J. FREDERICK MOTZ, District Judge.
Novell, Inc. instituted this action against Microsoft under § 4 of the Clayton Act, 15 U.S.C. § 15, alleging violations of the Sherman Act, 15 U.S.C §§ 1-2.
That claim was tried to a jury. After an approximately eight-week trial, the jury hung and was unable to return a verdict.
In the mid-1990s, Microsoft, which for many years had possessed a monopoly in the PC operating systems market, was developing a new operating system. This operating system became known as Windows 95 and was released for sale to the public on August 24, 1995.
WordPerfect, a word processing application, was originally developed by WordPerfect Corporation, a company located in Orem, Utah. WordPerfect competed with Word and was particularly popular with customers who used character-based DOS operating systems. In the mid-1980s and early 1990s, WordPerfect was the acknowledged "king of the hill" on the DOS platform. (Peterson, Trial Tr. at 4667, Dec. 7, 2011; see Middleton, Trial Tr. at 4178 (Dec. 13, 2008 Video Dep.), Dec. 5, 2011). Although Microsoft manufactured and sold a character-based operating system known as MS-DOS, as early as 1984 it began concentrating on an alternative type of operating system, known as a graphic user interface ("GUI") system. (Gates, Trial Tr. at 2709, 2713, Nov. 21, 2011). As Microsoft developed Windows, its GUI-based operating system, Bill Gates, the Chief Executive Officer of Microsoft, attempted to persuade Pete Peterson, then a principal of WordPerfect, to write for Windows.
Novell announced its intent to purchase WordPerfect on March 21, 1994 and completed the purchase on June 24, 1994. At the same time, Novell purchased Quattro Pro, a spreadsheet application that had been developed by Borland, Inc., a company located in Scotts Valley, California. (Frankenberg, Trial Tr. at 1080, Nov. 7, 2011). Robert Frankenberg, who was the Chief Executive Officer of Novell after its purchase of WordPerfect and Quattro Pro, saw that GUI operating systems were the wave of the future, and he was more inclined than Peterson to work with Microsoft and to develop versions of its applications for Windows. (Id. at 1040). The WordPerfect products for Windows had generally not been well received by the market, and Frankenberg recognized the need to improve those products. In November 1994, Novell released WordPerfect 6.1 for Windows, which met with favorable industry reviews. (See Pl.'s Exs. 239 & 241). Frankenberg also saw the need for Novell to market an office productivity suite, and in late December 1994 WordPerfect successfully released a Windows-compatible suite, PerfectOffice 3.0.
On June 10, 1994, shortly before Novell finalized its purchase of WordPerfect Corporation and Quattro Pro, Microsoft provided independent software vendors ("ISVs"), including WordPerfect, with the Milestone 6 beta version of Windows 95. (See Harral, Trial Tr. at 434-35, Oct. 24, 2011). Novell was extremely enthusiastic about Windows 95 when it was released in beta form to ISVs. Frankenberg testified that Windows 95 was a significant step forward and that Novell was "very excited and very interested" in it. (Frankenberg, Trial Tr. at 1225-26, Nov. 8, 2011). He further testified that Novell's business strategy was to "take advantage of the capabilities in Windows 95." (Id. at 1226). The testimony of other Novell witnesses was to the same effect. (Harral, Trial Tr. at 253-54, 256-57, Oct. 20, 2011; Richardson, Trial Tr. at 607, Oct. 25, 2011; Gibb, Trial Tr. at 788, Oct. 26, 2011; Noll, Trial Tr. at 1911, Nov. 15, 2011).
On October 3, 1994, Microsoft withdrew support for the part of this beta version making namespace extension application programming interfaces ("APIs") accessible to ISVs. (See Pl.'s Ex. 1). Prior to the hearing on the Rule 50 motion, Novell contended that the act of withdrawing support for the namespace extension APIs was itself an unlawful act because beta versions of a release cannot be modified except to fix bugs discovered during the course of the beta process. Novell did not, however, press this argument at the Rule 50 hearing. Its reason for not doing so is clear. Although one Novell software developer, Adam Harral, did so suggest, (Harral, Trial Tr. at 303, Oct. 20, 2011), all of the other evidence was to the contrary. Microsoft's licensing agreement with WordPerfect/Novell provided that the beta version "may be substantially modified prior to first commercial shipment," and that WordPerfect "assume[d] the entire risk with respect to the use of the beta." (Def.'s Ex. 18 at 1 ¶ 2; Def.'s Ex. 19 ¶ 2). Moreover, numerous witnesses, including Frankenberg and other persons who during the relevant time frame had been employed by Novell, testified that they understood the beta versions of Windows 95 "could change" and "might change" prior to commercial release. (Frankenberg, Trial Tr. at 1201, 1204-05, 1209, Nov. 8, 2011; Larsen, Trial Tr. at 3603, 3654-58, Nov. 30, 2011). Novell's expert witnesses likewise agreed with this proposition. (Alepin, Trial Tr. at 1555-56, Nov. 10, 2011; Noll, Trial Tr. at 1878, Nov. 15, 2011).
