For the past three years, the Department of Justice has been involved in antitrust litigation against Microsoft, the world’s largest software company. Because Microsoft's founder is the world’s richest man, and because Microsoft has achieved an unprecedented role in shaping the computer industry, this case has captured the attention of the popular press and mass media.

In previous work (Fagin 1997), I have made the argument that politics as practiced today is rife with irrationality, and that scientists should become more politically active because irrationality in politics is harmful. I believe the Microsoft trial is a case in point. Computer scientists in particular should exert what pressure they can to call for the dismissal of this case. It is an affront to rational thinking.

Axioms of Economics

Rational inquiry into the physical world requires self-evident axioms. For scientific study, these axioms include notions of regularity (observed physical constants, properties, and laws do not change with time) and objectivity (there is a world outside of us that we perceive but do not create). The fact that they are regarded as self-evident, however, does not mean they are taken on faith. These axioms have been observed to be true throughout time since the development of the scientific method, but honest scientists would admit that they are at least in principle falsifiable, and should be abandoned in the face of evidence to the contrary.

Similarly, any study of human action (economics) ought to include the principle of mutual benefit: honest, consensual transactions between mentally competent adults leave both parties better off than they were before. This axiom is not a religious statement, but one formed from direct and verifiable experience. Economists should abandon it if faced with evidence that people were in general poor judges of their own welfare, that some people had better knowledge of what others really wanted, that people normally hand over important decisions about their lives to others, and so forth. To date, such evidence is not forthcoming.

Economics also has a second axiom: the principle of scarcity. Human wants are as infinite as human imagination, but the resources available to humans are not. Accordingly, human beings must make decisions about the allocation of resources in the presence of imperfect knowledge, since at any given time our state of knowledge about the world is suboptimal. Thanks to science and human ingenuity, our knowledge of the world and our ability to discover how to satisfy human wants improves with time. However, because of scarcity, human beings face important tradeoffs on how to reallocate resources in the face of new knowledge.

These principles have important implications for public policy, and for the Microsoft case in particular. A skeptical reading of the Department of Justice’s complaint against Microsoft suggests a lack of awareness of these principles, as well as contradictory premises from which anything can be deduced. These difficulties arise from trying to apply the contradictions and irrationality of the 1890 Sherman Antitrust Act to computer science, a field that demands strict rationality of its practitioners and that could not have been envisioned a century ago. Let us see why this is so.

Background

Operating systems are required for any computer to be usable by a non-specialist. The most popular operating system is Microsoft Windows, currently released as Windows 98. Other operating systems include the MacOS (found in all Macintosh computers), Unix (a public domain system with its roots in academic and industrial research), and IBM’s OS/2. Differences between operating systems are the primary sources of differences between computers in the marketplace. It is the operating system, more than anything else, that is responsible for the characteristic "look and feel" of a particular computer.

But if look and feel were all that mattered, Microsoft would still be a struggling startup and Bill Gates would be just another college droupout. Operating systems also provide standardization, the ability to run the same program on different computers. Standardization is extremely important to both producers and consumers of software. Accordingly, it is valued highly in the marketplace, and there is a lot of money to be made in providing it.

How Operating Systems Provide Standardization

In addition performing tasks that users work with directly, the operating system provides important functions that are invisible to users but are relied upon by application programs (like Netscape Communicator, Word, and Doom). Operating systems provide literally hundreds of services that these application programs can use: drawing pictures, printing characters on the screen, providing standard question and answer boxes for interacting with users, and so forth.

These services provide two important functions. First, they reduce the time it takes to write programs. Software engineers only have to write code that is unique to their application. It’s the same advantage architects enjoy when designing a home: they can specify how the kitchen is to be built without having to build a range, microwave, and refrigerator from scratch.

Second, operating system services provide standardization. When a computer company makes the specifications of its operating system known to the computing community, software developers know that if their program uses those services correctly it will run on any computer with that operating system. This means that they need make only one version of their program for each operating system, instead of one version for every computer. This saves an enormous amount of time. It also provides significant convenience for consumers, who can now purchase a software product with the expectation that it will run on any computer with the same operating system, instead of one and only one machine.

Standardization of software is now so much a part of our lives that we don’t even think about it any more. It’s ludicrous to even imagine a word processing program that runs on one PC but not another. Those of us in the computing field who have been around a while, however, know that things weren’t always this way. There is nothing about computing, no scientific law, that compels standardization. Today’s computing professionals could write literally hundreds of operating systems tomorrow, all different from each other, all competing with each other, and all completely incompatible with each other. Fortunately, we don’t waste our time on such an obviously unproductive task.

