Opinion of Judge Wald.
Concurring in Part and Dissenting in Part.
USA v. Microsoft, USDC No. 94-1564, USCA No. 97-5343.
Date: June 23, 1998.
Source: U.S.C.A.
This document has been converted into an HTML document.  Several features were lost from the original, including pagination, fonts, and formatting. This document has not been edited for content.
See also, Majority Opinion.

Wald, Circuit Judge, concurring in part and dissenting in part: I depart from my colleagues only as to their interpretation of the consent decree, which I believe unnecessarily narrows the scope of the inquiry that the district court may conduct on remand. First, the majority opinion appears to decide that there is only one reasonable interpretation of section IV(E)(i), notwithstanding the fact that we are remand-ing for further factual development that may well be relevant to the most faithful interpretation of the section. Second, although the majority claims to have rooted its interpretation in antitrust law in accordance with the intent of the parties, it interprets section IV(E)(i) in a way that is, in fact, inconsistent with at least some governing precedent. For these reasons, I write separately to suggest that there may be an interpretation of section IV(E)(i) more consonant with the intent of the drafters and the weight of antitrust law. If facts are found on remand to support such an alternative interpretation, it should not be foreclosed by the majority opinion.

Under the majority's interpretation, the proviso of section IV(E)(i), which says that section IV(E)(i) "in and of itself shall not be construed to prohibit Microsoft from developing inte-grated products," is too safe a harbor with too easily naviga-ble an entrance: So long as Microsoft has created a design to combine functionalities in a way that offers the ultimate user some "plausible" advantage otherwise unavailable, Microsoft may require OEMs to install the resulting creation in its entirety, without fear of running aground on the main prohi-bition of section IV(E)(i) (which prohibits Microsoft from entering into any license agreement "in which the terms of that agreement are expressly or impliedly conditioned upon" the licensing of any "other product"). To my mind, this reading does not impose nearly enough scrutiny on "integration" and renders the central prohibition of section IV(E)(i) largely useless. Cf. Beal Mortgage, Inc. v. FDIC, 132 F.3d 85, 88 (D.C. Cir. 1998) (noting the "cardinal inter-pretive principle that we read a contract to give meaning to all of its provisions and to render them consistent with each other") (internal quotation omitted). The majority's interpretation would not, for example, appear to prevent Microsoft from requiring OEMs to license the right to sell computer peripherals (e.g., mice) as a condition of licensing Windows 95 so long as Microsoft had the prescience to include code in Windows 95 that made the cursor more responsive to the end- user's touch than it would be with other mice. I think the majority would agree, however, that under any reasonable reading of section IV(E)(i), the tying of mice to Windows 95 would clearly be prohibited.¹  I fail to see why the analysis should be so different for software than for peripherals or why our "evaluation of a claim of integration must be [any more] narrow and deferential," Maj. Op. at 25, for software than for other computer-related products. See, e.g., Fire-stone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 112 (1989) (courts construe contractual provisions "without deferring to either party's interpretation"). Our role, after all, is to interpret the consent decree as we would a contract: in a manner that is "grounded in the text of the agreement and contemporaneous understandings of its purposes, not in our own conception of wise policy." United States v. Western Elec. Co., Inc., 846 F.2d 1422, 1427 (D.C. Cir. 1988).

Another--some might say more--reasonable reading of section IV(E)(i) would give much greater weight to the main prohibition of the section. On its face, that prohibition for-bids Microsoft from requiring OEMs to accept under one agreement any offering that is in reality two products: a "Covered Product" and, for purposes of this case, an "other product." The proviso, on the other hand, permits Microsoft to develop and license "integrated" products.² Read together, I think the prohibition and the proviso could reasonably be construed to state that Microsoft may offer an "integrated" product to OEMs under one license only if the integrated product achieves synergies great enough to justify Microsoft's extension of its monopoly to an otherwise distinct market.

As I explain below, and as I read the majority opinion to agree, the consent decree was drafted against a backdrop of antitrust law. Under antitrust law, two products are considered distinct if there exists "sufficient consumer demand so that it is efficient for a firm to provide [the first product] separately from [the second]." Eastman Kodak Co. v. Image Technical Servs., Inc., 504 U.S. 451, 462 (1992). The difficul-ty in this case is that technological evolution can change the boundaries of what is "efficient." For example, Eastman Kodak cites cameras and film as examples of two functionally linked products for which there exist separate markets. See id. at 463. But antitrust law presumably would not bar the development of digital cameras, which do not require film in any conventional sense.