The parties dispute the reasons that Microsoft decided, on October 3, 1994, to withdraw support for the namespace extension APIs. Microsoft argues it had legitimate "stability" and "robustness" concerns, particularly because, as Novell's own technical expert, Ronald Alepin, conceded, an ISV that wrote a bad program using the namespace extension APIs could cause Windows 95 to crash. (Alepin, Trial Tr. at 1601-02, Nov. 10, 2011; see Gates, Trial Tr. at 2781-82, Nov. 21, 2011). Moreover, according to Microsoft, there was an internal debate at Microsoft between the Windows team and another team, working on what was known as the "Cairo Shell," about the design of the namespace extension APIs. (Gates, Trial Tr. at 2792-93, Nov. 21, 2011; Muglia, Trial Tr. at 3385-90, 3397-3400, Nov. 29, 2011; Nakajima, Trial Tr. at 3763-64, 3768-71, Dec. 1, 2011; Belfiore, Trial Tr. at 4269-71, 4278-80, Dec. 5, 2011). Finally, Microsoft contends that the namespace extension APIs did not provide the functionalities Gates had contemplated. (Gates, Trial Tr. at 2786-87, 2800-04, Nov. 21, 2011; Pl.'s Ex. 1).
Novell argues that all of the reasons Microsoft articulated for the October 3, 1994 decision were pretextual, and that, as revealed in an internal email Gates authored, the real reason for withdrawing support for the namespace extension APIs was to provide a competitive advantage to Microsoft's own office productivity suite, Office. (See Pl.'s Ex. 1). In the October 3, 1994 email announcing the withdrawal decision, Gates stated: "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 [a Lotus product], WordPerfect to achieve, and which will give Office a real advantage." (Id.).
According to Novell, Gates's decision to withdraw support for the namespace extension APIs was consistent with a plan that had been discussed and adopted by Microsoft executives in June 1993, at a retreat at Gates's estate on Hood Canal. That plan called for Windows 95 to be shipped with limited extensibility, reserving for Microsoft's applications the functionality that the namespace extension APIs provided. (See Pl.'s Ex. 61). The plan's purpose was to give Microsoft's applications "a very significant lead over [Microsoft's] competitors, and make [Microsoft's] competitors' products look `old.'" (Id. at MS 00971223). Microsoft was concerned that if the extensibility were not withheld from ISVs, Word and Excel would have to "battle against their competitors on even turf. Given that Lotus and WordPerfect have largely caught up, [Word and Excel] almost certainly lose ground — if not in market share, then in margins." (Pl.'s Ex. 62 at MX 1389851).
After Microsoft withdrew support for the namespace extension APIs in October 1994, Novell was confronted with three options: (1) to continue to write its own programs using the namespace extension APIs that were no longer supported by Microsoft but were still accessible to ISVs in possession of the beta version; (2) to use Microsoft's common file open dialog (instead of the namespace extension APIs), which remained available to ISVs after the withdrawal of support for the namespace extension APIs; or (3) to write its own customized file open dialog. (See Harral, Trial Tr. at 342-44, Oct. 20, 2011). Although it explored the first option, Novell concluded that it could not use the previously released documentation for the namespace extension APIs because Microsoft was no longer providing any assistance for writing programs based upon those APIs. (Id. at 345-46). Novell also concluded that the second option was unsatisfactory because to use it would "impose the standards of the operating system" on developers. (Id. at 271). Specifically, using Microsoft's common file open dialog would have caused Novell to stop providing features that it had provided to its customers in the past, thereby "disenfranchis[ing] [its] customer base." (Harral, Trial Tr. at 504, Oct. 24, 2011; Richardson, Trial Tr. at 629-30, Oct. 25, 2011; see Gibb, Trial Tr. at 891, Oct. 26, 2011). Therefore, in January 1995 Novell stopped trying to reproduce the lost functionality and chose the third option, beginning to write its own customized file open dialog by having two developers, Harral and Richardson, work around the clock, sometimes in excess of 80 hours per week, under Gibb's supervision. (See Harral, Trial Tr. at 350-54, Oct. 20, 2011).
When Gates decided to withdraw support for the namespace extension APIs, some Microsoft executives were concerned that the decision would not be well received by ISVs. In that regard, Brad Silverberg wrote an email to other Microsoft executives, including Gates, on October 5, 1994, predicting there would be a "firestorm of protests" from ISVs, including WordPerfect, that were using the shell extensions. (Pl.'s Ex. 220 at MX 5103184). However, shortly before Silverberg's email was written, Brad Struss, who headed the Windows 95 team for Microsoft's Development Relations Group and worked closely with Novell, reported that Novell "ha[d] not begun any work on IShellFolder, IShellView, etc.," i.e. the namespace extensions. (Def.'s Ex. 17 at MX 6109491). Struss's email also included the results of a survey Microsoft conducted of major ISVs to ascertain if ISVs in fact were using the namespace extension APIs. (Id.). On October 12, 1994, after the decision to withdraw support for the namespace extension APIs was made, Struss advised Microsoft executives by email that Microsoft was "notifying ISVs about the namespace api changes." (Def.'s Ex. 3 at MX 6055840). He reported that "WP [WordPerfect] . . . appear[s] to be OK with this." (Id.). Struss also testified at trial that he told others at Microsoft "what [he] knew to be true or what had been communicated to [him] from WordPerfect, which is that they were not using it and they were not dependent upon it." (Struss, Trial Tr. at 3270, Nov. 28, 2011).
