctly and if any licensees aren't satisfying them. If the legal action does arrive at that conclusion, a lot of people and companies will be running for cover.

If it turns out that SCO has no legal leverage over pre-existing licensees, the debate over the obligations of those licensees could become moot. It may turn out that only Novell can hold those licensees accountable for their actions and, as a part-IBM-owned distributor of Linux, that's not going to happen. If this turns out to be case, the rug will have been pulled out from under most, but not all of SCO's legal pursuits.

If the Unix IPR is enforceable (an issue that's discussed shortly), one doesn't have to be a licensee of Unix to misappropriate it. It's physically possible for anyone to cut, paste, compile, and distribute the source code. If the case reaches a point where Novell's interpretation is upheld, SCO will look to establish that it owns enough of the enforceable IPR to pursue those who have misappropriated it (regardless of whether they are a licensee).

This is where the highly controversial and oft cited as never having been publicly produced "copied code" comes into play. According to SCO spokesperson Blake Stowell, there are three categories of misappropriation: (1) letter-for-letter, word-for-word, line-by-line literal copying of copyrighted source code without proper attribution; (2) a masked or obfuscated version of (1) where the code is not a direct match, but it can be proven that copyrighted source code was copied and then altered in an effort to hide the fact that it was copied; and (3) copied code that can be traced to derivative work.

IBM's version of Unix (AIX), which includes proprietary code that was developed by IBM (I'll refer to that code as Unix extensions), is said to be a "derivative." SCO is not claiming ownership of the entire derivative or of IBM's extensions. But SCO does claim that the licensing agreements that allow licensees to build derivatives also prevents them from licensing their proprietary extensions into Unix clones like Linux without the blessing of Unix's rightful owner. This third type of misappropriation is, for the most part, relevant to the "obligations" portion of the previously described first vector.

What matters for this second vector are the first two forms of misappropriation. Despite the refutations of many Unix historians, SCO maintains that it has demonstrated to the public and to the court that certain copyrighted material that does not belong in Linux is or at one time was a part of Linux, and that the copyrighted material in question is the sole property of SCO. In my interviews with SCO, company officials contend that some of this code made its way into Linux by way of the Berkeley derivative of Unix (known as BSD). To Linux users, SCO has offered a letter that lists some of the copyrighted material that can be found in Linux.

Forgetting for a moment what constitutes a derivative and whether code that's unique to a derivative can be re-licensed into a clone of Unix (a very touchy undercurrent to this case), Open Source Initiative's Eric Raymond has openly maintained that the reason that SCO's claims are frivolous is because, in previous case, AT&T was unable to convince a court that BSD was improperly derived from its Unix intellectual property. According to some experts, the 1993 settlement between AT&T and Berkeley, which was sealed by the court, may indeed contain proof that AT&T's IPR was unenforceable. In other words, maybe Novell bought the Brooklyn Bridge as well, and just passed a part of it along to SCO. The reason AT&T may have wanted it sealed is that if Unix licensees discovered the unenforceability of the IPR, they might have stopped paying their licensing fees.

If BSD was indeed "in the clear," as Raymond contends, that would have paved the way for some of the BSD technology to find its way into Linux in a way that SCO can't even say "boo!" But SCO officials say Raymond's account is a revisionist's history. According to SCO's vice president of SCOSource Chris Sontag, the settlement of that case is indeed sealed, but, anybody can look at the BSD source and see that the files in question actually bear the AT&T copyright. That "evidence," claims Sontag, verifies that AT&T was able to maintain its copyright and the enforceability of its IPR while Berkeley got a hall pass to build a free version of Unix (something that no other provider of Unix has managed to extract from a Unix intellectual property holder). Sontag challenges anyone to verify the existence of the AT&T copyrights in the relevant BSD source code, and then to find the same source code in Linux that is mysteriously missing the copyright.

Meanwhile, some of the experts I've spoken with say it may come down to the specific language in the sealed settlement. If, for example, the language indicates that and no other third parties were entitled to the exemption given to BSD, it could bolster SCO's case. If the settlement is missing such language, it could weaken SCO's case.

As with the first vector, a judge must first agree with SCO that the source code in question (the code found in Linux) is indeed SCO's intellectual property and that it's enforceable. If it does, SCO will then have to prove who the responsible party was for allowing that code to end up as a part of the Linux source code. Already, SCO has gone after Silicon Graphics for such a "leak", and in a letter to Linux users, Rich Altmaier, SGI vice president of software, admits that some fragments of code that SGI considers to be somewhat inconsequential to the case were indeed copyrighted material that SGI decided to replace.

This raises two questions. Who else is responsible for code leakage and is the remedy as simple as replacing the infringing code (as Red Hat has promised to do, in lieu of any indemnification)?

This is where the entire derivative issue comes up again. As a part of its many claims, SCO contends that the final version of Linux doesn't have to contain any copyrighted Unix source code in order for it to qualify as a derivative that, according to the AT&T language, cannot be relicensed without SCO's consent. All Linux had to do was rely on misappropriated IP somewhere in its life to get it to where it is now, and it automatically became a qualified derivative.

SCO's Sontag argues that between the files listed in its letter to Linux users, the SGI admission, and a separate Jan. 12, 2004 court filing that detailed the misappropriation of additional derivative works by IBM, SCO will have no difficulty establishing that Linux either directly or indirectly relied on Unix to become the operating system that it is.

Under SCO's interpretation of the AT&T language (an interpretation that is also being challenged), Linux qualifies as a derivative to which the enforceable Unix IPR applies. From SCO's point of view, there's no going back. Once a derivative, always a derivative, regardless of whether code is replaced or not. It's an issue that could end up coming back to haunt SGI (which is also one of SCO's targets and Red Hat, both of which have taken the position that the replacement of infringing code is at least a part of the remedy if the court sides with SCO on SCO's various interpretations. Again, virtually all of SCO's interpretations are being challenged.

