IN THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLUMBIA
Direct Testimony of Franklin M. Fisher
TABLE OF CONTENTS
Appendix A. Curriculum Vitae
Appendix B. Declarations and C.I.D. Depositions
DIRECT TESTIMONY OF FRANKLIN M. FISHER
1. I am professor of economics at the Massachusetts Institute of Technology, where I have taught for more than 38 years. I am also a director of the National Bureau of Economic Research and the chair of the Middle East Water Project - a project supported by the government of the Netherlands that involves American, Dutch, Israeli, Jordanian, and Palestinian participation. I received my A.B. from Harvard University in 1956, and my Ph.D. in Economics from Harvard University in 1960.
2. I am a fellow and past President of the Econometric Society and for nine years was the editor of that society's journal, Econometrica. I am a member of the American Economic Association, from which I received the John Bates Clark Award; a fellow of the American Academy of Arts and Sciences; and a recipient of a John Guggenheim Fellowship.
3. My fields of specialization within economics are industrial organization, microeconomics, and econometrics. I am the author of 15 books and well over 100 articles. In the course of my scholarly research and my consulting work, I have studied issues of competition and monopoly in a large number of industries. I have written extensively in the area of antitrust economics. I have provided expert consultation and testimony in numerous antitrust cases and have testified at trial, in deposition, or by affidavit in more than 40 cases in the last 10 years. These cases are listed in my curriculum vitae, which is attached as Appendix A to this testimony.
4. I was for many years IBM's chief economic witness in US v. IBM (dismissed, 1982) and associated cases. This work is discussed in two of my books.
5. I have been asked to provide my expert economic opinion about the competitive effects of a collection of Microsoft's actions centered around its Internet browser. As part of my analysis, I have considered transcripts and exhibits from the depositions of Microsoft employees, of employees of other companies in the computer industry, and of others; Microsoft agreements with third parties; internal documents from Microsoft and from others; published reports and data; and other related documents and information. I have also read direct testimony and the trial transcript through December, 1998.
6. New and additional evidence and data have recently been made available, and I understand that more evidence and data may become available. To the extent that new data and evidence are relevant to my opinions, I will, to the extent possible, include them in my analysis.
7. Appendix B contains pages from declarations and C.I.D. depositions I have cited in my report.
8. I have been asked to consider four basic economic questions. First, does Microsoft possess monopoly power in the market or markets for personal computer operating systems? Second, has Microsoft maintained its monopoly power by anti-competitive conduct? Third, has Microsoft used its monopoly power in an anti-competitive way to distort competition or achieve monopoly power in markets other than the market or markets for personal computer operating systems? Fourth, has Microsoft engaged in unreasonable restraints of trade?
9. I note at the outset that the answer to the first question is significant in answering each of the remaining three questions. As discussed below, there are several reasons this is so. One important reason is that the effect of certain conduct by a firm depends on whether the enterprise engaging in it has monopoly power. For example, if an enterprise without monopoly power engages in tying (or other restrictive practices), customers who would prefer not to purchase the tied combination can decline to buy from the company with the tying enterprise. If there is effective competition in the market for the tying product, and if there is a separate demand for a component of a tied combination sufficient to make it efficient to supply that component separately, I would expect competitors to offer both the tying and the tied components of the combination separately. However, if an enterprise that is engaged in tying possesses monopoly power over the tying product, customers will not have realistic competitive alternatives and will be unable practicably to procure the tying product separate from the combination. Moreover, if the enterprise engaged in tying both has monopoly power over the tying product and ties a sufficient quantity of the tied product so that there is no longer sufficient demand to support viable alternative suppliers of the tied product, the enterprise engaged in tying will achieve a dominant position in the supply of the tied product as well. (If there are significant barriers to entry in the market for the tied product, that dominant position will result in the enterprise's achieving monopoly power over the tied product also.)
10. Similarly, analyzing the effect of conduct on either maintaining existing monopoly power or on securing monopoly power where such power does not exist may be useful in assessing whether the conduct is anti-competitive or exclusionary. This is not, of course, to say that all conduct which secures or maintains monopoly power is anti-competitive. Certain conduct (e.g., non-predatory price competition or product improvements) is so important to the competitive process, and the potential costs of interference are so high, that it is considered competitive (and not anti-competitive) even where it results in securing or maintaining a monopoly. However, certain conduct that may be benign (or at least tolerated) if it does not maintain or create monopoly power, is clearly recognized to be anti-competitive where it has such an effect.
11. For example, it is sometimes asserted that tying even by a monopolistic entity will not have an economically inefficient effect, since the monopolist only has a given quantum of monopoly power and it is irrelevant whether that power is exercised directly by charging a profit-maximizing price for the monopoly product or indirectly by requiring customers to take tied products that they would not otherwise have acquired from the monopolist. Without here specifying all of the reasons that general assertion is wrong, it is apparent that the assertion does not apply where the act of tying maintains an existing monopoly or secures a new monopoly (or to put it more generally, where the act of tying increases the monopoly power the enterprise would otherwise have possessed as compared with merely exploiting an unchanging, unthreatened quantum of monopoly power). (There may well be significant economic and social costs of permitting a monopolist to exploit a given quantum of monopoly power even if that exploitation neither protects an existing monopoly nor is likely to result in a new monopoly. For the reasons indicated above and discussed in more detail below, it is not necessary to address those costs in the present case.)
12. As another example, there is an arguable case for prohibiting enterprises from engaging in predatory pricing or predatory product design even if the enterprise does not have monopoly power and has no reasonable prospect of obtaining it. However, the limited (and, perhaps, speculative) cost of permitting predatory behavior under those circumstances coupled with the costs of possible misdiagnosis lead economists generally to favor non-intervention in cases of predatory pricing or product design in the absence of monopoly power, or the reasonable prospect of monopoly power.1 By contrast, where predatory pricing or product design can be reasonably expected to maintain existing monopoly power, or to obtain monopoly power in a new market, the economic costs of non-intervention are too high to tolerate.
13. In that context, I make one more general observation before turning to the answers to the four specific questions I have been asked to address. Because of the central and essential role the personal computer (PC) operating system plays (and is expected to play) in both commercial and consumer endeavors (including access to the Internet and the World Wide Web), the costs of improperly maintaining monopoly power over the operating system, and the danger that Microsoft's existing monopoly power will be used to monopolize other critical markets that are linked to the operating system, are very great.
14. For example, if Internet browsers and/or Java in fact threaten eventually to undermine Microsoft's operating system monopoly by eroding the applications programming barrier to entry that protects that monopoly, the economic cost of permitting Microsoft to rebuild that barrier to entry by stifling non-Microsoft browsers and Java will be substantial.
15. Similarly, if Microsoft is permitted to use its existing monopoly power over PC operating systems to monopolize the market for Internet browsers (and, thereby, to put itself in the position to control and ultimately exact a monopoly toll for efficient access to the Internet), the economic costs to consumers and the economy will again be substantial.
16. I now turn to a summary of my basic conclusions.
17. Microsoft has achieved monopoly power in the market for operating systems for Intel-compatible desktop personal computers.
18. Microsoft has foreseen the possibility that the dominant position of its Windows operating system will be eroded by Internet browsers and by Java, which are capable of supporting software applications that are operating-system independent.
19. Microsoft has taken anti-competitive actions to exclude competition in Internet browsers in order to protect the current dominance of its Windows operating system. Microsoft's conduct includes: anti-competitive agreements with PC manufacturers that require the manufacturers to acquire Microsoft's Internet browser as a condition of acquiring Microsoft's Windows operating system, that hinder the manufacturers' promotion of competing browsers, and that restrain PC manufacturers from removing Microsoft's browser or substituting an alternative browser; the predatory pricing and distribution of Microsoft's browser; and anti-competitive and exclusionary agreements with Online Services (OLSs), Internet Service Providers (ISPs), Internet Content Providers (ICPs), and others. Taken together, Microsoft's actions as to browsers are not profit-maximizing in themselves but are profitable only because of their adverse effects on competition.
20. Microsoft has also taken anti-competitive actions to restrain the use and availability of the Java technology in order to protect the current dominance of the Windows operating system. Further, Microsoft has engaged in a number of anti-competitive acts and solicitations designed to convince other firms not to compete against Microsoft in platform-level software.
21. Microsoft is using its monopoly power over PC operating systems to secure monopoly power over Internet browsers.
22. As we shall see in more detail, Microsoft's conduct has created, preserved, and increased barriers to entry into both the PC operating system market and the Internet browser market. Microsoft's conduct to create, preserve, and increase barriers to entry includes:
23. The principal effect of Microsoft's anti-competitive conduct is the maintenance of Microsoft's operating systems monopoly. If Microsoft's conduct is not checked, it is very likely to create a world in which entry into browsers is difficult or impossible (partially as a result of network effects). In that world, platforms that do not use a Microsoft standard will never prosper, and a critical opportunity for innovation that reduces or eliminates Microsoft's power will have been lost.
24. Further, if Microsoft is successful in its anti-competitive actions, that success will serve as a disincentive to other firms to innovate in areas that Microsoft may stake out as its own property. If software developers believe that Microsoft will engage in anti-competitive acts to impede any innovation that threatens its monopoly, they will have substantially reduced incentives to innovate in competition with Microsoft. As a result, the range of software products from which consumers can choose will be limited, ultimately further reducing consumer welfare.
25. It is important at the outset to distinguish the features of competition from those of monopoly. Under competition, any attempt by a firm to restrict output (or reduce quality) and to charge supra-normal prices will be defeated, as other firms offer lower prices or higher quality to attract customers. The business of the firms that do not offer lower prices or superior products will be bid away, and customers will benefit. In monopoly, by contrast, such beneficial effects will not occur, or will occur only to a significantly more limited extent. A monopolist can charge supra-competitive prices because the constraints imposed by rivals are loose and ineffective.
26. Economists use the term "market power" to refer to a departure from conditions of competition. Market power to a seller is the ability profitably to maintain prices above competitive levels.
27. Market power is a matter of degree. The possession of a small degree of market power is common in the real world. For example, a corner grocery store may possess a small degree of market power. When present only to a small degree, market power should not be of antitrust concern.
28. Monopoly power is a substantial degree of market power. While a firm with a slight degree of market power may find it profitable to charge supra-normal prices for a short time or to charge prices that are only slightly supra-normal, a firm with monopoly power will find it profitable (a) to charge a price significantly in excess of competitive levels and (b) to do so over a significant period of time. I believe this definition is consistent with the oft-cited legal definition of monopoly power as "the power to control prices or exclude competition."
29. Success achieved through legitimate means such as innovation, superior marketing, or historical accident may naturally give rise to market power or even monopoly power. Whether or not this is true of software (and, in particular, of PC operating systems), the very fact that the software industry is so innovative, together with its immense and growing importance in the American economy, makes it crucial that success be restricted to success on the merits, and that monopoly power be confined to that which results from that success. Even a firm that has attained monopoly power through legitimate means and natural economic effects must not be permitted to retain or extend that power through artificial, anti-competitive means. (Because of my conclusion that Microsoft has improperly maintained its PC operating system monopoly and attempted to monopolize the Internet browser market, it has not been necessary for me to examine the allegations of anti-competitive actions by Microsoft that underlie the earlier monopolization case against Microsoft.)
30. An analysis of competition and monopoly as applied to this case requires discussion of the following points:
31. After discussing these points, I proceed to apply them to an analysis of Microsoft's situation and behavior.
32. As the preceding discussion suggests, the hallmark of monopoly power is the absence or ineffectiveness of competitive constraints on price, output, product decisions, and quality. The issue of monopoly power is conventionally addressed by defining "the relevant market" and assessing shares in that market. A large share of a properly defined market is an indication of a firm's ability successfully to raise price. Because its purpose is the identification of monopoly power, if it exists, the "relevant market" should include all those products that reasonably serve to constrain the behavior of the alleged monopolist. Such constraints arise from three sources: substitution by consumers to other products (demand substitutability), substitution by producers to other products (supply substitutability), and entry of new productive capacity. As market power is a matter of degree, so too is the extent to which each of these factors imposes a constraint.
