144. Very soon after it recognized the need to gain browser usage share at Navigator's expense, Microsoft identified pre-installation by OEMs and bundling with the proprietary client software of IAPs as the two distribution channels that lead most efficiently to browser usage. Two main reasons explain why these channels are so efficient. First, users must acquire a computer and connect to the Internet before they can browse the Web. Thus, the OEM and IAP channels lead directly to virtually every user of browsing software. Second, both OEMs and IAPs are able to place browsing software at the immediate disposal of a user without any effort on the part of the user. If an OEM pre-installs a browser onto its PCs and places an icon for that browser on the default screen, or "desktop," of the operating system, purchasers of those PCs will be confronted with the icon as soon as the operating system finishes loading into random access memory ("RAM"). If an IAP bundles a browser with its own proprietary software, its subscribers will, by default, use the browser whenever they connect to the Web. In its internal decision-making, Microsoft has placed considerable reliance on studies showing that consumers tend strongly to use whatever browsing software is placed most readily at their disposal, and that once they have acquired, found, and used one browser product, most are reluctant -- and indeed have little reason -- to expend the effort to switch to another. Microsoft has also relied on studies showing that a very large majority of those who browse the Web obtain their browsing software with either their PCs or their IAP subscriptions.
145. Indeed, no other distribution channel for browsing software even approaches the efficiency of OEM pre-installation and IAP bundling. The primary reason is that the other channels require users to expend effort before they can start browsing. The traditional retail channel, for example, requires the consumer to make contact with a retailer, and retailers generally do not distribute products without charging a price for them. Naturally, once Microsoft and Netscape began offering browsing software for free, consumers for the most part lost all incentive to pay for it.
146. The relatively few users who already have a browser but would prefer another can avoid the retail channel by using the Internet to download new browsing software electronically, but they must wait for the software to transmit to their PCs. This process takes a moderate degree of sophistication and substantial amount of time, and as the average bandwidth of PC connections has grown, so has the average size of browser products. The longer it takes for the software to download, the more likely it is that the user's connection to the Internet will be interrupted. As a vanguard of the "Internet Age," Navigator generated a tremendous amount of excitement in its early days among technical sophisticates, who were willing to devote time and effort to downloading the software. Today, however, the average Web user is more of a neophyte, and is far more likely to be intimidated by the process of downloading. It is not surprising, then, that downloaded browsers now make up only a small and decreasing percentage of the new browsers (as opposed to upgrades) that consumers obtain and use.
147. The consumer who receives a CD-ROM containing a free browser in the mail or as a magazine insert is at least spared the time and effort it would take to obtain browsing software from a retail vendor or to download it from the Web. But, just as the consumer who obtains a browser at retail or off the Web, the consumer who receives the software unsolicited at home must first install it on a PC system in order to use it, and merely installing a browser product takes time and can be confusing for novice users. Plus, a large percentage of the unsolicited disks distributed through "carpet bombing" reach individuals who do not have PCs, who already have pre-installed browsing software, or who have no interest in browsing the Web. In practice, less than two percent of CD-ROM disks disseminated in mass-distribution campaigns are used in the way the distributor intended. As a result, this form of distribution is rarely profitable, and then only when undertaken by on-line subscription services for whom a sale translates into a stream of revenues lasting into the future. The fact that an OLS may find it worthwhile to "carpet bomb" consumers with free disks obviously only helps the vendor of browsing software whose product the OLS has chosen to bundle with its proprietary software. So, while there are other means of distributing browsers, the fact remains that to a firm interested in browser usage, there simply are no channels that compare in efficiency to OEM pre-installation and IAP bundling.
148. Knowing that OEMs and IAPs represented the most efficient distribution channels of browsing software, Microsoft sought to ensure that, to as great an extent as possible, OEMs and IAPs bundled and promoted Internet Explorer to the exclusion of Navigator.
149. Consumers determine their software requirements by identifying the functionalities they desire. While consumers routinely evaluate software products on the basis of the functionalities the products deliver, they generally lack sufficient information to make judgements based on the designs and implementations of those products. Accordingly, consumers generally choose which software products to license, install, and use on the basis of the products' functionalities, not their designs and implementations.
150. While the meaning of the term "Web browser" is not precise in all respects, there is a consensus in the software industry as to the functionalities that a Web browser offers a user. Specifically, a Web browser provides the ability for the end user to select, retrieve, and perceive resources on the Web. There is also a consensus in the software industry that these functionalities are distinct from the set of functionalities provided by an operating system.
151. Many consumers desire to separate their choice of a Web browser from their choice of an operating system. Some consumers, particularly corporate consumers, demand browsers and operating systems separately because they prefer to standardize on the same browser across different operating systems. For such consumers, standardizing on the browser of their choice results in increased productivity and lower training and support costs, and permits the establishment of consistent security and privacy policies governing Web access.
152. Moreover, many consumers who need an operating system, including a substantial percentage of corporate consumers, do not want a browser at all. For example, if a consumer has no desire to browse the Web, he may not want a browser taking up memory on his hard disk and slowing his system's performance. Also, for businesses desiring to inhibit employees' access to the Internet while minimizing system support costs, the most efficient solution is often using PC systems without browsers.
153. Because of the separate demand for browsers and operating systems, firms have found it efficient to supply the products separately. A number of operating system vendors offer consumers the choice of licensing their operating systems without a browser. Others bundle a browser with their operating system products but allow OEMs, value-added resellers, and consumers either to not install it or, if the browser has been pre-installed, to uninstall it. While Microsoft no longer affords this flexibility (it is the only operating system vendor that does not), it has always marketed and distributed Internet Explorer separately from Windows in several channels. These include retail sales, service kits for ISVs, free downloads over the Internet, and bundling with other products produced both by Microsoft and by third-party ISVs. In order to compete with Navigator for browser share, as well as to satisfy corporate consumers who want their diverse PC platforms to present a common browser interface to employees, Microsoft has also created stand-alone versions of Internet Explorer that run on operating systems other than 32-bit Windows, including the Mac OS and Windows 3.x.
154. In conclusion, the preferences of consumers and the responsive behavior of software firms demonstrate that Web browsers and operating systems are separate products.
155. In contrast to other operating system vendors, Microsoft both refused to license its operating system without a browser and imposed restrictions -- at first contractual and later technical -- on OEMs' and end users' ability to remove its browser from its operating system. As its internal contemporaneous documents and licensing practices reveal, Microsoft decided to bind Internet Explorer to Windows in order to prevent Navigator from weakening the applications barrier to entry, rather than for any pro-competitive purpose.
156. Before it decided to blunt the threat that Navigator posed to the applications barrier to entry, Microsoft did not plan to make it difficult or impossible for OEMs or consumers to obtain Windows without obtaining Internet Explorer. In fact, the company's internal correspondence and external communications indicate that, as late as the fall of 1994, Microsoft was planning to include low-level Internet "plumbing," such as a TCP/IP stack, but not a browser, with Windows 95.
157. Microsoft subsequently decided to develop a browser to run on Windows 95. As late as June 1995, however, Microsoft had not decided to bundle that browser with the operating system. The plan at that point, rather, was to ship the browser in a separate "frosting" package, for which Microsoft intended to charge. By April or May of that year, however, Microsoft's top executives had identified Netscape's browser as a potential threat to the applications barrier to entry. Throughout the spring, more and more key executives came to the conclusion that Microsoft's best prospect of quashing that threat lay in maximizing the usage share of Microsoft's browser at Navigator's expense. The executives believed that the most effective way of carrying out this strategy was to ensure that every copy of Windows 95 carried with it a copy of Microsoft's browser, then code-named "O'Hare." For example, two days after the June 21, 1995 meeting between Microsoft and Netscape executives, Microsoft's John Ludwig sent an E-mail to Paul Maritz and the other senior executives involved in Microsoft's browser effort. "[O]bviously netscape does see us as a client competitor," Ludwig wrote. "[W]e have to work extra hard to get ohare on the oem disks."
158. Microsoft did manage to bundle Internet Explorer 1.0 with the first version of Windows 95 licensed to OEMs in July 1995. It also included a term in its OEM licenses that prohibited the OEMs from modifying or deleting any part of Windows 95, including Internet Explorer, prior to shipment. The OEMs accepted this restriction despite their interest in meeting consumer demand for PC operating systems without Internet Explorer. After all, Microsoft made the restriction a non-negotiable term in its Windows 95 license, and the OEMs felt they had no commercially viable alternative to pre-installing Windows 95 on their PCs. Apart from a few months in the fall of 1997, when Microsoft provided OEMs with Internet Explorer 4.0 on a separate disk from Windows 95 and permitted them to ship the latter without the former, Microsoft has never allowed OEMs to ship Windows 95 to consumers without Internet Explorer. This policy has guaranteed the presence of Internet Explorer on every new Windows PC system.