One of these challenges is coming from the free software community, in particular the Free Software Foundation's general counsel Eben Moglen, who is also a professor of law at Columbia University. Throughout this ordeal, supporters of the Free Software Foundation's GNU General Public License have argued that once SCO (or the same company operating under any other name such as "Caldera") offered a Linux distribution of its own under the GPL (as all Linux distributions are offered), it lost any right to claim that parts of Linux contained misappropriated IP. It's a view that Moglen supports and one that's akin to calling SCO a hypocrite.

Here we have SCO accusing parties of wrongfully licensing (under the GPL) misappropriated intellectual property while the company itself appeared to have done the exact same thing with the same intellectual property as late as the Summer of 2003 (well after SCO began its legal endeavors). In his last reference to two of SCO's exhibits of IP misappropriation, Moglen says, "Both 'demonstrations' actually showed that SCO was claiming copyright infringement in material whose copyright it didn't own." If you put as much stock in the statements of chief presiding counsel over the GPL as most members of the open source community do, it does appear as though SCO has put itself between a rock and a hard place.

But, here again, legal interpretation may end up playing a huge role. In response to "the GPL position," SCO's Sontag has several points that may factor into final outcome.

Sontag doesn't dispute the fact that SCO distributed Linux under the GPL. But in my interview with him, Sontag claimed that you can't give away your copyright by mistake--not by U.S. law and not even according to the rules of the GPL. "U.S. copyright law doesn't allow a copyright to be inadvertently assigned" said Sontag. "In addition, the GPL says the legal copyright holder has to put a legal notice in the beginning of the source code that says you're contributing it to the GPL. Section 0 of the GPL says this and GPL even gives examples of how to do this."

Indeed, the following text that Sontag is referring to clearly states that the copyright holder must be the one to have placed the GPL notice into the code: "This License applies to any program or other work which contains a notice placed by the copyright holder saying it may be distributed under the terms of this General Public License." Sontag says, "We did not knowingly distribute Linux with the code that we have the copyright to, nor were we the ones to place the GPL notice on the code."

Secondly, when it comes to some of the derivative works in question (and to which, in part, Moglen's statement refers), SCO is not claiming it owns the copyrights by virtue of the fact that the derivative code can have its heritage traced back to the official Unix. It is only claiming that, according to SCO's interpretation of the AT&T language, the copyrighted work found in Unix derivatives is subject to certain requirements, one of which is that it can't be redistributed outside of the copyright holder's officially licensed derivative without the consent of Unix's owner. In other words, in SCO's interpretation, the owner of Unix has a significant amount of control over a licensee's derivative work. However, as SCO freely concedes, it doesn't own it.

Once again, we come back to the interpretation of the sale. Just about every one I've spoken with has their own interpretation or cites the interpretations of others. In the end, it appears that a judge is going to have to make the call.

This "legalese" is a strange language. Even though these contracts and amendments are written in black & white, on paper, there are all kinds of interpretations even among lawyers of what it all really means, and it will eventually come down to a judge to tell everyone what the judge thinks it all really means.

Henceforth -- all contracts should be written in plain english!

It is only claiming that, according to SCO's interpretation of the AT&T language, the copyrighted work found in Unix derivatives is subject to certain requirements, one of which is that it can't be redistributed outside of the copyright holder's officially licensed derivative without the consent of Unix's owner. In other words, in SCO's interpretation, the owner of Unix has a significant amount of control over a licensee's derivative work. However, as SCO freely concedes, it doesn't own it.

I'd like the author (or the poster) to describe exactly how SCO claims to have this 'control' over the derivative.

In the case of IBM's UNIX license, for example, IBM owns (in the full sense of the word) the derivative works, including the right to distribute as IBM sees fit(by licensing or otherwise) the derivative work. And the right to distribute had no express limitations in the license grant itself from ATT.

It's important to understand that ATT granted 'ownership' of IBM-created derivative works to IBM -- and since ATT did not retain ownership, then it is logically inherent as part of the bundle of rights contained in a grant of ownership, as opposed to a license, that the owner -- IBM --can do as it pleases (sell, distribute, etc) with the "thing" that it owns.

The only "restriction" on IBM's ownership (and relied on by SCO in bringing the action) was in an associated Confidentiality Clause executed between IBM and ATT that stated (in a very unclear way, and such lack of clarity would be interpreted by a court against the interests of the drafter, meaning ATT, and therefore SCO) that derivative works must be treated as "ATT-Confidential" information.

The confidentiality clause was unclear because, first, it is unclear that a Confidentiality restriction would mean, by itself, that distribution would be prohibited (because there are other ways to protect 'confidentiality" per se but still permit distribution). In any event, in order to eliminate the prior confusion over interpretation of the clause, ATT publicly stated to its licensees that the Confidentiality provision (because it was subject to confusing interpretations) should be officially interpreted to apply ONLY to the actual UNIX code that might be physically embedded in a derivative work, but NOT to the derivative itself -- and this is of course consistent with the license grant from ATT, which states that licensee (IBM) OWNS (!!) the derivative work.

Note again that ATT's licensees included IBM, and, by the legal principle of estoppel, SCO's claims are bound and limited by ATT's own words to the extent that ATT intended its licensees to rely on such words.

ATT's clarification was intended to settle the issue, to clarify that the right to distribute (sell, etc) is inherent in the grant of ownership.

In my view, what we're seeing from SCO is reminiscent of David Bois' legal arguments in the 2000 Florida election case -- to wit, an argument that is full of bluster, but held together with the legal equivalent of baling wire and duct tape, hoping to get in front of a biased judge to have any hope of succeeding.

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.