OPINION AND ORDER

LAKE, District Judge.

At issue in this action is the liability of Microsoft Corporation under Texas tort law and federal antitrust law for the demise in sales of the Cache86 disk cache program. Pending before the court is Microsoft's Motion for Summary Judgment. For reasons explained below the motion will be granted in part and denied in part.

I. BACKGROUND

An appreciation of the legal issues in this case requires a basic understanding of computer operating systems in general and Microsoft's Windows95 operating system in particular.

A. Computer Operating Systems

Modern personal computers consist of a microprocessor and several components ("peripheral devices"), including volatile and nonvolatile memory chips; data input devices, such as keyboards and mouses; data storage devices, such as hard and floppy disk drives; output devices, such as printers and monitors; and communication interfaces, such as fax/modems and ethernet connections. Computers only perform operations when told to do so. An operating system is a program that coordinates activities among the various components of the computer and controls the flow of data among them. It is a computer's operating system, for example, that instructs the computer to download data from memory to a disk or to execute the necessary procedures to open a program when requested to do so by the user. While personal computer users could conceivably write their own customized operating system programs, commercially available operating systems are far more desirable to most users.

Microsoft Corporation is a software manufacturer that sells numerous computer programs, including the MS-DOS operating system. As personal computers proliferated, MS-DOS became the operating system of choice among users of computers modeled on IBM's personal computer design. Software manufacturers began tailoring their programs to run in conjunction with MS-DOS. Many software companies also wrote programs, known as utilities, to improve the performance of operating systems in general, and MS-DOS in particular. Plaintiffs developed a utility program called Cache86 that accelerated the disk-writing functions of MS-DOS. David L. Aldridge, the owner of The David L. Aldridge Company, originally wrote the computer code for this program. Technical Partners owns the program, and The David L. Aldridge Company markets and sells it. (For simplicity, the court will refer to the three plaintiffs collectively as Aldridge.)

B. Disk Cache Programs

Cache86 is a disk cache program. Disk cache programs are designed to improve the speed at which computers function. An application program such as a word processor will often direct the operating system to download ("write") or upload ("read") data from the computer's hard or floppy disk. Because of the great speed of modern microprocessors the computer cannot physically perform these read-write operations as fast as the microprocessor can process the data. Read-write operations occur more slowly because disk drives are mechanical devices that can only move so quickly, while the rest of a computer's operations occur electrically and are theoretically only limited by the speed of light. The microprocessor must thus often wait idle while the read-write operations are completed.

Disk caches solve this problem by intercepting information that the operating system is reading from or writing to the computer's hard disk and temporarily using vacant portions of the computer's RAM as a pseudo-disk. This allows the operating system to perform the read-write operations electrically instead of using the hard disk. The disk cache program sets aside a portion of RAM to temporarily store data; this set-aside area of memory is called a "cache." When an application program reads data from a disk, the disk cache program copies the data into the set-aside portion of RAM. The next time the application program needs that same data it can read that data electrically from the cache in RAM rather than having to read mechanically from the hard disk. When the application program writes data onto the disk, the disk, cache program directs the data to the set-aside cache in RAM. Because reading from and writing to RAM is an electrical operation, the computer can perform these operations much faster than it can with a mechanical disk drive.

The way the Cache86 disk cache utility program functions is important to this case. Aldridge designed Cache86 to improve the performance of MS-DOS. MS-DOS (and versions of Windows that run in conjunction with MS-DOS) performs read-write operations to the hard disk by "calling" a "device driver" — a program used by operating systems to control peripheral devices such as a hard disk drive. The hard disk drive driver then sends commands to an interface on the microprocessor known as "Interrupt 13." Cache86 monitors Interrupt 13 and intercepts the information and instructions that pass through that interrupt (a process referred to as "hooking"). Cache86 thus reroutes read-write operations to the cache in RAM by hooking Interrupt 13.

C. Windows95

For years Microsoft developed gradually more sophisticated versions of its MS-DOS operating system. Eventually, Microsoft introduced a supplemental operating system, Windows, to run in conjunction with MS-DOS. Windows allowed users to perform tasks by "pointing and clicking" on small pictures ("icons") displayed on the computer's video screen. This "graphical user interface" enabled users to operate their computers without having to enter typewritten commands into the computer. The evolution of operating systems and extensive advances in computer technology ultimately caused Microsoft to replace MS-DOS and Windows with a new, fundamentally different operating system called Windows95. This new system combined the graphical user interface made popular by Windows with more advanced programming techniques.

Windows95 can operate in three different modes: DOS-compatibility mode, real-mode DOS, and protect mode. When Windows95 operates in DOS-compatibility mode it emulates MS-DOS by calling the hard disk drive in the same manner as MS-DOS. Microsoft included this mode so that Windows95 could run most MS-DOS programs as if they had been designed for use with Windows95. When operating in DOS-compatibility mode, however, Windows95 cannot use several of the new features designed to improve the operating system over its predecessors. Real-mode DOS is a hybrid between DOS-compatibility mode and protect mode. When operating in real-mode DOS the computer remains in protect mode but runs DOS as an application program. In other words, the user has opened DOS as an application program to run under the guidance of protect mode.

When in protect mode the Windows95 operating system functions in a new, fundamentally different way. Among numerous changes and improvements, protect mode uses a different file system and does not use device drivers; instead, in protect mode Windows95 interacts directly with peripheral devices. While the file system in MS-DOS cannot function while MS-DOS performs other tasks, the file system in protect mode can operate simultaneously while the operating system performs other tasks, a property known as "multitasking." Moreover, while MS-DOS only allows data to flow in one direction at a time between the operating system and the hard drive, protect mode permits data to flow in both directions at the same time, a property known as "asynchronous input-output."

How protect mode accomplishes these features is not relevant to this action except for one important point: Protect mode does not perform read-write operations only through Interrupt 13. Protect mode also uses other interrupts. Because Cache86, like many other disk cache programs, operates under the assumption that the computer will transfer all read-write data through Interrupt 13, Cache86 only monitors and hooks the data passing through Interrupt 13. Since computers running under Windows95 protect mode do not exclusively use Interrupt 13, a disk cache program like Cache86 will not reroute all of the data sent between the microprocessor and the hard disk. Programs such as Cache86 that hook. Interrupt 13 thus posed a serious concern for the designers of Windows95.

For example, while a disk cache program performs read-write operations of the data passing through Interrupt 13 to the cache in RAM, Windows95 protect mode could be sending data to the hard disk directly or through other data pathways. If that occurred, the data stored in RAM might no longer be synchronized with the corresponding information on the hard disk. This would create the danger of a problem known as "data corruption" that could cause the computer to lose all of the information stored on the hard disk. Microsoft wanted to avoid this problem while ensuring that computer owners could continue to use their existing MS-DOS and Windows-compatible programs.

Because of the popularity of MS-DOS and Windows, over the years software developers sold thousands of programs specifically designed to function with the MS-DOS and Windows operating systems. When Microsoft designed its new operating system it wanted to ensure that computer users could still use their MS-DOS-and Windows-compatible programs with the new system; in other words, Microsoft intended to make Windows95 "backwardly compatible." The problem Microsoft faced was to design Windows95 with the capability of running MS-DOS-and Windows-compatible programs, while preventing programs that hook Interrupt 13 (or certain other interrupts) from corrupting hard disk data. The way that Microsoft solved this problem ultimately led to this lawsuit.

First, Microsoft designed Windows95 to detect programs that hooked certain interrupts, including Interrupt 13. Some programs that hook Interrupt 13, such as data encryption or antivirus programs, remain valuable on computers using Windows95. Others, such as some older computer games, will not function in protect mode; to use them, Windows95 must operate in an alternative mode. In addition, disk cache utility programs remain useful when Windows95 is running in DOS-compatibility mode or real-mode DOS. In these modes Windows95 performs read-write operations in the same manner as MS-DOS and Windows, and a disk cache program operating in conjunction with either mode offers the same advantages it provides to a computer that uses MS-DOS. Microsoft therefore designed Windows95 not to delete programs it detected hooking Interrupt 13.

