Microsoft's Overreaching Definition of Operating System
Corporations, Securities & Antitrust Practice Group Newsletter - Volume 2, Issue 1, Spring 1998
May 1, 1998James Packard Love
Microsoft's Overreaching Definition of Operating System1
Microsoft's core product is the software operating system ("OS") Windows.2 A classical definition of an OS is a software program that "controls the operation of a computer and directs the processing of programs ... by assigning storage space in memory and controlling input and output functions."3 Microsoft's definition is far more expansive. Microsoft's Executive Vice President, Steve Ballmer, refers to the "Windows Experience." According to his paradigm, Windows is more than a collection of device drivers and a traffic cop for applications; it is a user interface, a set of links to publishers and electronic commerce, Internet navigation tools and additional functions that Microsoft chooses to include.
In a competitive market, few would care how a company defined its operating system. But when one firm has a 90 percent market share worldwide, many care.
If computer manufacturers "need" the OS, they can be forced to purchase or distribute products that are tied to (or components of) the OS. The appropriate definition of the OS thus has become a battleground in antitrust policy.
A computer OS is a collection of files. The computer I work on at home uses Windows 95. Every time I install a new application, my Windows directory grows. At present, I have 2,658 files in 145 folders, taking up 187 megabytes of space in the c:\Windows directory. The crucial c:\Windows\System subdirectory has 911 files, taking up 96 megabytes of disk space.
The U.S. Department of Justice ("DOJ") is currently litigating Microsoft’s efforts to expand the definition of the OS to include the entire range of applications which are distributed as Microsoft Internet Explorer ("MSIE"). In the MSIE4 release, these applications include, besides the Internet browser itself, computer "push" technology for publishing and electronic commerce, electronic mail clients, news group readers and dozens of multimedia and Internet add-ons. These additional applications and add-ons alone use nearly as much disk space as the basic Windows 95 files.
Depending upon what is happening in the antitrust case, Microsoft makes different claims about which files belong to the OS, and which files belong to particular applications. In December 1997, Microsoft claimed that several key files once distributed as Windows files were now part of MSIE, and that the OS would not work properly unless MSIE was installed. These claims flow from Microsoft's strategy of "spreading" the OS over key applications, thus requiring consumers to purchase the applications in addition to the standard OS.
Some of Microsoft’s competitors say this strategy is used with Microsoft Office, the suite of office productivity programs which includes such products as MS Word, Excel and Powerpoint. The MS Office CD installs 40 to 50 megabytes of files in the c:\Windows directory. Some of these files are updated versions of files first distributed with the OS. Some are new files which extend the general functionality of the OS, and enhance performance.
Windows is constantly updated, and Microsoft can modify the OS to make its applications work better or to make rivals products work poorly.4 Some of the changes in the OS are undocumented and only known to Microsoft’s developers. Even when Microsoft does disclose information about the code, Microsoft’s developers are the first to know what has changed, what will change in the future and the first to have the rights to distribute the changes on shipped products. Under Microsoft’s management, the OS is more like shifting sand or quicksand than a "level playing field."
There are countless other ways that Microsoft has been able to "leverage" the OS into neighboring markets, and to then leverage these products into even more markets. Software experts believe Microsoft is designing its NT server technology to give Microsoft’s browser clients an edge in performance. Microsoft is requiring companies who want certain Microsoft licenses to deploy proprietary technologies on Web pages, which will only work with Microsoft products. Microsoft takes lackluster products and bundles them with essential products, or gives them away, to deprive rivals of revenues, deter entry and, ultimately, to become the only product that receives continuous R&D.
Many of these problems are not unique to Microsoft. For years IBM engaged in similar tactics in mainframe computer markets. Netscape’s decision to bundle an email client with its browser had a negative effect on Eurdora and other email products. Netscape also engaged in an embrace and extend strategy toward HTML standards that is similar to Microsoft’s current efforts to destroy Java as a cross platform development tool. Ashton Tate, before it was acquired by Borland, tried to prevent competitors from developing command compatible database products. While Microsoft has a well deserved reputation as an innovator in the area of anticompetitive actions, it has also learned from its rivals.
Many of the problems which bedevil the software industry are related to the fact that computer software and hardware products need to interoperate with each other. Given the opportunity, firms will manipulate interoperability issues as weapons in competitive struggles.
Consumers obtain utility from interoperability. When markets are competitive, firms are sometimes driven by consumer preferences to include features which enhance interoperability, such as support for public domain HTML code or FTP protocols. The Internet was built upon a commitment to promote interoperability across different hardware and software platforms and it triumphed over rival proprietary networks which lacked interoperability features. But firms sometimes deliberately limit interoperability for strategic purposes.5
A number of economists are calling attention to factors which give rise to increasing returns to scale in high technology markets, including software markets. Not only are the costs of software development largely insensitive to the number of units shipped, but there are network effects which make market leaders good choices for consumers, creating positive feedback mechanisms, which can lead to highly concentrated markets. The domination of VHS over Beta is in part a function of the fact that it is more efficient for video stores to stock titles in a single format. This would be true even if the technical merits of the two systems were exactly equal. The fact that most computer users have chosen Windows 95 is a good reason for new buyers to choose it also because it will facilitate training users, maintaining software and sharing data. Software developers like to develop products which run on the largest installed base, and consumers like to have computers which support the largest selection of software titles.