159. Microsoft knew that the inability to remove Internet Explorer made OEMs less disposed to pre-install Navigator onto Windows 95. OEMs bear essentially all of the consumer support costs for the Windows PC systems they sell. These include the cost of handling consumer complaints and questions generated by Microsoft's software. Pre-installing more than one product in a given category, such as word processors or browsers, onto its PC systems can significantly increase an OEM's support costs, for the redundancy can lead to confusion among novice users. In addition, pre-installing a second product in a given software category can increase an OEM's product testing costs. Finally, many OEMs see pre-installing a second application in a given software category as a questionable use of the scarce and valuable space on a PC's hard drive.
160. Microsoft's executives believed that the incentives that its contractual restrictions placed on OEMs would not be sufficient in themselves to reverse the direction of Navigator's usage share. Consequently, in late 1995 or early 1996, Microsoft set out to bind Internet Explorer more tightly to Windows 95 as a technical matter. The intent was to make it more difficult for anyone, including systems administrators and users, to remove Internet Explorer from Windows 95 and to simultaneously complicate the experience of using Navigator with Windows 95. As Brad Chase wrote to his superiors near the end of 1995, "We will bind the shell to the Internet Explorer, so that running any other browser is a jolting experience."
161. Microsoft bound Internet Explorer to Windows 95 by placing code specific to Web browsing in the same files as code that provided operating system functions. Starting with the release of Internet Explorer 3.0 and "OEM Service Release 2.0" ("OSR 2") of Windows 95 in August 1996, Microsoft offered only a version of Windows 95 in which browsing-specific code shared files with code upon which non-browsing features of the operating system relied.
162. The software code necessary to supply the functionality of a modern application or operating system can be extremely long and complex. To make that complexity manageable, developers usually write long programs as a series of individual "routines," each ranging from a few dozen to a few hundred lines of code, that can be used to perform specific functions. Large programs are created by "knitting" together many such routines in layers, where the lower layers are used to provide fundamental functionality relied upon by higher, more focused layers. Some preliminary aspects of this "knitting" are performed by the software developer. The user who launches a program, however, is ultimately responsible for causing routines to be loaded into memory and executed together to produce the program's overall functionality.
163. Routines can be packaged together into files in almost any way the designer chooses. Routines need not reside in the same file to function together in a seamless fashion. Also, a developer can move routines into new or different files from one version of a program to another without changing the functionalities of those routines or the ability to combine them to provide integrated functionality.
164. Starting with Windows 95 OSR 2, Microsoft placed many of the routines that are used by Internet Explorer, including browsing-specific routines, into the same files that support the 32-bit Windows APIs. Microsoft's primary motivation for this action was to ensure that the deletion of any file containing browsing-specific routines would also delete vital operating system routines and thus cripple Windows 95. Although some of the code that provided Web browsing could still be removed, without disabling the operating system, by entering individual files and selectively deleting routines used only for Web browsing, licensees of Microsoft software were, and are, contractually prohibited from reverse engineering, decompiling, or disassembling any software files. Even if this were not so, it is prohibitively difficult for anyone who does not have access to the original, human-readable source code to change the placement of routines into files, or otherwise to alter the internal configuration of software files, while still preserving the software's overall functionality.
165. Although users were not able to remove all of the routines that provided Web browsing from OSR 2 and successive versions of Windows 95, Microsoft still provided them with the ability to uninstall Internet Explorer by using the "Add/Remove" panel, which was accessible from the Windows 95 desktop. The Add/Remove function did not delete all of the files that contain browsing specific code, nor did it remove browsing-specific code that is used by other programs. The Add/Remove function did, however, remove the functionalities that were provided to the user by Internet Explorer, including the means of launching the Web browser. Accordingly, from the user's perspective, uninstalling Internet Explorer in this way was equivalent to removing the Internet Explorer program from Windows 95.
166. In late 1996, senior executives within Microsoft, led by James Allchin, began to argue that Microsoft was not binding Internet Explorer tightly enough to Windows and as such was missing an opportunity to maximize the usage of Internet Explorer at Navigator's expense. Allchin first made his case to Paul Maritz in late December 1996. He wrote:
Allchin followed up with another message to Maritz on January 2, 1997:
If you agree that Windows is a huge asset, then it follows quickly that we are not investing sufficiently in finding ways to tie IE and Windows together. This must come from you. . . . Memphis [Microsoft's code-name for Windows 98] must be a simple upgrade, but most importantly it must be killer on OEM shipments so that Netscape never gets a chance on these systems.
167. Maritz responded to Allchin's second message by agreeing "that we have to make Windows integration our basic strategy" and that this justified delaying the release of Windows 98 until Internet Explorer 4.0 was ready to be included with that product. Maritz recognized that the delay would disappoint OEMs for two reasons. First, while OEMs were eager to sell new hardware technologies to Windows users, they could not do this until Microsoft released Windows 98, which included software support for the new technologies. Second, OEMs wanted Windows 98 to be released in time to drive sales of PC systems during the back-to-school and holiday selling seasons. Nevertheless, Maritz agreed with Allchin's point that synchronizing the release of Windows 98 with Internet Explorer was "the only thing that makes sense even if OEMs suffer."
168. Once Maritz had decided that Allchin was right, he needed to instruct the relevant Microsoft employees to delay the release of Windows 98 long enough so that it could be shipped with Internet Explorer 4.0 tightly bound to it. When one executive asked on January 7, 1997 for confirmation that "memphis is going to hold for IE4, even if it puts memphis out of the xmas oem window," Maritz responded affirmatively and explained,
Thus, Microsoft delayed the debut of numerous features, including support for new hardware devices, that Microsoft believed consumers would find beneficial, simply in order to protect the applications barrier to entry.
169. Allchin and Maritz gained support for their initiative within Microsoft in the early spring of 1997, when a series of market studies confirmed that binding Internet Explorer tightly to Windows was the way to get consumers to use Internet Explorer instead of Navigator. Reporting on one study in late February, Microsoft's Christian Wildfeuer wrote:
Microsoft's survey expert, Kumar Mehta, agreed. In March he shared with a colleague his "feeling, based on all the IE research we have done, [that] it is a mistake to release memphis without bundling IE with it."
170. Microsoft's technical personnel implemented Allchin's "Windows integration" strategy in two ways. First, they did not provide users with the ability to uninstall Internet Explorer from Windows 98. The omission of a browser removal function was particularly conspicuous given that Windows 98 did give users the ability to uninstall numerous features other than Internet Explorer -- features that Microsoft also held out as being integrated into Windows 98. Microsoft took this action despite specific requests from Gateway that Microsoft provide a way to uninstall Internet Explorer 4.0 from Windows 98.
171. The second way in which Microsoft's engineers implemented Allchin's strategy was to make Windows 98 override the user's choice of default browser in certain circumstances. As shipped to users, Windows 98 has Internet Explorer configured as the default browser. While Windows 98 does provide the user with the ability to choose a different default browser, it does not treat this choice as the "default browser" within the ordinary meaning of the term. Specifically, when a user chooses a browser other than Internet Explorer as the default, Windows 98 nevertheless requires the user to employ Internet Explorer in numerous situations that, from the user's perspective, are entirely unexpected. As a consequence, users who choose a browser other than Internet Explorer as their default face considerable uncertainty and confusion in the ordinary course of using Windows 98.
172. Microsoft's refusal to respect the user's choice of default browser fulfilled Brad Chase's 1995 promise to make the use of any browser other than Internet Explorer on Windows "a jolting experience." By increasing the likelihood that using Navigator on Windows 98 would have unpleasant consequences for users, Microsoft further diminished the inclination of OEMs to pre-install Navigator onto Windows. The decision to override the user's selection of non-Microsoft software as the default browser also directly disinclined Windows 98 consumers to use Navigator as their default browser, and it harmed those Windows 98 consumers who nevertheless used Navigator. In particular, Microsoft exposed those using Navigator on Windows 98 to security and privacy risks that are specific to Internet Explorer and to ActiveX controls..