Novell presented no witness to rebut this testimony or to contradict what was said in the October 12 email. There also is no evidence in the record that Novell ever complained to Microsoft about Microsoft's decision to withdraw support for the namespace extension APIs. Further, there is no evidence that any top-level executive at Novell was involved in the decision-making process concerning the choice of the third option, i.e., writing of customized code for WordPerfect. Frankenberg testified that any action that could jeopardize the timely release of WordPerfect or Quattro Pro 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 Glenn Mella, Vice President of Marketing. (Frankenberg, Trial Tr. at 1140-42, 1179-80, Nov. 7-8, 2011). Moreover, Frankenberg agreed that in any business or organization faced with an important decision, a formal memorandum would normally be presented to the senior executives laying out the concerns, issues, and considerations facing that business and making some strategic or tactical choice. (Id. at 1181, Nov. 8, 2011). Novell presented no evidence that any such memorandum was ever written, and Frankenberg testified that he knew of no evidence that Calkins, Mella, Moon, or Rietveld were presented with a decision about how to respond to Gates's decision to withdraw support for the namespace extension APIs. (Id. at 1181-82).
Microsoft sought commitments from several of the most important ISVs to ship their Window 95-compatible applications within 90 days of the release of Windows 95. (See Pl.'s Ex. 148). Frederick Warren-Boulton, Novell's economic expert, testified that if Novell's product was shipped according to that timeframe, the application would have been "in pretty good shape," and that his damage calculations are based on the assumption that Novell could have released their product according to the agreement, but for the alleged anticompetitive conduct. (Warren-Boulton, Trial Tr. at 2417-24, Nov. 17, 2011).
According to Novell, because it took one year for Novell to write its customized file open dialog, Novell was unable to meet that timetable. (Novell Mem. Opp'n Rule 50 Mot. ("Novell Opp'n") at 69, ECF No. 501). Microsoft strongly disputes this assertion. It argues Quattro Pro—the spreadsheet component of PerfectOffice—was not ready to be released to manufacturing until well into 1996, and that this explains the delay in Novell's marketing of PerfectOffice. (See Microsoft Mem. Supp. Rule 50 Mot. ("Microsoft Mem.") at 48, ECF No. 494). An internal Novell memorandum indicated that 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 [was] barely achievable with all their resources and with no additional functionality." (Def.'s Ex. 211). On March 1, 1995, Bruce Brereton, Novell's Vice President of the Business Applications Unit, wrote in an email that because Quattro Pro believed that "December 30 is a more realistic date," Novell had decided to "move . . . the Storm [PerfectOffice for Windows 95] RTM [release to manufacturer] date back by one month (to December 30) and have put WP on the same time-line as Storm." (Def.'s Ex. 221).
In light of Brereton's email, Frankenberg testified that as of March 1, 1995 the plan became to get PerfectOffice out, i.e., released to manufacturing, on December 30, 1995. (Frankenberg, Trial Tr. at 1220-21, Nov. 8, 2011). In March 1995 a "markets requirement report" prepared by the applications group ranked "Quattro Pro delivering late" as the highest "overall risk" for the PerfectOffice development project. (Def.'s Ex. 223 at 41). David LeFevre, who was then an employee of Novell (he subsequently became an employee of Microsoft), testified that he attended daily meetings in 1995 with Gibb and others where the matters discussed included "all the different product challenges" of releasing PerfectOffice in a timely manner. (LeFevre, Trial Tr. at 4037, Dec. 2, 2011). According to LeFevre, "the product that was causing the biggest problem was Quattro Pro." (Id. at 4045-47).
Karl Ford, then the lead developer for the user interface in WordPerfect for Windows 95, also attended regularly scheduled meetings in 1995, and he learned that "the schedule" was at risk because of Quattro Pro. (Ford, Trial Tr. at 3691-92, 3699-3700, Nov. 30, 2011). The problem of having Quattro Pro ready to be released to manufacturing became even more severe toward the end of 1995, when many developers left Quattro Pro for other employment in Silicon Valley, the area where Scotts Valley—home of Quattro Pro—was located. (See Def.'s Ex. 230). On December 23, 1995, four months after the release of Windows 95, Brereton wrote an email to Frankenberg and others reporting that "this past Thursday/Friday about 15 additional people [at Quattro Pro] submitted their resignations," leaving the Quattro Pro development team in Scotts Valley, California with "just 2 people." (Id.). In January 1996, Nolan Larsen, an executive of Novell, traveled to Scotts Valley 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 little bit shell shocked. So it was — it was very chaotic." (Larsen, Trial Tr. at 3620, Nov. 30, 2011). Larsen and LeFevre testified that Quattro Pro was not ready to be shipped by March 1996. (Id. at 3624-25; LeFevre, Trial Tr. at 4062-63, Dec. 2, 2011).
