Microsoft has filed its Memorandum in Support [PDF] of its Renewed Motion for Judgment as a Matter of Law in Novell v. Microsoft, the antitrust litigation Novell brought regarding WordPerfect.
It's 137 pages long. Yes. 137 pages. So somebody nominate me for sainthood, because I have done it as text for you, including all 77 footnotes, all of which add up to Microsoft saying it didn't do anything wrong to Novell, and it didn't owe Novell anything anyhow, so no reasonable jury could find for Novell, and even if Novell thinks Microsoft did some things a little harsh or deceptive, that's not covered by antitrust law. So Microsoft suggests that the judge bypass a second jury trial and just find for Microsoft as a matter of law. If you've ever wondered why Microsoft never seems to change for the better, I think this document explains it. They feel antitrust law lets them do the things they do. It would like to get off on a technicality, or many technicalities, so it offers more than one the judge can hang his hat on if he's so inclined. And Microsoft clearly believes he just might be.
'Unreasonable' jurors might not know the law and how much a monopolist can get away with, but the judge knows. And he even expressed at trial that Novell's case was weak, so why go through a second trial?
Jump To Comments
The general rule, Microsoft reminds the court is that with a limited exception “a monopolist generally has a right to refuse to cooperate with a competitor.” The Supreme Court itself has ruled that the general rule is that even a monopolist has "no antitrust duty to deal with its rivals at all". And, as one court put it, even "an act of pure malice by one
business competitor against another does not, without more, state a claim under the federal
antitrust laws." It even quotes from a 7th Circuit decision from 1986 -- they are in the 10th Circuit here -- that the "lawful monopolist should be free to compete like everyone else; otherwise the antitrust laws would be holding an umbrella over inefficient competitors.” Novell shot its own self in the foot by being inefficient. Microsoft's actions are not the reason they lost out, and it didn't harm competition in the PC operating system market, so it shouldn't be held responsible for any damages.
Whatever Microsoft did, they claim, they had a good business reason, and "Under controlling Tenth Circuit law, a legitimate business justification need be no
more than a desire to protect the profitability of one's business," Microsoft writes, adding: Here, the trial evidence established that Microsoft had at least three justifications for withdrawing support for the namespace extension APIs—all of which served the purpose of enhancing the value of Microsoft’s Windows 95 operating system: (1) third-party applications that used the namespace extension APIs could cause the Windows 95 operating system to crash; (2) the design of the namespace extension APIs was not compatible with future versions of Microsoft Windows under development; and (3) the namespace extension APIs did not achieve the functionality that Gates had anticipated. As for Novell's allegation that the issue was Microsoft's withdrawal of support for the namespace extension APIs after "evangelizing" them constitutes deception in violation of the Sherman Act (see the infamous Bill Gates email [PDF] about his decision to withdraw support), Microsoft says deception isn't "cognizable" under antitrust laws:
Deceiving a competitor does not give rise to an antitrust claim under Section 2 of the Sherman Act. Deception sounds in tort, and the Supreme Court has stated that the federal antitrust laws “do not create a federal law of unfair competition or ‘purport to afford remedies for all torts committed by or against persons engaged in interstate commerce.’”...“Even an act of pure malice by one business competitor against another does not, without more, state a claim under the federal antitrust laws.” So malice isn't actionable under antitrust laws. It is "nothing more than a tort claim" and this isn't one. That is Microsoft's argument. It couldn't find any cases that held that deception could be an antitrust violation. The jurors didn't think Microsoft hadn't done anything. And one lawyer explained to me that antitrust law can actually be anything. Not finding a case on one's precise fact pattern doesn't prove that it can't qualify as an antitrust violation. Microsoft quotes from a case without noticing the part that matters most, so I'll highlight it: "While the disparagement of a rival . . .
may be unethical and even impair the opportunities of a rival, its harmful effects on competitors
are ordinarily not significant enough to warrant recognition under § 2 of the Sherman Act." I read that case as saying that if the effects are significant, it absolutely could warrant recognition under the Sherman Act. Microsoft seems to acknowledge that, but says even if it could reach over that bar, Novell didn't provide any evidence of deception. It didn't even use the word in its complaint. Note how carefully this sentence is structured: There was no evidence that the decision to withdraw support for the namespace
extension APIs was made in order to harm Novell's development efforts, or that it even had
anything to do with Novell. Microsoft lawyers didn't write:The decision to withdraw support for the namespace
extension APIs was not made in order to harm Novell's development efforts, nor did it have
anything to do with Novell. Do you see the difference? If you ask one of your kids if he ate the cookies, if he answered, "There is no evidence that I did," you'd make a note of that, would you not, as being more equivocal than a simple, "No." Ditto here. There is no *evidence* Microsoft ate the cookies. Also, if you compare that claim by Microsoft with the Bill Gates email, linked above, note that he wrote this about why he decided to withdraw support: "I have decided that we should not publish these extensions. We should wait until we have a way to do a high level of integration that will be harder for the likes of Notes, Wordperfect to achieve, and which will give Office a real advantage." So. What do you think? That there was *no* evidence the decision was made to harm Novell or about anything to do with Novell?
If you were wishing to have more details about what happened at the trial, this document is for you, because Microsoft laboriously goes over the testimony, from its own perspective, of course. We'll hear from Novell on this later, although I really feel for them, having to quickly respond to 137 pages in a quick turn-around.
As you recall, the trial back in November of 2011 ended with a hung jury, so there should be a re-do with a new jury, but Microsoft is trying to end run that second trial by asking the judge to end it with a favorable ruling on Microsoft's motion instead of letting it go to a new jury. Then, if Novell appeals, it can go to the 10th Circuit Court of Appeals and let Microsoft get this mess wiped up that way, saying it would be a more "efficient and sensible" way "toward a final resolution". But that makes no sense, unless one assumes that the court of appeals will uphold a decision in favor of Microsoft. Maybe it would, but then again, maybe it wouldn't. It didn't last time this same judge tried to end the litigation by fiat, so to speak. If it didn't a second time, it'd be back to a jury trial anyway. So I fail to see the efficiency. Microsoft and I have trouble meshing perfectly in the logic department. Or the ethics department. But for sure, this document is for the appeals court or above. Here's the original motion [PDF] Microsoft filed in November during the first trial, in case you'd like to compare, with the Memorandum in Support [PDF], which is only about half as long as this new one.
Here's a bit of Microsoft's current argument that no reasonable jury could find against it: To prevail at trial, Novell was required to prove (a) that Microsoft's decision to
withdraw support for the namespace extension APIs was anticompetitive conduct prohibited by
Section 2 of the Sherman Act; (b) that this conduct "delay[ed] . . . Novell's development of the
versions of WordPerfect and Quattro Pro that were optimized for Windows 95," Novell, Inc. v.
Microsoft Corp., 699 F. Supp. 2d 730, 743 (D. Md. 2010); and (c) that the delay caused by
Microsoft's conduct "also caused anticompetitive harm in the PC operating system market." Id. In plain English, Microsoft says Novell doesn't have sufficient evidence that what Microsoft did is the sort of thing the law specifically forbids.
Since all 12 jurors at the first trial agreed that Microsoft had misbehaved, it's a bit of a stretch that no *reasonable* jury could so find, but this judge has ruled in Microsoft's favor over and over, so naturally the lawyers will try for a quick ending. And it could work, but if it does, Novell will be able to appeal again. I doubt it would fail to do so, having come this far.
As you are reading this document, if you happen to remember an exhibit from the Comes v. Microsoft collection that disproves or rebuts what Microsoft has written, please note it in your comments, so I can take a look.
The filings:
01/31/2012 - 489 - NOTICE OF INTENT TO REDACT 461 Transcript,,,, filed by Defendant Microsoft. (Attachments: # 1 Redaction Request)(Jardine, James) Modified by changing text from motion to notice on 2/3/2012 (rks). (Entered: 01/31/2012)
02/03/2012 - 490 - MOTION for Leave to File Under Seal filed by Defendant Microsoft. (Attachments: # 1 Text of Proposed Order)(Jardine, James) (Entered: 02/03/2012)
02/03/2012 - 491 - MEMORANDUM in Support re 490 MOTION for Leave to File Under Seal filed by Defendant Microsoft. (Jardine, James) (Entered: 02/03/2012)
02/03/2012 - 492 - NOTICE OF CONVENTIONAL FILING of Exhibit A and Exhibit Q filed by Defendant Microsoft re 491 Memorandum in Support of Motion, 490 MOTION for Leave to File Under Seal (Jardine, James) (Entered: 02/03/2012)
02/03/2012 - 493 - Modification of Docket: Error: Notice of Intent to Redact Transcript was filed as a motion to redact instead of the notice event and the Redaction Request was attached as an exhibit. Per the court's transcript policy the redaction request should have been filed separately so that it would be a restricted document. Correction: Termed the motion, changed the text to read as a notice of intent to redact, restricted the Redaction Request. re 489 MOTION to Redact 461 Transcript. (rks) (Entered: 02/03/2012)
02/03/2012 - 494 - MEMORANDUM in Support re 396 MOTION for Judgment as a Matter of Law (Renewed) filed by Defendant Microsoft. (Jardine, James) (Entered: 02/03/2012)
02/03/2012 - 495 - AFFIDAVIT/DECLARATION of Steven L. Holley in Support re 396 MOTION for Judgment as a Matter of Law (Renewed) filed by Defendant Microsoft. (Attachments: # 1 Exhibit A, # 2 Exhibit B, # 3 Exhibit C, # 4 Exhibit D, # 5 Exhibit E, # 6 Exhibit F, # 7 Exhibit G, # 8 Exhibit H, # 9 Exhibit I, # 10 Exhibit J, # 11 Exhibit K, # 12 Exhibit L, # 13 Exhibit M, # 14 Exhibit N, # 15 Exhibit O, # 16 Exhibit P, # 17 Exhibit Q)(Jardine, James) (Entered: 02/03/2012)
02/03/2012 - 496 - NOTICE OF FILING of Appendix re 494 Memorandum in Support of Motion, 495 Affidavit/Declaration in Support of Motion, 396 MOTION for Judgment as a Matter of Law (Renewed) filed by Defendant Microsoft. (Attachments: # 1 Appendix Part 1, # 2 Appendix Part 2, # 3 Appendix Part 3, # 4 Appendix Part 4, # 5 Appendix Part 5, # 6 Appendix Part 6, # 7 Appendix Part 7, # 8 Appendix Part 8, # 9 Appendix Part 9)(Jardine, James) (Entered: 02/03/2012)
02/03/2012 - 497 - **SEALED DOCUMENT** Exhibit A re 495 Affidavit/Declaration in Support of Motion, filed by Defendant Microsoft. (asp) (Entered: 02/06/2012)
02/03/2012 - 498 - **SEALED DOCUMENT** Exhibit Q re 495 Affidavit/Declaration in Support of Motion, filed by Defendant Microsoft. (asp) (Entered: 02/06/2012)
[ Update: The parties agree that the schedule should be altered, so Novell files its opposition on March 9 and then Microsoft replies on March 30:
02/14/2012 - 499 - NOTICE
OF FILING of Letter to the Honorable J. Frederick Motz filed by
Defendant Microsoft. (Jardine, James) (Entered: 02/14/2012)
From the document:The parties have agreed on a minor modification to the briefing
schedule on Microsoft's renewed motion for judgment as a matter of law.
Subject to the Court's approval, Novell will file a memorandum in
opposition to Microsoft's motion on March 9 and Microsoft will file its
reply memorandum on March 30. Even with the requested modification, the
motion will be fully briefed well in advance of the mediation session
with Chief Magistrate Judge Grimm, which is scheduled to take place on
April 24. That tells us also that there will be a mediation session on April 24. - End Update.]
Here's the Memorandum in support of Microsoft's motion, as text:
**************************
David B. Tulchin
Steven L. Holley
Sharon L. Nelles
Adam S. Paris
SULLIVAN & CROMWELL LLP
[address, phone]
James S. Jardine (A1647)
RAY QUINNEY & NEBEKER
[address, phone]
Steven J. Aeschbacher (A4527)
MICROSOFT CORPORATION
[address, phone]
Attorneys for Microsoft Corporation
______________________
IN THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF UTAH
CENTRAL DIVISION
____________________
NOVELL, INC.,
Plaintiff,
-v-
MICROSOFT CORPORATION,
Defendant.
________________________
MICROSOFT'S MEMORANDUM IN
SUPPORT OF ITS RENEWED MOTION
FOR JUDGMENT AS A MATTER OF
LAW
Civil No. 2:04 CV 1045
Honorable J. Frederick Motz
____________________________
February 3, 2012
TABLE OF CONTENTS
Page
PRELIMINARY STATEMENT ..............................2
SUMMARY OF THE ARGUMENT ............................4
STATEMENT OF FACTS ..................................................11
A. The Complaint ..............................................11
B. Pre-Trial Proceedings ........................................14
C. Trial ..........................................18
1. The Development of Windows 95 and the M6 Beta .................................18
2. Microsoft’s Decision to Withdraw Support for the Namespace Extension APIs ....................................................24
3. The October 3 Decision Was Not Made to Harm Novell .........................30
(a) Before October 3, Novell Said that It Was Not Using the Namespace Extension APIs......................................31
(b) After October 3, Novell Said It Was “OK” About the Decision to Withdraw Support .....................................31
4. There Are No Novell Documents Showing that The October 3 Decision Hurt Novell .................................34
5. WordPerfect Was Often Tardy and Novell Struggled to Catch Up
After Its 1994 Acquisition of WordPerfect ............................38
6. PerfectOffice 3.0, a Product Unaffected by The October 3 Decision, Was Not a Success in the Marketplace .................................45
7. Quattro Pro Delays Caused the Delay in Releasing PerfectOffice
for Windows 95 ...........................................48
8. Novell Chose the Most Difficult and Time-Consuming Path
Toward Release of Its Products for the Windows 95 Platform ................53
D. The Rule 50(a) Motion ...........................................57
ARGUMENT.....................................................58
TABLE OF CONTENTS
(continued)
Page
I. Microsoft's Withdrawal of Support for the Namespace Extension APIs Did Not
Harm Competition ................................................59
A. No Reasonable Jury Would Have a Legally Sufficient Evidentiary Basis to
Find that Microsoft Harmed Competition Under Either of Novell's Theories ......59
1. Novell's Franchise Applications Theory Was Unsupported by, and
Contrary to, the Evidence ...........................................60
2. Novell's Software Lacked All Three Required Elements of
Middleware .......................................69
(a) Novell's Software Was Not Cross-Platform .................................70
(b) Novell's Software Was Not Available on All or Nearly All
PCs ..............................................73
(c) Novell's Software Did Not Expose Sufficient APIs to Allow
ISVs to Write General-Purpose Personal Productivity
Applications...................................................75
B. The Applicable Causation Standard Is Whether Microsoft's Withdrawal of
Support for the Namespace Extension APIs "Contributed Significantly" to
Maintenance of Microsoft's Monopoly in the PC Operating System Market,
and Novell Came Nowhere Close to Meeting that Standard ................................82
C. The Allegedly Wrongful Conduct Could Not Have Harmed Competition in
the PC Operating System Market Because the Evidence Showed that the
Timely Release of PerfectOffice Would Have Enhanced Microsoft's
Monopoly ......................................................87
II. A Reasonable Jury Would Not Have a Legally Sufficient Evidentiary Basis to Find
that Microsoft Engaged in Anticompetitive Conduct ....................................90
A. Microsoft's Decision to Withdraw Support for the Namespace Extension
APIs Does Not Fall Within the Limited Aspen Skiing Exception ........................90
1. The Withdrawal of Support for the Namespace Extension APIs
Did Not "Terminate" Microsoft's Relationship with Novell ....................94
2. The Withdrawal of Support for the Namespace Extension APIs
Did Not Deny Novell Information or Support Available to All
Other ISVs ....................................................97
ii
TABLE OF CONTENTS
(continued)
Page
B. Common Practice in the Software Industry and the Terms of the Relevant
License Agreements Permitted Microsoft to Withdraw Support for APIs in
a Beta Version of Windows 95 .........................................99
C. Microsoft's Decision to Withdraw Support for the Namespace Extension
APIs Was Based on Legitimate Business Justifications .....................................105
1. A Third-Party Application Using the Namespace Extension APIs
Could Crash Windows .........................106
2. Supporting the Namespace Extension APIs Would Lock Future
Microsoft Operating Systems Into the Design of the Windows 95
Shell .....................................................109
3. The Namespace Extension APIs Did Not Provide the Functionality
Bill Gates Had Contemplated .............................................109
4. Novell's Experts Failed to Rebut Microsoft's Justifications for
Withdrawing Support for the Namespace Extension APIs .....................110
D. Novell's Attempt to Base Its Claim on a "Deception" Theory Has No Basis
in Law or in Fact ..............................................111
1. Novell's Purported Claim for Deception Is Not Cognizable Under
the Antitrust Laws .......................................111
2. There Was No Evidence of Any Deception ...............................113
III. Microsoft's Withdrawal of Support for the Namespace Extension APIs Did Not
Cause a Delay in the Release of PerfectOffice for Windows 95 ....................................115
A. Quattro Pro Caused the Delay in Releasing PerfectOffice for Windows 95 ......115
B. In Any Event, By Choosing the Most Time-Consuming and Difficult
Option, Novell Cannot Blame Microsoft for the Delay ......................................117
IV. Novell Is Not Entitled to Any Damages as a Matter of Law ..........................................118
A. Novell's Damages Models Depended on the Assumption that PerfectOffice
Would Have Been Released Within 60 Days of the Release of Windows 95 ....119
B. Because Warren-Boulton Failed to Account for Novell's Own
Responsibility for the Delay, Novell Is Entitled to No Damages .......................120
iii
TABLE OF CONTENTS
(continued)
Page
V. Novell Suffered No Cognizable Antitrust Injury .......................................123
VI. Novell Released Any Claim for Harm to PerfectOffice ..................................126
A. Novell Released Any Claim for Harm to PerfectOffice. .....................................127
B. Novell Released Any Claims for Harm to Other Products Not Pled in Its
Complaint ..........................................128
VII. Novell Sold Its Claim to Caldera ......................................130
VIII. Novell's Claim Is Barred by the Statute of Limitations ..............................133
CONCLUSION ...............................................137
iv
TABLE OF AUTHORITIES
Page(s)
CASES
American Professional Testing Service, Inc. v. Harcourt Brace Jovanovich
Legal & Professional Publications, Inc., 108 F.3d 1147 (9th Cir. 1997) .............................112
Aspen Skiing Co. v. Aspen Highlands Skiing Corp.,
472 U.S. 585 (1985) ........................................... passim
Associated General Contractors, Inc. v. California State Council of Carpenters,
459 U.S. 519 (1983) .........................................................125
Atlantic Richfield Co. v. USA Petroleum Co.,
495 U.S. 328 (1990) ..............................................123, 124
Bankers Trust Co. v. Lee Keeling & Associates, Inc.,
20 F.3d 1092 (10th Cir. 1994) ................................59
Bell v. Dow Chemical Co.,
847 F.2d 1179 (5th Cir. 1988) ................................110
Bigelow v. RKO Radio Pictures,
327 U.S. 251 (1946) .......................................119, 121
Blue Shield v. McCready,
457 U.S. 465 (1982) ...................................................125
Brooke Group Ltd. v. Brown & Williamson Tobacco Corp.,
509 U.S. 209 (1993) .....................................99, 111
Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc.,
429 U.S. 477 (1977) ...................................................124
Century 21 Real Estate Corp. v. Meraj International Investment Corp.,
315 F.3d 1271 (10th Cir. 2003)........................................59
Christy Sports, LLC v. Deer Valley Resort Co.,
555 F.3d 1188 (10th Cir. 2009) ................................ passim
Compliance Marketing, Inc. v. Drugtest, Inc.,
2010 U.S. Dist. LEXIS 34315 (D. Colo. April 7, 2010) ...........................................91
Conwood v. U.S. Tobacco Co.,
290 F.3d 768 (6th Cir. 2002) ......................................112
v
TABLE OF AUTHORITIES
Page(s)
CASES
(continued)
Daisy Mountain Fire District v. Microsoft Corp.,
547 F. Supp. 2d 475 (D. Md. 2008) ....................................92
Daviscourt v. Columbia State Bank,
2009 U.S. Dist. LEXIS 16815 (D. Colo. Feb. 20, 2009) ..............................125
Elliott Industries Ltd. v. BP America Production Co.,
407 F.3d 1091 (10th Cir. 2005) .....................11, 123, 124, 125, 126
Four Corners Nephrology Associates, P.C. v. Mercy Medical Center
of Durango, 582 F.3d 1216 (10th Cir. 2009) ..................... passim
Full Draw Productions v. Easton Sports, Inc.,
82 F.3d 745 (10th Cir. 1999) ..................................................124
Greater Rockford Energy & Tech. Corp. v. Shell Oil Co.,
790 F. Supp. 804 (C.D. Ill. 1992) .....................................123
Gregory v. Fort Bridger Rendezvous Association,
448 F.3d 1195 (10th Cir. 2006) ..............................98
Haynes Trane Service Agency, Inc. v. American Standard, Inc.,
51 F. App'x 786 (10th Cir. 2002) .......................124
Haynes Trane Service Agency, Inc. v. American Standard, Inc.,
573 F.3d 947 (10th Cir. 2009) ...................................132
Herrera v. Lufkin Industries, Inc.,
474 F.3d 675 (10th Cir. 2007) ...................................59
Homans v. City of Albuquerque,
366 F.3d 900 (10th Cir. 2004) ....................................125
Image Technical Services, Inc. v. Eastman Kodak Co.,
125 F.3d 1195 (9th Cir. 1997) ...................................122
In re Independent Service Organizations Antitrust Litigation,
989 F. Supp. 1131 (D. Kan. 1997) ..............................92
v
TABLE OF AUTHORITIES
Page(s)
CASES
(continued)
In re Microsoft Corp. Antitrust Litigation,
274 F. Supp. 2d 743 (D. Md. 2003) .......................................92
In re Warfarin Sodium Antitrust Litigation,
1998 U.S. Dist. LEXIS 19555 (D. Del. Dec. 7, 1998) .............................112
In2 Networks, Inc. v. Honeywell International,
2011 U.S. Dist. LEXIS 117589 (D. Utah Oct. 12, 2011) ................................92
Insignia Systems, Inc. v. News American Marketing In-Store, Inc.,
661 F. Supp. 2d 1039 (D. Minn. 2009) ..........................122
Intergraph Corp. v. Intel Corp.,
195 F.3d 1346 (Fed. Cir. 1999) ...............................98, 99, 103, 111
International Travel Arrangers, Inc. v. Western Airlines, Inc.,
623 F.2d 1255 (8th Cir. 1980) ................................112
Jackson v. State of Alabama State Tenure Commission,
405 F.3d 1276 (11th Cir. 2005) ............................132
Leh v. General Petroleum Corp.,
382 U.S. 54 (1965) ...........................................11, 133, 134
MCI Communications Corp. v. American Telephone and Telegraph Co.,
708 F.2d 1081 (7th Cir. 1983) .......................119, 121, 122
Midwest Underground Storage, Inc. v. Porter,
717 F.2d 493 (10th Cir. 1983) ...................................112
Multistate Legal Studies v. Harcourt Brace Jovanovich Legal & Professional
Publications, 63 F.3d 1540 (10th Cir. 1995) ............................110
New York v. Microsoft,
224 F. Supp. 2d 76 (D.D.C. 2002) ....................................111
Novell, Inc. v. Microsoft Corp.,
2005 U.S. Dist. LEXIS 11520 (D. Md. June 10, 2005) ...........................15, 133
Novell, Inc. v. Microsoft Corp.,
429 F. App'x 254 (4th Cir. 2011) ...............................15, 127, 128, 132
vi
TABLE OF AUTHORITIES
Page(s)
CASES
(continued)
Novell, Inc. v. Microsoft Corp.,
505 F.3d 302 (4th Cir. 2007) ......................................15, 79, 123, 125, 133
Novell, Inc. v. Microsoft Corp.,
699 F. Supp. 2d 730 (D. Md. 2010) ............................... passim
Olympia Equipment Leasing Co. v. Western Union Telegraph Co.,
797 F.2d 370 (7th Cir. 1986) .....................................100
Pacific Bell Telephone Co. v. linkLine Communications, Inc.,
555 U.S. 438 (2009) ...................................4, 91, 99
R.C. Dick Geothermal Corp. v. Thermogenics, Inc.,
890 F.2d 139 (9th Cir. 1989) ...................................125
Re/Max International v. Realty One, Inc.,
900 F. Supp. 132 (N.D. Ohio 1995) ........................123
Reazin, M.D. v. Blue Cross & Blue Shield of Kansas, Inc.,
899 F.2d 951 (10th Cir. 1990) ................................126
Smith v. Aztec Well Servicing Co.,
462 F.3d 1274 (10th Cir. 2006) ...............................58
Telex Corp. v. International Business Machines Corp.,
510 F.2d 894 (10th Cir. 1975) ..........................9, 99, 102
Trace X Chemical, Inc. v. Canadian Industries, Ltd.,
738 F.2d 261 (8th Cir. 1984) ..................................102
United States v. Colgate & Co.,
250 U.S. 300 (1919) .............................................91
United States v. Microsoft Corp.,
231 F. Supp. 2d 144 (D.D.C. 2002) .........................83
United States v. Microsoft Corp.,
253 F.3d 34 (D.C. Cir. 2001) ..........................................62, 82
vii
TABLE OF AUTHORITIES
Page(s)
CASES
(continued)
United States v. Syufy Enterprises.,
903 F.2d 659 (9th Cir. 1990) ...........................................100
Verizon Communications, Inc. v. Law Offices of Curtis V. Trinko, LLP,
540 U.S. 398 (2004) ........................................8, 91, 92, 99
Webco Industries, Inc. v. Thermatool Corp.,
278 F.3d 1120 (10th Cir. 2002) ..............................59, 117
Wessel v. City of Albuquerque,
463 F.3d 1138 (10th Cir. 2006) .......................125, 132
STATUTES
15 U.S.C. § 15b ...................................11, 14, 61, 133
15 U.S.C. § 16(i) .........................................61, 133
28 U.S.C. § 1292(b) .................................................15
RULES
FED. R. CIV. P. 50(a)............................................58
FED. R. CIV. P. 50(b) ................................................58
OTHER AUTHORITIES
2 PHILLIP E. AREEDA & HERBERT HOVENKAMP, ANTITRUST LAW (3d ed. 2011) .......................123
2A PHILLIP E. AREEDA & HERBERT HOVENKAMP, ANTITRUST LAW (3d ed. 2011) ....................122
3 PHILLIP E. AREEDA & HERBERT HOVENKAMP, ANTITRUST LAW (3d ed. 2011) .........83, 110, 112
9B CHARLES ALAN WRIGHT & ARTHUR R. MILLER, FEDERAL PRACTICE &
PROCEDURE (3d ed. 2011) .............................................58
18B CHARLES ALAN WRIGHT & ARTHUR R. MILLER, FEDERAL PRACTICE &
PROCEDURE (3d ed. 2011)........................................123
ix
Microsoft Corporation ("Microsoft") respectfully submits this Memorandum in
Support of its Renewed Motion for Judgment as a Matter of Law on the claim asserted by
Novell, Inc. ("Novell") under Section 2 of the Sherman Act for unlawful monopolization of the
PC operating system market. Based on the evidence at trial, a reasonable jury would not have a
legally sufficient evidentiary basis to find in Novell's favor. The Court therefore should enter
judgment as a matter of law in favor of Microsoft.
To prevail at trial, Novell was required to prove (a) that Microsoft's decision to
withdraw support for the namespace extension APIs was anticompetitive conduct prohibited by
Section 2 of the Sherman Act; (b) that this conduct "delay[ed] . . . Novell's development of the
versions of WordPerfect and Quattro Pro that were optimized for Windows 95," Novell, Inc. v.
Microsoft Corp., 699 F. Supp. 2d 730, 743 (D. Md. 2010); and (c) that the delay caused by
Microsoft's conduct "also caused anticompetitive harm in the PC operating system market." Id.
at 748 (emphasis in original). As to the third element (harm to competition), Novell's theory at
trial was based on the alternative assertions (a) that PerfectOffice, WordPerfect and Quattro Pro
were so popular that, if they were available on other operating systems, competition in the PC
operating system market would increase, and (b) that the same three Novell products were cross-platform middleware that exposed a sufficient number of application programming interfaces
("APIs") to enable independent software vendors ("ISVs") to write applications that called those
APIs, rather than the APIs exposed by the Windows operating system, thereby also leading to
increased competition in the PC operating system market. As shown below, Novell's evidence
was entirely insufficient on all three elements.
PRELIMINARY STATEMENT
This is an unusual case in many ways. It took Novell ten years after the allegedly
wrongful act occurred to file its Complaint, meaning that the sole claim that survived for trial is
barred by the statute of limitations unless it is "based in whole or in part" on United States v.
Microsoft Corp. (the "Government Case"). As a result, Novell can prevail only if its claim is
based on the claims asserted by the U.S. Department of Justice in May 1998. In addition, Novell
(a) released Microsoft from all antitrust claims except "the Claims set forth in" its 2004
Complaint, and (b) sold, and thus has no standing to prosecute, "any and all claims or causes of
action" that were "associated directly or indirectly with" Novell's PC operating system called
DR DOS. Novell is thus unable to prevail unless its trial proof conforms tightly to the theory
and allegations set forth in the Complaint, and Novell has no standing to proceed if its claim is
associated even indirectly with DR DOS. These limitations gave Novell a very narrow channel
through which to navigate.
These are not by any means the only difficulties for Novell. Novell has cited no
case, and we are aware of none, where a private antitrust plaintiff has obtained a money
judgment based on a "cross-market" theory of anticompetitive harm such as the one that Novell
advances--that conduct in one market injured competition in an entirely different market.
Although the Court of Appeals for the Fourth Circuit permitted Novell to survive a motion to
dismiss based on such a theory, a trial is about proof, not a lawyer's ability to articulate a clever
hypothetical. At trial, there was no proof of harm to competition in the PC operating system
market--the essence of any private antitrust action--especially in view of Bob Frankenberg's
testimony (with which Novell's other fact witnesses implicitly agreed) that in the "but-for"
world, Microsoft's market share in that market would have been higher (not lower) than it was in
the actual world. The Novell theories about harm to competition never advanced beyond pure
-2-
theory--there was no data or real-world facts that made the theoretical into anything
approaching a plausible scenario.
Novell failed in every respect to provide a legally sufficient evidentiary basis for a
reasonable jury to find in its favor on any element of its claim. Rather, the evidence at trial
established, among other things, that:
1. The sole act about which Novell complains--Microsoft's decision in October 1994 to
withdraw support for the namespace extension APIs--did not harm competition in the PC
operating system market under any causation standard;
2. Microsoft's October 1994 decision was not the cause of Novell's delay in releasing its
products for Windows 95;
3. The decision to withdraw support for the namespace extension APIs was not
anticompetitive--especially in light of the overwhelming evidence from more than a half
dozen witnesses that it is common practice in the industry for a software developer to
make changes to a beta version of an upcoming software release and in view of the
license agreement between Novell and Microsoft.
There is no dispute that designing and developing any operating system--and
particularly Windows 95, which was significantly more complex and advanced than its
predecessors--involves significant tradeoffs, and that during the development process there are
inevitably questions of how and to what extent to include or exclude features in view of
deadlines and design goals. Against this backdrop, Novell's claim hinges on the notion that--despite the well-understood practice in the software industry that beta releases are subject to
change and the explicit language in the license agreement--Microsoft could not modify its own
operating system if doing so disadvantaged Novell (and even if Microsoft was unaware that
Novell might be harmed by such a change). This cannot form the basis of a viable antitrust
claim. Any rule that in effect prohibits a software developer from making changes to a beta
would not only be entirely unworkable, but would stifle innovation in the software industry.
Such a rule would also collide with black letter antitrust law that a company--even one with
-3-
monopoly power--"has no antitrust duty to deal with its competitors" and "certainly has no duty
to deal under terms and conditions that the rivals find commercially advantageous." Pacific Bell
Telephone Co. v. linkLine Communications, Inc., 555 U.S. 438, 449-50 (2009). Novell's claim
does not come close to fitting within the narrow exception to that rule, which requires a plaintiff
to prove that the conduct (a) amounted to the termination of a profitable, long-term relationship
and (b) was without any economic justification. Four Corners Nephrology Associates, P.C. v.
Mercy Medical Center of Durango, 582 F.3d 1216 (10th Cir. 2009). See also Christy Sports,
LLC v. Deer Valley Resort Co., 555 F.3d 1188 (10th Cir. 2009).
The fact that the jury was unable to come to a unanimous verdict is not a reason to
deny the present motion. For one thing, the Court expressed some uncertainty during trial as to
the proper legal standard for the necessary effect on competition in the PC operating system
market (i.e., "reasonably capable" or "substantially contributing"). If the present motion for
judgment as a matter of law is granted, the Court of Appeals for the Tenth Circuit may have
occasion to clarify this issue. For another, as the Court indicated during the trial (but always
outside the presence of the jury), Novell's case is very weak in several respects, and a lengthy
second trial need not be conducted if the Court finds that no reasonable jury would have a legally
sufficient evidentiary basis to find for Novell on the present trial record. A review of such a
decision by the Tenth Circuit is likely to be the most efficient and sensible way toward a final
resolution of the case.
SUMMARY OF THE ARGUMENT
Based on the trial evidence, a reasonable jury would not have a legally sufficient
evidentiary basis to find, and as a matter of law there is no basis to find, that Novell met any of
-4-
the required elements of its claim. The Court should therefore enter judgment as a matter of law
in favor of Microsoft for each of the following reasons.
Section I: Microsoft's conduct did not contribute significantly to the
maintenance of Microsoft's monopoly in the PC operating system market and, in fact, could not
have harmed competition in that market under any causation standard.
First, each of Novell's theories of harm to competition in the PC operating system
market was entirely unsupported by the evidence at trial. The evidence established that the
applications barrier to entry protecting Microsoft's monopoly was a function of the thousands of
applications developed for Windows. As a result, Novell's theory that a mere three applications
(PerfectOffice, WordPerfect and Quattro Pro) could possibly have had an impact on competition
in the relevant market ("Novell's franchise applications theory") makes no sense.
1 In addition,
PerfectOffice, WordPerfect and Quattro Pro ("Novell's Three Products") had market shares so
low that a reasonable jury would not have a legally sufficient evidentiary basis to conclude that
they were sufficiently popular to induce PC users to move to operating systems other than
Windows. Further, the premise of Novell's franchise applications theory is refuted by the fact
-5-
that in the late 1980s and early 1990s--when WordPerfect was very successful and also was
available on many non-Microsoft operating systems--WordPerfect's popularity did not lead to
success for those non-Microsoft operating systems or hinder the success of Windows.
Novell also failed to establish that WordPerfect, AppWare, OpenDoc and
PerfectFit (or some combination of them) were middleware that could have affected competition
in the PC operating system market (the "Middleware Theory"). For Novell's software to have
any potential impact on competition, they must have possessed three characteristics: the
software must have (a) been cross-platform (e.g., Noll, Nov. 15 Trial Tr. at 1925-26), (b) been
available on all or nearly all PCs (e.g., Noll, Nov. 15 Trial Tr. at 1923-26), and (c) exposed a
sufficient number of APIs to allow for the development of general-purpose personal productivity
applications that relied on those APIs as opposed to APIs exposed by Windows (e.g., Findings of
Fact 28, 68, 74).
2
Novell's software had none of these three characteristics. At trial, perhaps as a
means of tryng to confuse the jury, Novell elicited from some of its witnesses an entirely
different definition of middleware--any software that sits between the operating system and an
application and that exposes APIs (referred to hereinafter as "Middle Software"). (Harral, Oct.
20 Trial Tr. at 234; Gibb, Oct. 26 Trial Tr. at 782-83.) But as Alepin, Novell's technical expert,
testified, Middle Software cannot "constitute any sort of threat to Windows" and thus could not
affect competition in the relevant market--"[t]here's got to be more." (Alepin, Nov. 9 Trial Tr.
at 1461-62.) Neither Professor Noll nor any other witness provided any plausible explanation or
real-world data to show that Novell's products, had they been released in a more timely fashion,
-6-
could somehow have diminished Microsoft's 90% share of the market. Even taking Novell's
theory at face value, the trial evidence comes nowhere close to supporting it.
