IBM Says SCO Willfully Failed To Detail Evidence

Robert wrote to mention a piece on CBR Online where the latest volley in the SCO case is covered. IBM is now accusing SCO of having acted in bad faith when they opened the trial against IBM, by being purposefully vague in their evidence. From the article: "All in all, according to IBM, SCO's evidence filing makes it impossible for the company to defend itself. 'By failing to provide adequate reference points, SCO has left IBM no way to evaluate its claims without surveying the entire universe of potentially relevant code and guessing ... Since only SCO knows what its claims are, requiring such an exercise of IBM would be as senseless and unfair as it would be Herculean.'"

Grocklaw's take

[Simon Brooke]

Grocklaw's take here, and it makes good reading: [groklaw.net]

"What an extraordinary response to the court's orders. As IBM points out, because SCO fails to "identify with specificity the versions, files and lines of System V, AIX, Dynix and Linux material that IBM is alleged to have misused," as a practical matter, it just isn't possible to evaluate SCO's claims. We're talking about a lot of code. IBM references a Declaration of Todd Shaughnessy, which we don't yet have, which says "there are at least 11 versions, 112,622 files and 23,802,817 lines of System V code potentially implicated by SCO's claims. There are at least 9 versions, 1,079,986 files and 1,216,698,259 lines of AIX code potentially implicated by SCO's claims. There are at least 37 versions of the base operating system, and 472,176 files and 156,757,842 lines of Dynix code potentially implicated by SCO's claims. And there are at least 597 versions, 3,485,859 files and 1,394,381,543 lines of Linux code potentially implicated to SCO's claims." Precisely where in this massive pile of code should IBM start digging?

...

"I feel sure we'll hear more on this topic at the hearing coming up. I have this vague memory that SCO told Magistrate Judge Wells, when she asked them at a recent hearing if they'd found anything of use in those materials, that they had."

Re:Two Words for IBM--Edit Distance

[AstrumPreliator]

I don't know much about the BLAST and FASTA algorithms, but the Levenshtein distance just compares strings. Simply rename everything and tweak the structure of your code somewhat and it can give false results. I know here at university they run all source code through a program (forget the name) that analyzes control flow and other program characteristics and then compare it to those of other students, past students, and source code from the web.

I don't know how effective the program is as I don't cheat, but I do know a few students in the department that have nearly been suspended.

However, your last statement is spot on, the judge should throw them out of court ;).

Re:Two Words for IBM--Edit Distance

[bjpowers39]

If you check out the post refering to Groklaw down the page, this is more complex than a simple "compare two assignments" problem. Although, IBM definitely has enough computer scientists and hardware to tackle the problem. Given that, it is not in their interest to do this. SCO brought the lawsuit and has the burden of proof. A defendent is not obligated to compare the code and notify the court (and SCO) or areas of possible infringement. That would effectively let SCO off the hook and potentially give SCO more ammunition in their lawsuit. If SCO wants to sue IBM for infringement, then SCO needs to provide evidence of that in court. Specifically they need something better than "it is somewhere in this mess".

Re:Nasty tactics

[Theatetus]

After all, SCO isn't actually producing anything

You forgot their cell-phone spamming multilevel marketing scheme [slashdot.org].

Re:Two Words for IBM--Edit Distance

[avdp]

Thus the word disprove.

It is plaintiff to make a case, and submit proof for it (evidence).

Then the defendent gets to disprove the case (if there is one to disprove), you know, show how all the evidence is a whole lot of BS, or provide counter-evidence.

Re:Two Words for IBM--Edit Distance

[FellowConspirator]

In fact, it's quite simple... Reduce the two code bases to a lexical parse of the actual code. Compare the structure rather than the arbitrary names of the symbols that compose it. It's quite simple to do.

Simple as it is (IBM even writes several tools to do such a thing and markets them to various niche markets), it wouldn't be helpful in this case. SCO no longer maintains that there is any "SCO" code in Linux. They now claim that certain "technological concepts" related to UNIX were improperly used, but they make the assertion without clear explanation of what they mean by that.

Searching for plagiarism would be cake... some ambiguous intellectual abstraction? Now that's hard!

If you look at the claims SCO started with, and what they are now attempting to argue in court, there's no relation. How the case has played out so long without being thrown out is anyone's guess.