This is a text version of the official court filing found at
For related documents, see

The PDF file includes some opaque white rectangles overlaying and obscuring some of the text. Those are not reproduced in this text version.

For tips on avoiding such redaction embarrassments, here is a selection of relevant references, from the perspectives of spooks, nerds, and clerks:

Case 2:03-cv-00294-DAK-BCW          Document 672      Filed 04/28/2006     Page 1 of 24

 Alan L. Sullivan (3152)
 Todd M. Shaughnessy (6651)
 15 West South Temple
 Gateway Tower West
 Salt Lake City, Utah 84101-1004
 Telephone: (801) 257-1900
 Facsimile: (801) 257-1800

 Evan R. Chesler (admitted pro hac vice)
 David R. Marriott (7572)
 Worldwide Plaza
 825 Eighth Avenue
 New York, New York 10019
 Telephone: (212) 474-1000
 Facsimile: (212) 474-3700

 Attorneys for Defendant/Counterclaim-Plaintiff
 International Business Machines Corporation


                           FOR THE DISTRICT OF UTAH

                                                  REBUTTAL DECLARATION
                                                    OF RANDALL DAVIS
                                                    Civil No. 2:03CV-0294 DAK
                                                     Honorable Dale A. Kimball
                                                  Magistrate Judge Brooke C. Wells

                                                  FILED IN REDACTED FORM

Case 2:03-cv-00294-DAK-BCW             Document 672         Filed 04/28/2006       Page 2 of 24


        1. I have been asked by counsel for IBM to respond to the Declaration of Marc

 Rochkind submitted by The SCO Group, Inc. ("SCO") in opposition to IBM's motion to

 limit SCO's claims. This declaration is limited to responding to the issues raised by Mr.


        2. Like my declaration dated March 29, 2006, this declaration is based on my

 experience working in the field of computer science and evaluating allegations of

 intellectual property violations. I further base the facts and opinions set out in this

 declaration upon careful review of SCO's Final Disclosure of Allegedly Misused Material

 dated December 22, 2005 (the "Final Disclosures").

        3. In summary, the Rochkind declaration talks past the conclusions set out in my

 previous declaration. While Mr. Rochkind states that he "strongly disagree[s] with [my]

 assertion (at paragraph 11) that SCO has failed to identify with specificity 198 challenged

 Items in the December submission" (Rochkind Decl.  7), he fails to directly confront the

 facts set out in my declaration and his conclusion is clearly based on a very different

 inquiry. Put differently, Mr. Rochkind reached a different conclusion than I did because

 he answered a different question.

        4. If the question is whether SCO provided version, file and line information for

 each of the 198 Items at issue, then the answer is unquestionably "No". Although

 Mr. Rochkind uses words that might be understood to suggest that these coordinates have

 been provided where possible (Rochkind Decl.  11-12), they have not been. Notably,

 he makes no effort to show that they have. Without version, file and line information

 concerning SCO's allegations, it is simply not possible fully to understand its claims,

 which puts IBM at an extraordinary disadvantage.

Case 2:03-cv-00294-DAK-BCW            Document 672        Filed 04/28/2006        Page 3 of 24

        5. If the question is whether version, file, and line information is still needed even

 where the allegations concern misuse of methods and concepts (rather than copying of

 code), the answer is unquestionably "Yes." It is entirely possible for the party making the

 allegations of misuse to assemble such information: Where a method or concept is in fact

 used in a program, there must be lines of source code in the program that implement the

 method or concept. The alleging party need simply cite the program version, file, and

 lines in which that source code appears.

        6. In the next section of this declaration, I explain some basic background that

 helps in understanding why Mr. Rochkind's position is untenable. Section III details the

 fundamental disagreement between my original declaration and the Rochkind declaration.

 In Section IV, I respond to Mr. Rochkind's assertions that the 198 Items are essentially all

 methods and concepts claims and that such claims do not require version, file and line

 information. In Section V, I address Mr. Rochkind's assertion that IBM has more than

 enough information to defend itself. Finally, in Section VI, I consider Mr. Rochkind's

 claim regarding "willfulness".


        7. Several basic facts concerning this case are worth reviewing briefly to help

 make clear the difficulties presented by SCO's position on disclosure.

        8. First, SCO has alleged misuse of its intellectual property, claiming, among

 other things, that its System V Unix code was copied by IBM into AIX and/or Dynix and

 then contributed, with and without AIX and Dynix code, by IBM to Linux. For these

 allegations to be understood, SCO must (a) specify the System V code that was allegedly

 copied, (b) specify where in AIX and/or Dynix the code allegedly derived from System V

 appears, and (c) specify where in Linux the allegedly infringing code appears.