Thus, antitrust law cannot avoid determining whether a particular technological development has occurred because it is efficient or merely because it permits a monopolist to extend its monopoly to a new market. Software code is a particularly stark example of why such analysis is essential if antitrust concepts are to survive at all. Here, the majority effectively exempts software products from antitrust analysis by stating that "[s]oftware code by its nature is susceptible to division and combination in a way that physical products are not." Maj. Op. at 28. But this to me is an argument for closer, rather than more relaxed, scrutiny of Microsoft's claims of integration. An operating-system designer who wished to turn two products into one could easily commingle the code of two formerly separate products, arranging it so that "Windows 95 without IE's code will not boot," id. at 23 n.11, so that Windows 95 without Internet Explorer would "represent a disabled version of Windows 95," id. at 23, and so that Internet Explorer instructs the Add/Remove function to leave so much of that program in place that "four lines of programming" will suffice to activate it, see id. at 29 n.17. This is not to say that commingling of code is per se pernicious or even suspicious. Rather, the point is that comming-ling alone is not sufficient evidence of true integration; the courts must consider whether the resulting product confers benefits on the consumer that justify a product's bridging of two formerly separate markets.

Although this task is difficult, it is by no means the impossible project that the majority suggests. As I explain below, traditional antitrust analysis, and the usual methods of the law, provide the courts with a range of ways to address the ultimate question of efficiency. See, e.g., PSI Repair Servs., Inc. v. Honeywell, Inc., 104 F.3d 811, 817 (6th Cir. 1997), cert. denied, 117 S. Ct. 2434 (1997) (citation omitted) ("While we are mindful of the various efficiency gains that can accrue to both suppliers and consumers from bundling certain products together, we are confident that the antitrust laws provide the tools to distinguish between meritorious and non-meritorious claims."). As I see it, the efficiency calculus takes two factors into account. The first is evidence that there are real benefits to the consumer associated with integrating two software products; I call these benefits "syn-ergies." ³ The second is independent evidence, of the type that is usually employed in antitrust analysis, that a genuine market exists for the two products provided separately. For example, Windows 95 includes a built-in calculator program with relatively few functions. Although there may be few synergies associated with building this program into the operating system, there is also not likely to be much of a market for this program provided separately, and this factor must be taken into account. Market evidence of demand for independent products will spare the courts from the need to speculate in the abstract about considerations of efficiency.

Taken together, then, these two factors generate a balancing test. The greater the evidence of distinct markets, the more of a showing of synergy Microsoft must make in order to justify incorporating what would otherwise be an "other" product into an "integrated" whole. If the evidence of dis-tinct markets is weak, then Microsoft can get by with a fairly modest showing (although perhaps not the minimal showing required by the majority). But if there are clearly two distinct markets, then Microsoft would need to demonstrate substantial synergies in order to compel OEMs to accept a new "integrated" product that bridges those markets. In other words, the decree does not purport to chill Microsoft's technological development of its products by prohibiting a product outright merely because it incorporates new fea-tures--the proviso makes clear that Microsoft is free to design such products and to market their benefits to OEMs such that OEMs overwhelmingly choose the new product over a competitor's product.⁴  The proviso thus becomes a safe harbor only for those integrations in which the "other prod-uct" has been (legitimately) technologically subsumed in a greater whole.⁵

Thus, the real question for purposes of determining if a violation of the degree has taken place is whether a new combination of formerly separate functionalities still contains an "other product" or if the two functionalities have been legitimately blended, or "integrated," and so have lost their former identities and become one product to which the prohibition no longer applies. (This, of course, is a factual question to be explored on remand.) As I have already suggested, and as the majority agrees, given the context in which section IV(E)(i) arose, it is appropriate to look to antitrust law as a guide to determining when such integration occurs. Although the majority opinion claims that its construction is consistent with antitrust law, see Maj. Op. at 22, it does not, in my view, give due weight to the Supreme Court's holding in Jefferson Parish Hospital District No. 2 v. Hyde, 466 U.S. 2 (1984), the leading guide to the separate product determination.⁶ In Jefferson Parish, the Court considered whether anesthesiological services, which a hospital had required patients to take only from certain anesthesiologists, were in fact separate products from the other services provided by the hospital or, rather, were part of what the hospital claimed was a "functionally integrated package of services." Id. at 19.