First, the relevant versions of PerfectOffice,3 WordPerfect and Quattro Pro were
not cross-platform--they were written solely for Windows. Novell never released any version of
PerfectOffice for any other operating system. Second, Novell's products were not available on
all or nearly all PCs--their market share was very low. Third, all witnesses agreed that general-purpose personal productivity applications had not been and could not be written to the APIs
exposed by Novell's products, and thus any application written to run on those products would
necessarily be written to the APIs exposed by Windows. (Noll, Nov. 15 Trial Tr. at 1922-23;
Alepin, Nov. 9 Trial Tr. at 1489-90; Alepin, Nov. 10 Trial Tr. at 1533-35, 1538-40.) In fact,
Alepin explained that no ISV would even bother to attempt to write a general-purpose personal
productivity application on top of WordPerfect because this "would not be the best use of [an
ISV's] time." (Alepin, Nov. 9 Trial Tr. at 1480.)
Microsoft's withdrawal of support for the namespace extension APIs did not
contribute at all, let alone significantly, to the maintenance of Microsoft's monopoly. As
Novell's CEO admitted, had Novell utilized the namespace extension APIs and timely released
its products for Windows 95, the position and market share of Windows in the PC operating
system market would have been enhanced, not reduced. The allegedly wrongful act did not harm
competition.
-7-
Section II: As a matter of law, Microsoft's October 3, 1994 decision to withdraw
support for the namespace extension APIs (which, for ease of reference, will often hereinafter be
referred to as the "October 3 Decision") was not an anticompetitive act. All agree that Microsoft
was not required to provide Novell with access to or any information about the namespace
extension APIs or to cooperate with Novell at all. Verizon Communications, Inc. v. Law Offices
of Curtis V. Trinko, LLP, 540 U.S. 398, 408 (2004); Aspen Skiing Co. v. Aspen Highlands Skiing
Corp., 472 U.S. 585 (1985). As a result, Novell's claim is dependent upon its ability to fit within
the narrow exception to the general rule laid out in Aspen Skiing.
The evidence at trial conclusively demonstrated that Novell's claim falls far
outside the Aspen Skiing exception under the controlling Tenth Circuit standard. First, a
reasonable jury would not have a legally sufficient evidentiary basis to find that Microsoft
"terminated a profitable business relationship" with Novell and did so "without any economic
justification." Four Corners, 582 F.3d at 1225. The evidence was overwhelming that Novell's
Three Products remained compatible with Windows 95 and that Novell could have released
versions of these products for Windows 95 on a timely basis. Former Novell developers testified
that they could have used the Windows 95 common file open dialog, but that Novell made the
conscious decision to continue to develop its own custom file open dialog for its own business
reasons. Moreover, the evidence also established that, rather than seek to block Novell,
Microsoft continued to assist Novell in its efforts to release its products for Windows 95. In
addition, Microsoft did not "deny" to Novell any technology that was made "available to all
other" ISVs. Id. (emphasis in original). Microsoft's October 3 Decision applied to all ISVs.
Beyond that, Microsoft's conduct was not anticompetitive because (a) it was
widely understood in the software industry, including at Novell, that software developers can and
-8-
do modify beta versions before commercial release of the product, and (b) the applicable contract
provided that the operating system was still under development and all features of Windows 95
were subject to change. (DX 18, Microsoft Corporation Non-Disclosure Agreement (Pre-release
Product) with WordPerfect Corporation, May 24, 1994 at 1, § 2.)4 As an "ordinary business
practice[] typical of those used in a competitive market," Microsoft's decision to modify the
features of its Windows 95 operating system prior to its final release is not anticompetitive
conduct. Telex Corp. v. International Business Machines Corp., 510 F.2d 894, 925-26, 928
(10th Cir. 1975). Moreover, because Novell was "aware that the relationship was temporary and
subject to [Microsoft's] business judgment," the October 3 Decision does not constitute
anticompetitive conduct as a matter of law. Christy Sports, 555 F.3d at 1196.
The evidence at trial also established that Microsoft had legitimate business
justifications for its October 3 Decision, which, as a matter of law, precludes a finding that it was
anticompetitive behavior. Four Corners, 582 F.3d at 1225. Support was withdrawn because the
namespace extension APIs (a) posed a risk to the stability and reliability of Windows 95;
(b) raised compatibility concerns for future versions of Microsoft's operating systems; and
(c) failed to achieve the level of integration that Bill Gates had hoped to achieve.
There is also no basis in fact or law for the assertion that the Section 2 antitrust
claim here at issue can be premised on a "deception" theory. A competitor has no claim under
the antitrust laws for deception and, in any event, the evidence at trial established that Novell
was not and could not have been "deceived" because, among other reasons, it was well aware
-9-
that the namespace extension APIs were subject to change based on the explicit language in the
beta license agreement and common industry practice.
Section III: The evidence at trial overwhelmingly demonstrated that the
October 3 Decision was not the cause of delay in the release of PerfectOffice, WordPerfect or
Quattro Pro for Windows 95.5 Even as of January 1996, Quattro Pro--"an essential element" of
PerfectOffice without which Novell could not release the suite (Frankenberg, Nov. 7 Trial Tr. at
1143)--was "[n]ot by any stretch of the imagination" ready to go. (Larsen, Nov. 30 Trial Tr. at
3624; accord LeFevre, Dec. 2 Trial Tr. at 4062-63; Bushman, Nov. 28 Trial Tr. at 3192-93.)
Further, it is undisputed that Novell could easily have released the relevant products on time if it
had used the Windows 95 common file open dialog. Microsoft is not liable for Novell's decision
to pick a far more difficult route.
Section IV: Novell's damages theories were all dependent on the assumption that
but for Microsoft's withdrawal of support for the namespace extension APIs, Novell would have
been able to release PerfectOffice for Windows 95 within 30 to 60 days of the August 24, 1995
release of Windows 95. This assumption was shown conculsively at trial to be false. As a result,
Novell is not entitled to recover any damages as a matter of law.
In addition to these reasons, there are four additional independent legal issues that
also compel the granting of Microsoft's motion. First, Novell has no standing to assert its claim
because, even if Novell's Three Products had been harmed by the October 3 Decision, Novell
-10-
suffered no cognizable antitrust injury that had an "adverse effect on competition or consumers"
in the PC operating system market. Elliott Industries Ltd. v. BP America Production Co., 407
F.3d 1091, 1125 (10th Cir. 2005). Second, the evidence at trial established that Novell did not
need the namespace extension APIs for WordPerfect and Quattro Pro (the only two products at
issue in the Complaint), but rather for five other products (including Novell's QuickFinder
search engine and Soft Solutions document management system). As a result, any purported
claim based on harm to these products or to PerfectOffice was released in Novell's
November 2004 settlement with Microsoft. Third, the claim Novell submitted to the jury was
dependent on assertions about the installed base of WordPerfect on the DOS platform, and is
thus "associated directly or indirectly with" the claim that Novell sold Caldera in 1996. Fourth,
by disavowing a highly important element of the theory of the Government Case and proceeding
on a different theory at trial, Novell's claim at trial bears no "real relation" to the matters
"complained of in the government suit," Leh v. General Petroleum Corp., 382 U.S. 54, 59
(1965), and therefore is barred by 15 U.S.C. § 15b.
STATEMENT OF FACTS
A. The Complaint
On November 12, 2004, Novell filed its Complaint in this action, asserting six
counts under Sections 1 and 2 of the Sherman Act, all based on harm allegedly suffered by
Novell's WordPerfect word processing application and Quattro Pro spreadsheet application in
the period in which Novell owned them (June 24, 1994 to March 1, 1996).
On November 8, 2004, four days before it filed the Complaint, Novell and
Microsoft entered into a settlement agreement which released Microsoft from "any and all
Claims that Novell ever had or has as of the date of this Agreement in law or in equity, known or
-11-
unknown, of any kind whatsoever (including without limitation any antitrust or similar Claims of
any kind), except for . . . (iii) the Claims set forth in the draft WordPerfect complaint . . . ."
(Nov. 8, 2004 Settlement Agreement, at ¶2(a), Holley Decl. Ex. A.) This exception referred to a
draft complaint that was identical to the Complaint actually filed on November 12.6
Count I of the Complaint alleged that Microsoft engaged in anticompetitive
conduct directed at WordPerfect and Quattro Pro7 in order to maintain its monopoly in the PC
operating system market.8 (Compl. ¶¶ 151-55.) Novell alleged that Microsoft's withdrawal of
support for the namespace extension APIs "forced Novell to expend an entire year" developing
an alternative technology and thus prevented Novell from releasing WordPerfect and Quattro Pro
for Windows 95 on a timely basis. (Compl. ¶¶ 72-78.)
According to the Complaint, Microsoft "evangelized the benefits of using the
browsing extensions" before the release of Windows 95 but then "ripped out these programming
interfaces without warning to Novell" only "a few months" before the final release of
Windows 95. (Compl. ¶¶ 73, 75.) The Complaint also alleged that Microsoft's acts "degraded
the functionality of Novell's applications"9 in that "Novell was suddenly unable to provide basic
-12-
file management functions in WordPerfect; in many instances, a user literally could not open a
document he previously created and saved."10 (Compl. ¶¶ 75, 78.) The Complaint alleged that
versions of WordPerfect written to run on Windows 3.0 and Windows 3.1 were incompatible
with Windows 95,11 and that "[a]s a consequence," it was imperative to release new versions of
WordPerfect and Quattro Pro around the same time as Microsoft released Windows 95 because
when consumers purchased Windows 95, they would "almost simultaneously switch to
applications" that were compatible with the new operating system. (Compl. ¶ 70.)
The Complaint set forth two theories as to how the alleged harm to WordPerfect
and Quattro Pro contributed to Microsoft's maintenance of its monopoly in the PC operating
system market. First, Novell alleged that WordPerfect was so popular that its availability on
other operating systems would popularize those non-Microsoft operating systems and thus
provide a bridge across the applications barrier to entry that protected Microsoft's PC operating
system monopoly. (Compl. ¶ 52.) Second, Novell alleged that its AppWare and OpenDoc
technologies, when integrated with WordPerfect (not PerfectOffice), were a "middleware" threat
here, if the conduct at issue, the decision to withdraw support for the namespace extension APIs,
did not cause any delay"); see also Noll, Nov. 14 Trial Tr. at 1838-40.)
-13-
because they exposed APIs that would enable ISVs to write applications that called those APIs
rather than the APIs exposed by Windows. (Compl. ¶¶ 46-51.) Novell alleged that the
"portfolio of OpenDoc, AppWare, and WordPerfect software posed a competitive threat to
Microsoft's operating systems monopoly similar to that described in the Government Suit."
(Compl. ¶ 51.)12
Counts II through V of Novell's Complaint alleged that Microsoft unlawfully
monopolized or attempted to monopolize markets for word processor software and spreadsheet
software--the markets in which WordPerfect and Quattro Pro actually competed.
(Compl. ¶¶ 156-73.) Count VI alleged that Microsoft entered into agreements with original
equipment manufacturers ("OEMs") and others to exclude Novell's word processing and
spreadsheet applications from important software distribution channels, in violation of Section 1
of the Sherman Act. (Compl. ¶ 175; see also id. ¶ 24.)
B. Pre-Trial Proceedings
On January 7, 2005, Microsoft moved to dismiss all six counts of Novell's
Complaint. On June 10, 2005, the Court dismissed Counts II through V as barred by the
applicable four-year statute of limitations under 15 U.S.C. § 15b, holding that the Government
Case did not operate to toll the limitation period. Novell, Inc. v. Microsoft Corp., 2005 U.S. Dist.
-14-
LEXIS 11520, at *9-14 (D. Md. June 10, 2005). The Fourth Circuit affirmed dismissal of
Counts II through V in 2007. Novell, Inc. v. Microsoft Corp., 505 F.3d 302, 320-23 (4th Cir.
2007).
With respect to Counts I and VI, Microsoft argued in its motion to dismiss that
Novell (a) lacked standing because the Complaint failed to adequately allege antitrust injury, and
(b) had sold those claims to Caldera in 1996. The Court denied the motion as to Counts I and VI,
2005 WL 1398643, at *1-3, but certified its rulings pursuant to 28 U.S.C. § 1292(b). The Fourth
Circuit granted Microsoft's petition for leave to appeal only with respect to the first issue
(antitrust standing), and subsequently affirmed the denial of Microsoft's motion to dismiss.
505 F.3d at 307, 319.
In 2009, after the completion of discovery, Microsoft moved for summary
judgment on the merits of Counts I and VI and Novell moved for summary judgment seeking to
eliminate certain affirmative defenses advanced by Microsoft. In response to Novell's motion,
Microsoft filed a cross-motion for summary judgment asserting that Novell lacked standing to
bring the claims asserted in Counts I and VI because it sold to Caldera all claims "associated
directly or indirectly with" Novell's DR DOS operating system in 1996, and further that the
doctrine of res judicata barred Novell from asserting those claims.
On March 30, 2010, the Court dismissed Counts I and VI on the ground that
Novell had sold those claims to Caldera. 699 F. Supp. 2d at 739. The Court also held that "had
Novell not assigned them to Caldera, Count I would have survived Microsoft's Motion for
Summary Judgment and Count VI would not have." Id. at 740. On appeal, the Fourth Circuit
reversed with respect to Count I, holding that Novell's sale to Caldera did not include the claim
asserted in Count I. Novell, Inc. v. Microsoft Corp., 429 F. App'x 254, 260-62 (4th Cir. 2011).
-15-
As to Count VI, the Fourth Circuit noted that "Novell's opening brief did not argue that the
district court erred by granting summary judgment as to Count VI, and Novell confirmed in oral
argument that it was not pursuing" Count VI on appeal. Id. at 258 n.7.
In its March 30, 2010 decision, this Court held that in order to succeed on
Count I, Novell (a) "must prove not only that the defendant's conduct was anticompetitive, but
also that it caused anticompetitive harm in the relevant market," 699 F. Supp. 2d at 747-48, and
(b) "must [also] prove that the specific Microsoft conduct which caused injury to Novell's
applications also caused anticompetitive harm in the PC operating system market." Id. at 748
(emphasis in original). The Court further held that Novell must prove "that the conduct that
harmed its software applications contributed significantly to Microsoft's monopoly in the PC
operating system market." Id. at 750. The Court recognized that "a monopolist generally has a
right to refuse to cooperate with a competitor," but ruled that whether the conduct at issue fell
within the Aspen Skiing exception was an issue for trial. Id. at 745-47.13
Prior to trial, Novell sought collateral estoppel on certain Findings of Fact made
by Judge Jackson and Legal Rulings made by the D.C. Circuit in the Government Case and by
the Fourth Circuit in this action. In October 2011, the Court gave collateral estoppel effect to 52
-16-
Findings (Court's Oct. 4 Letter, Dkt. #163; see also D. Md. Dkt. #76), the majority of which
concerned background information (Findings of Fact 2, 4, 6-10, 17), characteristics of the market
(Findings of Fact 18, 20, 33-35, 59-60), and a description of the "applications barrier to entry"
(Findings of Fact 28-32, 36-39, 68-70, 73-74).
At trial, on October 18 and November 14, counsel for Novell read to the jury the
collaterally estopped Findings of Fact (Oct. 18 Trial Tr. at 143-57; Nov. 14 Trial Tr. at 1675-97),
which the Court instructed had "binding" effect (Oct. 18 Trial Tr. at 143). Included among them
were several Findings that have direct bearing on Novell's theories about harm to competition.
Among these was "[t]he fact that a vastly larger number of applications are written for Windows
than for other PC operating systems [which] attracts consumers to Windows, because it reassures
them that their interests will be met as long as they use Microsoft's product" (Finding of Fact 37)
and that, as a result, "[t]he large body of applications thus reinforces demand for Windows,
augmenting Microsoft's domina[nt] position and thereby perpetuating ISV incentives to write
applications principally for Windows" (Finding of Fact 39). The Findings also included
statements that (a) "no middleware product exposes enough APIs [as of November 1999, when
the Findings were issued] to allow independent software vendors, ISVs, profitably to write full-featured personal productivity applications that rely solely on those APIs" (Finding of Fact 28),
and (b) "it remains to be seen [also as of 1999] whether server- or middleware-based
development will flourish at all. Even if such development were already flourishing, it would
still be several years before the applications barrier eroded enough to clear the way for the
relatively rapid emergence of a viable alternative to . . . incumbent Intel-compatible PC operating
systems" (Finding of Fact 32).
-17-
C. Trial
Trial commenced on October 17 and concluded on December 16. Novell called
four fact witnesses to the stand and introduced the deposition testimony of eleven additional
witnesses. Novell also called three experts: Mr. Ronald Alepin on technical issues, Professor
Roger Noll on competition issues, and Dr. Frederick Warren-Boulton on damages.
Microsoft called eleven fact witnesses to the stand and introduced the deposition
testimony of two others. Microsoft also called three experts: Professor John Bennett on
technical issues, Professor Kevin Murphy on competition issues, and Professor Glenn Hubbard
on damages. A total of 642 exhibits were admitted into evidence.
The evidence at trial established as follows:
1. The Development of Windows 95 and the M6 Beta
In 1991, Microsoft began developing a 32-bit PC operating system, codenamed
"Chicago," that was eventually released as Windows 95. (Gates, Nov. 21 Trial Tr. at 2754.)
Bill Gates testified that the development of Windows 95 was "one of the toughest engineering
tasks ever done" and a "bigger challenge" than developing prior 16-bit operating systems, in part
because of the "engineering complexity" of making Windows 95 compatible with the many
software and hardware products that had entered the market after the 1992 release of Windows
3.1. (Id. at 2759-60.)14
In 1993, Microsoft began providing ISVs with information about user interface
features that Microsoft was planning to include in Chicago. In July 1993, at a "Design Preview,"
-18-
Microsoft provided WordPerfect Corporation15 and other ISVs with preliminary information
about Chicago's new user interface, including common dialogs and various common controls.
(PX 63, Trip Report: Chicago User Interface Design Preview, July 8-9, 1993.) At a December
1993 conference, Joe Belfiore, the lead program manager for the Windows 95 shell (Belfiore,
Dec. 5 Trial Tr. at 4227), gave a presentation to ISVs describing plans for the user interface of
Chicago. (PX 113, New Windows "Chicago" UI: What It Means for Your Application, Dec.
1993.) Belfiore's presentation described various "Shell Extensibility" mechanisms (PX 113, at
NOV 00734378, NOV 00734389), only a small subset of which depended on the namespace
extension APIs (see Belfiore, Dec. 5 Trial Tr. at 4249-50). The presentation noted that the
extensible shell in Chicago would allow ISVs to take advantage of features such as "Drag-and-Drop," "Property Sheet Extensibility" and "Explorer UI Integration." (PX 113, at NOV
00734389.) With regard to the last item, the presentation described the ability to add a "custom
container" to the Windows Explorer (id. at NOV 00734389-90), which was a new general-purpose viewer included in Windows 95.
Two former software developers from Novell's shared code group,16 Adam Harral
and Greg Richardson, testified at trial that they received written slides from this presentation--although Harral did not attend the presentation and Richardson said he did not remember
-19-
attending (Harral, Oct. 24 Trial Tr. at 415; Richardson, Oct. 25 Trial Tr. at 606-07)--which
described, in part, the functionality later provided by the namespace extension APIs. (Harral,
Oct. 20 Trial Tr. at 296-99; Richardson, Oct. 25 Trial Tr. at 590-91.) Notably, Belfiore's written
presentation warned that the namespace extensions were (a) "[n]ot for most applications!,"
(b) should be used "[o]nly . . . if your application displays a pseudo folder: electronic mail,
document management, etc," and (c) "should NOT [be used to] edit documents with an explorer
extension!" (PX 113, at NOV 00734390.)
Consistent with this warning, Harral testified that Novell did not need the
namespace extension APIs in order to create a version of WordPerfect for Windows 95, stating:
"I don't know anything that WordPerfect [the] word processor needed to do for a Namespace
extension. They did have shell extensions, but I don't recall a NameSpace extension that they
needed to do." (Harral, Oct. 20 Trial Tr. at 327.) Rather, Harral and Richardson explained that
the shared code group intended to use the namespace extension APIs in order to augment
Windows 95 by embedding Novell's QuickFinder search engine, Soft Solutions document
management system, e-mail client, Presentations clip-art gallery and FTP/HTTP browser directly
in the Windows 95 shell.17 (Harral, Oct. 20 Trial Tr. at 268-70; Harral, Oct. 24 Trial Tr. at
372-74; Richardson, Oct. 25 Trial Tr. at 629-30, 638, 690-92.)
-20-
As Novell was warned in 1993 (PX 113), the namespace extension APIs were not
designed to be used by word processing or spreadsheet applications. Satoshi Nakajima, the
inventor of this technology, testified that it never made sense for technical reasons for a word
processor or spreadsheet to use those APIs. (Nakajima, Dec. 1 Trial Tr. at 3864-65.) Indeed,
Nakajima's patent plainly stated that a "NameSpace extension should not be used . . . to expose
the contents of a spreadsheet or word processing document in the shell." (PX 364, United States
Patent No. 5,831,606, Nov. 3, 1998, issued to Nakajima et al., at 54-55.) Gates also testified that
the namespace extension APIs were designed to be called by "an e-mail client or some type of
system utility, not [] a word processor [or] spreadsheet." (Gates, Nov. 21 Trial Tr. at 2796-97.)
In the spring of 1994, Microsoft sent to ISVs a "'Chicago' Reviewer's Guide"
(Alepin, Nov. 10 Trial Tr. at 1624-25), which provided additional information about features
then planned for Windows 95 (PX 388, Microsoft Windows "Chicago" Reviewer's Guide). The
cover page of the Reviewer's Guide expressly cautioned ISVs that it did "not represent a
commitment on the part of Microsoft for providing or shipping the features and functionality in
the final retail product offerings of Chicago." (Id. at MSC 00762731.)
On June 10, 1994, Microsoft provided WordPerfect/Novell with the Milestone 6
("M6") beta version of Windows 95--the first beta version of the new operating system provided
to ISVs in general and to WordPerfect/Novell in particular.18 (Harral, Oct. 24 Trial Tr. at 434-35.) The license agreements with WordPerfect and Novell provided that the beta "may be
substantially modified prior to first commercial shipment," and that WordPerfect and Novell
"assume[] the entire risk with respect to the use of the" beta. (DX 18, at 1 § 2; DX 19, Microsoft
-21-
Corporation Non-Disclosure Agreement (Pre-release Product) with Novell, Inc., executed Dec.
10, 1993, at 1 § 2.)19
Novell's CEO, developers and other former employees testified that they
understood that Microsoft was entitled to modify the beta version of Windows 95. At trial, Bob
Frankenberg, Novell's CEO from 1994 to 1996, testified that Novell then understood that beta
versions of Windows 95 both "could change" and "might change" prior to commercial release.
(Frankenberg, Nov. 8 Trial Tr. at 1201, 1209.) Frankenberg testified that it "was widely
understood in the software industry" that beta versions of software products may change, that
such software products may never be released at all, and that the entire risk arising from use of a
beta version is borne by the beta tester. (Id. at 1204-05.)
Likewise, Nolan Larsen, Novell's Director of Human Factors and later a Quattro
Pro developer (Larsen, Nov. 30 Trial Tr. at 3567), testified that "the definition of a beta" is that
"there can be and almost certainly will be changes" (id. at 3603; see also id. at 3654-58). Dave
LeFevre, the Director of Marketing for PerfectOffice for Windows 95 at Novell from 1994 to
1996 (LeFevre, Dec. 2 Trial Tr. at 4017-18), agreed that a "[b]eta by definition is an early release
or a prerelease of a product that is subject to change" and that "there is no promise that what is in
an early beta or even a late beta will be in the final product (Id. at 4031). Novell's expert
witnesses also conceded this same point. (Noll, Nov. 15 Trial Tr. at 1878 ("all beta versions of
-22-
all software are provisional, and they are not guarantees of what the program will contain upon
final release"); Alepin, Nov. 10 Trial Tr. at 1555-56 ("[t]he expectation" with a beta release "is
that the software is being worked on").) Gates, Belfiore and Brad Struss had the same
understanding--that it was common in the software industry for beta versions to change. (Gates,
Nov. 22 Trial Tr. at 3124-25; Belfiore, Dec. 5 Trial Tr. at 4238-39; Struss, Nov. 28 Trial Tr. at
3257.) Indeed, the only witness who hedged this unqualified testimony about industry practice
was Harral, who testified that "the Beta is to hammer out the problems; not, at that point, to do
new features or change features." (Harral, Oct. 20 Trial Tr. at 303.)20 Harral's testimony ignores
the obvious--that "fixing" a "problem" could require change to a feature.
Novell's "Software Developer's Kit" (distributed to ISVs with beta versions of
Novell's software) also warned ISVs that "Novell does not guarantee that Beta Products will
become generally available to the public or that associated products will be released" and that
"[t]he entire risk arising out of your use of Beta Product remains with you." (DX 618, Novell
Software Developer's Kit License, copyrighted 1994 and 1995, at NOV-B07520262.)
Frankenberg acknowledged that these provisions were "certainly pretty much similar" to those in
Microsoft's license agreement. (Frankenberg, Nov. 8 Trial Tr. at 1208-09.) And LeFevre
testified that when he "ran the beta program" for WordPerfect 5.1 for Windows, WordPerfect
"cut a number of features in WordPerfect 5.1" during the beta testing process. (LeFevre, Dec. 2
Trial Tr. at 4033-34.)
Finally, a formal memorandum from the Novell Corporate Development Group
dated October 18, 1994--just fifteen days after the October 3 Decision--set forth without
-23-
equivocation Novell's understanding that during the beta phase "the product and features may
still change dramatically." (DX 612A, Memo to Files from Steven W. Bentley, Novell
Corporate Development Group, Oct. 18, 1994, at 4.) The Novell Corporate Development Group
memorandum also stated that these changes "may include removal of [an] entire feature" from a
beta. (Id. at 2.) This October 1994 memorandum completely refutes Harral's testimony that
changes to a beta should not include "removal" of a "feature" cannot be reconciled with the
Novell memorandum.
2. Microsoft's Decision to Withdraw Support for the Namespace
Extension APIs
On October 3, 1994, Bill Gates decided to withdraw support for the namespace
extension APIs. (Gates, Nov. 21 Trial Tr. at 2792.) Gates testified that "serious problems with
robustness and compatibility . . . led to my decision" (Gates, Nov. 22 Trial Tr. at 3039), as well
as concerns that the namespace extension APIs did not provide the level of functionality he had
anticipated (Gates, Nov. 21 Trial Tr. at 2800-04).
The evidence at trial established that the namespace extension APIs presented
significant robustness concerns. Nakajima testified that due to memory constraints with
Windows 95, he designed the namespace extension APIs to allow a third-party application to run
in the "same process" as the Windows 95 shell (the Windows 95 user interface). (Nakajima,
Dec. 1 Trial Tr. at 3758-61, 3834; see also Belfiore, Dec. 5 Trial Tr. at 4270-71.) Although this
approach was "efficient," Nakajima testified that "there is a risk that goes with" allowing third-party applications to run in the same process as the Windows 95 shell because if the third-party
-24-
application "crashes with a bug or virus or whatever the reason, then the whole system will
crash." (Nakajima, Dec. 1 Trial Tr. at 3761.)21
There is no dispute that poorly behaved third-party applications calling the
namespace extension APIs could crash the Windows 95 shell. Gates testified that "when the
[Windows] Explorer would call these other applications [using the namespace extension
APIs] . . . , if there was a problem in that piece of software . . . if it crashed or anything, it would
crash the whole system because you were running in the shell." (Gates, Nov. 21 Trial Tr. at
2781-82.) Paul Maritz, the Microsoft executive in charge of all operating systems, testified that a
program written by an ISV that called the namespace extension APIs "could bring down the
shell." (Jan. 9, 2009 Maritz Deposition at 129, Dkt. #279, used at trial on Oct. 27.) James
Allchin, the Microsoft executive in charge of Windows NT, testified that because of the way the
namespace extension APIs were designed, "if an application had an error in it, it could take down
or corrupt the user experience overall." (Jan. 8, 2009 Allchin Deposition, Nov. 8 Trial Tr. at
1297.) Bob Muglia, the group program manager for Windows NT, also testified that "if an
application misbehaved, if it crashed or hung, it could bring the whole operating system down."
(Muglia, Nov. 29 Trial Tr. at 3386-87; see also id. at 3395-97 ("[W]hen an ISV extended the
shell, it did so in the same process that the shell actually ran in. What that means is that if that
-25-
application crashes, it can crash the shell and it can bring the system down from the end user
perspective.").)
The namespace extension APIs were "more risky" than other shell extension
mechanisms because, as Nakajima testified, there was "no limit" on the number of namespace
extensions that could be running on Windows 95 at any one time and "no limit" on the size or
complexity of applications that could run in the Windows Explorer process space. (Nakajima,
Dec. 1 Trial Tr. at 3761-63, 3766.) Thus, it was possible for an unlimited number of ISVs each
to add an unlimited number of custom containers that would then appear in the Windows
Explorer tree view and in the Windows 95 common file open dialog. (Id. at 3764-65.) Alepin
acknowledged that Microsoft had no ability to "impose quality control standards on third-party
developers . . . whose products called the namespace extension APIs." (Alepin, Nov. 10 Trial
Tr. at 1593-94; see also Richardson, Oct. 25 Trial Tr. at 659.) Microsoft's inability to control the
number and complexity of the namespace extensions that could run on Windows 95 at one time
further aggravated the robustness problem. (Nakajima, Dec. 1 Trial Tr. at 3766.)
Novell's witnesses did not dispute that an error in an application calling the
namespace extension APIs could crash the Windows 95 shell and bring down the operating
system. Richardson and Alepin testified that the namespace extension APIs could have crashed
the Windows 95 shell "because at that time namespace extensions were running in the same
process as Windows explorer and the rest of the shell" (Richardson, Oct. 25 Trial Tr. at 756-57),
and thus an error in an application calling the namespace extension APIs "had the potential to
make the system unresponsive." (Alepin, Nov. 10 Trial Tr. at 1589.) Indeed, Professor Noll
opined that "one valid reason for not documenting an API" is "where those APIs are unstable."
(Noll, Nov. 15 Trial Tr. at 1872-73.)
-26-
The evidence also showed that Microsoft was concerned about compatibility with
other versions of Windows as well. Gates, Belfiore, Nakajima and Muglia all testified that the
October 3 Decision was the culmination of an internal debate at Microsoft between teams
developing Windows 95, Windows NT, and a future version of Windows code-named Cairo.
(Gates, Nov. 21 Trial Tr. at 2792-93; Belfiore, Dec. 5 Trial Tr. at 4269-71, 4278-80; Nakajima,
Dec. 1 Trial Tr. at 3763-64, 3768-71; Muglia, Nov. 29 Trial Tr. at 3385-90, 3397-3400.) In early
1994, these teams were working on "three different shell efforts" and each had differing design
specifications for different types of customers. (Gates, Nov. 21 Trial Tr. at 2784.) Belfiore
explained that "the Cairo team, because it was focused on higher end PCs and was more oriented
around reliability and work station like behavior," believed "that the architecture of having a
third-party application that can bring down the entire shell was unacceptable for their goals of
reliability." (Belfiore, Dec. 5 Trial Tr. at 4269-71, 4279.) Likewise, the NT team, which was
working on "a more powerful version" of Windows "that shipped particularly to business
customers or scientific workstation type customers" (Gates, Nov. 21 Trial Tr. at 2780), was
"very keen that there never be crashes, because their system was being used for things like stock
exchange trading, or other things where you just wouldn't want the software to stop working at
any time" (Id. at 2781; see also Alepin, Nov. 10 Trial Tr. at 1608 ("it was the goal to have
Windows NT be more reliable than Windows 95")).
In early to mid-1994, the NT and Cairo teams were concerned that they would
have to support the namespace extension APIs, which they viewed to be unreliable, in their shell
development efforts and future versions of Windows. As Muglia testified, once the Chicago
team shipped its product, the Cairo team would be required "to support what they [Chicago] did"
and would not be "able to go forward with the Cairo shell as [they] had planned." (Muglia, Nov.
-27-
29 Trial Tr. at 3399.) Belfiore, a member of the Chicago team, agreed that a "critically
important" issue in the debate over the namespace extension APIs was the "notion of
compatibility," that is, ensuring that end-users who bought an application that ran on
Windows 95 would also be able to run the same application on Cairo and other future Microsoft
operating systems. (Belfiore, Dec. 5 Trial Tr. at 4278-79; see also id. at 4272-73 ("the Cairo
team cared what--what our software did because the way we built it, in order to have
compatibility, they would have to accept and run as well").) As Alepin explained, one reason for
not documenting an API is because "you do not wish customers to attach to them, lest you be
obligated to support them in the future." (Alepin, Nov. 10 Trial Tr. at 1560-61.)
Gates testified that prior to October 3, 1994, "NT was saying" that Microsoft
should "either redesign these things to be right or don't do them," and "Cairo was just saying
don't do them. They take things in the wrong direction and it is just going to be a mess." (Gates,
Nov. 21 Trial Tr. at 2792-93.) As a result, Nakajima was asked "to defend" the APIs in a
meeting in the "board room" with "Bill Gates and senior executives." (Nakajima, Dec. 1 Trial
Tr. at 3771.) Prior to this meeting, the Cairo team sent Nakajima "100 pages of document[s] to
prove . . . why Cairo's approach was better" and after reviewing these materials, Nakajima chose
not to defend his technology in the boardroom meeting, concluding that the Cairo objections
"were right." (Id. at 3772-73.)
In September 1994, in the midst of this debate, Gates transferred the members of
the Cairo shell team to what was called the Ren group, a team within Microsoft Office, because
the Cairo team had been unsuccessful in producing a prototype of its object-oriented shell and
Gates wanted to see if the Ren group "could do a better job making progress on it." (Gates, Nov.
21 Trial Tr. at 2782.) Around the same time, the Windows NT shell effort was terminated when
-28-
Gates decided "to use a common s[h]ell across Windows 95 and Windows NT" (id. at 2784-85;
see also Belfiore, Dec. 5 Trial Tr. at 4359), but before the design of a common shell could be
finalized, Microsoft "needed to solve the problem" of robustness (Gates, Nov. 21 Trial Tr. at
2785). In addressing this issue, Gates and others "looked hard at the objections" from the NT
and Cairo groups. (Id.)
Thus, on October 3, 1994, Gates decided to withdraw support for the namespace
extension APIs (Gates, Nov. 22 Trial Tr. at 3057) and distributed an e-mail to others at Microsoft
announcing his decision (PX 1). Contemporaneous evidence confirmed that the reasons for the
October 3 Decision were robustness and compatibility. On October 4, 1994, one day later,
Muglia wrote that the decision was "very good news for BSD [Business Systems Division]"
because "these interfaces introduce significant robustness issues," and "[s]ince Bill has decided
these interfaces won't be published, NT development does not have to expend precious energy
on implementing these for NT." (DX 21, E-mail from Robert Muglia, Oct. 4, 1994.)
On October 12, 1994, Scott Henson of the Microsoft Developer Relations Group
("DRG") (Struss, Nov. 28 Trial Tr. at 3266) circulated a set of proposed answers to questions
that ISVs might ask about the October 3 Decision (DX 3). The answers stated that because the
namespace extension APIs were "design[ed] to [be] part of the system," these APIs would "run
in the explorer's process space" and, as a result, "[b]adly written name space extension[s] could
cause the reliability of Windows 95 to be less th[a]n what it should." (Id. at MX 6055843.)
DX 3 also explained that Microsoft had "determined that it will be very difficult to support these
API's for applications as we move forward with our operating systems," and that Microsoft "did
not want to encourage ISV's to support interfaces that would go away in the future." (Id.)
-29-
An additional reason for the October 3 Decision was that the namespace extension
APIs did not provide the level of integration Gates had anticipated. (Gates, Nov. 21 Trial Tr. at
2786-87, 2800-04.) Gates explained that his statement in PX 1 that "[w]e should wait until we
have a way to do a high level of integration that will be harder for [the] likes of Notes,
WordPerfect to achieve, and which will give Office a real advantage" was a reference to his
desire to wait until the Cairo vision had been achieved. (Id. at 2803-04.)22 As Gates told the
jury, "the namespace extension APIs [we]re not rich enough to give you the ability to do this
kind of information browser shell" that he had envisioned. (Id. at 2803.) As a result, Gates
explained that it was not worth causing "problems for the NT and Cairo teams" by supporting
this technology. (Id. at 2804.)
