|
IN THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLUMBIA
UNITED STATES OF AMERICA,
Plaintiff,
v.
MICROSOFT CORPORATION,
Defendant.
STATE OF NEW YORK ex rel.
Attorney General DENNIS C. VACCO, et al.,
Plaintiffs,
v.
MICROSOFT CORPORATION,
Defendant.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Civil Action No. 98-1232 (TPJ)
Civil Action No. 98-1233 (TPJ)
|
DIRECT TESTIMONY OF
FREDERICK R. WARREN-BOULTON
TABLE OF CONTENTS
I. BACKGROUND................................................................................................................. 1
II. INTRODUCTION ................................................................................................................2
III. SUMMARY OF CONCLUSIONS......................................................................................6
IV. MICROSOFT HAS MONOPOLY POWER IN THE PC OPERATING SYSTEM
MARKET .............................................................................................................................9
A. PC Operating Systems Comprise a Relevant Antitrust Market.............................10
1. Background Industry Facts.........................................................................10
2. Principles of Market Definition.................................................................13
3. PC Operating Systems Comprise a Relevant Market................................16
B. Microsoft Possesses Monopoly Power In the PC Operating System Market........20
1. Market .......................................................................................................20
2. Barriers to Entering the PC Operating System Market are High...............21
3. Microsoft's Monopoly Power is Evidenced by its Use..............................28
4. in the Market Value of its Equity .............................................................28
V. INTERNET WEB BROWSERS POSE A THREAT TO MICROSOFT'S PC
OPERATING SYSTEM MONOPOLY ............................................................................29
A. Browsers and Operating Systems Comprise Separate Products............................32
1. Efficiently be Provided Separately.............................................................34
2. Internet Explorer and Windows 95/98 are Separate Products...................38
B. System Monopoly..................................................................................................39
VI. MICROSOFT'S EXCLUSIONARY CONDUCT.............................................................41
A. Microsoft's Tying is Exclusionary.........................................................................42
B. Microsoft's OLS and ISP Restrictions are Exclusionary.......................................46
1. Microsoft's OLS Agreements are Exclusionary........................................46
2. Microsoft's Restrictions on Other ISPs are Exclusionary..........................49
C. Microsoft's ICP Restrictions are Exclusionary......................................................53
D. Browsers ................................................................................................................ 56
E. Exclude Rival Browsers ........................................................................................60
VII. THE CUMULATIVE EFFECT OF MICROSOFT'S PRACTICES IS SIGNIFICANTLY
TO IMPAIR COMPETITION AMONG COMPETING BROWSERS.............................63
A. Anticompetitive Practices......................................................................................63
B. Anticompetitive Practices Have Substantially Foreclosed Competition...............64
C. Anticompetitive Practices Have Substantially Foreclosed Competition...............66
VIII. THERE IS A DANGEROUS PROBABILITY THAT MICROSOFT WILL GAIN
MONOPOLY POWER IN THE BROWSER MARKET; AND, EVEN IF IT DOES NOT,
IT COULD SIGNIFICANTLY REDUCE OR DELAY THE THREAT TO THE OS
MONOPOLY..................................................................................................................... 69
A. Dangerous Probability of Monopolization.............................................................69
B. System Monopoly May be Strengthened...............................................................70
IX. MICROSOFT'S EXCLUSIONARY CONDUCT LACKS LEGITIMATE BUSINESS
JUSTIFICATION ...............................................................................................................70
A. Microsoft's Prohibition on OEMs' Removing Internet Explorer is Unjustified...71
B. Windows Platform.................................................................................................74
C. Restrictions on OEM Modification of Desktop and StartUp Sequence.................78
1. User Experience.........................................................................................78
2. of the User's Windows Experience............................................................81
D. Achieve any Legitimate Business Purpose............................................................82
X. MICROSOFT'S ACTIONS TO INCREASE ITS BROWSER USAGE SHARE ARE
PREDATORY ....................................................................................................................83
XI. THE COSTS TO CONSUMERS AND COMPETITION FROM MICROSOFT'S
EXCLUSIONARY CONDUCT ARE SIGNIFICANT AND FAR REACHING..............86
.
Page 1.
DIRECT TESTIMONY OF
FREDERICK R. WARREN-BOULTON
I. BACKGROUND
1. My name is Frederick R. Warren-Boulton. I am a Principal with MiCRA
(Microeconomic Consulting and Research Associates, Inc.), a Washington-based economics
consulting and research firm specializing in antitrust and regulatory matters. I hold a B.A. degree
from Yale University, a Master of Public Affairs from the Woodrow Wilson School of Public
and International Affairs at Princeton University, and a Ph.D. in Economics from Princeton
University.
2. From 1972 to 1983, I was an Assistant and then Associate Professor of Economics
at Washington University in St. Louis. From 1983 to 1989, I served as the chief economist for
the Antitrust Division of the U.S. Department of Justice, first as the Director of its Economic
Policy Office and then as the Deputy Assistant Attorney General for Economic Analysis. Since
leaving the Department of Justice, I have served as a Resident Scholar at the American Enterprise
Institute, a Visiting Lecturer of Public and International Affairs at the Woodrow Wilson School
at Princeton University, and a Research Associate Professor of Psychology at The American
University.
3. My area of specialization is in the economics of industrial organization. I have
authored numerous publications, primarily in the application of industrial organization
economics to antitrust and regulatory issues, including a number of papers dealing with aspects
of the computer industry. A complete description of my background and papers can be found in
my Curriculum Vita, a copy of which is attached to this testimony as Attachment 1.
4. I have been asked by the plaintiffs in these actions to perform an economic
Page 2.
analysis of certain actions by Microsoft Corporation ("Microsoft") with regard to PC operating
system software and Internet browser software. In conducting that analysis, I have considered (1)
public documents containing relevant information; (2) certain confidential Microsoft and
third-party documents supplied to me by the plaintiffs; (3) deposition transcripts of personnel
from Microsoft and other industry participants; (4) relevant publications from the economics
literature; and (5) filings made in these actions, including the reports of other expert witnesses.
As additional material becomes available that is relevant to the testimony offered here, I will, to
the extent that is possible, seek to include it in my analysis.
II. INTRODUCTION
5. Based on my analysis of Microsoft's conduct and its competitive effects, I have
drawn the following three basic conclusions:
6. First, Microsoft has monopoly power in a relevant market. Using the basic
methodology for defining markets supplied by the 1992 Horizontal Merger Guidelines1 – an
accepted approach for delineating antitrust markets in Sherman Act cases – I conclude that
operating systems compatible with Intel x86/Pentium architecture personal computers (PCs)
comprise a relevant market.
7. The evidence also demonstrates that Microsoft possesses monopoly power in that
market. Microsoft for several years has enjoyed, and is projected for several years to retain, a
market share in excess of 90%.2 This market share is protected by substantial barriers to entry.
Page 3
Operating system markets exhibit "network effects." The greater the number of users a particular
operating system enjoys, the more likely it is that software developers will write applications for
that operating system. This, in turn, makes the operating system still more attractive to users and,
as a consequence, makes software developers more likely to develop applications for that
operating system, and, given scarce resources, less likely to develop applications for alternative
operating systems. The end result is that in such markets, the very network effects that drive
users toward the dominant firm make displacing that firm difficult.
8. Second, Microsoft has engaged in a number of practices that significantly impede
the commercial opportunities of rival producers of Internet web browsers, a product that is a key
element of a threat to Microsoft's operating system monopoly. Because of the nature of the
barriers to entry created by network effects, the most likely long-term threat to Microsoft's
monopoly power does not come directly from other operating systems, but rather from the spread
of cross-platform technologies, that can serve (like Microsoft's operating system) as a platform to
which application developers write.
9. Internet web browsers are a key component of precisely such a threat. PC users
demand browsers principally to locate, access, display, use, and navigate applications located on
the Internet's World Wide Web. At the same time, although browsers may never develop into
full-fledged operating systems, browsers can serve as a platform to which application developers
Page 4
write. Should application vendors use a browser platform other than the Windows platform, the
applications barrier to entry that protects Microsoft's monopoly could be diminished, and
competition in the PC operating system market created. Microsoft itself recognizes the threat
non-Microsoft web browsers pose. As Bill Gates stated: "They [Netscape] are pursuing a
multi-platform strategy where they move the key API into the client [browser] to commoditize
the underlying operating system."3
10. Subsequent to Netscape's launching of its browser, Microsoft implemented a
number of contractual restrictions in its licenses with Original Equipment Manufacturers
("OEMs"), Internet Service Providers ("ISPs"), Online Services ("OLSs") and Internet Content
Providers ("ICPs"). These restrictions bias consumers' choice of browser in favor of Microsoft's
browser, Internet Explorer ("IE"), disadvantage rival browsers, such as Netscape Navigator, and
restrain effective competition between them.
11. Microsoft's practices are exclusionary; that is, they impede the commercial
opportunities for rival web browsers. Microsoft's tying of Internet Explorer as a condition of
licensing its Windows operating system to OEMs and Microsoft's restrictions on OEMs' ability
to customize the Windows desktop and start-up sequence inhibit OEMs from offering or
promoting non-Microsoft browsers with their PCs. Microsoft's agreements with ISPs and OLSs
severely restrict (or, until recently, significantly restricted) the ability of these firms to distribute
or promote non-Microsoft browsers. And Microsoft's agreements with ICPs not only place
significant restrictions on the distribution and promotion of competing web browsers, but also
restrict the ability of ICPs to pay competing browser suppliers for promoting their services. The
restrictions, moreover, are mutually reinforcing and, when considered in the aggregate, make it
Page 5
significantly more difficult for Netscape and others to gain usage of their browser products.
12. Microsoft's restrictions have constricted, and continue to constrict the distribution
channels available to Netscape and other independently-supplied browsers. The majority of users
obtain a browser through OEMs, ISPs, or OLSs. Downloading of browsers from Internet sites is
increasingly time-consuming and fraught with technical difficulties. Although Netscape can
distribute disks containing its browser software to potential users through the mail and other
channels, this marketing strategy is not an effective substitute for the OEM, ISP, and OLS
channels.
13. The impairment of rivals caused by Microsoft's agreements matters; Microsoft's
constriction of distribution channels has significantly impaired the ability of Netscape and other
non-Microsoft browsers to effectively offer consumers a choice between their browser and
Microsoft's. Since 1996, when Microsoft imposed most of its restrictions, Netscape's overall
share of browser usage has fallen markedly. More importantly, in the wake of Microsoft's
exclusionary practices, Netscape's share of new browser users has declined even more.
According to Microsoft's estimate of the "run rates" for browsers – the percentage of new
Internet connections that use a particular browser – IE's current run rate is approximately 62%,
and is projected to rise to 70%.4
14. Third, Microsoft's practices cannot be justified on efficiency grounds.
Microsoft's ISP, OLS, and ICP restrictions are not necessary to further pro-competitive purposes.
Similarly unjustified is Microsoft's tying of Internet Explorer to OEMs. If users want the combination of operating system and IE provided by Windows 95 or Windows 98, they will
select the combination on their own volition. Microsoft's restrictive practices, however, prevent
Page 6.
operation of a "market test," in which OEMs would be free to offer the product configuration of
their choice and those product configurations would be tested against consumer preferences in
the market. Furthermore, Microsoft's restrictions on OEM modification of the Windows desktop
and start-up sequence are more restrictive than reasonably necessary to achieve any
pro-competitive goals.
15. The costs to consumers from Microsoft's maintenance of its PC operating system
monopoly, and its creation of a new monopoly in Internet browsers, through its exclusionary
practices would be substantial. The price paid for personal computer systems likely will be
higher than it otherwise would have been. Moreover, and perhaps more importantly, the artificial
barriers to entry erected by Microsoft's conduct will slow or halt the natural tendency of the
marketplace to provide alternative technologies in the affected markets, and Microsoft's
operating system monopoly would be further entrenched.
III. SUMMARY OF CONCLUSIONS
16. I have reached the following conclusions:
(1) First, operating systems ("OSs") for the x86 architecture PCs comprise a relevant
antitrust market within which Microsoft has, and has exercised, monopoly power.
(2) Second, Internet browsers and PC operating systems, including Microsoft's
Internet Explorer browser and Windows 95 and 98 operating systems, comprise separate
products. Products are separate in an economic sense when they would be provided separately
under competitive conditions; that is, when there is sufficient demand such that separate
provision is efficient. In competitive markets, we would observe products being provided
separately when that is efficient, and one could therefore ordinarily determine whether doing so
Page 7
is efficient simply by looking at what products are supplied in the market. Here, however,
Microsoft's use of its monopoly power to tie the browser to the operating system has prevented
such a "market test." Under the circumstances, therefore, it is appropriate to examine other
evidence to determine whether the requisite separate demand exists. And the available evidence
of demand for IE alone and for the Windows 95 and 98 OSs without IE, and of the costs of
providing them separately, supports the conclusion that browsers and operating systems are
separate products.
(3) Third, independently-developed browsers represent, as Microsoft itself has
recognized, a threat to the continued dominance of its Windows operating system product. This
threat arises from the ability of the browser to provide a user interface and platform for
applications so that, either by itself or in combination with cross-platform technologies such as
Java technologies, widespread use of independent browsers can facilitate the availability of a
substitute for the Windows platform and operating system. Thus, while browsers and operating
systems are complements in demand (e.g., in terms of features as experienced by users), the
widespread usage of non-Microsoft browser products can serve to facilitate the entry and
expansion of a substitute for Microsoft's operating system.5
(4) Fourth, Microsoft has acted to forestall this threat to its operating system
monopoly by imposing a set of restrictive agreements on, or taking advantage of restrictions in
existing agreements with, other market participants, including PC manufacturers, Internet service
Page 8
providers, on-line services providers and Internet content providers. Microsoft's doing so has
had the purpose and effect of biasing consumers' choice of browsers in favor of Microsoft's
browser, Internet Explorer; hindering the distribution of rival browsers, notably Netscape
Navigator; and injuring effective competition between them.
(5) Fifth, the cumulative effect of these restrictive agreements has been to make it
more difficult for competing browsers to acquire new users. These agreements, together with
Microsoft's tying of IE to its operating system, both in upgrade sales and in sales to OEMs, and
its pricing at zero to the installed base of Windows users, have increased IE's market share
significantly.
(6) Sixth, under current conditions, Microsoft's share of browser usage can be
expected to continue to increase rapidly. Should Microsoft's anticompetitive conduct continue,
there is a dangerous probability that Microsoft will monopolize the browser market.
(7) Seventh, Microsoft could significantly reduce or delay the threat to its operating
system monopoly even without fully monopolizing the browser market, a relevant antitrust
market. Preserving the barrier to entry created by the large stock of Windows applications
requires only that Microsoft's expected or actual browser usage share be high enough to induce
independent software developers ("ISVs") not to develop a stock of cross-platform applications
sufficient to encourage entry or expansion of competing operating systems.
(8) Eighth, the imposition and use by Microsoft of contractual restrictions on OEMs,
ISPs, OLSs, and ICPs to exclude competition are not justified by the need to achieve significant
efficiencies or are more restrictive than reasonably necessary to achieve such efficiencies.
(9) Ninth, based on the evidence I have seen, I conclude that Microsoft's distribution
of its browser at a zero price, the tying of its browser to its operating system, and its other
Page 9.
exclusionary contractual restrictions with ISPs, OLSs, ICPs, and OEMs were undertaken without
regard to whether those actions were profit-maximizing -- or even profitable -- absent the future
revenue gains from weakening rival browsers and thereby preserving Microsoft's Windows
operating system monopoly and from gaining a monopoly in the browser market. Microsoft
regarded the goal of winning the browser "war" as an overriding strategic objective, driven by the
need to preserve its PC operating system monopoly and was prepared to pursue this goal despite
the large costs it incurred.
(10) Tenth, if Microsoft is permitted to crush the incipient threat to its PC operating
system monopoly that independent browsers and cross-platform technologies pose, the adverse
consequences for competition and innovation are likely to be substantial. They include the
reinforcement of barriers to entry into the operating system market and the reduction of
incentives to innovate by other firms, especially new technologies whose independent provision
may be regarded as a substitute for, or threat to, the Windows operating system.
IV. MICROSOFT HAS MONOPOLY POWER IN THE PC OPERATING SYSTEM
MARKET
17. The starting point for an antitrust analysis of Microsoft's actions is the question:
Does Microsoft have monopoly power in a properly defined relevant market? In answering this
question, it is important to keep in mind that the purpose of defining markets in this case is to
determine whether Microsoft's conduct has the potential to harm consumer or social welfare. If
Microsoft lacks monopoly power in any relevant market, any profit-maximizing unilateral
conduct is unlikely significantly to harm consumer welfare.