173. Microsoft's actions have inflicted collateral harm on consumers who have no interest in using a Web browser at all. If these consumers want the non-browsing features available only in Windows 98, they must content themselves with an operating system that runs more slowly than if Microsoft had not interspersed browsing-specific routines throughout various files containing routines relied upon by the operating system. More generally, Microsoft has forced Windows 98 users uninterested in browsing to carry software that, while providing them with no benefits, brings with it all the costs associated with carrying additional software on a system. These include performance degradation, increased risk of incompatibilities, and the introduction of bugs. Corporate consumers who need the hardware support and other non-browsing features not available in earlier versions of Windows, but who do not want Web browsing at all, are further burdened in that they are denied a simple and effective means of preventing employees from attempting to browse the Web.
174. Microsoft has harmed even those consumers who desire to use Internet Explorer, and no other browser, with Windows 98. To the extent that browsing-specific routines have been commingled with operating system routines to a greater degree than is necessary to provide any consumer benefit, Microsoft has unjustifiably jeopardized the stability and security of the operating system. Specifically, it has increased the likelihood that a browser crash will cause the entire system to crash and made it easier for malicious viruses that penetrate the system via Internet Explorer to infect non-browsing parts of the system.
175. No technical reason can explain Microsoft's refusal to license Windows 95 without Internet Explorer 1.0 and 2.0. The version of Internet Explorer (1.0) that Microsoft included with the original OEM version of Windows 95 was a separable, executable program file supplied on a separate disk. Web browsing thus could be installed or removed without affecting the rest of Windows 95's functionality in any way. The same was true of Internet Explorer 2.0. Microsoft, moreover, created an easy way to remove Internet Explorer 1.0 and 2.0 from Windows 95 after they had been installed, via the "Add/Remove" panel. This demonstrates the absence of any technical reason for Microsoft's refusal to supply Windows 95 without Internet Explorer 1.0 and 2.0.
176. Similarly, there is no technical justification for Microsoft's refusal to license Windows 95 to OEMs with Internet Explorer 3.0 or 4.0 uninstalled, or for its refusal to permit OEMs to uninstall Internet Explorer 3.0 or 4.0. Microsoft's decision to provide users with an "uninstall" procedure for Internet Explorer 3.0 and 4.0 and its decision to promote Internet Explorer on the basis of that feature demonstrate that there was no technical or quality-related reason for refusing to permit OEMs to use this same feature. Microsoft would not have permitted users to uninstall Internet Explorer, nor would consumers have demanded such an option, if the process would have fragmented or degraded the other functionality of the operating system.
177. As with Windows 95, there is no technical justification for Microsoft's refusal to meet consumer demand for a browserless version of Windows 98. Microsoft could easily supply a version of Windows 98 that does not provide the ability to browse the Web, and to which users could add the browser of their choice. Indicative of this is the fact that it remains possible to remove Web browsing functionality from Windows 98 without adversely affecting non-Web browsing features of Windows 98 or the functionality of applications running on the operating system. In fact, the revised version of Professor Felten's prototype removal program produces precisely this result when run on a computer with Windows 98 installed.
178. In his direct testimony, Felten provides a full technical description of what his prototype removal program does. This description includes a list of the twenty-one methods of initiating Web browsing in Windows 98 that were known to Felten when he developed his program. When the revised version of Felten's program is run on a computer with Windows 98 and no other software installed, Web browsing is not initiated in response to any of these methods.
179. James Allchin tried to show at trial, by way of a videotaped demonstration, that the functionality of Internet Explorer could still be enabled, even after the prototype removal program had been run, by manually adding a new entry to the Windows Registry database. During Felten's rebuttal testimony, one of Microsoft's attorneys directed Felten to perform a second demonstration intended to show that the functionality of Internet Explorer could still be enabled, even after the prototype removal program had been run, by hitting the "control" and "N" keys simultaneously after running the Windows Update feature. Neither of these methods of initiating Web browsing was among the twenty-one documented methods known to Felten when he developed his program. Furthermore, the latter demonstration was hardly a reliable test of Felten's program, because the Encompass shell browser and other applications had been installed on the Windows 98 PC system used in the demonstration. At most, the two demonstrations indicate that Felten did not know all of the methods of initiating Web browsing in Windows 98 when he developed his program, and that he did not include steps in his program to prevent the invocation of Internet Explorer's functionality in response to methods of which he was unaware. Microsoft has special knowledge of its own products, and it alone chooses which functionalities in its products are to be documented and which are to be left undocumented. Felten was aware of this fact, and he himself noted that his own documentation of initiation methods was not exhaustive.
180. Allchin also attempted to show that Felten's program causes performance degradations in Windows 98, as well as malfunctions in certain Windows 98 applications and the Windows Update feature of Windows 98. Those demonstrations, however, were performed on a PC on which several third-party software programs had been installed in addition to Windows 98, and which had been connected to the Internet via a dial-up connection. Felten's program was not intended to be definitive and had not been verified under preconditions other than those for which it was designed. Thus, there was no reason to expect that his program would operate flawlessly during Allchin's demonstrations, and nothing can be inferred from any failure to do so.
181. In fact, the revised version of Felten's program does not degrade the performance or stability of Windows 98 in any way. To the contrary, according to several standard programs used by Microsoft to measure system performance, the removal of Internet Explorer by the prototype program slightly improves the overall speed of Windows 98.
182. Given Microsoft's special knowledge of its own products, the company is readily able to produce an improved implementation of the concept illustrated by Felten's prototype removal program. In particular, Microsoft can easily identify browsing-specific code that could be removed from shared files, thereby reducing the operating system's memory and hard disk requirements and obtaining performance improvements even beyond those achieved by Felten.
183. Microsoft contends that Felten's prototype removal program does not remove Internet Explorer's Web browsing functionalities, but rather "hides" those functionalities from the perspective of the user. In support of that contention, Microsoft points out that Felten's program removes only a small fraction of the code in Windows 98, so that the hard drive still contains almost all of the code that had been executed in the course of providing Internet Explorer's Web browsing functionalities. Some of that code is left on the hard drive because it also supports Windows 98's operating system functionalities. Microsoft did not offer any analytical basis, however, for distinguishing this sharing of code from the code sharing that exists between all Windows applications and the operating system functionalities in Windows 98.
184. While Microsoft's observation suggests that Felten's program does not greatly reduce Windows 98's "footprint" on the hard disk, that point is irrelevant to the question of whether Felten's program removes Internet Explorer's functionalities from Windows 98. This is because the functionalities of a software product are not provided by the mere presence of code on a computer's hard drive. For software code to provide any functionalities at all the code must be loaded into the computer's dynamic memory and executed. To uninstall a software program or to remove a set of functionalities from a software program, it is not necessary to delete all of the software code that is executed in the course of providing those functionalities. It is sufficient to delete and/or modify enough of the program so as to prevent the code in question from being executed.
185. This deletion and modification is precisely what Felten's program does to Windows 98. After Felten's program has been run, the software code that formerly had been executed in the course of providing Web browsing functionalities is no longer executed. Web browsing functionalities are not merely "hidden" from the user. To the contrary, Felten's program deletes and modifies enough of Windows 98 so as to prevent the necessary code from being executed altogether. Since code that is not to be executed does not need to be loaded into memory, Felten's program is able to reduce the memory allocated to Windows 98 by approximately twenty percent.
186. As an abstract and general proposition, many -- if not most -- consumers can be said to benefit from Microsoft's provision of Web browsing functionality with its Windows operating system at no additional charge. No consumer benefit can be ascribed, however, to Microsoft's refusal to offer a version of Windows 95 or Windows 98 without Internet Explorer, or to Microsoft's refusal to provide a method for uninstalling Internet Explorer from Windows 98. In particular, Microsoft's decision to force users to take the browser in order to get the non-Web browsing features of Windows 98, including support for new Internet protocols and data formats is, as Allchin put it, simply a choice about "distribution."
187. As Felten's program demonstrated, it is feasible for Microsoft to supply a version of Windows 98 that does not provide the ability to browse the Web, to which users could add a browser of their choice. Microsoft could then readily offer "integrated" Internet Explorer Web browsing functionality as well, either as an option that could be selected by the end user or the OEM during the Windows 98 setup procedure, or as a "service pack upgrade."
188. Unlike a "pocket part" supplement to a book, a software upgrade need not consist only of new material. A service pack upgrade may install a combination of new software files and/or replacements for existing software files. The use of such service packs to distribute new functionality is a standard feature of Windows applications generally. Microsoft could offer "integrated" Internet Explorer Web browsing functionality as a service pack upgrade that would locate the relevant software and replace it with the current Windows 98 software. In this way, any consumer who wished to do so could easily acquire all of the functionality, features, and performance of the current version of Windows 98 by obtaining the browserless operating system package and the service pack upgrade and then installing them together.