In rejecting the hospital's argument that such a package did not involve a tying arrangement, the Court held that "the answer to the question whether one or two products are involved turns not on the functional relation between them, but rather on the character of the demand for the two items." Id. Thus, the fact that the hospital provided "the space, equipment, maintenance, and other supporting services neces-sary to operate the anesthesiology department," purchased the necessary drugs and supplies, and furnished the required nursing personnel, id. at 6, did not prevent a finding of separateness, nor did the district court's conclusion that the hospital believed that a "closed system" anesthesiology department "resulted in the best quality of patient care." Hyde v. Jefferson Parish Hosp. Dist. No. 2, 513 F. Supp. 532, 540 (E.D. La. 1981). In other words, despite the overlap between the services provided by the hospital and those provided by the anesthesiologist, and despite the claimed benefits of an "integrated" relationship, the Court held that the proper focus of analysis was whether the arrangement tied two distinct markets for products that are separate from the buyer's perspective. See Jefferson Parish, 466 U.S. at 19-20 (citing Times-Picayune Publishing Co. v. United States, 345 U.S. 594 (1953), and Fortner Enters. v. United States Steel Corp., 394 U.S. 495 (1969)). It is not clear to me why this analysis should be markedly less applicable in the technological context; indeed, the post-Jefferson Parish trend is to apply its test even in the technological realm. See, e.g., Allen-Myland, Inc. v. International Bus. Machs. Corp., 33 F.3d 194, 211-12 (3d Cir. 1994) (computer parts and installa- tion); Service & Training, Inc. v. Data Gen. Corp., 963 F.2d 680, 684 (4th Cir. 1992) (diagnostic software and maintenance/repair service); Digidyne Corp. v. Data Gen. Corp., 734 F.2d 1336, 1339 (9th Cir. 1984) (central processing units and operating system); cf. Jefferson Parish, 466 U.S. at 25 n.42 ("In the past, we have refused to tolerate manifestly anticompetitive conduct simply because the health care industry is involved.").

The tying analysis is, of course, a pragmatic one. For example, no one would claim that tying law was violated by the practice of selling shoes in pairs despite the possible existence of some market for only left shoes (among those with only one foot, for example, or with differently sized feet). Likewise, it is in all likelihood not a tying violation for Jefferson Parish's hospital to require that patients accept the hospital's receptionists (instead of bringing their own) and accept the cleaning services and meals provided in their rooms (instead of making other arrangements) and to charge patients for these services. This is so even though there might be a limited group of patients who would prefer to make their own arrangements for receptionists, cleaning, and meals. Cf. Jefferson Parish., 466 U.S. at 22 n.36 (noting that the antitrust analysis might differ for "radiologists, patholo-gists, and other types of hospital-based physicians"); see also Jack Walters & Sons Corp. v. Morton Bldg., Inc., 737 F.2d 698, 703 (7th Cir. 1984) (noting that "[t]he practice has been to classify a product as a single product if there are rather obvious economies of joint provision"). In the case of the shoes, the receptionist, and the cleaning, a judgment is made that the benefits of joint provision clearly predominate over what is undoubtedly a minimal separate market (if one can be said to exist at all). In the case of the anesthesiologist, by contrast, the Court found that the claimed benefits--24-hour anesthesiology coverage, flexible scheduling, and facilitation of work routine, professional standards, and equipment maintenance--were not sufficient to justify joint provision because there was a very substantial market for anesthesiologists' services and because these benefits could be achieved without the forced tie (by, for example, promoting the benefits of the hospital's anesthesiologists to patients and setting standards of compatibility). See Jefferson Parish, 466 U.S. at 25 n.42. Under this doctrine, then, an "integrated" product cannot simply be one where some benefit exists as a result of joint provision, since the hospital easily met this standard. Rather, "integration" must mean something more: a combination of functionalities in which the synergies created predominate over the existence of a separate market--in other words, where the benefits of the combination dissuade consumers from seeking and suppliers from providing the alleged "tied" product.7

This alternative interpretation of section IV(E)(i) also ac-commodates the majority's "paradigm" case. The mere provision of Windows 3.11 and DOS 6.22 as a single product may have created some benefits--reduction of transaction costs, guaranteed compatibility, and a single source of customer support services, for example. These benefits were not, however, enough to overwhelm the existence of a separate operating system market--indeed, the very presence of Novell's product as an operating system to be used with Windows 3.11 suggests the existence of such a market. See, e.g., J.A. 845 (DG IV Statement of Objections) ("Such tying affects the competitive freedom of the licensee to find a better substitute or obtain better terms for the operating system to be used with Windows."). Windows 95, however, is a different matter. In that case, the whole is clearly greater than the sum of its parts--as the majority notes, "it is not simply a graphical user interface running on top of MS-DOS." Maj. Op. at 23. And it is clear on the present record that Microsoft, at least, understood Windows 95 to provide such benefits. See, e.g. J.A. 1101 (Microsoft's "Windows 95 Feature Review") ("When you first boot Windows 95 it is immediately apparent that the old world of Windows running on top of MS-DOS is no more."). Thus, the synergistic benefits appear to have been large enough so that it would have been reasonable for DOJ to agree to treat Windows 95 as an integrated product.⁸