3. The October 3 Decision Was Not Made to Harm Novell
There was no evidence that the decision to withdraw support for the namespace
extension APIs was made in order to harm Novell's development efforts, or that it even had
anything to do with Novell. To the contrary, Gates testified that at the time he made the
decision, he knew nothing "at all about the specifics of whether they [Novell] were using them
[the namespace extension APIs] or not." (Gates, Nov. 21 Trial Tr. at 2811; see also id. at 2828.)
-30-
(a) Before October 3, Novell Said that It Was Not Using the
Namespace Extension APIs
The contemporaneous documentary evidence is in full accord with Gates'
testimony. On September 22, 1994, Brad Struss, who led the Windows 95 team for DRG and
worked closely with Novell (Struss, Nov. 28 Trial Tr. at 3246-47, 3252-53), reported on the
results of a survey conducted to determine the extent to which ISVs were using the namespace
extension APIs (DX 17, E-mail from Brad Struss, at MX 6109491). Struss testified that DRG
"did the initial overall survey of shell extensibility overall, which included the namespace
extension APIs," and that DRG "reached out" to Novell "specifically to talk to them about those
APIs to understand what they were using and what the implications of changing those [APIs]
may be." (Id. at 3268.)23 In DX 17, Struss reported that as of September 22, 1994, Novell
"ha[d] not begun any work on IShellFolder, IShellView, etc." (i.e., the namespace extension
APIs). (DX 17, at MX 6109491.) In the same document, Struss put in quotation marks Novell's
response upon hearing that Microsoft might withdraw support for this technology: "we'll figure
it out if it's not documented." (Id; see Struss, Nov. 28 Trial Tr. at 3270.)
(b) After October 3, Novell Said It Was "OK" About the Decision to
Withdraw Support
The evidence also showed without ambiguity or contradiction that, after
October 3, Novell said that it was "OK" with the decision to withdraw support for the namespace
extension APIs. On October 12, 1994, Struss reported in an e-mail that "we're now in the
process of proactively notifying ISVs about the namespace api changes (will not document them
and they'll go away/change)" and "[s]o far Stac, Lotus, WP, Oracle, SCC appear to be OK with
-31-
this." (DX 3, at MX 6055840.)24 Not a single witness for either side ever testified that Novell
complained that the October 3 Decision had harmed Novell's development efforts, and there is
no document to that effect either.25
It is also undisputed that even after the October 3 Decision, Microsoft continued
to offer assistance to Novell so that it could build applications for Windows 95. Struss testified
that he and others in DRG worked "extra hard" to assist Novell because "WordPerfect was a
major software application" and Microsoft believed it was "critical" to have WordPerfect's
support for Windows 95. (Struss, Nov. 28 Trial Tr. at 3253-54.) On October 21, 1994, Struss
reported in an e-mail that although WordPerfect was "focus[ing] on 16-bit product revision this
fall," he was "[w]orking with their [Novell's] sr. management to see about getting more focus on
their 32-bit release." (DX 2, at MX 6062581.)
Former Novell employees conceded that Microsoft assisted Novell's development
of applications for Windows 95. Frankenberg testified that he was "sure" that "people in the
[operating] systems group at Microsoft were trying to help WordPerfect/Novell produce a great
application for Windows 95." (Frankenberg, Nov. 7 Trial Tr. at 1131.) Frankenberg added that
Novell software developers worked with Microsoft's operating system developers "on a regular
basis," and that Microsoft developers generally "endeavored to be helpful to Novell."
-32-
(Frankenberg, Nov. 8 Trial Tr. at 1217.)26 LeFevre, who was a Novell employee in 1994 and
1995, testified that "starting in 1994 all through 1995, we had an employee at Microsoft who
lived in Utah County whose job it was to support us in this development effort," and that "[h]e
was at our offices so frequently that we finally gave him an office with a telephone so he could
come in and work when he needed to." (LeFevre, Dec. 2 Trial Tr. at 4029.) LeFevre added that
he and Tom Creighton, who was the Director of PerfectFit Technology (PX 372, Business
Applications Development Organization, Feb. 16, 1995, at 2), traveled to Redmond and "spent
an entire day in building 22 of the Microsoft campus meeting with the development team for
Windows answering some critical questions that Tom had about the product" (LeFevre, Dec. 2
Trial Tr. at 4029-30). LeFevre recalled that Microsoft was "very happy to do this" and "even
paid for our flight and everything to get up to Redmond and spend the day." (Id. at 4030.)
The October 3 Decision conferred no benefit on Microsoft Office or the
component applications of the Office suite. To the contrary, Gates testified that no commercially
released version of Office, Excel, Word, PowerPoint or Access ever used the namespace
extension APIs, and Alepin agreed (at least during the 1994 through 1996 time period, which
was as far as he looked). (Gates, Nov. 21 Trial Tr. at 2826; Alepin, Nov. 10 Trial Tr. at 1641-43;
-33-
see also Belfiore, Dec. 5 at 4280-81 (testifying that Office 95 and Office 97 did not use the
namespace extension APIs).) Professor Bennett testified that "Microsoft Office 95, Microsoft
Office 97, and Microsoft Office 2000 did not use the NameSpace extension APIs." (Bennett,
Dec. 12 Trial Tr. at 4990-91.) Not a single witness disagreed.27
Maritz's November 7, 1994 e-mail to Gates reported that the only Microsoft
software that had planned to use the namespace extension APIs were two parts of Windows 95,
Marvel (the MSN client) and Capone (the e-mail client), but that Capone "ha[d] found ways not
to use them." (DX 82; see also Gates, Nov. 22 Trial Tr. at 3085-86.) Marvel and Capone were
not separate products; they were technologies that were "only shipping with Windows 95."
(Gates, Nov. 21 Trial Tr. at 2815, 2817; Alepin, Nov. 9 Trial Tr. at 1415.)
4. There Are No Novell Documents Showing that The October 3 Decision
Hurt Novell
Novell introduced no evidence that anyone at Novell ever contacted anyone at
Microsoft (other than Harral's vague testimony about a few calls to Premier Support) to
complain about the October 3 Decision or to say that the decision had or might have an adverse
impact on Novell's development efforts. Indeed, the evidence was to the contrary. As shown
above, DX 3 shows that in October 1994, Struss was told "WP [WordPerfect] . . . appear[s] to be
OK with this." (DX 3, at MX 6055840; see Struss, Nov. 28 Trial Tr. at 3272-73.) Struss
testified that, despite regular contact between Novell and DRG (see, e.g., DX 22, E-mail from
-34-
Mark Calkins, March 6, 1995), Novell raised no complaints about the namespace extension APIs
to anyone else at DRG (Struss, Nov. 28 Trial Tr. at 3276, 3281-82).28
Gates testified that he attended "lots of meetings with ISVs in late 94, early 95
and at none of those meetings did anyone come forward and say either that they were using them
[the namespace extension APIs] or that they had any issues related to them whatsoever." (Gates,
Nov. 21 Trial Tr. at 2815-16.) In fact, at a November 13, 1994 dinner hosted by Gates and
attended by Ad Rietveld and Dave Moon of Novell, neither of them expressed any concerns
about the namespace extension APIs. (Id. at 2821-23; DX 84, E-mail from Brad Struss to Bill
Gates, Nov. 12, 1994, at MX 9025187.) Even Frankenberg, who met with Gates on January 10,
1995, testified that he did not recall any discussion about the namespace extension APIs.
(Frankenberg, Nov. 7 Trial Tr. at 1121-22.) The detailed 8-page minutes taken by Dave Miller
of Novell of this January 10, 1995 meeting--a meeting held in part so that Novell could
complain to Microsoft about other issues--also contained no mention of the namespace
extension APIs. (DX 636, Memo from Dave Miller to Frankenberg and others, Jan. 10, 1995.)
Thus, the contemporaneous evidence showed that Novell complained not at all
and instead told Microsoft that it (Novell) was "OK" with the decision.29 Frankenberg testified
-35-
that, although he "complained aggressively to Microsoft" about other issues, Novell made no
complaint in writing to Microsoft about the namespace extension APIs. (Frankenberg, Nov. 8
Trial Tr. at 1269-70.)30 If the October 3 Decision had in fact injured Novell, it is
incomprehensible that Novell would have remained silent about it.
Moreover, Novell introduced no writing of any kind indicating that anyone at
Novell urged management to complain to Microsoft about this issue. This is notable for two
reasons. First, the internal silence contradicts Novell's claim at trial that the decision caused
significant harm to the company or delayed its development efforts. Second, as Frankenberg
testified, any action that could jeopardize the timely release of WordPerfect or Quattro Pro--which "had some real important consequences for Novell" (Frankenberg, Nov. 8 Trial Tr. at
1180)--would have been referred to some or all of four senior executives: Ad Rietveld,
Executive Vice President of the Novell Applications Group; Dave Moon, Senior Vice President
of the Business Applications Group; Mark Calkins, Vice President and General Manager of the
Business Applications Group; and Glen Mella, Vice President of Marketing. (Frankenberg, Nov.
7 Trial Tr. at 1140-42; Frankenberg, Nov. 8 Trial Tr. at 1179-80.) Frankenberg also agreed that
"[i]n any business organization faced with an important decision," a formal memorandum would
normally be presented to the senior executives "laying out the concerns and the issues and the
-36-
considerations facing that business in making some strategic or tactical choice." (Frankenberg,
Nov. 8 Trial Tr. at 1181.)
For example, the evidence at trial included three separate documents addressed to
Novell executives discussing whether Novell would participate in the Windows 95 logo licensing
program or on what terms: (a) a January 12, 1995 memorandum from Calkins to Frankenberg,
Rietveld, Moon, Mella and others discussing possible Novell responses to Microsoft's logo
requirements (DX 155), (b) a February 2, 1995 e-mail from Todd Titensor to Ryan Richards
describing Frankenberg's decision to oppose one of the logo requirements or refuse to participate
in the logo licensing program (DX 157), and (c) a March 6, 1995 e-mail from Calkins to Chase
and Silverberg of Microsoft, copying Frankenberg, Mella and others, requesting an exemption
from Microsoft's logo program requirements (DX 22). Frankenberg conceded that the January
12, 1995 memo (DX 155) was the type of formal memorandum that would "normally" be written
when Novell was faced with an important strategic or tactical decision. (Frankenberg, Nov. 8
Trial Tr. at 1181.) Nevertheless, Novell introduced no such document referring to the important
choices it faced after the October 3 Decision or how best to proceed in view of the "vital[]
importan[ce]" of these issues. (Frankenberg, Nov. 7 Trial Tr. at 998-99.)31
-37-
In fact, Frankenberg testified that he had never seen any memorandum regarding
the October 3 Decision, and that he was never consulted about the shared code group's decision
to spend almost a year attempting to write a custom file open dialog. (Frankenberg, Nov. 7, Trial
Tr. at 1132-34; Frankenberg, Nov. 8 Trial Tr. at 1180-81.) Each of Harral, Richardson and Gary
Gibb, the director of PerfectOffice, testified that he never spoke with any Novell senior executive
regarding what Novell should do in light of Microsoft's October 3 Decision. (Harral, Oct. 24
Trial Tr. at 401-02; Richardson, Oct. 25 Trial Tr. at 703; Gibb, Oct. 26 Trial Tr. at 869.)
Frankenberg also knew of no "evidence whatsoever that any of the four [executives], Calkins,
Mella, Moon or Rietveld ever were presented with a decision about how to respond to Gates'
decision to withdraw support for the namespace extension APIs." (Frankenberg, Nov. 8 Trial Tr.
at 1181-82.) No such evidence was ever introduced at trial, and Novell called none of these four
former senior executives to the witness stand.
5. WordPerfect Was Often Tardy and Novell Struggled to Catch Up After Its
1994 Acquisition of WordPerfect
Novell’s claim that Microsoft is to blame for Novell’s tardiness in releasing products for the Windows 95 platform must be evaluated in light of the uncontested evidence that WordPerfect/Novell had consistently been late to develop and release products. WordPerfect’s failures to anticipate and prepare for the two major shifts in the software industry in the late 1980s and early 1990s—the shift from character-based to graphical user interface (“GUI”) operating systems and the shift from sales of standalone applications to office productivity suites—caused WordPerfect to gain the well-deserved reputation of being behind the curve. Even after Novell acquired WordPerfect in June 1994, Novell struggled to catch up while simultaneously grappling with difficulties arising out of the acquisition.
-38-
In the mid-1980s and early 1990s, WordPerfect was the acknowledged "king of
the hill on the [character-based] DOS platform." (Peterson, Dec. 7 Trial Tr. at 4667; see also
Dec. 13, 2008 Middleton Deposition, Dec. 5 Trial Tr. at 4178.) WordPerfect enjoyed about 75%
of the market for word processing software, and WordPerfect for DOS accounted for 80%-90%
of the company's revenues. (Peterson, Dec. 7 Trial Tr. at 4665-66.) By contrast, Microsoft had
as far back as 1984 concentrated on "building the operating system [t]hat became Windows 1.0,"
a GUI-based operating system (Gates, Nov. 21 Trial Tr. at 2709), and tried to persuade ISVs to
write applications for Windows (id. at 2713). Pete Peterson recalled that in October or
November of 1989, Gates personally "stopped me [at a conference] and said you need to write
for Windows." (Peterson, Dec. 7 Trial Tr. at 4708.)
But WordPerfect Corporation failed to heed Gates' advice. WordPerfect resisted
writing software for Windows because, as Peterson testified, the company "would rather have
someone else besides Microsoft, our main competitor, own the operating system" market. (Id. at
4670-71.) Craig Bushman, WordPerfect International Product Marketing Manager, testified that
Peterson's personal antipathy toward Microsoft and Gates partly motivated the decision not to
devote sufficient resources to Windows development, because Peterson "was not going to put
any effort into producing a product that would put another penny in Bill Gates' pocket."
(Bushman, Nov. 28 Trial Tr. at 3152-53.)
Thus, when Microsoft released Windows 3.0 in May 1990--a product that
Professor Noll called a "revolutionary technological leap" (Noll, Nov. 15 Trial Tr. at 1910)--WordPerfect was caught unprepared. Windows 3.0 became immensely popular and caused a
major shift to GUI-based operating systems. (See Frankenberg, Nov. 7 Trial Tr. at 1040-43; see
also Noll, Nov. 15 Trial Tr. at 1909-10.) WordPerfect did not come out with a product for
-39-
Windows 3.0 until November 1991, about 18 months later. (Dec. 13, 2008 Middleton
Deposition, Dec. 5 Trial Tr. at 4187.) As Frankenberg acknowledged, such a delay in the
software industry could prove to be a "big disadvantage." (Frankenberg, Nov. 7 Trial Tr. at
1059-63; see also Dec. 13, 2008 Middleton Deposition, Dec. 5 Trial Tr. at 4198.) Thus, by the
time WordPerfect reacted, the company was "suddenly behind the curve a little bit and trying to
play catchup," which "certainly changed the perception" of the company "as a technological
leader." (Larsen, Nov. 30 Trial Tr. at 3587-88.)
In the early 1990s, there was a second major shift--also missed by
WordPerfect--from sales of individual standalone applications to sales of office productivity
suites. Microsoft created the concept of the office suite by releasing Microsoft Office 1.0 in
1990. (Frankenberg, Nov. 7 Trial Tr. at 1080; Gibb, Oct. 26 Trial Tr. at 823.) Frankenberg
testified that this gave Microsoft a "huge head start" in the suite market. (Frankenberg, Nov. 7
Trial Tr. at 1060-64.) According to Bushman, Microsoft Office was "a stunning development"
and a "brilliant move," which WordPerfect/Novell quickly realized would "put us in a very
difficult position." (Bushman, Nov. 28 Trial Tr. at 3153.) Frankenberg acknowledged that by
the mid-1990s, "customers were buying suites rather than individual products." (Frankenberg,
Nov. 7 Trial Tr. at 1068.) See also pp. 45-46 n.32, infra.
Novell's own internal documents acknowledged that Microsoft Office was of very
high quality, rating "Word and Excel" as "the strongest combination in the industry." (DX 7,
PerfectOffice Business Review Exercise, July 15, 1994, at 3.) Peterson acknowledged that
Microsoft Office was "tough competition" because Excel "looked like a really nice product" and
WordPerfect had no spreadsheet. (Peterson, Dec. 7 Trial Tr. at 4679-80.) The high quality of
Office and the fact that it was first to market gave Microsoft "enormous momentum," and by
-40-
April 1995, Novell estimated that Microsoft held 86% of the suite market. (DX 271, Novell
Business Applications Business Plan, April 3, 1995, at 6, 9.) In the same document, Novell
estimated that Lotus SmartSuite had 13% of the suite market. (Id. at 7.)
Indeed, WordPerfect's first two suite products, Borland Office 1.0, released in
May 1993--nearly three years later than Microsoft Office--and Borland Office 2.0, released in
January 1994, were "incomplete suite[s]" that were poorly received in the marketplace. (DX
267, 1994 Business Plan, Aug. 17, 1993, at 2.) The trial record unanimously established that
Borland Office lacked integration and interface consistency, and functioned instead as "separate
applications that were put together in a cardboard box." (Acheson, Dec. 1 Trial Tr. at 3892.)
Larsen testified that "there really was no consistency to how [WordPerfect and Quattro Pro]
behaved or how they looked," and that WordPerfect "had dropped the ball" in developing
Borland Office. (Larsen, Nov. 30 Trial Tr. at 3590, 3595.) Gibb admitted that Borland Office
was merely a "stopgap product." (Gibb, Oct. 26 Trial Tr. at 826-27.) In the face of "a very slick
strategy from Microsoft," WordPerfect could only offer "some cobbled products together" in lieu
of a true suite. (Bushman, Nov. 28 Trial Tr. at 3154.) According to David Acheson,
WordPerfect's Sales Director of Enterprise Accounts, the introduction of Microsoft Office
caused a dramatic decline in sales of WordPerfect. (Acheson, Dec. 1 Trial Tr. at 3891.)
The quality of WordPerfect's products for Windows, as a general matter, also was
not good. Peterson testified that when WordPerfect Corporation finally released
WordPerfect 5.1, its first product for Windows, in November 1991, "[t]he reviews were
lukewarm" and the product was "not as well received as we had hoped." (Peterson, Dec. 7 Trial
Tr. at 4677.) Charles Middleton, the director of WordPerfect 5.1 for Windows, testified that
WordPerfect released the product "to get us to the market with something as quick as we could
-41-
because we were so far behind" and thus, even at the time of the release, WordPerfect had "a lot
of things that we had discovered that we knew that we wanted to fix." (Dec. 13, 2008 Middleton
Deposition, Dec. 5 Trial Tr. at 4198-99.)
The next major version of WordPerfect, WordPerfect 6.0 for Windows 3.1,
released in October 1993, also was not well-received and was "perceived as a slow and buggy
product." (DX 271, at 9.) WordPerfect acknowledged that 6.0 "was considered by the press and
many users (in its initial release) as too slow for their current hardware and as compared to the
competition and containing too many bugs to be considered sufficiently stable." (DX 259,
WordPerfect for Windows "Eliot" Marketing Requirements Document, Dec. 21, 1993, at 2.)
Gibb agreed that WordPerfect 6.0 was "reviewed as having some bugs and slow." (Gibb, Oct.
26 Trial Tr. at 895-96.) These "performance issues and quality issues" damaged WordPerfect's
reputation and partly accounted for Novell's falling sales. (Frankenberg, Nov. 7 Trial Tr. at
1065, 1067-68.)
As a result, WordPerfect/Novell was forced to spend almost all of 1994 working
on its products for Windows 3.1, rather than developing for Windows 95. Indeed, before the
merger with Novell, WordPerfect recognized that it could not "afford to have Eliot
[WordPerfect 6.1] go out in a similar state" as 6.0 and that "[e]very effort must be made" to fix
the product. (DX 259, at 6-7.) Accordingly, Frankenberg's "objective for the last six months of
1994" was to create a suite compatible with Windows 3.1, and thus the company put "most" of
its "resources . . . on what became . . . Perfect Office 3.0." (Frankenberg, Nov. 7 Trial Tr. at
1068.) Gibb agreed that "in 1994 WordPerfect had to work really hard and work really long
hours to get out another version of WordPerfect for Windows 3.1 that ran faster and with fewer
bugs." (Gibb, Oct. 26 Trial Tr. at 845; see also LeFevre, Dec. 2 Trial Tr. at 4038.) Indeed,
-42-
Steve Weitzel, Director of WordPerfect for Windows, tearfully told his software engineers in
early 1994 that "this is basically our last chance to get this thing right." (Ford, Nov. 30 Trial Tr.
at 3680.)
On March 21, 1994, Novell announced that it would acquire WordPerfect
Corporation and Borland's Quattro Pro application. (Frankenberg, Nov. 7 Trial Tr. at 1037.) In
the two days following Novell's announcement, Novell's stock price dropped by 20%.
(Frankenberg, Nov. 7 Trial Tr. at 990.) The decline in Novell's market capitalization for the two
days was about $1.5 billion, almost exactly equal to the $1.55 billion purchase price for
WordPerfect and Quattro Pro. (Id. at 1095; Hubbard, Dec. 6 at 4463-64.) Novell offered no
explanation for Novell's stock price drop other than the market's recognition that Novell had
overpaid and that adding new product lines might cause Novell's management to lose focus on
its core products. (Warren-Boulton, Nov. 16 Trial Tr. at 2229-31, 2235-36.)
On June 24, 1994, Novell completed its purchase of WordPerfect and Quattro
Pro, hoping to combine these products into a well-integrated suite. (See Compl. ¶ 37.) But,
instead of devoting resources toward development of new products for Windows 95, Novell
spent most of 1994 working on versions of its products written for the old platform,
Windows 3.1. Even in August 1994, Novell had "very few resources on Chicago." (DX 4,
Novell/WordPerfect/Quattro Pro Unification Plan, Aug. 3, 1994, at 5.) In October 1994, Novell
informed Microsoft that "[d]ue to focus on 16-bit product revision this fall [i.e., WordPerfect and
PerfectOffice for Windows 3.1], there are limited resources working on next years 32-bit
release." (DX 2 at MX 6062581.) WordPerfect 6.1 was finally released in November 1994 and
PerfectOffice 3.0 was released in December 1994. (Frankenberg, Nov. 7 Trial Tr. at 1008,
-43-
1013.) By this point, Novell had about eight months to develop and test versions of its products
for Windows 95.
This late start in developing products for Windows 95 was nothing new. Novell
reported in a Form 10-K it filed in early 1995 that it "has experienced delays in its product
development and 'debugging' efforts, and the Company can be expected to experience similar
delays from time to time in the future" because such delays were "common in the computer
software industry." (DX 380, Novell Form 10-K for Fiscal Year Ended Oct. 29, 1994, filed
Jan. 25, 1995, at 10.) Frankenberg agreed that it was "common in the software industry for
companies to experience delays in developing new software products." (Frankenberg, Nov. 7
Trial Tr. at 1073.) Alepin also testified that software projects "tend to be late and they don't
meet their deadlines, their announced deadlines" and that "some organizations" are "overly
optimistic" and "frequently miss their release dates." (Alepin, Nov. 10 Trial Tr. at 1544-45; see
also Noll, Nov. 15 Trial Tr. at 1881 (acknowledging that "[d]elays happen" in software
development).)
In this same 1994-95 period, Novell also encountered difficulties arising from the
acquisition itself. For example, Novell's layoffs of legacy WordPerfect personnel were
"particularly painful because they cut deeply into our development teams, into our testing team,
and probably most significantly into our sales organization." (LeFevre, Dec. 2 Trial Tr. at 4022-23.) Historically, WordPerfect's legacy sales force "had very good customer relationships with
these large accounts," which proved "critical to the sales ultimately." (LeFevre, Dec. 2 Trial Tr.
at 4024-25.) According to LeFevre, Novell generated 85% of its revenue through these close
customer relationships, and Novell "lost that when we lost the sales force." (Id. at 4024-26.)
Novell instead imposed its own value added reseller ("VAR") sales distribution model, which
-44-
meant that "the customer face-to-face contact was reduced dramatically." (Acheson, Dec. 2 Trial
Tr. at 3972.) The VAR model proved unsuccessful in replacing personal customer relationships,
and sales of WordPerfect software applications "decreased dramatically after those lay-offs in
1994." (Id. at 4025-26; see also id. at 3972.)
The layoffs also had a significant impact on employee morale. About 93% of
those laid off by Novell in August 1994 were legacy WordPerfect employees. (Frankenberg,
Nov. 7 Trial Tr. at 1097-98; DX 15, Novell/WordPerfect Integration Memorandum from Joe
Marengi to Frankenberg, Aug. 19, 1994.) Former Novell employees testified about the
"significant lack of morale on the WordPerfect legacy employee side" (Bushman, Nov. 28 Trial
Tr. at 3161-62) and the "total demoralization of the team" (Acheson, Dec. 2 Trial Tr. at 3968).
A survey of Novell employees conducted in spring 1995 showed that "48% of Employees
Originally Hired at WordPerfect Are Thinking About Quitting." (DX 16, Novell Employee
Survey, May 3, 1995, at 40.) Even Frankenberg acknowledged that the layoffs "did have an
impact on morale." (Frankenberg, Nov. 7 Trial Tr. at 1098.)
6. PerfectOffice 3.0, a Product Unaffected by The October 3 Decision, Was
Not a Success in the Marketplace
The second shift, from standalone word processing and spreadsheet applications
to office suites,32 was a huge problem for Novell given the poor quality of the two earlier
-45-
releases of Borland Office. See pp. 40-41, supra. As Frankenberg explained, the "major reason"
for WordPerfect's decline in 1994 was that "we didn't have a suite to offer customers, and
customers were buying suites rather than individual products." (Frankenberg, Nov. 7 Trial Tr. at
1068.) Professor Murphy agreed that Novell's low sales in the suite market had "particular
significance because the market was moving more and more to buying things in suites."
(Murphy, Dec. 7 Trial Tr. at 4760; see also Hubbard, Dec. 6 Trial Tr. at 4444-50.) Accordingly,
in 1994, Novell put "most of its efforts" into PerfectOffice 3.0 (a product written for the old
Windows 3.1 platform). (Frankenberg, Nov. 7 Trial Tr. at 1068.)
Novell released PerfectOffice 3.0 in December 1994. (Frankenberg, Nov. 7 Trial
Tr. at 1008.) The product received favorable reviews in the trade press33 and Frankenberg
testified that he was pleased with PerfectOffice 3.0 and the "great integration" of its components.
(Frankenberg, Nov. 7 Trial Tr. at 1012.) Nolan Larsen agreed that Novell had finally produced a
quality suite product. (Larsen, Nov. 30 Trial Tr. at 3643-46; see also Bushman, Nov. 28 Trial
Tr. at 3214.) Despite all this, sales of PerfectOffice 3.0 in the first seven months of 1995
(January 1 through July 31) were very small--according to Noll, it captured about 8% of the
suite market. (Noll, Nov. 15 Trial Tr. at 1915.)34
-46-
Of course, the poor sales of PerfectOffice 3.0--which was written for the 16-bit
Windows 3.1--cannot be blamed on the allegedly wrongful act (which could have affected only
products written for Windows 95). In 1995, Novell attributed the low sales of PerfectOffice to,
among other things, Novell's reputation for producing "slow and buggy" products, Microsoft's
head start in the suite market, and Microsoft's superior products. A Novell business plan, dated
April 3, 1995, stated that Novell was "[s]till recovering from WordPerfect 6.0 for Windows,
which was perceived as a slow and buggy product" (DX 271, at 9; see also id. at 6), and
Frankenberg testified that such a reputation can "stick around for a long time" and affect future
sales (Frankenberg, Nov. 7 Trial Tr. at 1068, 1091).
Novell introduced no evidence whatsoever that PerfectOffice for Windows 95
would have fared any better in the "but-for" world than PerfectOffice 3.0. In fact, Noll testified
that his "expectation" was that PerfectOffice for Windows 95 would have had about the same
(low) market share as that of PerfectOffice for Windows 3.1. (Noll, Nov. 15 Trial Tr. at 1911-12.) Further, because Windows 95 was backward compatible, PerfectOffice 3.0--as well as
WordPerfect 6.1--would run just fine on Windows 95. (Alepin, Nov. 10 Trial Tr. at 1581;
Gates, Nov. 21 Trial Tr. at 2754.) But sales of PerfectOffice 3.0 dropped to 1.7% of the suite
market by the third quarter of 1995. (Hubbard, Dec. 6 Trial Tr. at 4483.) Market forces, not any
wrongful conduct, caused Novell to fail.
-47-
7. Quattro Pro Delays Caused the Delay in Releasing PerfectOffice for
Windows 95
Further, the overwhelming evidence at trial established that delays with Quattro
Pro, the spreadsheet component of PerfectOffice, caused Novell to be late to release
PerfectOffice for Windows 95. Quattro Pro "was an essential element" of PerfectOffice and
"[i]n order to get PerfectOffice out to market, [Novell] needed to have Quattro Pro ready to go."
(Frankenberg, Nov. 7 Trial Tr. at 1143.) As it turned out, Quattro Pro for Windows 95 was not
ready until well into 1996. Because Novell could not release PerfectOffice for Windows 95
without Quattro Pro, the October 3 Decision--even assuming it delayed the work of the shared
code group--did not harm Novell. The delays in developing a version of Quattro Pro for
Windows 95, which had nothing to do with the October 3 Decision, meant that PerfectOffice was
not ready to be released until 1996 in any event.35
Following Novell's acquisition of the Quattro Pro spreadsheet from Borland in
June 1994, some Quattro Pro developers working on Quattro Pro resigned, and the others
remained in Scotts Valley, California (where Borland had its headquarters), rather than moving
to Utah. (Frankenberg, Nov. 7 Trial Tr. at 1070-71.) A Novell memorandum dated August 3,
1994 indicated that a "key issue" for Novell was "[g]etting company resources focused on
supporting Quattro Pro." (DX 4, at 5.)
-48-
After Novell released Quattro Pro 6.0 for Windows 3.1 in October 1994, the
Quattro Pro team subsequently spent "many months" creating localized (foreign language)
versions of Quattro Pro 6.0. (LeFevre, Dec. 2 Trial Tr. at 4046.) As Bushman explained,
Quattro Pro had been written in such a way that creating localized versions was a "significant
development effort." (Bushman, Nov. 28 Trial Tr. at 3181; see also Larsen, Nov. 30 Trial Tr. at
3614-15.) As a result, the Quattro Pro team "didn't get started on their Windows 95 efforts until
well into 1995," and LeFevre realized "there was no chance they were going to hit" the release
dates Novell was targeting. (LeFevre, Dec. 2 Trial Tr. at 4046.)36
Although in late 1994 Novell had been considering a September 30, 1995 release
date for PerfectOffice for Windows 95, the Quattro Pro team then "believe[d] this is barely
achievable with all their resources and with no additional functionality." (DX 211, Project
Proposals for `Storm,'37 at NOV-B01491217; see also Noll, Nov. 15 Trial Tr. at 1885; Gibb,
Oct. 26 Trial Tr. at 867-68.) As shown below, the target date was soon pushed back for reasons
having nothing to do with the shared code group's work.
As of February 1995, the "Quattro Pro folks [were] still working on International
versions of QP 6.0" and "[e]xpect[ed] to finish that [localization work] by end of March," and
only then (after March) would they "begin on [the] next version of QP." (DX 219, Notes from
-49-
Storm Coordination Meeting, Feb. 2, 1995, at NOV-B 6655277.) On March 1, 1995, Bruce
Brereton (Vice President of the Business Applications Unit at Novell) stated in an e-mail that
because Quattro Pro believed that "December 30th is a more realistic date," Novell had decided
to "move[] the Storm RTM date back by one month (to December 30th) and have put WP on the
same time-line as Storm." (DX 221, at NOV-B13528783.) Frankenberg explained that "RTM"
means "release to manufacturing," and thus, he agreed that as of "March 1st of '95 the plan
became to get PerfectOffice out, released to manufacturing, not even to the market," on
December 30, 1995. (Frankenberg, Nov. 8 Trial Tr. at 1220-21.)
Despite pushing back the target RTM date to end-of-year 1995, Novell was
concerned that Quattro Pro would cause PerfectOffice to be delayed even beyond that. A March
1995 "Market Requirements Document" prepared by the Applications Group, ranked "Quattro
Pro delivering late" as the highest "overall risk" for the PerfectOffice development project:
-50-
(DX 223, at 41.) On May 25, 1995, "Quattro Pro delivering late" was still ranked as the highest
"overall risk" for the PerfectOffice development project. (DX 226, Project Development Plan
for Storm, at NOV-B01425535.) LeFevre testified that he attended "daily" meetings in 1995
with Gibb, Weitzel and Creighton where they discussed "all the different product challenges" in
releasing PerfectOffice for Windows 95 in a timely manner, and "the product that was causing
the biggest problem was Quattro Pro." (LeFevre, Dec. 2 Trial Tr. at 4037, 4045-47.) Karl Ford,
the Lead Developer for the User Interface in WordPerfect for Windows 95, attended "regularly
scheduled meetings every week or so" in 1995 to discuss the "risks and features" of
PerfectOffice for Windows 95 and learned that "the schedule" was at risk because of Quattro
Pro. (Ford, Nov. 30 Trial Tr. at 3691-92, 3699-70.)
As the difficulties with Quattro Pro became increasingly apparent in 1995, Novell
gave half-hearted consideration to the idea of shipping the PerfectOffice suite without Quattro
Pro, and including instead a coupon that would allow the customer to obtain the Quattro Pro
functionality later. (See, e.g., DX 223, at 41.) LeFevre, who was responsible for marketing
Windows products at Novell, testified that such a plan "wasn't a serious proposal," and that
Novell "never seriously considered that." (LeFevre, Dec. 2 Trial Tr. at 4047.) Bushman agreed
that a voucher plan was "simply not feasible." (Bushman, Nov. 28 Trial Tr. at 3186-87.) Even
Gibb acknowledged that he was "being a little facetious" in describing a voucher program as a
"contingency plan." (Gibb, Oct. 26 Trial Tr. at 866.)38
-51-
By December 23, 1995, four months after the release of Windows 95, Quattro Pro
was still not ready. On that day, Bruce Brereton wrote an e-mail to Frankenberg and others at
Novell reporting that "this past Thursday/Friday, about 15 additional people submitted their
resignations," leaving the Quattro Pro development team in Scotts Valley with "just 2 people."
(DX 230, E-mail from Bruce Brereton, Dec. 23, 1995.) After reviewing DX 230, Frankenberg
conceded that Quattro Pro "wasn't released to manufacturing" even as of December 23, 1995
and "clearly the product wasn't [yet] complete." (Frankenberg, Dec. 7 Trial Tr. at 1145.)
Contrary to this powerful evidence, Gibb testified that Quattro Pro was "basically
code completed" in December 1995. (Gibb, Oct. 26 Trial Tr. at 808.) On direct examination,
Gibb offered vague testimony that "early on" Novell thought "Quattro Pro might be Critical
Path," but that "they were very conservative in their estimates and kind of over delivered."
(Id. at 806.) Gibb offered this testimony with no dates, no explanation, and no reference to any
document whatsoever. Even assuming that Quattro Pro was "basically code completed" in
December 1995, it is clear that a product is not ready to ship merely because it is "code
complete." For example, DX 231, a document used by Novell's counsel in closing rebuttal
argument, showed that PerfectOffice 3.0 was code complete on July 22, 1994 (DX 231,
Development Project Status, Jan. 11, 1996, at NOV00161055), but was not released to the
market until December 1994 (Frankenberg, Nov. 7 Trial Tr. at 1008), five months later. And,
although no witness addressed or explained DX 231, it states on its face that the "RTM" date for
Quattro Pro was "3/31/96." (DX 231, at NOV00161055.)
In January 1996, shortly after the mass exodus of the Quattro Pro developers,
Larsen traveled to Scotts Valley (Larsen, Nov. 30 Trial Tr. at 3619) and found that "it was kind
of a train wreck" and that "[t]hose people who had not resigned were kind of walking around a
-52-
little bit shellshocked . . . [s]o it was § it was very chaotic" (id. at 3620). Larsen explained that
Quattro Pro was not ready to be shipped in January 1996 "by any stretch of the imagination."
(Id. at 3624.) Larsen testified that they could not even locate the source code in order to create
an interim build of the product. (Id. at 3622-24.) As DX 231 also showed, Novell estimated in
January 1996 that Quattro Pro would be ready to be released to manufacturing on March 31,
1996, more than seven months after the release of Windows 95. (DX 231, at NOV00161055.)