33. Demand substitutability refers to the ease with which customers, faced with an attempt by the alleged monopolist to exercise power and earn supra-normal profits, could turn to products other than those of the alleged monopolist.
34. Supply substitutability refers to the ease with which firms that do not currently supply demand-substitutable products could do so by reallocating their existing productive capacity in the event of an attempt by the alleged monopolist to exercise power and earn supra-normal profits.
35. An example may help to clarify what is involved in the consideration of supply substitutability. The bottoms of boats used in seawater need to be painted with a copper-based, anti-fouling paint (to protect against barnacles). Suppose that there were only a single manufacturer of such paint but that it were very easy for other paint manufacturers to produce and market anti-fouling paint. Clearly, the single current producer of anti-fouling paint does not possess monopoly power. One could come to this conclusion even if one found that this firm had a market share of a hundred percent, observing that the hundred percent share of the alleged monopolist was meaningless because of high supply substitutability. But such a share also could lead to an erroneous inference of monopoly power from market share.
36. Market shares are useful indicators of market power only if they reflect supply substitutability as well as demand substitutability. This can be accomplished in the anti-fouling paint example either by defining "the relevant market" as all paint, or by defining "the relevant market" as just anti-fouling paint, but assigning shares in that market to paint manufacturers that do not currently sell such paint based on their capacity to sell it. Obviously, there will often be no bright line between defining products as in the market on the basis of supply substitutability and leaving them out while remembering that the firms that do not produce them can enter fairly readily. But the lack of such a clear line will not matter, so long as one remembers that market definition need not be precise and that its purpose is to assist in analyzing the constraints on the behavior of the alleged monopolist.
37. These principles have long been recognized. Since 1982, the Department of Justice's Merger Guidelines have approached market definition by asking in part whether a single, profit-maximizing firm controlling a candidate market could raise price by a significant amount (i.e., 5 percent), for a non-negligible time period. ( U.S. Department of Justice, Merger Guidelines § I (1982), reprinted in 4 Trade Reg. Rep. (CCH) ¶13,102; U.S. Department of Justice and Federal Trade Commission, Horizontal Merger Guidelines § 1.0 (1992), reprinted in 4 Trade Reg. Rep. (CCH) ¶13,104. The Merger Guidelines generally consider raising price from the prevailing level, because the issue presented by a merger case typically is whether the proposed merger would create or enhance market power. When the issue is instead whether a particular firm possesses market power or monopoly power, it is necessary to consider raising price from the competitive level.)
38. Having defined an appropriate market, one then goes on to consider market share and the ability of firms not in the market to enter, in the event of an attempt by the alleged monopolist to earn supra-normal profits through an exercise of power. As noted above, a key distinguishing feature of monopoly power is its durability. If the attempt by a firm to earn supra-normal profits by pricing above competitive levels would be rapidly frustrated by entry, that firm does not possess monopoly power.
39. Economists generally refer to factors that would prevent entry in the face of supra-normal profits as "barriers to entry." (Such factors also limit expansion of existing firms. I use the term "barriers to entry" to refer both to limits on entry of new firms and to limits on the expansion of existing firms.) Where there are significant barriers to entry, monopoly power can be present. Where there are no barriers or barriers are low, monopoly power cannot exist.
40. The barriers to entry in the present case include two phenomena known respectively as economies of scale and network effects.
41. An economy of scale is a phenomenon in which the average cost of production falls as more units of a product are produced. The software business exhibits substantial economies of scale, because most of the costs of software production come in the creation of the software and are independent of the number of copies that are produced.
42. A network effect is a phenomenon in which the attractiveness of a product to customers increases with the use of that product by others. This means, for example, that the fact that many applications are written for a given operating system and cannot easily run on the other operating systems will make that operating system more attractive to users a network effect. (This may be reinforced by other network effects related to the operating system and applications the desirability of using a common word-processing system, for example.)
43. Where network effects are present, a firm that gains a large share of the market, whether through innovation, marketing skill, historical accident, or any other means, may thereby gain monopoly power. This is because it will prove increasingly difficult for other firms to persuade customers to buy their products in the presence of a product that is widely used. The firm with a large share may then be able to charge high prices or slow down innovation without having its business bid away.
44. Such a circumstance can occur "naturally" in the sense that if network effects are sufficiently strong, a firm may acquire monopoly power without engaging in anti-competitive conduct. However, the fact that the successful firm has acquired monopoly power with a "natural" barrier to entry does not justify its taking anti-competitive acts to extend that power to another market or, in particular, its engaging in anti-competitive acts that serve to buttress and protect its power in the original market.
45. In the absence of anti-competitive conduct, market forces and developments can erode monopoly power based solely on network effects. This is in fact why Microsoft felt it necessary to take the actions discussed below with respect to Java and internet browsers.
46. Clearly, it is important to understand what an "anti-competitive act" is in the context of this case. In the case of a single firm, anti-competitive acts typically involve the taking of measures that are more restrictive of competition than necessary. When this involves actions that would not be profitable without their effect on competition, those actions can be called "predatory." (What is under discussion here are anti-competitive acts by a single firm. I am not discussing agreements, such as the division of markets, between two firms.)
47. This definition of a predatory anti-competitive act can be spelled out in two parts. The first part is deceptively simple: A predatory anti-competitive act is an act that is not expected to be profitable in the long run without accounting for the supra- normal profits that can be earned because of the adverse effects on competition.
48. The second part of the definition is as follows. A predatory anti-competitive act is one that is expected to be profitable in the long run only when taking into account the supra-normal profits to be earned because of the adverse effects on competition.
49. Some discussion is now in order. First, what matters is what is expected (or can reasonably be expected) at the time the action in question is taken. In an uncertain world, firms will often take actions that turn out to be unprofitable. Hence, the second part of the definition is useful to distinguish a predatory anti-competitive act from one that merely turns out to lose money ex post.
50. Second, in effect, a predatory anti-competitive act is one that involves a deliberate sacrifice of profit in order to secure or protect monopoly power. To understand the nature of this sacrifice, it is important to understand the concept of "opportunity cost."
51. Opportunity cost is the cost of foregoing an opportunity. Suppose, for example, that a certain manufacturing firm also owns the land on which its plant is situated. Suppose further that the land involved is favorably situated and would, if sold or rented, command a high price. The manufacturing firm that chooses to use the land itself does not, in an economic sense, obtain it for free. In so choosing, it incurs an opportunity cost equal to the amount it could have gotten by leasing the land to others. In considering its manufacturing business, such a firm will make the correct profit calculation only if it treats that opportunity cost as a real cost to its manufacturing business.
52. A second example is even more common. Firms that own their own plant and equipment are not profitable in terms of economics unless they earn an amount sufficient to return to their investors what their capital could earn in alternative uses. For example, a firm with $100 million in invested capital that earns $100 over current operating costs is, nevertheless, an economically unprofitable firm. Its invested capital could earn far more than that if merely invested in government securities. The correct calculation of economic profits requires treating as a cost what the invested capital could earn on the outside (at comparable risk). This is an opportunity cost. It is universally accepted in economic analysis to be a true cost. The firm in this example is incurring economic losses. If those losses continue, the firm will wish to leave the business, since the firm is not profitable.
53. To return to the analysis of predatory anti-competitive acts, a firm that takes an action not expected to be profit-maximizing, save for its effects on competition, is sacrificing a profit regardless of how the results are reported on its books. Such a firm is using its assets in a way that incurs an opportunity cost a sacrifice of profits that could have been made had the firm taken a profit-maximizing action. If the firm does this in order to earn later supra-normal profits dependent on the effects of its actions on competition, then that firm has taken an action that is not profitable except for those effects.
54. The standard tests of predatory anti-competitive acts in the literature are not at odds with the definition that I have given. Often, they are attempts to make that definition operational in the form of relatively simple tests. The well-known Areeda-Turner article on predatory pricing, for example, takes essentially the same view as do I.2 It ends up with a simple test for a single-product firm in terms of price and average variable cost, but that test is an important application of the general principle. A predatory anti-competitive act is one that is deliberately not profit-maximizing, save for the supra-normal profits to be earned because of the effects on competition. With opportunity costs properly understood, this is the same thing as an act that is deliberately unprofitable, save for the supra-normal profits to be earned because of the effects on competition.
55. A firm with monopoly power may choose to exercise that power in ways other than by immediately charging monopoly prices. In particular, such a firm, even while earning supra-normal profits in a given market, may choose to exercise its power to gain an advantage or even a monopoly in a second market. It is important to understand the circumstances under which such an action can have anti-competitive consequences.
56. Suppose that product A is used to make product B and begin by supposing that such use involves fixed proportions; that is, a fixed amount of A must be used to make each unit of B. In such a case it might appear that the A monopolist has nothing to gain by extending its monopoly into B. But such an appearance would be illusory because there are circumstances in which the extension of the A monopoly into the market for B could enable the original A monopolist to garner additional supra-competitive profits. For example, if A has other uses than the production of B, then the charging of a high price for A risks losing customers for those other uses. The monopolist may be able to do better by charging a relatively low, uniform price for A and ensuring that it is the only source for B. (This is an example of the classic "metering" case in which the monopolist of A and B price discriminates, charging a low price for A to keep the business of highly price-responsive consumers of A, and extracts additional profit from less price-responsive users of A and B together by charging a high price for B.)
57. Suppose, alternatively, that while A is used to make B, it is not used in fixed proportions, so that the amount of A per unit of B can be varied. In such a case, the A monopolist may very well be able to do better by becoming a B monopolist than by simply charging a high price for A. In effect, a monopolist may be able to use its power in A to organize the market in B, creating a new monopoly there and obtaining additional monopoly profits.
58. Another possibility is that by extending its monopoly into the market for B, the original monopolist may be in a position to earn supra-competitive profits in related markets in which B is used as an input.
59. Beyond all this, there is another very important possibility. If extending a monopolist's power from A to B can prevent others from entering A, then such an action will serve to maintain the A monopolist's original power. This could happen as follows. Suppose firms that produce B learn technology or know-how or gain access to customers that assist them in producing (or selling) A. Or suppose that B could be used by potential competitors of A to overcome barriers to entry in the supply of A. In such cases, and others, gaining control of the B business can create a barrier to entry in the A market.
60. Indeed, such an example is directly on point for the present case. Suppose that firms can somehow use B to facilitate the creation of a substitute for A. Then the A monopolist will gain from keeping or driving firms out of the production of B. If it does so through acts that are not profitable save for the preservation of the A monopoly, then those acts are predatory and anti-competitive.
61. In general, extending one monopoly to encompass a second market can be anti-competitive under either of two circumstances. The first such circumstance is one in which a firm uses its power in one market to organize a second market, permitting the earning of supra-normal profits that would not otherwise be available. The second circumstance is one in which the extension of power to a second market serves to protect monopoly power in the first market by inhibiting entry. As I shall explain, both circumstances are present here.
62. Microsoft possesses monopoly power in the market for operating systems for Intel-compatible desktop personal computers. There are no reasonable substitutes for Microsoft's Windows operating systems for Intel-compatible desktop PCs. Operating systems for non-Intel-based computers are not a reasonable substitute for Microsoft's Windows operating system.
63. As numerous representatives from personal computer OEMs have testified, OEMs do not believe they have any alternative to the acquisition and installation of Microsoft's Windows operating system. For example:
64. Microsoft's share of personal computer operating systems is very high and has remained stable over time. Microsoft's worldwide share of shipments of Intel-based operating systems has been approximately 90 percent or more in recent years. (See, e.g., Pl. Ex. 1.) Even if operating systems for non-Intel-based computers are included in the market definition, Microsoft's share is still very high and stable.