189. Microsoft contends that a service pack must necessarily be deemed part of the operating system when it replaces and adds a large number of core operating system files in the process of upgrading the operating system to a higher level of functionality. This contention is false. Both Microsoft Word, an application program, and Norton Utilities, a suite of utility and application programs, replace and add files to Windows without thereby becoming part of the operating system.
190. Microsoft's actual use of a service pack upgrade to offer integrated Internet Explorer Web browsing functionality (Internet Explorer 4.0) separately from the Windows 95 operating system illustrates the feasibility of this approach. In fact, it produces results remarkably similar to those that could be achieved by offering integrated Internet Explorer Web browsing functionality as a separate service pack upgrade to a browserless Windows 98 operating system. When installed together by the end user, the combined software provides nearly all of the features that Microsoft attributes to the "integrated" design of Windows 98. Of the missing features, all but WebTV for Windows can be obtained by thereafter installing a separately obtained copy of Internet Explorer 5.0. Microsoft has presented no evidence that the WebTV functionality could not easily be included in the stand-alone version of Internet Explorer 5.0.
191. Therefore, Microsoft could offer consumers all the benefits of the current Windows 98 package by distributing the products separately and allowing OEMs or consumers themselves to combine the products if they wished. In fact, operating system vendors other than Microsoft currently succeed in offering "integrated" features similar to those that Microsoft advertises in Windows 98 while still permitting the removal of the browser from the operating system. If consumers genuinely prefer a version of Windows bundled with Internet Explorer, they do not have to be forced to take it; they can choose it in the market.
192. Windows 98 offers some benefits unrelated to browsing that a consumer cannot obtain by combining Internet Explorer with Windows 95. For example, Windows 98 includes support for new hardware technologies and data formats that consumers may desire. While nevertheless preferring to do without Web browsing, Microsoft has forced Windows users who do not want Internet Explorer to nevertheless license, install, and use Internet Explorer to obtain the unrelated benefits. Although some consumers might be inclined to go without Windows 98's new non-browsing features in order to avoid Internet Explorer, OEMs are unlikely to facilitate that choice, because they want consumers to use an operating system that supports the new hardware technologies they seek to sell.
193. Microsoft's argument that binding the browser to the operating system is reasonably necessary to preserve the "integrity" of the Windows platform is likewise specious. First, concern with the integrity of the platform cannot explain Microsoft's original decision to bind Internet Explorer to Windows 95, because Internet Explorer 1.0 and 2.0 did not contain APIs. Second, concern with the integrity of the platform cannot explain Microsoft's refusal to offer OEMs the option of uninstalling Internet Explorer from Windows 95 and Windows 98 because APIs, like all other shared files, are left on the system when Internet Explorer in uninstalled. Third, Microsoft's contention that offering OEMs the choice of whether or not to install certain browser-related APIs would fragment the Windows platform is unpersuasive because OEMs operate in a competitive market and thus have ample incentive to include APIs (including non-Microsoft APIs) required by the applications that their customers demand. Fourth, even if there were some potential benefit associated with the forced licensing of a single set of APIs to all OEMs, such justification could not apply in this case, because Microsoft itself precipitates fragmentation of its platform by continually updating various portions of the Windows installed base with new APIs. ISVs have adapted to this reality by redistributing needed APIs with their applications in order to ensure that the necessary APIs are present when the programs are launched. To the same end, Microsoft makes the APIs it ships with Internet Explorer available to third-party developers for distribution with their own products. Moreover, Microsoft itself bundles APIs -- including those distributed with Internet Explorer -- with a number of the applications that it distributes separately from Windows.
194. Microsoft also contends that by providing "best of breed" implementations of various functionalities, a vendor of a popular operating system can benefit consumers and improve the efficiency of the software market generally, because the resulting standardization allows ISVs to concentrate their efforts on developing complementary technologies for the industry leaders. Microsoft's refusal to offer a version of Windows 98 in which its Web browser is either absent or removable, however, had no such purpose. Rather, it had the purpose and effect of quashing innovation that exhibited the potential to facilitate the emergence of competition in the market for Intel-compatible PC operating systems.
195. Furthermore, there is only equivocal support for the proposition that Microsoft will ultimately prove to be the source of a "best of breed" Web browser. In fact, there is considerable evidence to the contrary. Both Microsoft and the plaintiffs have used product evaluations to support their claims about the relationship between innovations in Web browser technology and consumer choices regarding the use of Web browsers. These product evaluations generally compare Internet Explorer with Navigator by identifying the beneficial and detrimental features of each. Because the evaluations disagree as to which features are most important, there is no consensus as to which is the best browser overall. When read together, the evaluations also do not identify any existing Web browser as being "best of breed" in the sense of being at least as good as all others in all significant respects. Moreover, there is nothing in the evaluations, nor anywhere else in the evidence, to suggest that further innovation efforts by vendors other than Microsoft in the field of Web browser technology are no longer necessary or desirable. To the contrary, many of the product reviews suggest further innovations in both Microsoft and non-Microsoft Web browsers that would benefit consumers.
196. Despite differences in emphasis, the product evaluations do generally concur as to which browser features are beneficial, which browser features are detrimental, and why. Thus, the evaluations provide extensive detailed information about consumer preferences that can be used to predict likely directions in the evolution of Web browser technology.
197. First, the evaluations suggest that, although most Web publishers charge nothing for access to their sites, consumers recognize that there are search and communication costs associated with Web transactions. Accordingly, consumers prefer, and benefit from, innovations in Web browser technology that reduce these costs. Second, consumers recognize that the Web contains a vast and growing range of digital information resources, many of which contain viruses that are capable of causing devastating and irreversible harm to their security and privacy interests. Accordingly, consumers prefer, and benefit from, innovations in Web browser technology that help them identify and avoid harmful Web resources. Third, consumers recognize that they frequently lack adequate information to enable them to assess accurately the costs, risks, and benefits of performing a particular Web transaction. Accordingly, consumers prefer, and benefit from, innovations in Web browser technology that help them assess these costs, risks, and benefits prior to performing the transaction.
198. The reduction of search and communication costs, the identification and avoidance of harmful Web resources, and the provision of more accurate information as to the costs, risks, and benefits of performing Web transactions are just three of the many possible areas of innovation in the field of Web browser technology. Far from demonstrating that Internet Explorer is currently a "best of breed" Web browser, the evidence reveals Microsoft's awareness of the need for continuous improvement of its products. For example, Microsoft frequently releases "patches" to address security and privacy vulnerabilities in Internet Explorer as they are discovered. In sum, there is no indication that Microsoft is destined to provide a "best of breed" Web browser that makes continuing, competitively driven innovations unproductive.
199. Since the World Wide Web was introduced to the public in 1991, the resources available on the Web have multiplied at a near-exponential rate. The Internet is becoming a true mass medium. Every day Web resources are published, combined, modified, moved, and deleted. Millions of individuals and organizations have published Web sites, and Web site addresses are pervasive in advertising, promotion, and corporate identification.
200. The economics of the Internet, along with the flexible structure of Web pages, have made the Web the leading trajectory for the ongoing convergence of mass communications media. Many television and radio stations make some or all of their transmissions available on the Web in the form of static multimedia files or streaming media. Many newspapers, magazines, books, journals, public documents, and software programs are also published on the Web. Multimedia files on the Web have emerged as viable substitutes for many pre-recorded audio and video entertainment products. Web-based E-mail, discussion lists, news groups, "chat rooms," paging, instant messaging, and telephony are all in common use. In addition to subsuming all other digital media, the Web also offers popular interactive and collaborative modes of communication that are not available through other media.
201. The use of Web browsers to conduct Web transactions has grown at pace with the growth of the Web, reflecting the immense value that subsists in the digital information resources that have become available on the Web. Consumer demand for software functionality that facilitates Web transactions, and the response by browser vendors to that demand, creates a market for Web browsing functionality. Although Web browsers are now generally not licensed at a positive price, all Web transactions impose significant costs on consumers, and all browser vendors, including Microsoft, have significant economic interests in maximizing usage of the browsing functionality they control.