Gibb, however, testified that Quattro Pro was not "critical path" because the delay in the development of shared code on which Harral and Richardson were working was the only issue holding up the release. (Gibb, Trial Tr. at 804-07, Oct. 26, 2011). Ultimately, Quattro Pro and PerfectOffice were not released until May 1996, after the sale to Corel Corporation in March 1996. As described in section III, infra, according to Novell, this delay prevented WordPerfect and PerfectOffice from obtaining a sizeable share of the Windows 95-compatible word processing and suite markets. This, in turn, Novell argues, prevented Novell products from becoming successful "middleware," which could have been an effective competitor with Windows 95 in the operating systems market.
On November 8, 2004, Novell and Microsoft entered into a settlement agreement under which any claims not pled in Novell's draft complaint in this action, filed Nov. 12, 2004, were released. (See Settlement Agreement, Holley Decl. Supp. Microsoft Mem., Ex. A, ECF No. 497). The complaint in this action asserted claims for damages suffered by Novell's "lost sales of office productivity applications and a diminution in value of Novell's assets, reputation and goodwill in amounts to be proven at trial." (Compl. ¶ 155). The complaint further alleged, in the prayer for relief, that Microsoft maintained its monopoly in three specific markets, the operating systems market, the word processing applications market, and the spreadsheet applications market. (Id. at 67).
A Rule 50 motion "test[s] whether there is a legally sufficient evidentiary basis for a reasonable jury to find for the moving party." Keylon v. City of Albuquerque, 535 F.3d 1210, 1215 (10th Cir. 2008). Judgment as a matter of law is appropriate "only if the evidence points but one way and is susceptible to no reasonable inferences which may support the opposing party's position." Tyler v. RE/MAX Mountain States, Inc., 232 F.3d 808, 812 (10th Cir. 2008) (citing Finley v. United States, 82 F.3d 966, 968 (10th Cir. 1996)); see Miller v. Auto. Club of N.M., Inc., 420 F.3d 1098, 1131 (10th Cir. 2005) ("Judgment as a matter of law is only appropriate when `a party has been fully heard on an issue and there is no legally sufficient evidentiary basis for a reasonable jury to find for that party on that issue.'") (quoting Fed. R. Civ. P. 50(a)(1)). All reasonable inferences are drawn in favor of the non-moving party, and the court does not "weigh the evidence, pass on the credibility of witnesses, or substitute [its] conclusions for that of the jury." Miller v. Eby Realty Grp., LLC, 396 F.3d 1105, 1110 (10th Cir. 2005). The decisive question is whether "the plaintiff has arguably proven a legally sufficient claim." Dillon v. Mountain Coal Co., 569 F.3d 1215, 1219 (10th Cir. 2009).
Novell's theory may be briefly stated. According to Novell, Microsoft violated § 2 of the Sherman Act when it decided in October 1994 to withdraw support for the namespace extension APIs, a decision that Novell asserts led to Microsoft's maintenance of its monopoly in the operating systems market.
Microsoft argues that this theory is unique and unprecedented because it is based upon conduct that occurred in one market, the software applications market, to assert a Sherman Act § 2 claim in an entirely different market, the operating systems market. (See id. at 6). That alone, according to Microsoft, is fatal to Novell's claim. (Id.). Novell counters that although Microsoft, through Gates, was motivated by a desire to disadvantage Novell's applications in favor of Microsoft's own applications, its conduct was intentional, and a jury could appropriately find that its conduct had anticompetitive effects in the operating systems market. (Id. at 165-68). According to Novell, that is sufficient to establish a § 2 violation.
I need not decide this issue because, even assuming Novell's argument is correct, its claim nevertheless fails for three separate and independent reasons: (1) Microsoft's conduct was not anticompetitive within the meaning of the Sherman Act; (2) Novell did not present sufficient evidence from which a jury could find that its products would have been successfully developed as middleware; and (3) there is no underlying business reality to the claims. Each of these reasons is examined in section V. Before discussing those issues, however, I will first address two arguments made by Microsoft that I believe lack merit.
Microsoft argues it has articulated legitimate business reasons for the decision Gates made on October 3, 1994, to withdraw support for the namespace extension APIs and that the fact-finder is not to weigh the factors that led to a particular decision in assessing whether that decision was a violation of antitrust laws. (See Microsoft Mem. at 110 (citing Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585, 597, 605 (1985); Multistate Legal Studies v. Harcourt Brace Jovanovich Legal & Prof'l Publ'ns, 63 F.3d 1540, 1550 (10th Cir. 1995))). The fallacy in that contention is that in this case, the jury was not asked to weigh the factors that Microsoft articulated in support of the October 3, 1994 decision, but rather to decide whether those articulated reasons were pretextual in light of Gates's email suggesting the decision was made to give Microsoft application developers time to catch up with Microsoft's major competitors, Novell and Lotus. Fact-finders are frequently called upon to answer this question of pretext, particularly in the field of employment discrimination law. See, e.g., Reeves v. Sanderson Plumbing Prods., 530 U.S. 133, 153-54 (2000) (holding that a plaintiff "introduced enough evidence for a jury to reject [the defendant's explanation]" as pretextual); Townsend v. Lumbermens Mut. Cas. Co., 204 F.3d 1232, 1243-44 (10th Cir. 2002) (holding that a jury "could reasonably have found" a defendant employer's stated reasons for demoting an employee "were pretextual, potentially entitling him to a finding of discrimination").