Instead, Microsoft created a list of all known hooking programs that could potentially corrupt the hard disk in a file named SETUPC.INF. When Windows95 begins "booting up" on a computer (the process by which an operating system loads itself into the computer when it is turned on) it refers to this file and looks for any of the listed programs. If Windows95 finds a program listed in SETUPC.INF, it disables the program.

Because it was impossible to identify all existing hooking programs, and because it was likely that new hooking programs would be marketed after the release of Windows95, Microsoft also added a second program to protect against data corruption from unrecognized hooking programs. When Windows95 detects a program hooking Interrupt 13 that is not listed in SETUPC.INF., it refers to this second file, IOS.INI. This file contains a list of known hooking programs that will not cause data corruption while Windows95 runs in protect mode. If the detected hooking program appears on this "safe" list Windows95 continues operating in protect mode; if not, Windows95 automatically shifts from protect mode to DOS-compatibility mode. In DOS-compatibility mode the operating system performs read-write operations in the same way as MS-DOS and Windows, and read-write data and commands pass only through Interrupt 13. Because DOS-compatibility mode functions the same way that a disk cache program like Cache86 assumes the operating system will function, there is no risk that the disk cache program will corrupt data. Succinctly stated, Windows95 protects against data corruption from unrecognized programs that hook Interrupt 13 by shifting into DOS-compatibility mode.

A computer operating in DOS-compatibility mode loses the efficiencies and advancements provided by Windows95's preferred mode of operation, protect mode, and is no more efficient than one using MS-DOS. To alert users of the shift to DOS-compatibility mode Microsoft programmed Windows95 to display a series of messages when Windows95 detects that an unfamiliar program has triggered the shift in modes. These messages are the basis of all of Aldridge's claims against Microsoft.

D. Windows95's Treatment of Cache86

Microsoft did not list Cache86 in either the disabling file (SETUPC.INF) or the safe list (IOS.INI). Thus, upon detecting Cache86, Windows95 will not disable it because it is not listed in SETUPC.INF. Because Cache86 does not appear on the safe list in IOS.INI, however, Windows95 will shift into DOS-compatibility mode.

Microsoft programmed Windows95 to display a generic warning to the user the first time an unknown program causes a shift to DOS-compatibility mode. The warning appears in a box on the video screen captioned "Performance Warning." The box contains a triangle with an exclamation point. Next to the triangle is a statement to the user. (The statement is the same for any program that causes such a shift, but Windows95 inserts the filename of the particular triggering program into the statement.) The statement that Windows95 displays when it detects Cache86 and shifts to MS-DOS compatibility mode says

A new MS-DOS resident program named `C86EXTEN' may decrease your system's performance. Would you like to see more information about this problem?

Below this statement are a "yes" box and a "no" box. (The court will refer to this message box as Message One.)

If a user clicks on the yes button, a new message box appears on screen, captioned "System Properties" and subcaptioned "Performance." This box includes a statement that

C86EXTEN in AUTOEXEC.BAT forces MS-DOS compatibility mode. Compatibility mode paging reduces overall system performance.

(Message Two)

Users can obtain additional information by highlighting either sentence in Message Two and clicking on the "details" box. A new message box, captioned "Windows Help," then appears on the screen with Message Two. If the user highlights the first sentence of Message Two and clicks on the details box the following explanation of the first sentence appears in the Windows Help box:

[Filename] in AUTOEXEC.BAT forces MS-DOS compatibility mode.

Windows was unable to identify the specified memory-resident program which was loaded in your Autoexec.bat file. If you no longer need this program, remove it from your Autoexec .bat file. Otherwise, contact the manufacturer of this program to see if an updated version is available. Windows will not perform optimally while this program is loaded.

(Message Three)

If a user highlights the second sentence in Message Two and clicks on the "details" box the following explanation of the second sentence appears in the Windows Help box:

Compatibility mode paging reduces overall system performance.

Windows was unable to identify a real-mode driver or memory-resident program loaded in your Config.sys or Autoexec .bat file. You will see another message that lists the name of the driver or program causing the problem. Because Windows could not identify the driver or program, it has switched to MSDOS compatibility mode to ensure the program will run. However, this slows down overall performance. To improve performance, remove the program or driver causing the problem, or contact its manufacturer for an updated version.

(Message Four)

Windows95 only displays Message One once for each program that causes Windows95 to shift into DOS-compatibility mode. If the user takes no action, either by removing the program or by installing an updated version compatible with Windows95, the computer shifts into DOS-compatibility mode, but does not warn the user each time it again detects the previously discovered hooking program. Windows95 will not display Messages Two, Three, or Four again without directions to do so from the user.

Microsoft included two of its own disk cache programs with Windows95: VCACHE and SMARTDRV. VCACHE is a disk cache designed by Microsoft to operate safely while Windows95 functions in protect mode. Like Windows95, it functions according to principles and assumptions fundamentally different from those of Cache86 and SMARTDRV. For purposes of this action it is not relevant how VCACHE functions, except to note that rather than assuming that all read-write operations will pass through Interrupt 13, VCACHE takes advantage of Windows95's asynchronous input-output system. Aldridge does not allege that Cache86 competes with VCACHE.

Like Cache86, SMARTDRV caches read-write data by hooking Interrupt 13. Windows95's treatment of SMARTDRV is different, however, from its treatment of Cache86. When Windows95 detects SMARTDRV it does not shift into DOS-compatibility mode. Microsoft developed a way to allow Windows95 to remain in protect mode despite the presence of its SMARTDRV disk cache program on the system without risking data corruption. Microsoft reprogrammed SMARTDRV so that it can determine if Windows95 is running in protect mode. If it is, SMARTDRV disables itself by reducing the size of the reserved cache in RAM to zero. Because the cache size is zero it does not save any data, and the read-write data passes directly to the hard disk. The data hooked by SMARTDRV therefore remains synchronized with the hard disk. By eliminating the danger of asynchronization Microsoft avoided the possibility of data corruption and placed SMARTDRV on the safe list in IOS.INI. Aldridge alleges that Cache86 competes with SMARTDRV.

E. Events Leading to this Action

Although Microsoft did not release Windows95 for sale to the public until August of 1995, it distributed trial versions of the operating system as part of a series of design tests, called "beta tests," before the public release. Some 400,000 computer users received test versions of Windows95 as part of this program. These "beta testers" could run Windows95 on their computers and report any problems to Microsoft. Microsoft then examined the reports generated by the beta testers in an effort to correct any problems with Windows95. Once Microsoft was satisfied that it had identified and corrected the problems, it began publicly selling Windows95.

Although Aldridge had previously beta tested earlier versions of Windows, Microsoft did not send him a beta-test version of Windows95, nor did Aldridge ask for one. During one of the last rounds of beta testing Aldridge obtained a beta-test version of Windows95 from a third party. In April of 1995, four months before the general release of Windows95, Aldridge discovered that Windows95 would display messages when it detected Cache86 and would shift into DOS-compatibility mode.

Sales of Cache86 began to decline even before the general release of Windows95. Aldridge later developed an upgraded version of Cache86 (version 6.0) that functioned like SMARTDRV and was compatible with Windows95. The release of version 6.0 had little impact on the plummeting sales of Cache86, however. In addition to the decline in sales of individually packaged units of Cache86, Aldridge claims to have lost at least one contract with a major software designer, Quarterdeck, to include Cache86 in a software program designed to improve performance of computer games.

Aldridge blames Microsoft for the lost sales of Cache86. Aldridge alleges that Microsoft is liable under Texas law for

(1) business disparagement,

(2) defamation,

(3) tortious interference with contract, and

(4) tortious interference with business relations.

Aldridge also alleges that Microsoft is guilty of actual and attempted monopolization in violation of section 2 of the Sherman Antitrust Act, 15 U.S.C. § 2. In support of the antitrust claims Aldridge relies on the doctrines of product disparagement and essential facilities.

F. Microsoft's Motion for Summary Judgment

Microsoft has moved for summary judgment, asserting truth as a defense to all claims. With respect to Aldridge's antitrust claims Microsoft also argues that any damages from product disparagement were de minimis, that Windows95 is not an essential facility, and that Microsoft did not deny access to Windows95.