The majority's interpretation, however, departs from this precedent by accepting any "plausible claim" that the combination (i.e., the design) offers "some advantage." Maj. Op. at 26-27. Of course, both the majority's interpretation and my alternative proposal would discredit any specious claims of integration--Microsoft's claim that it could simply put two disks in the same box and claim integration, for example, see id. at 21, would hardly merit a second thought. But the majority's considerable deference to Microsoft's plausible claims of advantage, coupled with Microsoft's privileged knowledge of the inner workings of its operating system, barely raises the bar of section IV(E)(i) above ground level. It is difficult to imagine how Microsoft could not conjure up some technological advantage for any currently separate soft-ware product it wished to "integrate" into the operating system.⁹ And for the majority, the chase ends there: Internet Explorer, it contends, shares code with the operating system in a way that other browsers do not, and therefore it is integrated. But the fact that parts of Internet Explorer share code with the operating system and thus with other applications should not end the analysis any more than did the fact that the anesthesiologists in Jefferson Parish shared hospital equipment and personnel with the hospital and its staff (or that the hospital could identify some minor practical benefits to requiring the use of only certain anesthesiologists). The analysis must also consider whether Internet Explorer is a separate product under antitrust law, that is, whether "consumers differentiate between [Internet Explorer] and [Windows 95]" such that consumers desire to pur-chase--and hence that manufacturers desire to supply--a substitute for Internet Explorer from another manufacturer; in other words, whether there is "a distinct product market in which it is efficient to offer [the tied product] separately from [the tying product]." Jefferson Parish, 466 U.S. at 22; see also Digidyne, 734 F.2d at 1339 ("[t]he undisputed facts summarized in the district court's opinion establish that a demand existed for NOVA instruction set CPUs [central processing units] separate from defendant's RDOS [operating system], and that each element of the NOVA computer system could have been provided separately and selected separately by customers if defendant had not compelled purchasers to take both"). Whether such a market exists, and whether it is significant enough to outweigh the particular synergies associated with integrating IE 3.0 and/or IE 4.0 into Windows 95, is, of course, a factual determination within the province of the district court. Relevant indicators in the market analysis, however, would surely include (1) whether manufacturers of other operating systems require OEMs to include a particular browser, see, e.g., Jefferson Parish, 466 U.S. at 23 n.39 (noting that "other hospitals often permit anesthesiological services to be purchased separately"); X Areeda, Antitrust Law p 1746, at 225 (1996) (suggesting comparison of alleged tie with practices in analogous competi-tive markets); (2) whether Microsoft's own actions reflect a perception of a competitive market for "Internet Explorer" separate from the market for Windows 95, see, e.g., Allen-Myland, 33 F.3d at 208-09 (noting probative value of internal reports in determining distinct product markets); and even (3) the very existence of competitor browser manufacturers, see, e.g., Eastman Kodak, 504 U.S. at 462 (noting that "the development of the entire high-technology service industry is evidence of the efficiency of a separate market for service"). The majority opinion, however, relies on none of these consid-erations. By discounting the relevance of such analysis, the majority in fact shorts traditional antitrust law.