Here, although I am granting Microsoft's Rule 50 motion for the reasons stated in section V, I believe the text of Gates's email provides sufficient evidence upon which a jury could find that the reasons for the October 3, 1994 decision to withdraw support for the namespace extensions were pretextual. Therefore, if the only question raised by the Rule 50 motion were whether the jury was asked to weigh the factors that led Microsoft to make the October 3, 1994 decision, I would deny the Rule 50 motion.
As stated in section I, supra, on November 8, 2004, Microsoft and Novell entered into a settlement agreement pursuant to which Novell released any and all claims it had against Microsoft with the exception of claims asserted in a draft complaint that became the complaint in this action. In Count I of the complaint (the only count remaining in the case), Novell alleged that as a result of Microsoft's alleged misconduct, it "was damaged by, without limitation, lost sales of office productivity applications and a diminution in the value of Novell's assets, reputation, and goodwill in amounts to be proven at trial." (Compl. ¶ 155). The term "office productivity applications" is defined earlier in the complaint as "word processing and spreadsheet applications." (Id. ¶ 24).
Microsoft argues that because the only Novell applications allegedly damaged by Microsoft's alleged misconduct were the word processing application, WordPerfect, and spreadsheet application, Quattro Pro, Novell released Microsoft from any claim that it had for damages to Novell's office productivity suite, PerfectOffice. (See Microsoft Mem. at 126-27). Microsoft also points out that the complaint does not refer in any way to the suite market. (See id. at 128-30).
In entering into the release—which certainly was a significant business transaction— Microsoft had the right to rely upon the language of the release and the language of the draft complaint referred to in the release. Those documents made no reference whatsoever to PerfectOffice. Moreover, in deciding to enter into the settlement agreement, Microsoft may have considered the omission of any reference to PerfectOffice to be significant because, as the evidence at trial established, by 1994 it was clear that the office suite market stood separate and apart from the word processing and spreadsheet applications markets. (See Gibb, Trial Tr. at 823, Oct. 26, 2011; Frankenberg, Trial Tr. at 1080, Nov. 7, 2011). As Frankenberg's testimony highlighted during the trial, Novell as well as Microsoft understood the importance of the suite market in 1994. Frankenberg testified that in 1994 approximately three-quarters of the suite market was open, and Novell was planning to make substantial inroads into it. (See Frankenberg, Trial Tr. at 1008-09, 1063, Nov. 7, 2011; Pl.'s Ex. 412 at 2).
Nevertheless, Microsoft's argument fails. The complaint referred not only to damages resulting from lost sales of office productivity applications but also damages resulting from "a diminution in the value of Novell's assets, reputation, and goodwill in amounts to be proven at trial." (Compl. ¶ 155). Novell's damages expert, Frederick Warren-Boulton, did not calculate damages on a product-by-product basis but instead on the basis of the decrease in the sales price paid to Novell by Corel, allegedly caused by Microsoft's withdrawal of support for the namespace extension APIs. (See Warren-Boulton, Trial Tr. at 2094-98, Nov. 16, 2011). This decrease in sales price falls well within the "diminution in value" category of damages referred to in the complaint and thus was not released by the agreement entered into November 8, 1994.
I will now discuss the three reasons Microsoft's Rule 50 motion should be granted.
A monopolist is free "to exercise [its] own independent discretion as to parties with whom [it] will deal." Verizon Commc'ns, Inc. v. Law Offices of Curtis V. Trinko, LLP, 540 U.S. 398, 408 (2004); see United States v. Colgate & Co., 250 U.S. 300, 307 (1919). It is well established that a monopolist generally has no duty to cooperate with its competitors. See, e.g., Trinko, 540 U.S. at 408; Aspen Skiing, 472 U.S. at 600; Image Technical Servs., Inc., v. Eastman Kodak Co., 125 F.3d 1195, 1209 (9th Cir. 1997). In accordance with this principle, Novell's counsel properly conceded during the hearing on Microsoft's Rule 50 motion that it would not have been unlawful for Microsoft to keep the namespace extension APIs to itself and never disclose them to ISVs. (Rule 50 Hr'g Tr. at 191-92). Novell argues, however, that Microsoft's conduct was anticompetitive within the meaning of § 2 of the Sherman Act because when it decided to withdraw support for the namespace extension APIs in October 1994, it intentionally made a decision that had an adverse effect upon competition in the operating systems market. (See id. at 165-68).
The distinction Novell seeks to draw is unpersuasive. A decision not to publish the namespace extension APIs in the first place is as "intentional" as a decision to withdraw support for the namespace extension APIs after they have been published. Therefore, the principle that a monopolist generally has no duty to cooperate with a competitor governs unless the act of withdrawing support for the namespace extension APIs was itself anticompetitive.