A motion for summary judgment should be granted if the record, taken as a whole, "together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to judgment as a matter of law." Fed.R.Civ.P. 56(c); New York Life Ins. Co. v. Travelers Ins. Co., 92 F.3d 336, 338 (5th Cir. 1996); Rogers v. International Marine Terminals, Inc., 87 F.3d 755, 758 (5th Cir. 1996). The Supreme Court has interpreted the plain language of Rule 56 as mandating "the entry of summary judgment, after adequate time for discovery and upon motion, against a party who fails to make a showing sufficient to establish the existence of an element essential to that party's case, on which that party will bear the burden of proof at trial." Celotex Corp. v. Catrett, 477 U.S. 317, 322, 106 S.Ct. 2548, 2552, 91 L.Ed.2d 265 (1986).

As the party moving for summary judgment Microsoft "must `demonstrate the absence of a genuine issue of material fact,' but need not negate the elements of the nonmovant's case." Little v. Liquid Air Corp., 37 F.3d 1069, 1075 (5th Cir. 1994) (en banc) (emphasis in original) (quoting Celotex, 477 U.S. at 323-25, 106 S.Ct. at 2553). "The movant accomplishes this by informing the court of the basis for its motion, and by identifying portions of the record which highlight the absence of genuine factual issues." Rizzo v. Children's World Learning Ctrs., Inc., 84 F.3d 758, 762 (5th Cir. 1996) (citing Topalian v. Ehrman, 954 F.2d 1125, 1131 (5th Cir. 1992)). If the moving party "fails to meet this initial burden, the motion must be denied, regardless of the nonmovant's response." Little, 37 F.3d at 1075.

If Microsoft meets this burden Rule 56 requires Aldridge, as the nonmovant, to go beyond the pleadings and to show by affidavits, depositions, answers to interrogatories, admissions on file, or other admissible evidence that specific facts exist over which there is a genuine issue for trial. EEOC v. Texas Instruments, Inc., 100 F.3d 1173, 1180 (5th Cir. 1996); Wallace v. Texas Tech Univ., 80 F.3d 1042, 1046-47 (5th Cir. 1996). The nonmovant's burden may not be satisfied by conclusory allegations, unsubstantiated assertions, metaphysical doubt as to the facts, or a mere scintilla of evidence. Wallace, 80 F.3d at 1047; Little, 37 F.3d at 1075.

Factual controversies are to be resolved in favor of the nonmovant, "but only when there is an actual controversy, that is, when both parties have submitted evidence of contradictory facts." Wallace, 80 F.3d at 1047; accord, S.W.S. Erectors, Inc. v. Infax, Inc., 72 F.3d 489, 494 (5th Cir. 1996). The court will not, "in the absence of any proof, assume that the nonmoving party could or would prove the necessary facts." McCallum Highlands, Ltd. v. Washington Capital Dus, Inc., 66 F.3d 89, 92 (5th Cir.), as modified, 70 F.3d 26 (5th Cir. 1995). Unless there is sufficient evidence for a reasonable jury to return a verdict in the nonmovant's favor, there is no genuine issue for trial. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 249-51, 106 S.Ct. 2505, 2511, 91 L.Ed.2d 202 (1986); Texas Instruments, 100 F.3d at 1179.

II. STATE LAW CLAIMS

A. Burden of Proof

Microsoft argues that none of Aldridge's state law claims have merit because all of the Windows95 messages are true. Whether falsity is an element of Aldridge's state law claims, or an affirmative defense to these claims, is an important issue because this distinction governs which party has the burden of proof. If truth is an affirmative defense Microsoft has the burden of proof at trial, and to prevail on its summary judgment motion Microsoft must prove that the messages are true. See Exxon Corp. v. Oxxford Clothes, Inc., 109 F.3d 1070, 1074 (5th Cir.), cert. denied, ___ U.S. ___, 118 S.Ct. 299, 139 L.Ed.2d 231 (1997). On the other hand, if falsity is an element of Aldridge's causes of action, Aldridge must produce summary judgment evidence that would support a jury finding of falsity to avoid summary judgment for Microsoft. See EEOC v. Texas Instruments, Inc., 100 F.3d 1173, 1180 (5th Cir. 1996). In this context truth is a negative defense. See Emmons v. Southern Pac. Transp. Co., 701 F.2d 1112, 1118 (5th Cir. 1983) (explaining that negative defenses are those tending to disprove at least one element of a plaintiff's cause of action).

1. Business Disparagement

To recover damages for the tort of business disparagement under Texas law a plaintiff must prove

(1) publication by the defendant of disparaging words,

(2) that were false,

(3) that were published with malice,

(4) the absence of any privilege, and

(5) that plaintiff suffered special damages.

See Hurlbut v. Gulf Atl. Life Ins. Co., 749 S.W.2d 762, 766 (Tex. 1987). Because falsity is an element of a business disparagement claim, truth is a negative defense. Aldridge must raise a fact issue as to the falsity of the Windows95 messages to avoid summary judgment on this claim.

2. Defamation

There are two forms of defamation; libel and slander. In Texas libel is defined by statute:

A libel is a defamation expressed in written or other graphic form that tends to blacken the memory of the dead or that tends to injure a living person's reputation and thereby expose the person to public hatred, contempt or ridicule, or financial injury or to impeach any person's honesty, integrity, virtue, or reputation or to publish the natural defects of anyone and thereby expose the person to public hatred, ridicule, or financial injury.

Tex.Civ.Prac. Rem. Code Ann. 73.001 (Vernon 1986). Although Texas has no parallel statutory provision for slander, Texas courts recognize a common-law cause of action for slander. See Cain v. Hearst Corp., 878 S.W.2d 577, 580 (Tex. 1994). Because Microsoft's messages appear in graphic form they are governed by the law of libel.

The Texas libel statute does not list falsity as an element, and in a later section of the statute truth is identified as a defense. See Tex.Civ.Prac. Rem. Code Ann. § 73.005 (Vernon 1986). Because falsity is not a statutorily required element of a libel claim, but an independent ground for denying recovery, the statutory scheme implies that truth is an affirmative defense on which a defendant has the burden of proof at trial.

Texas appellate courts are divided on whether truth is an affirmative defense to a libel claim. Compare Frank B. Hall Co. v. Buck, 678 S.W.2d 612, 623 (Tex.App. — Houston [14th Dist.] 1984, writ ref'd n.r.e.) ("For many years the courts of Texas have held in libel and slander actions that truth of the defamatory statements is an affirmative defense, and the burden of proving truth by a preponderance of the evidence is on the defendant."), with A.H. Belo Corp. v. Rayzor, 644 S.W.2d 71, 79-80 (Tex.App.-Fort Worth 1982, writ ref'd n.r.e.) (holding that the First Amendment requires the plaintiff to prove falsity when claiming defamation).

The Texas Supreme Court has not squarely addressed the issue. In Cain v. Hearst Corp., 878 S.W.2d 577 (Tex. 1994), the court was asked to decide if Texas recognized the tort of false-light invasion of privacy. One of the reasons the court held that Texas does not recognize that tort was because falselight invasion of privacy duplicates the existing cause of action for defamation. In particular, the court observed that both torts require proof of falsity. Cain, 878 S.W.2d at 580.

In Randall's Food Markets, Inc. v. Johnson, 891 S.W.2d 640, 646 (Tex. 1995), however, the court characterized truth as an affirmative defense to a slander action. Johnson worked as a store manager for Randall's. One day she bought several items from the store but failed to pay for a large wreath she held in her hands at the checkout station. The checkout clerk charged Johnson for the other items and asked her if there was anything else to ring up; Johnson said no. Randall's investigated the incident and eventually suspended Johnson, who then sued on a variety of theories, including slander. Johnson alleged that statements made by employees during the investigation were slanderous. Because the statements were true and protected by the employer's qualified investigative privilege, the Texas Supreme Court affirmed summary judgment in favor of Randall's. The court made no reference to the statement in Cain that "defamatory statements must be false in order to be actionable." Cain, 878 S.W.2d at 580. Instead, the court stated that "[i]n suits brought by private individuals, truth is an affirmative defense to slander." Randalls, 891 S.W.2d at 646 (footnote omitted).