18. My assessment of whether Microsoft possesses monopoly power proceeds in the
Page 10.
following steps. First, I will define the relevant market. I begin by setting out certain relevant
industry facts and defining some terms for the purpose of the ensuing discussion. I then briefly
summarize the principles of market definition I believe are appropriate here. Next, I apply those
principles to the facts, and explain how they lead me to the conclusion that PC operating
systems are an antitrust market. Second, having defined the market, I then assess Microsoft's
power within it. I show that Microsoft possesses monopoly power in the PC operating system
market, and that this power, in large measure, is secured by barriers to entry created by the large
number of applications available exclusivelyfor Microsoft operating systems.
19. I then explain how non-Microsoft browsers, both by themselves and in
conjunction with Java cross-platform technologies, threaten that monopoly. I will show that
browsers and Java threaten to eliminate the most significant of the entry barriers that shields
Microsoft's monopoly from effective competition and, therefore, that Microsoft stands to gain
significantly from eliminating the browser threat.
A. PC Operating Systems Comprise a Relevant Antitrust Market
1. Background Industry Facts
20. Personal computers are computers designed to be used by one person at a time.
They include desktop and laptop models. Because personal computers are actually computer
systems, they are made up of many components, each of which must be technically compatible
with the others for the system to function properly. A typical personal computer includes at
least one CPU ("Central Processing Unit"), dynamic memory, a hard disk drive, a floppy drive,
a keyboard and monitor, and an operating system. The operating system is "the software that
controls the allocation and usage of hardware resources such as memory, central processing unit
time, disk space, and peripheral devices. The operating system is the foundation on which
Page 11
applications are built."6
21. When desirable for clarity in this testimony, I distinguish the operating system
itself from the operating system product sold to consumers, which may include software or
features that are not part of the operating system. For example, Microsoft's Windows 95
operating system product includes a software Solitaire card game which is not part of the
Windows 95 operating system. Operating system software is software that can be part of the
operating system (e.g., either substitutable for part of the operating system, or closely related to
the functioning of the operating system – for example, a hard disk clean-up utility is closely
related to the operating system and would be operating system software). Operating system
software may be sold as a product that is separate from the operating system and produced by
vendors other than the OS vendor.
22. Applications are software programs designed to assist in the performance of a
specific task. They include such things as spreadsheets, word processors, and database
management programs. Applications are said to "run on top" of the operating system. In
particular, applications must communicate with the operating system to request services from
the operating system. Applications do so by using (or "calling") the operating system's
application programming interfaces ("APIs").
23. The components of PCs are assembled by computer makers or OEMs. The great
majority of operating systems installed on PCs are installed on new machines by OEMs.7 The
Page 12
OEM stage of the PC industry is competitive, as indicated by the large number of computer
makers, thin profit margins, and the absence of a dominant firm.8
24. Both businesses and households purchase PCs. Businesses and households have
different preferences and make different purchase decisions. Customers who purchase an OS as
an upgrade (e.g., the Windows 98 upgrade) to the operating system already on their PC have
different demand characteristics from customers who purchase an OS for a new PC either at
retail or pre-installed from an OEM.
25. IBM introduced the original PC in November, 1981 and offered Microsoft's MS-
DOS as one of its operating systems. Since then, Microsoft has become the leading supplier of
operating systems for PC OEMs. In the early 1990s, Microsoft began to enjoy widespread
acceptance of its "Windows" graphical user interface ("GUI") that runs on MS-DOS. Windows
and MS-DOS were often pre-installed on OEM PCs. In 1995, Microsoft introduced a successor
self-contained operating system product called "Windows 95."9 In June 1998, Microsoft
released Windows 98, the successor to Windows 95. As with Windows 95 prior to June 1998,
OEMs believe it is commercially necessary to offer Windows 98 on the PCs they sell to end
users.10
2. Principles of Market Definition
26. The first step in ascertaining whether a firm possesses monopoly power is to
Page 13.
define the market. For this task, I draw upon the principles for defining markets set forth in the
1992 Horizontal Merger Guidelines, promulgated jointly by the U.S. Department of Justice and
the Federal Trade Commission. The Guidelines supply a well-accepted method for delineating
markets and arecommonly relied upon by both economists and courts to define relevant
antitrust markets.11
27. The basic idea underlying the Guidelines is to find the smallest group of
products, and smallest geographic area, over which a monopolist in those products could
exercise market power by profitably imposing "at least ‘a small but significant and
nontransitory' increase in price."12 This test, known as the "hypothetical monopolist" test,
reflects the underlying rationale for defining markets: to assess whether conduct has the
potential to cause anticompetitive effects. If a "hypothetical monopolist" in the relevant product
could not effectively exercise monopoly power, then we can be confident that attempts by firms
operating in such a market to impose anticompetitive restraints could effectively be checked by
other forces, such as competing firms, entry by new firms, or a shift by consumers to a substitute
product. In contrast, if the "hypothetical monopolist" can exercise power over the product in
question, then a dominant producer of that product might well be able to inflict harm on
consumers through its conduct.
28. Just as it is important to define the market broadly enough to ensure that a
monopolist over all the products in that market would cause significant harm, it is also
important not to define the market too broadly, for that might understate the power of the firm
Page 14
whose conduct is being examined. Thus, the Guidelines' hypothetical monopolist inquiry
begins with a product (and geographic area) that the relevant firm produces and asks if the
hypothetical monopolist could exercise power over those products. If the answer is "no," the
market has not been drawn broadly enough; other forces would defeat the hypothetical
monopolist's attempt to exercise market power. In such circumstances, the next closest
substitute for the product considered is added, and the question of whether the hypothetical
monopolist could exercise power in this possible market asked again. This process is repeated
until the answer is "yes"; at that point, the market has been properly defined. Completing the
analysis at the earliest point at which the hypothetical monopolist could exercise power is
known as the "smallest market principle."
29. There is an important distinction I should mention about applying the Guidelines
to this case. As I have explained, the Guidelines inquire as to the profitability to a hypothetical
monopolist of raising price. In a merger case, the prices used as the starting point in the analysis
are prevailing or existing prices. This is because, in a merger case, the concern is not so much
whether the merging firms presently are exercising market power, but rather whether the merger
will increase the market power of the merging firms.
30. In contrast, in a monopolization case such as this, the question is whether the
firm in question already possesses monopoly power. In such circumstances, the appropriate
benchmark is not the prevailing price but rather the competitive price; and the question is
whether a hypothetical monopolist of the candidate market could profitably charge a price in
excess of the competitive level. Because the prevailing price might be a monopoly price, it is
not meaningful to ask whether a hypothetical monopolist could profitably increase the price
above the prevailing price; even a monopolist cannot charge greater than a monopoly price
Page 15
without having its conduct constrained sufficiently to make that price increase unprofitable.
Using the competitive price thus yields a proper market definition in the circumstances of this
case.
31. In applying the hypothetical monopolist test to possible markets (that is,
determining whether such a firm could profitably impose a "small but significant and
nontransitory" price increase over the competitive price), an economist ideally would like to
have reliable and precise estimates of what is known as the "own price elasticity" (considered
over the relevant price range) for the product or group of products that comprise the possible
market.13 The own price elasticity, together with the marginal cost, essentially tells an
economist when the imposition of a particular price increase will be profitable.
32. Such data, however, are often not available -- either to economists or to antitrust
tribunals.14 It is thus appropriate to rely, as the Guidelines explain, on "all relevant evidence"15
in applying the hypothetical monopolist test. This includes, but is not limited to, the perceptions
of market participants and the characteristics of the industry in question.
3. PC Operating Systems Comprise a Relevant Market
33. Applying the hypothetical monopolist test, it is my opinion that PC operating
systems comprise a relevant antitrust market. I reach this conclusion based on two distinct
Page 16.
types of evidence. First, the characteristics of the product in question, and the demand for it,
readily support the conclusion that operating systems designed for other (non-PC) hardware
platforms would not constrain, and have not constrained, the ability of a monopolist of PC
operating systems to exercise monopoly power. Second, evidence from OEMs, the major direct
purchasers of PC operating systems, confirms that even a large increase in the current price of
the dominant PC operating system, Windows, would not result in significant switching by them
or their consumers to other PC operating systems, let alone to operating systems designed for
other processors.
34. As an initial matter, it is important to observe that consumers do not demand
operating systems simply for the sake of having an operating system. Rather, consumers
demand computers, for which an operating system is one (albeit a key) component. In economic
terms, this means that demand for an operating system is a derived demand: demand for an
operating system is derived from demand for a computer system. Moreover, an operating
system is essential on every PC and consumers generally demand only one operating system per
personal computer.
35. A consequence of these facts -- that operating systems are demanded in fixed
proportions and that demand for operating systems derives from consumer demand for PCs -- is
that consumers faced with an increase in the price of PC operating systems can effectively
substitute away from PC-compatible operating systems only by substituting away from PCs.
Such substitution, however, would impose significant costs on end users and on
PC-manufacturers.
36. Both end users contemplating switching to alternatives to PCs and new users
considering such alternatives would face significant costs or disadvantages. End users purchase
Page 17
computers, not for the operating system itself, but rather to run applications; and applications
are typically designed for a particular operating system. Because the PC platform is dominant,
other platforms (and, as discussed below, non-Microsoft operating systems that also run on the
PC platform) typically have far fewer applications available. Thus, both new end users and
users considering switching to another platform would face the disadvantage of a smaller
portfolio of available applications. Switchers would also need to expend time and money
learning how to use a computer designed for a different processor. And both switchers and new
users would have to bear costs resulting from any incompatibility or impaired compatibility
between their computer and PCs used by colleagues or others with whom the users may wish to
communicate or share files.
37. Because there is generally one operating system on a PC, the increase in the price
of the PC that would result from a given percent increase in the price of the operating system
would be at most the increase in the price of the operating system times the share of the
operating system cost in the price of a PC. But the operating system for PCs accounts for only a
small share of the price of the PC – on average about 2.5% – and at most 10% for very
inexpensive PCs.16 Thus, even a 10% increase in the price of the OS would result at most in a
1% increase in the price of even inexpensive PCs. Given the cost to users of switching to
another platform, such a small increase in the price of the PC platform would not be expected to
result in a large reduction in the demand for PCs, and thus for PC operating systems. In
economic terms, the price elasticity of derived demand for PC operating systems must be very
low for at least a significant price range above the competitive price for PC operating systems,
leading me to conclude that PC operating systems are a separate market.
Page 18
38. The large costs that PC manufacturers and end users would incur in substituting
away from the PC platform reinforces this conclusion. OEMs make a significant investment in
developing the machines they sell; to substitute to another hardware platform (one not based on
the Intel-chip), such as the "PowerPC" chip, would require incurring significant costs.17
39. The testimony of, and documents authored by, OEM executives further supports
this conclusion. These executives explain that, if confronted with a 10% increase in their
Windows license, they would not switch to operating system products for other hardware
platforms.18 To the contrary, they make clear that preloading Microsoft's Windows operating
system is commercially necessary.19
40. OEM executives also explain that Intel-compatible operating system products
that are designed, not for personal computers, but rather to operate "servers" are not viable
substitutes for a desktop operating systems.20 Server operating systems are generally more
expensive yet do not provide the features consumers demand when they purchase PC operating
systems.21 Therefore, the existence of server operating systems would not constrain the ability
Page 19
of a monopolist of operating systems for PCs to exercise monopoly power. Accordingly, I
conclude that operating systems for Intel-compatible personal computers comprise a relevant
product market.
41. I should add that, even if the market were defined more broadly to include
operating system products for all personal computers – such as those offered by Apple and some
vendors of UNIX based operating systems that do not use an Intel-compatible microprocessor –
my conclusion that Microsoft possesses monopoly power in a relevant market would still stand.
B Microsoft Possesses Monopoly Power in the PC Operating System Market
42. Once the market is properly defined, the next step is to determine whether
Microsoft possesses monopoly power within it. Monopoly power is the ability of a firm
profitably to raise market price above the competitive level for an extended period of time or to
exclude competition.
43. There are four reasons why I believe that Microsoft possesses monopoly power
in the PC operating systems market. First, Microsoft's share of this market has been at a very
high level since at least the early 1990s and is expected to remain high. Second, barriers to
effective entry into the PC operating system market are high, and other PC operating systems
cannot easily increase their shares of that market because the huge stock of applications written
for Windows 95/98 will not run on those systems. Third, Microsoft has engaged in conduct that
would not be effective or profitable if it did not have monopoly power. Fourth, the pattern of
OS prices, Microsoft's margins, and the market value of Microsoft equity are themselves
consistent with Microsoft possessing monopoly power.
Page 20.
1. Microsoft has an Overwhelming Share of the PC Operating System
Market
44. The first step in determining whether a firm has monopoly power is usually to
determine the level and stability of its share of the relevant market. According to Microsoft's
figures, 80.8% out of the estimated 209.2 million PCs shipped worldwide since July of 1995
include a Microsoft operating system.22 During this period, naked PCs, e.g., PCs shipped
without any operating system at all, accounted for 31.5 million units, or 15% of the total PCs
shipped. In other words, of the PCs shipped with an operating system, Microsoft's share was
95.1%.23
45. This high market share has been remarkably stable. As shown in Pl. Ex. 1,
Microsoft's share of PCs shipped with an operating system has been above 90% since at least
the early 1990s and this dominance is forecast through at least 2001.
2. Barriers to Entering the PC Operating System Market are High
46. Microsoft's high market share has been protected by high -- indeed formidable --
barriers both to the entry of new PC operating systems and to the ability of rival PC operating
systems to acquire market share, even in response to prices above competitive levels. First, as I
will explain, PC operating systems are characterized by scale economies and sunk costs that
make the cost of entry high. Second, switching operating systems would impose significant
costs on users; this tends to "lock" users into Microsoft operating systems. Third, operating
systems exhibit network effects, a consequence of which is greatly to increase the costs of, and
Page 21.
reduce the probability of, a successful challenge to Microsoft's market dominance.
47. First, operating systems in particular, and software in general, are characterized
by economies of scale. The bulk of the costs are development costs -- the costs that must be
expended to create the software, irrespective of how many copies ultimately are sold. These
costs include, for example, writing, testing and debugging program code. The cost of producing
and marketing individual copies of the product ("the marginal costs") are, by comparison, quite
small.
48. The costs of developing an operating system to compete with Microsoft's
Windows operating system would be immense.24 For instance, IBM, a company that is very
experienced in developing competitive software, was reported to have spent a staggering
amount on the OS/2 project through 1996.25 Moreover, these cost are what in economic terms
are called "sunk" because only a small portion of either initial fixed development costs and any
subsequent negative cash flow could be recovered if the entrant were to exit the market.
Moreover, competition between two suppliers, each with very high fixed costs and very low
marginal costs, would likely result in a decrease in prices, further reducing the profitability of
entry to the would-be entrant. Entry into head-to-head operating system competition with
Microsoft thus would be time consuming, risky, and costly; profiting from such entry would be
at best very uncertain and long in coming.
49. A second barrier both to entry and to expansion by an existing competitor is that
users tend to become "locked in" to a particular operating systems. As discussed above, users
Page 22
are reluctant to switch from Windows to another operating system, even another PC operating
system, because to do so requires them to replace application software, to convert files, and to
learn how to operate the new software. Often, switching operating systems also means
replacing or modifying hardware. Businesses can face even greater switching costs, as they
must integrate PCs using the new operating systems and application software within their PC
networks and train their employees to use the new software. Accordingly, both personal and
corporate consumers are extremely reluctant to change PC operating systems. The software
"lock-in" phenomenon creates a barrier to entry for new PC operating systems to the extent that
consumers' estimate of the switching costs is large relative to the perceived incremental value of
the new operating system.
50. Additional switching costs arise from the fact that, for most users, operating
systems are only a means to an end – it is the application software that was designed to work
with the operating system that users want.26 Once they have purchased an operating system,
users are naturally reluctant to consider a different operating system. Unless their current
operating system product prevents them from using new applications or hardware, they are
likely to continue to use that operating system; for operating systems, unlike other goods, do not
wear out.
51. A third entry barrier is created by the well-understood fact that operating systems
are characterized by "network effects." Network effects occur when the value of an item to a
Page 23
user increases as the total number of users increases. Examples of products subject to network
effects include fax machines and both local and long distance telephone networks. A fax
machine is not particularly valuable until a large number of other people have fax machines that
use the same standards and protocols. Similarly, the more users within a given region that a
local phone network serves, the more valuable that network is to each user.
52. Operating systems are subject to network effects because, among other things,
the value of an operating system product is largely dependent on the number of applications
available for it. Today, applications written for one vendor's operating system product generally
will not work on another vendor's operating system. Thus, to write an application for more than
one operating system, a software developer generally must incur additional development costs,
and doing so could also divert scarce resources from the task of enhancing the application for
use with the first operating system. As a consequence, ISVs are often reluctant to "port" (or
convert) their software from one operating system to other operating systems.