Case 2:03-cv-00294-DAK-BCW            Document 672         Filed 04/28/2006       Page 4 of 24

         9. Without such specification, how can IBM respond, much less prepare a

 defense? How, for instance, can it determine whether the code allegedly copied from

 System V is in fact protectable, or instead is unoriginal; an idea, process or procedure;

 dictated by externalities; or in the public domain? Such an analysis must proceed from

 the specific code, and absent an indication of what code is in question, the analysis cannot

 even begin.

         10. Second, the volume of code at issue in this case is so enormous as to make it

 pragmatically impossible to determine what code might be in question unless the version,

 file and lines are specified. To put this in more familiar terms, consider that a single

 recent version of Linux contains about 6 million lines which if printed would be about

 110,000 pages. In other words, a single recent version of Linux is the equivalent of a 218

 volume encyclopedia.1 Now consider that there are 597 distinct versions, and think of each

 version as an edition of the encyclopedia. Hence in the absence of a specification of

 version, file and line information for the allegedly misused code, SCO is essentially

 saying "somewhere in the 597 distinct editions of this multi-(in many cases 100+)volume

 encyclopedia you have misused some of our property."

         11. Without a specification of the System V code that has allegedly been copied,

 IBM cannot know with specificity even what IBM stands accused of misusing. Without a

 specification of where the accused code allegedly appears in Linux and AIX or Dynix,

 IBM faces a pragmatically impossible task of finding it. For the record, there are:

       Assuming 500 pages to a volume. Even the earliest and smallest version of Linux
 contains over 175,000 lines of text, the equivalent of over 3000 pages.

Case 2:03-cv-00294-DAK-BCW            Document 672         Filed 04/28/2006      Page 5 of 24

            At least 11 versions of System V code totaling almost 24 million lines of text2;

            At least 9 versions of AIX totaling almost 1.2 billion lines;

            At least 25 versions of Dynix totaling almost 157 millions lines; and

            At least 597 versions of Linux totaling almost 1.4 billion lines.

         12. In the case of the 198 challenged items, SCO thus has offered an impossibly

 non-specific accusation, attempting to leave both the interpretation of the allegations and

 the finding of the evidence (should there be any) as an exercise for IBM. Requesting

 version, file, and line numbers for all the code in question is no more unreasonable on the

 face of it than an encyclopedia publisher asking that an allegation of plagiarism be

 specified in terms of the edition, volume and page where the accused text appears, as well

 as a listing of the text from which it was allegedly copied.

         13. In the absence of such information, allegations are impossible even to analyze:

 imagine the publisher of the Encyclopedia Americana telling the publisher of the

 Encyclopedia Britannica "your encyclopedia contains material that was copied from us"

 and then refusing to specify what was copied (what text from which edition, volume and

 page of the Americana) or where it appears (which edition, volume and page of the



         14. Mr. Rochkind disagrees with my conclusion that the 198 Items are not

 disclosed with the requisite specificity. The primary reason for this disagreement is that

      Each complete version of an operating system is typically given a distinct "release
 number," as, for example, version 2.6.9 of Linux. The version counts given above list the
 number of distinct versions shown in the Declaration of Todd Shaughnessy, dated April 4,
 2006; the total lines of text cited report all text contained in both the complete versions
 and any additional "patches" (i.e., incremental changes), as listed in the Shaughnessy

Case 2:03-cv-00294-DAK-BCW              Document 672         Filed 04/28/2006        Page 6 of 24

 he used a very different test to evaluate specificity than I did. When the test that I

 understand to be the correct test is applied, the 198 Items come nowhere close to passing


           15. I was instructed by counsel for IBM to evaluate the 198 Items based on the

 language of the Court's orders of December 12, 2003, March 3, 2004, and July 1, 2005.

 As described in Exhibit A, I understand the orders to require the disclosure of the

 allegedly misused material by version, file and line of code. That is the standard (and

 most precise) means of identifying the code, methods and concepts, and concepts of an

 operating system.

           16. The Court's Order of December 12, 2003, states that SCO is required:

           (1) "To identify and state with specificity the source code(s) that SCO is claiming
           form the basis of their action against IBM." ( 4.)

           (2) "To respond fully and in detail to Interrogatory Nos. 1-9 as stated in IBM's
           First Set of Interrogatories" ( 1), which provide, for example, as follows:

                  Interrogatory 1: "Please identify, with specificity (by product, file
                  and line of code, where appropriate) all of the alleged trade secrets
                  and any confidential or proprietary information that plaintiff alleges
                  or contends IBM misappropriated or misused, . . . ."

                  Interrogatory 3: "Please . . . describe, in detail, . . . all places or
                  locations where the alleged trade secret or confidential information
                  may be found or accessed."

                  Interrogatory 4: "Please describe, in detail, . . . with respect to
                  any code or method plaintiff alleges or contends that IBM
                  misappropriated or misused, the location of each portion of such
                  code or method in any product, such as AIX, in Linux, in open
                  source, or in the public domain."