65. Microsoft's high market share is an indication that it possesses monopoly power. The analysis of barriers to entry confirms that monopoly power exists.
66. Operating systems are characterized by network effects, and all software is characterized by economies of scale. Users want the operating system that will permit them to run all the applications programs they want to use; developers tend to write applications for the most popular operating system; and applications software written for a specific operating system cannot run on a different operating system without extensive and costly modifications or add-ons. (Operating systems provide application programming interfaces (APIs) through which applications interact with the operating system and through the operating system with the computer hardware. Applications developers must write their programs to interact with a particular operating system's APIs. The time and expense of then "porting" the application to a different operating system can be substantial. An API set to which applications may be written is often referred to in the industry as a "platform.")
67. There are other network effects as well. For example, the presence of a common interface may enable firms to avoid training costs when personnel are moved within the firm or new personnel are hired from outside. This gives firms an incentive to have the same user interface throughout its own computers and the same interface that is widely used by other firms. Other network effects include the ease of exchanging files and the opportunity to learn from others. According to Jon Kies of Packard Bell NEC, his customer base "is very concerned about training and standardization. They prefer to standardize on a single interface and not have to retrain any of their end users to learn a new interface .... they are very loath to change anything in the interface." (Jon Kies 9/11/98 Dep. Tr. 31-32.)
68. As discussed above, there is nothing inherently anti-competitive about network effects. However, to the extent that anti-competitive conduct by Microsoft exists, network effects increase the risk that such conduct will further entrench Microsoft's monopoly.
69. The existence of network effects also implies that the effect of anti-competitive contracts or conduct will not be dissipated merely by terminating the anti-competitive contracts or stopping the anti-competitive conduct.
70. As the result of economies of scale and network effects, Microsoft's high market share leads to more applications being written for its operating system, which reinforces and increases Microsoft's market share, which in turn leads to still more applications being written for Windows than for other operating systems, and so on. This in turn supports the conclusion, also supported by Microsoft's internal documents and other evidence, that this share is not likely to be eroded by new entry as long as the applications programming barrier to entry remains strong.
71. There is abundant document and deposition testimony on the importance of network effects, scale economies, and the applications programming barrier to entry. For example:
72. Microsoft does not appear to consider other operating systems vendors as a material constraint on its pricing of the Windows operating system. For example, Joachim Kempin, Senior Vice President of OEM Sales at Microsoft, was asked if he considered any other vendors in setting the Windows 98 prices. He testified that he set the royalty rates for Windows 98 and that he never thought about looking at other vendors. (Joachim Kempin 3/18/98 Dep. Tr. 75-78.)
73. In a memorandum in December 1997 about potential competition that could affect Microsoft's OEM pricing strategy, Mr. Kempin described for Bill Gates the barriers to entry for OEMs that might be interested in entering the market:
"Our high prices could get a single OEM( Compaq might pay us 750M$ next year) or a coalition to fund a competing effort (say in India). While this possibility exists I consider it doubtful even if they get a product out that they can market it successfully, leapfrog us and would not deviate from their own standard to differentiate. Could they convince customer to change their computing platform is the real questions. The existing investments in training, infrastructure and applications in windows computing are huge and will create a lot of inertia. No bundling of OS on low end systems would be the easiest way to hurt us- but who would want to start with this and loose business?" (12/16/97 re "As promissed OEM pricing thoughts," Joachim Kempin to Bill Gates, et al.: Pl. Ex. 365, p. MS7 007196.)
74. Microsoft argues that it faces competition from its own installed base (i.e., the copies of earlier versions of its operating system already in the hands of users). (Richard Schmalensee 9/4/98 Expert Report pp. 6-7.) Even if that were true because of the absence of other competition, it does not follow that whatever constraint its own installed base poses is sufficient to prevent Microsoft from having monopoly power; indeed, the contrary is the case.
75. New operating systems are principally acquired in connection with the purchase of new computers and only secondarily in connection with upgrades. At best, Microsoft's installed-base argument relates to its pricing of upgrades. It does not apply to the more important channel of new computers.
76. New computers are bought largely to take advantage of developments in hardware or software. The fact that a given user has an old operating system will not do much to keep that user from changing computers when hardware or software advances, and a new computer is required to use those advances.
77. Moreover, Microsoft has taken action to ensure that installed-base competition is minimal. (For examples of these agreements, see Pl. Ex. 1113; Compare Kempin 3/18/98 Dep. Tr. 74-75.) Microsoft's licenses preclude customers from transferring their licenses to use Windows separately from their PCs; the ability to so transfer licenses would be a necessary condition for the development of a secondary market for such licenses that would permit OEMs to acquire such licenses as an alternative to licensing the use of Microsoft's newest version of its operating system. Microsoft's contracts with OEMs also generally prohibit them from shipping PCs to consumers with earlier versions of Microsoft's operating system once a new version is released.
78. Despite all the evidence set out above, Microsoft denies that it has monopoly power. Professor Schmalensee's expert report uses a well-known formula for profit-maximizing pricing and concludes that it shows that Microsoft does not have monopoly power. (Richard Schmalensee 9/4/98 Expert Report p. 6.) The proper conclusion from Schmalensee's argument cannot be that Microsoft lacks monopoly power. If any conclusion can be drawn, it would be that Microsoft is not maximizing its short-run profits.
79. It is probable that in the absence of intervention, Microsoft will obtain monopoly power in the market for Internet browsers.
80. There is a market for Internet browsers. Before Microsoft gave away its browser for free, a price for browsers was determined in this market and the market could have continued to perform this function. There is substantial demand for browsers that is separate from the demand for operating systems. Browsers are distributed separately from the operating system by ISPs and by retailers. There is demand for operating systems without browsers and for operating systems with a choice of browsers.
81. Barriers to entry (including network effects and the results of Microsoft's conduct) exist which prevent companies that might be able to produce a browser from entering the browser market. Indeed, by bundling its browser with its operating system and giving away its browser for "free," Microsoft effectively prevents companies from successfully entering the browser market unless they successfully enter the operating system market at the same time.
82. Microsoft has recognized that the dominant position of its Windows operating system is threatened by Internet browsers that are capable of supporting applications that are operating-system independent. The dominant position of Microsoft's operating system is protected by the applications programming barrier to entry described in some detail above. To the extent that browsers support applications independent of the operating system, they could erode the applications programming barrier to entry that protects Microsoft's monopoly in operating systems. In a 1995 memo to his "Executive Staff and direct reports," Microsoft CEO Bill Gates noted that Netscape was "pursuing a multi-platform strategy where they move the key API into the client to commoditize the underlying operating system." (5/26/95 "The Internet Tidal Wave," Pl. Ex. 20, p. MS98 0112876.3.) The following discussion explains how this could occur.
83. The browsers produced by Netscape run on many different operating systems, including Windows, the Apple Macintosh operating system, and various flavors of the UNIX operating system. (A list of these operating systems appears in Pl. Ex. 13.)
84. Netscape's browsers contain their own set of APIs (as well as a set of Java APIs) to which applications developers can write their applications. As a result, applications can be developed that will run on browsers regardless of the underlying operating system.
85. Because applications running on the browser are not operating-system specific, the Netscape browser could undermine the network effects and applications programming barriers to entry that currently protect Microsoft's operating system monopoly. In particular, by lessening the reliance on the operating system, the browser, while not performing all the traditional functions of an operating system, could provide opportunities for competing operating systems by reducing the applications programming barrier to entry that protects Microsoft's operating system monopoly.
86. With enough applications written to be operating-system independent, users might cease to care, or care as much, whether they had the same operating system as other users. As Gateway's James Von Holle testified, an interface that sits on top of the operating system reduces the need for a specific underlying operating system:
"If the user is not required to boot or launch an application directly from Windows and there was some sort of an underlying layer that sat between the interface and the PC operating system that abstracted those commands, ... then there would definitely be a threat to Microsoft Windows... The requirement for Windows would not be there." (James Von Holle 9/19/97 Dep. Tr. 32.)
87. Similarly, browsers could reduce the power of the operating system monopoly by facilitating the expansion of network computing, in which users with "thin clients" use a network to access applications residing on a server computer, rather than hosting the application on the PC itself. For example, in May 1998, Baer Tierkel, CEO of PeopleSoft Business Network, explained how the thin-client model is applied with his firm's enterprise software for large and medium-sized organizations: "All a customer must have for PeopleSoft's Internet solution to be available is a PeopleSoft-certified browser, along with a PeopleSoft-certified operating system.... We have created the Web client to run in a browser and is written using '100% Pure' Java. One reason we are making this effort is to reduce the cost to the end-user of maintaining client-based software." (BaerTierkel Decl. 5/11/98,1J5-7.)
88. Microsoft's own documents show an awareness of browsers as a serious threat to Microsoft's operating system monopoly, and its executives have expressed in both depositions and internal documents their concern that browsers could weaken Microsoft's control of the platform. As noted above, in 1995 Mr. Gates wrote that Netscape's strategy threatened to "commoditize" the operating system. Similarly:
89. The browser could also threaten the operating system monopoly by providing an alternative user interface that would reduce users' reliance on the dominant operating system interface. By providing a popular alternative user interface, browsers could reduce consumers' resistance to non-Windows operating systems and enable businesses that utilize the alternative interface to use different operating systems without increasing their training and support costs. In turn, this would reduce Microsoft's power to exploit the value of its interface real estate by requiring other companies to promote Microsoft's products through exclusive agreements.
For example, Brad Chase of Microsoft testified that Netscape was "... also trying to control and define user interface, which is another important element of an operating system..." (Brad Chase 3/25/98 Dep. Tr. 39.)
90. Microsoft was also concerned that browsers could ultimately develop into operating systems. For example:
91. Microsoft responded to the Netscape threat by adopting a strategy aimed at extending its dominance over PC operating systems to Internet browsers, and ultimately the Internet itself.
92. Microsoft recognized that it could protect its dominant position in the PC operating systems market by gaining and keeping a large share of the business in Internet browsers and by preventing any other browser from having a sufficient share to threaten Microsoft's platform dominance or remain viable. For example:
93. If Microsoft were to exclude competition in browsers, it would not be compelled by competitive pressure to ensure that its Internet Explorer browser could run on operating systems other than Windows. Nor would it be compelled by competitive pressure to support standards and technologies that are not tied to the Windows operating system.
94. Even though Microsoft faced browser competition from Netscape, the desirability of cross-platform activities was questioned at the highest Microsoft executive levels. In February 1997, James Allchin asked:
"Why are we doing so many things cross-platform? Are there more Macs, OS/2 or Unix clients today as a % or less than a year ago?
95. Moreover, if IE were the dominant browser and Microsoft decided to support only Windows-based technology, developers would have little incentive to create applications that were not Windows-based. In May 1995, Bill Gates noted:
"Another place for integration is to eliminate today's Help and replace it with the format our browser accepts including exploiting our unique extensions so there is another reason to use our browser...
96. Microsoft took a number of anti-competitive actions to exclude competition in Internet browsers. Taken together, it is clear that these actions are anti-competitive and that Microsoft would not have undertaken them except to exclude and foreclose competition.
97. Microsoft's activities to prevent the emergence of the browser as a platform threat are part of a consistent course of conduct to prevent other firms from developing any platform software that threatens the Windows operating system monopoly.
98. One of the first actions Microsoft took to stem the incipient threat to its monopoly posed by browsers was to solicit its emerging competitor, Netscape, to engage in a market allocation scheme. According to Netscape CEO James Barksdale, in June 1995 Microsoft proposed that the two competitors draw a line by "dividing the market between Windows 95-based browsers and all the others" with Netscape agreeing not to compete in supplying browsers for Windows 95. (James Barksdale 7/16/98 Dep. Tr. 236.)