202. Since the release of Internet Explorer 1.0 in July 1995, Microsoft has distributed every version of Windows with Internet Explorer included. Consequently, no OEM has ever (with the exception of a few months in late 1997) been able to license a copy of Windows 95 or Windows 98 that has not come with Internet Explorer. Refusing to offer OEMs a browserless (and appropriately discounted) version of Windows forces OEMs to take (and pay for) Internet Explorer, but it does not prevent a determined OEM from nevertheless offering its consumers a different Web browser. Even Microsoft's additional refusal to allow OEMs to uninstall (without completely removing) Internet Explorer from Windows does not completely foreclose a resourceful OEM from offering consumers another browser. For example, an OEM with sufficient technical expertise (which all the larger OEMs certainly possess) could offer its customers a choice of browsers while still minimizing user confusion if the OEM were left free to configure its systems to present this choice the first time a user turned on a new PC system. If the user chose Navigator, the system would automatically remove the most prominent means of accessing Internet Explorer from Windows (without actually uninstalling, i.e., removing all means of accessing, Internet Explorer) before the desktop screen appeared for the first time.
203. If OEMs removed the most visible means of invoking Internet Explorer, and pre- installed Navigator with facile methods of access, Microsoft's purpose in forcing OEMs to take Internet Explorer -- capturing browser usage share from Netscape -- would be subverted. The same would be true if OEMs simply configured their machines to promote Navigator before Windows had a chance to promote Internet Explorer, Decision-makers at Microsoft believed that as Internet Explorer caught up with Navigator in quality, OEMs would ultimately conclude that the costs of pre-installing and promoting Navigator, and removing easy access to Internet Explorer, outweighed the benefits. Still, those decision-makers did not believe that Microsoft could afford to wait for the several large OEMs that represented virtually all Windows PCs shipped to come to this desired conclusion on their own. Therefore, in order to bring the behavior of OEMs into line with its strategic goals quickly, Microsoft threatened to terminate the Windows license of any OEM that removed Microsoft's chosen icons and program entries from the Windows desktop or the "Start" menu. It threatened similar punishment for OEMs who added programs that promoted third-party software to the Windows "boot" sequence. These inhibitions soured Microsoft's relations with OEMs and stymied innovation that might have made Windows PC systems more satisfying to users. Microsoft would not have paid this price had it not been convinced that its actions were necessary to ostracize Navigator from the vital OEM distribution channel.
204. Although Microsoft's original Windows 95 licenses withheld from OEMs permission to implement any modifications to the Windows product not expressly authorized by Microsoft's "OEM Pre-Installation Kit," or "OPK," it had always been Microsoft's practice to grant certain OEMs requesting it some latitude to make modifications not specified in the OPK. But when OEMs began, in the summer of 1995, to request permission to remove the Internet Explorer icon from the Windows desktop prior to shipping their PCs, Microsoft consistently and steadfastly refused. As Compaq learned in the first half of 1996, Microsoft was prepared to enforce this prohibition against even its closest OEM allies.
205. In August 1995, Compaq entered into a "Promotion and Distribution Agreement" with AOL whereby Compaq agreed to "position AOL Services above all other Online Services within the user interface of its Products." An addendum to the agreement provided that Compaq would place an AOL icon -- and no OLS icons not controlled by AOL -- on the desktop of its PCs. Pursuant to its obligations, Compaq began in late 1995 or early 1996 to ship its Presario PCs with the MSN icon removed and the AOL icon added to the Windows desktop. At the same time, Compaq removed the Internet Explorer icon from the desktop of its Presarios and replaced it with a single icon representing both the Spry ISP and the browser product that Spry bundled, i.e., Navigator. Compaq added this icon in part because it recognized Navigator to be the most popular browser product with its consumers; it removed the Internet Explorer icon because it did not want its PCs desktops to confuse novice users with a clutter of Internet-related icons.
206. When Microsoft learned of Compaq's plans for the Presario, it informed Compaq that it considered the removal of the MSN and Internet Explorer icons to be a violation of the OPK process by which Compaq had previously agreed to abide. For its part, AOL informed Compaq that it viewed the addition of an icon for Spry as a violation of their 1995 agreement. AOL did not object to the presence of a Navigator icon; what concerned AOL was the fact that clicking on this icon brought the user to the Spry ISP. Despite the protests from Microsoft and AOL, Compaq refused to reconfigure the Presario desktop. Finally, after months of unsuccessful importunity, Microsoft sent Compaq a letter on May 31, 1996, stating its intention to terminate Compaq's license for Windows 95 if Compaq did not restore the MSN and Internet Explorer icons to their original positions. Compaq's executives opined that their firm could not continue in business for long without a license for Windows, so in June Compaq restored the MSN and IE icons to the Presario desktop.
207. Microsoft did not further condition its withdrawal of the termination notice on the removal of the AOL and Navigator icons; AOL, however, did protest both the continued presence of a Spry icon and the reappearance of the MSN icon. After AOL sent Compaq a formal notice of its intent to terminate the Promotion and Distribution Agreement in September 1996, Compaq removed the Spry/Navigator icon. For reasons discussed below, Compaq did not then replace the Spry/Navigator icon with an icon solely for Navigator.
208. In its confrontation with Compaq, Microsoft demonstrated that it was prepared to go to the brink of losing all Windows sales through its highest-volume OEM partner in order to enforce its prohibition against removing Microsoft's Internet-related icons from the Windows desktop.
209. If the only prohibition had been against removing Microsoft icons and program entries, OEMs partial to Navigator still would have been able to recruit users to Navigator by configuring their PCs to promote it before the Windows desktop first presented itself. This is true because the average user, having chosen a browser product, is indisposed to undergo the trouble of switching to a different one. With the release of Windows 95, some of the high-volume OEMs began to customize the Windows boot sequence so that, the first time users turned on their new PCs, certain OEM-designed tutorials and registration programs, as well as "splash" screens that simply displayed the OEM's brand, would run before the users were presented with the Windows desktop.
210. Promoting non-Microsoft software and services was not the only, or even the primary, purpose of the OEM introductory programs. The primary purpose, rather, was to make the experience of setting up and learning to use a new PC system easier and less confusing for users, especially novices. By doing so, the OEMs believed, they would increase the value of their systems and minimize both product returns and costly support calls. Since just three calls from a consumer can erase the entire profit that an OEM earned selling a PC system to that consumer, OEMs have an acute interest in making their systems self-explanatory and simple to use. A secondary purpose motivating OEMs to insert programs into the boot sequence was to differentiate their products from those of their competitors. Finally, OEMs perceived an opportunity to collect bounties from IAPs and ISVs in exchange for the promotion of their services and software in the boot sequence. Thus, among the programs that many OEMs inserted into the boot sequence were Internet sign-up procedures that encouraged users to choose from a list of IAPs assembled by the OEM. In many cases, a consumer signing up for an IAP through an OEM program would automatically become a user of whichever browser that IAP bundled with its proprietary software. In other cases, the IAP would present the user with a choice of browsers in the course of collecting from the user the information necessary to start a subscription.
211. In addition to tutorials, sign-up programs, and splash screens, a few large OEMs developed programs that ran automatically at the conclusion of a new PC system's first boot sequence. These programs replaced the Windows desktop either with a user interface designed by the OEM or with Navigator's user interface. The OEMs that implemented automatically loading alternative user interfaces did so out of the belief that many users, particularly novice ones, would find the alternate interfaces less complicated and confusing than the Windows desktop.
212. When Gates became aware of what the OEMs were doing, he expressed concern to Kempin, the Microsoft executive in charge of OEM sales. On January 6, 1996, Gates wrote to Kempin: "Winning Internet browser share is a very very important goal for us. Apparently a lot of OEMs are bundling non-Microsoft browsers and coming up with offerings together with Internet Service providers that get displayed on their machines in a FAR more prominent way than MSN or our Internet browser." Less than three weeks later, Kempin delivered his semi-annual report on OEM sales to his superiors. In the report, he identified "Control over start-up screens, MSN and IE placement" as one interest that Microsoft had neglected over the previous six months. The ongoing imbroglio with Compaq was prominent in Kempin's thinking, but he also recognized that establishing control over the boot process was necessary to ensure preferential positioning for MSN and Internet Explorer.