The Fifth Circuit has interpreted Texas law as requiring the plaintiff to prove falsity as an element of a cause of action for defamation. See Brown v. Petrolite Corp., 965 F.2d 38, 43 (5th Cir. 1992) ("A defamation plaintiff must prove that the allegedly defamatory language is false."). Because this court is bound to follow Fifth Circuit precedent on the interpretation of Texas substantive law, the court will require Aldridge to prove that the messages were false to establish his defamation claim.

3. Tortious Interference with Contract and with Business Relations

To recover on a claim of tortious interference with contract a plaintiff must prove

(1) the existence of a contract,

(2) a willful and intentional act by the defendant that interfered with the contract,

(3) the willful and intentional act proximately caused damage to the plaintiff, and

(4) actual damages.

See Texas Beef Cattle Co. v. Green, 921 S.W.2d 203, 210 (Tex. 1996); Browning-Ferris, Inc. v. Reyna, 865 S.W.2d 925, 926 (Tex. 1993); Victoria Bank Trust Co. v. Brady, 811 S.W.2d 931, 939 (Tex. 1991); see also Thrift v. Hubbard; 44 F.3d 348, 356 (5th Cir. 1995).

To recover for tortious interference with business relations a plaintiff must prove

(1) a reasonable probability that the plaintiff would have entered a contract,

(2) willful and intentional acts by the defendant effectively interfered with the prospective business relationship, and

(3) the plaintiff suffered actual harm. See Kiepfer v. Beller, 944 F.2d 1213, 1220 (5th Cir. 1991).

Legal justification is an affirmative defense to both causes of action. Id. (tortious interference with business relations); Texas Beef Cattle Co., 921 S.W.2d at 210 (tortious interference with contract); Sterner v. Marathon Oil Co., 767 S.W.2d 686, 690 (Tex. 1989) (tortious interference with contract). The affirmative defense of legal justification protects a defendant who provides truthful information about a plaintiff to third parties. Tarleton State Univ. v. Rosiere, 867 S.W.2d 948 (Tex.App.-Eastland 1993, writ dism'd by agr.). As the Restatement of Torts explains, "[o]ne who intentionally causes a third person not to perform a contract or not to enter into a prospective contractual relation with another does not interfere improperly with the other's contractual relation, by giving the third person truthful information." Restatement (Second) of Torts § 772 (1979); accord id. § 772 cmt. b ("There is of course no liability for interference with a contract or with a prospective contractual relation on the part of one who merely gives truthful information to another.").

Because the falsity of Microsoft's messages is an element of Aldridge's business disparagement and defamation claims, to avoid summary judgment on these claims Aldridge must present evidence that the messages are false. Conversely, because truth is an affirmative defense to Aldridge's tortious interference claims, Microsoft must prove that the messages are true to prevail on its motion for summary judgment on these claims.

B. The Evidence

Aldridge generally alleges that the four messages falsely state that Cache86 reduces overall system performance. He also alleges that the following specific statements are false:

(1) Cache86 is a new program (as to Message One),

(2) Cache86 may decrease overall system performance (as to Message One),

(3) Cache86 is a problem (as to Messages One and Four), (4) Cache86 forces MS-DOS compatibility mode (as to Messages Two and Four),

(5) MS-DOS compatibility mode reduces overall system performance (as to Messages Two and Four),

(6) Cache86 is unknown to Windows95 (as to Messages Three and Four),

(7) Windows95 will not perform optimally while Cache86 is on the system (as to Message Three),

(8) Windows95 switched to MS-DOS compatibility mode to ensure that Cache86 will run (as to Message Four), and

(9) performance will improve if Cache86 is removed (as to Messages Three and Four).

Because many of Aldridge's allegations represent only minor semantic differences, and because neither Aldridge nor Microsoft discusses all of the allegations separately in their motion papers, the court will group them into three general categories:

(1) Was Cache86 new and unknown to Windows95,

(2) does Cache86 decrease system performance, and

(3) does Windows95 call Cache86 a problem, and if so, is this statement true?

1. Was Cache86 new or unknown to Windows95?

Aldridge argues that Message One calls Cache86 a new program, despite the fact that Cache86 has been on the market since 1989. Windows95 considers "new" any program installed after Windows95 had last been run on the computer. Aldridge admits that Windows95 only labels Cache86 as new when a user has "newly introduced" the program on a computer running Windows95. The evidence also establishes that Cache86 was unknown to Microsoft. The software engineer responsible for including the warning messages testified that he had never heard of Cache86 before this lawsuit. Windows95 will generate the warning messages only if a program that hooks Interrupt 13 is not on the safe list. Because Cache86 is not on the safe list it is accurate for Windows95 to characterize Cache86 as unknown. Because it is accurate for Windows95 to call Cache86 new and unknown, these statements cannot support any of Aldridge's state law claims.

2. Does Cache86 decrease system performance?

Microsoft argues that results of tests conducted both by Microsoft and Aldridge confirm that Cache86 decreases performance of the Windows95 operating system. First, Microsoft offers the results of tests conducted by one of its employees, Steve Goan, and repeated by its expert, Jeffrey Brumfield. Goan used a pair of computer programs, Winstone 96 and Winbench 96, commonly employed to evaluate computer performance. Microsoft argues that these tests show that computers using Windows95 do not perform as well with Cache86.

Aldridge's expert, Gavin Clarkson, questions the reliability of these test results. Goan ran the tests on computers having only eight megabytes of RAM. Clarkson states that users of disk cache programs usually have more than eight megabytes of RAM on their computers. Because a computer's performance is directly related to the amount of RAM, Clarkson states that the amount of memory in the computer influenced the performance scores on Microsoft's tests. Although Microsoft protests that Aldridge presents no evidence that a typical computer user had more than eight megabytes of RAM when Windows95 was first marketed, Clarkson, who was himself a disk cache program user, had at least sixteen megabytes of RAM on his computers at that time. This testimony raises a fact issue whether Cache86, or the amount of memory on the computers, caused the low Winstone and Winbench test scores.

Clarkson also states that Winstone and Winbench were designed to test hardware performance and are not appropriate for evaluating the performance of software. Microsoft counters that Clarkson once worked for a company that used the programs to test software, that Microsoft used Winstone and Winbench during the development of Windows95, and that Aldridge also used these test programs to evaluate software performance. Although Microsoft's evidence undercuts Clarkson's opinion, he has nevertheless raised a fact issue whether Winstone and Winbench are proper testing tools.

Clarkson also criticizes Goan for not properly controlling the testing environment. Clarkson states that "running the same Winstone test on the same machine twice should net the same result." Yet, two separate Winstone tests yielded somewhat different scores. One test produced a score of 12.70 on an undefined rating scale, while another test produced a score of 12.47. Without an explanation of how Winstone calculates its results, however, the court cannot determine whether the 0.23 difference in test scores has any statistical significance or is within the expected scientific error. The court therefore concludes that Aldridge has not shown that this difference undermines the integrity of the Winstone test results.

Microsoft also argues that tests run by Aldridge confirm that Cache86 decreases the performance of the Windows95 operating system. David Aldridge and his staff set up a series of tests to measure the time for a computer to perform twenty-eight common functions run two times on two different hard-disk drives (a total of 112 tests). With Cache86, Windows95 performed each series of tests faster (by 13, 4, 23, and 15 seconds, respectively) than Windows95 alone. Although at first blush these results appear to contradict its argument, Microsoft asks the court to disregard the first, or "boot-up," function in each series of Aldridge's tests. Since users boot-up a computer only once a day, Microsoft argues that booting up is not a normal operation, and that test times that include boot-up time do not accurately reflect overall system performance. If the test times are recalculated to exclude boot-up, Windows95 with Cache86 performed slower in the first, second, and fourth series of tests. The court is not persuaded by this argument because the senior Microsoft software designer admitted that boot-up time is relevant to system performance. At the very least, Aldridge has raised a fact issue on this point.

However, Microsoft does raise a salient point about Aldridge's tests. Aldridge focuses on the aggregate time for each series of functions and ignores the fact that many individual functions were slower with Cache86. Cache86 caused slower performance in nine of the first series of tests, seventeen of the second series of tests, eight of the third series of tests, and thirteen of the fourth series of tests, for a total of 47 out of 112 slower functions.