99. Microsoft's attempt to enter into a horizontal agreement with Netscape to eliminate Netscape as a competitor supplying browsers for Windows 95 is worth some elaboration. The attempt is significant, first, because if Netscape had agreed, Microsoft would have succeeded in eliminating its only serious browser competitor and in monopolizing the market for browsers.
100. This attempt is also significant because it helps reveal the purpose and effect of actions taken by Microsoft when Netscape refused to agree to divide markets.
101. I have read Mr. Gates's deposition testimony in which he denied any participation in preparing for the June 1995 meetings with Netscape and in which he testified that he first heard of a June 1995 attempt to divide markets when a story appeared in the Wall Street Journal. (Bill Gates 8/27/98 Dep. Tr. 265-66.) (Mr. Gates also testified he did not know until the Wall Street Journal story that a claim of attempted market division was included in the Complaints of the United States and the state Attorneys General.) (Gates 8/27/98 Tr. 271-72.)
102. However, the testimony of participants in the June 1995 meetings and contemporaneous documents make quite clear that an attempt to divide markets between Microsoft and Netscape was made in June 1995.
103. For example, Mr. Barksdale (who participated in the June 21, 1995, meeting with Microsoft) and Netscape Chief Technology Officer Marc Andreessen (who also participated in the June 21 meeting) both testified that Microsoft tried to convince Netscape to divide the browser market by drawing a line between browsers for Windows 95 and all other platforms, including Windows 3.1, with Netscape and Microsoft agreeing that Netscape would stop marketing browsers for Windows 95.
104. As noted above, Microsoft proposed that the two competitors draw a line "dividing the market between Windows 95-based browsers and all the others." (Barksdale Tr. 236.) Mr. Barksdale also testified, "...we would be advised or allowed to sort of go to the market that wasn't Windows 95." If Netscape didn't do that, it would be "crushed." (Barksdale Tr. 236.) Mr. Barksdale's testimony is confirmed by Mr. Andreessen. (Marc Andreessen 7/15/98 Dep. Tr. 463-72.)
105. The testimony of Netscape's participants in the June 21, 1995, meeting concerning what happened at the meeting is also confirmed in important respects by the testimony of Microsoft participants. For example, Microsoft's Christopher Jones testified:
106. Contemporaneous documents confirm that the June 21, 1995, meeting included an effort by Microsoft to induce Netscape to agree to draw a line between Windows 95 browsers and other browser-related products and to induce Netscape to agree not to compete in supplying browsers for Windows 95. For example:
107. Microsoft's contemporaneous notes also confirm Microsoft's attempt to get Netscape to agree not to compete. For example:
108. The contemporaneous notes of Microsoft and Netscape are also confirmed by contemporaneous memoranda of persons not affiliated with either company who received reports concerning this extraordinary meeting from participants. For example, on June 22, 1995 (the day after the June 21 meeting), AOL's David Kaiser passed on to his superior at AOL Marc Andreessen's report of Microsoft's attempt to use a combination of threats and inducements to get Netscape to agree not to engage in platform competition with Microsoft. Mr. Kaiser wrote:
"Microsoft was at Netscape yesterday.... They wanted:
109. Microsoft has engaged in similar conduct with Intel and with Apple. It appears that when Intel proposed offering certain platform-level software, Microsoft objected on the grounds that this platform conflicted with Microsoft's platform plans. Microsoft threatened, among other things, to withhold support for Intel's new generations of processors if Intel proceeded with its plans. In the words of Intel chairman Andy Grove, Intel ultimately "caved" and withdrew the effort, at least under its own brand, explaining, "Introducing a Windows-based software initiative that Microsoft doesn't support. . . well, life is too short for that." (7/8/96 Brent Schlender, "A Conversation with the Lords of Wintel," Fortune: Pl. Ex. 559, p. 8.)
110. Another Intel employee, Steven McGeady, Vice President of the Content Group and Director of Intel's Health Technology Initiative, explained that Microsoft also discouraged Intel from supporting Netscape or Java as an alternative platform. (Steven McGeady 8/10/98 Dep. Tr. 11-12, 18-20, 34-37, 57-60, 64-67.) In August 1995 Mr. McGeady wrote:
"On August 2, 1995, in a meeting of Intel and Microsoft executives, Bill Gates told Intel CEO Andy Grove to shut down the Intel Architecture Labs. Gates didn't want IAL's 750 engineers interfering with his plans for domination of the PC industry. Gates made vague threats about support for other platforms, and on the same day he announced a major program to support Digital Equipment's Alpha microprocessor, an Intel competitor. Gates was livid about IAL's investments in the Internet, and wanted them stopped. All of this was said in the presence of executives from both companies." (8/28/95 Pl. Ex. 280, p. INT 0386.)
111. Microsoft's internal documents, including particularly confidential messages from Bill Gates personally, confirm Microsoft's attempt to convince Intel to agree not to engage in platform competition with Microsoft. For example:
112. Microsoft documents also confirm that Microsoft used its relationship with Intel to discourage Intel from supporting Java or Netscape's browser. For example:
113. Microsoft's internal documents also confirm that Microsoft engaged in extensive efforts to convince Intel not to support competing technologies, even when those competing technologies would enhance the performance of Windows PCs. For example, Microsoft's General Manager for Internet Multimedia, Eric Engstrom, wrote to his superiors that he was working "aggressively" to convince "Intel to stop helping Sun create Java Multimedia APIs, especially ones that run well (ie native implementations) on Windows." (5/26/97 Eric Engstrom to John Ludwig and David Cole, et al.: Pl. Ex. 235, p. MS7 027416.)
114. Microsoft engaged in similar conduct with Apple. Timothy Schaaff, Senior Director of the Interactive Media Group at Apple, described how Apple promoted QuickTime, its multimedia streaming technology, as an audio/visual content creation/authoring and playback mechanism on the Windows operating systems. (Timothy Schaaff 8/28/98 Dep. Tr. 37-38.) Microsoft, however, considered audio/visual streaming technologies to be part of a "growing collection of technologies" that "were a threat to the Windows platform." (Benjamin Slivka 9/3/98 Dep. Tr. 244-245.)
115. Beginning in 1997, Microsoft representatives informed Apple that "Microsoft wanted to have control over the user interface ... and that Microsoft was determined that the essential APIs that were the foundation of the operating system should all come from Microsoft and not come from a third party." (Schaaff 8/28/98 Tr. 42.) Microsoft offered to forego competing in the multimedia authoring tools market if Apple would scale back its efforts to establish QuickTime as a multimedia platform on the Windows operating system. (Schaaff 8/28/98 Tr. 38-41.) In addition to this inducement, Microsoft also set forth a threat: in the absence of an accommodation, Microsoft could devote 100 to 150 engineers to competing against Apple in the authoring tools market. (Schaaff Tr. 60.)
116. Microsoft representative Eric Engstrom told Mr. Schaaff that, although "Bill Gates was not interested in creating multimedia authoring software .... if controlling the authoring was a necessary element of ultimately owning the playback story, that that would make it an acceptable strategic move, even if it didn't make sense from a business standpoint." (Schaaff Tr. 58, 60-61.)
117. As these incidents indicate, Microsoft will respond immediately to prevent any other firm from writing platform-level software. This is true even though this software could increase the functionality and performance of, and thus the demand for, Windows-based PCs.
118. Microsoft's conduct with respect to Intel and Apple is useful in analyzing the purpose and effect of Microsoft's efforts to prevent browsers from surviving as a source of APIs that could be used to develop cross-platform applications software. For example:
119. As discussed above, taking action that does not "make sense from a business standpoint" in order to restrict competition (and thereby profit from the reduced competition) is the essence of predatory anti-competitive conduct.
120. Microsoft's response to the prospect of a successful Netscape browser with cross-platform APIs that could erode the applications programming barrier to entry is a textbook example of predatory conduct.
121. Microsoft recognized the potential threat posed by Netscape's browser. As discussed above, in May 1995, Microsoft CEO Bill Gates warned his top executives that the browser could "commoditize" the operating system, and in June 1995, Microsoft attempted to agree with Netscape that Netscape would not offer a browser for Windows 95.
122. At the same time, Microsoft began devoting at least $100 million per year to developing its own browser. (Microsoft's Answers to Interrogatories, Civil Investigative Demand No. 18140, Interrogatory 4.) An early step in that process was Microsoft's licensing of the right to use the code of the Mosaic browser. Microsoft then undertook a massive development effort to expand and improve the Mosaic code. Microsoft also spent tens of millions of dollars a year marketing and promoting Internet Explorer.
123. Despite the huge browser-related costs it was incurring, Microsoft distributed its browser at a negative price. The IE browser was not only given away free; companies were also paid money and given valuable concessions to accept, use, distribute, and promote IE. Examples include:
124. Microsoft's internal documents make clear that Microsoft undertook its browser development not to make money from browsers, not because doing so would "make sense from a business standpoint" on its own, but to prevent Netscape's browser from facilitating competition with Microsoft's monopoly operating system. This is the essence of predatory monopoly maintenance. For example:
125. Microsoft, at Bill Gates's personal direction, undertook detailed studies of Netscape's sources of revenue and what Netscape required to survive as an effective competitor. Microsoft estimated that at the time Microsoft made its decision to supply IE without charge, from 20 percent to 50 percent of Netscape's revenues came from licensing its browser. (Bill Gates 8/27/98 Dep. Tr. 236.) Microsoft's decision to price its browser below cost (indeed, at a zero or even negative price) was thus made when it knew that Netscape was charging for its browser and that Netscape depended on those revenues to continue to compete effectively. For example:
126. It is important to note that this is not merely colorful language that could be interpreted either as aggressive competition or as evidencing a predatory intent (for example, language like: "Our goal is to get 100% of the business" or even like "Let's kill the competition"). This is language that accurately describes the purpose and effect of Microsoft's conduct distribute its browser at a zero or negative price in order to eliminate competition.
127. Without the gain to Microsoft that will result from preserving its highly profitable operating system monopoly and from monopolizing the browser market, Microsoft's conduct does not "make sense from a business standpoint." It is giving away, indeed paying people to take and distribute, something that it has spent a lot of money to develop and distribute and something for which the leading competitor was charging. Netscape's revenues from browsers for clients fell sharply in 1997 to zero in the first quarter of 1998, as shown in Pl. Ex. 9 and Pl. Ex. 10.
128.It is only when Microsoft's gains from preserving and extending its monopoly are included that Microsoft's conduct is profitable.
129. Microsoft in this litigation argues that its conduct is profitable without considering gains from reducing competition because the wide distribution of its browser will cause more people to buy PCs to browse the Internet, with the result that Microsoft will sell more copies of its Windows operating system. There are many reasons why this argument is wrong. They include:
I am told that a judge once said that judges should be careful not to forget as judges what they know as men and women. The same is true of economists.
130. There are many difficult questions that can arise in analyzing whether a price is or is not predatory. It is not necessary to answer those questions in the present case. For example:
131. The potentially difficult questions that arise when analyzing whether a price is or is not predatory arise because of a desire to provide companies with an objective "safe harbor" in situations where pricing could reasonably be argued to be above average variable cost and profit-maximizing. In the usual case, the analysis is also complicated by the issue of whether the company whose pricing is challenged is a monopolist or merely a competitor competing aggressively on the merits. In the present case:
132. Given such facts, classifying Microsoft's pricing as predatory and anticompetitive does not require reaching the difficult questions that are usually faced in predatory pricing cases. (It is worth noting that the desire for an "objective" test that is sometimes used to argue against a profit-maximizing test (as opposed simply to comparing a product's revenues with its costs of production) does not apply either where the absence of profit-maximizing pricing is explicit (as it is here) or where proof of revenues from projected ancillary products or services is necessary in any event (as will always be the case where price is zero or negative).)