           (3) "To respond fully and in detail to Interrogatory Nos. 12 and 13 as stated in
           IBM's Second Set of Interrogatories" ( 2), which provide, for example, as

                  Interrogatory 12: "Please identify, with specificity (by file and
                  line of code), (a) all source code and other material in Linux . . . to

Case 2:03-cv-00294-DAK-BCW             Document 672         Filed 04/28/2006       Page 7 of 24

                which plaintiff has rights; and . . . how the code or other material
                derives from UNIX."

                Interrogatory 13: "[P]lease . . . describe in detail how IBM is
                alleged to have infringed plaintiff's rights. . . ."

        17. The Court's Order of March 3, 2004, required SCO to: (1) "provide and

 identify all specific lines of code that IBM is alleged to have contributed to Linux from

 either AIX or Dynix" ( 2), (2) "provide and identify all specific lines of code from Unix

 System V from which IBM's contributions from AIX or Dynix are alleged to be derived"

 ( 3), and (3) "provide and identify with specificity all lines of code in Linux that it claims

 rights to" ( 4, emphasis added). It is difficult to imagine instructions that are any clearer,

 more specific, or more unambiguous.

        18. The Court's Order of July 1, 2005 (at 4) reiterated SCO's obligations to

 specify its claims and ordered it to update its interrogatories accordingly.

        19. Note that the Court's orders required no more of SCO than SCO required of

 IBM. In its First Request for Production of Documents, SCO defined the term "identify"

 as follows:


        The term "identify" shall mean: . . .

        e. in the case of alleged trade secrets or confidential or proprietary
        information, whether computer code, methods, or otherwise, to give a
        complete and detailed description of such trade secrets or confidential or
        proprietary information, including but not limited to an identification of the
        specific lines and portions of code claimed as trade secrets or confidential
        or proprietary information, and the location (by module name, file name,
        sequence number or otherwise) of those lines of code within any larger
        software product or property." (Exhibit B (emphasis added).)

 I understand that SCO subsequently incorporated this identical definition in eight

 additional document requests, five additional sets of interrogatories, seven 30(b)(6)

Case 2:03-cv-00294-DAK-BCW             Document 672       Filed 04/28/2006        Page 8 of 24

 deposition notices, and three requests for admission, the latest of which was served on

 March 10, 2006. Thus, SCO itself has continuously demanded the same degree of

 specificity ordered by the Court and requested by IBM.

        20. Despite the language of the Court's orders, and of SCO's own discovery

 requests, the Final Disclosures do not provide version, file and line information for each

 of the 198 Challenged Items. As is illustrated in my original declaration (Addendum B),

 and summarized in the following table, SCO provides version, file and line information

 for very few of the Challenged Items:

                              Version(s)               File(s)             Line(s)

         System V                 1                       1                   0

         AIX                      1                       1                   0

         Dynix                    2                       3                   0

         Linux                    27                     149                  3

 Note that there is not even one Item for which SCO provides a complete set of


        21. Mr. Rochkind does not seem to disagree that SCO has not provided a

 complete set of coordinates for each of the 198 Items. Instead, he asserts that, with

 respect to many of the Items, SCO has provided sufficient detail relating to claims

 because it has summarized its allegations of misuse, provided documents relating to the

 alleged misuse, identified persons involved in the alleged misuse and/or pointed IBM to

 source code. (Rochkind Decl.  9.)

Case 2:03-cv-00294-DAK-BCW            Document 672        Filed 04/28/2006        Page 9 of 24

        22. It is true that SCO has, for most of the Items, summarized its allegations,

 listed persons involved in the alleged misuse and referred IBM to certain documents.

 That is simply not the appropriate measure of compliance, as I understand the Court's

 orders. Nor would it be the appropriate measure of compliance under SCO's own

 discovery requests. Mr. Rochkind's declaration defines his own standard of specificity

 and asserts that SCO's Final Disclosures, of which Mr. Rochkind claims to be the primary

 author, meet the standard.

        23. Putting aside the language of the Court's orders, it is difficult to consider the

 information SCO has provided as sufficiently specific when (1) many of the summaries

 are extremely general (e.g., Item 180 claims only that IBM misused the "internals" of

 System V Release 4, without any mention of which part of the several-million-line

 operating system was misused); (2) the documents provided are mostly documents that

 IBM provided to SCO, and they tell IBM little more than it would have known before

 SCO filed its complaint; (3) for many of the Items, SCO does not identify any individuals,

 it says only "IBM"; and (4) according to Exhibit B to Mr. Rochkind's declaration, SCO

 identified code with respect to no more than 16 of the 198 Items.