To summarize the relevance of the parties' test results on the truthfulness of the Windows95 messages about Cache86's performance, Aldridge has raised a fact issue as to the reliability of Microsoft's Winstone and Winbench tests. The tests performed by Aldridge show that Cache86 does slow the performance of Windows95 in some circumstances.

As an alternative to arguing that tests prove that Cache86 always improves performance, Aldridge argues that Cache86 improves performance when programs other than Cache86 cause Windows95 to shift into DOS-compatibility mode. Aldridge claims that since DOS-compatibility mode performs read-write operations the way MS-DOS and Windows do, Cache86 necessarily provides the same performance advantages to that mode as it does for DOS and Windows. Aldridge also claims that Cache86 always improves performance in real-mode DOS.

Because Aldridge's argument only focuses on whether Cache86 improves performance of the two subordinate modes, the court concludes that it is not relevant to the broader question whether Cache86 decreases overall Windows95 system performance. Aldridge's argument would be relevant only to the extent that any of the four Microsoft messages can be read as stating that Cache86 always decreases Windows95 performance. Having examined the parties' summary judgment evidence of the effect of Cache86 on Windows95's performance, the court turns to the actual text of each message to determine if the statements about performance are false.

(a) Message One

Message One states:

A new MS-DOS resident program named C86EXTEN may decrease your system's performance. Would you like to see more information about this problem?

The parties agree that Microsoft defines system performance as "the overall speed at which a computer performs a wide variety of functions." Message One states that Cache86 may decrease performance; it does not affirmatively declare that Cache86 will do so. Clarkson performed three different tests designed to simulate typical uses of a personal computer. These tests measured the speed at which the computer operated a spreadsheet, a word processor, and web graphics program. Windows95 ran faster with Cache86 than without it in the spreadsheet and word processing tests. However, Windows95 ran 2.4859 seconds slower with Cache86 on the web graphics test. Cache86 thus improved performance for two "typical" uses, but decreased performance in the third. The tests run by David Aldridge's employees also showed that Cache86 decreases performance in many circumstances. While Cache86 yielded faster aggregate times for each of the four series of tests, Cache86 caused slower performance in many of the individual functions. Because both David Aldridge's and Clarkson's tests show that Cache86 sometimes decreased performance, the court concludes that Aldridge has not shown that Message One falsely states that Cache86 decreases performance.

(b) Message Two

Message Two states:

C86EXTEN in AUTOEXEC.BAT forces MS-DOS compatibility mode. Compatibility mode paging reduces overall system performance.

Windows95 shifts into MS-DOS compatibility mode when it detects an unfamiliar program, such as Cache86, hooking Interrupt 13. The very manner in which Cache86 functions compels Windows95 to shift modes to avoid the danger of data corruption. Message Two does not state that Cache86 reduces overall system performance; it states that " [c]ompatibility mode paging reduces overall system performance." Aldridge argues that the court should assume that most users will not understand the term "compatibility mode paging." This argument may support a claim that the message is confusing, but not that it is false. The court concludes that Aldridge has failed to present evidence that Message Two is false.

(c) Message Three

Message Three states:

Windows was unable to identify the specified memory-resident program which was loaded in your Autoexec.bat file. If you no longer need this program, remove it from your Autoexec .bat file. Otherwise, contact the manufacturer of this program to see if an updated version is available. Windows will not perform optimally while this program is loaded.

This message tells users that Windows95 will not perform optimally because of Cache86. Neither party has explained what "optimal" performance is. The Oxford English Dictionary defines optimal as "[b]est or most favorable, most satisfactory." 10 Oxford English Dictionary 876 (2d ed. 1989). A fair reading of Message Three tells users that Windows95 will not perform at its best as long as Cache86 remains on the system. Unlike the conditional language of Message One, this is a more universal statement. Aldridge's tests and those conducted by his expert raise a fact issue whether this statement is always true.

(d) Message Four

Message Four states:

Windows was unable to identify a real-mode driver or memory-resident program loaded in your Config.sys or Autoexec.bat file. You will see another message that lists the name of the driver or program causing the problem.

Because Windows could not identify the driver or program, it has switched to MS-DOS compatibility mode to ensure the program will run. However, this slows down overall performance. To improve performance, remove the program or driver causing the problem, or contact its manufacturer for an updated version.

Message Four notifies users that Windows95 has switched modes and explains why. Thus far, the message is true. It then says that this switch in modes slows down overall performance. However, Aldridge's tests and those of his expert demonstrate that in some instances performance does not decline. For example, in the four series of twenty-eight tests run by Aldridge's employees, performance time did not change in 46 of the 112 tests. In 19 of the tests performance time improved when Cache86 ran in DOS-compatibility mode. Aldridge has thus produced some evidence that Message Four is false when it unequivocally states that overall performance slows down.

3. Does Windows95 call Cache86 a problem, and if so, is this statement true?

Aldridge argues that Messages One and Four falsely characterize Cache86 as the problem that causes decreased performance. "This problem" as used in Message One clearly refers to the problem of decreased system performance, not to Cache86. The Microsoft software designers responsible for drafting the message confirmed that they intended the phrase "this problem" to refer to decreased system performance, not to Cache86 or any other program. Furthermore, even if the message could be read as calling Cache86 a problem, it only characterizes Cache86 as a potential problem. The message states that Cache86 " may decrease" performance; it does not say that Cache86 will decrease performance. As explained above, such a statement is true.

Message Four makes two references to a problem. The first paragraph states "Windows was unable to identify a real-mode driver or memory-resident program loaded in your Config.sys or Autoexec.bat file. You will see another message that lists the name of the driver or program causing the problem" (emphasis added). The "problem" here is the inability to identify the triggering program. Cache86 is not on the disabling list in SETUPC.INF or on the safe list in IOS.INI. Thus, the statements in this paragraph are true.

The second paragraph of Message Four states:

Because Windows could not identify the driver or program, it has switched to MS-DOS compatibility mode to ensure the program will run. However, this slows down overall performance. To improve performance, remove the program or driver causing the problem, or contact its manufacturer for an updated version. (Emphasis added.)

It is unclear whether the problem referred to in this paragraph is the inability of Windows95 to identify Cache86 or slow overall performance. Which meaning Microsoft intended is an issue of fact for the jury. While a statement that the inability to identify Cache86 was a problem would be true, a statement that Cache86 causes the problem of slow overall performance would raise an issue of fact as to its truthfulness. The court concludes that Aldridge has raised an issue of fact as to Message Four.

C. Conclusion

There is no issue of material fact as to Messages One and Two. Aldridge has failed to support the business disparagement and defamation claims by producing evidence of the falsity of Messages One and Two, and Microsoft has met its affirmative burden on the tortious interference claims by establishing the truth of these two messages. The court will therefore grant Microsoft's motion for summary judgment on all of Aldridge's state law claims as to Messages One and Two. Because issues of material fact preclude summary judgment as to Messages Three and Four, and because Microsoft relies solely on its defenses of truth in its motion for summary judgment on all of the state-law claims, the court will deny summary judgment as to Messages Three and Four.

III. ANTITRUST CLAIMS

Aldridge alleges that Microsoft violated section two of the Sherman Act, 15 U.S.C. § 2, under the theories of product disparagement and essential facilities. As a preliminary matter Microsoft challenges Aldridge's standing to assert an antitrust claim.

A. Antitrust Standing

The Clayton Act, 15 U.S.C. § 15(a), limits the class of private persons who may sue on antitrust violations to "any person who shall be injured in his business or property by reason of anything forbidden in the antitrust laws." To demonstrate antitrust standing Aldridge must show

(1) injury-in-fact,

(2) antitrust injury, and

(3) status as a proper plaintiff.

Doctor's Hosp. of Jefferson, Inc. v. Southeast Med. Alliance, Inc., 123 F.3d 301, 305 (5th Cir. 1997) (citing McCormack v. National Collegiate Athletic Ass'n, 845 F.2d 1338, 1341 (5th Cir. 1988)). In analyzing Aldridge's standing under this test the court should assume that Microsoft has committed the antitrust violations alleged and then determine whether Aldridge has shown the three elements of standing. Id. at 306.