133. Microsoft's predatory pricing was part of, and should be evaluated in connection with, a broader campaign to eliminate Netscape's browser and Java as sources of potential competition to Microsoft's operating system monopoly a campaign characterized by actions by which Microsoft lost money in order to raise rivals' costs and exclude them from the market;5 by actions that Microsoft recognized internally did not "make sense from a business standpoint," except for their anti-competitive effects; and by Microsoft's agreements with customers and competitors that require them to refuse to deal with Netscape and other browser competitors or to do so only on unfavorable terms.
134. For example, Microsoft sought to further deprive Netscape of revenue by inducing ICPs to agree not to pay Netscape for carrying or promoting the ICPs' content or logos. Moreover, Microsoft was prepared to give away valuable concessions to ICPs to secure such agreements.
135. Microsoft also entered into agreements with companies such as Intuit, a leading software applications supplier that competes with Microsoft in the supply of applications software, in which the companies agreed to "Bundle IE3 (Quicken) and IE4 (other products) will all new 97 and 98 releases of Intuit products," and to "Not enter into marketing/promo agreements with Other Browser manufacturers for distribution or promotion of Intuit content." (4/17/97 "Intuit Terms Agreed," Will Poole to Brad Chase: Pl. Ex. 206.)
136. Microsoft's dealings with Apple are evidence of how far Microsoft was willing to go to limit Netscape's opportunities and to stifle Java. One of Bill Gates's explicit "key goals" was to get Apple "to embrace Internet explorer in some way." (6/23/96 "Apple meeting," Bill Gates to Paul Maritz et al.: Pl. Ex. 260, p. MS98 0113116.) In June 1996, Mr. Gates proposed a "deal" to "top Apple executives" in which the first element of what "Microsoft gets" was "Apple endorses Microsoft Internet explorer technology." (6/23/96 "Apple meeting," Bill Gates to Paul Maritz et al.: Pl. Ex. 260.) Microsoft's determination to get Apple to agree to work "against Sun and Netscape" and its willingness to engage in conduct that didn't "make sense from a business standpoint" to accomplish that purpose are illustrated by numerous Microsoft documents. For example:
137. Microsoft's determination to restrict the support and distribution of Netscape's browser by Apple is particularly significant when one considers that Apple represents the main potential alternative to desktop PCs running Microsoft's Windows.
(Although that alternative is not sufficient to keep Microsoft from having monopoly power.) Whatever the relevance of Microsoft's arguments about why it wanted to make Internet Explorer available to sell more copies of Windows, those arguments cannot apply to Microsoft's efforts to force Apple to distribute Internet Explorer. As Mr. Gates candidly told his Executive Staff on August 20, 1997, he "doesn't care that much about being aligned with apple, he just wants them split from other potential allies." (8/21/97 re "conversations with billg last nite," John Ludwig to Don Bradford: Pl. Ex. 255.) In addition, there is no legitimate justification for Microsoft and Apple (two competitors) entering into an agreement "to undermine SUN." (8/8/97 re "post-agreement," Bill Gates to Paul Maritz et al.: Pl. Ex. 265.)
138. Actions taken by Mr. Gates where he "doesn't care that much about being aligned with apple, he just wants them split from other potential allies" are yet another example of actions that did not "make sense from a business standpoint" on their own merits but which were undertaken with the purpose and effect of excluding competition.
139. Although it is obvious, it is worth emphasizing that the Microsoft documents that reveal Microsoft's predation are not documents from low-level employees or employees likely to be misinformed about the purpose and effect of the company's conduct. Many of the most significant documents are documents to or from CEO Bill Gates, personally.
140. Microsoft made a deliberate decision to "tie" or "bundle" its IE browser with its Windows operating system.
141. Moreover, Microsoft's bundling was not restricted to OEM licensing. Although IE was not originally bundled with the retail version of Windows 95 when it was first released in the summer of 1995, Microsoft soon bundled IE with Windows 95 in distributing Windows 95 to OEMs, and IE is now bundled with all Windows 95 and Windows 98 operating systems that Microsoft distributes through retail or OEM channels.
142. In an effort to weaken Microsoft's browser competition in order to protect Microsoft's monopoly in operating systems, Microsoft made the decision to bundle IE and Windows even though there is demand for browsers separate from the demand for operating systems.
143. Microsoft made its decision to combine its browser and operating system not to achieve efficiencies but to foreclose competition. For example:
144. Microsoft also recognized that OEMs wanted the ability to develop their own screens and substitute Netscape's browser for IE. As Bill Gates wrote to his staff in January 1996:
"Winning Internet browser share is a very very important goal for us.
145. As a result of Mr. Gates's initiative, Microsoft later that year imposed screen and start-up restrictions to prevent OEMs from developing their own first screen or positioning competing browsers more favorably than IE.
146. In connection with its tying of IE and Windows, Microsoft has required the distribution of IE and restricted the distribution of other browsers by entering into restrictive agreements with PC original equipment manufacturers. The agreements have required PC manufacturers who wanted to preinstall Windows 95 or Windows 98 on their machines also to preinstall Microsoft's Internet Explorer. The agreements also limited the ability of OEMs to promote other browsers, or to substitute other browsers for IE.
147. Until changes were prompted by an early 1998 stipulation between Microsoft and the Department of Justice, the agreements typically required that licensees not modify or delete any of the product software. This prevented OEMs from removing any part of IE, including the visible means of user access to the IE software, such as the IE icon on the Windows desktop or the IE entry in the "Start" menu. This was true despite the fact that many OEM executives expressed an interest in removing the IE icon from the desktop. For example:
148. Typically, Microsoft's agreements with OEMs stipulated that licensees may not modify or obscure the appearance of the start-up or desktop screens as they appear during the initial boot. While licensees may add icons or folders on the desktop, the icons (folders) must be the same size as icons (folders) already on the desktop and of similar shape. The effect is to limit the ability of OEMs to promote other browsers by, for example, highlighting the existence of another browser with a large desktop icon or modifying the start-up sequence to give users an opportunity to make a non-IE browser their default browser or replace IE with a competing browser. This, again, was in spite of the fact that OEM executives had expressed an interest in modifying the startup sequence to give users a choice of browsers or achieve other benefits for users. For example:
149. James Barksdale of Netscape testified about the effects of the OEM restrictions on his company:
150. Licensees were not contractually restricted from loading other browsers on the desktop. However, some OEMs preferred to load only one browser to avoid user confusion and the resulting consumer support costs, and to avoid increased testing costs. For example:
151. In addition, some OEMs viewed the desktop and/or disk space as scarce real estate and were generally reluctant to preinstall more than one software title in each functional category. For example, Stephen Decker, Director of Software Procurement at Compaq, was averse to loading a second browser because it would "take up additional real estate on our hard drive." (Stephen Decker 10/17/97 Dep. Tr. 22.)
152. Microsoft's restrictions on the startup screen have been somewhat modified so that OEMs had somewhat more flexibility than when the restrictions were imposed. OEMs are also able to select ISPs to be in the Microsoft referral service (8/7/98 Cameron Myhrvold Decl. ¶8.) However, IE must still be installed on every PC and the IE icon cannot be removed. Thus, since Microsoft's tying arrangement ensures that IE is on every Windows PC, the result is a significant exclusionary effect that ensures that IE is the only browser on most PCs shipped by OEMs.
153. Microsoft also entered into a restrictive agreement with Apple that requires Apple to make IE its default browser on all of its Macintosh operating systems. This agreement forced Apple to place all competing browsers in a folder (i.e., banishes other browsers from the Macintosh desktop) and limited Apple's ability to promote other browsers. (8/5/97 "Technology Agreement between Apple Computer, Inc., and Microsoft Corporation": Pl. Ex. 1167, pp. MAC 0044-45.) In order to induce Apple to enter this contract, Microsoft, among other things, threatened to stop development of its Office application suite for the Macintosh. As Microsoft knew, withdrawal of support for this crucial application would have had a devastating effect on the viability of the Macintosh operating system. (Avadis Tevanian 7/17/98 Dep. Tr. 135-42.)
154. As Avadis Tevanian, Senior Vice President of Software Engineering at Apple, has indicated, whatever the merits of IE as a browser there were "certainly no dependencies or it wasn't necessary to have IE be the default." (Tevanian Tr. 142-43, 149.) This suggests strongly that Microsoft's actions are not merely technology- or efficiency-driven.
155. Microsoft has proffered a number of justifications for its conduct, but none suggests that Microsoft's primary motive was anything other than to restrict competition in browsers.
156. Microsoft has designed interdependencies between IE and Windows, and claims that this is a rationale for its bundling practices. (Microsoft Corporation's Response to Plaintiffs' First Joint Set of Written Interrogatories to Microsoft Corporation, 6/27/98, pp. 27-28; Richard Schmalensee 9/4/98 Expert Report pp. 29-30; Defendant Microsoft Corporation's Memorandum in Support of its Motion for Summary Judgment, pp. 6-9.) But even if two products as designed cannot readily be separated, the bundling or tying of the two can raise the same anti-competitive concerns that contractual bundling or tying would raise. Moreover, such concerns are not automatically overcome merely because the bundle brings some amount of consumer benefit to certain consumers. (For my purposes it does not matter whether one calls Microsoft's actions "bundling" or "tying" or both. Here, Microsoft does not merely offer IE at no separate charge but effectively forces customers (OEMs) to take it. Of course, the anti-competitive effects of Microsoft's actions are independent of the nomenclature.)
157. Virtually every product design, particularly in the area of computer software, can make a plausible claim for some efficiency or benefit. Many software products can be combined in such a way that they share certain code; if code is shared there is some plausible efficiency (although perhaps very slight), and separating the two products once they have been combined may be very difficult.
158. If combining two products in a way that produces plausible efficiencies (however slight), or that makes it difficult to separate the products, were an absolute defense to a claim that the combination was anti-competitive, software commerce would be essentially immune from tying scrutiny. In the present case, the anti-competitive effects are large; the technological benefits appear to be small or non-existent.
159. Microsoft has argued that it must force OEMs to take IE because the absence of IE may undermine the quality of the operating system, to the detriment of users. However, several facts contradict this suggestion. For example, Microsoft provided ways to remove IE in Windows 95 a function that would most likely not have been provided if it led to a decrease in the quality of the operating system.
160. Also, as I understand it, it is possible within Windows 98 to remove the ability to browse the Web with IE and to replace IE with another browser with no appreciable decline in the quality of the Windows 98 operating system. As Edward W. Felten, Assistant Professor of Computer Science at Princeton University, explained, "It is possible to construct a mechanism for removing Web browsing from Windows 98....This demonstrates that Microsoft could have produced a version of Windows 98 without Web browsing in a way that did not endanger the functionality of the operating system." (Edward W. Felten 9/1/98 Expert Report pp. 13-14.)
161. Professor Felten further stated that:
"I have run the prototype removal program on a computer, and then installed Navigator. A computer that has undergone these procedures suffers no apparent loss of stability or functionality and provides the user the full Web browsing experience offered by Navigator." (Edward W. Felten 9/1/98 Expert Report p. 15.)
162. In fact, Microsoft has permitted Dell to remove IE from the desktop for Windows 95 at the request of the OEM's large customers. Joseph Kanicki, Strategic Commodity Manager at Dell testified:
"Dell negotiated for and obtained an exception to this requirement for instances in which a customer requested a Windows 95 or Windows NT computer without Internet Explorer." (Joseph Kanicki, Jr. 4/29/98 Decl. ¶2)163. Presumably, Microsoft would not allow this kind of exception if it undermined the quality of the operating system. Likewise, OEMs would not negotiate to remove IE if the operating system would be adversely affected, since a poorly operating computer would reflect poorly on the OEM and would be likely to increase the number of customer support calls; also, large customers would not request an operating system with IE removed if they felt this system would be adversely affected.