53. As an operating system gains popularity, the incentive to develop software for
that operating system increases because the larger number of users for the operating system
product implies a greater potential market for software developers. The development of yet
more applications for that operating system, in turn, increases the value of the operating system
to end users who, as explained, purchase operating systems in significant part based upon the
quality and variety of applications available for it.27 As Hewlett-Packard's Frank Santos
explained, demand for an operating system is driven by the availability of "applications that run
Page 24
on the operating system."28 The operating system's market share, therefore, is likely to increase,
and that, in turn, is likely to cause software developers to devote yet more resources to writing
applications for that operating system product.
54. This phenomenon – known in economics as "positive feedback" – creates what is
best termed the "applications barrier to entry." Simply put, an operating system product can rise
to dominate the market, and once that dominance is achieved maintain it, because of both the
large number of complementary software applications available for it and the flow of new
applications that are written to it. This too is well-recognized by Microsoft. As observed by Dr.
Nathan Myhrvold, chief technology officer of Microsoft, "the laws of positive feedback govern
any system where compatibility with other users is either directly or indirectly a key factor in the
utility of a product or service," and these laws "ha[ve] been responsible for the phenomenal
strength of leading software products in both applications and operating systems [products]."29
Microsoft has further stated: "The availability of a rich variety of quality applications software
that will run on a particular operating system [product] is fundamental to its success. This fact
has been recognized by publishers of operating systems [products] for years."30 This was
reiterated by Microsoft's Brad Chase, who explained: "speaking very generally, a developer's
interested in building applications when there's a lot of users who could run those applications
on the platform."31
Page 25
55. The applications barrier to entry is supplemented by other barriers to entry that
derive from network effects. Books, publications, training, user groups, and news groups for the
incumbent operating system product provide a large sense of community for its users. Users can
exchange files, and perhaps more readily use their computers to communicate, with other
members of the group. Finally, when the incumbent operating system is installed at work, it
leads users to select the same operating system product for use at home.
56. It is clear that the applications barrier to entry sustains Microsoft's dominance,
critically contributes to its monopoly power, and helps explains why other Intel-compatible
operating systems, such as OS/2 and Linux, have persistently small market shares. As I
previously explained, because of economies to scale, the marginal costs of producing additional
copies of these operating systems is very low. Thus, one might expect that, if Microsoft
attempted to exercise monopoly power, vendors of these operating systems would flood the
market with their product and constrain Microsoft's behavior. This, however, has not occurred.
No rival has succeeded in mounting a sustained effective threat to Microsoft's market
dominance.
57. One reason for the lack of success is that alternative operating systems lack the
installed base of applications that Microsoft's PC operating system products enjoy. To offer a
product that a significant number of consumers wish to have installed on their PCs, vendors of
these operating systems would have to create, or induce others to create, an extensive set of
compatible software applications. This would be not merely expensive, but also very risky
because it would involve significant sunk costs, as explained above.
58. The failed attempt of IBM to enter the PC operating system market with its OS/2
operating system illustrates the strength of the applications barrier to entry and confirms that
Page 26
Microsoft's dominant share of the operating system market is indicative of monopoly power.
IBM's OS/2 operating system, first released in 1987, was designed to replace Microsoft's DOS
and DOS with Windows. In 1994 and 1995, IBM, which also competes as an OEM, installed
OS/2 on its own Aptiva line of computers. Retailers, and consumers, however, routinely
demanded application software that was not available for OS/2.32 Applications written for
Windows 3.x could also run on OS/2, but not the increasing number of applications written for
Windows 95. In response, IBM abandoned its practice of installing OS/2 on the Aptiva
computer, choosing instead to license Windows from Microsoft. As Mr. Kozel of IBM
explained, "the market . . . moved to the Windows 95 platform," so IBM "ship[ped] a hundred
percent of [its] Aptiva products with Windows 95."33
59. Microsoft itself recognizes that the failure of a new entrant, such as OS/2, to
attract customers is attributable to the "applications barrier to entry." As Joachim Kempin,
Microsoft's Senior Vice President for OEM Sales, explained in discussing what might "derail"
Microsoft's pricing strategy in January 1997:
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 could 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 [sic] to change
their computing platform is the real questions [sic]. 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 [sic] business.34
Page 27
3. Microsoft's Monopoly Power is Evidenced by its Use
60. Microsoft has engaged in conduct that it could not profitably pursue unless
it possessed monopoly power. For instance, when one OEM removed the IE icon from the
Windows 95 desktop, Microsoft responded by threatening to terminate that OEMs'
Windows 95 license. The OEM capitulated to Microsoft's demands.35 It is plain it did so
because, as OEMs universally explain, a Windows license is essential to remaining
competitive in the OEM market.36 This capitulation is itself evidence of Microsoft's
monopoly power.
4. Margins, and in the Market Value of Its Equity
61. Although accurate historical data on Microsoft's operating system product
license fees are not readily available, it is my understanding that since at least 1987 the
operating system has accounted for a steadily increasing share of the cost of a PC. An
internal Microsoft document acknowledges that it has increased its operating system product
"prices over the last ten years [while] other components' prices [for PC computers] have
come down and continue to come down. This is particularly true of CPU prices."37
Page 28.
62. Microsoft's monopoly power in operating systems has translated into
extraordinarily high net profit margins that have been increasing over time.38 Even more
telling is Microsoft's extraordinarily high market capitalization. With a price/earnings ratio
more than double the S&P 500 average, the financial markets are signaling very optimistic
investor expectations regarding Microsoft's future growth in earnings.
63. My analysis thus far has shown that Microsoft possesses monopoly power
in the desktop operating system market. Although the barriers to entry and to the growth
of other PC operating systems that protect Microsoft's monopoly power are formidable, they
are not impenetrable. There is no reason to believe that the market, if left to function
properly, will not in time generate alternatives to Microsoft's operating system that will be
sufficiently superior to overcome the entry barrier advantage that Microsoft enjoys.
64. Microsoft, however, has interfered with these market forces. As I explain
below, Microsoft has engaged in a course of conduct with the apparent purpose, and
evident effect, of reinforcing its monopoly power in the PC operating system market.
V. INTERNET WEB BROWSERS POSE A THREAT TO MICROSOFT'S PC
OPERATING SYSTEM MONOPOLY
65. Given the natural barriers to entry described above, a competitive threat to
Microsoft's operating system monopoly is less likely to come from other operating system
products than from extensions to complements of Windows that also can serve as platforms
Page 29.
to which ISVs write applications programs. Although a PC operating system cannot
successfully compete against Microsoft's operating systems without first overcoming
formidable barriers to entry, the situation is different for a product (e.g., browsers or Java
technology) that is both initially a complement from an end user perspective and a potential
substitute for the Windows 95/98 platform to which applications developers can write.
Because applications written to such a complement are compatible with Windows, their
developers can sell their applications to users of the Windows operating system.
Eventually, a sufficient number of such applications may become available to support an
alternative platform to Windows. This alternative platform can then be combined with other
operating systems to offer a complete and viable substitute for Windows.
66. To be sure, if the technology is to realize its full potential as a substitute for
Windows, it would eventually need to become an attractive platform for many applications.
However, because such a technology can gain wide marketplace acceptance based on its
value as a complement to Windows, the hurdle of attracting application developers is
substantially reduced. The wide dissemination of the complement among PC end users
means that application developers can reach a broader base of potential customers by
writing to it than by writing to an operating system that competes directly with Windows
95/98 and starts with very low market penetration and installed base.
67. As more applications are written that work with the new platform, it will
become more attractive to users and ISVs as a substitute for Windows. Ultimately, the
choice between the Windows platform and the alternative platform may be made by users
on the basis of the price and other characteristics of those platforms, rather than on the
availability of compatible applications.
Page 30
68. Internet browsers are complements that pose precisely this sort of threat to
Microsoft's operating system monopoly. A browser is software that enables computer users
to navigate and view content on the World Wide Web.39 Today the typical browser product
includes additional related software such as an e-mail program, a web-authoring tool and
a news group reader. Browser products support sophisticated security/encryption and may
include the ability to run Java programs.40 Competition has resulted in the porting of the
most popular browsers to a variety of desktop platforms.41 Competition among browsers
for PCs, principally between Microsoft's Internet Explorer and Netscape's Navigator and
Communicator products,42 has benefitted consumers significantly.43
69. Internet browsers may serve as an alternative software development platform
by exposing APIs that other software developers may use to perform particular functions.
Page 31
An API is essentially a method pre-defined by one software developer that enables other
software developers to access functions provided by the original software developer's
product. A platform for the development of software is a sufficiently rich set of APIs that
ISVs can use them to supply applications. Some browsers expose APIs that enable ISVs
to create "plug-ins" that expand the capability of the browser, independently access the
functions of the browser itself, and, through the Java Virtual Machine that comes with the
browser, provide other useful applications such as word processors and spreadsheets.44
A. Browsers and Operating Systems Comprise Separate Products
70. An issue of central concern in this case is whether IE is a "separate product"
from the Windows 95 and 98 operating systems. Microsoft contends that IE is an integrated
feature or "technology" of Windows 95/98 that cannot be "removed" without significantly
degrading the performance or value of Windows 98. Therefore, according to Microsoft, IE
and the Windows 95/98 operating system are a single integrated product.
71. The appropriate economic definition of a separate product is an item for
which there is sufficient demand such that it is efficient to offer that item separately from
other items. This test is a "demand based" or "market" test. Thus, for example, the
Windows operating system and IE are separate products if there is sufficient demand for one
without the other such that competing firms (including possibly Microsoft itself) could
profitably supply a product consisting of the Windows operating system but, from the user's
perspective, without IE. The "market" test is an appropriate separate product test because
it ascertains what product configurations would be supplied in a market where no firm could
Page 32.
exercise monopoly power.45
72. Applying the market test when firms producing the items in question lack
monopoly power is straightforward. In the absence of monopoly power, the product
configurations marketed by firms can be assumed to be efficient and comport with
consumer demand.
73. When, however, a firm has monopoly power over one product, that firm may
have an anticompetitive motive to tie the monopolized product to a complementary product.
One circumstance under which anticompetitive tying is especially likely to occur is when
the complementary product is also a "partial substitute" for the monopolized product.46
74. When a firm possessing monopoly power in a product employs it to force
customers to take another product, it prevents the market from determining whether the
components of the combined package – in this case, Windows 95/98 and IE – are one
product or two. It then becomes necessary to look to other evidence to determine whether
the components are one product or two. That evidence would include both evidence of user
demand for one without the other and of the cost of providing them separately.
75. As the following discussion explains, the available evidence strongly supports
the conclusion that IE and the Windows 95/98 operating systems are separate products. IE,
like other browsers, has been, and continues to be, offered as a separate product, and users
regularly acquire IE and other browsers as separate products. The evidence also shows that
there is substantial demand for Windows 95 and Windows 98 without IE and that it would
Page 33
inexpensive for Microsoft to provide Windows 95 and Windows 98 in a way that would
satisfy this consumer demand.
1. be Provided Separately
76. Internet browsers are currently provided separately from operating systems.
Netscape Navigator for several different operating system products is offered as a stand-
alone product through the retail channel, through the download channel, and supplied with
access services by some ISPs. Similarly, Opera, which has limited presence in some
distribution channels, is distributed independently of an operating system product. Mosaic,
the first browser with a graphical user interface ("GUI") is still available as a stand alone
product. Indeed, there are a large number of lesser known browsers still distributed
independently from any operating system.
77. Similarly, Microsoft has provided Internet Explorer separately from its
Windows operating system in the past and continues to do so to this day. Moreover, as
Microsoft executives have explained, the company's decision to produce an "unintegrated"
version of Internet Explorer for non-Microsoft operating systems, including a version of
Internet Explorer for the Apple Macintosh and Sun Solaris operating systems, indicates that
there is a demand for Internet Explorer separate from those operating systems and that it is
efficient to provide the browser separately in order to meet that demand. For instance,
Microsoft Vice President Brad Chase testified:
Q. Going back to IE4. Why did Microsoft develop a version of IE4 for those
other operating systems -- the Mac, Solaris, Win 3.1?
A. We developed business for other operating systems because customers
requested it and for some customers they didn't want to roll out Internet
Page 34.
Explorer unless they had cross-platform versions and because for certain of
the -- and because that also impacts developers.47
78. Microsoft tracks the market usage share of its browser separately from
Windows. As Brad Silverberg, then-Senior Vice-President of Microsoft's Applications and
Internet Group testified:
Q. Within Microsoft, after Windows 95 was released, did the company track the
share of usage of IE?
A. Yes, it did.
Q. And it tracked that separately from usage of Windows?
A. Yes, it did.
Q. Why? Why did you track share of IE?
A. See how -- compare IE share with competitive products, competitive technologies.
Q. And what competitive technologies were you comparing IE usage to?
A. Navigator, Cyberdog, others.48
79. Consistent with Microsoft's marketing of Internet Explorer to meet a demand
separate from that for an operating system, many OS vendors that do not view the browser
as a threat and that distribute a browser application with their OS view the browser as a
separate product.49 At one time Apple distributed Netscape Navigator as its default
Page 35
browsers for the Mac OS, and Apple currently distributes IE as its default browser.50 In
either case, the browsers were identified as separate products with their own brand names.
Most UNIX vendors have also distributed browsers with their operating system products.51
These browsers are always identified as separate products with separate brand names. Most
important, OEMs that license operating systems from these vendors are not required to
pre-install browsers on their desktop computers, indicating that operating systems and
browsers are separate products.52
80. There is also evidence of separate demand for operating systems generally,
and for Windows 95 and Windows 98 in particular. First, many corporate users purchase
only an operating system and do not want a browser at all. For instance, Joseph J. Kanicki,
Jr., Strategic Commodity Manager, Dell Computer Corporation, stated in a recent
declaration:
Some business and government customers prefer not to have
Internet Explorer preinstalled on their computer because: (1) the
customer may have its own software or software standards which do not
include the latest version of Internet Explorer, (2) the customer may
wish to install a competitive browser instead of Internet Explorer, or (3)
the customer may wish to prevent its employees from accessing or
attempting to access the Internet or the World-Wide Web.53
81. Mal Ransom of Packard Bell expressed the same point in a recent deposition.
Mr. Ransom stated: "In the commercial end of the business depending upon what part . . .
Page 36
you're selling to, on some days all they want is the operating system. They don't want other
things. They don't want any access to the Internet because companies don't want their
employees potentially being able to get on the Internet or play games."54
82. Second, whether in response to perceived corporate demand or for other
reasons, several OEMs requested that Microsoft permit them to remove an end-user's visible
means of accessing IE from both Window 95 and Windows 98.55
83. Microsoft responded to this corporate demand for operating systems without
IE by designing Windows 95 so that Internet Explorer could be "uninstalled" using the
Add/Remove programs utility that comes with the Windows 95 operating system. This
utility removes the ability of end users to use Internet Explorer to browse the web (web
browsing functionality) and, thus, from an economic perspective, unties the browser from
the operating system. Confirming that the "uninstall" capability for Internet Explorer was
created in response to consumer demand for a browserless operating system product, David
Cole, the Microsoft Vice President responsible for overseeing the development of Internet
Explorer, testified that the capability was added in response to "feedback from corporate
customers that wanted to prevent access to the Internet, so that when they . . . buy a new
machine from a PC manufacturer they want the ability to remove easy access to the Internet
so their employees, you know, aren't spending their time out on the Web doing whatever."56
Page 37
84. Microsoft's creation of the "uninstall" utility that, from an economic
perspective, removes IE, shows that it was efficient for Microsoft to meet the demand for an
operating system without a browser in the case of Windows 95. Indeed, having designed the
utility, it would have been virtually costless for Microsoft to permit OEMs, in effect, to use
the utility to remove IE.
85. It is my understanding that Windows 98 does not contain the same ability as
Windows 95 to "uninstall" Internet Explorer. Nonetheless, I understand that Professor Felten
has shown that it is easy to remove from Windows 98 the ability to browse the Web using
Internet Explorer and to substitute a different browser in its place.57 It would thus be
efficient for Microsoft to provide, or to permit OEMs to provide, the operating system
without the ready means of invoking IE.
2. Internet Explorer and Windows 95/98 are Separate Products
86. Because the evidence shows that there is separate demand for IE and the
Windows 95 and Windows 98 operating systems (e.g., demand for each without the other)
and because the evidence shows that it would be inexpensive for Microsoft to provide both
those products separately, I conclude that it would be efficient for Microsoft to meet that
separate demand. In a competitive market, therefore, I would expect that OEMs would have
the option of licensing or distributing Windows 95 and Windows 98 without the visible
means of access for IE, just as OEMs have that option from PC operating system vendors
Page 38.
that do not have monopoly power.58 I therefore conclude that IE and Windows 95/98 are
separate products.