To establish the first element, injury-in-fact, Aldridge must show an injury to its business or property proximately caused by the conduct of Microsoft. See Doctor's Hospital, 123 F.3d at 305; United Indus., Inc. v. Eimco Process Equip. Co., 61 F.3d 445, 448 (5th Cir. 1995); Ancar v. Sara Plasma, Inc., 964 F.2d 465, 468 (5th Cir. 1992). Aldridge's evidence shows a significant decline in sales of Cache86 after November of 1994 when Micrsoft began releasing early versions of Windows95 for beta testing and establishes the necessary causal link between the claimed injury (lost sales) and the alleged antitrust violations (denial of access to Windows95 and disparagement of Cache86).

Aldridge must next show antitrust injury. This is different from the requirement that a plaintiff show market-wide injury to competition, which is often a component of substantive antitrust liability. Doctor's Hosp., 123 F.3d at 305. To demonstrate the antitrust injury element of standing a plaintiff must show anticompetitive effects of the defendant's actions; in other words, antitrust injury should be the "type of loss that the claimed violations of the antitrust laws would be likely to cause." Zenith Radio Corp. v. Hazeltine Research, Inc., 395 U.S. 100, 125, 89 S.Ct. 1562, 1577, 23 L.Ed.2d 129 (1969); see Bell v. Dow Chem. Co., 847 F.2d 1179, 1182 n. 4 (5th Cir. 1988). Lost sales and declining market share, as alleged by Aldridge and supported by Aldridge's summary judgment evidence, are the types of losses that would likely result from an anticompetitive campaign of product disparagement or denial of access to an essential facility.

To establish the third element of antitrust standing Aldridge must show that it is a proper plaintiff. In evaluating this element the court must examine

(1) whether Aldridge's injuries or their causal link to Microsoft are speculative,

(2) whether other parties have been more directly harmed, and

(3) whether allowing Aldridge to sue would risk multiple lawsuits, duplicative recoveries, or complex damage apportionment.

B. Product Disparagement

Microsoft argues that Aldridge cannot succeed on the product disparagement claim because the allegedly disparaging information is true and because the anticompetitive effect of any disparaging remarks is de minimis. An evaluation of whether Aldridge's damages were de minimis encompasses the question of whether the statements were false. The court will therefore consider Microsoft's truth defense as part of its evaluation of Microsoft's de minimis defense.

Even if a plaintiff shows falsity it must overcome a presumption that the disparagement had only a de minimis effect on competition. American Prof'l Testing Serv., Inc. v. Harcourt Brace Jovanovich Legal Prof'l Publications, Inc., 108 F.3d 1147, 1151, 1152 (9th Cir. 1997); Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263, 288 n. 41 (2d Cir. 1979); see IIIA Areeda Turner, supra, § 782b, at 260-61 (rev. ed. 1996). Courts and commentators presume a de mmimis effect for three reasons:

(1) it is generally quite difficult to distinguish mere puffery from patently false statements relied upon by a consumer;

(2) the effects of disparagement are speculative at best, especially when the disparagement is not systematic; and

(3) most consumers view statements about a competitor cynically, recognizing the inherent bias and lack of objectivity of such statements.

At the suggestion of Professors Areeda and Hovenkamp, courts have adopted a strenuous test for rebutting the de minimis presumption. To satisfy this test a plaintiff must prove that the allegedly disparaging statements were

(1) clearly false,

(2) clearly material,

(3) clearly likely to induce reasonable reliance,

(4) made to consumers having little understanding of the subject matter,

(5) continued for extended time periods, and

(6) not readily susceptible to counter statement, explanation, or other neutralizing effort or offset by the plaintiff.

Aldridge concedes that falsity is an essential element of its product disparagement claim. The alleged false statement or misrepresentation must amount to more than mere puffery by the defendant. General Communications Eng'g, Inc. v. Motorola Communications Elecs., Inc., 421 F. Supp. 274, 289-90 (N.D.Cal. 1976); see IIIA Areeda Hovenkamp, supra, § 782d, at 263 (rev. ed. 1996). Because Aldridge has failed to show that Message One and Two are false, proof of Aldridge's product disparagement claim must rely on Messages Three and Four. The court is persuaded that a fact issue exists as to the clear falsity of these two messages.

In addition to showing clear falsity, Aldridge must show that the challenged messages are clearly material. A material fact is one "that is of consequence to the determination of the action." Fed.R .Evid. 401; see 22 Charles Alan Wright Kenneth W. Graham, Jr., Federal Practice Procedure § 5164, at 40 (1978). Aldridge argues that the messages meet this factor because they bring into question the marketability of Cache86 by telling users that it may or does decrease performance. The court agrees that Aldridge has raised a fact issue as to the clear materiality of the four messages.

Aldridge must next show that the challenged messages were clearly likely to induce reliance. Message One is captioned as a "Performance Warning," and the message box includes an exclamation point, which implies urgency. The message invites the user to view further information, suggesting that the operating system has anticipated these circumstances and can explain what is happening. Message Two appears in a message box captioned as "System Properties." In addition to the message about Cache86, the box contains diagnostic information about the computer, such as the amount of available memory, the current mode in which Windows95 is operating, and whether certain auxiliary hardware devices (PCMCIA cards, for example) are running. Messages Three and Four appear in boxes captioned "Windows Help," implying that the information contained in the messages will assist the user. Microsoft acknowledges that it sought to make all four messages understandable to assist the user in knowing what to do. The court concludes that Aldridge has raised a fact issue as to the reliance factor.

Aldridge also has established the fourth factor — that Microsoft published the messages to consumers who had little understanding of the subject matter. Microsoft admits that it assumed users would not understand terms like DOS-compatibility mode.

To satisfy the fifth factor Aldridge must show that the messages were displayed for long periods of time. Aldridge argues that "the warnings are continued for a prolonged period because Windows95 generates them every time it detects CACHE86 on the computer." The court does not agree. Message One appears only once. When the user acknowledges Message One it disappears from the video screen and will not reappear unless the user takes Cache86 off the computer and later reinstalls it. None of the three other messages automatically appear on the screen; the user must seek them out. None of the other messages remain on display after the computer is turned off. Because the messages do not appear for Extended periods of time, the court concludes that Aldridge has failed to raise a fact issue as to the fifth rebuttal factor.

The sixth factor requires Aldridge to show that the messages were not readily susceptible to explanation or neutralization. Aldridge argues it had no way of neutralizing these warnings until it was too late, and that it did not receive a copy of Windows95 until April of 1995, four months before the general release of Windows95. Aldridge also argues that Microsoft should have informed it about the new read-write system and told it how to avoid triggering the switch to DOS-compatibility mode.

The court is not persuaded by these arguments. Microsoft did not have an affirmative duty to predisclose the messages or the new design of Windows95 to Aldridge. As the Second Circuit has observed:

[A] firm may normally keep its innovations secret from its rivals as long as it wishes, forcing them to catch up on the, strength of their own efforts after the new product is introduced. . . . It is the possibility of success in the marketplace, attributable to superior performance, that provides the incentives on which the proper functioning of our competitive economy rests. If a firm that has engaged in the risks and expenses of research and development were required in all circumstances to share with its rivals the benefits of those endeavors, this incentive would very likely be vitiated.

Messages Three and Four encourage users to contact the manufacturer of Cache86 for an updated version of Cache86 as an alternative to removing the program Aldridge could have countered these messages by giving information to users who called. Even if no users responded, Aldridge could have designed an updated version of Cache86 that would operate under Windows95 protect mode without triggering the messages. (In fact, Aldridge eventually did market such a product, known as Cache86 version 6.0.) The fact that Aldridge took no action, despite knowing that Windows95 was being developed and despite learning from beta-testers at least a year before Windows95 was introduced that it would display warning messages about Cache86, severely undercuts Aldridge's argument that the messages were not susceptible to explanation or neutralization. The court concludes that Aldridge has failed to raise a fact issue as to the sixth factor.