164. As noted above, Microsoft is allowing OEMs slightly more flexibility on the first screen and the ISP registration process. It seems unlikely that either Microsoft or the OEMs believe these changes will lead to significant deterioration in the quality of the operating system.
165. Microsoft has also argued that its bundling of IE is necessary to provide a uniform platform for software developers. But, as noted above, Professor Felten demonstrated that Windows 98 could have been produced without IE while still allowing independent software vendors (ISVs) to write applications. In any event, in light of the different versions of Windows and IE that Microsoft has put in the marketplace, developers that rely on system services or code found in IE must redistribute the necessary IE code anyway to ensure that the proper version of the necessary DLL (dynamic link library) or file is present to support their applications. For example:
166. Microsoft has argued that it is justified in restricting OEMs from altering the start-up process to preserve the quality and speed of the start-up process and to give each user a consistent experience. However, the fact that Microsoft has granted exceptions to these restrictions to certain OEMs suggests that the concern for quality, speed, and consistency is not Microsoft's primary motive for enforcing these restrictions. (See, e.g., Microsoft Windows Initial Boot Process, letters from Karen Hurlbut, Microsoft's General Manager OEM Operations: 5/26/98 to Marina Morrilla, Hewlett-Packard: Pl. Ex. 773, p. MS98 0113871; 5/26/98 to Senior Business Manager, Software Supply Management, Gateway: Pl. Ex. 774; and 6/22/98 to Joe Kanicki, Strategic Commodity Manager at Dell Computer Corporation: Pl. Ex. 1196.)
167. This conclusion is supported by internal Microsoft documents establishing the restrictions that admit that third parties can "be part of the original feature set if they work closely with us to promote our platforms and thus help us strategically." (9/17/96 Kempin to Gates, Maritz and Ballmer: Pl. Ex. 304, p. MSV 0009376 A.)
168. If Microsoft were a small company with a small share of operating systems, its bundling provisions would be harmless. Given Microsoft's dominance, these types of provisions are anti-competitive. They inhibit PC manufacturers from preinstalling and promoting competing browsers. Their purpose and effect are to weaken Microsoft's browser competition in order to protect Microsoft's business in operating systems. The benefit gained by creating interdependencies between IE and Windows would have to be great to counterbalance the anti-competitive effects of bundling.
169. Microsoft has also required the promotion and distribution of IE and restricted the promotion and distribution of other browsers by striking deals with ISPs in order to protect Microsoft's business in operating systems. ISPs and the OLSs are, after OEMs, the largest distributors of browsers. In December 1996, Sam Jadallah and Cameron Myhrvold of Microsoft wrote that "ISPs Drive Browser Market share. 35% of end-user Internet access customers get their browser from an ISP." (12/18/96 Memorandum Re: Plan of Record: Working with ISPs in North America (DRAFT): Pl. Ex. 200, p. MS6 6011450; Cameron Myhrvold 4/24/98 Dep. Tr. 43.)
170. Microsoft's approach to ISPs was to license IE for free:
"If an ISP is willing to make IE the preferred browser and agree to a few other requirements in our license agreement we offer to license IE and its add-on components for free. We allow them to distribute another browser if they wish but it is very important that IE is the preferred browser. We will not sign deals were that is not the case." (9/9/96 re "ISP marketing update," Bjorn Hovstadius to Cameron Myhrvold et al.: Pl. Ex. 93, pp. MS6 5003741-42.)
171. Because of the monopoly position of its Windows operating system, Microsoft understood that ISPs would be very interested in having favorable placement on the Windows desktop. As part of its effort to exploit its Windows advantage, Microsoft designed a special access method called the Internet Connection Wizard to assist users in signing up for ISPs. Only a few ISPs could be accessed through the Internet Connection Wizard. Initially there were twelve, including some of the largest ISPs.
172. By mid-August of 1996, Microsoft had signed "IE Preferred" distribution agreements with about 2,500 ISPs. Among these were most of the largest ISPs in the United States. These agreements usually specified that IE would be the preferred and default browser. While the ISPs could distribute other browsers, Microsoft expected a large fraction of them to distribute only IE. "Of the 2,500 signed, I'll bet 2,000 only distribute our browser." (8/13/96 re "IE 3.0 Sales Status," Cameron Myhrvold to Steve Ballmer: Pl. Ex. 193.)
173. Some ISPs had agreements that allowed them to distribute IE and Netscape without preferences; Microsoft's documents use the term "IE Parity" to identify these companies.
174. Microsoft also created another desktop folder for ISPs which were on-line services providers (OLSs) and entered into agreements with America Online, CompuServe, Prodigy, and AT&T to appear in it.
175. Brad Silverberg, formerly Senior Vice President of Applications and Internet Client Group at Microsoft, described the advantages of Microsoft's mechanisms for signing up Internet and online services subscribers, such as the Internet Connection Wizard and online services folder. In the context of questioning about Microsoft's negotiations with AT&T, Mr. Silverberg testified that:
"We made it very easy for AT&T to acquire customers and sign up and have them configured. And you wouldn't have to have a CD mailed to you." (Brad Silverberg 4/14/98 Dep. Tr. 176.)
176. Microsoft used the strong demand by OLSs for access to Microsoft's Windows operating system to extract promises from the services not to deal with Netscape or to do so only on very unfavorable terms. Mr. Silverberg testified that he told AT&T:
"You want to be part of the Windows box, you're going to have to do something special for us. There are very, very few people we allow to be in the Windows box. If you want that preferential treatment from us, which is extraordinary treatment, we're going to want something very extraordinary from you." (Brad Silverberg 4/14/98 Dep. Tr. 159.)
177. Silverberg explained that this meant giving IE "exclusive or very, very preferential treatment (ala what we have with AOL). Parity is completely unacceptable for them to be in the box." (3/15/96 re "AT&T Meeting (3/13) Trip Report," Brad Silverberg to Steve Wells et al.: Pl. Ex. 183, p. MS6 6010581.)
178. While Bill Gates initially was "very, very uncomfortable about shipping AOL" on Windows because of its own online service, MSN (Brad Silverberg 4/14/98 Dep. Tr. 187), Microsoft's agreement with AOL turned out to be an important deal for increasing IE's share. At the time the agreement was signed, in early 1996, AOL was being installed on PCs by many of the OEMs. AOL also had by far the largest subscriber base of all the OLSs. Up to that point it had used its own proprietary browser, based on Booklink. (David Colburn 3/6/98 Dep. Tr. 20-24.)
179. David Colburn, Senior Vice President of Business Affairs at AOL, testified that the "AOL client would be on the Microsoft desktop in the online services folder for a specified period of time and that we essentially had to make Microsoft the default browser as what we included in the AOL client." (Colburn Tr. 24.) AOL was also subject to percentage restrictions on shipments of other browsers as well as other restrictions on the promotion or distribution of other browsers. Mr. Colburn also testified that "we had...some requirements about distributing updated Internet explorers on an accelerated basis to our clients." (Colburn Tr. 48.)
180. AOL's performance under the restrictive agreement with Microsoft had strong positive effects on Microsoft's browser share. As of November 1997, a Microsoft document reported that "straight" IE users gave IE 24 percent share, while AOL IE users provided an additional 17.9 percent share. (1/12/98 re "AOL now at 92% share according to SVP at AOL," Bill Koszewski to Cameron Myhrvold et al.: Pl. Ex. 220, p. MS98 0120897.) In January 1998 Microsoft stated that "IE is at 92% share at AOL That is up from 85% in November." (1/12/98 re "AOL now at 92% share according to SVP at AOL," Mauricio Gonzalez de la Fuente to Laura Torina, Cameron Myhrvold, et al.: Pl. Ex. 220, p. MS98 0120897.) As noted below, restrictions on AOL and the other OLSs were not waived by Microsoft earlier this year when restrictions were modified for the ISPs.
181. Microsoft explicitly recognized that the decision to grant OLSs favorable access to Windows (particularly AOL, which was the leading competitor to Microsoft's MSN on-line service) was an expensive one. Bill Gates decided that the lost opportunity was less important than its over-riding goal of winning "the browser battle" and protecting its core monopoly. For example:
182. While Microsoft charged a referral fee for customers the ISPs acquired through the Windows 95 desktop, browser share, not revenue, was the object of the agreements. Microsoft also made valuable concessions, directly or indirectly, to the ISPs. These varied across ISPs but included joint marketing programs, pricing deals, and discounts from referral fees for users switched from competitive browsers. As Mr. Myhrvold testified at his deposition, the ISP referral server did not pay for itself. (Cameron Myhrvold 4/24/98 Dep. Tr. 137.)
183. A Microsoft document describes how Mr. Silverberg dealt with AT&T's $17 million minimum commitment for purchases of Netscape's browser:
"...they'd really like to be browser neutral and are strongly motivated to preserve their partnership with Netscape. But bradsi [Brad Silverberg] has told them that to get in the box they need to give us preferred status; bradsi also told them he'd let them use our bounties to pay down the Netscape min commits if we got preferred status." (3/14/96 Dan Steele to Brad Chase et al.: Pl. Ex. 179.)
184. While there was some variation in the restrictions imposed on the OLSs and other ISPs, these agreements with Microsoft limited the ISPs' ability to promote and distribute third-party browsers. In general these agreements stated that Microsoft would provide users with access to ISP services from the desktop, and in return, ISPs were not only required to promote IE, but they were also required not to promote other browsers. (For examples, see Pl. Ex. 1115.) Typically, such restrictive provisions included:
185. For example, according to Stephen Von Rump, Vice President of Enterprise Services at MCI, MCI's Internet-Sign Up Wizard agreement with Microsoft signed in July 1996, required that MCI not display the logo of or maintain a link to a browser other than IE. MCI was restricted in its ability to advertise or otherwise promote Netscape Navigator, and could not ship Navigator unless specifically requested to do so. Moreover, MCI could not tell customers that other browsers were available to them. In addition, 75 percent of the browsers it shipped had to be IE; if not, Microsoft could delete MCI from the ISP Phone Book, in which Microsoft kept MCI's name, telephone number, and other information. (Pl. Ex. 1138, p. MS6 5000870) To Mr. Von Rump's knowledge, MCI complied with the terms of its agreement and Microsoft expected MCI to comply. (Stephen Von Rump 4/28/98 Dep. Tr. 9-13, and 9/3/98 Dep. Tr. 25-26, 29-31.)
186. In April of 1998, Microsoft issued a statement to certain ISPs with which it had restrictive agreements, waiving some of the restrictions in their agreements. For example, in a letter to Earthlink, Microsoft committed not to enforce provisions concerning distribution volumes or percentages, discussion, promotion, or advertising of IE and the use of IE as a standard or default browser. In addition, restrictions, performance obligations, and qualifications for referral fees were removed. (Cameron Myhrvold 8/7/98 Decl. ¶4; 4/21/98 Cameron Myhrvold of Microsoft to Leland C. Thoburn: Pl. Ex. 374, pp. MS98 0106631-32.)
187. However, ISPs in the Internet Connection Wizard were (and are) still prohibited from distributing and promoting Navigator with "preference." IE must be discussed, promoted, or advertised so that in its entirety, its treatment is no less prominent and favorable than that accorded to Navigator.