        24. If the Court's orders required only that SCO provide some minimal, additional

 information about its allegations, then I agree with Mr. Rochkind that it has done that. If

 they required that SCO provide the standard coordinates for identifying allegedly misused

 aspects of an operating system (code, methods and concepts), then SCO's disclosures

 regarding the 198 Items fall far short. For some of the Items (e.g., Item 93), the Final

 Disclosures reveal little more than the minimal description found in SCO's Complaint.

Case 2:03-cv-00294-DAK-BCW               Document 672     Filed 04/28/2006      Page 10 of 24


         25. Rather than disagree with the fact that SCO has not provided version, file and

  line information regarding any of the 198 Items, Mr. Rochkind devotes the better part of

  his declaration to rationalizing SCO's decision not to provide the information. Contrary

  to Mr. Rochkind's suggestion, however, there is no reason SCO could not have provided

  the missing information with respect to its methods and concepts-misuse allegations, as

  well as its code-misuse allegations.

         26. To begin, Mr. Rochkind seems to suggest that virtually all of the 198 Items

  concern methods and concepts rather than source code. (Rochkind Decl.  8-9.)

  According to Mr. Rochkind, less information is required to evaluate a method than is

  required to evaluate code. (Rochkind Decl.  10.) Thus, Mr. Rochkind states, there was

  no need for SCO to identify version, file and line information relating to methods and

  concepts. (Rochkind Decl.  10.) Putting aside the fact that the Court's orders -- on their

  face -- require version, file and line information for methods and concepts as well as code,

  Mr. Rochkind is mistaken both as to the number of Items that concern methods and

  concepts and the information needed fully to evaluate operating-system methods and


         27. Contrary to Mr. Rochkind's suggestion, a significant portion of the 198 Items

  concern the alleged misuse of code. As described in Exhibit C, the language of many of

  the challenged Items themselves relate to the alleged misuse of code. For example:

          Item 17:     "Port of discontiguous memory code from ptx to Linux 2.5".

          Item 22:     "Port of ptx NUMA code to Linux".

          Item 27:     "Transferring ptx source code to AIX developers".

Case 2:03-cv-00294-DAK-BCW            Document 672         Filed 04/28/2006       Page 11 of 24

  Thus, it is simply wrong for SCO's Mr. Rochkind to imply that the only Items in dispute

  concern methods and concepts.

         28. Mr. Rochkind suggests that all Items of allegedly misused of code are

  disclosed by SCO with appropriate line specificity. That is unfortunately patently false

  and Mr. Rochkind is ignoring dozens of the 198 challenged Items that do concern alleged

  code misuse. In fact, many of the Items that clearly concern the alleged misuse of code

  comprise SCO's most imprecise allegations. In 39 of the Items (Items 232 to 270), for

  example, SCO accuses IBM of making improper reference to Dynix source code as a

  basis for writing additional code, while providing essentially no further information. Each

  of these 39 items has an "Improper Disclosure" claim of the form: "Use of ptx [i.e.,

  Dynix] programming experience (and a fortiori exposure to related aspects of Unix

  System V) in programming [or `implementing'] _________," where the blank contains

  things such as "MP preemption and synchronization code", "i686 large-memory SMP

  systems", "code for SCSI Mid-layer Multi-Path IO", and so forth. That is, SCO is

  specifically accusing IBM of referring to Dynix code and System V code, and then using

  that as the basis for creating additional code (e.g., "MP preemption and synchronization

  code"). Yet there is absolutely no specification of any kind (no version, file, or line

  numbers) of which Unix code was allegedly referenced, or of which Dynix code was

  allegedly referenced. IBM is left to guess as to which of the 470,000-plus files and 156

  million-plus lines of Dynix code included within SCO's vague claims are in fact

  challenged by these Items.

         29. As if to further justify SCO's failure to provide version, file and line

  information, Mr. Rochkind suggests that it is not possible to identify version, file and line

  coordinates with respect to methods and concepts. (Rochkind Decl.  10.) That is simply

Case 2:03-cv-00294-DAK-BCW            Document 672        Filed 04/28/2006       Page 12 of 24

  incorrect. The methods and concepts employed in an operating system (or any computer

  program) are in the source code. It could not be otherwise: The source code of a

  program specifies all of its possible behavior. If that behavior truly embodies a method,

  that method must be expressed in specific lines of the source code; there is just no other

  way to do it.

         30. Consider, for example, Item 146, which alleges (among other things) that

  IBM improperly disclosed a method called "differential profiling". Simply put, the

  method suggests ways of finding performance bottlenecks by counting the events that

  happen inside a program and then analyzing those counts. But the counting and analyzing

  can be done only by code, i.e., source code written to keep track of the number of times an

  event happens and written to analyze the counts as explained in the method. Any time a

  method is used, it can only be because there is source code that implements it. It really is

  that simple. Hence, if System V, AIX, Dynix or Linux used that method, they must

  contain source code that implements it, and SCO ought to cite the specific lines of code.

