Here's SCO's Reply Brief in Further Support of its Supplemental Memorandum Regarding Discovery [PDF], which has now been refiled in redacted form. When I read it, I understood several things:
1) why Judge Wells granted them all the discovery she originally did
After all, if the only way to find copyright "infringement" in Linux is to sort through all *unreleased* versions of AIX and Dynix going back ten years to try to find some abstruse, nearly impossible to find nonliteral infringement that can only be proven by tracing methods and concepts mini steps through a decade of AIX and Dynix development that not even SCO can put its finger on after two years of effort and mostly because of contract terms that they argue give them more rights than copyright law, it seems obvious that no normal person would blame Linus or the Linux development process for not noticing something as hard to discover as all that.
2) what SCO's strategy is
3) that much of SCO's need for discovery is to defend themselves from IBM's counterclaims, and
4) that even if they were to win the IBM contract claims, which I doubt, Linux has now been vindicated once and for all, ironically enough *because* of all the discovery SCO obtained.
And SCO still hasn't shown any evidence of infringement two years into the process and after having access to all versions of everything, released and unreleased, despite their public claims in 2003 that they already had in their hands such evidence. Now they tell us they are still looking.
This filing was part of SCO's ultimately largely successful strategy to obtain discovery from Judge Brooke Wells, asking her over and over in every way for unfettered access to IBM's CMVC and the Dynix equivalent. She didn't give them that, but she gave them plenty. If you found that incomprehensible at the time, reading this Reply Brief will help you understand at last.
As you will see, they filled her ears with indignant stories about how evil IBM had misused their code in AIX on Power, stating they only just found out about this wicked deed, and they needed discovery. Later, Judge Kimball ruled that SCO or its predecessor appeared to know or should have known all about AIX on Power years ago, so he refused to allow them to amend their complaint to include that issue. And he also batted down SCO's arguments about IBM's Ninth Counterclaim. So as you read this filing, you can safely ignore all the bluster about AIX on Power and the Ninth Counterclaim. They lost on both of those points in the end, but they used both to argue their need for more discovery, and in that, they were more successful. Why? Because it isn't Judge Wells' job to decide who is right, just who needs what in discovery if they claim whatever they are claiming. And SCO was claiming with what turns out to be mock outrage that they only just discovered the AIX on Power use and they vitally needed the discovery to investigate the particulars. But they also stressed to her that to defend against IBM's counterclaims, they needed all the code. And so she said OK.
You will see they portrayed IBM as refusing to produce materials that SCO had earlier asked for but the court had never ordered. It's a smarmy technique, because it is misleading, but I've given up trying to teach SCO anything about morals.
So here they waxed poetic about IBM's alleged sins. She, as the magistrate, didn't know it was all a tall tale. She presumably knows now, and she may be kicking herself a bit after seeing all the evidence about AIX on Power and the ultimate ruling against SCO. But at the time, she likely felt she had no choice but to give them what they asked for. They told her, in effect, that they couldn't defend themselves without it.
This document lays the strategy all out in a clarity I haven't seen before, maybe because they knew they'd be filing it as a sealed document. They didn't know that their pals G2 would ask the court to unseal everything and that this document would see the light of day. Be careful what you ask for, eh? It's possible I just didn't clue in to SCO's strategy before, so if you figured this out years ago, more power to you. But it only just clicked into place for me when I read this filing.
