IMAGINE A corporation whose ambitions are unbounded and whose main product seems indispensable. Imagine that it has threatened to withhold this product in order to impose its will on competitors and manufacturers who require it. Imagine that it has in its coffers $30 billion in cash, and that its profits in one recent year came to almost $8 billion. And now imagine that this company has provoked so much hostility in its industry that it is regularly referred to as the “Evil Empire,” and that one competitor has spent $3 million lobbying the U.S. Department of Justice to take action against it.
The company, of course, is Microsoft, and in response to that lobbying, and other complaints, the Justice Department did indeed bring an antitrust suit against it in 1998. Last year, after a trial that lasted 76 days and generated 13,466 pages of transcript and 2,695 exhibits, the district court in Washington, D. C., presided over by Judge Thomas Penfield Jackson, ruled that Microsoft, with its Windows program, possessed monopoly power over a particular kind of operating system. (An operating system is the crucial program that controls all a computer’s activities.) Indeed, so unshakably secure was Microsoft’s position that, according to Judge Jackson, no competitors were likely to arise in the foreseeable future. In protecting and expanding this monopoly, the company had been, he said, “untrustworthy,” “disingenuous,” “oppressive,” and “predacious,” and had done “violence” to the creative process. For thus violating the Sherman Antitrust Act, Microsoft would be broken up, split into two separate companies.
This past June, an appellate court reviewed Judge Jackson’s harsh conclusions. In a decision hailed by Microsoft and accepted by most of the news media as a setback for the Justice Department, the court rejected the proposed “remedy,” declared one of the major accusations unproved, and attacked Judge Jackson for holding confidential press interviews during the trial in which he had likened Microsoft to drug traffickers and “gangland killers”; in so doing, the appellate court declared, the judge had violated the judicial code of conduct in ways that were “deliberate, repeated, egregious, and flagrant.”
Initial impressions notwithstanding, however, the appellate court’s decision was hardly good news for Microsoft. The court refrained from challenging most of judge Jackson’s “findings of fact”; it found the company to be in possession of monopoly power; and it declared Microsoft guilty of maintaining that power by anticompetitive means. As of now, it is unclear whether some settlement will be reached, or whether a lower court, in order to determine appropriate “remedies,” will conduct the full “evidentiary hearings” that Judge Jackson dispensed with, or whether the case will be decided by the Supreme Court. But neither a settlement nor any future judgment is likely to call into question the assertion that Microsoft is a monopoly or even remotely to challenge the findings of fact. Any compromises and penalties demanded of the company are bound to be substantial.
Have we, then, arrived at a just conclusion? Not in the least. Despite everything we have learned through two stages of adjudication, this case remains one that, in my view, should never have been brought at all. Moreover, I believe that the Justice Department and the eighteen state attorney generals who remain committed to the case seriously misjudged the nature of the technological landscape and, in the name of protecting competition, acted to inhibit it.
Some of my objections have been raised by Microsoft in its own defense or have appeared in articles or books; others have not. Under close scrutiny, for example, many of the presumptions of the case seem very peculiar. Antitrust law in the United States, as the company itself has pointed out, is designed to protect consumers, not competitors, yet consumers did not seem eager for protection; before the trial began, a New York Times/CBS News poll found that only 2 8 percent of computer users had an unfavorable opinion of Microsoft and a full 77 percent of all Americans believed it made quality products. (Attorney General Janet Reno said at the time that part of the government’s job was to “convince people” it was acting in their interest in this case.)
As far as the consumer and the industry are concerned, moreover, market dominance is not necessarily a drawback. In creating a defacto standard, Microsoft has made computer use easier for consumers and more economical for software developers. Then, too, while Windows is, without question, the most important piece of software in any computer, it accounts for less than 5 percent of the average computer’s cost; in fact, the price of Windows today is not significantly higher than the price of earlier versions a decade ago. In addition, the Windows operating system, whatever its flaws, is regularly improved, with new versions leading to faster speeds and more sophisticated applications. In pursuit of such improvements, Microsoft has spent over $5 .3 billion on research and development in the last year alone; its latest version, Windows 2000, required 5,000 workers, 500 many years of testing, and a $2 billion investment.
The polemical drift of these arguments is clear: what sort of monopoly can it be that does not dispense with, or whether the case will be decided by the Supreme Court. But neither a settlement nor any future judgment is likely to call into question the assertion that Microsoft is a monopoly or even remotely to challenge the findings of fact. Any compromises and penalties demanded of the company are bound to be substantial.