         31. Although, as Mr. Rochkind states (Rochkind Decl.  10), methods and

  concepts are sometimes discussed in text books without reference to source code, such

  discussions are, most often, at a high level of generality. The mere fact that a method can

  be discussed generally without referring to source code does not mean that its

  corresponding source code cannot be identified. It can, and SCO--having alleged that

  System V, AIX, Dynix or Linux code somewhere embody a method--bears the

  responsibility of identifying the specific code it claims embodies that method.

         32. The disclosure of the corresponding source code also greatly aids in

  understanding the method, as Mr. Rochkind's own text illustrates. Despite his attempt to

  make the identification of source code seem irrelevant to the identification of methods and

Case 2:03-cv-00294-DAK-BCW            Document 672        Filed 04/28/2006        Page 13 of 24

  concepts (Rochkind Decl.  10), his own book on the subject of operating systems,

  Advanced Unix Programming (2d ed. 2004), (which he asserts, "explain[s] in detail how

  to use UNIX system calls" ( 5)) devotes considerable space to describing methods and

  concepts with reference to source code. For example, it states that "this new book

  includes thousands of lines of example code". (xii). Indeed, his chapter devoted to

  "Fundamental Concepts" describes UNIX concepts using, in many cases, nearly full-page

  excerpts of source code and even refers back to his own website to offer complete code

  listings where the excerpts are not enough. (See, e.g., id. at 24-38. (Exhibit D).)

         33. SCO's Chief Technology Officer, Sandeep Gupta, testified concerning the

  importance of having version, file and line information with respect to methods and

  concepts. Mr. Gupta was asked the following questions and provided the following


         Q.     "Okay. How would you determine whether a particular description
         was specific enough to describe an aspect of System V as a method?

         A.      "I have to look at the source code."

         Q.      "Okay. What would you do if you looked at the source code?

         A.    "I look at various steps that are taken, specific for that particular

         Q.     "Okay. So in order to determine what a particular method or
         concept is, you would actually have to look at the source code?

         A.      "In some cases, yes."

         Q.   "Okay. I mean, I -- I understand you just articulated a few from
         memory and --

         A.      "Yeah."

         Q.      "-- I'm impressed with that, but in general, would you have to look
         at the source code to be able to accurately describe a method or concept in

Case 2:03-cv-00294-DAK-BCW              Document 672       Filed 04/28/2006      Page 14 of 24

          A.     "That's my opinion, yes." (Exhibit E (03/17/2006 Gupta Dep. Tr.
          at 266-67) (objections omitted).)

          34. Furthermore, as stated, SCO itself specifically demanded that IBM identify

  methods and concepts with reference to files and lines of code. It did that, no doubt,

  because the standard means of identifying an operating system method with specificity is

  by file and line of code. I assume SCO would not have demanded that IBM provide

  information that could not be provided.

          35. In truth, it is even more important to have version, file and line information

  regarding methods and concepts claims than it is to have the information for code claims.

  When specific lines of source code are identified by a plaintiff who alleges they have been

  improperly copied, a defendant can at least automate the process of looking for literal

  infringement: he can set a computer to work searching through his own code to see if it

  contains the lines identified by the plaintiff.

          36. But the same cannot be said for methods and concepts. Consider once again

  the alleged "Improper Disclosure" in Item 146: "The idea is to compare corresponding

  buckets of the profile data to determine which portion of the code is most responsible for

  the slowdown." There are no automated techniques for finding the lines of code that

  embody that method. Because they are abstractions, methods and concepts must instead

  be located by manual review of the code, and given that there are between tens of millions

  (System V) and billions (AIX, Linux) of lines to be searched, locating such methods and

  concepts are simply untenable here. Given the size of the code base here, manual review

  is, as a practical matter, an impossible task.3 Hence, without a specification by SCO of

       Returning to our analogy of the two encyclopedias, imagine that the Americana
  accused the Britannica of copying a specific passage of Americana's text. Britannica

Case 2:03-cv-00294-DAK-BCW            Document 672        Filed 04/28/2006       Page 15 of 24

  the location of the code implementing the method, the claim cannot be adequately



         37. I stated in my original declaration that SCO's failure to provide version, file

  and line information makes it impossible, practically speaking, for IBM to defend itself.

  Mr. Rochkind disagrees. (Rochkind Decl.  7.) However, his view is supported only by

  naked assertions and does not survive even the weakest scrutiny.

         38. The kinds of questions that must be asked to defend against SCO's allegations

  are not a secret. They have been involved (more or less) in each of the 30-plus cases in

  which I have been retained as an expert to deal with alleged misappropriation of

  intellectual property, including in Computer Associates v. Altai, in which I served as an

  expert appointed by and for the Court.