If you turn with me to page 11 of the PDF, in the argument under the heading "The Discovery SCO Seeks is Relevant to SCO's Claims and Defenses to IBM's Counterclaims on Several Distinct, and Independently Sufficient Grounds" (that begins on page 10), we find the following, and I've set off the quoted section in colored text, to distinguish it from what I write about it afterward:
First, the discovery is unmistakably relevant to SCO's core claims concerning the IBM and Sequent license agreements. 9
The plain language of the IBM and Sequent Software License Agreements required that any derivative or modification of the original UNIX System V code was to be treated as if it was "part of the original SOFTWARE PRODUCT," and thus subject to the same restrictions on use and disclosure as the original licensed UNIX source code itself. 10 They mentioned the Gates Rubber case in an earlier memorandum [PDF] in support of their requests for more discovery but without really explaining as clearly as here how they thought it applied. You can read the decision in the Gates case here. In this filing, they spell out exactly what they are thinking. Their footnote 11 states:
One way in which SCO will prevail on its contract claims is by tracing the AIX or Dynix code that IBM contributed to Linux back to UNIX. The outstanding discovery concerning the development history of AIX and Dynix that IBM has refused to produce is clearly relevant to this proof because it documents among other things: (1) the names of the programmers who were involved with developing, modifying, and/or contributing the AIX and Dynix source code in question; (2) the dates of those developments, modifications, and/or contributions; and (3) the precise developments, modifications, and/or contributions that were made, including the precise source code involved and other relevant programmer notes pertinent to these issues. See SCO Reply Mem. Regarding Discovery (July 12, 2004) at 11-13.11 . . . .
Second, independent of SCO's own claims, the discovery at issue is plainly relevant to two separate counterclaims that IBM inserted into this case on March 29, 2004 -- after this Court entered its last discovery-related Order on March 3. IBM's Tenth Counterclaim, for example, seeks a blanket declaration that Linux does not infringe any SCO copyright. See IBM 2d Am. Countercl. ¶173. There is no merit to IBM's attempt to shield the development history of AIX and Dynix from discovery based on its repeated refrain that the proof of its Linux copyright claim requires a comparison of only UNIX System V code on the one hand and Linux code on the other. SCO does not dispute that its defense to IBM's Tenth Counterclaim will necessarily involve proof of literal and/or non-literal copies of UNIX code that were ultimately dumped into Linux. What IBM ignores is that its Linux dump did not occur directly from the original licensed UNIX product; rather, before the original UNIX code was copied, literally or non-literally, into Linux (within the past five years), it traveled for longer than a decade through the development process of the UNIX-based AIX and Dynix programs. It is for this precise reason that the development history of AIX and Dynix -- the discovery that IBM has withheld for over a year -- is critical to SCO's defense to the Tenth Counterclaim.
In particular, and as SCO has previously detailed for the District Court in its opposition to IBM's first summary judgment motion and at the recent argument on that motion, the outstanding programming-history discovery may contain programmer admissions and comments on which courts have heavily relied in copyright infringement cases. See, e.g., Computer Assocs. Int'l v. Quest Software, Inc., No. 02 C 4721, 2004 WL 1459495, at *9 (N.D. Ill. June 28, 2004)(Exh. A)(granting preliminary injunction in computer software infringement case based on such evidence). See, e.g., Gates Rubber Co. v. Bando Chem. Indus. Ltd., 9 F. 3d 823, 835-36 (10th Cir. 1993)(noting that the "structure, sequence, organization" of a program, and not only its "literal" source and object code, may be copyrightable). Automated computer programs, while useful in detecting verbatim copying, are not perfect, and are virtually useless in identifying non-literal copying. And manual, side-by-side comparison of millions of lines of System V and Linux source code in a search for non-literal copying would be incredibly burdensome and time-consuming. The discovery SCO has requested, on the other hand, will allow SCO to determine which elements of UNIX were directly incorporated into AIX and Dynix in the first instance and thereby to trace the migration of protected, non-literal elements of UNIX, through AIX and Dynix, to Linux. In turn, SCO will be able to identify "hot spots" in Linux (those portions of AIX and Dynix that can ultimately be traced back to UNIX) on which it can effectively focus its Linux copyright infringement investigation in this case.