188. Moreover, restrictions were not waived for ISPs who were OLSs. For example, the contractual restrictions in Microsoft's agreement with AOL, the largest provider of Internet access, were not waived. Based on his talks with executives of AOL, James Barksdale, Netscape's CEO, was convinced in July of this year that AOL was prohibited from distributing Navigator because of its contract with Microsoft. (James Barksdale 7/16/98 Dep. Tr. 34-37.) Brad Chase of Microsoft confirmed that the restrictions in the OLSs agreements were not removed. (Brad Chase 3/25/98 Dep. Tr. 169-170, 173-176.)
189. At least one ISP was not aware of Microsoft's asserted waivers of ISP restrictions. Mr. Von Rump of MCI testified at his deposition that he had discussed an agreement with Microsoft that would remove some of the restrictions, but that neither party had yet signed it. He was unaware of Microsoft's waiver announcement and had not been notified of it. He also testified that he believed the unsigned agreement would not allow MCI to make Navigator its default browser because that would give it preferential treatment compared to IE. (Stephen Von Rump 9/3/98 Dep. Tr. 33-34, 40.)
190. According to Microsoft, its 1998 agreements with ISPs will not contain restrictions. However, Microsoft remains free to reimpose even the waived restrictions, and whatever the extent of Microsoft's waiver, it did not undo the harm to competition that had already occurred.
191. As I explain below, relegating Microsoft's competitors to distribution through decidedly inferior channels has serious consequences in foreclosing its competitors and raising their costs.
192. In these agreements, Microsoft offered ISPs valuable space on its desktop. But rather than trading desktop space for financial remuneration, Microsoft placed requirements on ISPs that hindered their ability to promote or distribute Netscape Navigator. Again, given Microsoft's position in operating systems, these provisions are anti-competitive. Their purpose and effect are to reduce the ability of competing browser manufacturers to distribute and promote their browsers through leading ISPs. Regardless of whether such provisions would be anti-competitive in themselves if put in place by a company with a small share of operating systems, they are certainly anti-competitive when Microsoft uses them to protect its dominant position in operating systems.
193. ICPs create program content for the World Wide Web. Microsoft has also promoted the use of IE and restricted the promotion and distribution of other browsers in its agreements with ICPs for its Channel Bar. (Examples appear in Pl. Ex. 1114.)
194. ICPs valued the opportunity to have a channel on the Microsoft desktop, because it encouraged users to visit the ICPs' Web sites, which in turn increased the ICPs' ability to promote their own products and to sell advertising space on their Web pages. For example:
195. Typically, the general nature of the agreements was that, in return for a prominent position on Microsoft's Channel Bar, ICPs agreed not only that they would promote IE, but also that they would not promote or distribute competing browsers. Some of the more restrictive provisions typically included in the agreements are as follows:
196. Examples of these restrictions for particular ICPs include:
197. Yusuf Mehdi of Microsoft stated that the ICP agreements have not been commercially successful, and therefore they will not be renewed after they expire on September 30, 1998 (Yusuf Mehdi 8/8/98 Decl. ¶14.) However, again, damage to the competitive process has already occurred.
198. Microsoft also has restrictive agreements with other software vendors. For example, its "Internet Explorer Marketing, Distribution & Promotion Agreement" (6/6/97) with Intuit and its "Marketing, License, Distribution & Promotion Agreement" (12/7/96) with PointCast generally prohibit Intuit and PointCast from distributing other browsers with their most popular software products. (Pl. Ex. 1157, p. MS98 0100570; Pl. Ex. 1153, p. MS98 0100811.)
199. Again, given Microsoft's position in operating systems, these provisions are anti-competitive, because they preserve Microsoft's large share of business in operating systems by hindering competition from other browsers. In particular, the provision that prevents ICPs from compensating a company that produces other browsers for carrying or promoting the ICPs' content or logos can have no purpose other than that of damaging those browser suppliers. It is not a profitable act by Microsoft independent of its effect of weakening the competition. Regardless of whether such provisions would be anti-competitive in themselves if put in place by a company with a small share of operating systems, they are certainly anti-competitive when Microsoft uses them to protect its dominant position in operating systems.
200. Microsoft's anti-competitive actions restrain the use and availability of Java technology in order to protect the current dominance of the Windows operating system.
201. A Java Runtime Environment (JRE), which consists of a Java Virtual Machine (JVM), the Java platform core classes, and supporting files, is a software layer with its own API set that resides on top of an operating system and is designed to allow applications written in Java to function on different operating systems.
202. Microsoft has created its own implementation of Java technology that includes its own JRE. Michael Homer, Netscape Executive Vice President and NetCenter General Manager, explained in his deposition that "...Microsoft's implementation of Java is nonstandard and inconsistent with the reference version that Sun has published." (Michael Homer 8/4/98 Dep. Tr. 261.)
203. Browsers (i.e., non-Microsoft browsers) are an important distribution channel for JREs, as Curtis Sasaki, Group Marketing Manager of SUN Microsystems, testified:
204. As discussed above, Microsoft recognized Sun Microsystems' Java as a threat to its operating system monopoly because Java, like browsers, offered the potential for eroding the applications programming barrier to entry.
205. Indeed, following, and based on, a meeting with Bill Gates in early 1997, Microsoft's Ben Slivka described Java as "...the biggest threat to Microsoft" and wrote to Mr. Gates that "...clearly the work the Java team is doing has hit a raw nerve with you." (4/14/97 re "java review with you," B. Slivka to Bill Gates: Pl. Ex. 58.)
206. Microsoft concluded that there exists a need "... to fundamentally blunt Java/AWT momentum" to "..protect our core asset Windows." (6/20/96 re "windows & internet issues," Paul Maritz to Brad Silverberg et al.: Pl. Ex. 42, p. MS6 6010347.) To this end Microsoft sought:
207. Microsoft undertook two basic approaches to eliminating the potential competitive threat posed by Java. First, Microsoft, recognizing that Netscape's browser was the primary distribution method for Java, sought to eliminate Java by eliminating Netscape's browser as a viable alternative. For example:
208. Second, Microsoft took actions to impede the cross-platform potential of Java by developing an interface called J/Direct. Any application that uses "J/Direct will run only on the Microsoft virtual machine." The default way of writing applications and applets for Microsoft's virtual machine causes some of those applications and applets not to be able to run properly on non-Windows platforms or even on non-Microsoft virtual machines running on Windows (James Gosling 9/10/98 Decl. ¶16). Microsoft's intent is revealed by its documents. For example:
209. Microsoft did not seek to "kill cross-platform Java" merely by developing its own version of Java and marketing it on the merits in competition with cross-platform Java. Instead, Microsoft used its monopoly powers over PC operating systems to induce and require industry participants to accept J-Direct (and IE which included J-Direct) instead of Java (and Netscape's browser which included Java).
210. Microsoft's conduct has prevented browser competitors from effectively competing on the merits for new business, artificially raised barriers to entry into both the browser and the operating system markets, and preserved Microsoft's operating system monopoly. Microsoft's conduct also threatens to monopolize the browser market.
211. As Microsoft and others have recognized, the vast majority of browser users tend to stay with the browser they receive with their PC if there is one or, if not, the browser provided by their ISP. For example, as Kumar Mehta wrote to Brad Chase and Yusuf Mehdi on March 27, 1997, "80% of those who do not use IE say they have no plans to switch to it." Mr. Mehta goes on to note that this "means that if we take away IE from the o/s, most nav users will never switch to us." (3/27/97 re "ie data," K. Mehta to Brad Chase et al.: Pl. Ex. 204.)
212. By ensuring that virtually all new users receive Microsoft's browser either with their PC or from their ISP or both, Microsoft effectively excludes Netscape and other browser competitors from the market, limiting them to a declining base of existing users.
213. As discussed above, Microsoft recognized that it would not be able to compete successfully against Netscape on the merits of IE alone. (2/24/97 Christian Wildfeuer to Adam Taylor et al.: Pl. Ex. 202, p. MS7 004346.) This was in part because while no company is perfect, and while Netscape (like Microsoft) made mistakes, Microsoft recognized the strengths of Netscape's product offerings. For example:
214. Microsoft's response was to exclude Netscape and other browser competitors from what Microsoft considers to be the two most important channels of distribution OEMs and ISPs. OEMs and ISPs are critical to browser distribution because many users get their browser from one or the other and because few users switch from one browser to another unless they buy a new computer or switch ISPs.
215. Microsoft has succeeded in effectively excluding Netscape almost completely from the personal computer OEM distribution channel one of the most important channels of browser distribution. (Cameron Myhrvold 4/24/98 Dep. Tr. 43- 44.) OEMs that license Windows are required to take (and not remove) IE and for most OEMs, including the largest, that means including only IE with the PCs they ship.
216. Another important browser distribution channel is through ISPs (including OLSs). (Cameron Myhrvold 4/24/98 Dep. Tr. 43.) Here, Microsoft's restrictive agreements with AOL and CompuServe alone have tied up ISPs/OLSs with 65 percent of the subscribers to ISPs/OLSs considered to be in the "Top 80" by Microsoft at year-end 1997 ("Netscape Competitive Analysis": Pl. Ex. 835, pp. MS98 0112834-36). More than 95 percent of subscribers to ISPs in the "Top 80" subscribe to ISPs that were contractually required to distribute IE preferentially. (See Pl. Ex. 12.) In this context, Microsoft's claim that there is little foreclosure because only eleven ISPs out of thousands are in the Internet Connection Wizard clearly mischaracterizes the extent of foreclosure. (Richard Schmalensee 9/4/98 Expert Report p. 23, fn. 79.)
217. Microsoft has asserted its anti-competitive practices do not result in foreclosure because users can download browsers for free from the Internet. The most important point to remember here, though, is that users prefer to get their browsers installed on their computers because consumers pay in terms of time and trouble to download a browser from the Internet. Browsers have become so large that they can take up to two hours to download over a typical user's modem, and the attempt to download can often be unsuccessful.
218. For example, as Michael Homer, Netscape's Executive Vice President and NetCenter General Manager, explained in his deposition, a download may be interrupted, or, even if the download is successful, it still may not be feasible to install the software. (Michael Homer 8/4/98 Dep. Tr. 54-55.) Moreover, according to Homer, "...the installation process can be confusing and difficult unless the users are fairly skilled users." (Homer Tr. 51-52.)
219. Thus, users are likely to settle for the browser that is already on their operating system. In fact, Microsoft's own studies show that most Internet users have never downloaded a browser. For example, Kumar Mehta of Microsoft reported to Brad Chase and Yusuf Mehdi in March 1997 that "Almost 60% of all surfers have never downloaded any software from the web. my sense is that these people are not very likely to download anything, let alone a browser that takes 2 hours to download, from the web." (3/27/97 re "ie data," Kumar Mehta to Brad Chase et al.: Pl. Ex. 204.) Carl Stork, Microsoft general manager of hardware evangelism and business development, summarized this tendency by saying, "Well, I prefer furniture that is already assembled over furniture I have to assemble myself." (Carl Stork 8/11/98 Dep. Tr. 125.)
220. What is important is not whether users can download a competitor's browser, but whether users will download a competitor's browser under prevailing market conditions.
221. Microsoft claims that competitors can distribute browsers effectively and that Netscape distributed or will have distributed by its partners 250 million to 270 million copies of its browser in 1997 and 1998. (Defendant Microsoft Corporation's Memorandum In Support of Its Motion For Summary Judgment, 8/10/98, 9-10.) However, Netscape does not itself intend to do any of this distribution, and according to Michael Homer, Netscape's distribution by CD-ROM has been "almost none." (Homer Tr. 70.) James Barksdale also testified that Netscape will not distribute any copies; its 10,000 "partners" will distribute 150-170 million copies. (James Barksdale 7/16/98 Dep.Tr. 91-93.)