         39. Among the many questions IBM must answer are the following:

                        Did IBM offer the Item to Linux?

                        Did the Item originate in or derive from System V and AIX or

                        Was the Item accepted into Linux and, if so, when and to what

                        Is the Item copyrightable (or is it unoriginal; a mere idea, process or
                        procedure; dictated by externalities; or in the public domain)?

                        Has the disclosure of the Item or its inclusion in Linux had a
                        negative effect on SCO or a positive impact on IBM?

  could do an automated search for that text. But imagine instead if Americana accused
  Britannica of using what Americana claimed to be its proprietary "non-Eurocentric
  method of describing history" (i.e., ensuring a more global, inclusive world view), and
  then refused to give any information about which edition(s), volume(s) or page(s) in
  Britannica had done that. Consider the nature and difficulty of the task Britannica would
  face in trying to find places that had used that method.

Case 2:03-cv-00294-DAK-BCW             Document 672        Filed 04/28/2006       Page 16 of 24

  I do not understand Mr. Rochkind to dispute the relevance of these questions, which must

  be answered on a line-by-line basis.

          40. Many thousands of persons have contributed to the development of Linux,

  and IBM has made many contributions to Linux, some of which represent only a few lines

  of code in a file comprised of hundreds of lines of code. The only way to know whether

  IBM made a given contribution is to know precisely what the alleged contribution is.

  Similarly, whether a given contribution originated in, or is derived from, System V, AIX

  or Dynix is a line-specific inquiry. One line may have; another may not have. Version,

  file and line information is no less critical to determining whether a line of code--and

  especially a method--is in Linux, since it is composed of millions of lines of codes and

  many thousands of methods and concepts and concepts.

          41. To determine whether an Item is copyrightable requires line information

  because that is the only way to assess originality, determine whether the line is merely an

  idea, process or procedure, evaluate whether the Item is dictated by programming

  practices, governed by standards, or in the public domain. These questions simply defy

  generalized examination. In a given file, one line might be original, whereas another

  might not; one might be in the public domain, whereas another might not; and so on. For

  these same reasons, it is also not possible to evaluate whether a method has a positive

  impact on Linux and IBM (or a detrimental impact on SCO) without understanding

  precisely what it is.

          42. Absent the production of the version, file and line information referenced in

  the Court's orders, it is very difficult, if not impossible, to answer these questions. As

  described in the Declaration of Todd Shaughnessy, dated April 4, 2006, the size of the

  code bases implicated by SCO's claims is enormous.

Case 2:03-cv-00294-DAK-BCW               Document 672      Filed 04/28/2006       Page 17 of 24

          Operating            Version(s)                File(s)              Line(s)

          System V                 11                    112,622            23,802,817

          AIX                       9                   1,079,986          1,216,698,259

          Dynix                    37                    472,176            156,757,842

          Linux                    597                  3,485,859          1,394,381,543

          Total:                   654                  5,150,643          2,791,640,461

         43. Mr. Rochkind does not disagree that the implicated code base is enormous.

  Nor does he disagree that SCO's Chris Sontag provided sworn testimony early in the case

  that it would take 25,000 person-years to review a code data base .2% the size of the stack

  of code at issue. Mr. Rochkind states only that he has helped SCO to provide enough

  information for IBM to find the 198 needles in the haystack. Having helped to decide

  what the needles are, Mr. Rochkind may well feel as if he knows what they are. But I do

  not. Nor do I believe that other independent experts would.

         44. Examining the only Item specifically mentioned in Mr. Rochkind's

  declaration, Item 146, makes the point. In Item 146, SCO complains about IBM's "Use

  of Dynix/ptx for Linux development" by reference to: (1) an email asking for help with a

  performance problem, (2) an email response with a suggested analysis technique

  (differential profiling), (3) a technical paper written by Paul McKenney, (4) a URL

  reference to scripts that might be of help, and (5) a list of 11 Linux files (names only, no

  versions or lines). Contrary to Mr. Rochkind's claim that I ignored these materials, in fact

  it was by examining them closely that I concluded, as stated in my original declaration,

  that SCO has provided no meaningful information about what IBM is alleged to have

  done wrong.

Case 2:03-cv-00294-DAK-BCW              Document 672       Filed 04/28/2006       Page 18 of 24

          45. The claim in Item 146 is sufficiently vague as to lead to several different

  interpretations. As it takes several pages to analyze all of the possible interpretations, and

  to point out all of Mr. Rochkind's errors, I have put the analysis in Exhibit F. The details

  can be found there; the summary points are simple enough:

             Mr. Rochkind points out that the email cited in Item 146 contains "an actual

             Linux patch" and "a specification for the files and lines being patched". He

             conveniently overlooks the fact that the patch, and the files and lines being

             patched, are all the "before" version of the code. That is, the patch and

             associated files contain the code that didn't work well enough, the code that

             the application of differential profiling was supposed to help repair. There is

             in fact no code cited that is alleged actually to contain the use of the method.

