imported_BrianW

No problem, Colin. Since I've been following the story for a few months now, it wasn't that hard for me to slap the above post (including the “Crap, crap, crap!” part) together.

Another thing worth noting is that IBM asked SCO to identify each of the trade secrets and Unix code SCO thinks have been misappropriated, why SCO claims ownership of these trade secrets and code, and how they came into possession of these trade secrets or code (whether developed by them, or acquired by them, etc.). IANAL, but as far as I can tell, this is standard. When someone accuses you of stealing something, the accuser must identify the item that was stolen and offer some evidence that he actually owns it. The accusation must be specific enough to be investigable. He can't just make a general accusation and be granted court authority to go rummaging through your house looking for something – anything – that he thinks he can get away with claiming as his own.

In response to this request for specificity, SCO returned a list of 591 Linux source files that “may or may not” contain misappropriated code.

It may not be obvious, but this is not what IBM asked for. Linux source code is available for anyone to look at, and singling out 591 files of open-source Linux kernel code does nothing to identify misappropriated trade secrets or stolen Unix code. Unless SCO identifies SysV Unix code, it's not identifying anything that it can demonstrate it owns.

And it gets better. Never mind the fact that IBM is entitled to know exactly what it's accused of; SCO has been blasting IBM in public, claiming over one million lines of SysV Unix code have been copied directly into the Linux kernel. Furthermore, SCO has taken investors, analysts, and members of the media behind closed doors, and under protection of a Non-Disclosure Agreement (NDA) purportedly showed them examples of SysV Unix source code side-by-side with Linux kernel source code, proving to them that infringement has indeed taken place. (Not surprisingly, this has had an incredible positive effect on SCO's stock price.) Darl himself has proclaimed that SCO had hired teams of NASA rocket scientists and MIT mathematicians to perform “spectral analysis” to compare Linux kernel source code with SCO's SysV Unix code, and, according to SCO, the results indicate that Linux is almost a verbatim, line-for-line copy of Unix.

Yet, when it comes time in a court of law to put up or shut up, SCO won't have the courtesy to show any of this proof of infringement to IBM, the company who stands accused of misappropriating the Unix code.

So they didn't put up. And they haven't shut up, either. They've recently proclaimed that they're going to sue a prominent Linux user within 90 days, even though they have yet to publicly prove that they own even a shred of Linux.

Oh, but it gets even better. Much, much better.

Remember those 591 Linux files identified by SCO that “may or may not” contain misappropriated trade secrets? Well some geeks (I'm a geek, as if you all didn't know) have been taking a look at it and, well, laughing, mostly.

For example, consider the following, which is the entire contents of include/asm-arm/spinlock.h, one of the files on SCO's infringement list: It's a simple, six-line source file. And it “may or may not” contain a SCO trade secret.

Now if you're a geek, you're probably laughing already. If you're not a geek, go find one, show him this code, and tell him that there's a trade secret in it worth three billion dollars. (Make sure you take away his Mountain Dew so he doesn't spew it all over your monitor.) See how hard he laughs? (It's okay, you can laugh, too, if you want. The geek will think you're laughing at the code. )

In English, what this code does is produce a compiler error, issuing the message that “ARM architecture does not support SMP spin locks.” That's all. Why is this funny? Because, since this code produces a compiler error, there is no way it can be included in ANY working Linux kernel.

But, to be fair, this is the list of files that “may or may not” contain infringement. So what was it about this file that the NASA rocket scientists and MIT mathematicians thought was infringement? The answer is three simple letters: SMP SCO is claiming, if not ownership, then at least contractual control over several technologies: SMP, JFS, RCU, and NUMA. (Never mind what they are, just go with me on this.) So, on a hunch, a bunch of geeks used a program called grep to search the Linux kernel source code for any of these acronyms. It takes just a few seconds to execute. They got a list much larger than SCO's 591 Linux files. So they had to eliminate some. On another hunch, they modified their grep command to eliminate those files that mentioned “SCO” and “Caldera” (SCO's former company name), assuming that SCO wouldn't want to implicate itself in a court document.

Guess what? The grep command, when instructed to identify files with SMP, JFS, RCU, NUMA, and eliminate from that list those files that contain SCO or Caldera, produces the exact same list of 591 files that SCO produced for the court. (This was with one of the 2.5 releases of Linux. Othe releases had slightly different results.)

Exactly the same! And it takes less than a minute! And to think that SCO wasted all that time and money on NASA rocket scientists and MIT mathematicians performing all that high-tech “spectral analysis” in order to find misappropriated trade secrets, when they could have just used a simple grep command!

Yeah, sure. You can scroll back up and laugh at the code now that you understand it. If you've stuck with me this long, you're entitled.

Now I must volunteer that this in no way proves that SCO used a simple grep command to produce its list of “may or may not be” infringing Linux files. But that this list includes six-line header files that actually prevent any reference to alleged trade secrets, shows, I believe, that no human examination was performed on these files.

Not surprisingly, IBM has subpoenaed investors who have publicly claimed to have been shown proof of code infringement. As IBM's court document says (paraphrasing), “If SCO can show them, SCO can show us.”

Now before you go thinking that SCO doesn't have a case, there was a smaller list of Linux files that it produced, claiming that the files in that list definitely have misappropriated trade secrets and/or Unix code. No “may or may not” about it. Where did these files come from? Well, as it turns out, it's just a wholly-contained subset of the 591 “may or may not” files. What's different about them? They contain the strings “IBM” and “International Business Machines.” Many of the "IBM" references are copyright notices, but a surprisingly large number of the "IBM" references are comments by programmers who ridicule IBM for its hardware design. IBM had nothing to do with these files, except to inspre frustration in software engineers by providing a less-than-adequate interrupt architecture in the PC. Again, the absence of any human examination of the files is glaringly obvious.

How often is it that a company will litigate a three-billion-dollar lawsuit claiming code infringement, and not even look at the code it's submitting to the court as evidence?

Well, apparently, it's not “never.”