The decision to withdraw support for the namespace extension APIs therefore did not constitute a violation of § 2 of the Sherman Act unless it was made deceptively for an anticompetitive purpose or unless, by making it, Microsoft terminated a previously existing profitable relationship.
Any deception claim necessarily is based upon the premise that Microsoft withdrew support for the namespace extension APIs because it knew that Novell was using those APIs in the development of its applications and that, by withdrawing support for those APIs, Microsoft knew that Novell would fall behind schedule. There is no evidence to support this premise.
The claim that Microsoft purposely destroyed a preexisting profitable business relationship is equally flawed. Preliminarily, it should be noted that Aspen Skiing, upon which Novell bases its argument, has been characterized by the Supreme Court as being "at or near the outer boundary of § 2 liability." Trinko, 540 U.S. at 409. The facts in Aspen Skiing were unusual, involving a defendant's exploitation of a situation largely bestowed upon it by Mother Nature. The defendant owned three mountains in a ski resort area and engaged in sharp business practices to prevent the owner of the fourth mountain from competing effectively for skiing business. See Aspen Skiing, 472 U.S. at 589-95. In contrast, what Novell is asserting here is that Microsoft could not withdraw from Novell a product that was the result of Microsoft's own ingenuity and that Novell believed would give it optimal advantage in its competition with Microsoft in the applications market.
There is no evidence that Microsoft withdrew support for the namespace extension APIs for the purpose of terminating its relationship with Novell. The final version of Windows 95 made available to all ISVs, including Novell, contained Microsoft's common file open dialog, and Novell never advised Microsoft that this dialog was insufficient for its own purposes. (See Harral, Trial Tr. at 502, Oct. 24, 2011; Gibb, Trial Tr. at 847-49, Oct. 26, 2011; Alepin, Trial Tr. at 1604, Nov. 10, 2011). Moreover, after it withdrew support for the namespace extension APIs, Microsoft continued to provide assistance to Novell and never terminated their relationship. (Struss, Trial Tr. at 3253-54, 3259-69, Nov. 28, 2011; Def.'s Ex. 2 at MX 6062581; see Frankenberg, Trial Tr. at 1131, Nov. 7, 2011 (stating that he is "sure" people in the operating system group at Microsoft were trying to help WordPerfect/Novell produce a great application for Windows 95)).
For these reasons, Novell has not created a jury question on the issue of whether Microsoft's conduct was anticompetitive.
The evidence is undisputed that during 1995 Novell's focus was on writing office productivity programs that would run on Windows 95. In Novell's view, Microsoft's new operating system was a "significant step forward" and a "wonderful evolution" in technology. (Harral, Trial Tr. at 253-54, 256-57, Oct. 20, 2011; Frankenberg, Trial Tr. at 1225-26, Nov. 8, 2011; see Gibb, Trial Tr. at 788, Oct. 26, 2011; Noll, Trial Tr. at 1911, Nov. 15, 2011). Therefore, as Frankenberg acknowledged, in the short term Microsoft's share of the operating systems market would have increased, not decreased, if it had not withdrawn support for the namespace extension APIs. (See Frankenberg, Trial Tr. at 1226-28, Nov. 8, 2011; Noll, Trial Tr. at 1949-50, Nov. 15, 2011).
Therefore, Novell's theory necessarily is that the anticompetitive effects it alleges did not occur until sometime after 1995. Novell has never specified exactly when those effects did occur. Novell sold WordPerfect to Corel in March 1996, and there is no evidence that by that time there was any operating system on the market comparable to Windows 95 to which Novell planned to write its applications or for which those applications could have been written. That fact, however, may not be dispositive because Novell's damages claim is based upon the alleged diminution in the value of its assets caused by Microsoft's conduct. Presumably, Corel would have paid more for what it bought if by that time Microsoft had not, as Novell alleges, effectively destroyed Novell's applications and prevented them from being cross-platformed to other operating systems then under development. Therefore, I will assume, as Novell implicitly argues, that within a reasonable period of time after Windows 95 was put on the market (say two or three years), another operating system would have come into existence to which Novell's office productivity products could have been written.
In asserting that it could and would have written to an operating system other than Windows 95 within a reasonable period of time after 1995, Novell relies upon two theories: the "franchise theory" and the "middleware theory."
In advancing the franchise theory, Novell posits that its applications were so popular that if Microsoft had not withdrawn support for the namespace extension APIs, Novell's applications would have flourished in the market, regardless of the operating system on which they ran. There is no evidence to support this theory. In fact, it is contradicted by the record.
First, Novell recognized that Windows 95 was a superior operating system, constituting a "significant step forward." (Frankenberg, Trial Tr. at 1225-26, Nov. 8. 2011). In light of that fact, there is no basis for inferring that office productivity applications Novell developed that did not draw upon the superior functionality of Windows 95 would have been as successful as the applications that ran on Windows 95.