Finally, and even more obviously, the outstanding programming history is independently relevant to IBM's Ninth Counterclaim, which seeks a declaration that IBM does not infringe "any SCO copyright through the reproduction, improvement, and distribution of AIX and Dynix." IBM 2d Am. Countercl. ¶ 167 (emphasis added). That history is relevant entirely apart from the specific evidence that SCO has recently uncovered to establish IBM's infringement of SCO's copyrights through AIX (i.e., evidence that in connection with its work on "Project Monterey," IBM used its CMVC system as an instrument to misappropriate SCO's proprietary code for inclusion in a specialized version of AIX (AIX for Power)). Given its plain language -- and notwithstanding IBM's attempt now to recast its scope, see Part II, below -- IBM's Ninth Counterclaim necessarily directly implicates the programming history of AIX and Dynix, an issue on which the CMVC and RCS systems constitute direct evidence.
9These claims include SCO's breach of contract claims (Counts One through Four) as well as its copyright claim (Count Five), which centers on IBM's improper continued use of AIX and Dynix following SCO's termination of the IBM and Sequent license agreements.
10 Similarly, Section 7.06(a) of the Sequent license agreement specifically prohibited the disclosure of "any or all of such SOFTWARE PRODUCTS (i.e., UNIX System V and, pursuant to Section 2.01, any modifications or derivatives based on that contents of System V] (including methods or concepts utilized therein) to anyone." Thus, any disclosure of any method or concept of anything in UNIX System V (or anything derived from UNIX System V) would breach the license agreement. SCO's access to the requested programming history will permit SCO to locate the proof of such breaches.
11 IBM has previously mischaracterized SCO's contract interpretation argument to be limited to showing that AIX and Dynix are, as a whole, derivatives of the original licensed product. As SCO has repeatedly explained, although there is much to support this contractual reading -- including IBM's own repeated admissions -- SCO's theory of the contract is not so limited. Rather, SCO also seeks to show that the particular contributions that IBM made to Linux violated the license agreements because those specific contributions represented derivatives or modifications that were required to be treated as if they were part of the original licensed Unix System V product.
12 After SCO showed during the September 15 hearing before the District Court how the development history of AIX and Dynix is a roadmap permitting SCO to investigate potential copyright infringement in IBM's Linux contributions, IBM counsel opined that the roadmap for which SCO asks "might as well be the road map for China" because "The only road map that matters, Your Honor, is what's in Unix and what's in Linux." Tr. at 115-16. This argument is misguided. Because IBM's contributions to Linux were taken from AIX and Dynix, the UNIX-derived products, the roadmap for IBM's contributions to Linux exists in the programming history to those programs. IBM's counsel's argument also misapprehends the governing discovery standard, under which SCO is entitled to any information reasonably calculated to lead to the discovery of admissible evidence.
SCO also seeks to show that the particular contributions that IBM made to Linux violated the license agreements because those specific contributions represented derivatives or modifications that were required to be treated as if they were part of the original licensed Unix System V product. Why would they need to look to the licenses instead of copyright law? Because the Gates case precludes methods and concepts from copyrightability:
Some concern has been expressed lest copyright in computer programs should extend protection to the methodology or processes adopted by the programmer, rather than merely to the “writing” expressing his ideas. Section 102(b) is intended, among other things, to make clear that the expression adopted by the programmer is the copyrightable element in a computer program, and that the actual processes or methods embodied in the program are not within the scope of the copyright law. I believe they are hoping, or were when this all began, to extend copyright law interpretation to include methods and concepts (remember their book analogy?), but that exalted hope may have taken a back seat, and may now be sitting next to the GPL-is-unconstitutional silliness. But as of now, this isn't about copyright infringement in the line by line sense. The best they are hoping for, if they can find it, is indirect copying, obfuscation, that and the AIX-post "termination", a termination Novell has waived anyhow. Good luck with that, by the way. No, it's primarily about the licenses. It's a convoluted way to try to claim control of the code by coming in the back door via contract. They appear to understand, despite what they told the media, that they have no copyright infringement claim against IBM regarding Linux that would match what they told the world, the mountains of code baloney. That is a claim they have never substantiated with evidence, and I gather from this filing, they never will be able to do so. The claim is foundationally contract-based. The rest is still just a hope.