Consistent with this, Larsen, LeFevre and Bushman testified that Quattro Pro was not ready to be
shipped even in March 1996. (Larsen, Nov. 30 Trial Tr. at 3624-25; LeFevre, Dec. 2 Trial Tr. at
4062-63; Bushman, Nov. 28 Trial Tr. at 3192-93; see also Larsen, Nov. 30 Trial Tr. at 3664-65.)
Ultimately, Quattro Pro and the PerfectOffice suite were released by Corel on May 29, 1996.39
8. Novell Chose the Most Difficult and Time-Consuming Path Toward
Release of Its Products for the Windows 95 Platform
Harral and Richardson testified that in October 1994, upon learning of
Microsoft's decision to withdraw support for the namespace extension APIs, Novell had three
development options: (1) to continue using and relying on the namespace extension APIs;
(2) to use the Windows 95 common file open dialog (provided to ISVs at no cost); or (3) to build
a custom file open dialog that would, in Novell's view, be superior to the Windows 95 common
file open dialog. (Harral, Oct. 20 Trial Tr. at 342-43; Richardson, Oct. 25 Trial Tr. at 602-04.)
At trial, Novell claimed that its work on the file open dialog was delayed as a result of the
October 3 Decision, and that this delay in turn delayed the release of PerfectOffice for
Windows 95. However, setting aside the Quattro Pro problems, PerfectOffice for Windows 95
-53-
could have been released without any delay--according to Harral and Gibb--by choosing
Option 1 or Option 2. (Harral, Oct. 24 Trial Tr. at 502-04; Gibb, Oct. 26 Trial Tr. at 847-48.)
Harral explained that Option 1 "would be to continue to use the documentation
that we would have had for the APIs." (Harral, Oct. 20 Trial Tr. at 342.) Frankenberg
confirmed that one of the choices available to Novell was to use the namespace extension APIs
at Novell's risk: "That was the nature of undocumented APIs, yes." (Frankenberg, Nov. 7 Trial
Tr. at 1133.)40 Indeed, according to Novell's developers, by October 1994 Novell was already
finished or nearly finished developing a file open dialog that called the namespace extension
APIs. (Richardson, Oct. 25 Trial Tr. at 676-77, 687; Harral, Oct. 24 Trial Tr. at 436-37.)
As for the second option, Gibb testified that Novell had worked on developing
PerfectOffice using the Windows 95 common file open dialog, and that it would have been
"quite easy" for Novell to release its products using the file open dialog that Microsoft provided
for free. (Gibb, Oct. 26 Trial Tr. at 847-48.) Indeed, on September 22, 1994, Struss reported
that Novell had said that its "current plan [was] to use the MS [Microsoft] dialogs." (DX 17, at
MX 6109491.) Harral agreed that "Novell could have come out with a product in '95 that
utilized the Windows common file open dialog." (Harral, Oct. 24 Trial Tr. at 502.)
-54-
Ford recalled discussions about whether Novell should "just use the common
open dialog right now and use the new one in the next release." (Ford, Nov. 30 Trial Tr. at
3710.) Harral also testified that Novell considered the second option (using the file open dialog
provided by Microsoft) "many times" and "every time we had to consider an option, this . . .
came back on the table because it would have been an easier option than the third" and would
have entailed "less work for us" and been "less risky." (Harral, Oct. 24 Trial Tr. at 365-66.) In
July 1995--nine months after the October 3 Decision and one month before Windows 95 was
released--there were still "differences of opinion of how this dialog would be implemented" and
Novell was still considering using the Windows 95 common file open dialog. (DX 114,
PerfectFit 95: Open File Dialog--Functions and Issues, last modified July 11, 1995, at 1.)
Harral confirmed that, even in July 1995, "one of the decisions that Novell was facing was
whether to use the common file open dialog." (Harral, Oct. 24 Trial Tr. at 488.) Ford said: "I
remember in the summer . . . we had meetings discussing the open dialog, where they were at on
their schedule, whether they should use it, if they should use the common open dialog that
Microsoft provided in their APIs for Windows 95." (Ford, Nov. 30 Trial Tr. at 3709-10.)
Ford testified that he told Gibb and Weitzel in 1995 that the "safest route" was to
use "the common open dialog" provided by Windows if "they were concerned about schedule"
because of uncertainties of whether the project could "be completed on time" if Novell chose
Option 3. (Ford, Nov. 30 Trial Tr. at 3710-11.) Ford testified that he also heard others
recommend the same course: "let's just use the common open dialog right now and use the new
one [Novell's custom file open dialog] in the next release." (Id. at 3710.) Likewise, LeFevre
testified that he "became convinced that . . . what we needed to do was just abandon that entire
effort [to create Novell's own file open dialog] and use the standard Windows dialog . . . . So
-55-
with almost no work on our part, we can take advantage of those things" and thus he "became an
advocate for just going that route as opposed to trying to do all those customizations that seemed
to be taking way too much time." (LeFevre, Dec. 2 Trial Tr. at 4041-42.)
But, as Gibb testified, Novell decided to try to "do something cooler" and "exceed
what was the default stuff." (Gibb, Oct. 26 Trial Tr. at 848-49; see also Richardson, Oct. 25
Trial Tr. at 629-30.) Gibb agreed that Novell was "faced with the age old trade off" between, on
the one hand, "get[ting] out a product more quickly and sacrific[ing] features," or, on the other
hand, "delay[ing] until 1996 and try[ing] to build a cooler product." (Gibb, Oct. 26 Trial Tr. at
891-92.) Novell's shared code group therefore chose the most difficult and time-consuming
path: writing its own custom file open dialog rather than utilizing the common file open dialog.
(Harral, Oct. 20 Trial Tr. at 342-47.) By late 1994, Harral knew that attempting to build a
custom file open dialog "would be a significant commitment in resources." (Id. at 342.)
Novell even chose to take the most difficult path possible to creating its own
custom file open dialog. Alepin testified that Novell could have built its own custom file open
dialog (Option 3) using common controls provided by Microsoft in Windows 95. (Alepin,
Nov. 10 Trial Tr. at 1603-04, 1664.) By "just using the common controls in Windows 95,"
Novell could have "add[ed] whatever custom file locations . . . [it] wanted to add" to its file open
dialog. (Id. at 1664.) Novell's developers could "make use of these common controls to use
them for many different application purposes, one of which could be to pretend to be like the
Windows Explorer." (Id. at 1604.) Ultimately, Novell elected to build its own custom file open
dialog from scratch without using either the common file open dialog or the common controls
(both of which Microsoft provided for free). Novell also could have just used the file open
dialog ported from its Windows 3.1 products, with no loss in functionality. (Id. at 1579-81.)
-56-
It is clear that Novell could have released its products on time, but instead chose
the riskier and more time-consuming path of trying to "do something cooler." Had Novell
chosen to use the Windows 95 common file open dialog, the shared code group would have had
its work completed in plenty of time.
D. The Rule 50(a) Motion
At the close of Novell's case on November 17, 2011, Microsoft filed its Motion
for Judgment as a Matter of Law. (Dkt. #297; see also Nov. 18 Trial Tr. at 2565-66.) Among
other things, Microsoft argued that Novell had failed to introduce evidence that would provide a
legally sufficient evidentiary basis for a reasonable jury to find that Microsoft harmed
competition in the PC operating system market, both because the evidence showed that the
timely release of Novell's Three Products would have enhanced Microsoft's monopoly in the PC
operating system market and because the evidence refuted Novell's theories of harm to
competition. Microsoft also argued that Novell had failed to introduce evidence that would
provide a legally sufficient evidentiary basis for a reasonable jury to conclude that Microsoft's
withdrawal of support for the namespace extension APIs was anticompetitive conduct because
Microsoft's decision (a) did not constitute the termination of Microsoft's relationship with
Novell, (b) was consistent with industry practice and (c) had several legitimate business
justifications. Microsoft also argued that Novell had failed to show that Microsoft's withdrawal
of the support for the namespace extension APIs caused any delay in the release of Novell's
Three Products; and that Novell was not entitled to any damages because Warren-Boulton's
opinion on damages depended on the incorrect assumption that in the but-for world, Novell's
products would have been released within 30 or 60 days of the release of Windows 95.
The Court heard oral argument on Microsoft's motion on November 18 and
November 21, 2011. At the close of argument on November 21, the Court stated that it was
-57-
either "denying [Microsoft's Rule 50 motion] without prejudice of [its] being renewed at the
close of all the evidence" or "reserv[ing] ruling upon the Rule 50 motion until all the evidence is
in." (Nov. 21 Trial Tr. at 2931.) The Court stated that while "I think there are legitimate legal
reasons why the plaintiff may not prove the case," "it seems to me we ought to have a jury
verdict on that as a practical matter." (Id. at 2931-32.) At the close of all the evidence on
December 12, 2011, Microsoft renewed its Rule 50 motion "for all of the same reasons set forth
in our brief and oral argument of that motion." (Dec. 12 Trial Tr. at 5099.) The Court ruled that
the "motion is denied with the same--again I'm not sure whether it is denied or whether it is
deferred or whatever, it is going to the jury and then I'll focus upon it after that." (Id. at 5100.)
Given the jury's inability to come to a verdict and the Court's familiarity with all the facts and
circumstances, the Court should now give its fullest consideration to this motion.
ARGUMENT
Under Rule 50 of the Federal Rules of Civil Procedure, a motion for judgment as
a matter of law may be granted "[i]f a party has been fully heard on an issue during a jury trial
and the court finds that a reasonable jury would not have a legally sufficient evidentiary basis to
find for the party on that issue." Fed. R. Civ. P. 50(a)(1). Rule 50(b) authorizes a party to renew
a motion for judgment as a matter of law after judgment or after the jury is discharged without
reaching a verdict. The standard that applies to a motion for judgment as a matter of law is "the
same whether it arises in the procedural context of a motion for judgment as a matter of law prior
to the submission of the case to the jury under Rule 50(a) or in the context of a renewed motion
for judgment as a matter of law" under Rule 50(b). 9B CHARLES ALAN WRIGHT & ARTHUR R.
MILLER, FEDERAL PRACTICE & PROCEDURE § 2524 (3d ed. 2011); see also Smith v. Aztec Well
Servicing Co., 462 F.3d 1274, 1287 (10th Cir. 2006) ("[W]e review a judgment as a matter of
-58-
law under the same standard regardless of whether the judgment is rendered before or after the
jury renders its verdict.") (alteration in original) (quotation omitted).
To survive a Rule 50 motion, a plaintiff must present "substantial evidence" in
support of its case. Webco Industries, Inc. v. Thermatool Corp., 278 F.3d 1120, 1128 (10th Cir.
2002). "Substantial evidence is something less than the weight of the evidence, and is defined as
such relevant evidence as a reasonable mind might accept as adequate to support a conclusion,
even if different conclusions also might be supported by the evidence." Id. at 1128 (internal
quotation omitted). Under this standard, "`[t]he question is not whether there is literally no
evidence supporting the nonmoving party but whether there is evidence upon which a jury could
properly find for that party.'" Herrera v. Lufkin Industries, Inc., 474 F.3d 675, 685 (10th Cir.
2007) (quoting Century 21 Real Estate Corp. v. Meraj International Investment Corp., 315 F.3d
1271, 1278 (10th Cir. 2003)) (affirming grant of Rule 50(a) motion); see also Bankers Trust Co.
v. Lee Keeling & Associates, Inc., 20 F.3d 1092, 1099-1100 (10th Cir. 1994) (affirming grant of
Rule 50(b) motion).
I. Microsoft's Withdrawal of Support for the Namespace Extension APIs Did Not
Harm Competition.
A. No Reasonable Jury Would Have a Legally Sufficient Evidentiary Basis to
Find that Microsoft Harmed Competition Under Either of Novell's Theories.
Judgment as a matter of law should be entered in Microsoft's favor because a
reasonable jury would not have a legally sufficient evidentiary basis to find that, under either of
Novell's theories, Microsoft's withdrawal of support for the namespace extension APIs harmed
competition in the PC operating system market.
-59-
1. Novell's Franchise Applications Theory Was Unsupported by, and Contrary to,
the Evidence.
Novell was required to prove at trial that WordPerfect and Quattro Pro were such
popular applications that if available on rival operating systems, they would have increased
competition in the PC operating system market by "offer[ing] competing operating systems the
prospect of surmounting the applications barrier to entry and breaking Microsoft's operating
system monopoly." (E.g., Novell's Memorandum Regarding Proposed Final Jury Instructions
and Verdict Forms, filed Dec. 5, 2011, Dkt. #336, at 4.) For this theory to work, Novell would
need at the very least to introduce evidence that there was some plausible prospect that in the
but-for world, WordPerfect would have been highly popular and that end-users would abandon
Windows for other operating systems for some reason having to do with the availability of
WordPerfect on those competing operating systems.41 The evidence at trial disproved Novell's
theory for several independent reasons.
First, the applications barrier to entry that protects Microsoft's monopoly arises
from the very large and diverse population of applications developed for Windows. Indeed, the
Findings of Fact on which Novell sought and obtained collateral estoppel--and which therefore
have "binding effect here" (Oct. 18 Trial Tr. at 143)--demonstrated that the applications barrier
to entry arises from a "positive feedback loop" created by the tens of thousands of applications
written to run on Windows. (Findings of Fact 37-39.) Finding 37 explained that "[t]he fact that
a vastly larger number of applications are written for Windows than for other PC operating
systems attracts consumers to Windows, because it reassures them that their interests will be met
-60-
as long as they use Microsoft's product." Finding 39 established that "[t]he large body of
applications thus reinforces demand for Windows, augmenting Microsoft's dominant position
and thereby perpetuating ISV incentives to write applications principally for Windows." As the
Court instructed the jury before they were read into evidence, these Findings of Fact have
"binding effect" in "this case." (Oct. 18 Trial Tr. at 143.) Novell's counsel told the jury the
same thing. (E.g., Nov. 14 Trial Tr. at 1709 ("the Court, in this case, has decided that certain
facts are binding in this case, and we actually read quite a few of them this morning").)
These Findings of Fact are fatal to Novell's franchise applications theory. As
Professor Murphy explained, the applications barrier to entry is predicated on "not a few, but a
vastly large number of applications [that] are written for Windows than for other PC operating
systems, [which] attracts consumers to Windows because it reassures them that their interests
will be met as long as they use Microsoft's product." (Murphy, Dec. 7 Trial Tr. at 4744.)
Indeed, Novell's Complaint described the applications barrier in exactly this fashion, alleging
that "[a]s found by the courts in the Government Suit, . . . Microsoft's monopoly share of the
Intel-compatible PC operating systems market is protected by a barrier to entry arising out of the
much greater number of applications that operate only with Windows personal computer
operating systems." (Compl. ¶ 43 (emphasis added).)42 In addition, this barrier is made stronger
by the fact that these thousands of Windows applications were "in lots of categories . . . . Even
within a category, having more than one choice is important. So [consumers] didn't want to go
-61-
to an operating system where there is only one office productivity application, they liked having
two, three, whatever it is. If they don't like one, they can move to the other." (Murphy, Dec. 7
Trial Tr. at 4744.)
Thus, a franchise applications theory that asserts that a small number of
applications in a narrow category could affect competition in the PC operating system market is
contrary to the premise on which Novell's claim is predicated. Indeed, in the Government Case,
Microsoft contended "that software developers do write applications for other operating systems
[and] point[ed] out that at its peak IBM's OS/2 supported approximately 2,500 applications."
United States v. Microsoft Corp., 253 F.3d 34, 55 (D.C. Cir. 2001). Microsoft therefore argued
that an operating system could be "competitive" even if it had far fewer than 70,000 applications
(the number written to Windows, according to Finding 40). Id. The D.C. Circuit rejected this
argument as "miss[ing] the point" because "[a]s the District Court explained [in Finding 37], . . .
the applications barrier to entry gives consumers reason to prefer the dominant operating system
even if they have no need to use all applications written for it." Id. Here, despite the conclusive
nature of Finding 37, Novell argues that two or three applications nevertheless had the potential
to surmount the applications barrier.
No reasonable jury would have a legally sufficient evidentiary basis to reject the
Findings of Fact or permit Novell to depart from the applications barrier to entry underlying the
Government Case and set forth in the Complaint. The availability of PerfectOffice, WordPerfect
and Quattro Pro on non-Microsoft operating systems is insufficient as a matter of law to have
induced users to move from Windows to some other operating system. For this reason alone,
Novell's theory that its products could accomplish what, according to the D.C. Circuit, even the
-62-
existence of 2,500 diverse applications on a rival operating system could not, is alone sufficient
to reject Novell's franchise applications theory.
Second, even assuming that two or three highly popular applications could
possibly reduce the applications barrier to entry, no reasonable jury would have a legally
sufficient evidentiary basis to find that Novell's Three Products had anything close to the
necessary level of popularity. Novell had the burden to establish at trial that its products had
sufficient popularity to induce users to switch operating systems and that, in the but-for world,
competition in the PC operating system market would have been enhanced. Novell failed
entirely to meet this burden and, indeed, introduced no evidence showing that its products had
achieved a high market share or were highly "popular" under any meaning of the term. The only
independent market share document introduced by Novell showed that WordPerfect's share of
shipments of word processors in 1994 was 22%; and that even looking at installed base,
WordPerfect had 29.4% of the installed base on all Windows platforms and 36% of the installed
base on Windows and DOS platforms combined. (PX 599A, IDC Report, The Word Processing
Software Market Review and Forecast 1994-1999: DOS, Windows, OS/2, and Macintosh, at
Table 13.) There was no evidence that Novell's products for the Windows platform ever
achieved a level of popularity high enough to affect the competition in the market in which
Windows competed.
For example, WordPerfect's share of the word processing market on Windows
peaked at around 30% in 1992 and subsequently declined to "the low 20s and going down" by
1994. (Murphy, Dec. 7 Trial Tr. at 4751.)43 WordPerfect/Novell's own internal documents
-63-
showed that WordPerfect's worldwide market share by revenue on the Windows platform was
33.8% in 1993 and was projected, as of July 1994, to fall to 22% for 1994 and remain at 22% for
1995. (DX 294, WordPerfect Corp. Market Share Analysis Using Internal and SPA Data,
undated, at NOV 00068402; DX 9, at 4; see also Microsoft's Demonstrative 307, shown at trial
on Dec. 13, Trial Tr. at 4749, 4788, Holley Decl. Ex. G (showing data from IDC reports that
WordPerfect's market share of sales on the Windows platform was 31% in 1992, 24% in 1994
and 16% in 1995).)44 With respect to Quattro Pro, "Novell and Borland were never really large
sellers of the spreadsheet software," and by 1994, Novell/Borland's spreadsheet product Quattro
-64-
Pro possessed a "very, very small fraction" of the spreadsheet market for Windows. (Murphy,
Dec. 7 Trial Tr. at 4756.)45 Novell submitted no concrete evidence that in the but-for world,
WordPerfect or Quattro Pro's market shares would have exceeded these low numbers.
PerfectOffice never had more than a miniscule portion of the suite market.46
According to WordPerfect/Novell documents, Novell had a 2% share of the suite market as of
August 1993 and a 2.5% share in 1994.47 (DX 223, at 2; PX 448, IDC Report, PC and Consumer
Software Office Suites Market Review and Forecast: Revised 1994 Market Sizing, at 3; see also
Microsoft's Demonstrative 311, shown at trial on Dec. 13, Trial Tr. at 4759, 4788, Holley Decl.
Ex. K (showing data from IDC reports that Novell's market share in the suite market was 2.2%
in 1993, 2.6% in 1994 and 3.6% in 1995).) In fact, even after Novell released its first high
quality suite product in December 1994, that product captured only 8% or less of the suite
market. Noll testified that PerfectOffice 3.0's market share in the first seven months of 1995 was
"roughly" 8%. (Noll, Nov. 15 Trial Tr. at 1915.) Other data show that PerfectOffice 3.0 had
-65-
only 4.5% of the suite market during the first three quarters of 1995 and that PerfectOffice's
revenues and market share decreased throughout 1995. (Hubbard, Dec. 6 Trial Tr. at 4483; DX
213, WordPerfect, The Novell Applications Group, Quarterly Review Background, Q1 1995, at
NOV 00725771 (PerfectOffice revenue in Q1 1995 was $38.4 million); DX 358, Novell
Applications Group Year to Date/Quarterly Review Background Q3 1995, at 16 (PerfectOffice
revenue in Q3 1995 was $20.2 million).)48
Professor Noll testified that had there been no delay in the release of
PerfectOffice, that product would have achieved a similar market share on Windows 95 as the
earlier version had achieved at the "end of the life of" Windows 3.1, which Noll defined as about
two years preceding the release of Windows 95. (Noll, Nov. 15 Trial Tr. at 1911-12.) By this
measure, PerfectOffice's market share on Windows 95 would have been (in the but-for world) in
the range of 2% to 8%.
This makes the franchise applications theory completely unviable. Indeed, the
notion that products with low market shares could "offer[] competing operating systems the
prospect of surmounting the applications barrier to entry and breaking the Windows monopoly"
(Novell's Memorandum Regarding Proposed Final Jury Instructions and Verdict Forms, filed
Dec. 5, 2011, Dkt. #336, at 4) is untenable. Because only a small percentage of end-users with
Windows PCs used Novell's products, only that small percentage would have the option of
switching to a different operating system even if the rest of Novell's theory had been proven (and
it was not). (See Murphy, Dec. 7 Trial Tr. at 4750.) Even if Novell had decided after the release
of Windows 95 to develop and release its products for competing operating systems, products
-66-
with a small share of the Windows market could not possibly change the competitive landscape
in the PC operating system market.
Third, the entire premise of Novell's franchise applications theory is refuted by
the fact that in the late 1980s and early 1990s there were versions of WordPerfect that ran on
many non-Microsoft operating systems, yet the availability of WordPerfect at that time--when it
was truly popular--on these other operating systems in no way caused them to become popular
or diminished Microsoft's large share of the PC operating system market. Rather, Microsoft
maintained at least a 90% share of the market throughout the relevant period. (E.g., Finding of
Fact 35 ("Every year for the last decade, Microsoft's share of the market for Intel-compatible PC
operating systems has stood above ninety percent."); see also Noll, Nov. 15 Trial Tr. at 1929-30
& Microsoft's Demonstrative 120, shown at trial on Nov. 15, Trial Tr. at 1930, Holley Decl. Ex.
M; Murphy, Dec. 7 Trial Tr. at 4722-23 & Microsoft's Demonstrative 301, shown at trial on
Dec. 7, Trial Tr. at 4722, Holley Decl. Ex. N.)
According to Harral, Gibb and Ford, in the early 1990s, WordPerfect wrote its
word processing software for a number of operating systems, including DR DOS, Macintosh,
OS/2, Unix, Amiga and VMS. (Harral, Oct. 20 Trial Tr. at 204; Gibb, Oct. 26 Trial Tr. at 775-76; Ford, Nov. 30 Trial Tr. at 3670.) By 1994, however, Novell began to focus development
exclusively on Windows. This was because Novell was very late in developing for Windows,
see pp. 39-40, supra, and because Novell fully understood by that time that the huge majority of
end-users were choosing Windows over other operating systems. (See, e.g., PX 599A, at Table
13 (showing that about 93% of all new shipments of word processors in 1994 in the PC operating
system market were shipments of word processors for the Windows platform).)
-67-
A Novell business plan dated August 3, 1994 stated that "all resources need to be
applied to Chicago" and that, as a result, Novell should "[r]educe resources on WordPerfect for
Macintosh and WordPerfect for Unix," and "phas[e] out WordPerfect for VMS." (DX 4, at 2, 5.)
Similarly, in July 1994, a formal business review by Novell recommended that Novell "reduc[e]
the number of developers from 30 to approximately 10" on the Unix platform and "[t]hat these
developers can be moved to Chicago and Tapestry to increase resources in these areas."49
(DX 326, Business Applications Business Unit, Business Review Exercise Summary, at NOV-25-006589.) With respect to Novell's software development efforts for OpenVMS, the same
document stated that "[t]he recommendation is to cancel 6.0 development immediately. This
will free up developers to move to Chicago and Tapestry." (Id.)
As a result, development efforts for non-Microsoft platforms were significantly
reduced or eliminated. A July 21, 1994 memorandum from Ad Rietveld (Executive Vice
President of the Novell Applications Group) to Frankenberg urged that Novell should "[c]lose
the VAX OpenVMS business; Evaluate Unix business based upon strategic importance to the
greater Novell." (DX 5, at NOV 00542198.) A 1995 Novell memorandum to "Bob" showed
that Frankenberg accepted these recommendations--it stated that in "August 94 . . . the feedback
from you appeared to be loud and clear--cut everything but Windows. Based on that, we started
outsourcing platforms--we outsourced WordPerfect for VMS totally and have WordPerfect for
Unix development outsourced . . . ." (DX 227, at NOV-B00642501 (emphasis added).) In
addition, notes from a Novell conference on June 13, 1995 state that "We've had to chop
everything that was having any negative affect on shipping Storm [PerfectOffice for Windows
-68-
95] ASAP. This caused us to stop the OS/2 integration disk work for 3.0." (DX 272, at 4.) As
Ford explained, WordPerfect "discontinue[d] working on new development and shipping new
products on [non-Microsoft] platforms" and focused its resources on Windows. (Ford, Nov. 30
Trial Tr. at 3671.)
In sum, as the Court observed at trial, the notion that the availability of Novell's
applications on rival operating systems would have popularized those operating systems "is
counterfactual," because WordPerfect had been available "on other operating systems since time
immemorial," but failed to popularize them. (Nov. 9 Trial Tr. at 1501.) Novell's franchise
applications theory is refuted by the historical evidence that WordPerfect's availability on other
operating systems did nothing to diminish Microsoft's market share.
2. Novell's Software Lacked All Three Required Elements of Middleware.
The evidence also established that WordPerfect--even combined with the
PerfectFit technology in PerfectOffice and AppWare (see Novell's Memorandum Regarding
Proposed Final Jury Instructions and Verdict Forms, Dec. 5, 2011, Dkt. #336, at 4)50--was in no
way a species of middleware that could possibly have had any impact on competition in the PC
operating system market. As Noll agreed and/or as required by the binding Findings of Fact,
only software that has all of the following three defining characteristics can even in theory
impact competition in the PC operating system market: the software must (1) be cross-platform
in the sense that it runs on multiple PC operating systems (e.g., Noll, Nov. 15 Trial Tr. at 1925-26); (2) be available on "all or nearly all PCs" of the "dominant operating system,"
-69-
i.e., Microsoft Windows (e.g., Noll, Nov. 15 Trial Tr. at 1923-26); and (3) expose a sufficiently
broad set of APIs to enable ISVs profitably to develop full-featured personal productivity
applications that rely solely on those APIs exposed by the middleware (e.g., Findings of Fact 28,
68, 74).51
(a) Novell's Software Was Not Cross-Platform.
It was undisputed at trial that to be middleware in the sense that it might impact
competition in the PC operating system market, middleware must be cross-platform--it must
expose the same set of APIs on different operating systems.52 As Professor Noll acknowledged,
in order to be able to serve as a vehicle for applications to run on multiple platforms and thus
"become a threat to the applications barrier to entry," the software "has to be available on a
number of alternative operating systems." (Noll, Nov. 15 Trial Tr. 1925-26; see also Noll, Nov.
14 Trial Tr. at 1717-18 (a "middleware" product "provid[es] the opportunity to run that
particular application or middleware product on numerous operating systems"); Murphy, Dec. 7
Trial Tr. at 4775 ("if it's going to enhance competition, [middleware must] run on some other
platform in the relevant market, some other operating system that's in the x86 marketplace").)
-70-
The versions of PerfectOffice, WordPerfect and Quattro Pro that purportedly
posed a threat to Microsoft's monopoly were not cross-platform--they were developed to run
solely on Windows 95. During the period from December 1994 (when Novell released
PerfectOffice 3.0 for Windows 3.1) until March 1996--when Novell sold to Corel--Novell was
not developing a version of PerfectOffice for any operating system other than Windows 95.
(Frankenberg, Nov. 8 Trial Tr. at 1168-69; Gibb, Oct. 26 Trial Tr. at 787.) Frankenberg also
acknowledged that, to his knowledge, after Corel purchased WordPerfect and Quattro Pro from
Novell in March 1996, "Corel never released any version of PerfectOffice for any other platform
except Windows." (Frankenberg, Nov. 8 Trial Tr. at 1169.) And, upon questioning by the
Court, Harral agreed that WordPerfect was going to run only on Windows 95:
Q. THE COURT: In fact, no matter what happened, you were
trying to connect WordPerfect and whatever it exposed in terms of
its own APIs or everything else, it was going to be operating on the
Windows 95 operating system?
A. Yes.
Q. No matter what happened. And if somebody could use what it
exposed in terms of APIs and use them eventually as whatever, it
was still going to be operating on the basis of the Windows 95?
A. Yes.
(Harral, Oct. 24 Trial Tr. at 559-60.) Professor Noll testified that in 1994 and 1995, Novell was
"devoting virtually all of their energy to being on Windows 95," and offered his vague
understanding (with no particulars) that Novell "had plans to develop it for other platforms" at
some unspecified point in the future. (Noll, Nov. 14 Trial Tr. at 1845-46.) Because Novell's
products were on Windows only, they did not expose "the same set of APIs on different
-71-
operating system platforms" and do not satisfy the first requirement of middleware. (Murphy,
Dec. 7 Trial Tr. at 4779.)53
Moreover, Novell's software could not have been
cross-platform in the but-for
world (where PerfectOffice utilized the namespace extension APIs) because the namespace
extension APIs were unique to Windows 95. Novell's counsel "agree[d]" there is no evidence
that Novell's technology "could have been easily ported to another platform" if it had utilized the
namespace extension APIs. (Nov. 15 Trial Tr. at 2060.) In addition, Alepin conceded that no
operating system other than Windows 95 exposed the same functionality as the namespace
extension APIs, because those APIs were "platform specific" to Windows. (Alepin, Nov. 9 Trial
Tr. at 1482-83; Alepin, Nov. 10 Trial Tr. at 1532-33.) Microsoft's experts testified to the same
effect. (Bennett, Dec. 12 Trial Tr. at 5023 (the "NameSpace extension APIs . . . [were] a unique
component of Windows 95"); see also Murphy, Dec. 7 Trial Tr. at 4783-84.) As a result, had
Novell utilized the namespace extension APIs to create its file open dialog, Novell's Three
Products would have become even more tightly tied to Windows, and thus more difficult to port
to non-Microsoft operating systems in the future. This destroys the entire Middleware Theory.
-72-
(b) Novell's Software Was Not Available on All or Nearly All PCs.
As Professor Noll explained, in order to "imperil" the applications barrier to entry
protecting Microsoft's monopoly, Novell's software had to run on "all or nearly all" PCs of the
"dominant operating system" (Noll, Nov. 15 Trial Tr. at 1923; see also id. at 1925-26), which
Professor Noll defined as "the category of Microsoft or Microsoft compatible operating systems"
(id. at 1924-26). Indeed, this requirement is a matter of logic because no software could possibly
serve as an attractive middleware platform to which ISVs would choose to write unless by so
doing, ISVs would reach all or nearly all of the end-users that could be reached by writing
directly to Windows. (See Murphy, Dec. 7 Trial Tr. at 4786-87 (explaining that "being available
on nearly all PCs is really important" in order to "provide[] a . . . good enough environment for
ISVs to evolve and to start attracting developers away from Windows").)
As shown above, by the time Windows 95 was released in August 1995, the
market share of Novell's products was small and on a downward trajectory. In 1995,
WordPerfect had 16% of the Windows market, PerfectOffice had 3.6% and Quattro Pro had 2%.
See pp. 63-66, supra. Moreover, for purposes of whether Novell's software was present on all or
nearly all Microsoft PCs, these low market share numbers must be further reduced by about 50%
because office suites or any of their component applications were installed on only half of all
PCs. (Murphy, Dec. 7 Trial Tr. at 4750, 4788-89.)
Thus, because Novell's software was available on far fewer than all or nearly all
PCs, an ISV "would be limiting his market enormously" by writing to the APIs exposed by
Novell's products instead of the APIs exposed by Windows (because the number of end-users
with Windows was several times the number of end-users with Novell's products). (Murphy,
Dec. 7 Trial Tr. at 4789-90.) That makes it entirely illogical to posit a world where Novell's
products become middleware in the sense that they could supplant Windows.
-73-
Even accepting Professor Noll's opinion that the relevant inquiry is the
percentage of the installed base of all PCs with WordPerfect (including DOS versions of
WordPerfect), the result is no different. According to Professor Noll, "WordPerfect still ha[d]
about half of the install[ed] base in 1995" if one includes DOS versions of WordPerfect. (Noll,
Nov. 14 Trial Tr. at 1762.) This does not help Novell's position.
First, Noll offered no data to support this 50%, and, as shown below, it is far too
high. In any event, that proportion is far from "all or nearly all" PCs. Second, as Professor
Murphy testified, only about half of all PCs then had an office suite or any of its component
applications installed on them. (Murphy, Dec. 7 Trial Tr. at 4750, 4788-89.) This places
Novell's installed base at 25% of all PCs. Third, Professor Noll's 50% estimate does not
comport with the actual data Novell introduced at trial, which established that, as of the end of
1994, WordPerfect had about 36% of the installed base of all PCs with word processors when
one includes WordPerfect's share of the DOS platform. (PX 599A, at Table 13;54 see also
Acheson, Dec. 2 Trial Tr. at 3993-94.) In other words, at the end of 1994, only about 18% of all
PCs had a copy of WordPerfect installed on them. (Murphy, Dec. 7 Trial Tr. at 4788-89.) By
any measure, this falls far short of the requirement that it be available on all or nearly all PCs.
In any event, WordPerfect's installed base on DOS is irrelevant for purposes of
Novell's Middleware Theory.55 Professor Noll conceded that his estimation of WordPerfect's
-74-
installed base was based on WordPerfect's historic success on the character-based DOS
platform. (Noll, Nov. 15 Trial Tr. at 1923-25; see also Hubbard, Dec. 6 Trial Tr. at 4442-43
("WordPerfect has a . . . very significant share of DOS, the older platform, but a smaller share,
and actually declining share in the Windows 3.x and then subsequently Windows 95, 98.").)
There was and is no dispute that Novell's historic success on the DOS platform would not and
did not translate into success on Microsoft Windows. An internal Novell memorandum dated
April 14, 1995 and titled "Market Requirements Document for `Storm'" explained:
WordPerfect for DOS possesses the single largest user installed
base in the word processing market. WordPerfect currently claims
that 10 Million users or 30% of the total word processor market are
WordPerfect for DOS users. Currently, only 30% of this
WordPerfect for DOS installed base is remaining with
WordPerfect as they transition to a Windows word processor.
(DX 224, at 20.) For a Middleware Theory dependent on the notion that users would switch
operating systems because of Novell's ubiquitous software, the presence of WordPerfect on DOS
cannot logically make any difference.
Moreover, the version of the WordPerfect word processor written to DOS did not
even include the Novell software that was purportedly middleware. (Murphy, Dec. 8 Trial Tr. at
4917-18; see also Noll, Nov. 15 Trial Tr. at 1946-47 (conceding that he was unaware of any
evidence that any of the purported middleware capability in PerfectFit was available on
WordPerfect for DOS).) As a result, Novell's installed base on DOS is irrelevant for purposes of
Novell's Middleware Theory. (See Murphy, Dec. 7 Trial Tr. at 4788.)
(c) Novell's Software Did Not Expose Sufficient APIs to Allow ISVs to
Write General-Purpose Personal Productivity Applications.
For software to affect competition in the PC operating system market, it must
expose sufficient APIs such that general-purpose personal productivity applications could be
written solely to those APIs rather than the APIs exposed by Windows. Finding of Fact 28
-75-
makes this clear: only software that "exposes enough APIs to allow independent software
vendors (`ISVs') profitably to write full-featured personal productivity applications that rely
solely on those APIs" can pose a threat to Microsoft. Indeed, Finding of Fact 74 explains that
the threat (there, merely a "nascent" threat) that Sun's Java technology posed to Microsoft's PC
operating system monopoly stemmed from Java's intended availability on non-Microsoft
operating systems:
The inventors of Java at Sun Microsystems intended the
technology to enable applications written in the Java language to
run on a variety of platforms with minimal porting. A program
written in Java and relying only on APIs exposed by the Java class
libraries will run on any PC system containing a JVM that has
itself been ported to the resident operating system.