Second, historical data disprove Novell's claim. WordPerfect was extremely popular in the 1980s and early 1990s when it ran on many non-Microsoft operating systems. (Peterson, Trial Tr. at 4667, Dec. 7, 2011; see Middleton, Trial Tr. at 4178 (Dec. 13, 2008 Video Dep.), Dec. 5, 2011). However, this popularity did not diminish Microsoft's share of the PC operating systems market, which was approximately 90% during that period. (Noll, Trial Tr. at 1329-30, Nov. 15, 2011; Murphy, Trial Tr. at 4722-23, Dec. 7, 2011; Finding of Fact ¶ 35). On this record, it cannot reasonably be inferred that if in the years after 1995 Novell's applications ran on operating systems other than Windows, that fact would have challenged Microsoft's monopoly in the PC operating systems market.
Novell's second argument is its middleware theory. For a middleware product to have an impact on competition in the PC operating systems market, the product (1) must be cross-platformed to various operating systems; (2) must be ubiquitous on the "dominant operating system"; and (3) must expose a sufficient number of APIs of its own to entice ISVs to write applications to it rather than to the operating system on which it sits. (See Noll, Trial Tr. at 1923-26, Nov. 15, 2011; Finding of Fact ¶ 28). Novell's office productivity applications did not meet any of these requirements.
As to the first requirement, assuming that within a reasonable period of time after 1995 effective operating systems would have come into existence to which Novell's office productivity applications could have been written, there is no basis for inferring that Novell's office productivity applications written for Windows 95 via the namespace extension APIs could have been effectively ported to those systems. This is so because the namespace extension APIs were, as Novell's own technical expert testified, "platform specific" to Windows. (Alepin, Trial Tr. at 1482-83, 1532-33, Nov. 9-10, 2011; see Murphy, Trial Tr. at 4783-84, Dec. 7, 2011; Bennett, Trial Tr. at 5023, Dec. 12, 2011 (stating that "namespace extension APIs . . . [were] a unique component of Windows 95")).
As to the second requirement, although the parties agree that the dominant operating system was Windows 95, they disagree as to the meaning of ubiquity. Microsoft contends it means that Novell's software had to run on "all or nearly all PCs running the `dominant operating system.'" (Microsoft Mem. at 73-75; Microsoft Reply Supp. Rule 50 Mot. at 37-39, ECF No. 503). Novell argues that "something less than that" might be sufficient "by weakening, though not eliminating, the applications barrier to entry." (Novell Opp'n at 90 (quoting Noll, Trial Tr. at 1926, Nov. 15, 2011)). Novell provided no evidence as to what this lesser threshold might be.
In any event, under either definition of ubiquity, the evidence is clear that Novell's office productivity applications would never have been ubiquitous on Windows 95. In 1995, prior to the release of Windows 95, WordPerfect had roughly a 15% share of the Windows-compatible word processing market, and PerfectOffice had less than a 5% share of the Windows-compatible suite market.
The parties also disagree about the meaning of the third requirement. Microsoft argues that to constitute middleware, an application must "expose a sufficiently broad set of APIs to enable ISVs profitably to develop full-featured personal productivity applications that rely solely upon those APIs exposed by the middleware." (Microsoft Mem. at 70 (emphasis added)). Novell, on the other hand, relying upon the testimony of Roger Noll, its antitrust expert, and Ronald Alepin, its technical expert, contends that the third element is satisfied if the application "expose[s] a wide range of APIs and sophisticated functionality to developers." (Novell Opp'n at 28). Novell concedes that if Microsoft's interpretation of the meaning of the third element is correct, Microsoft is entitled to judgment as a matter of law because Novell did not present evidence to show that its software exposed sufficient APIs of its own to allow ISVs to write full-featured personal products applications to it. (Trial Tr. at 5436-37, 5439, Dec. 15, 2011). Thus, on this issue, whether Microsoft is entitled to judgment in its favor on the Rule 50 motion turns on the meaning of the third requirement.
Microsoft's position is based upon the Findings of Fact made in the government case, upon which Novell's claim is founded. Judge Jackson found that "[c]urrently, no middleware product exposes enough APIs to allow independent software vendors ("ISVs") profitably to write full-featured personal productivity applications that rely solely on . . . APIs [of the middleware product itself]." Finding of Fact ¶ 28. In contrast, Novell argues the exposure of APIs that would result in "something less" than the writing of full-featured personal product applications is sufficient to constitute a threat to Microsoft's monopoly. (Novell Opp'n at 89-90). This argument is based on the concept, expressed by Noll, that diminishing, as opposed to nearly eliminating, the barrier to entry that protected Microsoft's monopoly in the PC operating systems market was itself sufficient. (Noll, Trial Tr. at 1926, Nov. 15, 2011). To the extent this testimony is based on the premise that other companies would produce similar middleware that, in combination with Novell's products, would diminish the barrier to entry, there is no evidence such other products existed.
Although in other circumstances conduct directed at reducing a barrier to entry might constitute a violation of § 2 of the Sherman Act, under the facts of this case, I find Novell's reasoning to be unpersuasive. The findings made in the government case made clear that Microsoft possessed its monopoly by virtue of what Judge Jackson described as a "chicken-and-egg" problem. As Judge Jackson described:
Finding of Fact ¶ 30. Other findings made by Judge Jackson are to the same effect. See Findings of Fact ¶¶ 36, 37, 39, 40.