B. Operating System Monopoly
87. As I explained above, Internet browsers produced by firms other than
Microsoft, by themselves and in combination with the Java technologies they distribute, pose
a threat to Microsoft's operating system. Microsoft internal documents show that the firm
understood this threat. As noted above, Bill Gates wrote in May 1995 that "[Netscape is]
pursuing a multi-platform strategy where [it] move[s] the key API into the client [browser]
to commoditize the underlying operating system."59 Similarly, Microsoft Vice President
Brad Chase warned that competing Internet browsers could eventually "obsolete
Windows,"60 and Group Vice President Paul Maritz worried that Netscape's browser might
make Windows "replaceable."61 As Brad Silverberg, another Microsoft Executive,
succinctly put it: "the Internet Battle" is "not about browsers. Our competitors are trying to
create an alternative platform to Windows."62 More recently, Microsoft's James Allchin
explained that, in his view, "[t]he goal, the stated goal of Netscape was" among other things
Page 39.
"to create a new set of APIs that developers would write to."63 When asked how Netscape
could threaten Windows, Mr. Allchin stated: "You get developers to write the APIs, you
cover up Windows, you've just got this layer running on top, and if the size [and]
performance was acceptable, it becomes irrelevant. Windows becomes irrelevant."64
88. Because competing browsers, such as Netscape Navigator, pose a threat to
Microsoft's operating system monopoly, Microsoft protects the profits it earns in the OS
market by dominating the browser market. By reducing the market share of competing
browsers to low levels, Microsoft could significantly diminish the possibility that
applications developers will write to those browsers' APIs. Microsoft's browser dominance
also would impede the distribution of a cross-platform Java technologies. Microsoft,
therefore, has a significant incentive to gain a large share of the browser market in order to
protect its desktop operating system monopoly.
89. Browsers, like operating systems, exhibit not only economies of scale, but also
network effects. Websites can be written to standards that favor one browser over another.
For instance, websites can use technologies that are accessible only by a particular browser
or work better with that browser. If Microsoft were to gain a dominant share of the browser
market, it might succeed in inducing website developers to write their content using
Microsoft-specific technologies. If a large number of websites are written to such a
technology, more end users would switch to IE, which in turn would increase the incentives
of website developers to embrace Microsoft-specific technology. The consequence of this
instance of "positive feedback" is that the browser market could tip to a Microsoft
Page 40
monopoly, in which the installed base of Microsoft specific web sites, along with switching
costs, create barriers to entry. Microsoft, of course, would benefit from such a monopoly
because it would mark the death knell of the threat posed by non-Microsoft browsers to its
operating system monopoly.
VI. MICROSOFT'S EXCLUSIONARY CONDUCT
90. The analysis thus far established has led me to two conclusions: First, that
Microsoft possesses monopoly power in the market for PC operating systems and, second,
that Internet browsers marketed by firms other than Microsoft pose a threat to Microsoft's
monopoly and that hindering the success of rival browsers would thus serve to protect and
extend that monopoly. I now turn to an examination of Microsoft's practices with respect
to browsers. For ease of exposition, I will examine each component of Microsoft's conduct
separately. However, the economic significance of any particular conduct cannot be
understood unless set in its proper context. Accordingly, I consider the effect of particular
practices in light of Microsoft's other conduct and its dominance in the PC operating system
market. My principal conclusion is that Microsoft's practices, taken as a whole, are
anticompetitive and are likely to facilitate monopolization in the browser market and the
preservation of Microsoft's monopoly in the PC operating system market.
A. Microsoft's Tying is Exclusionary
91. Microsoft has engaged in a number of practices with respect to the
distribution of its Windows operating system to OEMs that, taken together, have the effect
of excluding competing Internet browsers. First, although Microsoft did not bundle IE with
Page 41.
the initial retail version of Windows 95, it subsequently bundled Internet Explorer with its
Windows operating system for licensing to OEMs. By bundling, I mean that Microsoft
provides the browser with the operating system in a single "package" at no additional cost.
Second, Microsoft has tied the Internet Explorer browser to the operating system in both
Windows 95 and Windows 98. By tying, I mean that Microsoft requires OEMs to install
IE on, and prohibits them from removing IE or access to IE from, the PCs they sell. The
combined effect of these practices is exclusionary. They ensure that OEMs will pre-install
rival browsers on fewer machines.
92. It is true that Microsoft does not contractually prohibit OEMs from
pre-installing a competing browser, or even pre-installing a competing browser as the
default browser. But Microsoft's tying of IE to the Windows 95 and Windows 98 operating
systems has made it more costly and burdensome for OEMs to install other browsers and
has thus significantly, although not completely, deterred OEMs from doing so.
93. Pre-installing a second browser imposes significant costs on OEMs and
yields them few benefits when the second browser is not perceived to be of significantly
higher quality. Even if the OEM is not required to pay a license fee in order to install a
second browser, the costs of including a second browser along with IE may be significant.
They include increases in support costs that result from customer confusion and increased
testing costs.65
Page 42.
94. In particular, OEMs report that customers become confused by the presence
of two or more applications that perform similar functions and by desktop screens populated
with multiple icons.66 Because Microsoft requires OEMs to bear all support costs
associated with the PCs they sell, OEMs have a keen interest in reducing customer
confusion, which can lead to increased support costs. For this reason, OEMs pre-install
only a limited number of software programs with the PCs they ship and do not cover the
desktop screen with icons.67 Microsoft also recognizes that multiple products performing
similar functions may cause consumer confusion. For instance, Microsoft's OEM account
manager for Gateway, Gail McClain, explained in a recent deposition that "redundan[t]
icons on the desktop" may "be confusing to end users."68
95. Thus, Jim Von Holle of Gateway testified that Gateway sought permission
to remove the IE icon from Windows 98 because it sought to lessen support costs that would
result from also installing Navigator.69 Von Holle explained that "general usability studies
. . . indicate that the less cluttered the desktop . . . the less confusing it is for the customer to use
the product."70
Page 43
96. The value to an OEM from adding a product, such as Netscape Navigator,
that performs the same functions as Internet Explorer is small compared to the increased
costs imposed on OEMs and the resulting customer confusion. If OEMs were permitted
either not to install IE or to remove the visible means of access to it, their costs of including
a rival browser would be lower, and the benefits to consumers from having such a browser
would be greater.
97. Microsoft executives both in depositions and in its documents confirm that
the requirement that OEMs install IE, combined with the contractual prohibition on OEMs'
removing any part of that browser, can significantly deter OEMs from installing competing
browsers. Microsoft's Senior Vice President of OEM Sales, Joachim Kempin, has testified:
Q. ....[D]oes Microsoft sometimes essentially make this argument to [OEMs]; Why
do you need to incur the extra testing costs and the extra user education and maybe
undergo the longer loading time --
A. I believe we have.
Q. Is that sometimes successful in persuading OEMs that they don't really need to
distribute another browser because they already have Internet Explorer?
A. That is sometimes successful.71
98. The foreclosure of the Netscape browser as a result of Microsoft's tying of
its browser to its Windows operating system is significant. The OEM channel is one of the
two principal channels through which users obtain browsers.72 Moreover, because of
Page 44
reluctance to install new software themselves, many less sophisticated PC users simply use
whatever browser comes pre-installed on their machines. For example, one marketing study
explains that the response "`[i]t came with my computer' is the #1 reason people switch to
Internet Explorer" and concludes that "OEMs are the best vehicle to gain browser share."73
99. Contemporaneous Microsoft internal documents confirm this analysis. They
show that Microsoft believed that its forced licensing of Internet Explorer to OEMs would
serve to deny rivals access to the OEM channel and, by doing so, increase Internet
Explorer's market share at the expense of other browsers. For instance, Microsoft's Jim
Allchin wrote in January 1997: "I do not feel we are going to win on our current path. We
are not leveraging Windows from a marketing perspective . . . . We do not use our strength
-- which is that we have an installed base of Windows and we have a strong OEM shipment
channel for Windows . . . . I am convinced we have to use Windows -- this is the one thing
[Netscape] do[es]n't have."74 Christian Wildfeuer echoed this a month later. She wrote:
"It seems clear that it will be very hard to increase browser market share on the merits of
IE 4 alone. It will be more important to leverage the OS asset to make people use IE instead
of Navigator."75
Page 45
B. Microsoft's OLS and ISP Restrictions are Exclusionary
100. The second set of restrictions I examined were Microsoft's agreements with
Internet Service Providers, which are firms that provide PC users with access to the Internet.
Some important ISPs, such as America Online (AOL) also provide proprietary content to
users: these ISPs are known as Online Service Providers. Along with OEMs, OLSs and
ISPs comprise the two most significant channels through which users obtain browsers.76
As with Microsoft's restrictive OEM agreements, the terms of its contracts with ISPs
exclude competing browsers and significantly impede competition on the merits among
competing browsers.
1. Microsoft's OLS Agreements are Exclusionary
101. As I will explain in detail in connection with Microsoft's screen restrictions,
the Windows 95/98 desktop and boot-up sequence provide an attractive advertising vehicle
for OLSs. Placement on Windows screens is valuable to OLSs because, among other
reasons, it ensures that the OLS reaches many potential new subscribers at the precise time
when those new subscribers must open an account to secure access to the Internet. As
Kevin Knott, an executive of CompuServe, Inc., testified, placement in the Online Services
Folder was important to CompuServe because "Windows 95 reaches such a large number
of people and is on such a large percentage of new computers, it represents a very large
distribution opportunity for us . . . . No other single hardware [or] software company can
give it that level of distribution . . . . There are other ways to try to do that, but it would
require separate deals and probably at much greater expense to try to match what we could
achieve through Windows 95.77
Page 46.
102. Microsoft used this asset to induce OLSs to enter into agreements that
restricted the distribution and promotion of competing browsers. It did so by creating for
the Windows 95 desktop the "Online Service Folder," in which it featured participating
OLSs, and placed icons for them in the Windows "Start" menu. Because of screen
restrictions, which I describe below, OEMs could not remove these items or advertise other
OLSs in a more prominent way. Microsoft, the evidence shows, was aware of the value
that OLSs attached to favorable desktop placement, and sought to exchange this valuable
asset for exclusionary restrictions. For example, Brad Silverberg, former head of
Microsoft's Internet Group, told AT&T during negotiations:
You want to be part of the Windows box [desktop], you're going to have to
do something very 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.78
103. In exchange for favorable desktop placement, the OLSs typically agreed to
the following:
(a) To promote and distribute Internet Explorer as the "exclusive"
or primary browser;
(b) not to distribute a non-Microsoft browser unless specifically
requested by the customer or "express or imply" that a
competing browser is available;
(c) not to ship non-Microsoft browsers more than 15% of the time,
even upon customer request for alternative browsers; and
Page 47
(d) to restrict the ability of users to employ or download
non-Microsoft browsers.79
104. These agreements explicitly (although not completely) exclude competing
browsers from an important channel for acquiring new users and maintaining existing users.
And, as I will explain below, this exclusion matters. A substantial number of users obtain their browser through OLSs such as America Online (AOL). Indeed, recognizing the
importance of their agreements with OLSs to maintaining IE's overall market share,
Microsoft has not relaxed those restrictions in Windows 98.
105. It is ordinarily not anticompetitive for a firm, such as Microsoft, that has
created a valuable asset, like desktop real estate, to charge customers for the use of that
asset. My concern with Microsoft's conduct is not that it has extracted compensation for
Windows desktop real estate. Rather, my concern involves the form or nature of the
payment Microsoft has extracted; namely, agreements that raise the distribution cost of
competing browsers and tend to exclude those browsers from the market.
2. Microsoft's Restrictions on Other ISPs are Exclusionary
106. In addition to creating the Online Service Folder in Windows 95, Microsoft
developed a feature called the "Internet Connection Wizard" (ICW). In Windows 95, the
ICW was prominently displayed on the Windows 95 desktop and OEMs were not permitted
to remove it. If selected by a user, the ICW would display a list of ISPs generated by the
Microsoft Internet Referral Server. As with the Online Service folder, ISPs viewed the ICW
as an attractive way of promoting their service and acquiring new subscribers. As discussed
Page 48
below, Microsoft took advantage of the screen restrictions on OEMs to capture the full
value for itself of being the first to recommend ISPs to new users.
107. As with the Online Service Folder, Microsoft exchanged this desktop
placement in the ICW for exclusionary agreements. Indeed, Microsoft's Cameron
Myhrvold testified that Microsoft created the Internet Connection Wizard in part to induce
ISPs to grant preferences for Microsoft's browser.80 In order to secure placement in the
ICW, ISPs agreed to restrictions similar to those Microsoft imposed on OLSs. The ISPs
typically agreed to:
(a)make Internet Explorer the default or preferred browser;
(b)ship a rival browser no more than 15% of the time;
(c)not express or imply that a competing browser is available;
(d)limit the ways in which a user could link to a site promoting, or
download, a competing browser; and
(e) in exchange for discounts, employ technologies that made the service
function better with Internet Explorer than with rival browsers.81
108. The economic effects of the first four restrictions are essentially the same as
the restrictions imposed on OLSs. The provision regarding use of Microsoft-specific
technologies, however, requires separate discussion. The effect of this provision is to reward
ISPs that configure their services in a way that reduces the cross-platform threat to
Microsoft's operating system monopoly. The reason is that ISP use of Microsoft-specific
technologies reinforces the dominance of the Windows platform. For instance, one of the
Microsoft-specific technologies is known as "ActiveX." As I understand it, ActiveX is "a
Page 49.
set of technologies . . . built on Microsoft's Component Object Model (COM) . . . that
enables software components to interact with one another in a networked environment,
regardless of the language in which the components were created."82 ActiveX controls are
"reusable software components that incorporate ActiveX technology [and] can be used to add
specialized functions . . . to Web pages, desktop applications, and software development
tools."83 The crucial feature of ActiveX for my purposes is that it is operating system
(typically Windows) specific. Use of such technologies by ISPs serves to blunt the
cross-platform threat that, as explained above, rival browsers might pose.
109. I understand that, faced with the threat of this litigation, Microsoft last Spring
relaxed its ISP restrictions on certain ISPs, but not the most important ones -- OLSs'. The
only remaining requirement is that ISPs promote Internet Explorer on a par with
non-Microsoft browsers. For two reasons, however, the ISP restrictions nevertheless are
important to my analysis. First, the remaining restriction is itself exclusionary. Some ISPs
will choose to promote or support only a single browser at any given time in order to reduce
its costs.84 For such ISPs, a requirement of "parity" for Internet Explorer in order to secure
access to the ICW may amount to a defacto requirement that the ISP exclusively support
Page 50
Internet Explorer. Second, and perhaps more important, Microsoft's ISP restrictions had
significant anticompetitive effects while they were in place, and if Microsoft's statements
that some ISPs will support only one browser are accurate, those effects cannot be reversed
simply by removing the restrictions. Third, absent intervention, Microsoft would be free to
reimpose all the ISP restrictions.
110. In addition to entering agreements with ISPs that involved the Internet
Connection Wizard, Microsoft by late 1996 had entered into more than 2,500 "IE preferred"
licenses.85 For fifteen of the top seventy-five providers, these agreements required that a
certain percentage of browsers shipped be Internet Explorer.86 Moreover, more that fifty of
the top seventy-five providers had agreed to make Internet Explorer the "preferred" or
"default" browser.87 As with Microsoft's other agreements with OLSs and ISPs, these
agreements serve to bias browser distribution toward Internet Explorer in ways that may not
reflect consumer demand and impaired distribution of other browsers.
111. Testimony of ISP executives confirms that the restraints at issue restricted
their ability to distribute competing browsers that users might prefer. Kevin Knott of
CompuServe testified that CompuServe agreed to the restrictions even though "we prefer to
have flexibility in software that we use."88 Stephen von Rump of MCI testified that "there
are certainly users out there that prefer browsers and e-mail clients that are not Microsoft.
Page 51
And our ability to reach them and entice them to sign up for our service is presumably
enhanced by the ability to promote and distribute those."89 As discussed above, ISPs have
costs that may induce them to support only one browser, and some may choose to do so; but
the choice of which browser to support is one that ISPs try to retain. It is in their interest to
respond to consumer preferences in determining which browser to ship at any point in time.
112. Indeed, internal Microsoft documents confirm that ISPs, reflecting consumer
demand for competing browsers, resisted agreeing to Microsoft's restrictions. In a series of
E-mail messages, Cameron Myhrvold, who had a key role in dealing with the ISPs and OLSs
on this issue, noted that "ISPs have to swear allegiance to IE for typically 75% of browsers they distribute in order to get into the referral server." (emphasis in original) 90 But
Myhrvold emphasized that ISPs, for their own business reasons, resisted Microsoft's efforts
to bias their browser distribution in favor of IE: "ISPs are agnostic on the browser. It's
against their nature to favor a browser or even a platform. This has been damn hard for us
to influence... I have had a hard time guiding the ISPs to IE loyalty even when I made them
sign explicit terms and conditions in a legal contract."91
C. Microsoft's ICP Restrictions are Exclusionary
113. Microsoft developed for Internet Explorer 4 a feature known as the "Active
Desktop." The Active Desktop, if enabled, overlays the standard Windows desktop with
content that makes the desktop resemble a web page. Microsoft also created a number of
"channels" on the Active Desktop. These channels display the content of Microsoft-selected
Internet Content Providers.