213. In an effort to thwart the practice of OEM customization, Microsoft began, in the spring of 1996, to force OEMs to accept a series of restrictions on their ability to reconfigure the Windows 95 desktop and boot sequence. There were five such restrictions, which were manifested either as amendments to existing Windows 95 licenses or as terms in new Windows 98 licenses. First, Microsoft formalized the prohibition against removing any icons, folders, or "Start" menu entries that Microsoft itself had placed on the Windows desktop. Second, Microsoft prohibited OEMs from modifying the initial Windows boot sequence. Third, Microsoft prohibited OEMs from installing programs, including alternatives to the Windows desktop user interface, which would launch automatically upon completion of the initial Windows boot sequence. Fourth, Microsoft prohibited OEMs from adding icons or folders to the Windows desktop that were not similar in size and shape to icons supplied by Microsoft. Finally, when Microsoft later released the Active Desktop as part of Internet Explorer 4.0, it added the restriction that OEMs were not to use that feature to display third-party brands.
214. The several OEMs that in the aggregate represented over ninety percent of Intel- compatible PC sales believed that the new restrictions would make their PC systems more difficult and more confusing to use, and thus less acceptable to consumers. They also anticipated that the restrictions would increase product returns and support costs and generally lower the value of their machines. Those OEMs that had already spent millions of dollars developing and implementing tutorial and registration programs and/or automatically-loading graphical interfaces in the Windows boot sequence lamented that their investment would, as a result of Microsoft's policy, be largely wasted. Gateway, Hewlett-Packard, and IBM communicated their opposition forcefully and urged Microsoft to lift the restrictions. Emblematic of the reaction among large OEMs was a letter that the manager of research and development at Hewlett-Packard sent to Microsoft in March 1997. He wrote:
Our data (as of 3/10/97) shows a 10% increase in W[indows]95 calls as a % of our total customer support calls . . . .
Our registration rate has also dropped from the mid-80% range to the low 60% range.
There is also subjective data from several channel partners that our system return rate has increased from the lowest of any OEM (even lower than Apple) to a level comparable to the other Microsoft OEM PC vendors. This is a major concern in that we are taking a step backward in meeting customer satisfaction needs.
These three pieces of data confirm that we have been damaged by the edicts that [] Microsoft issued last fall. . . .
From the consumer perspective, we are hurting our industry and our customers. PC's can be frightening and quirky pieces of technology into which they invest a large sum of their money. It is vitally important that the PC suppliers dramatically improve the consumer buying experience, out of box experience as well as the longer term product usability and reliability. The channel feedback as well as our own data shows that we are going in the wrong direction. This causes consumer dissatisfaction in complex telephone support process, needless in-home repair visits and ultimately in product returns. Many times the cause is user misunderstanding of a product that presents too much complexity to the common user. . . .
Our Customers hold HP accountable for their dissatisfaction with our products. We bear [] the cost of returns of our products. We are responsible for the cost of technical support of our customers, including the 33% of calls we get related to the lack of quality or confusion generated by your product. And finally we are responsible for our success or failure in the retail PC market.
We must have more ability to decide how our system is presented to our end users.
If we had a choice of another supplier, based on your actions in this area, I assure you [that you] would not be our supplier of choice.
I strongly urge you to have your executives review these decisions and to change this unacceptable policy.
215. Even in the face of such strident opposition from its OEM customers, Microsoft refused to relent on the bulk of its restrictions. It did, however, grant Hewlett-Packard and other OEMs discounts off the royalty price of Windows as compensation for the work required to bring their respective alternative user interfaces into compliance with Microsoft's requirements. Despite the high costs that Microsoft's demands imposed on them, the OEMs obeyed the restrictions because they perceived no alternative to licensing Windows for pre-installation on their PCs. Still, the restrictions lowered the value that OEMs attached to Windows by the amount of the costs that the restrictions imposed on them. Furthermore, Microsoft's intransigence damaged the goodwill between it and several of the highest-volume OEMs.
216. Microsoft was willing to sacrifice some goodwill and some of the value that OEMs attached to Windows in order to exclude Netscape from the crucial OEM distribution channel. Microsoft's restrictions succeeded in raising the costs to OEMs of pre-installing and promoting Navigator. These increased costs, in turn, were in some cases significant enough to deter OEMs from pre-installing Navigator altogether. In other cases, as is discussed in the next section, OEMs decided not to pre-install Navigator after Microsoft brought still more pressure to bear.
217. Microsoft's license agreements have never prohibited OEMs from pre-installing programs, including Navigator, on their PCs and placing icons and entries for those programs on the Windows desktop and in the "Start" menu. The icons and entries that Microsoft itself places on the desktop and in the "Start" menu have always left room for OEMs to insert more icons and program entries of their own choosing. In fact, Microsoft leaves enough space for an OEM to add more than forty icons to the Windows desktop. Still, the availability of space for added icons did not make including a Navigator icon inexpensive for OEMs. Given the unavoidable presence of the Internet Explorer and MSN icons, adding a Navigator icon would increase the amount of Internet-related clutter on the desktop. This would lead to confusion among novice users, which would in turn increase the incidence of support calls and product returns. Microsoft made this very point clear to OEMs in its attempts to persuade them not to pre-install Navigator on their PCs. Furthermore, OEMs recognized that including multiple Navigator icons in an attempt to draw users' attention away from Internet Explorer would only increase the amount of clutter on the desktop, thus adding to user confusion. Although the Windows 98 OEM license does not forbid the OEM to set Navigator as the default browsing software, doing so would fail to forestall user confusion since, as the Court found in the previous section, Windows 98 launches Internet Explorer in certain situations even if Navigator is set as the default.
218. The restrictions on modifying the Windows boot sequence, including the prohibition against automatically loading alternate user interfaces, deprived OEMs of the principal devices by which to lure users to Navigator over the high-profile presence of Internet Explorer in the Windows user interface. An OEM remained free to place an icon on the desktop that a user could click to invoke an alternate user interface. Plus, once invoked, the interface could be configured to load automatically the next time the PC was turned on. This mode of presentation proved to be much less effective than the one Microsoft foreclosed, however, for studies showed that users tended not to trouble with selecting an alternate user interface; they were content to use the interface that loaded automatically the first time they turned on their PCs. Furthermore, while Microsoft's restrictions never extended to the interval between the time when the PC was turned on and the time when Windows began loading from the hard drive into RAM, developing anything more complicated than a simple splash screen to run in that period would have involved, at a minimum, the writing of a DOS utility and, at the maximum, the pre-installation of a second operating system. Such measures were simply not worth the cost. Finally, although the Windows 98 license does not prohibit an OEM from including on the keyboard of its PCs a button that takes users directly to an OEM-maintained site containing promotion for Navigator, such a configuration is extremely costly for an OEM to implement, and it represents a less effective form of promotion than automatically advertising Navigator in the initial boot process.
219. In the spring of 1998, Microsoft began gradually to moderate certain of the restrictions described above. The first sign of relaxation came when Microsoft permitted some fifty OEMs to include ISPs of their choice in Microsoft's Internet Connection Wizard. Then, in late May and early June 1998, Microsoft informed seven of the highest-volume OEMs that it was granting them the privilege of inserting their own registration and Internet sign-up programs into the initial Windows 98 boot sequence. If the user selected an IAP using the OEM program, Microsoft's Internet Connection Wizard would not run in the boot sequence. Microsoft subsequently extended these same privileges to several other OEMs, upon their request.
220. It is important to note that Microsoft's tractability emerged only after the restrictions had been in place for over a year, and only after Microsoft had managed to secure favorable promotion for Internet Explorer through the most important IAPs. Furthermore, while Microsoft permitted the OEMs to include in their registration and sign-up programs promotions for their own products (including OEM-branded shell browsers built upon Internet Explorer) and for ISPs (but only if and when those ISPs were selected by consumers in the sign-up process), Microsoft continued to prohibit promotions for any other non-Microsoft products, including Navigator. In a single exception, Microsoft granted Gateway's request that it be permitted to give consumers who used Gateway's sign-up process and selected Gateway.net as their ISP an opportunity to choose Navigator as their browser. Microsoft granted this permission orally, and it did not extend similar privileges to any other OEMs.
221. Microsoft asserts that the restrictions it places on the ability of OEMs to modify the Windows desktop and boot sequence are merely intended to prevent OEMs from compromising the quality and consistency of Windows after the code leaves Microsoft's physical control, but before PC consumers first begin to experience the product. In truth, however, the OEM modifications that Microsoft prohibits would not compromise the quality or consistency of Windows any more than the modifications that Microsoft currently permits. Furthermore, to the extent that certain OEM modifications did threaten to impair the quality and consistency of Windows, Microsoft's response has been more restrictive than necessary to abate the threat. Microsoft would not have imposed prohibitions that burdened OEMs and consumers with substantial costs, lowered the value of Windows, and harmed the company's relations with major OEMs had it not felt that the measures were necessary to maximize Internet Explorer's share of browser usage at Navigator's expense.