Despite the plausibility of this alternative interpretation, the extent to which antitrust law was intended to inform the interpretation of the decree--specifically the interpretation of the terms "integrated product" and "other product"--is another question best left open to the district court on remand. Everyone agrees, I believe, that section IV(E)(i) of the decree is ambiguous--indeed, had it been unambiguous, the district court could have fished or cut bait, i.e., it could have determined either that Microsoft's actions constituted contempt or that they did not violate the decree at all. Thus, I do not read the majority opinion to say at any point that there is a plain meaning of section IV(E)(i) that can be located in the text alone. To the contrary, we are relegated to ordinary principles of contract law: searching for the parties' intent and guided in that adventure by "conventional 'aids to construction,' including the 'circumstances surrounding the for-mation of the consent order, any technical meaning words used may have had to the parties, and any other documents expressly incorporated in the decree.' " United States v. Western Elec. Co., 894 F.2d 430, 434 (D.C. Cir. 1990) (quoting United States v. ITT Continental Baking Co., 420 U.S. 223, 238 (1975)). Under these principles, the interpretation of an ambiguous contract involves factual findings as to the parties' intent. See, e.g., Bennett Enters., Inc. v. Domino's Pizza, Inc., 45 F.3d 493, 497 (D.C. Cir. 1995); Carey Canada, Inc. v. Columbia Cas. Co., 940 F.2d 1548, 1553-54 (D.C. Cir. 1991). In non-plain meaning cases such as this one, appellate courts generally decline to embark on their own factfinding mission, deferring to the district court's role as the primary factfinder and reviewing its findings only for clear error. See, e.g., United States v. Insurance Co. of N. Am., 131 F.3d 1037, 1042-43 (D.C. Cir. 1997) (declining to determine intent of parties to contract on the basis of incomplete record and  remanding for further findings). While the district court below did conclude that "[r]eading [section] IV(E)(i) in light of its avowed purpose raises a logical inference that the parties anticipated the use of [tying law] antitrust precedents in determining the application of [section IV(E)(i)] to the conduct the government challenges here," United States v. Microsoft Corp., 980 F. Supp. 537, 542 (D.D.C. 1997)--a conclusion that suggests that Jefferson Parish, decided well before the parties' agreement, is indeed relevant to the construction of the decree--it is not inconceivable that, given notice and an adequate opportunity to present evidence and arguments, the parties will succeed in persuading the district court otherwise.¹⁰ The majority opinion, however, leaves little room for such efforts--its interpretation seems clearly meant to be the last word.¹¹

Finally, and relatedly, I note that when the parties' contem-poraneous understanding of section IV(E)(i) is ultimately revealed, there is further work to be done on how the parties intended that understanding to be applied. Should the district court conclude that whether something is an "other product" depends on the existence of a separate market for that product, it must still determine how that "something" is defined--in this case, "Internet Explorer." Here, again, the majority relies heavily on a presumption that "Internet Explorer" contains code that upgrades the operating system as well as code that enables end-users to access the Internet and therefore concludes that there is no "separate" product for any tie-in analysis. See Maj. Op. at 29.¹² The validity of that perception is not so evident to me. The fact that the supplies and equipment that made the anesthesiologist's job possible in Jefferson Parish remained at the hospital--that, in a manner of speaking, the anesthesiologist was something of an interface between the end-user patient and the hospital's "operating system"--did not prevent the Court from concluding that his portion of the service constituted a separate product. Unfortunately, perhaps due to the irregular nature of the preliminary injunction proceedings, the district court made little in the way of findings concerning what the linch-pin product "Internet Explorer" is intended to encompass.¹³ It would not be unreasonable, as the prior discussion suggests, to approach the problem from the perspective of a typical end-user, who most likely regards "Internet Explorer" as a particular vehicle for accessing information on the Internet, regardless of the underlying code associated with that process. The fact that, as the majority suggests, see Maj. Op. at 29-30, "Internet Explorer" distributes certain code to the operating system may simply suggest that some or all of this code should not be considered part of "Internet Explorer" at all but part of the operating system. Or perhaps the work that this code does could be done equally well by similar code written by Netscape or some other competitor, in which case the code's function is not necessarily an operating system function at all.¹⁴ Which approach to the meaning of "product" was the contemporaneous understanding of the parties to the decree, however, remains to be determined.

Furthermore, the record suggests that many of the benefits that the majority asserts for the incorporation of Internet Explorer into Windows 95, including customizing of "Start" menus and "thumbnail" previews, see id. at 27-28, are provided only by Internet Explorer 4.0 and not by Internet Explor-er 3.0. See, e.g., J.A. 490-95, 1664-69. The majority seems to conclude that IE 3.0 and Windows 95 are "integrated" on the basis of little or no evidence. Should more evidence on this point come to light, the district court thus cannot be bound by the majority's conclusions. As to IE 4.0, the district court's preliminary conclusions seemed to hinge on the fact that Microsoft had offered the program only on a separate disk and not on the technology involved, see Microsoft, 980 F. Supp. at 544, thus suggesting that more factfind-ing also needs to be done as to IE 4.0.

Because I believe there is significant further factfinding as to intent and operation to be accomplished by the district court on remand, I would not rule out its authority to reissue a preliminary injunction.15 That is why I am troubled by the seemingly authoritative nature of the interpretation of section IV(E)(i) in the majority opinion, which would appear to foreclose any other interpretation of the section and proviso that might evolve in further proceedings and justify either a preliminary or a permanent injunction. To that extent, I respectfully dissent from the majority opinion.