Page 52
114. Because the IE 4 Active Desktop was anticipated to be shipped on a
substantial number of PCs, the channels -- just like Microsoft's On-Line Service Folder and
Internet Connection Wizard -- provided ICPs with an attractive way of promoting their
service. For instance, Wadsworth of Disney explained that "entering in to a promotional deal
with Microsoft was highly valuable because of Microsoft's ability to create icons or
‘channels' that would be located on the Windows ‘desktop.'"92 Another ICP opined that "a
preferred position on the active desktop . . . is of almost incalculable value."93
115. And, just as with its agreements with ISPs and OLS, Microsoft conditioned
ICPs' right to placement on the desktop on their agreement to exclusionary terms. In
exchange for placement on the top level of channels, those immediately visible to end users
on the Active Desktop, ICPs agreed:
(a) not to promote or advertise any "Other Browser" product;
(b) not to pay compensation in any manner to the producer of an "Other
Browser";
(c) to use Internet Explorer "and no Other Browser" as part of any client the
ICP develops for the Windows or Macintosh operating systems;
(d) to implement certain Internet Explorer-specific and Windows-specific
technology in their web sites (such as ActiveX), the effect of which is to
make the web site best viewable with Internet Explorer.94
116. These restrictions impede the commercial opportunities of independent
Page 53.
browsers in several ways. First, they directly inhibit the promotion and dissemination of
non-Microsoft browsers in ways similar to the ISP restrictions. The economic effect is to
prevent Netscape and other rivals from effectively competing with Microsoft in promoting
their browser.
117. Second, ICPs are also prohibited from paying compensation to "Other
Browsers" (a term restricted to the top two non-Microsoft browsers by market share95). This
prohibition inhibits the continued development of Netscape's browser by depriving Netscape
of important ICP partners and revenues from promoting those ICPs.
118. Both of these effects are illustrated by Netscape's experience with Intuit.
Intuit has made clear that, absent its ICP agreement with Microsoft, it would have promoted
Netscape on its web site and would have "entered into an agreement with Netscape to
provide financial content on Netscape's Web Sites."96 Indeed, because Intuit also is an ISV,
the effect of its agreement with Microsoft was to force Intuit to abandon distributing
Netscape's browser with its popular Quicken software.97
119. Third, by conditioning access to the Windows desktop on ICPs agreeing to use
Windows-specific technologies in their web sites, Microsoft biases consumers' choice of
browser toward Internet Explorer and away from browsers that do not support their
Microsoft-controlled technologies. This is because the writing of web sites to IE- and
Windows-specific standards reduces consumer demand for other browsers.
120. In April 1998, Microsoft relaxed most of these exclusionary restrictions.
Page 54
Nonetheless, in my judgment, these restrictions, in combination with Microsoft's other
conduct, substantially contributed to the impairment of rivalry among competing browsers --
an impairment that is not eliminated by the relaxation of the restrictions. The ICPs subject
to Microsoft's agreements included some of the most popular content providers, such as
Disney, Time Warner, and Intuit. Their Microsoft-induced endorsement of both the IE
browser and IE-specific standards could be expected to significantly influence browser
adoption. Moreover, Microsoft documents shows that its personnel expected these restrictions
to have an exclusionary effect. For example, one document explains that "Windows
distribution -- for IE, and as a draw for partners" provided a "competitive lever" over
Netscape.98
D. Browsers
121. When a personal computer on which Windows 95 or Windows 98 has been
pre-installed is turned on for the first time, the user is presented with a series of screens and
menus that appear prior to the Windows desktop. This series of screens, known as the
"start-up" or "boot-up" screens or sequence, is a user's first point of "contact" with his PC.
Purchasers of new PCs generally will view the start-up sequence before they see the
Windows desktop. Software products (such as browsers) and services (such as ISPs)
promoted in the start-up sequence will be displayed to those end-users before they can
consider any competing software products or services the OEM might have pre-installed in
the Windows desktop. And, because of the very high share of new PCs that come with
Page 55
Windows pre-installed, this will be true for a correspondingly high share of purchasers of
new PCs.
122. For these reasons, the start-up sequence is a particularly effective vehicle for
promoting software or services, especially for firms interested in promoting their products to
first-time PC purchasers. With respect to browsers, it is apparent that the ability to reach new
users first offers an important advantage in acquiring browser market share. For example,
before Microsoft made the decision partially to lift its restrictions and permit OEMs to place
OEM-selected ISPs in the Inter Connection Wizard, Brad Chase wrote: "In order to protect
our position on the desktop and increase the likelihood that Internet Explorer gets the
prominent position with the end user, we should move the [Internet Connection Wizard] into
the boot up sequence."99 Mr. Chase went on to explain that moving the ICW into the boot
up sequence would "increase[] the likelihood that an end user gets the option to sign up for
solutions that promote IE before they get into the desktop or any customized shells that feature
other browser solutions."100
123. The principal reason why it is advantageous to have a "first-to-market"
position in the boot up sequence is simple: It is apparently quite difficult to induce new
browser users to switch to a competing browser. This is confirmed by Microsoft's own
documents, which conclude that "it is very hard and expensive to make people switch
[browsers]."101 The same conclusion emerges from market research conducted by Microsoft.
For example, a study of "Switcher Intentions" found that of approximately 6 million users
Page 56.
who were aware of Internet Explorer but not using it, "81% will not switch to IE."102 A
February 10, 1998 "Platforms-Desktop 3 Year Business Outlook" discussing browser share,
noting that "many customers see MS and NS [Netscape] as parity products, no strong reason
to switch," described as "critical success factors" in Microsoft's attempts to increase its
browser share "ISP, OEM, corp deployment, customer retention."103 In short, as one
Microsoft document explains, "since only 30% of Internet users have ever downloaded a new
browser (they use what comes with their pc or comes with their ISP sign up kit), the only real
chance IE has of getting them to switch is thru a new pc, an OS upgrade or a new ISP kit."104
124. Since the initial beta releases of Windows 95, Microsoft, in its contracts and
OEM pre-installation kits (OPKs), has restricted the ability of OEMs to customize the
Windows start-up sequence. These restrictions were strengthened in Microsoft's later
Windows 95 licenses.105 The two most important restrictions are:
OEMs cannot modify the "sequence or appearance of any screens"
displayed during the initial start-up sequence; that is, until the Windows
desktop screen has been displayed, and
OEMs cannot configure their PCs to run automatically any program
(including a program that would add screens at the end of the start-up
sequence or create an alternative desktop or "shell" for the end-user) at the
completion of the initial end-user boot.106
Page 57
125. hese restrictions have significant exclusionary effects. First, these restrictions
raise the cost of promoting and distributing non-Microsoft Internet browsers. Absent the
restrictions, OEMs could change the start-up sequence (or provide an alternative desktop
shell) and thereby differentiate and promote non-Microsoft Internet browser offerings at lower
cost, and thus increase the value to OEMs of carrying a non-Microsoft browser. Microsoft's
restrictions on OEMs' ability to customize the Windows 95 desktop similarly inhibit OEMs
from distributing or promoting other browsers on the desktop itself. Together, these
restrictions substantially reduce both the OEMs' incentive to offer browser choice and the
effectiveness of any such offering.
126. Second, because these restrictions mean that Microsoft alone controls access
to the most valuable placement – the start-up sequence – they create for Microsoft a unique
asset, which Microsoft bartered in exchange for anticompetitive, exclusionary agreements
with ISPs and OLSs. For instance, by prohibiting OEMs from inserting advertisements for
ISPs and OLSs in the boot-up process, Microsoft was able to increase the value to On-Line
Services (such as AOL) of placement in the On-Line Services folder, and to ISPs of placement
in the Internet Connection Wizard (which has been moved in Windows 98 into the start-up
sequence).
127. At the same time, however, these restrictions exacerbated the exclusionary
effect of Microsoft's agreements with ISPs and OLSs. If OEMs had been permitted to replace
Microsoft's ICW with their own ISP referral screens, and thus recommend ISPs to new PC
owners first, they might have promoted ISPs that did not have preferred or exclusive contracts
with Microsoft. Indeed, Microsoft's Cameron Myhrvold explained that he "definitely" was
concerned that "if OEMs were allowed to select the ISPs," those OEMs might "select ISPs
Page 58
that did not support IE as the primary browser."107 Further, these ISPs might induce the PC
owner to use a non-Microsoft browser, especially if that ISP received special inducements
from alternative browser producers. The OEM screen restrictions thus served to reinforce
other Microsoft-created distribution barriers to Netscape.
128. Microsoft has recently relaxed certain start-up and desktop restrictions. But
its doing so neither suggests that the restrictions were not anticompetitive nor undoes the
anticompetitive effects they have had to date – and, as I explain below, the effects have been
substantial. Microsoft has permitted six OEMs to insert their own Internet Connection
Wizard into the Windows 98 start-up sequence.108 These OEMs, and certain additional ones,
also have been allowed to include in Microsoft's Internet Connection Wizard the Internet
access providers of the OEMs' choice in place of those selected by Microsoft.109 But
Microsoft has still not permitted OEMs to promote third party browser brands in the start-up
sequence, nor has it allowed OEMs to configure a PC to boot automatically into an alternative
desktop screen (including one supplied by a competing browser.110
E. Desktop Exclude Rival Browsers
129. As a condition of licensing both Windows 95 and Windows 98, Microsoft
compelled OEMs to agree to a number of restrictions on the ways in which OEMs can
Page 59
customize the Windows desktop. The restrictions provided, among other things, that:
- OEMs may not delete any icons or folders from the Windows desktop, and
- OEMs may add icons or folders to the desktop only if such icons and
folders are the same size and substantially the similar shape as the icons
and folders included on the Windows 95 desktop by Microsoft.111
130. Microsoft's refusal to permit OEMs to delete icons or folders is exclusionary
for essentially the same reasons that Microsoft's refusal to permit OEMs to license and
distribute Windows without Internet Explorer is exclusionary. Because of Microsoft's license
agreements, OEMs must distribute Windows with the Internet Explorer icon on the desktop.
For the reasons explained above, the presence of that icon on the desktop both raises the costs
to OEMs of shipping their PCs with another browser and reduces the value to them of
including such a browser. Thus, the restrictions substantially inhibit OEMs from preinstalling
non-Microsoft browsers on the Windows desktop.
131. The requirement that folders and icons added to the desktop must be the same
size and substantially the same shape increases the exclusionary effect of the prohibition on
removing the Internet Explorer icon. Differentiating the appearance of a second browser's
icon on the desktop might reduce the customer confusion that results from having two
browsers on the desktop. In turn, this might reduce the OEM's support costs from including
the second browser and thus create a greater incentive to preinstall a second browser.
132. Microsoft introduced its Internet Explorer 4.0 browser in October 1997. This
browser includes the Active Desktop feature. The restrictions on the depiction of icons is
Page 60.
somewhat different for OEMs that ship Windows 98 with the Active Desktop feature as the
default desktop. The idea of the Active Desktop, as I explained earlier, is to display the
Windows desktop using World Wide Web standards and protocols. Doing so gives users
greater freedom to modify their desktops and enables them to place hyperlinks on the desktop
giving them easy access to various locations on the web or their hard drive. In addition, the
Active Desktop is able to display ActiveX controls and Java applets, which cannot be
displayed on the standard Desktop. The Active Desktop, which is layered over the "standard"
Windows desktop, also contains a "Channel Bar" that enables the user to access a number
of "channels" which display the content of certain ICPs. The Active Desktop remains a
feature of Internet Explorer and is included in the Internet Explorer browser shipped with
Windows 98.
133. Microsoft allows OEMs to customize the Active Desktop to some degree.
OEMs can add background "wallpaper" to the Active Desktop. They also can add to the
Active Desktop "items," which serve a purpose similar to icons and folders on the standard
desktop, and are permitted to include on the Active Desktop items of different shapes and
sizes. But OEMs are not allowed to include an icon featuring any third party (non-OEM, non-
Microsoft) brand on the Active Desktop (including, for instance, another browser).112 The
purpose of this limited restriction is to prevent the promotion of non-Microsoft software; it
cannot be related to any desire on the part of Microsoft to maintain a consistent "look and
feel" for the Windows product. The restrictions are thus exclusionary in the same way as
Microsoft's restrictions on the standard desktop. They prevent OEMs from differentiating the
appearance of a second browser and, therefore, inhibit them from pre-installing a second browser.
Page 61
VII. THE CUMULATIVE EFFECT OF MICROSOFT'S PRACTICES IS
SIGNIFICANTLY TO IMPAIR COMPETITION AMONG COMPETING
BROWSERS
134. Microsoft has prevented its browser competitors from competing on the merits
in the browser market. Because most browser users stay with the browser that they receive
either through their PC purchase or through their ISP, Microsoft practices that inhibit
Netscape's ability to have OEMs install, or ISPs distribute, its browser restrict Netscape's
ability to compete in the browser market.
135. The exclusionary impact of Microsoft's practices has been significant. This
is demonstrated by (1) Microsoft's own prediction that its conduct would foreclose browser
competition; (2) empirical evidence that Microsoft's conduct blocked available browser
distribution channels; and (3) data showing that Microsoft's anticompetitive contractual
practices have resulted in substantial foreclosure of the browser market. I discuss each of
these types of evidence below.
A. From its Anticompetitive Practices
136. Microsoft expected to gain browser market share in the aftermath of its
anticompetitive practices. Using an internal forecasting model, Microsoft estimated browser
sales and shares based on a variety of market and user characteristics, including the sales and
retirement rates of desktop computers in the home, business, educational, and international sectors.113
Page 62
137. Microsoft's browser market share was 6 percent at the end of Fiscal Year 1996
and 31 percent in 1997. In February 1998, Microsoft used a model to project browser market
shares for fiscal years 1998 through 2001. It estimated that its IE market share would increase
to 48 percent in 1998 and then to 65 percent by the end of Fiscal Year 2001.114 Microsoft's
actual share of browser users through 1997 and its projections are shown in Pl. Ex. 14.
138. Control over the OEM and ISP channels was critical for Microsoft's gains in
browser user share. As Microsoft's Randy Haas explained in an email to Brad Chase
discussing the importance of various modes of IE distribution, "[a] critical success factor in
gaining browser share is continued focus on ISP's, OEM's[,] and corporate deployments to
target the growth of new users."115
B. Anticompetitive Practices Have Substantially Foreclosed Competition
139. The "market share" measure I have discussed to this point is a "stock." It
shows IE's share of the "active installed base" of users, some of whom are using a browser
they installed several years ago. For purposes of evaluating many of the consequences of
Microsoft's anticompetitive restraints, a "flow" based share of new users is the more
appropriate measure because it shows a particular browser's share of new installations, either
Page 63.
by people who are using any browser for the first time or by those replacing an old browser.
Unlike a stock measure, a flow measure does not measure browsers previously installed,
which of course reflects only the browsers' past success. The flow measure is what
economists ordinarily use to measure market share in antitrust analysis. (Thus, for example,
General Motors' share of the U.S. automobile market is well expressed by its share of new
car sales, as opposed to the percentage of GM cars on the road.) The flow measure is also a
better guide to the future of the installed base: if a flow share is higher than installed base
share, the installed base share will rise toward the flow share.
140. Because there are no direct measures of browser user "flow", I have calculated
a quarterly flow measure using (1) data describing the "active installed base" by browser type;
(2) information on the numbers of users of the Internet; and (3) an estimate of the percentage
of people who switch from one browser type to another. The resulting flow measures of
browser market share are shown in Pl. Ex. 261. The data that underlie the figure are given
in Pl. Ex. 337, along with a set of notes that explains the derivation of the numbers in the
table.
141. The flow measure of user market shares shows that Netscape's share of new
users has declined dramatically since the second quarter of 1997 and is far less than its current
48 percent share of the installed base. Similarly, IE's flow-based share has increased dramatically
over the same period, and is well above its stock-based share of the installed base.
142. Of course, product improvements and other actions by Netscape and by
Microsoft have also affected browser shares. For example, during the fourth quarter of 1997
when Internet Explorer 4.0 was released, Internet Explorer's flow market share grew to
Page 64.
approximately 50 percent while Netscape Navigator's market share fell to approximately 48
percent. However, Netscape's market share increased to approximately 52 percent and
Internet Explorer's fell to 45 percent in the first quarter of 1998, when Netscape started
distributing its browser for free. Internet Explorer's market share began to increase again and
reached approximately 60 percent in the next quarter, while Netscape's declined to
approximately 38 percent.
143. The flow measures of IE market share show that Microsoft's share of new
browser users is high and increasing. Further, they suggest the real possibility that Microsoft
will gain monopoly power in the browser market.
C. Anticompetitive Practices Have Substantially Foreclosed Competition
144. The effect of Microsoft's exclusionary practices can be seen by comparing IE's
share among ISPs and OLSs that made IE their default browser with its share among ISPs and
OLSs that did not agree to make IE the default browser.
145. Browser usage market shares can be measured by examining data showing
visits to Internet sites ("hits") by IE, Netscape, and other browsers. The data are reported by
AdKnowledge, a company that markets web advertising management services.