222. Microsoft asserts that it restricts the freedom of OEMs to remove icons, folders, or "Start" menu entries that Microsoft places on the Windows desktop in order to ensure that consumers will enjoy ready access to the features that Microsoft's advertising has led them to expect. The Windows trademark would be blemished, Microsoft argues, if consumers could not easily find the features that impelled them to purchase a Windows-equipped PC. At the same time that it has put forward this justification, however, Microsoft has permitted OEMs to de-activate Microsoft's Active Desktop and its associated "channels" prior to shipment. More significant is the fact that Microsoft's license agreements require OEMs to bear product support costs. So if a consumer has difficulty locating a feature that he wants to use, he will call a customer service representative employed by the OEM that manufactured his PC. Since only a few calls erase the profit earned from selling a PC system, OEMs are loathe to do anything that will lead to consumer questions and complaints. Therefore, if market research indicates that consumers want and expect to see a certain icon on the Windows desktop, OEMs will not remove it. Since OEMs share Microsoft's interest in ensuring that consumers can easily find the features they want on their Windows PC systems, Microsoft would not have prohibited OEMs from removing icons, folders, or "Start" menu entries if its only concern had been consumer satisfaction. In fact, by forbidding OEMs to remove the most obvious means of invoking Internet Explorer, Microsoft diminished the value of Windows PC systems to those corporate customers, for example, who did not intend for their employees to browse the Web and did not want a browser taking up hardware resources. Incidentally, there is no merit in the hypothesis that OEMs might cause problems in the functioning of the rest of Windows by removing Internet Explorer's desktop icon and program entry, because Microsoft still allows users to do exactly that.
223. According to Microsoft, its restrictions on the ability of OEMs to insert programs into the initial Windows boot sequence are meant to ensure that all Windows users experience the product the way Microsoft intended it the first time they turn on their PC systems; after all, there would be little incentive to develop a high-quality operating-system product if OEMs were free to alter it for the worse before handing it over to consumers. This argument might be availing were it not for the fact that Microsoft currently allows several of the largest-volume OEMs to make major modifications to the initial Windows 98 boot sequence. Microsoft permits each of these OEMs to configure its own splash screens, tutorials, registration wizards, Internet sign-up wizards, and utilities so that they run automatically when the consumer first turns on a new PC system. Either Microsoft stopped caring about the consistency of the Windows experience in 1998, when it tempered its restrictions on modifications to the boot sequence, or preserving consistency was never Microsoft's true motivation for imposing those restrictions in the first place. With all the variety that Microsoft now tolerates in the boot sequence, including the promotion of OEM-branded browser shells, it is difficult to comprehend how allowing OEMs to promote Navigator in their tutorials and Internet sign-up programs would further compromise Microsoft's purported interest in consistency.
224. Although Microsoft has tolerated a variety of OEM modifications to the Windows boot sequence, it has never acquiesced to an alternate user interface that automatically obscures the Windows desktop after the PC system has finished booting for the first time. In demanding the removal of such automatically loading user interfaces, Microsoft has postulated that consumers who purchase Windows PCs expect to see the Windows desktop when their PC systems finish booting for the first time. If consumers instead see a different user interface, they will be confused and disappointed. What is more, Microsoft asserts, OEM shells have tended to be of lower quality than Windows. One OEM's version allegedly even disabled the ability of a Windows user to invoke functionality by clicking the right button of his mouse.
225. The alternate shells that OEMs have developed may or may not be of lower quality than Windows. One thing is clear, however: If an OEM develops a shell that users do not like as much as Windows, and if the OEM causes that shell to load as the default user interface the first time its PCs are turned on, consumer wrath will fall first upon the OEM, and demand for that OEM's PC systems will decline commensurately with the resulting user dissatisfaction. The market for Intel-compatible PCs is, by all accounts, a competitive one. Consequently, any OEM that tries to force an unwanted, low-quality shell on consumers will do so at its own peril. Had Microsoft's sole concern been consumer satisfaction, it would have relied more on the power of the market -- and less on its own market power -- to prevent OEMs from making modifications that lead to consumer disappointment.
226. At times, Microsoft has argued that the limitations it imposes on the ability of OEMs to modify Windows originate in a desire to prevent its platform from becoming fragmented, like UNIX. Microsoft believes that ISVs benefit from the fact that Windows presents the same platform for applications development, irrespective of the underlying hardware. Certainly, Microsoft has a legitimate interest in ensuring that OEMs do not take Windows under license, alter its API set, and then ship the altered version. This fact does not add credibility to Microsoft's stated justification, though, for two reasons. First, Microsoft itself creates some degree of instability in its supposedly uniform platform by releasing updates to Internet Explorer more frequently than it releases new versions of Windows. As things stand, ISVs find it necessary to redistribute Microsoft's Internet-related APIs with their applications because of nonuniformity that Microsoft has created in its own installed base. More important, however, is the fact that none of the modifications that OEMs are known to have proposed making would have removed or altered any Windows APIs.
227. To the extent Microsoft is apprehensive that OEMs might, absent restrictions, change the set of APIs exposed by the software on their PCs, the concern is not that OEMs would modify the Windows API set. Rather, the worry is that OEMs would pre-install, on top of Windows, other software exposing additional APIs not controlled by Microsoft. In the case of alternate user interfaces, Microsoft is fearful that, if these programs loaded automatically the first time users turned on their PCs, the programs would attract so much usage that developers would be encouraged to take advantage of any APIs that the programs exposed. Indeed, one user interface in particular that OEMs could configure to load automatically and obscure the Windows desktop -- Navigator -- exposes a substantial number of APIs. Therefore, Microsoft's real concern has not been that OEM modifications would fragment the Windows platform to the detriment of developers and consumers. What has motivated Microsoft's prohibition against automatically loading shells is rather the fear -- once again -- that OEMs would pre-install and give prominent placement to middleware that could weaken the applications barrier to entry.
228. Like most other software products, Windows 95 and Windows 98 are covered by copyright registrations. Since they are copyrighted, Microsoft distributes these products to OEMs pursuant to license agreements. By early 1998, Microsoft had made these licenses conditional on OEMs' compliance with the restrictions described above. Notwithstanding the formal inclusion of these restrictions in the license agreements, the removal of the Internet Explorer icon and the promotion of Navigator in the boot sequence would not have compromised Microsoft's creative expression or interfered with its ability to reap the legitimate value of its ingenuity and investment in developing Windows. More generally, the contemporaneous Microsoft documents reflect concern with the promotion of Navigator rather than the infringement of a copyright. Also notable is the fact that Microsoft did not adjust its OEM pricing guidelines when it lifted certain of the restrictions in the spring of 1998.
229. Finally, it is significant that, while all vendors of PC operating systems undoubtedly share Microsoft's stated interest in maximizing consumer satisfaction, the prohibitions that Microsoft imposes on OEMs are considerably more restrictive than those imposed by other operating system vendors. For example, Apple allows its retailers to remove applications that Apple has pre-installed and to reconfigure the Mac OS desktop. For its part, IBM allows its OEM licensees to override the entire OS/2 desktop in favor of a customized shell or to set an application to start automatically the first time the PC is turned on. The reason is that these firms do not share Microsoft's interest in protecting the applications barrier to entry.
230. Microsoft's restrictions on modifications to the boot sequence and the configuration of the Windows desktop ensured that every Windows user would be presented with ready means of accessing Internet Explorer. Although the restrictions also raised the costs attendant to pre-installing and promoting Navigator, senior executives at Microsoft were not confident that those higher costs alone would induce all of the major OEMs to focus their promotional efforts on Internet Explorer to the exclusion of Navigator. Therefore, Microsoft used incentives and threats in an effort to secure the cooperation of individual OEMs.
231. First, Microsoft rewarded with valuable consideration those large-volume OEMs that took steps to promote Internet Explorer. For example, Microsoft gave reductions in the royalty price of Windows to certain OEMs, including Gateway, that set Internet Explorer as the default browser on their PC systems. In 1997, Microsoft gave still further reductions to those OEMs that displayed Internet Explorer's logo and links to Microsoft's Internet Explorer update page on their own home pages. That same year, Microsoft agreed to give OEMs millions of dollars in co-marketing funds, as well as costly in-kind assistance, in exchange for their carrying out other promotional activities for Internet Explorer.