(See also Finding of Fact 68 (applications must be written "exclusively on middleware APIs" to
be able to "run . . . on any operating system hosting the requisite middleware").) The Court
instructed the jury that these Findings have "binding effect here" as the Court instructed the jury
(Oct. 18 Trial Tr. at 143), and Novell conceded the same thing (e.g., Nov. 14 Trial Tr. at 1709).
The evidence was undisputed that Novell's software lacked this defining
characteristic because ISVs could not write general-purpose personal productivity applications
that would run on top of Novell's software. Alepin conceded that Novell's purported
middleware (including WordPerfect, AppWare, OpenDoc and PerfectFit or some combination of
those products and technologies) did not expose a sufficiently broad set of APIs to enable
development of general-purpose personal productivity applications. (Alepin, Nov. 9 Trial Tr. at
1489-90; see also Alepin, Nov. 10 Trial Tr. at 1533-35, 1538-40.) The APIs exposed by
Novell's Three Products could only support applications "that worked with and were
compl[e]mentary to the WordPerfect system," such as a thesaurus application or a spell-checker
application. (Alepin, Nov. 9 Trial Tr. at 1479.) Alepin admitted that no ISV would even attempt
-76-
to write a general-purpose personal productivity application on top of WordPerfect--this "would
not be the best use of [an ISV's] time." (Id. at 1480.) Professor Noll agreed, explaining that
there were no third-party applications "that would threaten the applications barrier to entry that
were written to this platform, PerfectFit and/or appware." (Noll, Nov. 15 Trial Tr. at 1922-23;
see also Murphy, Dec. 7 Trial Tr. at 4729, 4789-90 (same).)
It is undisputed that ISVs could not write general-purpose personal productivity
applications to the APIs exposed by Novell's products. Indeed, Novell conceded that, if the jury
had been instructed that Novell's software must "expose[] enough APIs to allow independent
software vendors (ISVs) profitably to write full-featured personal productivity applications that
rely solely on those APIs," as stated in Finding of Fact 28, that "would be directing a verdict on
that portion of our theory." (Dec. 15 Trial Tr. at 5436-37, 5439.)
Perhaps in recognition of the fact that Novell could not prevail if the criteria of
Finding 28 is applied, Noll sought to water down the requirement at trial, contending that
software need not meet the middleware definition of the Government Case to constitute a threat
to the applications barrier to entry. Noll testified:
[I]f the middleware is exposing a certain number of API's, you can
write to those API's and be on--and access functions in multiple
operating systems. You may have to write additional code
separately for each operating system, but if the middleware reduces
the amount of code you have to write to be cross-platform, then it
makes being cross-platform more attractive. . . . Middleware can
begin to have an effect on competition in the operating system
market if it starts to be used because it's reducing porting costs
and, therefore, increasing the number of applications that are cross-platform, and thereby reducing the applications barrier to entry.
(Noll, Nov. 15 Trial Tr. at 1958-59.) This fails for several reasons.
First, Professor Noll conceded, as he must, that "the whole point of middleware is
to free the software vendor, the applications developer, from using any particular operating
-77-
system's set of API's" because the "middleware provider perform[s] the function of accessing
the operating system and [the ISV] just us[es] the applications programming interfaces of the
middleware product itself as a way to gain access to all operating systems." (Noll, Nov. 14 Trial
Tr. at 1736.) Thus, Novell's watered-down version of middleware failed to comport with
Professor Noll's own understanding of what middleware must accomplish in order to be able to
impact competition in the PC operating system market--if applications must rely on even in part
the Windows APIs, Windows remains essential and its monopoly position will not be eroded.
Moreover, the notion underlying Professor Noll's testimony that reducing the incremental
porting costs of applications would be sufficient to meet the API requirement is illogical: If an
application relies on a large number of Windows APIs as well as some PerfectOffice APIs, it
could only be used on the few Windows PCs that had PerfectOffice installed, and significant
porting costs would be incurred for that software to run on other operating systems. This could
not result in increased competition in the PC operating system market.
Rather, as Professor Murphy explained (consistent with the theory of the
Government Case), the only middleware that could impact competition is middleware that allows
an ISV "to write just to the APIs of middleware and not to the operating systems" and that also
exists "on another operating system" because if an ISV is not using the APIs of the underlying
operating system, then "more applications [may become] available for those other operating
systems." (Murphy, Dec. 7 Trial Tr. at 4772-73.) Because the evidence at trial established that
Novell's software did not expose sufficient APIs "to free the software vendor . . . from using any
particular operating system's set of APIs" (Noll, Nov. 14 Trial Tr. at 1736), Novell's software
had no ability to erode the applications barrier to entry.
-78-
Second, Paragraphs 44 and 48 of Novell's own Complaint adopted the definition
of middleware used in the Government Case, and in Finding of Fact 28 in particular:
44. As the U.S. Court of Appeals for the District of Columbia Circuit held
in affirming the district court's essential findings, Microsoft's Windows
monopoly was threatened by "middleware" such as Netscape's Navigator,
which is a browser application, and Sun Microsystems' implementation of
the "Java" technologies, both of which were not only able to function on
multiple operating systems, but were potentially able to provide platforms
for end-use applications, which made them a threat to replace Windows
itself as such a platform.
48. The District Court defined middleware as software that "relies on the
interfaces provided by the underlying operating system while
simultaneously exposing its own APIs to developers." Findings of Fact
§ 28. In the Government Suit, Netscape, when coupled with Java, is
described as having "the potential" to become a middleware platform on
which applications could be written to run on multiple operating systems.
Such cross-platform functionality undermines the applications barrier to
entry that helps protect Microsoft's operating system dominance.
(Compl. ¶¶ 44, 48 (emphasis added).) In fact, the Complaint described the alleged middleware
threat posed by AppWare as "a serious threat to Microsoft" specifically because "[p]rogrammers
could write programs using these APIs that could function on any AppWare installation
regardless of the operating system" and by "[w]riting to the AppWare APIs and not to the
Windows APIs would enable applications to run not only on Windows, but also on Macintosh
and other operating systems at no additional cost." (Compl. ¶ 50 (emphasis added).)56 Because
Novell's software did not possess the essential characteristic of exposing APIs that would permit
-79-
ISVs to write full-featured personal productivity applications solely to the APIs exposed by that
software, it could not have had any impact on competition in the PC operating system market.
Moreover, to the extent that Novell seeks to depart from the definition of
middleware used in the Government Case and incorporated into the Complaint, Novell released
any such claim in 2004. In the 2004 settlement agreement, Novell released Microsoft from "any
and all Claims that Novell ever had or has as of the date of this Agreement in law or in equity,
known or unknown, of any kind whatsoever (including without limitation any antitrust or similar
Claims of any kind) except for . . . the Claims set forth in the draft WordPerfect complaint . . . ."
(Nov. 8, 2004 Settlement Agreement, at § 2(a), Holley Decl. Ex. A.) See pp. 126-30, infra.
Because Novell's software possessed none of the three characteristics of
middleware required to have the potential to impact competition in the PC operating system
market, Novell failed to show any impact on competition in that market.
To summarize this Section:
- First, it is elementary and undisputed that, in order for middleware to have any potential to
increase competition to Windows, the software must be cross-platform. But Novell's
PerfectOffice suite containing the Perfect Fit purported technology (the alleged middleware)
was developed only for Windows. See pp. 70-72, supra.
- Second, it is also undisputed that, in order to have any potential to increase competition in the
PC operating system market, the cross-platform middleware must be present on "all or nearly
all" PCs of the dominant operating system. Otherwise, ISVs would have no incentive to
write their applications to middleware (and limit their potential customer base) when they
could write their applications to Windows and gain access to users of about 95% of all PCs.
In 1995, WordPerfect had 16% of the Windows market, Quattro Pro had 2% and
PerfectOffice 3.0 captured at most 8% of the suite market. Professor Noll testified that had
PerfectOffice been released on time, it would have achieved a similar market share as the
prior version achieved on Windows 3.1. See p. 66, supra. Moreover, even using
WordPerfect's 36% installed base of PCs that had a word processor installed (as of the end of
1994), that low number does not come close to "all or nearly all" PCs (and is only 18% of all
PCs because half of all PCs had no suite or word processor on them at all).
- Third, no software could pose a middleware threat to Windows unless it exposed enough
APIs to enable ISVs to write general-purpose personal productivity applications to its APIs
-80-
because, if ISVs must rely on the Windows APIs even in part, their applications would run
only on Windows. If an application relies on APIs exposed by both Windows and
PerfectOffice, then that application could be used only on the few Windows PCs that had
PerfectOffice installed, and significant porting costs would be incurred for that software to be
able to run on any other operating system. This could not conceivably reduce the
applications barrier to entry protecting Windows' monopoly.
* * *
In addition, and independently, Novell's two theories about the potential impact
on competition in the PC operating system market assume the existence of a viable alternative to
Windows during the relevant period. This assumption is critical, for in the absence of such a
viable alternative, Novell's products--no matter how popular or cross-platform and regardless of
whether they were or could have been middleware as that term is used in the Government
Case--could not have had the necessary impact on competition in the adjacent market for PC
operating systems. "[I]n order to make [Novell's theories] work," there must be "other operating
systems that [were] sufficiently attractive" to convince consumers in the "but for world to move
from Windows to an alternative operating system." (Murphy, Dec. 7 Trial Tr. at 4735.)
There was no evidence, however, that effective operating system competitors
existed in 1995. (Murphy, Dec. 7 Trial Tr. at 4735-36, 4763.) The evidence at trial was to the
contrary. Professor Noll admitted that Linux "wasn't really a competitor" to Windows during
the relevant period, because it only "became a full-fledged, commercial product" in 1996. (Noll,
Nov. 15 Trial Tr. at 1903, 1961.) Similarly, the version of OS/2 that IBM released in 1995 was,
according to Noll, "not an effective competitor" to Windows. (Id. at 1903.) In fact, during the
time that Novell owned WordPerfect and Quattro Pro, "there were no real strong competitors out
there to take the business away from Windows." (Murphy, Dec. 7 Trial Tr. at 4735; see also
Gates, Nov. 22 Trial Tr. at 3114-15 (same).) Even the combined market share of Unix, Linux,
-81-
and OS/2 together was "very small relative to Windows." (Noll, Nov. 14 Trial Tr. at 1781.)57
The lack of even a single meaningful competitor to Windows in the relevant time period is also
fatal to Novell's theories of harm to competition.
B. The Applicable Causation Standard Is Whether Microsoft's Withdrawal of
Support for the Namespace Extension APIs "Contributed Significantly" to
Maintenance of Microsoft's Monopoly in the PC Operating System Market,
and Novell Came Nowhere Close to Meeting that Standard.
As this Court held in 2010, to establish causation, Novell had to prove that
Microsoft's October 3 Decision "contribut[ed] significantly to [Microsoft's] continued monopoly
power." 699 F. Supp. 2d at 747-48. Despite this holding, Novell contended at trial that it only
needed to show that the October 3 Decision was "reasonably capable of contributing
significantly" to Microsoft's monopoly power.
The causation standard Novell sought, however, applies only to claims for
injunctive relief in enforcement actions brought by the Government, not in private actions
seeking treble damages. In the Government Case, the D.C. Circuit expressly recognized that it
was applying an "edentulous test for causation" and held that, as a result, the Government was
not required to show that "Java or Navigator would actually have developed into viable platform
substitutes." United States v. Microsoft Corp., 253 F.3d at 79. There, the Government needed to
show only that "the exclusion of nascent threats is the type of conduct that is reasonably capable
of contributing significantly to a defendant's continued monopoly power," and that Java and
Netscape were such "nascent threats at the time Microsoft engaged in the anticompetitive
-82-
conduct" alleged. Id. The D.C. Circuit "found a causal connection between Microsoft's
exclusionary conduct and its continuing position in the operating system market only through
inference," and "expressly did not adopt the position that Microsoft would have lost its position
in the [PC operating system] market but for its anticompetitive behavior." Id. at 106-07.
Where, as here, a party seeks treble damages, two-thirds of the amount sought is
entirely punitive and, accordingly, no such inference of causation is permitted because it is
"critical that treble damage remedies be strictly limited to those aspects of a plaintiff's injury that
were in fact caused by an unlawful exploitation of market power or an unlawful quest for such
power in attempt cases." 3 PHILLIP E. AREEDA & HERBERT HOVENKAMP, ANTITRUST LAW
§ 657a; see also United States v. Microsoft Corp., 231 F. Supp. 2d 144, 164 (D.D.C. 2002) (there
must be "proportionality between the severity of the remedy and the strength of the evidence of
the causal connection"). Indeed, as the Areeda treatise makes clear, a more stringent causation
requirement applies to actions for treble damages because "antitrust's mandatory treble damage
provision often awards damages that are `excessive,' in that the punitive two-thirds is assessed
even to actions that are not criminal in character, or where the defendant had made a reasonable
but mistaken judgment that it was doing nothing unlawful." 3 PHILLIP E. AREEDA & HERBERT
HOVENKAMP, ANTITRUST LAW § 657a. This, when coupled with the fact that "the line between
aggressive competitiveness and a § 2 violation is indistinct," makes "judicial judgments and
particularly juries prone to error." Id. Use of the "contributes significantly" standard--especially here, where the alleged harm to competition in the PC operating system market is
based on an attenuated cross-market theory--is therefore necessary to "moderate the treble
damage consequences of finding `exclusionary' conduct" to "be strictly limited to those aspects
-83-
of a plaintiff's injury that were in fact caused by an unlawful exploitation of market power . . . ."
Id. The edentulous causation standard provides no such assurance.
There was no basis for the jury to find that any delay of the release of Novell's
Three Products contributed significantly to Microsoft's PC operating system monopoly. With
respect to Novell's franchise applications theory, the evidence demonstrated that (a) the
applications barrier to entry was due to the very large and diverse population of applications
written to Windows, and (b) PerfectOffice, WordPerfect and Quattro Pro, which had low market
shares on Windows, were not even close to being popular enough that--if and when Novell
decided to develop and release a version of PerfectOffice for a rival operating system--users
would migrate to that other operating system. See pp. 60-69, supra. Novell did not come
anywhere close to establishing that the October 3 Decision contributed significantly to the
maintenance of Microsoft's PC operating system monopoly.
Likewise, the evidence demonstrated that Novell's software possessed none of the
three defining characteristics of middleware that could have had even the potential to impact
competition in the PC operating system market, and therefore the supposed delay in the release
of Novell's Three Products could not have "contributed significantly" to the maintenance of
Microsoft's monopoly. See pp. 69-81, supra. Moreover, the binding Findings of Fact make
clear that as of 1999--five years after Microsoft's decision to withdraw support for the
namespace extension APIs--there was no product yet in existence of the sort that Novell
hypothesized, i.e., a middleware product that allowed ISVs to write applications to the APIs
exposed by the middleware.58
-84-
Indeed, Professor Noll never opined that the withdrawal of support for the
namespace extension APIs had any substantial impact on competition in the PC operating system
market. Instead, he testified that Novell was merely making an "attempt" to be cross-platform,
which would at some later (and undetermined) point in the future give users "the ability to switch
platforms" and which theoretically "can have the effect of increasing competition." (Noll, Nov.
14 Trial Tr. at 1717-18, 1765-66; see also Noll, Nov. 15 Trial Tr. at 1924-25 ("[I]f you can retain
that install base, and then if you're the largest single entity in that install base you have a natural
advantage, although there are switching costs, and you can, if you then become middleware,
increase competition in the operating system market.").) Noll's vague and wholly theoretical
testimony, which included no attempt to quantify any supposed impact on competition in the PC
operating system market or to state when such an impact might occur, is not a legally sufficient
evidentiary basis for a reasonable jury to find that the October 3 Decision had any impact on
competition in the PC operating system market under any causation standard.
Professor Noll not only failed to opine that there was some measurable and non-trivial harm to competition in the PC operating system market caused by the October 3 Decision,
but he also conceded that the withdrawal of support for the namespace extension APIs by itself
"would not have affected competition in the operating system market." (Noll, Nov. 15 Trial Tr.
-85-
at 1905-06 (emphasis added).) Rather, Professor Noll testified that any harm to competition in
the PC operating system market came about only in combination with a set of different Microsoft
acts that harmed OS/2, Netscape and Java. (See id. at 1906-08.) Noll acknowledged, however,
that "in a world in which all the other people are still operating and competing, then the loss of
Novell would not have been a significant factor in effecting [sic] competition in the operating
system market." (Id. at 1907.)
This testimony is fatal to Novell's claim. The Court has been consistent and clear
that "Novell cannot piggyback on the anticompetitive harm caused by conduct directed at third
parties without actually showing the conduct which injured its applications had an
anticompetitive impact as well." 699 F. Supp. 2d at 750. (See also Nov. 14 Trial Tr. at 1815-16
("THE COURT: . . . Obviously Novell's case rises and falls with conduct directed by Microsoft
against Novell . . . .").) But this is precisely what Professor Noll's testimony requires--the
combination of alleged harm to Novell with alleged harm to other competitors. (See Noll,
Nov. 15 Trial Tr. at 1905-08.) Professor Noll's theory is thus nothing more than impermissible
"piggybacking" and did not provide Novell with the evidence necessary to demonstrate that the
sole act about which Novell complains--the October 3 Decision--"contributed significantly" to
the maintenance of Microsoft's PC operating systems monopoly or satisfied even a lesser
causation standard.59
-86-
C. The Allegedly Wrongful Conduct Could Not Have Harmed Competition in
the PC Operating System Market Because the Evidence Showed that the
Timely Release of PerfectOffice Would Have Enhanced Microsoft's
Monopoly.
In addition to the defects shown above, a reasonable jury would not have a legally
sufficient evidentiary basis to find that Microsoft's October 3 Decision harmed competition in
the PC operating system market because in the but-for world Novell would have utilized the
namespace extension APIs to enhance Windows 95 and thereby make it even more popular.
First, all of Novell's fact witnesses testified that, had there been no delay in the
release of versions of Novell's products for Windows 95, Windows 95 would only have become
stronger. Notably, Frankenberg testified that the market share of Windows 95 would have
increased in that situation:
Q. Was it your view at the time, in 1994 and 1995, that if
PerfectOffice, the new version of PerfectOffice for Windows 95
had been released by Novell, that that would have made Windows
95 even more desirable in the marketplace than it otherwise would
have been?
A. Definitely. It would have made Windows 95 more desirable in
the marketplace.
Q. It was your view at the time that if PerfectOffice for Windows
95 had been released by Novell, that would have been a benefit to
Microsoft for exactly the reason that you just said, it would have
made Windows 95 even more desirable for consumers?
A. That is true.
Q. If --
A. Especially those who used WordPerfect products. They would
be able to use Windows 95, and they wouldn't otherwise have been
able to do that if they wanted to continue using WordPerfect.
Q. If anything, that would increase the sales of Windows 95,
correct?
A. Yes.
-87-
Q. Having a good PerfectOffice product out there would make
Windows 95 even more popular than it turned out to be; true?
A. True.
Q. If PerfectOffice had been released in 1995 by Novell and had
been successful, and had gained a reasonably good share of the
market how, if at all, would that have effected [sic] sales of
Windows?
A. Presumedly [sic] it would have increased sales of Windows 95.
Q. And would have made Windows 95's market share even higher
than what it turned out to be, correct?
A. Yes.
(Frankenberg, Nov. 8 Trial Tr. at 1226-28 (emphasis added); see also Noll, Nov. 15 Trial Tr. at
1949-50 ("completely agree" that Frankenberg was "exactly right" that a timely release of
PerfectOffice would have increased Microsoft's share of the PC operating system market).) This
testimony by the CEO--that Windows 95 would have been "more desirable" and that its sales
would have "increased" if Novell had been able to use the namespace extension APIs--is fatal to
Novell's claim. It affirmatively disproves the assertion that the allegedly wrongful act had some
adverse effect on competition in the relevant market. In the but-for world, Microsoft's monopoly
power would have increased.
Novell's witnesses also established that Windows 95 was a huge step forward
technologically, and that Novell wanted to tie its products as closely as possible to Windows 95.
Frankenberg testified that Windows 95 was a "significant step forward" and that Novell was
"very excited and very interested" in it. (Frankenberg, Nov. 8 Trial Tr. at 1225-26.)
Frankenberg also explained that Novell's business strategy was to "take[] advantage of the
capabilities in Windows 95." (Id. at 1226.) Harral likewise testified that "Windows 95 was in
-88-
my view a significant step forward for the P.C." and that Novell was "excited about Windows
95" and the "wonderful evolution" in technology it provided. (Harral, Oct. 20 Trial Tr. at 253-54, 256-57; see also Gibb, Oct. 26 Trial Tr. at 788 ("Well, from a technology standpoint,
Windows 95 was a huge step forward."); Richardson, Oct. 25 Trial Tr. at 607 ("There were many
features in Windows 95 that we were very excited about."); Noll, Nov. 15 Trial Tr. at 1911
(Windows 95 was a "substantial step forward.").)
Harral and Richardson both testified that Novell's use of the namespace extension
APIs would have enhanced Windows 95. They planned to use the namespace extension APIs to
put five other Novell products (Novell's QuickFinder search engine, Soft Solutions document
management system, e-mail client, Presentations clip-art gallery, and a primitive FTP/HTTP
browser) in the tree view of the Windows Explorer so that these products would appear in the
Windows 95 Explorer and common file open dialogs once a user had installed Novell's Three
Products. (Harral, Oct. 20, Trial Tr. at 268-70; Harral, Oct. 24 Trial Tr. at 373-74, 515;
Richardson, Oct. 25 Trial Tr. at 592-93, 612, 629-30, 638, 691-92.) Adding these five Novell
products to the Windows 95 shell would have acted to "make Windows [95] the best version of
Windows that it could be." (Harral, Oct. 24 Trial Tr. at 372-74; see also Richardson, Oct. 25
Trial Tr. at 613 ("It was our intent to make the user's experience on Windows better because
they had WordPerfect installed.").)
Given that Windows 95 would have become a better and more desirable product
had Novell used the namespace extension APIs, Windows' market share would have been even
higher in the but-for world than it otherwise was. (Frankenberg, Nov. 8 Trial Tr. at 1226-28;
Noll, Nov. 15 Trial Tr. at 1949; Murphy, Dec. 7 Trial Tr. at 4797-98.) As the Court has
emphasized, the evidence showed that Novell "wanted to marry the two products, the operating
-89-
system and WordPerfect . . . both through 1996 and the foreseeable future." (Oct. 27 Trial Tr. at
928-29.) This disproves an element of Novell's claim--that "the specific Microsoft conduct
which caused injury to Novell's applications also caused anticompetitive harm in the PC
operating system market," 699 F. Supp. 2d at 748, under any causation standard.
II. A Reasonable Jury Would Not Have a Legally Sufficient Evidentiary Basis to Find
that Microsoft Engaged in Anticompetitive Conduct.
A. Microsoft's Decision to Withdraw Support for the Namespace Extension
APIs Does Not Fall Within the Limited Aspen Skiing Exception.
Novell concedes, as it must, that "Microsoft doesn't have a duty to provide us
with anything." (Nov. 18 Trial Tr. at 2587.) As Novell's counsel told the court, "I don't think
under the antitrust laws, certainly as we have them today, that that [i.e., refusing to provide
Novell with the namespace extension APIs] would have been actionable." (Id. at 2597-98.)
Nevertheless, according to Novell, having made the namespace extension APIs available to ISVs
in the June 10, 1994 beta version of Windows 95, Microsoft was obligated to continue
supporting them because once Microsoft "evangelize[d]" them "to us" and Novell supposedly
planned to use those APIs, Microsoft was precluded from withdrawing support for the APIs. (Id.
at 2599.) This claim is completely untenable as a matter of law.
As the Court explained in its March 2010 summary judgment decision, to prevail
on its Section 2 claim, Novell was required to prove at trial that Microsoft's October 3 Decision
fell within the limited exception provided in Aspen Skiing Co. v. Aspen Highlands Skiing Corp.,
472 U.S. 585 (1985) to the general rule that "a monopolist generally has a right to refuse to
cooperate with a competitor." 699 F. Supp. 2d at 745.
Of course, "as a general matter, the Sherman Act `does not restrict the long
recognized right of [a] trader or manufacturer engaged in an entirely private business, freely to
-90-
exercise his own independent discretion as to parties with whom he will deal.'" Verizon
Communications, Inc. v. Law Offices of Curtis V. Trinko, LLP, 540 U.S. 398, 408 (2004) (second
alteration in original) (quoting United States v. Colgate & Co., 250 U.S. 300, 307 (1919)). As
the Supreme Court has recognized, "[c]ompelling such firms to share the source of their
advantage is in some tension with the underlying purpose of antitrust law, since it may lessen the
incentive for the monopolist, the rival, or both to invest in those economically beneficial
facilities." Trinko, 540 U.S. at 407-08. This rule applies with equal force to a monopolist. Four
Corners Nephrology Associates, P.C. v. Mercy Medical Center of Durango, 582 F.3d 1216, 1221
(10th Cir. 2009) ("The Supreme Court has recently emphasized the general rule that a business,
even a putative monopolist, has `no antitrust duty to deal with its rivals at all.'") (quoting Pacific
Bell Telephone Co. v. linkLine Communications, Inc., 555 U.S. 438, 444 (2009)).
"[T]he Supreme Court has found a duty to deal in only one limited circumstance,"
Compliance Marketing, Inc. v. Drugtest, Inc., 2010 U.S. Dist. LEXIS 34315, at *53 (D. Colo.
April 7, 2010) (citing Trinko, 540 U.S. at 409); see also Four Corners, 582 F.3d at 1224-25
(quotation omitted), where a defendant "terminates a profitable relationship" with the plaintiff
and does so "without any economic justification," denying its rival terms "available to all other
consumers." Four Corners, 582 F.3d at 1225 (emphasis in original).
This is the so-called Aspen Skiing exception. The Aspen Skiing exception is
narrow "`because of the uncertain virtue of forced sharing and the difficulty of identifying and
remedying anticompetitive conduct by a single firm.'" Christy Sports, 555 F.3d at 1194 (quoting
Trinko, 540 U.S. at 408). Indeed, the Supreme Court has emphasized that "Aspen Skiing is at or
near the outer boundary of § 2 liability." Trinko, 540 U.S. at 409. As a result, "courts should
-91-
impose a duty to deal under Section 2 only `very cautious[ly].'" Four Corners, 582 F.3d at 1225
(quoting Trinko, 540 U.S. at 408).
As an initial matter, it is undisputed that Windows 95 was Microsoft's intellectual
property. Indeed, as Alepin acknowledged, the namespace extension APIs were invented by
Satoshi Nakajima who obtained a U.S. patent covering his invention. (PX 364; see also Alepin,
Nov. 10 Trial Tr. at 1625-26; Nakajima, Dec. 1 Trial Tr. at 3793.) As this Court has recognized,
"to require one company to provide its intellectual property to a competitor would significantly
chill innovation." In re Microsoft Corp. Antitrust Litigation, 274 F. Supp. 2d 743, 745 (D. Md.
2003) (Motz. J.) (citations omitted); see also Daisy Mountain Fire District v. Microsoft Corp.,
547 F. Supp. 2d 475, 489-90 (D. Md. 2008) (Motz. J.). The Court has also recognized that, in
light of the fact that "the software development industry is dynamic and involves continuous
innovation," it would be wrong to require Microsoft to "disclose significant information to its
competitors." In re Microsoft, 274 F. Supp. 2d at 745; see also Four Corners, 582 F.3d at 1221
("Allowing a business to reap the fruits of its investments is an important element of the free-market system: it is what induces risk taking that produces innovation and economic growth."
(quotation and citation omitted)).
Courts have refused to extend the Aspen Skiing exception to require a
technological innovator to provide its intellectual property to its rivals. See In re Independent
Service Organizations Antitrust Litigation, 989 F. Supp. 1131, 1139 (D. Kan. 1997) (pointing out
that "Aspen Skiing did not involve intellectual property rights" and explaining that extending
Aspen Skiing to require the provision of intellectual property to competitors would "seriously
undermin[e] the objectives of the intellectual property laws"); see also In2 Networks, Inc. v.
Honeywell International, 2011 U.S. Dist. LEXIS 117589, at *16 (D. Utah Oct. 12, 2011) (under
-92-
Christy Sports, there is no duty "to allow [a rival] to use [one's intellectual] property, like Deer
Valley was not required to invite competitors onto its property to rent skis") (dicta).
Under Tenth Circuit law, a plaintiff relying on the Aspen Skiing exception must
prove that a monopolist (a) "terminat[ed] a profitable business relationship" with plaintiff, and
(b) did so "without any economic justification." Four Corners, 582 F.3d at 1225. As the Tenth
Circuit stated, the "key fact" is that a "monopolist was willing to jettison a profitable short-term
business relationship and deny to a rival the retail prices available to all other consumers." Id.
(emphasis in original). In addition, for a termination of a business relationship to serve as the
predicate of an Aspen Skiing claim, it must not be "temporary" or "subject to [defendant's]
business judgment" because termination in those circumstances "does not reach the outer
boundary of § 2 liability, at which Aspen Skiing lies." Christy Sports, 555 F.3d at 1197 (internal
citation and quotation omitted).
Microsoft's October 3 Decision was not anticompetitive under these strict
standards, and no reasonable jury would have a legally sufficient evidentiary basis to find
otherwise. First, Microsoft's October 3 Decision did not "terminate" any business relationship
between Microsoft and Novell. See Four Corners, 582 F.3d at 1225 (citation omitted). The
opposite is true. Microsoft continued providing Novell with beta versions of Windows 95,
continued to provide information and assistance to Novell, and did nothing to prevent Novell
from developing versions of its products for Windows 95. See pp. 32-33, supra. There is also
no dispute that Novell could have used the Windows 95 common file open dialog to get its
products released to market in a timely manner. (Gibb, Oct. 26 Trial Tr. at 847-49; Alepin, Nov.
10 Trial Tr. at 1604; Harral, Oct. 24 Trial Tr. at 502; Belfiore, Dec. 5 Trial Tr. at 4264-66;
Bennett, Dec. 12 Trial Tr. at 5019.) No relationship was "terminated," and Novell's 1994
-93-
versions of PerfectOffice, WordPerfect and Quattro Pro remained compatible with Windows 95.
See pp. 13 n. 11 & p. 47, supra.
Second, Microsoft did not deny Novell access to any information "available to all
other consumers." Four Corners, 582 F.3d at 1225 (emphasis in original). On the contrary,
even assuming that Microsoft's decision could be deemed a "termination" at all, the evidence is
clear that Microsoft's decision to withdraw support for the namespace extension APIs applied to
all ISVs.60 Novell was not singled out for disparate treatment.
1. The Withdrawal of Support for the Namespace Extension APIs Did Not
"Terminate" Microsoft's Relationship with Novell.
Harral and Richardson both testified that, after Microsoft's decision to withdraw
support for the namespace extension APIs, Novell had three options. First, if the namespace
extension APIs were as important as Novell contends, Novell could have continued to call those
APIs, just as Richardson said Novell's shared code group had done between June 1994 and
October 1994. (Richardson, Oct. 25 Trial Tr. at 677-78.) Richardson and Harral agreed that the
namespace extension APIs remained in Windows 95, and that Novell could "continue to use the
documentation" it had received from Microsoft in June 1994 with the M6 beta. (Harral, Oct. 20
Trial Tr. at 342; Richardson, Oct. 25 Trial Tr. at 624.)
Second, Novell could have used the Windows 95 common file open dialog, which
Microsoft made available to all ISVs. Gibb testified that it would have been "easy" for Novell to
release WordPerfect and Quattro Pro using that Windows 95 common file open dialog. (Gibb,
Oct. 26 Trial Tr. at 847-48; see also Harral, Oct. 24 Trial Tr. at 502.) He said that Novell
-94-
decided against the easy route because he thought it "could do something cooler." (Gibb, Oct.
26 Trial Tr. at 848-49.)
Third, Novell could have written its own file open dialog without relying on the
namespace extension APIs, either using common controls supplied by Windows 95 or writing a
file open dialog from scratch, the latter being more time-consuming. (Harral, Oct. 20 Trial Tr. at
342-43.) Harral and Richardson both testified that they chose to pursue an onerous and difficult
version of Option 3 by attempting to replicate functionality provided by the Windows Explorer
in Novell's custom file open dialog. (Harral, Oct. 20 Trial Tr. at 344-47; Harral, Oct. 24 Trial
Tr. at 502-04; Richardson, Oct. 25 Trial Tr. at 628-30.)
Harral testified that he knew that Option 3 posed the greatest risk of delaying the
release of WordPerfect and Quattro Pro for Windows 95. (Harral, Oct. 20 Trial Tr. at 342-43.)
In fact, when the developers in Novell's shared code group were working to write a custom file
open dialog, others at Novell recommended in the interest of a timely release using the
Windows 95 common file open dialog instead. Ford and LeFevre each testified that they
advocated--in the interest of avoiding delay--use of the Windows 95 common file open dialog.
See pp. 55-56, supra.
The evidence also shows that Microsoft continued to assist Novell and never
terminated the relationship. Frankenberg testified that he was "sure" that "people in the systems
group at Microsoft were trying to help WordPerfect/Novell produce a great application for
Windows 95" (Frankenberg, Nov. 7 Trial Tr. at 1131; see also Frankenberg, Nov. 8 Trial Tr. at
1217), and, in fact, six months after Gates' October 3 Decision, Scott Nelson of Novell reported
in an April 7, 1995 e-mail that "the cooperation between Microsoft and Novell has been very
good" (DX 172). LeFevre--the Director of Marketing for PerfectOffice and one of four people
-95-
on the leadership team that was in charge of the development of PerfectOffice for Windows 95
(LeFevre, Dec. 2 Trial Tr. at 4034-35)--testified that throughout Novell's development efforts in
1994 and 1995, Microsoft provided assistance to Novell, including having a Microsoft employee
working at Novell's Orem campus to answer questions from the developers:
[S]tarting in 1994 all through 1995, we had an employee at
Microsoft who lived in Utah County whose job it was to support us
in this development effort. He was at our offices so frequently that
we finally gave him an office with a telephone so he could come in
and work when he needed to. And he was just down the hall in my
building, and we saw him frequently. We also had direct support
to Microsoft's developer relations group. So if we had questions,
we could contact them on the phone or e-mail and they would
answer our questions. We were very supported during this time.
One example I remember distinctly is that Tom Creighton and I
were able to fly to Redmond, and we spent an entire day in
building 22 of the Microsoft campus meeting with the
development team for Windows answering some critical questions
that Tom had about the product. They were very happy to do this.
They even paid for our flight and everything to get up to Redmond
and spend the day.
(Id. at 4029-30.)
In light of this evidence, no reasonable jury would have a sufficient evidentiary
basis to find that Microsoft "jettison[ed]" or "terminated a profitable relationship" between the
parties, which is required under Tenth Circuit law in order to bring a claim under the Aspen
Skiing exception. Four Corners, 582 F.3d at 1225. The relationship continued; it was Novell
that chose to try to get a competitive advantage by adding functionality to Windows 95 when it
could have used Microsoft technology without charge to release its products. This is far from the
facts in Aspen Skiing, where defendant refused to sell lift tickets to its rival, even at full retail
price. 472 U.S. at 592-93.
-96-
2. The Withdrawal of Support for the Namespace Extension APIs Did Not Deny
Novell Information or Support Available to All Other ISVs.
Given the evidence at trial, no reasonable jury would have a sufficient evidentiary
basis to find that Microsoft "den[ied] to [its] rival the [APIs] available to all other" ISVs. Four
Corners, 582 F.3d at 1225 (emphasis added). Microsoft did not treat Novell differently from
other ISVs creating applications to run on Windows 95. On the contrary, the October 3 Decision
applied to all ISVs--including the Microsoft Office team. (See, e.g., DX 3, at MX 6055841
("All applications within Microsoft which were originally implementing these interfaces have
been required to stop.").) See also pp. 33-34, supra.
Once Microsoft decided to withdraw support for the namespace extension APIs, it
promptly informed Novell and other ISVs in the First Wave Program. On October 12, 1994,
Struss reported that "we're now in the process of proactively notifying ISVs about the
namespace extension api changes." (DX 3, at MX 6055844.) Struss wrote that "[s]o far Stac,
Lotus, WP [WordPerfect], Oracle, SCC appear to be OK with this." (Id.) He testified that
"DRG proactively notified the software developers that we were working with about this
change . . . so that . . . they would know not to depend upon these interfaces moving forward."
(Struss, Nov. 28 Trial Tr. at 3270-71.)