AdKnowledge manages a set of servers that deliver web advertisements to browser users
when they visit particular web sites. To aid firms in monitoring the effectiveness of their web
advertisements, AdKnowledge maintains a database for each day that logs information every
time a web advertisement is delivered to, or "clicked on" by, an end user.116 This log records
Page 65
information about the user's "domain name," operating system, and browser type. In certain
cases, the information about a user's domain name may be used to determine the user's ISP.
146. To evaluate differential browser usage overall, an AdKnowledge data sample
was selected from the vast daily log of more than 5 million "ads served."117 A browser's
usage share in this database is simply the percentage of all "ads served" or "clicked on" by
users of that browser, divided by the total number of all ads served or clicked on by users of
all browsers. Calculated in this manner, Microsoft's share of the browser market started at
twenty percent in January 1997 and grew to forty-nine percent by August 1998. Over the
same period, Netscape's market share fell from seventy-seven percent to forty-eight percent,
just below Microsoft's share. This is shown graphically in Pl. Ex. 4.
147. Because the AdKnowledge database contains information about not only the
user's type of browser but also the user's domain name, it is possible to estimate browser
market usage share by individual ISP. These data were obtained for those ISPs that were
considered by Microsoft to have more than 10,000 subscribers,118 and for which domain name
information was available. Information about the two largest OLSs, AOL and CompuServe,
also was collected.
148. Pl. Ex. 5 shows Microsoft's monthly browser usage share from January 1997
through August 1998. The top line in the graph represents the market share of Internet
Explorer for America Online and CompuServe (now one company), both of which signed
agreements with Microsoft that contractually limited their ability to distribute Netscape
Page 66.
Navigator.119 From January 1997 to August 1998, Internet Explorer's share of browser usage
among the customers of these two firms grew from approximately 22 percent (a share which
was almost identical to the share for all Internet users) to more than 87 percent. As illustrated
by the middle line, Internet Explorer's overall share for all ISPs had grown by far less during
that period to 49 percent.
149. In a document prepared by Microsoft captioned the "Netscape Competitive
Analysis", Microsoft identified 12 ISPs as being in neither the "Netscape Preferred" nor the
"IE Preferred" category and listed them as having "IE parity." The average market share for
IE and Navigator among users of ISPs listed as "IE parity" is shown as the bottom line of the
Exhibit. IE's market share among these users was approximately 20 percent in January 1997,
just as it was among other ISPs' customers. These ISPs were not contractually required to
favor Internet Explorer. By August 1998, Internet Explorer's market share among these ISPs'
customers had increased only ten percentage points, to approximately 30 percent. These ISPs
can reasonably be used, in effect, as a control group that shows what would have happened
absent Microsoft's exclusionary agreements.
150. The difference between the IE market usage shares for the IE parity group on
the one hand and AOL and CompuServe (both of which were subject to Microsoft's
exclusionary agreements), and the average of all unconstrained ISPs (the control group), on
the other hand, can be used to show the effect of Microsoft's exclusionary agreements. If no
ISPs had been party to exclusionary agreements, assuming no other changes, it is reasonable
to expect that IE's market share would be its share of customers of these unconstrained IPSs
– approximately 30 percent in August 1998. The differences between that share and IE's
Page 67
actual overall share – 49 percent – shows the impact on the market of the Microsoft ISP
agreements.
151. Differences between the IE parity and the other two groups cannot readily be
explained by factors other than contractual restrictions; improvements in the IE browser, for
example, could have been expected, other things the same, to affect IE's market share for all
three ISP aggregations shown in the figure. Moreover, differences between the IE parity
group and the others groups, if anything, understate the exclusionary impact of Microsoft's
practices because the IE parity group itself may have been affected by Microsoft's
exclusionary conduct. 120
VIII. THERE IS A DANGEROUS PROBABILITY THAT MICROSOFT
WILL GAIN MONOPOLY POWER IN THE BROWSER MARKET;
AND, EVEN IF IT DOES NOT, IT COULD SIGNIFICANTLY
REDUCE OR DELAY THE THREAT TO THE OS MONOPOLY
A. Dangerous Probability of Monopolization
152. Market shares in the browser market have changed dramatically in just two
years. As explained above, IE's share of all browsers installed has increased from 6 percent
in 1996 to 49 percent today, and Microsoft itself projects that its share will soon exceed 60
percent. Moreover, Microsoft's share of current installations – its current market share –
already exceeds 60 percent and is projected soon to exceed 70 percent.121 Under these
circumstances, I thus conclude that there is a dangerous probability that Microsoft will gain
monopoly power in the browser market.
Page 68
B. System Monopoly May be Strengthened
153. The foreclosing effects on the browser market of Microsoft's tying of its
Internet browser to its Windows operating system are likely to be substantial. First, in my
opinion, anticompetitive foreclosure in this case does not require that Netscape be wholly
unable to distribute its product or unable profitably to maintain indefinitely a significant share
of the browser market. Microsoft can achieve its anticompetitive goal of preserving its
Windows operating system monopoly simply by discouraging or preventing ISVs from
developing a stock of cross-platform applications sufficient to encourage the development of
an alternative platform and thus of competing operating systems. This will occur if
independent browser suppliers either exit the browser market or decide to support only
Windows-specific software development technologies. Indeed, even if Netscape remains in
the browser market and does not support Windows-specific technologies, foreclosure is
achieved if the market share of independent browsers that do not support Windows-specific
technologies is small enough (or if ISVs believe it will shortly become small enough) to
discourage ISVs from writing cross-platform applications. In principle, therefore, Microsoft
can foreclose competition in the operating system market by foreclosing Netscape from only
a small share of the browser market.
IX. MICROSOFT'S EXCLUSIONARY CONDUCT LACKS LEGITIMATE
BUSINESS JUSTIFICATION
154. Thus far in my analysis, I have established three propositions based on the
Page 69.
evidence. First, Microsoft possesses monopoly power protected by substantial barriers to
entry. Second, browsers pose a threat to Microsoft's monopoly power which Microsoft
recognized. Third, Microsoft has engaged in practices that substantially impair the
commercial opportunities of competing browsers and thereby serve to blunt this threat.
155. Now, I will examine whether Microsoft's exclusionary practices – tying IE to
Windows 98 in the OEM channel; imposing screen restrictions on OEMs; adopting ISP
exclusivity agreements and adopting ICP exclusivity agreements – have legitimate business
justifications. I will analyze these practices by asking three questions. First, are there
legitimate business purposes that could explain Microsoft's practices? Second, do the
exclusionary practices actually further those purposes? And third, could any such legitimate
business purpose of the restriction be achieved without restricting other parties' distribution
and promotion of competing browsers? I conclude below that Microsoft's practices either do
not further efficiency objectives or are not reasonably necessary to achieve them.
A. Microsoft's Prohibition on OEMs' Removing Internet Explorer is Unjustified
156. Microsoft has argued that it has designed and tied IE to its Windows 95 and
Windows 98 operating systems in order to provide users with superior browsing functionality
that could not have been achieved if Windows 95 and Windows 98 and IE were treated as
distinct products. That proposition, I understand, is disputed, and if Microsoft is wrong, then
of course its forcing OEMs to license and distribute IE would not serve any legitimate
purpose.
157. Even if it were true that what Microsoft calls the "integrated" design of
Windows 95 and 98 does further some legitimate business purpose, Microsoft would still not
Page 70.
be justified in prohibiting OEMs from removing the icon or other ready means of access to
IE.
158. As an initial matter, it should be noted that Microsoft's claim cannot be that
forcing users to browse the web using Internet Explorer is efficient because Internet Explorer
is superior to other browsers. Even if Internet Explorer is preferred by some users, it is not
preferred by all users. Consumer welfare is maximized when the market is responsive to
consumer demand, not when a firm with monopoly power over one product requires
purchasers also to take an unwanted product or makes it difficult or costly for them to obtain
a related product they desire. Therefore, Microsoft's argument must be that compelling
OEMs to license and redistribute IE is necessary in order to provide those users who want IE
with all the features they desire.
159. This concern, however, plainly cannot justify Microsoft's practices with
respect to Windows 95. Microsoft itself designed Internet Explorer to be easily removed from
Windows 95 without impairing any non-browser related feature of the Windows package. It
is doubtful Microsoft would have done this if uninstalling Internet Explorer caused ill
effects.122 My understanding is that Internet Explorer is "hard wired" into Windows 98 to a
greater degree than in Windows 95. However, it is my understanding that Dr. Felten's work
demonstrates that it is possible to remove from Windows 98 most means of invoking Internet
Explorer to browse the Web and to replace Internet Explorer with another browser in a way
that achieves a comparable user experience and does not denigrate any operating system
Page 71.
function.123 Thus, even in the case of Windows 98, Microsoft's restrictions are not reasonably
necessary to ensure that users have the option of the benefits Microsoft claims result from its
combination of Internet Explorer and Windows.
160. In any event, even if Microsoft's combination of Windows and Internet
Explorer achieves a superior browser experience, there is no reason to believe that Microsoft's
exclusionary practices are necessary to achieve that experience. If consumers prefer the
Microsoft-supplied package of Internet Explorer and Windows to Windows combined with
another browser, the bundled Microsoft product should thrive in the marketplace. OEMs
would chose to license and distribute both IE and Windows on their own volition. As
Microsoft has acknowledged, OEMs are in the business of satisfying their customers. They
are exceedingly unlikely to market a product that does not meet user demand. Indeed,
Microsoft's General Manager for Multinational OEM Operations, Bengt Akerlind, conceded
that OEM actions that reduce the value of the product they offer to consumers will come home
to roost in the form of diminishing that OEMs' business in the future.124
161. Microsoft also has argued that its requirement that Internet Explorer be
pre-installed on every Windows machine OEMs ship is necessary to preserve "the same initial
user experience" across different OEMs.125 For reasons I will explain in more detail later, this
Page 72
justification is undermined by Microsoft's own conduct. Microsoft permits OEMs to vary the
user's initial experience in substantial ways. By doing so, Microsoft makes plain that
maintaining a consistent user experience is a minor concern and easily gives way when OEMs
create value by differentiating their products: for example, by pre-installing a particular set
of applications. There is no reason why it should not similarly give way when OEMs believe
that end-users will find their machines more attractive when they come with a non-Microsoft
browser, rather than Internet Explorer, pre-installed. This is particularly true because, as I
have explained, OEMs that decrease the value of their products will be punished by the
marketplace.
162. At bottom, Microsoft's forced licensing of Internet Explorer, and Microsoft's
prohibition on OEMs removing Internet Explorer, furthers no substantial efficiency. Rather,
the principal effect of the forced bundling, as the above analysis explains, is simply to exclude
non-Microsoft browsers. This is precisely what Internal Microsoft documents predict. For
example, Microsoft's Christian Wildfeuer wrote: "It seems clear that it will be very hard to
increase browser market share on the merits of IE 4 alone. It will be more important to
leverage the OS asset to make people use IE instead of Navigator."126
B. Windows Platform
163. Microsoft both compels OEMs to license Internet Explorer as a condition of
licensing Windows and prohibits OEMs from removing any part of Internet Explorer from the
PCs they ship to customers. The effect of this conduct, as I have explained, is substantially
Page 73
to impede the commercial opportunities of rival browsers by increasing the costs, and
reducing the benefits, to OEMs of pre-installing non-Microsoft browsers on the machines they
ship.
164. One possible justification for these restrictions, advanced in this litigation by
Microsoft, is to perform a standardization function. By prohibiting OEMs from modifying
the package of software Microsoft markets as Windows, the argument runs, Microsoft
preserves its ability to offer ISVs a consistent platform to which they can write applications.
If each OEM were to modify the Windows package in different ways so as to create different
versions of Windows, Microsoft argues, ISVs might have to support multiple versions of their
software. A consequence could be a reduction in the software developed for Windows and
an increase in its cost.
165. For several reasons, however, Microsoft's objective of supplying ISVs with
a consistent platform does not provide an economic justification for biasing OEMs' choice
of which browser to feature. In the first place, as Professor Farber will explain, Microsoft's
design decision was arbitrary; Microsoft could have put "platform" files (such as shared files)
entirely in the operating system and not included any such files in its browser product.127
166. Moreover, even if that was not the case – or, in any event, taking Microsoft's
design decision as given – Microsoft's platform fragmentation argument does not justify
Microsoft's practices. Microsoft's conduct is anticompetitive because it prevents OEMs from
removing the icon and other visible means of end-user access to Internet Explorer. If a user
cannot invoke a program, then from that user's perspective it is as if that program is not
Page 74.
present.128 And, if IE is not present in that sense, it is in effect removed, and the OEM will be
more likely to distribute other browsers to users who would rather purchase such a browser.
Removal of IE in this sense thus removes an obstacle to the distribution of other browsers.
And – this is the key point – removing IE in this way does not remove any of the shared files
or any of the platform files found in either Windows 95 or Windows 98.
167. Thus, removing other aspects of IE, but leaving behind any "shared program
libraries" upon which IE might rely, will not impair or fragment the Windows platform.
Microsoft recognizes this, and it instructs software developers to configure their programs to
be removed in precisely this manner. "Resources that other programs might use, such as
DLLs," Microsoft explains, "[are] better [left] behind."129
168. Internet Explorer can easily be removed from both Windows 95 and Windows
98 in just this fashion.130 In the case of Windows 95, Microsoft provided the ability to
remove Internet Explorer 3.0 and Internet Explorer 4.0 using the Windows-supplied
"Add/Remove" utility.131 David Cole testified that Microsoft configured Internet Explorer to
be removed from Windows using the "Add/Remove" utility to meet the demand of users who
wanted to remove Internet Explorer.132 Because invoking the Add/Remove utility to
"uninstall" Internet Explorer leaves behind shared program libraries upon which ISVs rely,
Page 75
configuring Windows 95 with Internet Explorer removed in this fashion would not impair the
consistency of the Windows 95 platform in any meaningful way. Indeed, if that were not the
case, it is doubtful Microsoft would have configured Internet Explorer to be so easily
removed.
169. In contrast to Windows 95, Microsoft has designed Windows 98 such that
Internet Explorer cannot be removed through invoking the Add/Remove utility. Nonetheless,
in his report, Dr. Felten explains that the ability to browse the Web using Internet Explorer
can easily be removed from Windows 98, and another browser can supply that functionality
in such a way that the consistency of the Windows platform for ISVs would not be
undermined.133
170. Furthermore, even if (contrary to my understanding) it were necessary to
remove certain shared program libraries from the Windows 98 package to remove Internet
Explorer, the effect on ISVs' practices is likely to be insignificant. There are millions of PCs
running earlier Windows releases that lack the latest versions of Windows 95 or Windows 98.
To ensure that the software they develop runs no matter which version of Windows a PC
contains, ISVs commonly redistribute necessary shared program libraries with their
software.134 In short, Microsoft's own practice of continually updating its platform means
that application developers must replicate part of the platform with the software they
distribute and, therefore, that the effect of an OEM removing certain parts of the "platform"
is likely to be small.
Page 76
C. Restrictions on OEM Modification of Desktop and Start-Up Sequence
171. Microsoft's restrictions on OEMs' modification of the Windows desktop and
start-up sequence, as I have explained, serve to exclude competing browsers and reinforce the
exclusionary effects of Microsoft's other restrictions. As with Microsoft's forced licensing
of, and prohibition on removing, Internet Explorer, these restrictions either serve no legitimate
business justification or any legitimate business justification could be achieved by
substantially less restrictive means.
1. the User Experience
172. Microsoft's principal justification for both the Desktop and Start-up
restrictions is that they promote consistency of user experience. But this concern cannot
justify the restrictions.
173. Taking the Desktop restrictions first, as an initial matter, Microsoft's own
conduct makes plain that this asserted justification cannot support forcing all OEMs to present
users with the exact same initial user experience because Microsoft permits OEMs to vary
substantially the "initial user experience" with their PC. Among other things:
OEMs are permitted to ship Windows 95 and Windows 98 with
Internet Explorer's "Active Desktop" either on or off, and depending
on whether the Active Desktop is on or off, the user will be presented
with a very different appearance.
OEMs that ship their PCs with the Active Desktop enabled may add
items of various different shapes and sizes.
OEMs that ship their PCs with only the standard Windows Desktop
may preinstall other applications and make other icons and folders
appear on the desktop.
A number of OEMs are permitted to populate the Microsoft-supplied
Page 77
Internet Connection Wizard with ISPs of the OEMs' choice.
A number of OEMs are permitted effectively to replace the
Microsoft-supplied Internet Connection Wizard with one supplied by
the OEM.
174. Because Microsoft permits this substantial variation among OEMs (and,
indeed, among different lines shipped by the same OEM), end users have different initial
experiences, a fact conceded by Microsoft.135 I therefore conclude that preserving "the same"
end user experience is an insubstantial justification and cannot support Microsoft's
restrictions on customizing the desktop.