232. Microsoft went beyond giving OEMs incentives to promote Internet Explorer. The company's dealings with Compaq in 1996 and 1997 demonstrate that Microsoft was willing to exchange valuable consideration for an OEM's commitment to curtail its distribution and promotion of Navigator. In early 1996, at around the same time that Compaq was removing the MSN and Internet Explorer icons and program entries from the Presario desktop, Compaq announced its intention to work with Netscape for its internal Internet needs and on Internet server initiatives. In response, Microsoft insisted that Compaq support Microsoft's Internet initiatives throughout its business. To make its displeasure felt, Microsoft initiated a series of cooperative ventures with some of Compaq's competitors, including DEC and Hewlett-Packard.
233. When Compaq eventually agreed to restore the MSN and Internet Explorer icons and program entries to the Presario desktop, it did so because its senior executives had decided that the firm needed to do what was necessary to restore its special relationship with Microsoft. On May 13, 1996, Compaq signed an addendum extending the firms' Frontline Partnership to the realm of network-related products. Pursuant to the addendum, Compaq agreed to ship Internet Explorer as the default browser product on all of its desktop and server systems, to adopt and promote Internet Explorer internally, and to focus the majority of Compaq's key network-oriented announcements and marketing activities on Microsoft's technologies and strategy. In September of the same year, Compaq agreed to offer Internet Explorer as the preferred browser product for its Internet products and to use two or more of Microsoft's hypertext markup language ("HTML") extensions in the home page for each of those products. Then in February 1997, Compaq committed itself to promote Internet Explorer exclusively for its PC products in exchange for Microsoft's agreement to pay Compaq a bounty for each user that signed up for Internet access using a Compaq PC. Despite the view of some within Compaq that the firm's goal should be "to feature the brand leader Netscape," Compaq elected not to resume the pre- installation of Navigator on its Presario PCs after it removed the joint Spry/Navigator icon. In fact, Compaq stopped pre-installing Navigator on all but very small percentage of its PCs.
234. In return for Compaq's capitulation and revival of its commitment to support Microsoft's Internet strategy, Microsoft has guaranteed Compaq that the prices it pays for Windows will continue to be significantly lower than the prices paid by other OEMs. Specifically, the operating system licenses signed by Compaq and Microsoft in March 1998 gave Compaq "[g]uaranteed better" pricing than any other OEM for Windows 95, Windows 98, and Windows NT Workstation (versions 4 and 5) until April 2000. Compaq's license fee for Windows is so low that other OEMs would still pay substantially more than Compaq even if they qualified for all of the royalty reductions listed in Microsoft's Market Development Agreements ("MDAs"). What is more, while Microsoft requires other OEMs to verify actual compliance with particular milestones in order to receive Windows 98 royalty reductions, Microsoft has secretly agreed to provide the full amount of those discounts to Compaq regardless of whether it actually satisfies the specified conditions. In addition to a guaranteed most-favorable price on Windows, Compaq has enjoyed free internal use of all Windows products for PCs since March 1998.
235. Microsoft's relations with Compaq beginning in late 1996 illustrate the blandishments that Microsoft is willing to extend to OEMs that ally with it to help it capture browser share. Microsoft's relations with Gateway and the IBM PC Company, by contrast, reveal the pressure that Microsoft is willing to apply to OEMs that show reluctance to cooperate on this front.
236. In February 1997 a Microsoft account representative told his counterpart at Gateway that Gateway's use of Navigator on its own corporate network was a serious issue at Microsoft. He added that Microsoft would not do any co-marketing and sales campaigns with Gateway if the firm appeared to be anything but pro-Microsoft. If Gateway would replace Navigator with Internet Explorer, Microsoft would compensate Gateway for its investment in Netscape's product. If Gateway refused, Microsoft might be compelled to audit Gateway's internal use of Microsoft products. Gateway was separately told by Microsoft representatives that its decision to ship Navigator with its PCs could affect its business relationship with Microsoft. Despite the pressure from Microsoft, Gateway refused to switch its internal use to Internet Explorer or to stop shipping Navigator with its PCs. Although Microsoft did not implement its more specific threats, Gateway has consistently paid higher prices for Windows than its competitors. Microsoft's actions not only corroborate the evidence of its interest in suppressing the usage of Navigator, they also demonstrate its ability to threaten recalcitrant customers without losing their business.
237. Similarly, in early 1997, Microsoft tried to convince the IBM PC Company to promote and distribute the upcoming release of its new browser, Internet Explorer 4.0. At a meeting with IBM executives in March 1997, Microsoft representatives threatened that, if IBM did not pre-load and promote Internet Explorer 4.0 to the exclusion of Navigator on its PCs, it would suffer "MDA repercussions." One of the Microsoft representatives in attendance, Bengt Ackerlind, stated that in return for IBM shipping its systems without any software that competed with Microsoft, IBM would receive "soft dollars," marketing assistance, improved access to the source code of Windows 95 and Microsoft's BackOffice product, and the ability to self-certify for Microsoft's Windows Hardware Quality Lab provisions. In a follow-up meeting three weeks later, Microsoft representatives again insisted that IBM distribute and promote Internet Explorer exclusively and again offered soft dollars, marketing assistance, and MDA reductions in return. Later that day, in a smaller meeting that Microsoft referred to as "secret discussions," Ackerlind stated Microsoft's desire that IBM promote Internet Explorer 4.0 exclusively and warned that if IBM pre-installed Navigator on its PCs, "We have a problem."
238. The IBM PC Company refused to promote Internet Explorer 4.0 exclusively, and it has continued to pre-install Navigator on its PCs. The difference in the ways that Compaq and IBM responded to Microsoft's Internet-related overtures in 1996 and 1997 contributed to the stark contrast in the treatment the two firms have since received from Microsoft.
239. Microsoft has largely succeeded in exiling Navigator from the crucial OEM distribution channel. Even though a few OEMs continue to offer Navigator on some of their PCs, Microsoft has caused the number of OEMs offering Navigator, and the number of PCs on which they offer it, to decline dramatically. Before 1996, Navigator enjoyed a substantial and growing presence on the desktop of new PCs. Over the next two years, however, Microsoft's actions forced the number of copies of Navigator distributed through the OEM channel down to an exiguous fraction of what it had been. By January 1998, Kempin could report to his superiors at Microsoft that, of the sixty OEM sub-channels (15 major OEMs each offering corporate desktop, consumer/small business, notebook, and workstation PCs), Navigator was being shipped through only four. Furthermore, most of the PCs shipped with Navigator featured the product in a manner much less likely to lead to usage than if its icon appeared on the desktop. For example, Sony only featured Navigator in a folder rather than on the desktop, and Gateway only shipped Navigator on a separate CD-ROM rather than pre-installed on the hard drive. By the beginning of January 1999, Navigator was present on the desktop of only a tiny percentage of the PCs that OEMs were shipping.
240. To the extent Netscape is still able to distribute Navigator through the OEM channel, Microsoft has substantially increased the cost of that distribution. Although in January 1999 (in the midst of this trial), Compaq suddenly decided to resume the pre-installation of Navigator on its Presario PCs, Compaq's reversal came only after Netscape agreed to provide Compaq with approximately $700,000 worth of free advertising.
241. In sum, Microsoft successfully secured for Internet Explorer -- and foreclosed to Navigator -- one of the two distribution channels that leads most efficiently to the usage of browsing software. Even to the extent that Navigator retains some access to the OEM channel, Microsoft has relegated it to markedly less efficient forms of distribution than the form vouchsafed for Internet Explorer, namely, prominent placement on the Windows desktop. Microsoft achieved this feat by using a complementary set of tactics. First, it forced OEMs to take Internet Explorer with Windows and forbade them to remove or obscure it -- restrictions which both ensured the prominent presence of Internet Explorer on users' PC systems and increased the costs attendant to pre-installing and promoting Navigator. Second, Microsoft imposed additional technical restrictions to increase the cost of promoting Navigator even more. Third, Microsoft offered OEMs valuable consideration in exchange for commitments to promote Internet Explorer exclusively. Finally, Microsoft threatened to penalize individual OEMs that insisted on pre-installing and promoting Navigator. Although Microsoft's campaign to capture the OEM channel succeeded, it required a massive and multifarious investment by Microsoft; it also stifled innovation by OEMs that might have made Windows PC systems easier to use and more attractive to consumers. That Microsoft was willing to pay this price demonstrates that its decision-makers believed that maximizing Internet Explorer's usage share at Navigator's expense was worth almost any cost.