We see then that standardization is an extremely important societal good. It represents the preservation of scarce resources invested in software development and computer use, most importantly human time and effort. Microsoft’s recognition of this basic fact and its attempts to capitalize on the high value of standardization in the marketplace are the single most important factors in its success. These have important implications for the DOJ’s case.

Standardization vs Innovation

But if standardization has some obvious advantages, it also has some problems. Over time, human knowledge increases. New discoveries and ideas reveal better ways to accomplish tasks that improve upon what was previously known. As new knowledge disseminates throughout society, old standards disappear and new methods are adopted.

This notion of economic progress follows scientific discovery. As new knowledge is gained through experimentation and study of the physical world, newer theories emerge that displace pre-existing ones. This does not mean that the older theories were "bad", only that they were the best available given existing knowledge. Similarly, as the discoveries and new ideas proposed by entrepreneurs in the marketplace become adopted, adherence to old standards and products becomes contrary to consumer preferences. This does not make older standards and products "bad", only less rational to pursue in light of newer knowledge. Were economic reality otherwise, we would still be driving Model T’s and using rotary phones.

But if we give up adherence to a standard, or decide to replace an older product with a newer one, we must reallocate resources. We are faced with a thorny problem: Under what circumstances should a standard be abandoned? At what point does it make sense to discard the resources invested in methods of older knowledge, and adopt those discovered more recently? We are faced with a standardization / innovation tradeoff, an important economic problem.

We all know from personal experience that this tradeoff exists. Consider, for example, the last time you bought a car. Let’s suppose that the latest issue of Consumer Reports has a profile on the latest model similar to yours, and their data indicate that it gets 10% more miles per gallon but costs the same. Should you buy it? Most of us probably wouldn’t, because the resources invested in our cars are significant, even though they were produced with less knowledge than that available today. Most people don’t have the resources available to purchase the newest car every year, even if is technologically superior. On the other hand, we may upgrade our computers more often, both because they represent a less significant investment of resources and because new knowledge is generated and put to use more rapidly in the computer industry.

Empirically, we may conclude that standards shift and new products are adopted when the benefits of moving to the new outweigh the costs of discarding the resources invested in the old. Profit-making firms attempt to influence when and where these shifts occur both by maximizing the benefits of their proposed innovations (increased ease of use, greater functionality, lower cost) and by minimizing the costs of their adoption (by offering compatibility with existing standards). Microsoft, for example, consistently adds features to Windows while fully supporting DOS, a much older operating system developed in the early days of computing. Intel’s latest chips including extremely sophisticated features for improved performance, but have always offered full compatibility with their original design of twenty years ago (Dulong 1998).

An understanding of the standardization / innovation tradeoff is essential to analyzing the economics of the computer industry. Unfortunately, it appears to be lacking in both the public statements and the legal documents on file in the Microsoft case.

The Internet, the WWW, and Browsers

Since the early 1960’s, computer scientists have investigated various methods for connecting computers together in a network. The advantages of networking including sharing information, sharing resources, and improved efficiency. Over the past two decades, a series of standards has evolved that permits computers to be connected to one another easily and predictably. These standards evolved into the well-known Internet Protocols.

In 1988, Tim Berners-Lee first proposed a general way of organizing information that made the Internet much easier to use, which led to the emergence of the World Wide Web. The first popular browser was written by Marc Andreesen, a graduate student in computer science at the University of Illinois. He went on to found Netscape Corporation, now the leading supplier of browser software.

Shortly after the explosion in popularity of the World Wide Web, a group of programmers at Sun began to work on a way to write platform-independent programs. Their work resulted today’s Java virtual machine and application development environment. The developers of Java negotiated arrangements with Netscape to support their efforts within Netscape’s browser. This meant that any computer that could access the web and run a browser could run programs written in Java, regardless of what kind of computer it was or what operating system it ran.

Microsoft, in response to the popularity of the WWW and Netscape’s browser, recognized that the benefits of newly discovered knowledge might now outweigh the costs of discarding consumer investment in existing standards. Bill Gates responded by redefining Microsoft as an "internet company", and not a software company. Microsoft embraced the WWW, Java, and the internet protocols as new standards, incorporating them into their existing products. They developed a competing browser to Netscape, Internet Explorer, and integrated it with their Windows operating system. By so doing, they sought to a) capture the tremendous benefits of innovation and ease of use of the WWW, and b) minimize the costs of abandoning standardization by offering full compatibility, including "look and feel", with DOS and Windows.