175. Nor can preserving a consistent user experience justify Microsoft's restrictions
on the start-up sequence. Although Microsoft has a legitimate interest in ensuring that users
receive basic warranty and registration information and in ensuring that the start-up does not
take an inordinate amount of time to complete, this interest cannot justify the exclusionary
aspects of the restrictions Microsoft imposed. First, these concerns cannot justify Microsoft's
prohibition on promoting third party brands in the start-up sequence because Microsoft
permits the OEM to promote its own products in that sequence. For example, Microsoft's
recently-signed license with Compaq provides that: "For COMPANY's ‘Presario' branded
Customer systems, COMPANY may insert an Internet Service Provider (‘ISP') sign up wizard
into the boot up process provided that . . . (2) MSCORP's WWW-based services (i.e.,
Expedia, CarPoint, Sidewalk.com, MSNBC, Internet Gaming Zone, MSN Investor, etc.), are
promoted . . . in such ISP sign up Wizard."136 If this promotion impaired legitimate interests,
Microsoft likely would not permit it.
Page 78.
176. Moreover, Microsoft could easily permit OEMs to advertise third party brands
in the start-up sequence, subject to guidelines specifically designed to protect its legitimate
interests in providing information to, and obtaining information from, end users. That
Microsoft easily could do this is confirmed by the fact that it permits certain OEMs effectively
to replace the Microsoft-supplied Internet Connection Wizard with one supplied by the
OEMs, sometimes with the requirement that the OEM promote Microsoft products in that
alternative ICW. This substitute Internet Connection Wizard must conform to Microsoft
guidelines, and Microsoft has conferred the right to include a substitute ICW in large part
based on the demonstrated ability of the OEM in question to implement it without causing
technical difficulties.137
177. Second, Microsoft's purported concern with preserving the consistency of the
initial-boot experience across end-users cannot justify its prohibition on OEMs' using
Windows to run certain programs, including alternative browsers, automatically upon
completion of the end-user boot. This interest can be fully protected by a labeling
requirement. Such a requirement would ensure that users get whatever experience they
choose, rather than the experience dictated by Microsoft. It would also remove an
unnecessary obstacle to the efficient distribution of non-Microsoft browsers.
178. Similarly, the efficacy of a labeling requirement demonstrates that the concern
with providing users a consistent experience cannot justify Microsoft's refusal to permit
OEMs to configure their PCs to boot into an alternative shell, a vehicle through which
competing browsers could be promoted. To be sure, some OEM shells in the past might have
been of poor quality. But Microsoft could meet this concern by establishing certain minimum
Page 79
requirements, including a labeling requirement, with respect to OEMs that decide to utilize
an alternative shell. Microsoft itself acknowledges that OEMs face competition in the
consumer segment of the PC market, and that competition creates for OEMs incentives to
provide only features of a kind and quality that users find desirable.138 As one OEM has
explained, the marketplace will "weed out" those OEMs that provide their users with a shell
that fails to meet with approval.139
2. the User's Windows Experience
179. Microsoft has also asserted that the screen restrictions are necessary to prevent
OEMs from "degrading" Microsoft's Windows product, as Microsoft contends happened
when Packard Bell substituted its own shell for the Windows shell. To a large degree, this
contention simply recapitulates Microsoft's (incorrect) argument that its restrictions are
necessary to preserve the consistency of the user experience. Just as Microsoft's concern with
providing a consistent user experience cannot justify its exclusionary restrictions, neither can
the concern with ensuring delivery of a quality product.
180. While the screen restrictions might have prevented the problem that developed
with Packard Bell's shell – namely, the failure of some APIs to be installed – they are not
necessary to prevent such problems. To maintain the integrity of its platform, Microsoft needs
to restrict or prevent only those modifications of the start up sequence that impair the ability
of ISVs to access the APIs provided by the Windows operating system product.
Page 80
181. Finally, even if OEM modifications of the desktop would in certain
circumstances reduce product "quality," Microsoft's restrictions are unnecessary to protect
that interest. As explained, because the markets in which OEMs operate are competitive,
OEMs have little incentive to take actions that disappoint their customers.140 The market will
serve to punish those OEMs that disappoint consumers by marketing sub-optimal product
configurations. A labeling requirement, moreover, will ensure that the market operates
efficiently. OEMs that modify the Windows desktop in ways Microsoft deems undesirable
could simply be required to advise customers of the modification.
D. achieve any Legitimate Business Purpose
182. Microsoft's restrictions on the ability of ISPs, OLSs, and ICPs to promote and
distribute competing Internet browsers are unrelated to any efficiency purpose; alternatively,
any legitimate efficiency purpose could be accomplished by substantially less restrictive
means.
183. One possible justification for Microsoft's restrictions against ISPs, OLSs, and
ICPs distributing or promoting competing browsers is to ensure that Microsoft is compensated
for the desktop real estate that it supplies those service providers. But this interest cannot
justify Microsoft's restrictions. There is no reason why Microsoft had to take its
Page 81.
compensation in the form of exclusionary agreements rather than a simple payment. That
ISPs commonly agree to pay for customer referrals and promotion of their services
demonstrates that selling desktop real estate does not involve prohibitive transaction costs.
184. The ICP agreements differ from the ISP and OLS agreements in that the ICP
agreements call for the employment of Internet Explorer-specific technologies. However, there
is no reason to think that compelling ICPs to make their web-sites work less well with
competing browsers is reasonably necessary to compensate Microsoft for placement on the
"channel bar." Microsoft could instead have charged ICPs a fee for such placement. Indeed,
one aspect of the ICP agreements – the prohibition on ICPs payment from Netscape for
promotional services – plainly serves no legitimate purpose. The only reason for that
restriction is to impede Netscape.
X. MICROSOFT'S ACTIONS TO INCREASE ITS BROWSER USAGE
SHARE ARE PREDATORY
185. The available evidence indicates that Microsoft pursued the practices I have
examined for the purpose of preserving its Windows operating system monopoly and gaining
monopoly power in the browser market, and pursued them without regard to whether they
would have been profitable in their own right. Accordingly, it is my opinion that Microsoft's
intent in engaging in this course of conduct, when considered as a whole, was predatory.
186. Conduct is predatory when it is expected to be profitable only if it excludes
rivals and thereby creates or enhances market power. The pertinent question in determining
whether Microsoft's conduct is predatory, therefore, is: would Microsoft's actions to increase
browser usage share have been profitable if they did not injure Netscape or other rivals and
enable Microsoft to gain or preserve market power? Based on the evidence I have seen, it is
Page 82.
my opinion that the answer to this question is no.
187. To be sure, Microsoft has a legitimate interest in ensuring that Windows users
are able to acquire high quality browsers at low prices, because that would increase the demand
for Microsoft's operating system. But even if achieving this objective were furthered by
Microsoft's decision to offer a quality browser product, its further efforts to increase IE's share
by excluding Netscape and making it more difficult for users to obtain Netscape's browser
could only reduce the value of its operating system to consumers.
188. Moreover, Microsoft's own documents show that its exclusionary restrictions
were implemented, not to increase the value of the browser as a complement or to obtain
ancillary revenues, but rather as part of a "jihad" to win the browser war and thereby eliminate
a threat to Microsoft's monopoly power. The documents make clear that Microsoft both feared
the browser would be a key component in a substitute to the Windows platform and viewed
the conduct examined here as instrumental in blunting that threat.
189. Microsoft's actions to increase IE's usage share and to exclude rival browsers
were expensive for Microsoft in at least two respects. First, Microsoft's OEM restrictions and
its exclusionary contracts with ISPs, OLSs, and ICPs imposed burdens on those parties, and
thus diminished the value of other consideration those parties were willing to pay to
Microsoft. As noted above, Microsoft's tying of IE to its operating system made distribution
of rivals browsers infeasible or more costly for OEMs and thus reduced the OEMs' demand
for Windows. Microsoft's agreements with ISPs and OLSs provided those firms with
preferential access on highly desirable terms to valuable Desktop real estate. This is a unique
asset; its value was enhanced by the OEM screen restrictions; and it could have generated
substantial direct revenue for Microsoft if it had been sold rather than bartered or exchanged
Page 83.
for exclusivity agreements.
190. Second, Microsoft's decision to give IE away free to the installed base of
Windows users meant sacrificing substantial revenue from two sources. First, Microsoft lost
revenue from not licensing IE at a positive price as a stand-alone application -- whether
through downloads directly to end users or through positive license fees to ISPs, OLSs and
ICPs.141 Second, Microsoft lost revenue from retail sales of Windows 98 upgrades because
providing IE free to the installed base reduced the demand for the Windows 98 upgrade and
the revenue Microsoft earns from that source.142
191. As is true for other software, the marginal production and marketing cost of
licensing IE to another user, once it has been developed, is very low. However, this does not
imply (nor does experience in the software industry indicate)that if there are two suppliers of
browsers, such as Microsoft and Netscape, the market price for browsers would invariably
equal zero. Indeed, the standard oligopoly models used by economists predict that price in a
market with two competing suppliers will not be zero even if marginal costs for both firms are
zero and the products are identical.143 And, importantly, we commonly observe instances in
which the prices for two or more similar software applications (such as WordPerfect and
Page 84
Word) are not zero.
192. The browser market arguably has characteristics that might reduce browser
prices further than would be expected for other software products. Specifically, Microsoft and
Netscape might price their browser products at a low level because higher browser usage share
could generate higher revenues from ancillary products, such as advertising on browser-related
web sites and sales of complementary software, such as server software or development
tools.144 Yet these same incentives were present when Netscape first decided to charge a
positive price for its browser. It was not the potential for the generation of ancillary revenue
that brought the market price of the browser down to zero, but rather Microsoft's actions.
193. Ideally, one would like to measure the costs Microsoft incurred through its
pricing policies and exclusionary agreements and to compare those to the revenues Microsoft
could have expected to gain absent any effect on the competitiveness of the browser and
operating system markets. There are, however, no data currently available to me that would
provide an accurate estimate of those costs and revenues. In particular, I have seen no
documents indicating that Microsoft ever performed such a calculation at the time these
decisions were made. (Indeed, if Microsoft performed such a calculation today and determined
that it earned substantial ancillary revenues from increasing its browser usage share, that result
would not be meaningful unless it could be shown to provide a reliable guide to what
reasonably could have been anticipated by Microsoft at the time of the decision was made.)
194. The evidence I have seen supports the inference that Microsoft took
Page 85
exclusionary actions and incurred costs without regard to whether its actions were profit-
maximizing – or even profitable – absent the future revenue gains from weakening rival
browsers and thereby preserving its Windows operating system monopoly and from gaining
monopoly power in the browser market. Instead, Microsoft viewed winning browser share at
almost any cost as being of overwhelming strategic importance.
195. Accordingly, on the basis of the available evidence, I conclude that Microsoft's
conduct, in the aggregate, was not expected to be profitable except for the market power
Microsoft expected to gain from the exclusion of browser rivals and therefore was predatory.
Page 86.
XI. THE COSTS TO CONSUMERS AND COMPETITION FROM
MICROSOFT'S EXCLUSIONARY CONDUCT ARE SIGNIFICANT
AND FAR REACHING
196. Consumers will be significantly harmed if Microsoft succeeds in crushing the
cross-platform threat that independent browsers pose to the Windows operating system
monopoly. The development of cross-platform technologies can be expected to bring
substantial benefits to end users. Because cross-platform technologies enable an application
to run on top of many different operating systems, they substantially reduce the applications
barrier to entry. This might have at least two important benefits.
197. First, compatibility between an application and multiple operating systems
makes it more practical or attractive for end users to choose operating systems that have been
optimized to their specific needs or requirements, rather than a general purpose OS like
Windows. For example, engineers, scientists, and even economists often run esoteric
applications programs using very large data sets, and an operating system can be specifically
designed for those applications. Similarly, other professionals do a large share of their
computer-related work with applications such as word processing or spreadsheets, and an
operating system also can be optimized to run those types of applications. Optimizing
operating systems in this way has the attractive feature of making the operating system simpler
and more "stable." At the same time, a large stock of cross-platform applications can run on
top of such operating systems. In effect, this compatibility allows users to achieve OS
"specialization" without sacrificing versatility.
198. Second, by reducing the demand for general purpose PC operating systems and
the barriers to entry into the PC OS market, cross-platform technologies can be expected to
reduce the price that end users pay for operating systems.
Page 87
199. There is no guarantee, of course, that independent browsers will bring these
benefits or reduce the monopoly power of Microsoft in the operating system market, even if
Microsoft did not engage in exclusionary conduct. That is a matter for the market – not
monopolists or engineers or economists – to decide. The important point is that the market
should not be prevented by Microsoft's anticompetitive practices from making that decision.
200. Indeed, the stakes in the current dispute go beyond the browser threat. If
Microsoft can use anticompetitive conduct to extinguish the threat to its operating system
posed by independent browsers, others are likely to expect it to use similar tactics to prevent
them from developing other types of cross platform technologies that would threaten the
Windows platform. Having seen Microsoft successfully leverage its Windows monopoly
against one threat, there will be less incentive to invest in other technologies that Microsoft
might view as a similar threat and, thus, less innovation in the software industry.
201. I declare under penalty of perjury that the foregoing is true and correct.
_______________________________
Frederick R. Warren-Boulton
FOOTNOTES
1See 4 Trade Reg. Rep. (CCH) ¶ 13,104 (1992).
2See Pl. Ex. 1 below, "Microsoft's Actual and Projected Share of the (Intel-based) PC Operating System Market."
3 MS6 5004553.
4See MS98 0203023-25.
5 Non-Microsoft browsers and Microsoft's operating system thus fit the meaning of the term "partial substitutes" as used by Areeda in his treatise. See 10 Phillip E. Areeda, et al., Antitrust Law ¶ 1747c, at 232-33 (1996) (explaining that "driving out producers of partial substitutes more likely" will "reduce the likelihood of entry into the tying market" because producers of partial substitutes "have the skill to be potential entrants in the tying market more frequently than producers of complete nonsubstitutes do").
6 Microsoft Press Computer Dictionary (Third Edition, 1997, Microsoft Press) (hereinafter "Microsoft Dictionary").
7 In 1997, 87.6% of all copies of the Microsoft's Windows 95 operating system product were installed by OEMs, while 7.3% were sold through retail channels as upgrades. Windows 95 is available at retail only as an upgrade from a Microsoft licensed operating system See Appendix B to Microsoft's Responses to Interrogatories, March 23, 1998.
8See Akerlind Dep. 78 (Aug. 26, 1998) (explaining that "[t]he end user segment [of the OEM market] is competitive").
9 Microsoft also develops and sells other operating systems targeted to more specialized markets, including operating systems for workstations (Windows NT Workstation), servers (Windows NT Server) and embedded and special purpose systems (Windows CE).
10See, e.g., Brown Dep. 8-11 (Mar. 5, 1998); McKinney Dep. 9-11 (Mar. 13, 1998).
11See, e.g., United States v. Baker Hughes Inc., 908 F.2d 981, 985-86 (D.C. Cir. 1990); Community Publishers, Inc. v. Donrey Corp., 892 F. Supp. 1146, 1153 (W.D. Ark. 1995), aff'd, 139 F.3d 1180 (8th Cir. 1998).
12 Guidelines § 1.0.
13 More precisely, the price elasticity for a product (or group of products) is the percentage reduction in unit sales for the product that would result from a one percent increase in the price of that product, holding all else constant.
14SeeU.S. Anchor Mfg., Inc. v. Rule Indus., Inc., 7 F.3d 986, 995 (11th Cir. 1993).
15 Guidelines § 1.11.
16See MS98 0113387 and MS7 007194.
17See Brownrigg Dep. 9-10 (Mar. 5, 1998) (explaining that "switch[ing] from manufacturing personal computers based on the Intel-based processor". . ."would be a very daunting task" because it would entail a "[s]ignificant amount of engineering work," . . . "[a] lot of retraining of technical support," and other costs).
18See Ransom Dep. 13-14 (Mar. 19, 1998); Ransom Dep. 16 (Aug. 7, 1998); Brown Dep. 10-11 (Mar. 5, 1998).
19See Ransom Dep. 9, 17 (Aug. 7, 1998); Brown Dep. 10-11 (Mar. 5, 1998); Romano Dep. 50-52 (Apr. 13, 1998); Von Holle Dep. 12-13 (Sept. 9, 1997); Kies Dep. 16-17 (Sept. 11, 1998); Santos Dep. 7-8 (Apr. 13, 1998); McKinney Dep. 11-12 (Mar. 13, 1998).
20SeeBrown Dep. 10-12 (Mar. 5, 1998)
21See Sanders Dep. 104-06 (Aug. 26, 1998).
22SeeMS7 000634.
23See MS7 000634.
24See Sparks Dep. 131-33 (July 9, 1998).
25 "Note on the Desktop Operating System Industry in 1996," Stanford University Graduate School of Business (August, 1997), p. 23.