The trend toward a dominant OS, Word Processor, browser or any number of software products would be driven by the efficiencies of standardization. Of course, interoperability and other network effects don’t have to be based upon a private monopoly. The success of the Internet, which was profoundly anti-monopoly, is a very important counter example.
In the new antitrust theory for these high technology markets, policy makers are being told that high market shares in a given product, like an OS or a Word Processor, are not necessarily problematic. However, antitrust policy should seek to limit a firm’s ability to leverage success in one market into an unfair advantage in neighboring markets and to avoid high barriers for new market entrants, thereby making the markets more contestable.
Next is the thorny problem of actually crafting rules that prevent such unfair leveraging without penalizing a company that has justly earned market share. Prohibitions on tying or the use of anticompetitive licensing practices is one part of the solution. Conduct rules regarding interoperability have never been applied to Microsoft, but they proved useful in making the areas of computer mainframes and telecommunications networks more competitive. If conduct rules are ineffective, divestitures (or structural separation) may be called for, as in the case of AT&T. It is worth noting that following the imposition of important conduct rules regarding interoperability for IBM and the break-up of AT&T, markets became more innovative, and both companies (including all components of AT&T) prospered.
The difficulty in the software area is that product cycles are very short and product definitions themselves often change. Moreover, consumers often derive utility from bundled and integrated products and they expect constant evolution of product features. These software-specific concerns are real and they should be addressed.
In some respects, the bundling issues are the most difficult because of the judgments that must be made regarding product definitions and functionality. Such obstacles are not insurmountable, however, particularly if the government chooses its targets carefully. Certainly there is no reason why MSIE can’t be distributed as a separate product, for example. One could envision unbundling of certain components of Microsoft Office, particularly because it has always been possible to separately install or uninstall each component. Indeed, many of the benefits of product integration can be retained, even if the products themselves are distributed separately. As software products are increasingly distributed over the Internet, consumers should find it ever easier to select and install individual products.
Remedies regarding interoperability should be, in our view, a high priority for government antitrust authorities. These could include requirements for timely disclosure of the Application Programming Interface ("API") to competitors, reasonable licensing terms for key technologies, disclosure of important data format information, requirements to support certain open standards and other factors which make it easier for competitors products to interoperate.
The target of such remedies should be products which by their nature, provide a platform for other products and which are essential platforms for developers. Certainly, Windows 95 and Window 98 are such products. The approach, however, should be more general. If Netscape had succeeded in monopolizing the browser market, these remedies would have prevented Netscape from engaging in the type of anticompetitive actions that Microsoft is often accused of with respect to the desktop OS.
There should also be triggers which lift conduct rules as firms face increased competition, and products become "less essential". For example, the FCC once considered AT&T a "dominant" carrier, requiring notice of changes in tariffs and other obligations. The dominant carrier status was eliminated as competition in long distance markets matured.
These suggestions, which present creative challenges to government regulators and company counsel alike, are preliminary efforts to think about a future in which software increasingly plays an important and inseparable role in our economy.
1. Microsoft's revenues in fiscal years 1995 through 1997 were $5.9, $8.7 and $11.4 billion respectively, an increase of 91 percent in two years. Microsoft is growing at an astonishing rate, but there are many larger companies. In 1996, General Motors reported revenues of $164 billion, nearly 19 times greater than Microsoft, and IBM, the world's largest computer company, reported revenues of $76 billion, nearly 9 times as high as Microsoft. Indeed, in 1996, Apple computer's revenues of $9.8 billion were still larger than Microsoft.
2. Windows includes Windows 3.1, Windows 95 and Windows NT.
3. This 1996 definition is from Webster's online dictionary. See also the Free On-line Dictionary of Computing, which describes as OS as "the low-level software which schedules tasks, allocates storage, handles the interface to peripheral hardware and presents a default interface to the user when no application program is running," or www.pewebopaedia.com, which describes the OS as "the most important program that runs on a computer...[to] perform basic tasks, such as recognizing input from the keyboard, sending output to the display screen, keeping track of files and directories on the disk, and controlling peripheral devices such as disk drives and printers."
4. Such as gave rise to the adage, "DOS isn’t done until Lotus won’t run."
5. When Microsoft Word was a minor player in the word processing market, it was designed to import and export files easily with WordPerfect, the market leader. Today Microsoft provides poor translation of WordPerfect documents, and even creates backward compatibility problems with older versions of Microsoft’s own word processor, in order to make WordPerfect less useful, and to promote upgrades of Microsoft’s older word processors.
James Packard Love is the director of the Consumer Project on Technology (http://www.cptech.org), a non-profit group founded by Ralph Nader that seeks to promote consumer awareness of technology’s impact on consumers and daily life.