222. Further, to the extent this distribution is to be by mail, it is a very inefficient distribution method. (James Barksdale 7/16/98 Dep. Tr. 227-228; see also Microsoft's Brad Silverberg's statement above quoted in paragraph 175.) Even though browsers on CD-ROMs distributed in the mail are free, customers must take time and trouble to install the software. Mr. Stork explained that using a CD-ROM is easier than downloading, but installing from a CD-ROM generates questions "...and has the potential for errors, especially if the customer has moved files... .loaded software on and futzed with" it. (Carl Stork 8/11/98 Dep. Tr. 42-43.) As a result, customers are unlikely to switch to another browser if they already have a browser that is up and running. Moreover, even to the extent that distribution by mail is a means of getting new browser users, it is a substantially more costly method. Relegating Netscape to such a method is an example of raising rival's costs.
223. Microsoft's foreclosure of Netscape and other browser competitors is shown by comparing Microsoft's share of browsers distributed by ISPs that have made IE their default browser with ISPs that have not made IE their default browser. (See Pl. Ex. 11.)
224. According to a Microsoft document, at year end 1997 Microsoft enjoyed a 94 percent weighted average share of browser shipments by ISPs who agreed to make IE their default browser, compared with a 14 percent weighted average share of browser shipments by ISPs who did not make IE their default browser. Microsoft's weighted average share of browser usage by subscribers to ISPs who made IE their default browser was over 60 percent; Microsoft's weighted average share of browser usage by subscribers to ISPs who did not make IE their default was less than 20 percent. If one is interested in the number of new browser users, then the fraction of existing subscribers accounted for by a given service is only a rough measure of the import of that service. However, as Microsoft's internal documents and analyses make clear, AOL and CompuServe are of great importance in acquiring new browser users.
225. The difference in IE usage across subscribers of different ISPs can also be analyzed by looking at IE's share of "hits" as reported by AdKnowledge, Inc., a company that develops and markets advertisement management products for the World Wide Web. AdKnowledge stores its clients' advertising on a server and delivers an advertisement when a particular page is accessed by a user. AdKnowledge keeps a record of the number of times a complete advertisement is sent from its server to a browser as well as the number of times a user clicks on a banner ad (each download or click counts as a "hit"). With every hit, AdKnowledge collects information about the user who accessed the page, including information about the user's browser and the domain from which the browser is surfing. The domain name can frequently be used to identify the ISP that the user uses to access the Internet.
226. A sample of AdKnowledge data was obtained in order to analyze how Microsoft's share of the browser market varies across ISPs, some of whom have entered into agreements to distribute IE preferentially, and some of whom have not entered into any such agreements. One data set reports all hits by users of all ISPs; these data are broken out by browser type. A second data set reports hits by users of particular ISPs; these data are broken out by ISP and browser type. Because millions of hits are logged every day, data collected on one day of each month (the second Wednesday) constitute the sample. In order to obtain data from AdKnowledge about a particular ISP, the domain name of the ISP must be known. A domain name is essentially the Internet address of a server. Domain names were determined using a search engine on an ISPs.com Web site (www.isps.com). While the AdKnowledge data are not as complete as one might wish,6 they show trends that are unmistakable.
227. Pl. Ex. 4 shows Microsoft's monthly share of browser usage by three categories of ISPs, from January 1997 through August 1998. The top line shows Microsoft's share of usage among subscribers to AOL and CompuServe rising sharply. Pl. Ex. 3 shows the same data as a three-month moving average. These companies (now merged) were chosen because they represent the largest ISPs (with a total of more than 11.5 million subscribers and about 65 percent of all subscribers to the services in the "Top 80" as of year-end 1997), and because AOL and CompuServe, as on-line service providers, were contractually restricted in their promotion and distribution of non-IE browsers to a greater extent than were most other ISPs. ("Netscape Competitive Analysis," Pl. Ex. 835, pp. MS98 0112834-36.) The middle line shows Microsoft's share for all ISPs. The bottom line shows Microsoft's share for the ISPs within the "Top 80" which Microsoft listed as having "IE Parity" (ISPs whose browser choice was not known to be contractually restricted), which had 10,000 or more subscribers, and for which data were available.
228. The effects are striking. Microsoft's share of "IE Parity" browser usage the category that is contractually neutral rises in twenty months from 20 percent to just under 30 percent. This rise includes the effects of technological improvement in IE. By contrast, the "All ISPs" line shows an increase in Microsoft's share from 20 percent to 49 percent. Finally, for AOL and CompuServe, Microsoft's share rose from just over 20 percent to over 87 percent. (It is worth noting that the dramatic jump in that share occurred before the introduction of IE4 in October 1997.)
229. The exclusion of Netscape and other browser competitors from the OEM channel has been even greater. Although several OEMs (including the largest, Compaq) have sought to replace IE with Netscape (see paragraph 147 above), none is now permitted to do so. And, as also discussed earlier (see paragraph 150), the fact that IE is required to be included means in most cases that only IE will be included.
230. Because of Netscape's innovations and success in creating and distributing the world's first widely used browser, Netscape initially had a very large share of the browser market (Pl. Ex. 5). Microsoft's browser share at the beginning of calendar year 1997 was approximately 20 percent (Pl. Ex. 5), and had been significantly lower earlier.
231. It is difficult to measure precisely what the changes in share have been in the last twenty months. This is true for three reasons. First, there is no universally accepted set of share statistics. Second, most of the statistics that do exist are browser usage shares that reflect the usage of all browsers whenever acquired; because of Netscape's large (and Microsoft's relatively small) share prior to 1997, present usage shares significantly understate Microsoft's share of current browser acquisitions. Third, usage shares are sometimes based on the number of browser users (in which case each browser used in the period measured is counted equally regardless of how often it is used in the period) and sometimes based on the number of times browsers are used in the period (in which case a browser is counted each time it is used).
232. Regardless, however, of how share is measured, it is clear that Microsoft's browser share has increased dramatically, and Netscape's browser share has fallen sharply, over the past two years. For example:
233. Microsoft's incremental share is even higher.7 For example:
234. Microsoft relied on its increasing browser market share, and the expected continued increase due to its practices, in trying to convince ICPs to abandon Netscape and agree to Microsoft's exclusivity provisions. For example, Microsoft, using forecasts from the Giga Information Group, told ICPs that its browser share had increased from 20 percent to 45 percent from 1996 to 1997, and would increase to 65 percent in 1998 and 75% in 1999. (Pl. Ex. 208; see also, Pl. Ex. 15.)
235. Thus there is a substantial probability that Microsoft's anti-competitive actions will permit Microsoft to retain its power over price in operating systems and will inhibit development of Microsoft-independent innovations. Both would harm consumer welfare.
236. Internet browsers that are capable of supporting applications that are operating-system independent are a threat to the Windows monopoly. If Microsoft minimizes competition from other browsers and chooses to support only Windows-based technology, Microsoft can maintain its monopoly power in operating systems with little threat of entry.
237. Microsoft's anti-competitive actions are aimed at hindering the success of non-IE browsers, but they are likely to send a message to all software developers: Microsoft will impede any innovation that threatens Microsoft's monopoly in operating systems. This will lessen developers' incentives to develop products that provide alternatives to the Windows platform. As a result, the range of software products consumers can choose from will be limited. Narrowed choice and slowed technological progress can never improve the welfare of consumers and are likely to decrease it. If Windows were truly a superior product, it would succeed on its merits. The actions Microsoft is taking will prevent that from being necessary.
238. Microsoft is using its monopoly power over PC operating systems to secure monopoly power over Internet browsers. While Microsoft has not yet succeeded in monopolizing browsers, Microsoft's browser market share grew significantly and Netscape's browser market share declined significantly from 1996 to the middle of 1998. As described in detail above, several sources indicate that Microsoft enjoys a browser market share of about 50 percent or more.
239. If Microsoft succeeds in acquiring a monopoly in Internet browsers, the monopoly will be protected by substantial barriers to entry. With ownership of the desktop, Microsoft can easily control the most common browser distribution channels, including distribution through OEMs and ISPs. Without an effective method of distribution, competitors' browsers pose little threat to IE. Moreover, natural barriers to entry would protect Microsoft's browser market share. Developers would tend to create Web sites that accommodate the dominant IE technology, which would increase users' demand for IE, generating a cycle that would reinforce IE's monopoly in the browser market.
240. As already discussed, Microsoft's monopoly in the market for Internet browsers would reinforce its monopoly over PC operating systems by preserving the barrier to entry created by network effects. Microsoft's dominance of the market for Internet browsers would also reinforce its monopoly over PC operating systems because a potential competitor in operating systems would need access to a compatible browser to be commercially viable. Thus, entry into the operating systems market would require either 1) entry into the browser market, where the entrant would face the network effects and other barriers to entry described above, or 2) the cooperation of Microsoft to make IE compatible with the competitor's operating system. If developers of competing operating systems did not have the open access to the IE technology that they would need to ensure compatibility, they would be at a constant disadvantage in providing a viable alternative to Windows in a timely fashion.
241. Microsoft has engaged in a number of anti-competitive actions. In particular, Microsoft's price for its browser, together with its other actions, is not profit-maximizing except for its effect of preserving Microsoft's operating system monopoly (and possibly gaining further monopoly profits by monopolizing the browser market and its ancillary revenues). Microsoft is providing its browser at zero price. Indeed, to the extent that Microsoft grants valuable concessions to firms to persuade them to promote IE, the price is effectively negative. Microsoft's browser pricing is thus another element of its anti-competitive behavior. Taken together with Microsoft's other actions, this pricing is anti-competitive to the extent that absent any deleterious effects on competition the pricing would not be profit-maximizing. Any foregone profits associated with its anti-competitive conduct can be recouped through the protection of Microsoft's operating system monopoly. Further, Microsoft may also gain from the reduction in competition in other markets. No later browser competitor will be able to enter to bid away such profits because of barriers to entry, including network effects and the threat that Microsoft can always again price its browser at zero.
242. If Microsoft's IE browser and Windows operating system are superior products, then competition will lead OEMs, ISPs, ICPs, and customers to choose them, and Microsoft need not artificially influence those choices. But Microsoft has engaged in anti-competitive conduct that has no compelling economic justification but for its effect of restricting competition. These actions will allow Microsoft to protect its monopoly in the market for operating systems and to establish a monopoly in the market for browsers. This situation can never make consumers better off than they would be with unfettered competition, and it is likely to make consumers worse off.
1 On the one hand, predatory pricing or predatory product design in a competitive market may be thought to benefit consumers by offering them lower prices (even if only temporarily) or an alternative product choice. On the other hand, where barriers to entry (or re-entry) exist, predatory pricing and product design can distort and reduce competition short of monopoly power. Moreover, the danger of misdiagnosis is a two-edged sword; if a finding of actual or prospective monopoly power is required to prohibit predatory behavior, there is the risk that a fact finder will understate the existence or risk of such power.
2 Phillip E. Areeda and Donald F. Turner. Predatory Pricing and Related Practices under Section 2 of the Sherman Act, 88 Harvard L. Review 697- 733 (1975). My analysis here concerns predatory anti-competitive acts generally. The Areeda-Turner article discusses predatory pricing, which is merely one type of predatory anti-competitive act.
4 The price is negative because Microsoft gives up valuable concessions such as space on the desktop (and the opportunity to earn money therefrom) in exchange for commitments to distribute its browser.
6 For example, not every ISP can be readily individually identified in the AdKnowledge data. The domain names, and thus the data for some of the ISPs could not be found. Further, hits by AOL subscribers are under-represented because of AOL's use of "caching," a device that makes repeated "hits" on a given page by the same or different AOL subscribers occur through AOL's own server rather than in a manner measured by AdKnowledge. In the way in which I have used the data, this phenomenon leads to an understatement of the effects of Microsoft's restrictive practices.
7 Incremental share of users is defined as the change in the number of IE users divided by the change in total users. Incremental share of browser usage is defined as the change in IE "hits" divided by the change in all "hits."