             Mr. Rochkind points out that there is "an exact URL reference to a 9-page

             technical paper by McKenney explaining the method and concept at issue."

             Indeed there is, and the paper was published in the open literature in 1995 (the

             1995 IEEE MASCOTS Symposium), six years prior to the email in question.

             As Mr. Rochkind points out, there are "11 Linux file paths" specified in Item

             146, but, as he omits to mention, no specific version of Linux is cited. Even

             so, unfortunately for his position, none of these files appears to deal with

             differential profiling.4

         Item 146 indicates yet another level of difficulty in deciphering SCO's claims: even
  where SCO does specify file names (but still not versions or line numbers), IBM still has
  to guess what SCO is talking about: 4 of the 11 Linux filenames in SCO's Item 146 are
  simply incorrect: there is no Linux file named arch/i386/oprofile/rmi_int.c,
  arch/i386/oprofile/rmi_int.c, arch/i386/oprofile/op_counter.c, or
  arch/i386/oprofile/op_x86_model.c. There are files whose names are close to these, and
  are likely what was intended, but this presents yet another step IBM must take to
  determine what SCO actually means.

Case 2:03-cv-00294-DAK-BCW             Document 672          Filed 04/28/2006       Page 19 of 24

          46. As described in Exhibit F, rather than clearly state its claims to Item 146, SCO

  identifies a series of dots and leaves IBM to try to connect them. The problem is they do

  not connect. At most they leave IBM to guess as to which of any number of claims SCO

  might actually be making. To defend itself, IBM is left to respond not just to what is at

  issue -- which is not clear -- but to all of the possibilities. For Items like Item 146, there

  are at least a handful of possibilities. As to other items, the possibilities are almost

  innumerable. When SCO accuses IBM of misusing the internals of System V (e.g.,

  Item 180) or of misusing its experience with Dynix/ptx, for example, SCO accuses IBM

  of misusing any one of the millions of lines of code and the thousands of methods and

  concepts contained in these operating systems.

          47. Even if IBM could feasibly chase all of the possibilities held open by the Final

  Disclosures (which clearly it could not do without years of additional effort), the

  generality, uncertainty and ambiguity inherent in the final disclosures are sure to lead to

  games of "where's the pea" during the expert and summary judgment phases of the case.

  Based on the information SCO has provided (or, more accurately, not provided), it is

  difficult to imagine any meaningful exchange of views among experts. Likewise, a court

  can hardly evaluate at summary judgment what cannot be defined. Had SCO provided

  full code coordinates for the allegedly misused material, games of "where's the pea"

  would not be possible. SCO's claims could have been understood and analyzed.

  Unintentional allegations could have been eliminated.

          48. While I do not believe that IBM can fairly defend itself absent version, file

  and line information for each of the Items, it would -- at the risk of stating the obvious --

  require a very significant period of time for IBM to conduct an investigation into the

  general allegations set out in the 198 Items. Without engaging a very large corps of

Case 2:03-cv-00294-DAK-BCW            Document 672        Filed 04/28/2006       Page 20 of 24

  experts, it would take years. Even then it is very unlikely that IBM could succeed in

  learning what is ultimately known only to SCO: its allegations.


         49. Finally, Mr. Rochkind addresses IBM's contention that SCO acted willfully in

  failing to provide version, file and line information. (Rochkind Decl.  16-18.)

  Mr. Rochkind claims that IBM is wrong to state that SCO acted "willfully in not

  specifying its claims" and wrong that "SCO has declined, as a practical matter, to tell IBM

  what is in dispute." (Rochkind Decl.  16.) Here again, Mr. Rochkind's view appears to

  turn on his own, self-defined view of the appropriate standard.

         50. I am not a legal expert, and do not pretend to be an authority on the meaning

  of the term "willfully" for purposes of a court's deciding whether a party should be

  limited in submitting evidence in support of its claims. In responding to Mr. Rochkind's

  assertion, however, I rely on the definition of the term used in the cases provided by

  counsel for IBM, e.g., Schroeder v. Southwest Airlines, 129 Fed. Appx. 481, 484-85, 2005

  WL 984495 (10th Cir. 2005) (holding that "[w]illful failure means `any intentional failure

  as distinguished from involuntary noncompliance. No wrongful intent need be shown'");

  F.D.I.C. v. Daily, 956 F.2d 277, 1992 WL 43488, at *3-6 (10th Cir. 1992) (same); and In

  re Standard Metals Corp., 817 F.2d 625, 628-29 (10th Cir. 1987) (same).