How does that vindicate Linux? In this filing, they say "SCO's copyright claim . . . relates only to IBM's distribution, reproduction, and use of AIX and Dynix without a license to do so." Oh, really? Then what was all that trash talk to the press since 2003? And what about their recent "explanation" about the 1999 study? Is there or is there not copyright infringement in Linux relative to IBM? If you look at their statement closely, you will see they mention obfuscation. So that is their fantasy, that they'll be able to find some of that. That explanation doesn't in any way support their having accused Linux developers of line by line infringement. Remember the part about the comments being identical? Laura DiDio found that convincing. But in the court case, where is any of that evidence? It seems to have gone walkabout.
Instead I see nothing but this strained and complex methods and concepts hunt, based not so much on copyright law but on contracts and a hope to find obfuscated code somehow, some day. If there isn't any copyright infringement except what is listed here, the post "termination" of the AIX license, which isn't about Linux, and if what they told the judge here was truthful, then why did they make the claims they did in 2003 and why are they telling the media even now that the 1999 study does show "problems" with Linux? If there were any such problems, why isn't IBM charged with copyright infringement regarding those "problems"? So which is it? Did they have in their hand in 2003 evidence of copyright infringement or didn't they? And by the way, what are those "problems", exactly? Do you really believe if they had anything at all to charge IBM with, that they would refrain?
What I now suspect happened is this: First, they found out from the 1999 study, which Davidson reported to Reg Broughton in August of 2002, that there was no copyright infringement. He told them so, plain as day. The study was useless. Any apparent similarities had turned out to be perfectly legal. At some point after that, SCO told us, they had three teams of programmers look at the code and they all, they claimed, said that they found literal copyright infringement, and yet SCO never, to my knowledge, presented any of these reports to the court as evidence and they've never released any of these studies to the media either. Why not? You may draw your own conclusions from SCO's behavior as to just how valuable those studies turned out to be, if they existed in the first place.
But, aside from the contract claims, the need for all the code was primarily driven by a need to beat back IBM's counterclaims, not to attack IBM for copyright infringement. Here's how SCO puts it:
Wholly apart from SCO's contract claims, the discovery that SCO seeks is now relevant, and indeed critical, to SCO's ability to defend against IBM's recently-added copyright claim. See what I mean? They need the discovery for two things: to try to establish their theory of the licenses, and they really need it to defend themselves. (Hope springs eternal in their breast, naturally, that if they keep looking, they'll find something that will at least appear to be copyright infringement.)
My point is simple: I think that whether or not SCO is successful in its claims regarding the contracts, it's now obvious that there was no breakdown in the Linux development process. SCO is desperately needing all that code to defend itself from IBM's counterclaims, especially regarding copyright infringement, not because SCO is accusing IBM of infringement at this point, other than the AIX termination stuff, but because IBM is asking the court for a clean bill of health. I'm sure, if SCO ever found any infringement, they'd use it aggressively, but that isn't why they needed that mound of code. They need it to save themselves.
Whatever contractual violations SCO hopes to someday find, it can't possibly reflect on Linux. Look what SCO has had to do to try to find any violations. They've been burrowing like little moles through this humungous pile of code, and so far, nothing. Linux isn't a party to any contract with SCO. If it's that hard to find, and SCO has had in hand both bodies of code from day one, Unix and Linux, since they sold both, how could Linus possibly screen out something that the owner of the code can't even identify after two years of serious looking? And the owner of the code appears to be going after methods and concepts in a contract claim precisely because there isn't any copyright infringement in Linux, and the only way to throw tacks in the road to try to slow down Linux adoption was this elaborate contract claim. Of course, the statements about "mountains of code" and "line by line infringement" will have to be explained one day, and justified, when IBM's other counterclaims about Lanham Act violations and common law torts have their day.
Obviously, there isn't a thing wrong with the Linux development process. So, Linux has been vindicated, ironically *because* of all the hoops SCO insisted on leaping through in discovery.