When Internet Explorer (IE) was first developed, Microsoft required all PC vendors to put its icon on the desktop and take other steps that favored IE as a condition for pre-installing the Microsoft Windows operating system. The Department of Justice believed this to be a violation of a previously obtained consent decree, and filed an antitrust complaint. The legal minutiae of the consent decree hinge on whether or not the browser is a "separate product" from its operating system, but the finer points do not concern us here. Our interest is in the rationality of the complaint, and the rationality of applying century-old antitrust law to the computer industry.
 

The DOJ Complaint and Public Statements

The civil action of United States of America v. Microsoft Corporation was filed on May 18^th, 1998. A browser-readable version is available on the web from the Department of Justice at http://www.usdoj.gov/atr/cases/f1700/1763.htm. Unless otherwise stated, all citations in this section are drawn from that document. Those who are skeptical as to my rendering of the government’s position may readily check the source above.

The DOJ’s argument, while elaborate, is essentially as follows:

1. Microsoft has a monopoly in the PC operating system market. This monopoly is protected by "network effects".
2. The emergence of the Internet, WWW standards, and Java are seen by Microsoft as a threat to that monopoly.
3. Microsoft’s actions in response to this threat (such as its promotion of Internet Explorer through restrictive tie-ins with the Windows operating system) are illegal violations of sections 1 and 2 of the Sherman Act of 1890. They are also anticompetitive and detrimental to consumer welfare.
4. Preventing Microsoft from engaging in these practices will promote innovation and competition in the browser market. This will enhance consumer welfare.

It is easily verified empirically that Microsoft has the largest share of the PC operating system market, generally estimated between 80 and 90%. As we have discussed previously, however, operating systems provide standardization, a tremendous consumer benefit. A smaller market share for Microsoft and more operating systems in the marketplace, while perhaps modeling a textbook notion of "competition" more closely, would impose tremendous costs on consumers. It is irrational to claim the primacy of consumer welfare while at the same time citing the dangers of a large market share. It is precisely that large market share that enhances consumer welfare, if consumer welfare is understood to include the value of the time and effort expended in using a computer.

Additionally, the complaint alleges that Microsoft’s "monopoly" is protected through "network effects". Network effects occur when a person’s benefit from using a product increases with the number of people who use it. A declaration filed in support of the DOJ notes correctly that "application software written for a specific operating system cannot run on a different operating system without extensive and costly modifications or add-ons", and that "Network effects have increased the desirability of Microsoft Windows [95 and 98] for consumers. Once enough users had been attracted to Windows, that very fact made Windows even more desirable to further users" (Fisher 1998). The complaint then claims that the number of software applications that must run on an operating system constitute a significant barrier to entry and that legal action is therefore justified.

This claim is contradictory to stated objectives of consumer welfare, because these very "barriers to entry" are what make Windows so attractive to consumers. It is precisely the large number of applications programs available under Windows and the standardization they provide that motivate consumers to purchase them. It is not rational to recognize the benefits of standardization for consumers while at the same time citing them as "barriers to entry" that harm consumers. These are contradictory premises from which any desired outcome could be plausibly argued. Accordingly, computer scientists should find them unpersuasive.

The Competitive Threat to Microsoft

One of the great ironies of US v. Microsoft is that the legal documents contain numerous admissions that market processes are working. Sections I.6 through I.10 of the complaint document very clearly that the combination of web-based standards and Java, developed by Microsoft’s competitors, do indeed pose a significant threat to Microsoft’s domination of the PC operating system market. Entrepreneurs and scientists at Netscape and Sun have generated new knowledge, with benefits that might outweigh the costs of discarding the existing standard. Even a standard as popular as Windows.

If Microsoft were truly a monopoly, then it would have no competition. It would not need to respond, it would not need to innovate, it would restrict output, and it would raise the prices of its products. In fact, as the complaint itself admits, Microsoft has done just the opposite. Far from ignoring the threat from Netscape, Microsoft developed a competing browser, and by the government’s own admission "spent hundreds of millions of dollars to develop, test, and promote [it]". Far from acting like a monopolist that can ignore consumers wishes and throttle innovation, Microsoft "released three subsequent versions [of IE] (2.0, 3.0, 4.0), in each case adding features and functionality to the product" . Far from pricing like a monopolist, it gives its browser away for free, and the price of the various releases of Windows have consistently dropped in real dollars while including more functionality. Rational individuals cannot consider this evidence of monopolistic behavior while at the same time regarding monopoly as characterized by arbitrarily high pricing and apathy towards consumers.