Because Aldridge has failed to show that Messages One and Two are clearly false and has failed to show that any of the four messages continued for extended periods or that they were not readily susceptible to explanation or neutralization, the court concludes that Aldridge has failed to rebut the strong presumption that any damage caused by Microsoft's allegedly disparaging messages was de minimis. Aldridge's claim of anticompetitive product disparagement therefore fails.

C. Essential Facilities

Aldridge alleges that Windows95 is an essential facility because all software must operate under an operating system, Windows95 is the most prevalent system, and if software cannot function under Windows95 it is not usable to most computer users. Aldridge argues that by disabling or disparaging all competing disk cache programs Microsoft denied Aldridge access to Windows95. Aldridge also contends that Microsoft prevented disk cache developers from updating their programs to operate under Windows95, and thereby continuing to compete in the disk cache market, by not disclosing design details about Windows95 to disk cache developers. Microsoft responds that Windows95 is not an essential facility, and even if it were, Microsoft did not deny Aldridge access to it.

Often criticized, the essential facilities doctrine nevertheless remains a viable part of the federal antitrust laws. The doctrine protects a competitor from having to enter an additional market with large capital requirements in order to compete. Fishman v. Estate of Wirtz, 807 F.2d 520, 540 (7th Cir. 1986). "The core concern of the doctrine is that a monopolist possesses a resource that is `essential' in some sense for the business of someone else, but that the monopolist refuses to share." IIIA Areeda Hovenkamp, supra, ¶ 772a, at 177 (rev. ed. 1996). By controlling the "essential" facility, a monopolist can potentially extend its monopoly power "from one stage of production to another, and from one market into another." MCI Communications Corp. v. AT T, 708 F.2d 1081, 1132 (7th Cir. 1983); accord Interface Group, Inc. v. Massachusetts Port Auth., 816 F.2d 9, 12 (1st Cir. 1987); IIIA Areeda Hovenkamp, supra, ¶ 771a, at 172 (rev. ed. 1996). To prevent this potential expansion of monopoly power the Sherman Act imposes a duty on the owner of an essential facility to open it to competitors on a nondiscriminatory basis. Ferguson v. Greater Pocatello Chamber of Commerce, Inc., 848 F.2d 976, 983 (9th Cir. 1988); Hendricks Music Co. v. Steinway, Inc., 689 F. Supp. 1501, 1509 (N.D.Ill. 1988); IIIA Areeda Hovenkamp, supra, ¶ 771a, at 172 (rev. ed. 1996).

To prevail on an essential facilities claim a plaintiff must establish that

(1) the defendant is a monopolist,

(2) the facility is essential,

(3) the defendant has the type of control over the facility that is forbidden by the Sherman Act,

(4) duplication of the facility is unreasonable or impractical,

(5) the defendant denied the plaintiff use of the facility, and

(6) providing access to the plaintiff was feasible.

See Twin Lab., Inc. v. Weider Health Fitness, 900 F.2d 566, 569, 570 (2d Cir. 1990); Ferguson, 848 F.2d at 983; MCI Communications Corp., 708 F.2d at 1132; City of College Station v. City of Bryan, 932 F. Supp. 877, 887 (S.D.Tex. 1996); TCA Bldg. Co. v. Northwestern Resources Co., 873 F. Supp. 29, 39 (S.D.Tex. 1995); International Audiotext Network, Inc. v. AT T Co., 893 F. Supp. 1207, 1220-21 (S.D.N.Y. 1994), aff'd, 62 F.3d 69 (2d Cir. 1995); IIIA Areeda Hovenkamp, supra, ¶ 773c, at 208 (rev. ed. 1996).

1. Is Microsoft a monopolist?

To establish that Microsoft is a monopolist Aldridge must show that Microsoft possesses monopoly power in the relevant geographic and product markets. In an essential facilities case the court must define several markets:

(1) the relevant product market for the alleged essential facility (the "primary product market"),

(2) the relevant product market for the potentially monopolized product (the "secondary product market"), and

(3) the relevant geographic markets.

Herbert Lyon, Aldridge's market share expert, concludes (and Microsoft does not dispute) that the relevant primary market is the worldwide market for personal computer operating systems, and the relevant secondary market is the United States market for disk cache programs. Aldridge must also show that Microsoft possessed the power to control prices or exclude competition in the market for personal computer operating systems. Lyon's report and affidavit conclude that Microsoft has such monopoly power, and Microsoft's summary judgment papers do not dispute this conclusion.

2. Is Windows95 essential?

The purpose of the antitrust laws is to promote competition. State Oil Co. v. Khan, ___ U.S. ___, ___, 118 S.Ct. 275, 282, 139 L.Ed.2d 199 (1997); Doctor's Hosp. of Jefferson, Inc. v. Southeast Med. Alliance, Inc., 123 F.3d 301, 306 (5th Cir. 1997). To show that a monopolist's facility is essential a plaintiff must first demonstrate that the facility is critical to the plaintiff's own competitive viability. See Fishman v. Estate of Wirtz, 807 F.2d 520, 539 (7th Cir. 1986); TCA Bldg. Co. v. Northwestern Resources Co., 873 F. Supp. 29, 39 (S.D.Tex. 1995); IIIA Areeda Hovenkamp, supra, ¶¶ 773a, 773b, at 199, 202 (rev. ed. 1996). The plaintiff must then demonstrate that sharing the facility is vital to enhancing competition in general. IIIA Areeda Hovenkamp, supra, ¶ 773b3, at 206 (rev. ed. 1996). A facility is essential under the antitrust laws only when it is vital to both the plaintiff's individual competitive viability and the viability of the market in general.

Aldridge only addresses the first requirement, arguing that without access to Windows95, "Cache86 cannot run, cannot be sold, and therefore cannot compete in the software market." The court agrees that Cache86, like any commercial software program, must be compatible with an operating system in order to function on a computer. However, Aldridge has not produced any evidence that Cache86 cannot run on other operating systems in the primary market, such as IBM's OS/2 system. In fact, Cache86 does run, as it was designed to do, on at least two other operating systems, MS-DOS and Windows. Lyon estimated that MS-DOS and Windows had a 21.5% share of the operating-system market in 1996, the year after Microsoft allegedly denied Aldridge access to Windows95. Aldridge offers no evidence that a 21.5% market share made those operating systems inadequate "facilities." Furthermore, although Lyon also predicts that the market share for MS-DOS and Windows will decline to 0.4% by 2000, Lyon presented no opinion or evidence about the present or projected market share of operating systems produced by firms other than Microsoft. Aldridge has failed to show that it cannot compete viably in the secondary market of disk cache programs by using operating systems other than Windows95.

Nor has Aldridge presented any evidence of the effect of Windows95 on competition in general in the secondary market for disk cache programs. While demand for Windows95-compatible disk cache programs may have declined sharply or even disappeared, Aldridge presents no evidence that the entire relevant secondary product market for disk cache programs has disappeared.

Lyon opines that by including its own disk cache software (SMARTDRV) with Windows95 and by disabling non-Microsoft disk cache programs, or displaying warnings about them, "Microsoft effectively owns 100% of the relevant market for disk cache utilities operating within the domain of the Windows95 operating system. But the relevant secondary market is disk cache programs in general, not disk cache programs operating within the domain of the Windows95 operating system. Although Lyon may be correct that there is effectively no competition in the "submarket" of Windows95-specific disk cache programs, that fact does not establish that Windows95 is vital to competitors in the disk cache program market.

Even had Aldridge brought evidence to support both requirements of this element, the court concludes as a matter of law that in this case Windows95 is not essential to the disk cache program market under the Sherman Act. Cache86 relies upon an imperfection in the design of MS-DOS and Windows — the practice of passing all read-write information between the hard disk and microprocessor through Interrupt 13. This imperfection caused slow performance, and Aldridge developed the Cache86 program to overcome it and to improve the speed of MS-DOS and Windows. Microsoft, the maker of those two operating systems, chose an alternative method for improving performance; it designed the new Windows95 operating system to avoid this imperfection in its primary mode. In essence, Microsoft's operating systems are essential to Aldridge only to the extent that the systems operate less efficiently.