Novell, like all other ISVs, was still able to utilize the thousands of other APIs
exposed by Windows 95 to build versions of WordPerfect and Quattro Pro for Microsoft's new
operating system. The decision to withdraw support for the namespace extension APIs--four
APIs out of thousands--is nothing close to the "terminat[ion] of a profitable relationship" on
terms that differed from those applicable to "all other competitors." Four Corners, 582 F.3d at
1225 (quotation omitted and emphasis in original). Indeed, in Christy Sports, the Tenth Circuit
held that defendant could terminate a 15-year business relationship, which put plaintiff out of
-97-
business, because plaintiff knew that the relationship was temporary and subject to change.
555 F.3d at 1196-97. Unlike the facts in Christy Sports, Novell was able to continue to develop
its Windows 95 products after the October 3 Decision and knowingly turned down options to
release its products on time.
In Intergraph Corp. v. Intel Corp., 195 F.3d 1346 (Fed. Cir. 1999), the Federal
Circuit held that the withdrawal of technical support is not anticompetitive under Section 2 of the
Sherman Act. There, defendant, a monopolist in the manufacture and sale of microprocessors,
provided plaintiff "with various special benefits, including proprietary information and
products," including "pre-release products." Id. at 1350-51. After a disagreement arose as to
licensing issues, defendant reduced the "technical assistance and other special benefits" it
previously had provided to plaintiff. Id. at 1350. Specifically, defendant "refus[ed] to authorize
help to Intergraph for removal of a `bug' or defect in a product," which "requir[ed] Intergraph to
spend substantial time and resources to solve the problem and delay[ed] Intergraph's product
entry into the market." Id. at 1365-66 (quotation omitted). The district court enjoined defendant
from refusing to provide assistance to plaintiff, and defendant appealed, "arguing that no law
requires it to give such special benefits" to plaintiff. Id. at 1351.
The Federal Circuit vacated the injunction, holding that "[t]he withdrawal of
technical service is not a violation of the antitrust laws." Id. at 1366. The Court further
explained that "[t]he federal antitrust laws do not create a federal law of unfair competition or
purport to afford remedies for all torts committed by or against persons engaged in interstate
commerce," id. at 1364 (quoted with approval in Gregory v. Fort Bridger Rendezvous
Association, 448 F.3d 1195, 1205 (10th Cir. 2006)), and that even "`an act of pure malice by one
business competitor against another does not, without more, state a claim under the federal
-98-
antitrust laws.'" Id. at 1366 (quoting Brooke Group Ltd. v. Brown & Williamson Tobacco Corp.,
509 U.S. 209, 225 (1993)).
Likewise, Microsoft's withdrawal of support for the namespace extension APIs--even if that act forced Novell "to spend substantial time and resources to solve the problem and
delay[ed] [Novell's] product entry into the market"--does not give rise to a cognizable claim
under Section 2 of the Sherman Act. Intergraph, 195 F.3d at 1366. As the Supreme Court
recently emphasized, a claim alleging "`insufficient assistance in the provision of service to
rivals' d[oes] not violate the Sherman Act" because Trinko "makes clear that if a firm has no
antitrust duty to deal with its competitors . . . , it certainly has no duty to deal under terms and
conditions that the rivals find commercially advantageous." Pacific Bell, 555 U.S. at 449-50
(quotation to Trinko omitted).
B. Common Practice in the Software Industry and the Terms of the Relevant
License Agreements Permitted Microsoft to Withdraw Support for APIs in a
Beta Version of Windows 95.
The very license agreement under which Novell received documentation for the
namespace extension APIs expressly provided, consistent with software industry practice, that
the beta version might change prior to the commercial release of Windows 95. This fact alone
defeats Novell's claim, for two reasons.
First, the license agreement warning Novell that Microsoft might make changes
to the beta version of Windows 95 was consistent with common software industry practice.
Under Tenth Circuit law, a monopolist is free to engage in "ordinary business practices typical of
those used in a competitive market," and cannot violate Section 2 for engaging in "the type of
competition prevalent throughout the industry." Telex Corp. v. International Business Machines
Corp., 510 F.2d 894, 925-26, 928 (10th Cir. 1975). If a defendant's conduct is "consistent with a
competitive market," then "the purpose of the antitrust laws is amply served." United States v.
-99-
Syufy Enterprises, 903 F.2d 659, 668-69 (9th Cir. 1990); Olympia Equipment Leasing Co. v.
Western Union Telegraph Co., 797 F.2d 370, 375 (7th Cir. 1986) ("[T]he lawful monopolist
should be free to compete like everyone else; otherwise the antitrust laws would be holding an
umbrella over inefficient competitors.").
The trial record established that it was the common understanding in the software
industry (including at Novell and Microsoft) that a software developer might make changes to
beta versions of products under development. Of most importance, Novell's CEO in 1994 and
1995, Frankenberg, testified that it "was widely understood in the software industry" that beta
versions of software products can and do change. (Frankenberg, Nov. 8 Trial Tr. at 1204-05.)
Larsen similarly testified that, based on his experience in the software industry for
more than 25 years (Larsen, Nov. 30 Trial Tr. at 3567, 3607), "the definition of a beta" is that
"there can be and almost certainly will be changes" (id. at 3603).61 LeFevre testified that, based
on his 20 years in the software industry, it was his understanding that a "company that develops
the beta software has the right to make any changes they deem necessary as a result of that
testing period" because a "[b]eta by definition is an early release or a prerelease of a product that
is subject to change." (LeFevre, Dec. 2 Trial Tr. at 4030-32.)
Alepin confirmed that when companies receive a beta version of a product under
development "they use it at their own risk" and they fully "expect[] . . . that the [beta] software is
being worked on." (Alepin, Nov. 10 Trial Tr. at 1555-56.) Noll agreed that "all beta versions of
all software are provisional, and they are not guarantees of what the program will contain upon
-100-
final release." (Noll, Nov. 15 Trial Tr. at 1878.) Gates likewise explained that the disclaimer
language included in Microsoft's license agreement for its M6 beta release was not only
"standard practice at Microsoft" but "I think everyone followed that practice." (Gates, Nov. 22
Trial Tr. at 3124-25.)
Novell's own practices were the same. As Frankenberg acknowledged, Novell's
license agreements for beta versions of its NetWare products included substantially similar
provisions to the Microsoft license agreement covering beta versions of Windows 95.
(Frankenberg, Nov. 8 Trial Tr. at 1202-05, 1208-09.) In fact, an internal Novell Memorandum
dated Oct. 18, 1994--a mere 15 days after Microsoft's October 3 Decision--explained that:
As with the Alpha phase, the Beta process typically uncovers
significant numbers of situations in which features do not meet
the conceptual design or the design is faulty requiring further
design, coding, and testing, or abandonment of the feature,
which also results in a new round of testing. Despite the
conceptual freeze, the product and features may still change
dramatically during this phase as problems are discovered.
(DX 612A, at 4 (emphasis in original).) Thus, Novell itself recognized in 1994 that "[f]eatures
may . . . change dramatically" during the beta process.
Novell and WordPerfect Corporation followed this practice themselves. Larsen
recalled instances in which Novell "made changes in our beta software where we would even
change the file formats and make other changes that could have potential negative impacts on the
customers. But we made those changes because we felt like it was in the overall best interest of
the product. So it's very common for changes to be made during the beta." (Larsen, Nov. 30
Trial Tr. at 3607.) LeFevre likewise testified that, during the beta testing process for
WordPerfect 5.1 for Windows, WordPerfect Corporation eliminated a number of features that
had been included in beta versions. (LeFevre, Dec. 2 Trial Tr. at 4033, 4080-81.)
-101-
In light of the evidence that the standard practice in the software industry was that
changes to beta versions were common and expected--and that Novell's CEO and developers
were familiar with this practice--a reasonable jury would not have a sufficient evidentiary basis
to find that Microsoft's decision to withdraw support for the namespace extension APIs was
anticompetitive. Telex, 510 F.2d at 925-26, 928; see also Trace X Chemical, Inc. v. Canadian
Industries, Ltd., 738 F.2d 261, 266 (8th Cir. 1984).
Second, as the Tenth Circuit held in Christy Sports, temporary business
relationships that are subject to change in accordance with a defendant's "business judgment"
cannot, as a matter of law, give rise to an antitrust claim. 555 F.3d at 1197. In Christy Sports,
plaintiff brought suit under Section 2 of the Sherman Act alleging that defendant's revocation of
permission for plaintiff to operate a ski rental facility under the terms of a restrictive covenant in
a lease was anticompetitive conduct under Aspen Skiing. Id. at 1196. Like Novell--which
concedes that Microsoft was not required to provide any information to ISVs about the
namespace extension APIs in the first place (Nov. 18 Trial Tr. at 2587, 2598-99)--plaintiff in
Christy Sports did "not seriously argue that it was impermissible for [defendant] to impose the
restrictive covenant back in 1990, or that it would have been impermissible for [defendant] to use
its ownership of the land to bar competition in the ski rental business from the beginning. " Id. at
1196. Nevertheless, just like Novell--which contended at trial that Microsoft could not
withdraw support for the namespace extension APIs once those APIs had been included in the
M6 beta in June 1994 (Nov. 18 Trial Tr. at 2599)--plaintiff in Christy Sports argued that
because defendant "allowed third parties to engage in the ski rental business for almost fifteen
years, [defendant] violated § 2 of the Sherman Act when it revoked that permission and took
-102-
over the ski rental business for itself." Id. at 1196. The Tenth Circuit rejected the argument
because the claim did not fall within the narrow Aspen Skiing exception.
The Court of Appeals held that the restrictive covenant, which prohibited plaintiff
from operating a ski rental facility without defendant's consent, made clear to plaintiff "that the
relationship could change at any time," and that plaintiff "should have been aware that the
relationship was temporary and subject to [defendant's] business judgment." Id. at 1197. The
Tenth Circuit further observed that it did "not see why an initial decision to adopt one business
model would lock [defendant] into that approach and preclude adoption of the other at a later
time." Id. at 1196. The Court stressed that "[t]he Sherman Act does not force [defendant] to
assist a competitor in eating away its own customer base," and concluded that even though
"[c]onceivably, such a change might lead to a claim under contract law or as a business tort,"
enforcing the restrictive covenant in the lease did not amount to anticompetitive conduct under
the Sherman Act. Id. at 1196-97; see also Intergraph, 195 F.3d at 1364-66 (holding that
withdrawal of technical information did not violate Section 2 of Sherman Act because
"proprietary information and pre-release products" were provided under "non-disclosure
agreements," which provided that "both parties may `cease giving Confidential Information to
the other party without liability,' and that either party can `terminate [the] Agreement at any time
without cause'").
Novell's claim regarding Microsoft's withdrawal of support for the namespace
extension APIs is even weaker than the claims in Christy Sports and Intergraph. Novell received
documentation for the namespace extension APIs less than four months before the October 3
Decision pursuant to a license agreement that expressly notified Novell that the product was still
under development and was subject to change:
-103-
2. PRE-RELEASE CODE. This PRODUCT consists of pre-
release code, documentation and specifications and is not at the
level of performance and compatibility of the final, generally
available product offering. The PRODUCT may not operate
correctly and may be substantially modified prior to first
commercial shipment. COMPANY assumes the entire risk with
respect to the use of the PRODUCT.
(DX 18, at 1, § 2.) Moreover, the M6 beta version of Windows 95 in June 1994 included clear
warnings that the documentation did "not represent a commitment on the part of Microsoft for
providing or shipping the features and functionality discussed in the final retail product offerings
of Chicago [Windows 95]." (PX 388, Microsoft Windows "Chicago" Reviewer's Guide, at
MSC 00762731.) In fact, Frankenberg acknowledged that Novell understood that Windows 95
"might change" and "could change" between the M6 beta version and the commercial release of
the new operating system. (Frankenberg, Nov. 8 Trial Tr. at 1201, 1209.)62
Indeed, Nakajima, Belfiore and Bennett all recounted instances where companies,
including Microsoft, withdrew support for a feature or interface in subsequent versions of
software. Nakajima testified that Apple removed an interface in its iOS operating system after
its release, and that Google also withdrew support for its Wave product after initially releasing it
as a beta version. (Nakajima, Dec. 1 Trial Tr. at 3735-37.) Belfiore also testified that in addition
to fixing bugs, beta releases of Windows reflected "lots of changes that [Microsoft] made to alter
the user interface to make it easier to learn how to use" the operating system. (Belfiore, Dec. 5
-104-
Trial Tr. at 4239.) Bennett testified that Sun Microsystems removed a communication protocol
that it had released as part of a beta version of its software. (Bennett, Dec. 12 Trial Tr. at 4967.)
Christy Sports establishes that a company can change an existing business
relationship--in that case, so as to put plaintiff out of business--without fear of incurring
Section 2 liability. 555 F.3d at 1197-98. The facts here are even worse for Novell than they
were for plaintiff in Christy Sports; not only did Novell know full well that features in the beta
version of Windows 95 were subject to change, but here, unlike the 15-year course of conduct in
Christy Sports, Novell first got documentation for the namespace extension APIs in the M6 beta
release of Windows 95 a mere four months prior to Microsoft's decision to withdraw support for
those APIs. And Novell had at least ten months (from October 1994 to August 1995) to release a
product and then knew that it could do so without delay by using the Windows 95 common file
open dialog that Microsoft provided for free.
C. Microsoft's Decision to Withdraw Support for the Namespace Extension
APIs Was Based on Legitimate Business Justifications.
In order to fit within the narrow Aspen Skiing exception, Novell must prove that
Microsoft's withdrawal of support for the namespace extension APIs was "without any economic
justification." Four Corners, 582 F.3d at 1225 (quotation omitted). As the Tenth Circuit has
emphasized, "[t]he critical fact in Aspen Skiing was that there were no valid business reasons for
the refusal." Christy Sports, 555 F.3d at 1197.
Under controlling Tenth Circuit law, a legitimate business justification need be no
more than a desire to protect the profitability of one's business. In Four Corners, for example,
defendant hospital terminated the credentials of unaffiliated nephrologists in order to protect the
hospital's budding nephrology practice. Four Corners, 582 F.3d at 1217-19. The Tenth Circuit
held that "the evidence here suggests that [defendant] refused to deal with [the plaintiff] to avoid
-105-
an unprofitable relationship, and that [defendant] pursued the course it did to protect and
maximize its chances of profitability in the short-term." Id. at 1225 (emphasis in original). The
Court held that "Aspen Skiing does not require more economic justification than [that] to avoid
Section 2 liability." Id.
Here, the trial evidence established that Microsoft had at least three justifications
for withdrawing support for the namespace extension APIs--all of which served the purpose of
enhancing the value of Microsoft's Windows 95 operating system: (1) third-party applications
that used the namespace extension APIs could cause the Windows 95 operating system to crash;
(2) the design of the namespace extension APIs was not compatible with future versions of
Microsoft Windows under development; and (3) the namespace extension APIs did not achieve
the functionality that Gates had anticipated.
1. A Third-Party Application Using the Namespace Extension APIs Could Crash
Windows.
Third-party applications that called the namespace extension APIs ran in the same
process as the Windows 95 shell and, as a result, if the third-party application crashed, the shell
would also crash. See pp. 24-26, supra. Withdrawal of support for the namespace extension
APIs increased the stability of the Windows 95 product and, in turn, the profitability of
Microsoft's business.
The evidence at trial on that point was overwhelming and largely unchallenged.
Nakajima, the inventor of the namespace extension mechanism in Windows 95, testified that a
misbehaving application calling the namespace extension APIs could cause the Windows 95
operating system to crash. (Nakajima, Dec. 1 Trial Tr. at 3757.) Several of Microsoft's other
witnesses, including Gates, agreed that the namespace extension mechanism in Windows 95
posed robustness problems. (Gates, Nov. 21 Trial Tr. at 2781-82; Jan. 9, 2009 Maritz Deposition
-106-
at 129, Dkt. #279, used at trial on Oct. 27; Jan. 8, 2009 Allchin Deposition, Nov. 8 Trial Tr. at
1297; Muglia, Nov. 29 Trial Tr. at 3394-97; Belfiore, Dec. 5 Trial Tr. at 4278-79; Struss, Nov.
28 Trial Tr. at 3329-30; Bennett, Dec. 12 Trial Tr. at 4993-94.) Moreover, the contemporaneous
documentary evidence is consistent with this testimony and confirmed that Microsoft was
concerned about robustness issues when it decided to withdraw support for the namespace
extension APIs. (DX 3, at MX 6055843.) See also p. 29, supra.
Not only did Novell fail to present any evidence at trial to undermine this
evidence, but Novell's own developers and expert witnesses aknowledged that the namespace
extension APIs posed a risk to the stability of the Windows 95 operating system. Richardson
agreed that "a namespace extension that was badly behaved . . . could have crashed the entire
Windows 95 shell, circa October 1994, because at that time namespace extensions were running
in the same process as Windows explorer and the rest of the shell." (Richardson, Oct. 25 Trial
Tr. at 756-57.) Alepin testified that the namespace extension APIs "had the potential to make the
system unresponsive." (Alepin, Nov. 10 Trial Tr. at 1589.) Noll agreed that a risk to the
stability of the operating system was "one valid reason for not documenting an API." (Noll,
Nov. 15 Trial Tr. at 1872-73.) This testimony eliminates any contention that Microsoft's
justifications for the withdrawal of support for the namespace extension APIs were a pretext.
The evidence also showed that robustness problems with the namespace extension
mechanism in Chicago posed significant problems for Windows NT, a high-end operating
system used in situations where a "crash" of the system was completely unacceptable. (Gates,
Nov. 21 Trial Tr. at 2781; see also Alepin, Nov. 10 Trial Tr. at 1607-08.) The Windows NT
team's objections to the namespace extension APIs were compounded by Microsoft's decision
"to use a common s[h]ell across Windows 95 and Windows NT." (Gates, Nov. 21 Trial Tr. at
-107-
2784-85; see also Alepin, Nov. 10 Trial Tr. at 1607.) Muglia testified that he "really work[ed]
hard" to ensure that the namespace extension APIs would "not be shipped" with Windows 95 in
order to avoid adopting the APIs as part of the Windows NT shell. (Muglia, Nov. 29 Trial Tr. at
3397; see also id. at 3386-87.) Indeed, a day after the October 3 Decision, Muglia applauded the
decision, noting that Windows NT would no longer have to expend resources trying to create a
robust implementation of the namespace extension APIs. (DX 21; Muglia, Nov. 29 Trial Tr. at
3420-21.)63
Alepin asserted that he did not believe that the risk to the stability of Windows 95
was a sufficient justification for the withdrawal of support for the namespace extension APIs
because "there were lots of ways to get Windows 95 to crash," and it is an ISV's "burden" to
write "good quality tested software that does not cause the system to fail." (Alepin, Nov. 9 Trial
Tr. at 1427-28.) That critique makes no sense. The existence of other ways to crash Windows
95 does not mean Microsoft was not justified in eliminating one such source of instability.
(Bennett, Dec. 12 Trial Tr. at 5006 ("[W]hether or not other . . . processes executing with the
operating system might or might not expose reliability issues is no reason not to address . . . the
one you know about.").) Moreover, Alepin acknowledged that Microsoft had no ability to
impose quality control requirements on ISVs whose applications called the namespace extension
APIs. (Alepin, Nov. 10 Trial Tr. at 1593-94). Microsoft clearly was justified for its own
business reasons in seeking to protect users of Windows 95. (Bennett, Dec. 12 Trial Tr. at 5006.)
-108-
2. Supporting the Namespace Extension APIs Would Lock Future Microsoft
Operating Systems Into the Design of the Windows 95 Shell.
The Cairo team, which was developing an advanced version of Windows,
objected to the namespace extension APIs because that team thought they were poorly designed
and because use of those APIs by ISVs would force the Cairo shell team to be compatible with
such applications. (Gates, Nov. 21 Trial Tr. at 2792-93, 2789-90; see also Muglia, Nov. 29 Trial
Tr. at 3453, 3399-3400; Belfiore, Dec. 5 Trial Tr. at 4278-79.)
Nakajima testified that the Cairo team's objections to the design of the Chicago
shell "were right." (Nakajima, Dec. 1 Trial Tr. at 3772-73.) Rather than force the Cairo team to
be compatible with the design of the Chicago shell, Gates decided to withdraw support for the
namespace extension APIs to address the Cairo team's objections. See pp. 24, 27-29, supra.
3. The Namespace Extension APIs Did Not Provide the Functionality Bill Gates Had
Contemplated.
The limited functionality of the namespace extension APIs was another reason
that Microsoft decided to withdraw support. Gates testified that he had hoped for an operating
system shell that provided a "rich new view" (Gates, Nov. 21 Trial Tr. at 2786-87), and that
Cairo was "trying" to develop "something very rich and flexible" (id. at 2788).
Gates wanted to "wait" until this Cairo-like "high level of integration" could be
achieved (PX 1) and, as Gates explained, "the namespace extension APIs [were] not rich enough
to give you the ability to do this kind of information browser shell" (Gates, Nov. 21 Trial Tr. at
2802-03). Indeed, several witnesses testified that the functionality provided by the namespace
extension APIs were neither important nor necessary to the development of an application for
Windows 95. (Belfiore, Dec. 5 Trial Tr. at 4263-64; Muglia, Nov. 29 Trial Tr. at 3389; Gates,
Nov. 21 Trial Tr. at 2786-87.) Gates therefore "decided that until we got to the Cairo capability
it really didn't change things enough to be worth the trouble to cause the problems for the NT
-109-
and Cairo teams, if I had gone the other way, would have caused." (Gates, Nov. 21 Trial Tr. at
2804.) Novell did not examine Gates about this reason for his decision, let alone offer any
evidence challenging its legitimacy.
4. Novell's Experts Failed to Rebut Microsoft's Justifications for Withdrawing
Support for the Namespace Extension APIs.
At bottom, Novell's only attempt to counter Microsoft's business justifications for
withdrawing support for the namespace extension APIs is Alepin's opinion that the purported
costs of that decision to Novell outweighed the benefits to Microsoft. (See Alepin, Nov. 9 Trial
Tr. at 1426-30; see also Noll, Nov. 15 Trial Tr. at 1874 ("If in fact they were sufficiently
unstable and the harms exceeded the benefits of documenting, then it would be perfectly valid to
withdraw them.").) That opinion is irrelevant as a matter of law. Once Microsoft has shown that
legitimate business justifications exist for the decision to withdraw support for the namespace
extension APIs, the inquiry is at an end. Aspen Skiing, 472 U.S. at 597, 605; Multistate Legal
Studies v. Harcourt Brace Jovanovich Legal & Professional Publications, 63 F.3d 1540, 1550
(10th Cir. 1995). A jury is not permitted to engage in a weighing of costs and benefits.
As the Fifth Circuit held in Bell v. Dow Chemical Co., a jury cannot "weigh the
sufficiency of a legitimate business justification against the anticompetitive effects of a refusal to
deal." 847 F.2d 1179, 1186 (5th Cir. 1988). Indeed, "[t]he fact determination that may be left to
a jury is whether the defendant has a legitimate business reason for its refusal, not whether that
reason is sufficient." Id. (emphasis in original) (citing Aspen Skiing, 472 U.S. at 597); see also 3
PHILLIP E. AREEDA & HERBERT HOVENKAMP, ANTITRUST LAW § 772c2 (3d ed. 2011) ("[T]he
Court [in Aspen Skiing] did not call for any balancing of social gains from refusing to deal or
cooperate with rivals based on legitimate business purposes against the losses resulting from that
refusal. Rather, the Court classified conduct or intention as either lawful or not on the basis of
-110-
the presence or absence of legitimate business purposes."). Thus, the existence of any economic
justification for withdrawal of support for the namespace extension APIs ends the inquiry.
D. Novell's Attempt to Base Its Claim on a "Deception" Theory Has No Basis in
Law or in Fact.
During the November 18 argument on Microsoft's Rule 50(a) Motion for
Judgment as a Matter of Law, Novell's counsel conceded that Section 2 of the Sherman Act does
not require Microsoft to share its intellectual property--in this case, the namespace extension
APIs--with Novell. (Nov. 18 Trial Tr. at 2587, 2599.) Novell's counsel also conceded that
under Aspen Skiing and its progeny, Microsoft has "no duty to cooperate" with Novell. (Id. at
2636.) Instead, Novell's counsel articulated a new theory--that Microsoft had a duty not to
"deceive" Novell. (Id. at 2587.) Novell contended that Microsoft's withdrawal of support for
the namespace extension APIs after "evangeliz[ing]" them constituted deception in violation of
the Sherman Act. (Id. at 2661-62.)
1. Novell's Purported Claim for Deception Is Not Cognizable Under the
Antitrust Laws.
Deceiving a competitor does not give rise to an antitrust claim under Section 2 of
the Sherman Act. Deception sounds in tort, and the Supreme Court has stated that the federal
antitrust laws "do not create a federal law of unfair competition or `purport to afford remedies for
all torts committed by or against persons engaged in interstate commerce.'" Brooke Group Ltd.
v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 225 (1993) (citation omitted); accord New
York v. Microsoft, 224 F. Supp. 2d 76, 138-39 (D.D.C. 2002). "Even an act of pure malice by
one business competitor against another does not, without more, state a claim under the federal
antitrust laws." Brooke Group Ltd., 509 U.S. at 225; see also Intergraph., 195 F.3d at 1354-55
("[T]he Sherman Act does not convert all harsh commercial actions into antitrust violations.").
The Tenth Circuit has also recognized that "unfair" conduct directed at a competitor does not in
-111-
and of itself violate the Sherman Act. Midwest Underground Storage, Inc. v. Porter, 717 F.2d
493, 496-97 (10th Cir. 1983). A claim for deception is nothing more than a tort claim, and, as
Novell's counsel has acknowledged, "[t]his is not a tort case." (Nov. 14 Trial Tr. at 1750.)
Indeed, Microsoft is not aware of any case in which liability was imposed under
the antitrust laws based on deceiving one's competitor. In Conwood v. U.S. Tobacco Co., 290
F.3d 768, 788 (6th Cir. 2002), the Sixth Circuit affirmed a finding of Section 2 liability based on
the defendant's pervasive abuse of imperfect information through misrepresentations made not to
its competitors, but to retailers. Similarly, the other decided cases involving deception concern
false advertising or conduct directed at third parties--such as distributors or consumers--and the
standard for imposing antitrust liability even in such cases is very high. See American
Professional Testing Service, Inc. v. Harcourt Brace Jovanovich Legal & Professional
Publications, Inc., 108 F.3d 1147, 1151 (9th Cir. 1997) ("While the disparagement of a rival . . .
may be unethical and even impair the opportunities of a rival, its harmful effects on competitors
are ordinarily not significant enough to warrant recognition under § 2 of the Sherman Act.");
International Travel Arrangers, Inc. v. Western Airlines, Inc., 623 F.2d 1255, 1260-63 (8th Cir.
1980); In re Warfarin Sodium Antitrust Litigation, 1998 U.S. Dist. LEXIS 19555, at *27-28 (D.
Del. Dec. 7, 1998), rev'd in part on other grounds, 214 F.3d 395 (3d Cir. 2000); see also 3
PHILLIP E. AREEDA & HOVENKAMP, ANTITRUST LAW § 782d (3d ed. 2011) ("We would go
further and suggest that such claims should presumptively be ignored.").
In any case, the Complaint makes no claim of deception.64 The word "deception"
appears exactly once in Novell's 68-page Complaint (and there, in a wholly different context
-112-
unrelated to the namespace extension APIs), and "deceit" and "deceive" appear not at all.
Moreover, under Tenth Circuit law, no such claim could lie. In Christy Sports, defendant leased
a parcel of land to plaintiff with a restrictive covenant prohibiting the operation of a ski rental
business. For 15 years, plaintiff ran a ski rental business on the property without objection.
When defendant finally sought to enforce the restrictive covenant, plaintiff brought suit under
Section 2 of the Sherman Act. The Tenth Circuit held that there was no basis for imposing
liability under Section 2 because defendant "had explicitly informed its competitor from the
beginning that the relationship could change at any time" and thus the plaintiff "cannot claim
unfair surprise." 555 F.3d at 1196-97. The court observed that it did "not see why an initial
decision to adopt one business model would lock [defendant] into that approach and preclude
adoption of the other at a later time." Id. at 1196. While acknowledging that such a change in
course "[c]onceivabl[y] . . . might lead to a claim under contract law or as a business tort,"
enforcing the restrictive covenant did not amount to anticompetitive conduct under the Sherman
Act. Id. at 1196-98.
Likewise, no claim for deception can arise when Novell received the M6 beta
under an express contractual provision that the operating system was still under development and
subject to change. See pp. 21-22, supra. As Frankenberg acknowledged, Novell understood that
the commercial version of Windows 95 "might change" and "could change" from the M6 beta
version. (Frankenberg, Nov. 8 Trial Tr. at 1201, 1204-05, 1209.) This is the very opposite of
deception.
2. There Was No Evidence of Any Deception.
Even if deception of a competitor could support an antitrust claim, Novell failed
to present any evidence at trial that Microsoft deceived Novell. Novell argued in its opening
statement and in response to Microsoft's Rule 50(a) motion that Gates' October 3 Decision "is
-113-
based on the exact same considerations laid out in the Hood Canal retreat in June 1993," that
"Chicago would not give ISVs the extensibility of the Namespace extensions" and that "the
purpose of the plan was to grant those benefits to Microsoft's Office productivity applications in
Office." (Nov. 18 Trial Tr. at 2644; see also Oct. 18 Trial Tr. at 47-48.)
Faced with no evidence to support that theory, Novell's counsel conceded during
summation that the Hood Canal plan "didn't go forward" (Dec. 13 Trial Tr. at 5324-25).
Consistent with that concession, every Microsoft witness asked about the Hood Canal retreat
testified that there was never a plan to make the Windows 95 shell non-extensible and to instead
ship an extensible shell with Microsoft Office after the release of Windows 95.
65 Gates testified
that "[t]here was no plan that I ever agreed with to make the Chicago shell non-extensible."
(Gates, Nov. 22 Trial Tr. at 2999; see also Gates, Nov. 21 Trial Tr. at 2770 ("We did not ever
decide to create a shell, an Office shell separate from the operating system."); id. at 2771 ("[W]e
didn't even do any work related to that, not to mention not shipping any such thing.").) Muglia
testified that such a proposal "was discussed in this brainstorming session" but "never went
anywhere." (Muglia, Nov. 29 Trial Tr. at 3402.) Maritz testified that "it never got beyond talk"
and "[t]here was never any . . . reality behind it." (Jan. 9, 2009 Maritz Deposition at 70, Dkt.
#279, used at trial on Oct. 27; see also id. at 65.) Silverberg, the Microsoft executive in charge
of Windows 95, testified that he "ha[d] a vague recollection that there was a discussion to also
-114-
ship another shell in Office, but . . . that never occurred." (Jan. 22, 2009 Silverberg Deposition
at 15, Dkt. #278, used at trial on Oct. 26.)
In addition, and even more fundamentally, there is no evidence that Microsoft
intended to withdraw support for the namespace extension at the time it evangelized them to
Novell or at the time Microsoft provided the M6 beta to Novell and other ISVs in June 1994.
Indeed, the evidence showed that Microsoft did not know that Novell was using the namespace
extension APIs. In his September 22, 1994 e-mail, Struss reported that WordPerfect had "not
begun any work on IShellFolder, IShellView, etc." (i.e., the namespace extension APIs). (DX
17.) See also p. 31, supra. Gates testified that he did not have "any awareness at all about the
specifics of whether [Novell developers] were using [the namespace extension APIs] or not"
(Gates, Nov. 21 Trial Tr. at 2811; see also Gates, Nov. 21 Trial Tr. at 2828), and there is no
evidence to the contrary.
III. Microsoft's Withdrawal of Support for the Namespace Extension APIs Did Not
Cause a Delay in the Release of PerfectOffice for Windows 95.
All agree that, as Professor Noll testified, "there can't be any harm to competition
under the facts here, if the conduct at issue, the decision to withdraw support for the namespace
extension APIs, did not cause any delay" in Novell's release of its PerfectOffice suite for
Windows 95. (Noll, Nov. 15 Trial Tr. at 1880-81; see also Noll, Nov. 14 Trial Tr. at 1839-40.)
The evidence at trial overwhelmingly demonstrated that the delay was due to other factors and
that Microsoft's withdrawal of support for the namespace extension APIs was not the cause of
any delay in the release of Novell's products. See pp. 34-45, 48-57, supra.
A. Quattro Pro Caused the Delay in Releasing PerfectOffice for Windows 95.
The evidence at trial established that PerfectOffice could not have been released
within 60 days (or even more) of the release of Windows 95 because Quattro Pro was not ready
-115-
until 1996. Because Novell could not have released PerfectOffice without Quattro Pro, the
October 3 Decision did not harm Novell.
The trial testimony and exhibits showed clearly that Novell had to delay the
release date of PerfectOffice several times as a result of problems with Quattro Pro, and that
Quattro Pro was not ready until March 1996 or later. Novell introduced not a single exhibit that
indicated that the delay was caused by the October 3 Decision.
Although Gibb testified that Quattro Pro "didn't turn out to be Critical Path" and
was "basically code completed" as of December 1995 (Gibb, Oct. 26 Trial Tr. at 808-09), he
nowhere testified that Quattro Pro was ready by October--or even November--1995 and, in any
event, the overwhelming testimonial and documentary evidence was to the contrary. See p. 52,
supra. Five former Novell employees, including Frankenberg, testified without ambiguity that
Quattro Pro--not the shared code group--was the cause of the delay. See pp. 48-53, supra.
Moreover, an e-mail from Brereton to Frankenberg clearly states that on December 23, 1995,
four months after the release of Windows 95, the Quattro Pro development team was left with
just two people and the product was not ready. (DX 230.) When shown DX 230 at trial,
Frankenberg agreed that "clearly the product wasn't complete" as of the end of December 1995.
(Frankenberg, Dec. 7 Trial Tr. at 1145.) Gibb's testimony cannot be reconciled with the mass of
evidence that pinpoints Quattro Pro as the cause of delay in the release of PerfectOffice for
Windows 95.
As the Court noted concerning Gibb's testimony, "there comes a point where
somebody comes in and they say it's like saying the world is flat. There . . . could not be clearer
evidence that Defendant's Exhibit 230 says that as of January 2006 Quattro Pro is not ready yet.
I mean, it couldn't be clearer." (Nov. 21 Trial Tr. at 2925; see also id. at 2906-07.) Gibb's
-116-
testimony falls far short of the "substantial evidence" necessary to defeat a motion under Rule
50. Webco Industries, 278 F.3d at 1128 ("substantial evidence . . . is defined as such relevant
evidence as a reasonable mind might accept as adequate to support a conclusion, even if different
conclusions also might be supported by the evidence") (quotation omitted). The documentary
evidence and the testimony of the other five former Novell employees--including Larsen who
worked as a developer on Quattro Pro in early 1996 and testified as a result of first-hand
experience based on his trip to Scotts Valley that the product was "[n]ot by any stretch of the
imagination" ready by March 1996 (Larsen, Nov. 30 Trial Tr. at 3624), and Bushman, who
testified that he recalled that the resignations on the Quattro Pro development team in December
1995 were a "death blow" to the product and its ship date (Bushman, Nov. 28 Trial Tr. at 3192-93)--far outweigh Gibb's unsupported testimony. See pp. 48-53, supra.
B. In Any Event, By Choosing the Most Time-Consuming and Difficult Option,
Novell Cannot Blame Microsoft for the Delay.
Assuming that Quattro Pro had been ready on time, Novell's own witnesses
acknowledge that Novell could have released PerfectOffice for Windows 95 without any delay
by using the Windows common file open dialog, but chose instead the time-consuming and
difficult option of writing its own custom file open dialog. The blame for this business choice
cannot be laid at Microsoft's door.
As an initial matter, Novell was late to start its development efforts for Windows
95--for reasons entirely unrelated to Microsoft's decision to withdraw support for the
namespace extension APIs. See pp. 38-45, supra. As Frankenberg said, Novell's "objective for
the last six months of 1994" was to create a suite written for the Windows 3.1 platform--what
became PerfectOffice 3.0, released in December 1994. (Frankenberg, Nov. 7 Trial Tr. at 1068.)
As a result, "very few resources" were devoted to developing products for Windows 95 (id.;
-117-
DX 4, at 5); see also pp. 42-45, supra, during the months after Microsoft provided the M6 beta to
Novell.
In any event, it would have been "easy" for Novell to have used the Windows 95
common file open dialog (Gibb, Oct. 26 Trial Tr. at 847-48), which would have been "less risky"
and "less work" than the path Novell chose (Harral, Oct. 24 Trial Tr. at 365-66). See also pp.