In light of these findings, it cannot be reasoned, as Novell argues, that Microsoft's operating systems monopoly would have been threatened by a middleware product that exposed only a limited number of APIs that permitted ISVs to write only a specialized set of applications to it. The barrier to entry that Judge Jackson found in the government case was created by a "chicken-and-egg" problem, and that problem arose because ISVs would write only to programs that supported full-featured personal productivity applications. See Finding of Fact ¶ 30. In other words, contrary to what Novell argues in support of what Microsoft has aptly described as its "watered-down version" of the third requirement, Microsoft's monopoly in the PC operating systems market was threatened not by a product that exposed only a limited number of its own APIs but only by a product that exposed sufficient APIs to entice full-featured applications to be written to it. (See Microsoft Mem. at 77-78). Otherwise stated, diminishment of the barrier to entry is not sufficient because mere diminishment would not have affected the PC operating systems market. In order to constitute a realistic threat to Microsoft's monopoly in that market, elimination (or, at least, near elimination) of the barrier to entry through development of full-featured applications using APIs from middleware that ran on operating systems other than Windows was required.
In sum, Novell did not present sufficient evidence from which a reasonable jury could find that its applications could have successfully developed into middleware that threatened Microsoft's monopoly in the operating systems market.
Claims asserted under the Sherman Act, like any other claim asserted in a court, are not cognizable simply because they are theoretically coherent. They must also be based on fact. Here, the absence of any evidence suggesting that the withdrawal of support for Microsoft's namespace extension APIs was the source of any contemporaneous urgency at Novell reflects that the claim Novell asserts is a lawyers' construct and is not based on an underlying business reality.
First, as stated in section I, supra, there is no evidence that anyone at Novell made any complaint to anyone at Microsoft who could have reversed the decision to withdraw support for the namespace extension APIs.
Second, as also stated in section I, there is no evidence that any top-level executive was involved in the decision-making process concerning the writing of shared code for WordPerfect. Frankenberg testified that any action that could jeopardize the timely release of WordPerfect or Quattro Pro would generally have been referred to some or all of four senior executives. (Frankenberg, Trial Tr. at 1140-42, 1179-80, Nov. 7-8, 2011). Novell presented no evidence that any memorandum was written to those executives, and Frankenberg testified that he knew of no evidence whatsoever that any of the four executives were presented with a decision about how to respond to Gates's decision to withdraw support for the namespace extension APIs. (Id. at 1181-82, Nov. 8, 2011). Instead, Novell assigned responsibility for writing shared code for WordPerfect and Quattro Pro to a middle manager, Gibb, who, in turn, assigned the responsibility to only two programmers, Harral and Richardson. However valiant the efforts of Harral and Richardson, if being ready to release WordPerfect, Quattro Pro, and PerfectOffice within 90 days of Microsoft's release of Windows 95 was as critical as Novell now claims it to have been, Novell clearly would have implemented a different development plan.
Third, although Gibb testified that Quattro Pro was "code complete" in time for the release of Quattro Pro and PerfectOffice within 90 days of Microsoft's release of Windows 95,
Fourth, if, as Novell now argues, the 90-day period after the release of Windows 95 was critical to the success of WordPerfect, Quattro Pro, and PerfectOffice, Novell could have released those products using Microsoft's common file open dialog. In fact, Ford and LeFevre testified that they urged Gibb to pursue that option. (Ford, Trial Tr. at 3710-11, Nov. 30, 2011; LeFevre, Trial Tr. at 4041-43, Dec. 2, 2011).
In short, no reasonable jury could find, on the basis of the evidence presented at trial, that Microsoft's withdrawal of support for the namespace extension APIs caused Novell's failure to develop its applications within 90 days of the release of Windows 95.
For these reasons, Microsoft's Rule 50 motion will be granted. A separate order to that effect is being entered herewith.
I recognize that this conclusion may appear somewhat disturbing because arguably it rewards Microsoft for unsavory behavior in the applications market. Nevertheless, the reasoning is sound. Although the proposition upon which Novell relies makes eminent sense as a matter of theory, it is based upon a factual inference about how a rational actor would behave, and here the evidence does not support the inference. Moreover, Novell has only itself to blame for the quandary in which it finds itself because it was its own delay in filing this action that resulted in its claims in the applications market being barred by limitations. See Novell, 2005 WL 1398643, at *5.
Novell referred to the email at trial and has likewise referred to it in opposing Microsoft's Rule 50 motion. In the event that the Tenth Circuit reverses my ruling that Microsoft's Rule 50 motion should be granted, it would be helpful in the retrial of this case if the Tenth Circuit were to decide whether the Raikes email was properly admitted. It would also be helpful, in the event of a reversal of my Rule 50 ruling, if the Tenth Circuit resolved a disagreement between the parties as to the proper causation standard. Microsoft contends that the standard that should be applied is whether its decision to withdraw support for the namespace extension APIs "contributed significantly to its continued monopoly power." (Microsoft Mem. at 82). Conversely, Novell contends that the appropriate standard is whether Microsoft's decision was "reasonably capable of contributing significantly" to Microsoft's monopoly power. (Novell Opp'n at 94-95). In reaching my decision, I have not found it necessary to resolve this issue.