         51. Using the definition of "willfully" set out in these cases, I have no difficulty

  concluding that SCO acted "willfully" in submitting its Final Disclosures and omitting the

  information called for in the Court's orders. The Court's orders clearly call for version,

  file and line information, with respect both to code and methods and concepts.

  Identifying code and methods and concepts by version, file and line of code is the

  standard method of identifying operating system source code and methods and concepts

Case 2:03-cv-00294-DAK-BCW             Document 672         Filed 04/28/2006       Page 21 of 24

  with specificity. SCO asked nothing less than this of IBM. There is no reason it could

  not be provided here. Indeed, without it, the 198 Items are too vague and indefinite to

  permit complete analysis.

         52. As Mr. Rochkind's declaration makes clear, SCO does not claim to have

  assembled the Final Disclosures unwittingly. It plainly did not, as evidenced by the fact

  that SCO provides version, file and line information for a number of Items that are not

  challenged in this motion. There is no dispute that SCO made a deliberate decision to

  provide the information it provided and the information it did not. (Rochkind Decl.  10.)

  And SCO deliberately created a different standard to apply to itself than it demanded of

  IBM, and the court required. SCO's failure to provide version, file and line information

  was not unknowing or inadvertent.

         53. Moreover, the information omitted from SCO's disclosures is unquestionably

  within SCO's control. (Rochkind Decl.  14 n.3.) The Court's orders, as I understand

  them, direct SCO (in substantial part) to make its allegations specific. For example, to the

  extent SCO claims that IBM improperly used Dynix code and methods and concepts in

  contributing to Linux (and the vast majority of SCO's allegations are of this type), the

  orders (on their face) require SCO to "describe, in detail, . . . with respect to any code or

  method plaintiff alleges or contends that IBM misappropriated or misused, the location of

  each portion of code or method in any product." Only SCO knows what it alleges. No

  amount of investigation by IBM can connect the dots. Yet SCO systematically omitted

  this information from the 198 Items as described in Addendum B to my initial declaration.

         54. In sum, Mr. Rochkind's claim that SCO did not willfully withhold

  information in its possession with respect to version, file and line of code misses the

  point. As has been demonstrated, it is possible to obtain version, file and line information

Case 2:03-cv-00294-DAK-BCW             Document 672         Filed 04/28/2006        Page 22 of 24

  with respect to methods and concepts if an effort to do so is undertaken. SCO, simply put,

  has willfully failed to undertake any such effort.


         55. Upon careful review of Mr. Rochkind's declaration, I find that he fails even to

  address the central assertions in my opening declaration. He does not -- and could not --

  dispute that SCO has not provided version, file and line information regarding each of the

  198 Items at issue in IBM's motion

         56. There is no reason SCO could not have provided this information, including

  with respect to methods and concepts, for which it is even more necessary, not less

  necessary. SCO's own discovery demands make the point.

         57. Without the missing information, IBM lacks precisely the kind of information

  needed to conduct a basic inquiry relating to the facts of SCO's claims. Given enough

  time, IBM might be able to discover some of the information SCO has failed to provide.

  It will never be able to find all of the information, however, because only SCO knows its


         58. It is for this reason, in significant part, that I have no difficulty disagreeing

  with Mr. Rochkind's statements regarding willfulness. To my knowledge, SCO has never

  argued (and could not credibly argue) that SCO's failure to provide version, file and line

  information was an oversight.

Case 2:03-cv-00294-DAK-BCW   Document 672   Filed 04/28/2006   Page 23 of 24


         59. I declare under the penalty of perjury that the foregoing is true and correct.

                                                      /s/ Randall Davis
                                                        Randall Davis

  Date:   28 April 06

  Place:  Cambridge, MA

Case 2:03-cv-00294-DAK-BCW              Document 672         Filed 04/28/2006       Page 24 of 24


                                 CERTIFICATE OF SERVICE
           I hereby certify that on the 28th day of April, 2006, I electronically filed the foregoing

   Rebuttal Declaration of Randall Davis, in redacted form, with the Clerk of Court using the

   CM/ECF system, which sent notification of such filing to the following:

               Brent O. Hatch
               Mark F. James
               HATCH, JAMES & DODGE, P.C.
               10 West Broadway, Suite 400
               Salt Lake City, UT 84101

I further certify that on the 1st day of May, 2006, a true and correct copy of the foregoing was
sent by U.S. Mail, postage prepaid, to:

               Stephen N. Zack
               Mark J. Heise
               100 Southeast Second Street, Suite 2800
               Miami, FL 33131

               Robert Silver
               Edward Normand
               333 Main Street
               Armonk, NY 10504

                                                      /s/ Todd M. Shaughnessy