Microsoft’s Response and the Sherman Act

Microsoft’s response to the threat from Netscape and Java are the key issues in this case. The primary concern is Microsoft’s requirement of favorable treatment of Internet Explorer as a condition for manufacturers to pre-install Windows on their machines. Microsoft also has negotiated arrangements with internet service providers and internet content providers, (ISPs and ICPs), specifying if they want their company to appear in Windows’ default internet connection startup screen, they must provide preferential treatment to Internet Explorer. Such preferential treatment might include stating the IE is the preferred browser, removing links from their main sites where competing browsers could be obtained, and so forth.

The DOJ is alleging that these practices are anticompetitive and illegal violations of sections 1 and 2 of the Sherman Act.

The Sherman Antitrust Act of 1890 forbids contracts "in restraint of trade". Sold to the public in response to perceived concerns about the economic power of corporations, it is the primary legal weapon in the antitrust arsenal. Given the paucity of rational debate that characterized the political processes of the 19^th century (and is still with us today) , and given the extraordinary difficulties in distinguishing restraint of trade from competition, it is not surprising that attempts to apply the Sherman Act 100 years later to a field that demands rationality and logic lead to contradictions and problems.

In this case, the problem is that in the computing industry attempts by firms to help address the standardization/innovation tradeoff may promote consumer welfare but still be interpreted as restraining trade. For example, Microsoft’s negotiated tie-in arrangements are the predictable and socially beneficial acts of a profit-making firm seeking to persuade consumers to adopt a new standard by minimizing their costs. Consumers may not care which browser they use, or they may not want browsers at all (more on this below). Consumers do care about convenience, and it may well be that the most valuable use of consumer time is found by having a simple, easy-to-use pre-installed operating system with simple, fast internet connection startup screens. Or it may be entrepreneurs will decide that Microsoft’s pre-installed vision of computing is not meeting consumer needs, and provide value by installing less restrictive versions of Windows on computers for resale. Or it may be that producers and consumers of software will embrace the Java paradigm of application programs and operating systems, discarding Windows altogether. We simply don’t know.

But there is no rational reason to prefer one specific outcome over another, as the DOJ complaint does, in the absence of the dynamic negotiations of the marketplace. The DOJ’s stated goal of "improved competition in the browser market", for example, assumes a particular point in the standardization / innovation tradeoff as most desirable, without any a priori evidence. In fact, due to the rapid generation of new knowledge by computer scientists and entrepreneurs, there may not be a browser market five years from now. Perhaps consumers will prefer the WWW and their desktop integrated into a single, easy to use view of computing, and won’t like having to use a separate program to search the web. Or perhaps new discoveries in the future will make operating systems obsolete. We simply don’t know. Given what we do know about computing and innovation, however, a focus on the browser market seems short-term and transitory. It is doubtful, for example, if consumer welfare would have truly been enhanced by intervention in the markets for buggy whips and ditto machines.
 

Rational Inquiry and Good Law

The problem with the Microsoft case is not inherent in the rule of law itself, but in the type of law being applied. Legal scholars have reached a surprising amount of consensus as to what distinguishes good laws from bad. In fact, the distinguishing characteristics of good law have a great deal in common with good science: (Pound 1960, Hayek 1960, Hayek 1979, Goldwin and Kaufman 1988, Peczenik 1989, Leoni 1991 and Posner 1995)

1. Axiomitization. The scientific method is predicated upon certain self-evident principles (see the introduction to this article). Similarly, the axiomatic principles of a lawful government are stated explicitly in a constitution or other founding documents: "We hold these truths to be self-evident …"
2. Consistency. Science seeks to replicate results in a consistent fashion. Experiments with similar inputs that produce similar outputs to those reported elsewhere provide strong evidence that a theory is correct. If this does not occur (as for example with cold fusion), then the original claim loses its legitimacy as a way of understanding the world.

 

Similarly, good law relies on precedent. Similar facts in a given case should lead to a similar decision. If not, the law is perceived as being applied in an arbitrary fashion and is therefore less legitimate in the eyes of its subjects.

3. Avoidance of contradictions. The elimination of contradictory hypotheses is one of the strongest impulses in modern science. Throughout history, the best scientific experiments were those designed to distinguish between two alternative explanations, showing one of them to be unambiguously false.