54-56, supra. As Harral said, "Novell could have come out with a product in '95 that utilized the
Windows common file open dialog." (Harral, Oct. 24 Trial Tr. at 502.) But Novell decided to
take the risk of trying to "do something cooler" and "exceed what was the default stuff." (Gibb,
Oct. 26 Trial Tr. at 848-49; see also Richardson, Oct. 25 Trial Tr. at 629-30.)
As Professor Noll testified, "to an antitrust economist, . . . there can't be any harm
to competition under the facts here if the conduct at issue, the decision to withdraw support for
the namespace extension APIs did not cause any delay." (Noll, Nov. 15 Trial Tr. at 1880-81.)
Novell's poor choice caused the delay.
IV. Novell Is Not Entitled to Any Damages as a Matter of Law.
Novell's damages expert, Warren-Boulton, calculated that Microsoft's October 3
Decision caused damages to Novell of between $976 million and $1.3 billion. (Warren-Boulton,
Nov. 16 Trial Tr. at 2103.) Each of the four damages theories presented by Warren-Boulton at
trial depended on the assumption that, but for Microsoft's October 3 Decision, Novell would
have been able to release PerfectOffice for Windows 95 within 30 to 60 days of the August 24,
1995 release of Windows 95 or "within a sufficiently short time period so that there would not
have been a significant effect on its sales" (Warren-Boulton, Nov. 17 Trial Tr. at 2418), which
neither Warren-Boulton or any other Novell witness ever put later than November 1995. That
-118-
assumption was demonstrated to be false by the evidence establishing that Quattro Pro was not
ready to be released until well into 1996. See pp. 48-53, supra.
Warren-Boulton attempted to salvage his damages theories by suggesting that the
jury could determine on its own the amount of "partial" damages if it found that PerfectOffice
for Windows 95 would not have been ready within 30 to 60 days of the release of Windows 95.
"This is precisely the type of `speculation or guesswork' not permitted for antitrust jury
verdicts." MCI Communications Corp. v. American Telephone & Telegraph Co., 708 F.2d 1081
1162 (7th Cir. 1983) (quoting Bigelow v. RKO Radio Pictures, 327 U.S. 251, 264 (1946)).
Further, Warren-Boulton's damages calculations were based largely on harm to PerfectOffice--an injury never alleged in the Complaint. See pp. 127-28 & n.70.
A. Novell's Damages Models Depended on the Assumption that PerfectOffice
Would Have Been Released Within 60 Days of the Release of Windows 95.
Warren-Boulton testified that Microsoft's withdrawal of support for the
namespace extension APIs "delayed the release of PerfectOffice until around May of 1996, far
past the date of release of Windows 95 in August of 1995." (Warren-Boulton, Nov. 16 Trial Tr.
at 2090.) To calculate damages resulting from the alleged delay, Warren-Boulton attempted to
construct "a world in which Microsoft did not pull the--did not deduct the namespace extensions
and therefore it would be a world in which Novell would have released PerfectOffice for 95 you
know at or close to the release date of Windows 7--or Windows 95." (Id. at 2096.) He testified
that "[i]t is my understanding also from that testimony that the expectation was that that was
[Novell's] goal, was to get it out within 30 or 60 days and that is my but-for world." (Warren-Boulton, Nov. 17 Trial Tr. at 2418.)
Warren-Boulton also testified that a timely release was critical to the success of
PerfectOffice, and that "everybody realized the really crucial importance of getting their suite out
-119-
or their product out . . . as soon as possible and certainly within some reasonable time period,
you know, 30, 60 something days . . . ." (Warren-Boulton, Nov. 16 Trial Tr. at 2190.) He added
that "[i]t is my understanding that . . . if you get it out within the prescribed time period . . . I
think it is 60 days in order to be part of the whole hype of Windows 95, if you make it in that
window you're in pretty good shape. It would [be] better to have it on August 23rd, but you're
in pretty good shape after a couple months." (Warren-Boulton, Nov. 17 Trial Tr. at 2419-20.)
According to Warren-Boulton, after the two month window of opportunity, "eventually,
however, as you delay, delay, delay, you're pretty dead, so it does not make very much
difference" if any additional delay occurs. (Id. at 2419.)
B. Because Warren-Boulton Failed to Account for Novell's Own Responsibility
for the Delay, Novell Is Entitled to No Damages.
As shown above, Quattro Pro was not ready even at the end of December 1995,
four months after the release of Windows 95. DX 230--the December 23, 1995 e-mail from
Bruce Brereton, the Vice President of the Business Applications business unit, to Frankenberg--makes clear not only that Quattro Pro was not ready to be released, but also that the resignation
of Quattro Pro developers in Scotts Valley meant that the Quattro Pro project for Windows 95
was in complete shambles. When shown DX 230, Frankenberg testified that "clearly the product
[Quattro Pro] wasn't complete" as of December 23, 1995. (Frankenberg, Nov. 7 Trial Tr. at
1145.) Because the essential assumption underlying all of Warren-Boulton's damages theories is
contrary to the evidence at trial, Novell is not entitled to any damages as a matter of law.
Asked how he would account for any delay beyond 60 days that was not the result
of Microsoft's withdrawal of support for the namespace extension APIs,
Warren-Boulton
admitted that his damages theories did not deal with that case, testifying that "I have to think
about this, and that is sort of off the cuff." (Warren-Boulton, Nov. 17 Trial Tr. at 2418-19.)
-120-
Warren-Boulton explained that "since my damage calculation basically assumes that in the but-for world they would have gotten this out within a reasonable time period, you know, if you
somehow present me with the fact that says that are not out until, you know, January 1st in the
but-for world I would say, yes, and I can do this if necessary, then I have to adjust my damages
for a different but-for world." (Id. at 2422-23.) But Warren-Boulton did not provide the jury
with any means to adjust any of his four damages theories for a different but-for world, instead
testifying that "if in fact somebody factually determines that even absent the bad acts the
products wouldn't have been out until January or February or March, then to that extent you [i.e.,
the jury] would need to modify the damages because it is partial." (Id. at 2421.)66 Novell
provided no basis for making such an adjustment.
This failure to provide the jury with any damages theory that took account of
factors other than Microsoft's October 3 Decision precludes Novell from being awarded
damages as a matter of law. As the ABA Model Jury Instructions explain, "[i]f you find that
plaintiff has failed to carry its burden of providing a reasonable basis for determining damages,
then your verdict must be for defendant." ABA Model Instruction F-16, Holley Decl. Ex. O. In
the antitrust context, providing the jury with a reasonable basis for determining damages requires
that the plaintiff disaggregate from the damages it seeks (i) any harm to the plaintiff attributable
to lawful conduct engaged in by the defendant and (ii) any harm to the plaintiff caused by factors
-121-
other than the defendant's anticompetitive conduct, including plaintiff's own business mistakes.
A leading treatise makes this clear:
[A]ny part of the plaintiff's loss that is due to the lawful business
practices of the defendant should not be part of the damage award.
In addition, some of the defendant's conduct may be unlawful but
not anticompetitive. . . . [T]he plaintiff's damage calculations must
[also] control for exogenous factors that also have an adverse
impact on the plaintiff's economic condition[,] . . . [such as] the
plaintiff's own mismanagement, a recessionary economy,
competition from other rivals, and other things unrelated to any
antitrust violation.
2A PHILLIP E. AREEDA & HERBERT HOVENKAMP, ANTITRUST LAW § 391g (footnotes and
citations omitted). "When a plaintiff improperly attributes all losses to a defendant's illegal acts,
despite the presence of significant other factors, the evidence does not permit a jury to make a
reasonable and principled estimate of the amount of damage." MCI Communications, 708 F.2d
at 1162. Warren-Boulton did exactly this--attributing all losses to Microsoft's October 3
Decision despite the presence of significant other factors, including Novell's own delays in
completing Quattro Pro for Windows 95. His failure to account for other factors leaves the jury
unable to make a "reasonable and principled estimate" of Novell's damages, meaning that Novell
cannot recover damages as a matter of law.
To be entitled to damages, an antitrust plaintiff "must segregate damages
attributable to lawful competition from damages attributable to . . . monopolizing conduct," and a
"failure to do so contravenes the commands of the Clayton Act." Image Technical Services, Inc.
v. Eastman Kodak Co., 125 F.3d 1195, 1224 (9th Cir. 1997) (citation omitted); see also Insignia
Systems, Inc. v. News American Marketing In-Store, Inc., 661 F. Supp. 2d 1039, 1054 (D. Minn.
2009) ("a plaintiff must disaggregate losses caused by the defendant's unlawful conduct from
losses caused by other factors."). Novell's failure to disaggregate the losses resulting from
Microsoft's October 3 Decision from the losses resulting from other factors, including Novell's
-122-
own responsibility for the delay, is fatal to its claim. "Several otherwise successful antitrust
claims have foundered on the expert's inability to show what portion of a plaintiff's loss resulted
from unlawful conduct." 2 PHILLIP E. AREEDA & HERBERT HOVENKKAMP, ANTITRUST LAW
¶ 309c (footnotes and citations omitted).
In addition, Warren-Boulton's damages calculations were based in large part on
harm to PerfectOffice. (Warren-Boulton Nov. 16 Trial Tr. at 2090, 2095-96.) As shown below
(see pp. 127-28, infra), the Complaint makes no claim for harm to PerfectOffice and thus Novell
released Microsoft from that claim. Warren-Boulton failed to disaggregate the harm to the
products that are the subject of the Complaint (WordPerfect and Quattro Pro), and thus Novell is
entitled to no damages for this reason as well. (See Warren-Boulton, Nov. 17 Trial Tr. at 2435-36.)
V. Novell Suffered No Cognizable Antitrust Injury.
Novell lacks standing to assert its claim because it suffered no cognizable antitrust
injury.67 A private plaintiff in a federal antitrust action must allege an "antitrust injury and must
have standing to bring an antitrust claim." Elliott Industries Ltd. v. BP America Production Co.,
407 F.3d 1091, 1124 (10th Cir. 2005) (citing Atlantic Richfield Co. v. USA Petroleum Co., 495
-123-
U.S. 328, 344 (1990)). The purpose of the antitrust injury requirement is to "ensure[] that the
harm claimed by the plaintiff corresponds to the rationale for finding a violation of the antitrust
laws in the first place." Atlantic Richfield, 495 U.S. at 342. An antitrust injury is an "injury of
the type the antitrust laws were intended to prevent and that flows from that which makes
defendants' acts unlawful." Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 489
(1977). Accordingly, the antitrust injury requirement "ensures that a plaintiff can recover only if
the loss stems from a competition-reducing aspect or effect of the defendant's behavior."
Atlantic Richfield, 495 U.S. at 344.
To establish antitrust injury in an unlawful monopolization claim, a plaintiff
"must show not only that [it] was harmed by [defendant's] conduct, but that the injury [it]
suffered involved harm to competition." Four Corners, 582 F.3d at 1225. In other words,
antitrust "plaintiffs ha[ve] to show that their alleged injuries reflect the anticompetitive effect of
the alleged violation." Haynes Trane Service Agency, Inc. v. American Standard, Inc., 51 F.
App'x 786, 803 (10th Cir. 2002). "Implicit in that requirement is the condition that they must
first show an actual anticompetitive effect." Id. Novell failed to do so at trial.
The injury Novell claims to have suffered--a delay in the release of Novell's
Three Products for Windows 95 (see Noll, Nov. 15 Trial Tr. at 1880-81; see also id. at 1839-40;
Noll, Nov. 14 Trial Tr. at 1775-76)--does not constitute an antitrust injury "because it has no
adverse effect on competition or consumers" in the PC operating system market. Elliott, 407
F.3d at 1125; cf. Full Draw Productions v. Easton Sports, Inc., 82 F.3d 745, 754 (10th Cir. 1999)
(finding antitrust injury where complaint alleged anticompetitive effect of boycott to be the loss
of competition through the elimination of one of two competitors in the market). In Elliott, the
Tenth Circuit upheld the district court's dismissal of a plaintiff's antitrust claims for failure to
-124-
allege antitrust injury where the only harm alleged was the "economic loss" allegedly suffered by
plaintiff, not harm to competition or to consumers. 407 F.3d at 1125; see also Four Corners,
582 F.3d at 1225 (where relief sought by plaintiff was to share defendant's monopoly, the court
held that "whatever injury [the plaintiff] may have suffered, . . . it is not one the antitrust laws
protect"). For the reasons stated above, see pp. 59-90, supra, the evidence at trial established
that a reasonable jury would not have a legally sufficient evidentiary basis to find that the
allegedly anticompetitive act harmed competition in the PC operating system market. Absent
such a showing, Novell cannot establish that it suffered antitrust injury and thus has no standing
to assert an antitrust claim.
In addition, Novell suffered no antitrust injury because it was "neither a consumer
nor a competitor" in the PC operating system market. Associated General Contractors, Inc. v.
California State Council of Carpenters, 459 U.S. 519, 539 (1983). Although the Fourth Circuit
did not follow the consumer-or-competitor rule, Novell, 505 F.3d at 311, courts in the Tenth
Circuit are not bound by the Fourth Circuit's holding because the law of the case doctrine "`is
discretionary rather than mandatory.'" Daviscourt v. Columbia State Bank, 2009 U.S. Dist.
LEXIS 16815, at *6 (D. Colo. Feb. 20, 2009) (quoting Homans v. City of Albuquerque, 366 F.3d
900, 904 (10th Cir. 2004)); see also Wessel v. City of Albuquerque, 463 F.3d 1138, 1143 (10th
Cir. 2006) ("`Unlike res judicata, the [law of the case doctrine] is not an inexorable command,
but is to be applied with good sense.'") (citation omitted).
In Reazin, M.D. v. Blue Cross & Blue Shield of Kansas, Inc., 899 F.2d 951, 963
(10th Cir. 1990), the Tenth Circuit cited Blue Shield v. McCready, 457 U.S. 465 (1982) for the
proposition that "an antitrust plaintiff need not necessarily be a competitor or consumer" to
establish antitrust injury, but subsequently in Elliott, the Tenth Circuit endorsed the consumer-or-
-125-
competitor rule, stating that "`[t]he requirement that the alleged injury be related to
anticompetitive behavior requires, as a corollary, that the injured party be a participant in the
same market as the alleged malefactors.'" 407 F.3d at 1125 (citing R.C. Dick Geothermal Corp.
v. Thermogenics, Inc., 890 F.2d 139, 148 (9th Cir. 1989) (en banc)). In Elliott, the Tenth Circuit
upheld the district court's opinion that antitrust injury was not established where plaintiff was
"neither a consumer of [defendant's] products, nor a competitor." Id. Under the consumer-or-competitor rule, Novell's claim is barred.
VI. Novell Released Any Claim for Harm to PerfectOffice.
Novell cannot base its claim on products not referred to in the Complaint because
under the terms of its November 8, 2004 settlement agreement with Microsoft, any claims not
pled in Novell's Complaint were released. That settlement agreement released Microsoft from
"any and all Claims that Novell ever had or has as of the date of this Agreement in law or in
equity, known or unknown, of any kind whatsoever (including without limitation any antitrust or
similar Claims of any kind)" with the exception of the following: "(i) patent Claims,
(ii) Revived Claims,
68 and (iii) the Claims set forth in the draft WordPerfect complaint . . . ."
(November 8, 2004 Settlement Agreement, at § 2(a), Holley Decl. Ex. A.) The third exception
refers to a draft complaint that was provided to Microsoft shortly before execution of the
settlement agreement, which turned out to be identical to the Complaint filed in this action on
November 12, 2004. Thus, any claim not asserted in the Complaint was released.
-126-
Consequently, Novell cannot now base its claim on harm to PerfectOffice, nor can
Novell now base its claim on harm to products such as its Soft Solutions document management
system and QuickFinder search engine, which are mentioned nowhere in Novell's Complaint.
A. Novell Released Any Claim for Harm to PerfectOffice.
Novell cannot base its claim on harm to PerfectOffice because any such claim
was released. Novell's Complaint alleged that the withdrawal of support for the namespace
extension APIs caused harm to WordPerfect and Quattro Pro. (E.g., Compl. ¶¶ 5, 75, 153.)
Indeed, the Complaint never mentions PerfectOffice in reference to the namespace extension
allegations and mentions PerfectOffice only twice in passing. (Compl. ¶¶ 81, 117.) In addition,
in defining the markets relevant to its case, Novell pled that "[t]hree markets are relevant to this
action: the market for Intel-compatible PC operating systems [in which Microsoft Windows had
monopoly power], the market for word processing applications [in which WordPerfect
competed], and the market for spreadsheet applications [in which Quattro Pro competed]."
(Compl. ¶ 24.) No mention was ever made of any suite market--the market in which
PerfectOffice competed. Moreover, the Complaint defined the term "office productivity
applications" to refer to "[w]ord processing and spreadsheet applications."69(Compl. ¶ 24.)
Prior to trial, Microsoft moved in limine to exclude Novell's theory that
PerfectOffice, alone or in combination with Java and Netscape, was a middleware threat to
Microsoft's PC Operating system monopoly. On October 6, 2011, this Court granted
-127-
Microsoft's motion, ruling that any middleware theory based on PerfectOffice is "a separate
claim which I don't think can be asserted" because it "was released." (October 6, 2011 Hearing
Tr. at 65.) When Novell attempted to resurrect the theory at trial during the examination of
Alepin, the Court instructed Novell, "Don't talk about PerfectOffice at this point, in light of my
prior ruling." (Nov. 9 Trial Tr. at 1409-10.) By the same token, Novell released any claim for
harm to PerfectOffice under the terms of the November 2004 settlement agreement.70
B. Novell Released Any Claims for Harm to Other Products Not Pled in Its
Complaint.
The Complaint alleged that the October 3 Decision delayed the release of
WordPerfect and Quattro Pro (e.g., Compl. ¶¶ 5, 75, 153) and that "WordPerfect and Novell's
other office productivity applications" were "the victims of [Microsoft's] anticompetitive
conduct." 71 (Compl. ¶ 21; see also Compl. ¶¶ 45, 54, 56, 105-06.) With respect to the October
3 Decision, the Complaint alleged that Microsoft's conduct "forced Novell to develop a costly
and difficult solution, delaying the shipment of WordPerfect for Windows." (Compl. ¶ 98; see
also Compl. ¶¶ 74-75.) In other words, the Complaint alleged only that WordPerfect and
Quattro Pro--and no other products--were harmed by the October 3 Decision.
At trial, however, Novell's first witness, Harral, refuted Novell's allegation
entirely, testifying that "I don't know anything that WordPerfect word processor needed to do for
-128-
a NameSpace extension." (Harral, Oct. 20 Trial Tr. at 327; see also id. at 270 (testifying that
"[t]he question isn't about WordPerfect's products").) Harral further testified that WordPerfect
"had no need for the namespace extension APIs" in order "[t]o ship their product." (Harral, Oct.
24 Trial Tr. at 476-77.) Novell also introduced no evidence at trial that Quattro Pro needed those
APIs.72 Indeed, Novell conceded in its summation that "WordPerfect understood that you don't
edit documents within an explorer extension. That was not what Microsoft was selling and that
was not what WordPerfect was planning." (Novell Summation, Dec. 13 Trial Tr. at 5172.)
Novell's witnesses further testified that, although Novell did not need the
namespace extension APIs for WordPerfect and Quattro Pro, it wanted to use the namespace
extension APIs to embed its QuickFinder search engine, Soft Solutions document management
system, e-mail client, Presentations clip-art gallery, and a primitive FTP/HTTP browser directly
in the Windows 95 shell. (Harral, Oct. 24 Trial Tr. at 372-74; Richardson, Oct. 25 Trial Tr. at
613, 629-30, 638-42, 664.) Novell intended to display these products in the tree view of the
Windows Explorer and Windows 95 common file open dialog, even when WordPerfect and
Quattro Pro were not running. (Harral, Oct. 20 Trial Tr. at 268-70; Richardson, Oct. 25 Trial Tr.
at 636-38, 691-97.) Consistent with this testimony, Novell's summation focused specifically on
the use of the namespace extension APIs for its Soft Solutions document management product
(not WordPerfect or Quattro Pro), describing "document management systems" as "the exact
-129-
use[]" that Novell planned for the namespace extension APIs. (Dec. 13 Trial Tr. at 5172; see
also id. at 5175, 5183.)
73
Four of the five products Novell now asserts to be the bases of its claim--including the Soft Solutions document management product that was the focus of Novell's
summation--were mentioned nowhere in the Complaint, and not one of the five were mentioned
in any way in relation to the namespace extension APIs.74 Further, the Complaint never alleged
(nor could it) that any of these five products competed in any of the three markets that the
Complaint lists as relevant to this action. (Compl. ¶ 24.) Having failed to plead in its Complaint
that any of these five products were harmed by Microsoft's conduct, Novell released any such
claims in its settlement agreement with Microsoft on November 8, 2004.
VII. Novell Sold Its Claim to Caldera.
In 1996, Novell transferred to Caldera, Inc. "any and all claims or causes of
action" that were "associated directly or indirectly" with DR DOS, Novell's PC operating
system. (Asset Purchase Agreement between Novell, Inc. and Caldera, Inc., July 23, 1996
-130-
("Asset Purchase Agreement") § 3.1, Holley Decl. Ex. P.) Caldera thereupon brought suit
against Microsoft asserting antitrust claims pertaining to the PC operating system market and
alleging that Microsoft had unlawfully monopolized that same market. The Caldera lawsuit
settled in 2000, providing a substantial recovery to Caldera and to Novell. In return, Microsoft
obtained a broad release covering "the Novell Claims and all claims asserted, or that could have
been asserted in the Action." (Settlement Agreement between Microsoft Corporation and
Caldera, Inc., January 7, 2000 § 2, Holley Decl. Ex. Q.)
75
In its November 13, 2009 cross-motion for summary judgment, Microsoft argued
that the Caldera release and the language of the Asset Purchase Agreement encompassed
Novell's present claim because the assignment to Caldera included all claims arising out of
Microsoft's allegedly anticompetitive conduct in the market in which DR DOS competed (the
PC operating system market), and the release covered "all claims asserted, or that could have
been asserted" in the Caldera action.
This Court agreed in 2010 and dismissed Novell's claim. 699 F. Supp. 2d at 739.
On appeal, Novell argued that its "DOS Products" were "entirely distinct" and in "different lines
of business" from WordPerfect and Quattro Pro, and that "the antitrust claims for injury to the
DOS Products were entirely distinct from those for injury to" WordPerfect and Quattro Pro.
(Novell Brief to the Fourth Circuit, Case No. 10-1482, Dkt. #19, Aug. 6, 2010, at 39-40.) The
Fourth Circuit accepted Novell's argument, and held that "[t]he mere existence of a possible
-131-
conceptual link between the DOS products and those applications does not mean that the
Agreement divested Novell of the claims alleged in Count I." 429 F. App'x at 261.
At trial, however, Novell made clear that its claim was intertwined with the
claims it sold to Caldera in 1996. Professor Noll repeatedly relied on and referred to the
presence of WordPerfect on the DOS platform and included WordPerfect's installed base on
DOS operating systems, saying that for his middleware theory the "relevant part is the installed
base of existing, running personal computers that are using a Microsoft operating system, all
right, or Microsoft compatible operating system" such as DR DOS. (Noll, Nov. 15 Trial Tr. at
1923-25.) In other words, Novell expressly relied on the success of its products on DOS
operating systems in order to support its theories of harm to the PC operating system market.76
As a result, the law of the case doctrine does not apply because the "evidence in a
subsequent trial is substantially different" than the evidence Novell presented at summary
judgment. Wessel, 463 F.3d at 1143 (quotation and citation omitted). Indeed, the Fourth
Circuit's interpretation of the Asset Purchase Agreement is no longer binding because "[w]hen
the record changes, which is to say when the evidence and the inferences that may be drawn
from it change, the issue presented changes as well." Jackson v. State of Alabama State Tenure
Commission, 405 F.3d 1276, 1283 (11th Cir. 2005); see also Haynes Trane Service Agency, Inc.
v. American Standard, Inc., 573 F.3d 947, 956-57 (10th Cir. 2009).
Novell's present claim is "associated directly or indirectly" with the claims
Novell sold to Caldera. Indeed, as this Court observed at trial, to the extent Novell "wanted to
-132-
back into the DOS system . . . then this claim is barred by . . . the asset purchase agreement."
(Oct. 24 Trial Tr. at 561-62.)
VIII. Novell's Claim Is Barred by the Statute of Limitations.
"Any action to enforce any cause of action" arising under the Clayton Act "shall
be forever barred unless commenced within four years after the cause of action accrued." 15
U.S.C. § 15b. The conduct about which Novell complains occurred ten years before Novell filed
its Complaint in 2004. There is no dispute that Count I is time-barred unless it was tolled by the
Government Case pursuant to 15 U.S.C. § 16(i). (E.g., Compl. ¶¶ 14-23.) Indeed, in dismissing
Counts II through V, the Court explained that those counts alleged harm to competition in
purported markets for word processing and spreadsheet software--"claims that were never
asserted by the government." Novell, Inc. v. Microsoft Corp., 2005 U.S. Dist. LEXIS 11520, at
*11 (D. Md. June 10, 2005). The Fourth Circuit affirmed. Novell, 505 F.3d at 320-23.
77
The test for determining the applicability of Section 16(i) is whether the matters
complained of in a private antitrust claim "bear a real relation" to the matters "complained of in
the government suit." Leh v. General Petroleum Corp., 382 U.S. 54, 59 (1965). As the Fourth
Circuit held in affirming dismissal of Counts II through V, the tolling provision should not be
construed "to permit private plaintiffs to `sit on their rights' and to assert, years after the
traditional statute of limitations has run, `claims so much broader than those asserted by the
government that they open entirely new vistas of litigation.'" Novell, 505 F.3d at 321-23
-133-
(quoting Novell, 2005 WL 1398643, at *5). As a result, "care must be exercised to insure that
reliance upon the government proceeding is not mere sham." Leh, 382 U.S. at 59.
Although Count I was "ingeniously designed to survive Microsoft's anticipated
limitations defense" by virtue of the cross-market nature of this claim, Novell, 699 F. Supp. 2d at
736, the evidence at trial established that Novell's claim bears no "real relation" to the claims
asserted in the Government Case. Leh, 382 U.S. at 59.
First, Novell repeatedly contended, both in summation and in oral argument on
Microsoft's Rule 50(a) motion for judgment as a matter of law, that what mattered was harm to
competition in applications markets. In summation, Novell's lawyer argued to the jury that
Microsoft devised a plan as early as 1993 to "gain a very significant lead over Microsoft's
applications competitors," supposedly recognizing that otherwise "Microsoft's Word and Excel
would be forced to battle against their competitors on even turf." (Dec. 13 Trial Tr. at 5163-64.)
Novell's counsel also told the jury that PX 1 showed that Microsoft's October 3 Decision was
"purely predatory action unrelated to any pro-competitive purpose" because Gates made the
decision to "give Office a real advantage." (Id. at 5184; see also id. (arguing that the Hood
Canal plan and the withdrawal of support for the namespace extension APIs were "all for the
purpose of disadvantaging Microsoft's application competitors").) Novell's counsel also argued
in summation that the October 3 Decision was not "consistent with competition on the merits"
because Gates "wanted to withhold this technology for Office 96, to deny Lotus the opportunity
to compete on the merits" and to ensure that Word and Excel would not be "forced to compete
on a level playing field." (Id. at 5216.) Novell even told the jury that "[y]ou won't find any
benefits to consumers in this decision," because "[i]f Microsoft had acted in a pro-competitive
-134-
manner, consumers would have had a choice of office productivity suites when Windows 95
came out." (Id. at 5216-17.)
Similarly, in oral argument on Microsoft's Rule 50(a) motion, Novell's lawyer
argued that Microsoft withdrew support for the namespace extension APIs "for the predatory
purpose of assuring that WordPerfect and Notes would fall behind." (Nov. 18 Trial Tr. at 2571.)
Throughout the trial, the Court expressed its "impatience" with Novell's theory, because "[t]his
case is not about trying § about the dominance that Word might have obtained over WordPerfect.
And it's not. That claim is time barred . . . ." (Oct. 24 Trial Tr. at 561; see also Oct. 27 Trial Tr.
at 939 ("Novell's apparent ideological position is to claim that they were attempting to
monopolize the Office suite market translates into them trying § you know, that that makes it the
same claim as trying to monopolize, maintain a monopoly in the operating system market. I
don't see that."); Oct 25 Trial Tr. at 575-76 ("It may be that Microsoft was using its knowledge
of Windows 95 and restricting what it was giving to competitors, application competitors so that
it could make Word and Office more dominate [sic] respectively in the word processing and
Office suite market. I understand that. But that is not the claim here. The claim here is
different. It has to work to the operating system.").) Although elsewhere Novell tried to wrap its
claim in the PC operating system flag, it injected into the trial this separate argument about harm
to competition in markets for word processing and spreadsheet software.
Second, the theory on which Novell tried the case differs significantly from (and
bears no "real relation" to) the theory of the Government Case. With respect to Novell's
franchise applications theory, although Novell's Complaint alleged that "[a]s found by the courts
in the Government Suit, . . . Microsoft's monopoly share of the Intel-compatible PC operating
systems market is protected by a barrier to entry arising out of the much greater number of
-135-
applications that operate only with Windows" (Compl. ¶ 43), Novell's theory at trial rested on
the notion that a mere three products could enable a rival operating system to surmount the
applications barrier to entry.
Although Novell's Complaint embraced the definition of middleware set out in
Finding 28 of the Government Case (Compl. ¶ 48 (quoting Finding of Fact 28)), Novell opposed
Microsoft's repeated requests that the Court instruct the jury that, in order to pose a threat to
Microsoft's operating system monopoly, a middleware product had to "expose[] enough APIs to
allow independent software vendors (`ISVs') profitably to write full-featured personal
productivity applications that rely solely on those APIs"--language drawn directly from Finding
of Fact 28. Novell went so far as to concede that giving such an instruction "would be directing
a verdict on that portion of our theory." (Dec. 15 Trial Tr. at 5436-37, 5439.) By seeking at trial
to vary from the definition of middleware used in the Government Case (and that was
incorporated explicitly into the Complaint), the allegations that saved the Complaint from
dismissal were thus discarded at trial. Because Novell's theory of harm to competition in the PC
operating system market was not "based in whole or in part" on the Government Case, Count I is
time-barred.
-136-
CONCLUSION
Microsoft respectfully requests that the Court enter judgment in Microsoft's favor
as a matter of law.
Dated: February 3, 2012
Respectfully Submitted,
/s/ James S. Jardine
James S. Jardine (A1647)
RAY QUINNEY & NEBEKER
[address, phone]
David B. Tulchin
Steven L. Holley
Sharon L. Nelles
Adam S. Paris
SULLIVAN & CROMWELL LLP
[address, phone, fax]
Steven J. Aeschbacher (A4527)
MICROSOFT CORPORATION
[address, phone, fax]
Attorneys for Microsoft Corporation
________________________
1
The Complaint alleged that the withdrawal of support for the namespace extension APIs
caused harm to WordPerfect and Quattro Pro but made no such allegation as to PerfectOffice.
(E.g., Compl. ¶¶ 5, 75, 153.) The Complaint also asserted that "[t]hree markets are relevant to
this action: the market for Intel-compatible PC operating systems, the market for word
processing applications [in which WordPerfect competed], and the market for spreadsheet
applications [in which Quattro Pro competed]." (Compl. ¶ 24.) Indeed, the Complaint defined
the term "office productivity applications" to refer to "[w]ord processing and spreadsheet
applications." (Compl. ¶ 24.) As a result, because Novell and Microsoft entered into a
settlement agreement four days before the Complaint was filed that released Microsoft from "any
and all Claims that Novell ever had or has as of the date of this Agreement . . . except for . . . the
Claims set forth in the draft WordPerfect complaint" (Nov. 8, 2004 Settlement Agreement, at
§ 2(a), attached as Exhibit A to the Declaration of Steven L. Holley, executed on February 3,
2012 ("Holley Decl.")), Novell released any claim for harm to PerfectOffice. See pp. 127-28,
infra. Accordingly, Novell's only claim is for alleged harm to WordPerfect and Quattro Pro.
Microsoft will nevertheless refer to all three products in portions of this memorandum so that
there is no doubt about the lack of merit to Novell's claim.
2
This third requirement was set forth in the binding Findings of Fact and incorporated
explicitly into the Complaint. See pp. 75-81, infra. The Findings of Fact that were given
collateral estoppel effect are attached as Exhibit B to the Holley Declaration.
3
On October 6, 2011, the Court granted Microsoft's motion in limine seeking to exclude
evidence of Novell's theory that PerfectOffice, alone or in combination with Netscape Navigator
or Sun's Java, was a form of "middleware" that threatened Microsoft's monopoly in the PC
operating system market. Moreover, any claim that PerfectOffice (alone or in combination with
other products) was middleware of the sort that might affect competition in the PC operating
system market was released. See p. 5 n.1, supra, and p. 14 n.12, infra.
4
All trial exhibits cited in this memorandum were admitted into evidence at trial, except
for Microsoft's Demonstrative Exhibits 218, 241, 301 and 307-311, to which reference is made
on pages 45, 64-67 and 82. The documents cited in this memorandum that were admitted at trial
are included in the accompanying Appendix. Other materials are attached to the Holley
Declaration.
5
Novell's definition of a delay that harmed its products was a delay of more than about 60
days after the August 24, 1995 release of Windows 95. (E.g., Oct. 18 Trial Tr. at 34-35
(Novell's opening statement: "The evidence will show that one of Novell's primary objectives
was to have PerfectOffice for Windows 95 on the store shelves within 30 to 60 days of the
release of Windows 95."); Warren-Boulton, Nov. 17 Trial Tr. at 2418 ("Novell's goal, was to get
it out within 30 or 60 days and that is my but-for world.").)
6
There is no dispute that the other two exceptions are inapplicable here.
7
The Complaint alleged that Microsoft's conduct caused harm to those two products and
no other. (Compl. ¶¶ 24, 72-78, 153.) It included no claim for any harm caused to
PerfectOffice. See p. 5 n.1, supra, and pp. 127-28, infra.
8
Novell also alleged in Count I that Microsoft "willfully and wrongfully obtained" its PC
operating system monopoly (Compl. ¶ 153), but dropped this allegation before trial (see
Proposed Pre-Trial Order, filed Sept. 27, 2011, Dkt. #152, at 2). The only issue at trial was
whether Microsoft unlawfully maintained its monopoly in the PC operating system market. (See
Verdict Form, Questions 4 and 5, provided to the jury on Dec. 14, Holley Decl. Ex. C.)
9
Professor Noll testified that the anticompetitive harm caused by Microsoft's conduct was
the delay in the release of its products for Windows 95--not degraded functionality. (Noll, Nov.
15 Trial Tr. at 1880-81 (agreeing that "there can't be any harm to competition under the facts here, if the conduct at issue, the decision to withdraw support for the namespace extension APIs, did not cause any delay"); see also Noll, Nov. 14 Trial Tr. at 1838-40.)
10 It was undisputed at trial that WordPerfect and Quattro Pro did not need to use the
namespace extension APIs to be compatible with Windows 95. Ronald Alepin testified that
Novell could launch its applications from either the Start menu or icons on the desktop without
use of the namespace extension APIs and that Novell's applications did not require the
namespace extension APIs to create a folder in the Windows 95 file system that would be the
default location for storing files created using WordPerfect and Quattro Pro or to place files on
the desktop that would automatically open the Novell application when selected by the user.
(Alepin, Nov. 10 Trial Tr. at 1577-79.) At argument on Microsoft's Rule 50(a) motion, Novell's
counsel falsely stated that "We never suggested that we were never able to run WordPerfect on
Windows 95." (Nov. 18 Trial Tr. at 2582-83.) The Complaint is to the contrary.
11 This allegation about incompatibility was false. Alepin acknowledged that Windows 95
was backward compatible and that versions of WordPerfect and PerfectOffice written to
Windows 3.0 and 3.1 would also run on Windows 95. (Alepin, Nov. 10 Trial Tr. at 1580-81; see
also Gates, Nov. 21 Trial Tr. at 2754.) No witness testified (or even suggested) the contrary.