26 As Professor Kenneth Arrow stated "The value of the operating system product is in its capability to run application software. The larger the installed base of a particular OS, the more likely it is that independent software vendors will write program that run on the OS, and, in this circular fashion, the more valuable the OS will be to customers." Declaration of Kenneth J. Arrow dated January 17, 1995, submitted in United States v. Microsoft Corp., No. 94-1564 (D.D.C. 1994), on behalf of the Government and in opposition to Amici Curiae.
27See Microsoft's Answer ¶ 58 (explaining that "(i) the popularity of an operating system [product] is to some extent a function of the number, variety and quality of application available for use with that operating system [product]; and (ii) software developers tend to write applications for operating systems [products] that are popular").
28SeeSantos Dep. 9 (Apr. 13, 1998);seealso Von Holle Dep. 9-10 (Sept. 19, 1997).
29 MS 154265-4279, 154268, Nathan Myhrvold, file attachment (InterOffice Memo) to email to Bill Gates, Peter Rinearson, and Jonathan Lazarus (July 24, 1993).
30See Confidential Submission of Microsoft Corporation to the Staff of the Antitrust Division of the United States Department of Justice 61 (Sept. 19, 1993).
31See Chase Dep. 97 (Mar. 25, 1997).
32As Brad Chase explained: "The OS/2 application market is small and getting smaller. Windows applications now account for almost 60% of all dollars spent worldwide on applications software, with $4 billion sold in 1993. In the same period sales of OS/2 applications have declined and in 1993 accounted for only 2% of the market, or $128 million." MS98 11434-44 (Aug. 8, 1994).
33 Kozel Dep. 11 (Sept. 19, 1997).
34 MS7 7193-7196, 7196.
35 Decker Dep. 18-21 (Oct. 17, 1997).
36See Ransom Dep. 9, 17 (Aug. 7, 1998); Brown Dep. 10-11 (Mar. 5, 1998); Romano Dep. 50-52 (Apr. 13, 1998); Von Holle Dep. 12-13 (Sept. 9, 1997); Kies Dep. 16-17 (Sept. 11, 1998); Santos Dep. 7-8 (Apr. 13, 1998); McKinney Dep. 9 (Mar. 13, 1998).
37 MS7 007194, Joachim Kempin to Bill Gates (Dec. 16, 1997).
38 For Microsoft as a whole, net revenue as a percentage of total revenue increased from 20% in FY86 to 26% in FY92, falling to 24% in FY95 before rising again to 30% in FY97 (See Microsoft Annual Reports). Among the Fortune 500 largest U.S. corporations, Microsoft ranks 137th in revenue, 165th in assets, 15th in profits, 7th in growth of earnings per share, 3rd in profits as percentage of assets, 2nd in market value, and 1st in profits as a percentage of revenues. Fortune (Apr. 27, 1998).
39 See, for example, Brad Chase deposition 206 (Mar. 25, 1998).
40 Browsers generally also contain a limited set of application programming interfaces ("APIs") to which software writers can "write" to extend the functionality of their application products to "Internet-oriented tasks." This set of API's is not a substitute for the set of API's provided by the operating system.
41 Netscape Communicator -- Standard Edition -- is available for almost all Windows products, Mac System 7.5 and above, all major UNIX desktop systems and soon for OS/2. See http://www.netscape.com/navigator/index.html. Microsoft's Internet Explorer 4.0 is available on all Windows products, on Macintosh OS 7.1 and above, and on UNIX Solaris 2.5 and above and other UNIX platforms. Seehttp://www.microsoft.com/ie/ download/sysreq.html; seealso Chase Dep. 98 (March 25, 1998); Mehdi Dep. 32, 45 (April 2, 1998).
42 Other browsers include Lynx, Mosaic, Opera, Web Explorer and WebSurfer.
43 Microsoft itself argues that there are consumer benefits deriving from the competitive race. In a memorandum entitled "The Internet PC" dated April 10, 1996, Bill Gates noted that Netscape Navigator "led the way with speed and features . . . . Netscape and Microsoft have overlapping visions of the future of the Internet. Each company is working as hard as it can, as fast as it can, to develop software that supports its approach. One consequence of this feature race is that browsers are evolving from relatively simple pieces of software into large programs, enhanced with various extensions . . . ." MS6 6012977-78.
44See, e.g., Smith Dep. 16-21 (Oct. 10, 1997).
45 This is consistent with the test the Supreme Court applied in the Jefferson Parish case. SeeJefferson Parish Hosp. Dist. No. 2 v. Hyde, 466 U.S. 2 (1984).
46See 10 Phillip E. Areeda, et al., Antitrust law ¶ 1747c, at 232-33 (1996)
47 Chase Dep. 99 (Mar. 25, 1998).
48 Silverberg Dep. 19 (April 14, 1998).
49See, e.g., Frasca Dep. 45 (July 24, 1998); seealso Clark Dep. 125 (July 22, 1998) ("You don't see Sun Microsystems integrating the browser into their UNIX operating system. You don't see DEC integrating it in their UNIX operating system. You don't see Hewlett Packard integrating it in their UNIX operating system, and you don't see Silicon Graphics integrating it in their UNIX operating system. They treat it as an application . . . .").
50See Trevanian Dep. 61-63 (July 17, 1998).
51See Sparks Dep. 74 (July 9, 1998) and Croll Dep. 71-72 (July 14).
52See Bergland Dep. 29-32 (Aug. 11, 1998).
53See Kanicki Decl.¶ 2 (Apr. 29, 1998); seealso Kanicki Dep. 26-29, 106-07 (Aug. 27, 1998).
54 Ransom Dep. 9-10 (Mar. 19, 1998).
55 Decker Dep. 17-21 (Oct. 29, 1997); Kies Dep. 10-12, 14-15 (Apr. 23, 1998); Kies Dep. 21-28 (Sept. 11, 1998); Von Holle Dep. 39-40, 59-60 (Oct. 6, 1998); McClain Dep. 43-44 (Aug. 7, 1998); Declaration of Eric Browning, ¶¶ 7-9 (October 14, 1997).
56 Cole Dep. 49-50 (Jan. 9, 1998)
57 Expert Report of Edward W. Felten ¶¶ 48-54.
58See Frasca Dep. 35-38, 86-87, 106-07 (July 24, 1998); Croll. Dep. 74, 188 (July 14, 1998); Limp Dep.123-25, 128 (July 30, 1998); Bergland Dep. 24-30 (Aug. 11, 1998).
59 MS6 5004550, 4553; Dunn Dep. 37-42 (Apr. 24, 1998).
60 MS7 004127 PI Ex. 15.
61 MS6 6008247, 8248 PI Ex. 35.
62 MS6 6005550 (PI Ex. 33).
63 Allchin Dep. 120 (Mar. 19, 1998).
64Id. at 121.
65 Webb Mckinney Dep, 27 (Mar. 13, 1998) ("Obviously it costs us money to bundle software, so we don't want to put a lot of software on a system that people don't value, and the amount of software on the system also increases our test burden and makes it harder for us to get to market quickly, so it's a pretty wide range of reasons to try to keep the amount of software on the system to what's really needed"); Kempin Dep. 37 (October 2, 1997). Mr. Kempin also testified that "[i]f the OEM wants to install the other browser, he can do it. And it's just a matter of how much cost he is willing to have. Because he will probably have to test all our product first then he has to test the other products..." Id. at 34.
66See McClain Dep. 51 (agreeing that "icons can cause clutter on the screen"). Indeed, Microsoft explained to OEMs that, although it permitted multiple icons for ISP sign-up on the desktop, it "recommend[s that] OEMs do not overwhelm the user by providing additional sign-up mechanisms." MS98 0109575 (4/16/98); seealso MS98 109900 (2/27/98) (Microsoft e-mail exchange reflecting concern that offering icons for two or more products or services in the same category on the desktop would cause consumer confusion).
67See McClain Dep. at 52-53; McKinney Dep. 27 (Mar. 13, 1998) ("[O]ne thing is that we found a lot of software required, you know, creates a lot of support, so it creates a support burden for us. It creates confusion for the consumer.").
68 McClain Dep. 122-23; id. at 57.
69 Von Holle Dep. 42, 61 (Oct. 6, 1998).
70 Von Holle Dep. at 42; seealso Brownrigg Dep. 14, 18 (Oct. 6, 1998).
71 Kempin Dep. 37 (Oct. 2, 1997).
72See, e.g., IE Market Review, April, 1997 (TXAG 0026734-770); MS7 006062; Myhrvold Dep. 43-44 (Apr. 24, 1998). The other principal channel is through OLSs and ISPs, the foreclosure of which is discussed below.
73 IE5 OEM Marketing Review, MS98 012655.
74 MS7 005526.
75 MS7 007468; MS7 006062-64 (Jeff Johnson e-mail) ("[I]t is pretty clear that current IE users are often getting it with their PC, and the focus group data . . . also stated that Netscape users aren't likely to switch to IE until it is integrated into the OS"); MS7 006352 (Jonathan Roberts e-mail) ("This distribution leads me to believe we are better off with a tighter tie to Windows.").
76See Myhrvold Dep. 43 (Apr. 24, 1998).
77 Knott Dep. 21-23 (Feb. 20, 1998). Indeed, Knott testified that CompuServe had tried but failed to put together a comparable package through OEMs. Id. at 22; seealso Von Rump Dep.14-15 (Apr. 28, 1998); Colburn Dep. at 29-33 (Mar. 6, 1998).
78 Silverberg Dep. 159 (Apr. 15, 1998).
79 MS6 5001199-1245 (AOL agreement); MS6 5000168-89 (CompuServe Agreement).
80 Myhrvold Dep. 64-65, 68 (Apr. 24, 1998).
81See, e.g., MS6 5001127-51 (Spry); MS6 50000920-47 (Mindspring).
82 Microsoft Dictionary 15.
83Id. at 15-16.
84 Testimony from ISP executives and internal Microsoft documents show that there are significant costs for ISPs which arise from distributing two browsers. See Von Rump Dep. 17 (Apr. 28, 1998) (expenses incurred for support of an additional browser include "marketing resources, whatever expenses are required in the promotion and distribution, advertising, literature, training for sales and customer service representatives"); Solnik Dep.81-82 (June 15, 1998) (explaining that "there are lots of costs" including testing, training, and support costs, for an ISP to distribute two browsers); seealso Beran Dep. 52 (Aug. 5, 1998); Schwartz Dep. 41-42 (Sept. 9, 1998); MS7 005526.
85 Silverberg Dep. 142-44 (Apr. 14, 1998).
86See MS98 0112834-36.
87Seeid.; MS6 6009919; MS7 000584. See also MS7 00591, a 1997 Mid-year review in which Microsoft noted that "46 of top 50 ISPs/OLSs [are] shipping IE as their preferred browser."
88 Knott Dep. 25 (Feb. 20, 1998)
89 Von Rump Dep. 16-17 (Apr. 28, 1998)
90 MS98 0151667-68.
91 MS98 01516667.
92 Wadsworth Decl.¶ 4 (Apr. 23, 1998).
93 ZD 0127.
94See, e.g, AOL 0000145-73 (AOL); CNET 00028-55 (CNET); TWDC 0704-55 (Disney); INT 00001-25 (Intuit).
95See, e.g., INT 00003; See Chase Dep. 206-08 (Mar. 3, 1998).
96 Dunn Dep. 37-42 (Apr. 24, 1998).
97Id.
98 MS 6003202-3227.
99 MSV 9137A.
100Id.
101 MS6 5005719-720 (Apr. 4, 1996, Chase Planning Memo).
102 MSV 10558.
103 MS7 000353-366-387.
104 MS7 006062.
105Compare MSV 000163 (IBM 8/24/95 Windows 95 License) with MSV 000203-04 (IBM 8/16/96 Windows 95 License).
106See, e.g., Amendment 12 to August 1, 1996, License agreement between Microsoft Corporation and AST Research, Inc., Ex. C1 (Aug. 8. 1996) (MSV 0006245).
107 Myhrvold Dep. 73-74 (Sept. 24, 1998).
108See, e.g., MS98 0113937; MS98 0113961; MS98 0113962; MS98 0113963.
109See Kempin Dep. 45-46, 49-50 (Sept. 9, 1998).
110See, e.g., MS98 0113849-52 (May 27, 1998 Letter to Packard Bell/NEC authorizing alternative ISP sign-up process); Kempin Dep. 94 (Sept. 9, 1998).
111See, e.g., Amendment 12 to August 1, 1996, License agreement between Microsoft Corporation and AST Research, Inc., Ex. C1 (Aug. 8. 1996) (MSV 0006245).
112See Kempin Dep. 96-98 (Sept. 9, 1998).
113 Microsoft made assumptions concerning the percentage of new Internet connections who use Internet Explorer in each of the sectors just described (the "run rate") and the rate at which people switch browser types (the "switching rate").
114 MS98 0203007.
115 MS98 0203007.
116 A user "clicks on" on an ad to get more information about the product.
117 Because of the size of this database, data were acquired for only the second Wednesday of each month.
118 "Netscape Competitive Analysis." MS98 0112834-36.
119 "Netscape Competitive Analysis." MS98 0112834-36.
120 The difference between the two groups (i) takes OEM restraints as given; (ii) measures a stock rather than flow of browser shares; and, (iii) does not take into account the possibility that network effects in the browser market will cause the effect of Microsoft's anticompetitive practices to accumulate over time.
121See MS98 0203023-25.
122See Kanicki Dep. 36, 40 (Aug. 27, 1998). It is also unlikely Microsoft would have agreed to a licensing agreement with Dell, see MS98 0128328, that lifts the screen restrictions, including the prohibition on removing the IE icon, when Dell is presented with a specific customization request by a large-volume customer, such as corporate customers, if such customization would have harmful effects.
123See Expert Report of Edward W. Felten ¶¶ 48-64 (hereinafter "Felten Report").
124See Akerlind Dep. 115 (Aug. 26, 1998) (agreeing that "[i]f Compaq's removal of the Internet Explorer icon caused . . . problems" that Compaq would "be punished by the marketplace."); seealsoid. at 87 ("I would agree . . . that if a company in this market or any market, you know, has weak products, that sooner or later they're either going to have to fix the problem or get out of the business, because what they try to sell is not going to be profitable business for them.").
125 Microsoft's Memorandum in Support of Summary Judgment at 58.
126 MS7 004246.
127 Expert Report of David J. Farber ¶ 23.
128 Expert Report of Glenn E. Weadock ¶ 26 ("The existence of a software product . . . depends on both the presence of a feature set and the means to use that feature set.") (hereinafter "Weadock Report").
129 Felten Report ¶ 52 (quoting Handbook for Applications 29).
130 Felten Report ¶¶ 48-57.
131 Weadock Report ¶ 24.
132 Transcript Jan. 14, 1998 a.m. 4, 48-50 (United States v. Microsoft, No. 97-1564 (TPJ))
133See Felten Report ¶¶ 48-57.
134See Weadock Report ¶ 27; seealso Gailey Decl ¶ 4 (Nov. 17, 1997); Allaire Decl. ¶ 2 (Nov. 17, 1997); Bourdeau Decl ¶¶ 3-4 (Nov. 17, 1997); Bickel Decl. ¶ 8 (Nov. 19, 1997) (all introduced in United States v. Microsoft, No. 94-1564 (TPJ)).
135See Akerlind Dep. 155-56 (Aug. 26, 1998).
136 MS98 0102953 (Apr. 1, 1998).
137See Kempin Dep. 84-85 (Sept. 9, 1998); Akerlind Dep. 150-55 (Aug. 26, 1998).
138See Akerlind Dep. at 83-88.
139 Akerlind Dep. at 83.
140 It is important to keep in mind that, from an economic perspective, the value of a product to consumers is determined by both its quality, including all of its features, and its price. For example, some consumers may prefer an Oldsmobile to a Cadillac because the additional features a Cadillac provides are not worth paying the higher price, or because they prefer the features the Oldsmobile provides and would chose it even if the price were the same. One cannot say that the Cadillac is superior simply because some or many of its features are not provided by, or are superior to counterparts provided by, the Oldsmobile.
141 Although early versions of IE might have had only modest revenue-generating potential, because they were widely regarded to be substantially inferior to Netscape's browser, that is no longer the case.
142See TXAG 0012832; MS7 005732.
143 For example, the Cournot model, the oldest and still most widely used model to predict behavior in oligopolistic markets, predicts that, with two suppliers producing an identical product with zero marginal cost, price will be below the monopoly level but will not equal zero. Although a zero price is possible under a Bertrand model of oligopoly that assumes that the products are identical and that the cost to users of switching products is zero, these assumptions do not comport with the realities of the browser market.
144 Microsoft has suggested in this litigation that it might set a non-predatory price of zero in the expectation that the resulting increase in IE usage would generate increased revenues from increased sales of its Windows 95 and Windows 98 operating system products. For reasons explained above, I do not believe that this argument can help explain Microsoft's pricing of IE to Microsoft's installed base.
|
U.S. Department
of Justice