 

Similarly, when courts arrive at contradictory decisions during their attempts to resolve legal problems, a higher level court is expected to rule on the issue. In fact, the existence of contradictory lower court decisions is a common reason for appellate courts to agree to hear cases.

4. Predictability. With a good experiment, you can tell in advance that something will be learned (though of course we cannot know the specific outcome). This simple notion is the basis for peer-reviewed scientific research all over the world. A scientist can read a description of proposed research project and know in advance that it will acquire knowledge. If instead a scientist cannot tell in advance what if anything will be learned, then the proposal should be marked as bad science and funding denied.

 

This notion of predictability applies to the law as well. Suppose you are contemplating a course of action, and are wondering whether or not it will break the law. One measure of the quality of a law is the ease with which this question can be answered. Laws that are clear and easy for citizens to obey are considered better than those which require extensive legal advice. Compare, for example, laws that prohibit murder to the regulations of the Internal Revenue Service.

5. Reductionism. Good science is reductionist. It seeks to understand the larger picture by breaking down systems into smaller and smaller components. Similarly, good law is reductionist. It focuses on the particular circumstances and behaviors of individuals, the smallest units with which the law can be concerned. Criminal law and tort law, for example, focus on the actions of individuals, and are well understood and accepted by the vast majority of the populace. Legislation that benefits a specific group or groups is, by contrast, controversial, fractious, and bitterly contested.
6. Preference for the actual over the hypothetical. Skepticism, to paraphrase Richard Feynman, is not concerned with what might be happening, but with what is actually happening. Scientists do not spend time worrying that UFOs that might be visiting Earth, homeopathic techniques might be effective, or that zero point energy might exist. They are concerned instead with what is actually happening in the world around them, and in constructing experiments that will shed light on this question. They are interested in what the world is actually doing, not in what it might be doing.

Finally, it is shows a strong preference for the hypothetical over the actual. The DOJ does not base its case on consumer harm caused by Microsoft, because there is none: Microsoft has consistently offered better products at lower prices to increase market share. Instead, the case documents recommend preemptive action based on Microsoft's potential to monopolize and cause harm (Sibley 1998, Fisher 1998):

Conclusions

I have argued in other work that irrationality in politics is harmful. The Microsoft trial has become a case in point:

But it is not correct. Computer scientists should urge the Department of Justice to drop this ill-considered action as an affront to rational thinking, and demand that DOJ limit its regulation of Microsoft (or any other company) to ensuring that it honors its contractual agreements with customers and shareholders. Our success would be a shining legacy of our profession for future generations to build on.
 

References

Berman, Harold et. al: The Nature and Functions of Law, The Foundation Press, © 1996.

The United States of America v Microsoft Corporation, civil action, available online at http://www.usdoj.gov/atr/cases/f1700/1763.htm.

Dulong, Carole: "The IA-64 Architecture at Work", IEEE Computer, July 1998, pp 24-32.

Fagin, Barry: "Liberty and Community Online", ACM 1998 Conference on Computing and Public Policy, Washington DC. Available online at http://www.rmi.net/~fagin/Papers/policy98.html.

Fagin, Barry: "Skepticism and Politics", Skeptical Inquirer, May/June 1997, available online at http://www.csicop.org/si/9705/politics.html.

Fisher, Franklin: Declaration of Prof. Franklin M. Fisher in support of plaintiff in USA v Microsoft, available online at http://www.usdoj.gov/atr/cases/f1700/1766.htm.

Goldwin, Robert and Kaufman, Art: Constitution Makers on Constitution Making, American Enterprise Institute for Public Policy Research, ©1988.

Leoni, Bruno: Freedom and the Law, Liberty Fund Press, © 1991.

Hayek, F. A.: The Constitution of Liberty, University of Chicago Press, © 1960.

Hayek, F. A.: Law, Legislation, and Liberty, University of Chicago Press, © 1979.

Peczenik, Alexander: On Reason and Law, Kluwer Academic Press, © 1989.

Pound, Roscoe: Law Finding Through Experience and Reason, University of Georgia Press, © 1960.

Posner, Richard: Overcoming Law, Harvard University Press, © 1995.

Rand, Ayn: Capitalism: The Unknown Ideal, The New American Library, 1967, p. 70.

Sibley, David: Declaration of Prof. David S. Sibley in support of plaintiff in USA v Microsoft, available online at http://www.usdoj.gov/atr/cases/f1700/1763.htm.