12 On September 21, 2011, Microsoft filed a motion in limine seeking to exclude evidence
of a middleware theory that Novell had advanced before trial--that PerfectOffice, alone or in
combination with Netscape Navigator or Sun's Java, was a form of middleware that threatened
Microsoft's monopoly in the PC operating system market. (Motion in Limine to Exclude
Evidence of Novell's New Middleware Theory, Dkt. #112.) Microsoft pointed out that Novell
never set out this theory in its Complaint and that, in the November 2004 settlement agreement
with Microsoft, Novell released all claims other than those set forth in the Complaint. On
October 6, 2011, this Court granted Microsoft's motion, ruling that Novell's middleware theory
concerning PerfectOffice is "a separate claim which I don't think can be asserted" because it
"was released." (October 6, 2011 Hearing Tr. at 65; see also Nov. 9 Trial Tr. at 1409-10
(excluding evidence about PerfectOffice's middleware capability and instructing Novell "[d]on't
talk about PerfectOffice at this point, in light of my prior ruling").
13 In opposing Microsoft's summary judgment motion, Novell asserted that Microsoft
engaged in three allegedly anticompetitive acts: "(1) Withdrawing access to information about,
and otherwise changing course regarding, the Windows 95 namespace extensions, thereby
delaying and impairing Novell's development of the versions of WordPerfect and Quattro Pro
that were optimized for Windows 95"; "(2) Misleading Novell about Windows 95 print
functionality, thereby increasing WordPerfect's costs and decreasing its functionality"; and
"(3) Refusing to grant a Windows 95 logo license for certain Novell software applications." 699
F. Supp. 2d at 743. At trial, Novell limited its claim to Microsoft's "withdrawal of support for
the namespace extension APIs" and made clear that it had abandoned its allegations with respect
to other conduct. (Novell's Objections and Suggestions Regarding the Court's Tentative Jury
Instructions and Verdict Form, filed Dec. 7, 2011, Dkt. #348, at 1; see also Novell's
Memorandum Regarding Proposed Final Jury Instructions and Verdict Forms, filed Dec. 5, 2011,
Dkt. #336, Ex. A at 4-5.)
14 Windows 3.1, the version of Windows prior to Windows 95, was a 16-bit operating
system. (Noll, Nov. 15 Trial Tr. at 1882.)
15 WordPerfect Corporation was acquired by Novell on June 24, 1994. See p. 43, infra.
16 The shared code group was headed by Tom Creighton. (PX 372, Business Application
Development Organization, Feb. 16, 1995, at 2.) Neither Creighton nor Jim Johnson (who
reported to Creighton during the relevant time) testified at trial. (Id.) Neither did Bruce
Brereton, who was Creighton's boss. (Id. at 1-2.) The shared code team was responsible for
writing features that would run across Novell's applications. (Harral, Oct. 20 Trial Tr. at 206,
209-10.) Harral explained that shared code allowed users to have the "same experience" when
using different products so that "if [a user] learned one product [he or she] would know how to
operate all of the other products in a similar way." (Id. at 206-07.) The shared code group was
responsible for developing, among other features, "all of the file handling," which included "how
you open a file in WordPerfect [and] how do you save it." (Id. at 209.)
17 The Complaint made no mention of Novell's Soft Solutions document management
system, e-mail client, Presentations clip-art gallery or FTP/HTTP browser, and the Complaint's
two references to Novell's QuickFinder search engine gave no indication that the October 3
Decision harmed QuickFinder in any way. (See Compl. ¶¶ 94-95.) In any lawsuit, the complaint
ordinarily defines the issues to be tried and the scope and nature of a plaintiff's claims. Here, the
Complaint is of far greater significance than in the usual case because Novell released Microsoft
from all claims other than those set forth in the Complaint. The Court has already dismissed any
claims related to the e-mail client, GroupWise. See p. 127 n. 69 & p. 130 n.72, infra.
18 Novell acquired WordPerfect Corporation on June 24, 1994. (Compl. ¶ 37.) Both Novell
and WordPerfect received the M6 beta pursuant to a nearly identical license agreement. (DX 18;
DX 19.)
19 In his opening statement, Novell's counsel stated that the evidence would show that
Microsoft adopted a plan in June 1993, referred to by Novell as the "Hood Canal plan," whereby
Microsoft would withhold shell extensibility from Windows 95 and ship an extensible shell with
Microsoft Office after the release of Windows 95 so as to give Office an advantage. (Oct. 18
Trial Tr. at 46-50.) With no evidence at trial to support that theory, see pp. 114-15, infra,
Novell's counsel conceded in his closing rebuttal that the "[t]he [Hood Canal] plan didn't go
forward because . . . what happened is the executives [in the] systems group said, no, no, no,
we're not doing that . . . . [T]he Office plan set forth in the radical extreme didn't go forward."
(Dec. 13 Trial Tr. at 5324-25.)
20 Harral testified that "[t]he designation of beta in the software industry is that it is for
ferreting out problems to be fixed, not for changes in the features." (Harral, Oct. 20 Trial Tr. at
336.) He did not offer further explanation for how this distinction could be drawn in practice.
21 Nakajima explained that having the namespace extension APIs running in the same
process as the Windows 95 shell was akin to having a pre-installed GPS device integrated into
the dashboard computer of a car, which is "simpler" and "less bulky" than having a separate
aftermarket GPS device sitting on the dashboard. (Nakajima, Dec. 1 Trial Tr. at 3754-56.)
However, the fact that it is "fully integrated" is also a "risk," because if an aftermarket GPS
malfunctions, the driver can still "operate the car," whereas if an integrated "GPS system
crashes, [the driver] cannot do anything." (Id. at 3756.) Nakajima also compared the risk to
permitting other chefs to come into a commercial kitchen, risking fires and contamination that
impacts the restaurant's ability to serve any food at all. (Id. at 3759-61.)
22 Gates also explained that he made reference in PX 1 to Notes, which was Lotus'
e-mail and workgroup collaboration product, because he thought that once the "Cairo level of
integration" had been reached, Microsoft "would be able to reconceptualize how e-mail and
WorkGroup was done." (Gates, Nov. 21 Trial Tr. at 2804.) This is because some of the
functionality that Cairo was expected to provide was the ability to run queries across applications
and file systems, which would only be useful for e-mail clients or "any type of WorkGroup thing
where you're dealing with a rich set of information." (Id. at 2805.) Thus, he explained that his
reference in PX 1 to WordPerfect, alongside Notes, was meant to be a reference to GroupWise,
Novell's e-mail client, and not to the word processing application. (Id. at 2804.)
23 This was, of course, during the time when Microsoft was debating internally about this
issue. (Struss, Nov. 28 Trial Tr. at 3264.)
24 On November 7, 1994, Paul Maritz wrote to Gates saying that "4 groups [were] using
these [namespace extension] interfaces," and listing them as "Capone, Marvel, Stac, Symantec."
(DX 82.) Novell was not on the list.
25
Nor is there any evidence of any intent on Microsoft's part to mislead Novell or anyone
else by releasing the M6 beta with the namespace extension APIs and then later withdraw
support for them.
26 Harral testified at trial that after the October 3 Decision, Microsoft's Premier Support
group was "starting to give [Novell] less and less information about the shell in general."
(Harral, Oct. 20 Trial Tr. at 345.) On direct examination, Harral testified that he spoke with
Premier Support three times, although he was unable to provide a date--or even a month--in
which any such call took place. (Id. at 329-31.) On cross examination, when asked to provide
the names of any people in Microsoft's Premier Support group with whom he spoke, Harral was
entirely unable to do so. (Harral, Oct. 24 Trial Tr. at 397, 399, 414.) Novell introduced no
evidence at trial of any e-mail sent to Microsoft referring to any such call; no internal Novell e-
mail or memorandum indicating that such a call took place or complaining about Microsoft's
lack of cooperation; and no document of any kind that could in any way confirm or even imply
that there was ever any such contact between the two companies about the namespace extension
APIs. And, other than Harral's vague testimony, Novell introduced no evidence at trial--
documentary or testimonial--reflecting communications between Novell and Microsoft's
Premier Support group.
27 Alepin testified that a pre-release version of a product code-named Athena, an internet
mail and newsgroup reader, called the namespace extension APIs. (Alepin, Nov. 9 Trial Tr. at
1435.) When confronted with the technical analysis conducted by Bennett showing that Athena
did not call the namespace extension APIs, Alepin admitted that he did not recall what analysis
he had done to support his testimony. (Alepin, Nov. 10 Trial Tr. at 1646; see also Bennett, Dec.
12 Trial Tr. at 4990-91 (explaining that "Athena did not use the NameSpace extension APIs").)
28 Accordingly, Struss' status reports to Microsoft executives in the months after October
1994 reflected no Novell complaints on that subject. (DX 2, at MX 6062581; DX 92, E-mail
from Brad Struss, Dec. 15, 1994, at 2.) Had Novell raised any complaints, Struss "absolutely"
would have included them in his status reports. (Struss, Nov. 28 Trial Tr. at 3277-78.)
29 There is ample record evidence that Novell employees frequently communicated with
Gates and Microsoft executives such as Brad Silverberg (Vice President of Systems Division),
Brad Chase (Vice President of Marketing), Bob Kruger (Systems Marketing and Standards
Director), David Cole (Group Program Manager for the Chicago Team) and Doug Henrich
(DRG Director). (Frankenberg, Nov. 7 Trial Tr. at 1027-29, 1125; Frankenberg, Nov. 8 Trial Tr.
at 1172, 1174-76; DX 22; DX 161, E-mail from Brad Silverberg, Nov. 18, 1993; DX 155,
Memorandum from Ryan Richards, Jan. 12, 1995; Jan. 8, 2009 Henrich Deposition, Nov. 8 Trial
Tr. at 1307-09.)
30
Frankenberg offered vague testimony on direct examination (Frankenberg, Nov. 7 Trial
Tr. at 1029-30), but cross examination clarified that while Frankenberg discussed the general
subject of "undocumented APIs," Frankenberg "never specif[ied] which APIs [he was] talking
about" (id. at 1117-18). Indeed, Frankenberg testified that he did not recall ever "sa[ying] to Mr.
Gates the problem is the namespace extension APIs," and that he also did not recall ever sending
or seeing any letter or e-mail to Gates that mentioned the namespace extension APIs. (Id. at
1118-19.) That is unsurprising because Frankenberg did not even know at the time what the
namespace extension APIs were. (Id. at 1127.)
31 The absence of any such document is particularly remarkable because, as the Court is
aware, despite its anticipation of litigation against Microsoft since the early 1990s, Novell's
preservation of relevant evidence has been entirely selective. "[T]here [wa]s, and there has been
maintained, a file of documents that has been referenced as Microsoft's bad acts," including
documents "that go back to probably about the 1994 time period after I had joined Novell, which
would have been the latter half of 1994." (Oct. 30, 2008 Rule 30(b)(6) Deposition of Novell,
Inc. by James F. Lundberg at 51-53, Holley Decl. Ex. D; see also Affidavit of Ryan Richards,
sworn on April 23, 2009, at p. 2 § 5, Holley Decl. Ex. E ("Since at least 1992--when I was
working for WordPerfect--I had been investigating Microsoft's unlawful conduct and had
determined that litigation was the likely avenue to seek redress against Microsoft.").)
Nevertheless, no document exists to support Novell's allegations concerning the purported
impact of the October 3 Decision on Novell.
32 By 1994, Novell recognized "how rapidly suites [were] overtaking the stand alone
Windows word processing market." (DX 9, WordPerfect for Windows Business Review
Exercise, July 15, 1994, at 6.) Indeed, according to Novell's own estimates, unit sales in the
suite market grew from "approximately 800,000 units" in 1992 (DX 267, at 1) to "exceeding 3
million" units in 1994 (DX 223, Storm Market Requirements Document, March 23, 1995, at 11).
Meanwhile, Novell estimated that revenue in the standalone word processor market fell from
$1.7 billion in 1993 (DX 5, Memorandum from Ad Rietveld to Bob Frankenberg, July 21, 1994,
at NOV00542218) to $1.04 billion in 1994 (DX 224, Storm Market Requirements Document,
April 14, 1995, at 47). According to a WordPerfect for Windows Business Review Exercise, as
of April 1994, 72.3% of word processors sold in North America for the Windows platform were
sold as part of suites. (DX 9, at 6.) Frankenberg agreed that "the market was moving quickly
from stand-alone products to suites" in 1994. (Frankenberg, Nov. 7 Trial Tr. at 1068.) Revenue
in the suite market continued to grow from "nearly $1 billion" in 1994 (DX 223, at 11) to "about
$3.8 billion" in 1996 (Hubbard, Dec. 6 Trial Tr. at 4484; see also Microsoft's Demonstrative
218, shown at trial on Dec. 6, Trial Tr. at 4444, Holley Decl. Ex. F (showing that, in 1995,
industry-wide revenue from the sale of suites was $3.12 billion while revenue from the sale of
word processors and spreadsheets sold as standalone products had dropped to $994 million)).
33 See, e.g., PX 297, InfoWorld Review, "PerfectOffice nearly lives up to its name," April
24, 1995, at NOV 00012602.
34 Noll testified that PerfectOffice 3.0's market share in the first seven months of 1995 was
"roughly" 8%. (Noll, Nov. 15 Trial Tr. at 1915.) Other data show even lower market share
numbers: 6.9% in Q1 of 1995, 6.1% in Q2 and 1.7% in Q3. (Hubbard, Dec. 6 Trial Tr. at 4483
(using market data contained in an independent market research firm's report).)
35 The delay in releasing Quattro Pro also delayed WordPerfect. A Novell project proposal
recognized that releasing WordPerfect before Quattro Pro would earn Novell a reputation as a
"standalone provider only" and recommended that both products be released together, even if a
simultaneous release would cause delay. (DX 211, Project Proposals for `Storm,' at NOV-
B01491220; see also DX 221, E-mail from Bruce Brereton, March 1, 1995, at NOV-B13528783
("After further discussion, we feel it will be much better to have WP . . . on the same schedule as
Storm [PerfectOffice]. . . . [W]e have moved the Storm RTM date back by one month (to
December 30th) and have put WP on the same time-line as Storm.").)
36 When Quattro Pro developers finally began work in 1995 on a version for Windows 95, they continued to struggle with localization issues, just as they had with Quattro Pro 6.0. (Bushman, Nov. 28 Trial Tr. at 3185-86.) Bushman testified that because of localization issues, “there were proposals put forth” to “ship[] the English Quattro Pro” and “get[] the localized version later,” but Bushman explained, “this just would not work” because at the time, 60% of Novell’s revenue came from international sales and many business customers wanted to buy fully-localized suites. (Id. at 3141-42, 3148-49, 3185-86.)
37 "Storm" was the code name for PerfectOffice for Windows 95. (Gibb, Oct. 26 Trial Tr.
at 790.)
38 Indeed, LeFevre made the obvious observation that "the challenge when you're shipping
a suite of products" is that "they kind of all have to be done at the same time." (LeFevre, Dec. 2
Trial Tr. at 4046-47; see also Harral, Oct. 24 Trial Tr. at 444-45 (agreeing Quattro Pro "needed
to be ready" for PerfectOffice to ship); see also Frankenberg, Nov. 7 Trial Tr. at 1143.)
39 Novell sold its Three Products to Corel on March 1, 1996. (DX 382, Novell Form 10-K
for Fiscal Year Ended Oct. 26, 1996, filed Jan. 2, 1997, at 2.)
40 Of course, as Frankenberg recognized, there was some risk in using unsupported APIs,
because those APIs might not be included in later versions of Windows. (Frankenberg, Nov. 7
Trial Tr. at 1133-34.) Microsoft advised ISVs not to use the namespace extension APIs for
exactly this reason. (DX 3, at MX 6055840-41.) There is no evidence of any deliberative
process at Novell concerning whether or not to choose Option 1, in contrast to the existence of
formal memoranda that addressed less important decisions. See pp. 36-38, supra. Although
there was risk in Option 1--the chance that one day in the future Microsoft might remove the
namespace extension APIs from Windows--there was even more risk in choosing Option 3
(because of the foreseeable delays in Novell trying to write its own file open dialog). Moreover,
choosing Option 1 would have been entirely rational, for as the Court noted, Novell could have
"temporarily used the [namespace extension] API[s]" in order to get out a product right away,
while continuing to work on its own file open dialog that could be utilized in a subsequent
version of PerfectOffice. (Oct. 27 Trial Tr. at 924-25.) The delay of which Novell complains
was of its own making.
41 The Complaint makes no claim in this regard about PerfectOffice and thus any such
claim was released. See p. 5 n.1, supra, and pp. 127-28, infra.
42
To the extent that Novell seeks to stray from the allegations in its Complaint with regard
to the fundamental nature of the applications barrier to entry, its claim is barred for two reasons.
First, Novell released any claim based on any theory not expressly set forth in the Complaint.
See pp. 126-30, infra. Second, Novell's claim would also be barred by the applicable four-year
statute of limitations set forth in 15 U.S.C. § 15b because rather than "based in whole or in part"
on the Government Case, see 15 U.S.C. § 16(i), it would instead run contrary to the Government
Case. See pp. 133-36, infra.
43 It makes no sense to contend that WordPerfect's installed base on all PCs--rather than
WordPerfect's market share on Windows--is relevant to determine whether WordPerfect could
be a "franchise application." As Novell itself said at trial, "[t]he thrust of Novell's argument is
that its popular applications . . . offered competing operating systems the prospect of
surmounting the applications barrier to entry and breaking the Windows monopoly" because
"during the relevant time period, they were the dominant office productivity applications in the
market." (Novell's Memorandum Regarding Proposed Final Jury Instructions and Verdict
Forms, filed Dec. 5, 2011, Dkt. #336, at 4 (emphasis added).) WordPerfect's share of the total
installed base share on all platforms (as compared to its low market share on Windows) was a
result of WordPerfect's historic success on the character-based DOS platform, which was
becoming irrelevant by 1995. (Noll, Nov. 15 Trial Tr. at 1923-24; Hubbard, Dec. 6 Trial Tr. at
4441-43.) Indeed, Novell's internal documents explain that "only 30% of this WordPerfect for
DOS installed base is remaining with WordPerfect as they transition to a Windows word
processor." (DX 224, at 20; see also Hubbard, Dec. 6 Trial Tr. at 4443 ("WordPerfect has a . . .
very significant share of DOS, the older platform . . . [that] did not translate into similar market
share success on the Windows platform."); Peterson, Dec. 7 Trial Tr. at 4706-07.) Thus, by no
stretch of the imagination could the installed base of WordPerfect on the DOS platform be a
reasonable proxy for how "popular" WordPerfect was during the "relevant time period." In any
event, the market data Novell introduced into evidence established that, as of the end of 1994,
even when including the DOS platform, WordPerfect was present only on about 36% of all PCs
that had a word processor installed--including all versions on the Windows, DOS and OS/2
platforms. (PX 599A, at Table 13.)
44 In fact, even including WordPerfect's market share on the character-based DOS platform,
WordPerfect's market share was relatively small well before the release of Windows 95. During
the early 1990s, the PC operating system market shifted from DOS to Windows, with Windows
capturing 80% of the PC operating system market by 1993, and more than 90% by 1996. (Noll,
Nov. 15 Trial Tr. at 1929-30.) As a result of the market shift from DOS to Windows, by 1994,
WordPerfect had 35.9% of the installed base on Windows and DOS combined, and only 25.4%
of "[n]ew [s]hipments" on those platforms. (PX 599A, at 26; see also Microsoft's
Demonstrative 308, shown at trial on Dec. 13, Trial Tr. at 4752-54, Holley Decl. Ex. H (showing
that WordPerfect's market share on Windows and DOS combined was 35% in 1994, declining to
27% in 1995).)
45 Borland released Quattro Pro for Windows in 1992, and the product "pick[ed] up around
five percent of the market." (Murphy, Dec. 7 Trial Tr. at 4756.) In 1993, Quattro Pro for
Windows earned $59 million in revenue, which amounted to 6% of the standalone Windows
spreadsheet market. (See DX 5, Quattro Pro Business Review Exercise, July 15, 1994, at NOV
00542227.) Going forward, Quattro Pro's market share in the standalone Windows spreadsheet
market was "always relatively small." (Murphy, Dec. 7 Trial Tr. at 4756; see also Microsoft's
Demonstrative 309, shown at trial on Dec. 13, Trial Tr. at 4755, Holley Decl. Ex. I (showing that
Quattro Pro's market share of the standalone Windows spreadsheet market was 5.5% in 1992,
7.0% in 1993, 2.0% in 1994, and 1.8% in 1995).) Quattro Pro's market share of the spreadsheet
market on the Windows and DOS platforms combined was similarly small, with only 6% in
1993. (See DX 5 at NOV 00542227, 30-31; see also Microsoft's Demonstrative 310, shown at
trial on Dec. 13, Trial Tr. at 4758, Holley Decl. Ex. J (showing that Quattro Pro's market share
on Windows and DOS was 2% in 1994 and 1995).)
46 WordPerfect Corporation and Novell never released a suite product for the DOS
platform. (Frankenberg, Nov. 8 Trial Tr. at 1169.)
47
By 1994, the market had shifted from sales of standalone word processors and
spreadsheets to office suites. For example, Novell's own documents estimated that in 1994,
more than 70% of word processors sold in North America for the Windows platform were being
sold as part of a suite. See pp. 45-46 n.32, supra.
48 See also Microsoft's Demonstrative 241, shown at trial on Dec. 6, Trial Tr. at 4483,
Holley Decl. Ex. L (showing that PerfectOffice's market share just before the release of
Windows 95 was 6.9% in Q1 of 1995, 6.1% in Q2 and 1.7% in Q3).
49 Frankenberg testified that Tapestry was a code name for the "next generation" of
Novell's office suite. (Frankenberg, Nov. 7 Trial Tr. at 997-98.)
50 As shown above, the Court excluded evidence of Novell's theory that PerfectOffice,
alone or in combination with Netscape Navigator or Sun's Java, was a form of middleware that
threatened Microsoft's monopoly in the PC operating system market. See p. 14 n.12, supra. As
a result, Novell cannot claim that PerfectOffice (alone or with other products) was middleware of
the sort that might affect competition in the PC operating system market.
51 Professor Noll stated the third criterion differently, testifying that as long as an ISV relies
on some of the APIs exposed by middleware, competition could have been enhanced. (E.g.,
Noll, Nov. 15 Trial Tr. at 1919.) This is illogical and refuted by the binding Findings of Fact,
which Novell's Complaint incorporated as the definition of middleware. See pp. 75-81, infra.
52 Novell's fact witnesses sometimes used the word middleware to refer to software that sits
between an operating system and applications and exposes APIs. Specifically, Novell's
witnesses described Novell's software as "something that is produced that sits in the middle"
between an operating system and applications (Harral, Oct. 20 Trial Tr. at 234), and as "a layer
that [ISVs] use to build . . . applications" (Gibb, Oct. 26 Trial Tr. at 782-83). This effort to sow
confusion should be rejected for, as Alepin acknowledged, "[t]here's got to be more than just the
exposure of API's or the encapsulation of meaningful abstraction of API's. You need more" to
have any potential impact on competition. (Alepin, Nov. 9 Trial Tr. at 1461-62.)
53 "[T]he middleware theory is about exposing the same APIs set on different platforms, not
saying a word processor that runs on two different platforms. It's about can ISVs write to
multiple platforms by writing once to the APIs set in the middleware." (Murphy, Dec. 7 Trial
Tr. at 4779.) The version of the WordPerfect word processor written on the character-based
DOS platform is irrelevant for Novell's Middleware Theory because "the DOS version wouldn't
have supported Windows applications" and the "applications written to run on top of Windows
wouldn't run on the DOS version of WordPerfect." (Murphy, Dec. 8 Trial Tr. at 4917; see also
Noll, Nov. 15 Trial Tr. at 1946-47 (conceding that there was no evidence that Novell's purported
middleware was available in WordPerfect for DOS).) Similarly, the version of WordPerfect for
Linux that Corel released in the spring of 1996 (Noll, Nov. 14 Trial Tr. at 1850) was an older
version of WordPerfect that did not contain the same shared code as the version of WordPerfect
developed for Windows 95 and included in the PerfectOffice suite, and therefore WordPerfect on
Linux could not have served as a middleware threat to Windows. (Murphy, Dec. 8 Trial Tr. at
4914-16.)
54 PX 599A shows that, as of the end of 1994, the total installed base of PCs with word
processing applications installed was 40,980,000. Of that number, 14,760,000 PCs had
WordPerfect installed. (PX 599A, at Table 13.) This is about 36%.
55
Novell's attempt to use its DOS product to establish harm to competition in the PC
operating system market makes it clear that its claim is "associated directly or indirectly with"
the PC operating system claim it sold to Caldera in 1996 and is thus barred. See pp. 130-33,
infra.
56 Novell's opening brief to the Fourth Circuit in 2010 made this same point:
"`Middleware' is a term used to refer to software products that have the
capability to serve as platforms for software applications themselves. They
expose, or make available, their own APIs, and theoretically, software
developers could rely upon these APIs rather than Windows's APIs . . . ."
Novell, 505 F.3d at 308 n. 14 (citations omitted).
(Novell Brief to the Fourth Circuit, Case No. 10-1482, Dkt. #19, Aug. 6, 2010, at 16 n. 5
(emphasis added).)
57 Windows maintained at least a 90% share of that market throughout the 1990s. (E.g.,
Finding of Fact 35 ("Every year for the last decade, Microsoft's share of the market for Intel-compatible PC operating systems has stood above ninety percent."); see also Noll, Nov. 15 Trial
Tr. at 1929-30 & Microsoft's Demonstrative 120, shown at trial on Nov. 15, Trial Tr. at 1930,
Holley Decl. Ex. M; Murphy, Dec. 7 Trial Tr. at 4722-23 & Microsoft's Demonstrative 301,
shown at trial on Dec. 7, Trial Tr. at 4722, Holley Decl. Ex. N.)
58 Finding of Fact 28 states that "[c]urrently [November 1999] no middleware product
exposes enough APIs to allow independent software vendors (`ISVs') profitably to write full-featured personal productivity applications that rely solely on those APIs." Finding of Fact 32
emphasizes that it still "remains to be seen whether server- or middleware-based development
will flourish at all." See also Finding of Fact 29 ("It remains to be seen, though, whether there
will ever be a sustained stream of full-featured applications written solely to middleware APIs.").
In short, "these middleware technologies have a long way to go before they might imperil the
applications barrier to entry." (Finding of Fact 77.) Consistent with these Findings of Fact,
Professor Noll testified that in "the period of this case plus the government case," there "never
was" a middleware product that "ran on various operating systems." (Noll, Nov. 15 Trial Tr. at
1929; see also id. at 1920 (agreeing that "as of 1999, there had never been any middleware that
could imperil the applications barrier to entry").)
59 Although this Court and the Fourth Circuit rejected a similar argument at the summary
judgment stage, the time for submission of evidence, rather than pure theory, was at trial. The
evidence at trial, including Professor Noll's testimony, made clear that the October 3 Decision
alone caused no harm to competition. Even accounting for harm to Java and Netscape, Professor
Noll offered no specifics or data and merely waved his hand over the issue with generalized
opinion.
60 Microsoft itself did not use the namespace extension APIs in any Microsoft applications
that competed with Novell's Three Products. See pp. 33-34, supra.
61 Larsen recounted on cross examination that while WordPerfect was working on a version
of its software for the Macintosh operating system, Apple evangelized a particular feature of its
operating system and then withdrew support for that feature without providing any explanation
for its decision. (Larsen, Nov. 30 Trial Tr. at 3656-58.)
62 Struss testified that DRG "never made promises about what would be in a version of the
operating system. For those people who have been in the technology industry, there is a good
solid understanding that a beta release of a product may or may not be everything that is in that
final release. It is really trying to meet the quality standards and the ship dates desired [that]
impacts what is in a final release versus what is in a beta release." (Struss, Nov. 28 Trial Tr. at
3257.)
63 It was not until the spring of 1996 that a plan was in place at Microsoft to modify the
design of the Chicago shell for use with Windows NT: this plan mitigated the robustness
problems posed by the namespace extension mechanism in Windows 95 by "rearchitect[ing] the
process slightly . . . to separate the Desktop/taskbar process from the rest of the explorer
extensions that live in the shell namespace." (DX 131A, E-mail from Joe Belfiore to Andrew
Schulman, March 21, 1996, at 2; see also Belfiore, Dec. 5 Trial Tr. at 4292-93.) In October
1994, however, no such solution existed for Windows NT.
64
Having failed to plead a claim of deception in its Complaint, any such claim was released
by Novell's November 2004 settlement agreement with Microsoft. See pp. 126-30, infra.
65 Novell conflates withdrawal of support for the namespace extension APIs, which were a
small subset of the shell extensions in Chicago (Nakajima, Dec. 1 Trial Tr. at 3752), with the
Hood Canal presentation, which called for shipping Windows 95 without any shell extensibility.
(PX 61, Office Shell Ideas and Issues, dated July 3, 1993, at 1.) The evidence at trial was
uncontroverted that Chicago was designed with full shell extensibility (Gates, Nov. 21 Trial Tr.
at 2773; Muglia, Nov. 29 Trial Tr. at 3403-04), and that the Chicago shell remained fully
extensible after the decision to withdraw support for the namespace extension APIs. (Gates,
Nov. 21 Trial Tr. at 2773, 2776.)
66 At one point, Warren-Boulton argued that "the jury can say, well, these numbers assume
that the product would have been out by, say November, it was actually out in May, six months,
you know, and if you decide that it would in fact--as a factual matter if [PerfectOffice for]
Windows 95 had come out in January, I guess the simplist [sic] thing to do is to take one-sixth
off the damages." (Warren-Boulton, Nov. 17 Trial Tr. at 2424.) "This is precisely the type of
`speculation or guesswork' not permitted for antitrust jury verdicts." MCI Communications,
708 F.2d at 1162 (quoting Bigelow, 327 U.S. at 264). The Court recognized the impropriety of
what Warren-Boulton was suggesting and struck the above-quoted testimony. (Warren-Boulton,
Nov. 17 Trial Tr. at 2424.)
67 The Fourth Circuit's determination that Novell adequately pled antitrust injury at the
motion to dismiss stage--"[t]aking Novell's allegations as true," 505 F.3d at 316--bears not at
all on the present inquiry. "[A]n appellate decision that a pleading is sufficient" does not bar a
subsequent "judgment that finds a lack of fact support." 18B CHARLES ALAN WRIGHT &
ARTHUR R. MILLER, FEDERAL PRACTICE & PROCEDURE § 4478.3 (3d ed. 2011); see also Re/Max
Int'l v. Realty One, Inc., 900 F. Supp. 132, 145 n. 3 (N.D. Ohio 1995) ("At the summary
judgment stage or at trial, the inquiry turns from the allegations to the evidence. At that time,
proof of antitrust injury which is necessary to prove a party's standing to bring a private cause of
action blurs with proof of the substantive elements of the underlying offense itself. Failure to
prove either antitrust injury or standing would result in judgment for the defendant as a matter of
law."); Greater Rockford Energy & Tech. Corp. v. Shell Oil Co., 790 F. Supp. 804, 809, 827
(C.D. Ill. 1992) (granting summary judgment motion against plaintiffs for failure to establish
standing following development of "voluminous" record and prior denial of motion to dismiss
for lack of standing).
68 The term "Revived Claims" is defined in the Settlement Agreement as "any
counterclaims, other than Claims arising out of the facts and circumstances described in Novell's
draft NetWare complaint delivered to Microsoft on September 24, 2004, that would constitute
compulsory counterclaims to the claim asserted by Microsoft." (Settlement Agreement § 2(a),
Holley Decl. Ex. A.)
69 The Fourth Circuit has already ruled that Novell's "office productivity applications"
refers only to WordPerfect and Quattro Pro. 429 F. App'x at 263. (See also Compl. ¶¶ 24, 25.)
In affirming the dismissal of Novell's GroupWise claim, the Fourth Circuit held that Novell's
"pleading expressly characterized" the products for which it sought to recover damages, and the
Complaint "intended to encompass" only "[w]ord processing and spreadsheet applications." 429
F. App'x at 263.
70 As a result, Novell's damages theories would not provide a reasonable jury with a legally
sufficient evidentiary basis to award any damages because Warren-Boulton made no attempt to
disaggregate the alleged harm to WordPerfect and Quattro Pro from the alleged harm to
PerfectOffice, but instead constructed a damages model in which he combined the projections of
revenue and profits of all three products. (See Warren-Boulton, Nov. 17 Trial Tr. at 2435-36.)
71
As mentioned above, the Fourth Circuit has already held that Novell's "office
productivity applications" refers only to WordPerfect and Quattro Pro. Novell, 429 F. App'x at
263.
72 Consistent with Harral's testimony, the evidence at trial showed that Microsoft clearly
warned ISVs that "[u]sers should NOT edit documents with an explorer extension!" (i.e., with a
word processor) and that the namespace extension APIs instead "[o]nly should be used if your
application displays a pseudo-folder: electronic mail, document management, etc." (PX 113, at
NOV 00734390.) See also pp. 18-20, supra. Satoshi Nakajima testified that it "doesn't make
sense to use [the namespace extension APIs] for word processing application or spreadsheet
application" (Nakajima, Dec. 1 Trial Tr. at 3865), and Joe Belfiore testified that he explained to
ISVs that the namespace extension APIs were not suitable for use by word processing or
spreadsheet applications (Belfiore, Dec. 5 Trial Tr. at 4261-63).
73 In the Court's summary judgment decision in 2010, the Court held that claims relating to
Novell's GroupWise e-mail product were not pled in the Complaint and therefore are not part of
this case. 699 F. Supp. 2d at 743-44, aff'd, Novell, Inc. v. Microsoft Corp., 429 F. App'x 254,
263 (4th Cir. 2011). Novell's expert, Warren-Boulton, testified at trial that Novell's Soft
Solutions document management system "is very, very similar to GroupWise and became part of
GroupWise." (Warren-Boulton, Nov. 16 Trial Tr. at 2160.) Accordingly, any claim based on
Novell's Soft Solutions product falls within the ambit of the Court's ruling excluding
GroupWise, and consequentially, such a claim cannot be asserted in this action. Novell, 699 F.
Supp. 2d at 743-44; Novell, 429 F. App'x at 263.
74 The Complaint made no mention of Novell's Soft Solutions document management
system, e-mail client, Presentations clip-art gallery, or FTP/HTTP browser, and the Complaint's
two references to Novell's QuickFinder search engine gave no indication that those products
were using the namespace extension APIs or that the withdrawal of support for the namespace
extension APIs harmed those products in any way. (See Compl. ¶¶ 94-95 (alleging that
"Microsoft made other inferior features de facto industry standards," which "prevented Novell
from presenting QuickFinder on the desktop").)
75 The "Novell Claims" is defined to include "any and all claims or causes of action held by
Novell as of July 23, 1996 `associated directly or indirectly with any of the DOS Products or
Related Technology' . . . including without limitation all such claims formerly held by Digital
Research, Inc." (Settlement Agreement between Microsoft Corporation and Caldera, Inc.,
January 7, 2000 at 1, Holley Decl. Ex. Q (quoting Asset Purchase Agreement).)
76 This is a far cry from Novell's argument to the Fourth Circuit that Novell's business
applications and its DR DOS operating system "were two entirely distinct group of products in
two different lines of business posing two different types of threats to two different Microsoft
operating systems." (Novell's Brief to the Fourth Circuit at 39.)
77 Microsoft stated in its 2005 motion to dismiss that "[a] comparison of Count I to the DOJ
Complaint reveals that Novell's claim involves different competitors, different products that
allegedly were injured and differences in the anticompetitive conduct alleged" and because any
apparent similarity between the Government Case and Novell's claim--which "Novell does not
own and has no standing to assert"--was merely a sham. (Microsoft's Memorandum in Support
of Its Motion to Dismiss Novell's Complaint, Civ. A. No. 04-1045, filed Jan. 7, 2005, at 15 n.9.)
-137-
CERTIFICATE OF SERVICE
I hereby certify that on the 3rd day of February, 2012, I caused a true and correct copy of
the foregoing Microsoft's Memorandum in Support of its Renewed Motion for Judgment as a
Matter of Law to be filed with the Clerk of Court using the CM/ECF system, which will send
notification of such filing to the following:
Max D. Wheeler
Maralyn M. English
SNOW, CHRISTENSEN & MARTINEAU
[address]
Jeffrey M. Johnson
Paul R. Taskier
Jason D. Wallach
DICKSTEIN SHAPIRO LLP
[address]
R. Bruce Holcomb
ADAMS HOLCOMB LLP
[address]
John E. Schmidtlein
WILLIAMS & CONNOLLY LLP
[address]
/s/ M. David Possick
|