Acceptance of Aldridge's arguments would mean that Microsoft could be punished under the antitrust laws for improving its own product. Such a result would inhibit, not promote, competition in the market. The antitrust laws do not require a competitor to maintain archaic or outdated technology; even monopolists may improve their products. See Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263, 281 (2d Cir. 1979) ("It is the possibility of success in the marketplace, attributable to superior performance, that provides the incentives on which the proper functioning of our competitive economy rests.").

The essentialness issue in this case presents a crucial distinction from previous cases applying the essential facilities doctrine. The plaintiffs here marketed a product, Cache86, whose sole purpose was to improve the operation of the allegedly monopolized product. In the usual essential facilities case the alleged facility is a conduit for the distribution of another product. Sports stadiums facilitate the display of indoor sports. See Fishman v. Estate of Wirtz, 807 F.2d 520, 532 (7th Cir. 1986); Hecht v. Pro-Football, Inc., 570 F.2d 982, 989 (D.C. Cir. 1977). Railroad bridges permit continuation of rail service and delivery of freight. See United States v. Terminal R.R. Ass'n, 224 U.S. 383, 392-94, 32 S.Ct. 507, 509, 56 L.Ed. 810 (1912). Telecommunications networks distribute information. See MCI Communications Corp. v. AT T, 708 F.2d 1081, 1093 (7th Cir. 1983). Ski mountains provide access to recreational skiing. See Aspen Highlands Skiing Corp. v. Aspen Skiing Co., 738 F.2d 1509, 1521 (10th Cir. 1984), aff'd on other grounds, 472 U.S. 585, 105 S.Ct. 2847, 86 L.Ed.2d 467 (1985).

While Windows95 does facilitate the use of application software, Cache86 is not such a product; it is a utility program designed to improve the distributing product. Windows95 is analogous to a long-distance telephone network. It is very doubtful that a long-distance carrier like AT T could be liable to a copper wire manufacturer under the essential facilities doctrine for switching from copper wire to faster fiber-optic lines. While it could be argued that access to AT T's long-distance network may be essential to competitors in the market for delivery of long-distance telephone service, the network would not be essential to the market for copper wire. Similarly, even were the court to assume for purposes of argument that Windows95 could be an essential facility for application software, such as a word processing program, it is not essential to a market whose sole purpose is to improve on imperfections in the facility at issue.

3. Does Microsoft have the type of control over the facility that is forbidden by the Sherman Act?

Courts have only applied the essential facilities doctrine to four situations:

(1) natural monopolies;

(2) facilities whose duplication is forbidden by law;

(3) publicly subsidized facilities that cannot practically be built privately; and

(4) the sole facility in the relevant geographic market, i.e., "the only facility in town."

Microsoft does not hold a natural monopoly over operating systems. Natural monopolies occur in industries having large initial fixed costs but declining marginal production costs. Greenwood Utils. Comm'n v. Mississippi Power Co., 751 F.2d 1484, 1502 (5th Cir. 1985). In such circumstances a single firm can produce more efficiently than multiple firms because the single firm can constantly expand its capacity at a much lower cost than the fixed costs required to enter the market. Fuchs v. Rural Elec. Convenience Coop. Inc., 858 F.2d 1210, 1212 n. 5 (7th Cir. 1988); Hecht v. Pro-Football, Inc., 570 F.2d 982, 990 n. 26 (D.C. Cir. 1977). While Microsoft may hold a natural monopoly over the distribution of its own products, that kind of monopoly does not offend the Sherman Act. See Sports Ctr., Inc. v. Riddell, Inc., 673 F.2d 786, 791 (5th Cir. 1982). To argue, as Aldridge appears to, that Microsoft holds a natural monopoly over Windows95 is therefore irrelevant. Aldridge has not shown that Microsoft holds a natural monopoly over the worldwide market for personal computer operating systems.

Nor has Aldridge shown that creation of a personal computer operating system is forbidden by law, that personal computer opererating systems are publicly subsidized facilities, or that Windows95 is the only operating system in the relevant geographic market. Instead, Aldridge argues that "[a]ntitrust violations based on essential facility claims are recognized whenever monopolists such as Microsoft deny equal access to a product that is essential for competition." The essential facilities doctrine is not that broad, however. The cases Aldridge relies on as authority for his argument confirm the narrowness of the doctrine because each falls within one of the four categories mentioned above. The court concludes that Aldridge has not established that Microsoft's control of Windows95 is the type of control that is forbidden by the Sherman Act.

4. Is duplication of an operating system unreasonable or impractical?

Because Microsoft does not address this factor, the court will assume that it is unreasonable or impractical for a disk cache program manufacturer to produce its own operating system in order to gain market access for its programs.

5. Has Microsoft denied Aldridge access to the Windows95 operating system?

Aldridge argues that Microsoft denied access to Windows95 by encrypting the Windows95 program code, by not disclosing functional details about the program to Aldridge before releasing Windows95, and by designing the operating system "to disable or disparage all competing disk caches." Microsoft responds that Aldridge never asked for any information about the design or operation of Windows95 and never asked to be a beta tester for the operating system.

To establish denial of access a plaintiff must have at the very least sought access to the facility. City of Vernon v. Southern Cal. Edison Co., 955 F.2d 1361, 1367 (9th Cir. 1992). David Aldridge admitted that he did not ask for any information that would have allowed him to upgrade Cache86 for Windows95. Nor did he ask for a beta version of Windows95. Instead, he expected Microsoft to invite him into the beta-testing program, as it had with respect to previous software products. "[A] manufacturer is under no obligation to pre-disclose or disclose its knowledge about its products so that competition may arise in the related peripheral hardware, software, and repair service markets." Data Gen. Corp. v. Grumman Sys. Support Corp., 761 F. Supp. 185, 192 (D.Mass. 1991), aff'd, 36 F.3d 1147 (1st Cir. 1994); accord Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263, 281 (2d Cir. 1979) ("Withholding from others advance knowledge of one's new products, therefore, ordinarily constitutes valid competitive conduct."). Because Microsoft could lawfully decline to reveal advances in technology, it did not offend the antitrust laws by failing to distribute design information about Windows95 to Aldridge or by encrypting the computer code for the operating system.

Aldridge fairs no better on his argument that "Microsoft denied access to Windows95 by designing it to disable or disparage all competing disk caches." Aldridge offers no evidence to show that Windows95 disables Cache86, and nothing in the messages prevents a user from employing Cache86 in DOS-compatibility mode or from contacting Aldridge for an updated version of the program. Aldridge has thus failed to show that Microsoft denied Aldridge access to the Windows95 operating system.

6. Was it feasible for Microsoft to provide Aldridge access to Windows95?

Aldridge argues that Cache86 runs as designed in DOS-compatibility mode and that Cache86 could have been updated to function in protect mode as well. Microsoft does not contest these arguments and, indeed, Aldridge did redesign Cache86 to disable itself in protect mode and to operate as designed when Windows95 runs other programs in DOS-compatibility mode. The court therefore concludes that Aldridge has shown that it was feasible to provide access.

7. Conclusion

While Aldridge has shown feasibility, and the parties have not contested that Microsoft is a monopolist and that duplicating Windows95 would be unreasonable or impractical, Aldridge has failed to establish the remaining elements required for an essential facilities claim. Aldridge has not shown that Windows95 is essential, that Microsoft has the type of control over Windows95 that is forbidden by the Sherman Act, or that Microsoft has illegally denied Aldridge access to Windows95. Accordingly, Aldridge cannot recover under the Sherman Act on the theory that Microsoft denied him access to an essential facility.

IV. CONCLUSION

Because Aldridge has only raised issues of fact as to the truth of Messages Three and Four, Microsoft is entitled to summary judgment on Aldridge's state law claims as to Messages One and Two. Because Aldridge has failed to rebut the presumption of de minimis damage in his antitrust product disparagement claim and has failed to establish an antitrust claim under the essential facilities doctrine, Microsoft is also entitled to summary judgment on those claims.

Microsoft's Motion for Summary Judgment (Docket Entry No. 26) is GRANTED IN PART and DENIED IN PART. Aldridge's antitrust claims will be dismissed with prejudice. Aldridge's state law claims will be dismissed with prejudice as to Messages One and Two.