Have we, then, arrived at a just conclusion? Not in the least. Despite everything we have learned through two stages of adjudication, this case remains one that, in my view, should never have been brought at all. Moreover, I believe that the Justice Department and the eighteen state attorney generals who remain committed to the case seriously misjudged the nature of the technological landscape and, in the name of protecting competition, acted to inhibit it.
The polemical drift of these arguments is clear: what sort of monopoly can it be that does not.
II
DOES, OR did, Microsoft hold monopoly power? Unless it did, what seemed to the courts
like anticompetitive behavior was merely competitive behavior.
The courts considered three major issues in establishing Microsoft’s monopoly power.
Defining the relevant market. The courts were very precise. Microsoft Windows was said to hold over 90 percent of one particular market, namely, the market for operating systems used in Intel-compatible PC’s (personal desktop computers that evolved out of a design now nearly twenty years old). Other forms of operating systems such as those used in handheld computers, Apple Macintoshes, network computers connected to other machines, or “servers” for hosting Internet sites were judged by the courts to constitute separate markets and thus to offer no competition. This is a necessary distinction, for if the relevant market had been defined as operating systems for all computing devices, Microsoft would be just another player in an actively competitive world.
Barriers to entry. According to Judge Jackson, Windows boasts some 70,000 “applications” programs, written specifically for it, ranging from word processors to video games. Any operating system is made stronger by the number of its applications the higher the number, said the courts, the more popular the system will become, and thus the more likely a system will be to preserve its strength by developing still more applications. And this constitutes a barrier to competitors trying to enter the market for who, in a short time, can match such a quantity of offerings? Thus, as the courts reasoned, IBM’s operating system, OS/2once a major competitor to Windows ended up failing because it could not match Windows’ number of applications.
“Switching costs” and “lock in.” Programs for one operating system will not run on another, which means that a consumer wishing to switch to a different operating system will have to purchase all new software and perhaps even new hardware. Such “switching costs” are high, said the courts, with the consequence that consumers are effectively “locked in” to Windows.
In sum, Microsoft (in the courts’ view) dominates a market the boundaries of which prevent competition from entering and consumers from leaving. This in turn means that, unfettered by competition, Microsoft does not need to succeed on the merits; it can charge a price for Windows higher than if it had to respond to competition.
These contentions have a certain truth to them, but a much larger modicum of untruth. Nor, separately or together, do they establish monopoly power. As the history of operating systems shows, the relevant markets are permeable and promise to remain so; the success of an operating system is due not to the absolute number of applications it supports but to the design of a handful of its essential applications; and consumers willingly undertake to pay “switching costs” when major improvements are offered.
A little over two decades ago, the entire computer arena was dominated by IBM and its “mainframes” large computers sold to institutions and corporations. No one in the mid-1970’s would have credited the possibility of a significant consumer market for operating systems alone. But then the world changed, desktop computers began to appear, and IBM, in developing its own such “personal computer” (PC), needed an operating system. Acting on its old understanding, according to which operating systems were mere appendages to the more important hardware, it did not design its own such system but turned to a fledgling company called Microsoft, from which it licensed a program called MSDOS. IBM’s failure to purchase this program outright was perhaps the most significant economic blunder in the history of the industry. The market had shifted, and IBM was caught unaware.
A similar transformation occurred in 1984 when Apple created the Macintosh the first consumer this constitutes a barrier to competitors trying to enter the market for who, in a short time, can match such a quantity of offerings? Thus, as the courts reasoned, IBM’s operating system, OS/2once a major competitor to Windows ended up failing because it could not match Windows’ number of applications. computer displaying different typefaces on the screen and using images to represent programs. So dramatic were these improvements that they not only helped displace earlier, non-graphic Apple models but lured users away from MSDOS as well. Until 1990, even though Apple’s own market share was small, Microsoft struggled unsuccessfully to compete, its awkward and inadequate designs leading to continual failure.
Then, at last, Microsoft made a breakthrough with Windows 3 .0. Black and white monitors were junked, new computers were sold with ever faster processors, and individuals and corporations, far from being “locked in” by their investments in hardware and software, or being deterred by “switching costs,” eagerly welcomed the new technology, spending millions of dollars in the process.
In today’s marketplace, the concepts of “lock in” and “switching costs” apply still less. Given the pace of technological change, computers hardly ever represent a onetime cost, but typically are replaced or upgraded as prices drop and power continuously increases. Judge Jackson in his findings of fact acknowledged that “PC users buy new PC systems relatively frequently,” but failed to draw the proper conclusion. A superior piece of software or hardware can quickly displace its rivals; the critical factor is quality.
In 1979, for example, a program called VisiCalc was developed that helped businesses create “spreadsheets,” studies of revenues and expenses. It was so powerful that it spurred the sale of PC’s themselves, and it became so established that it created an industry of ancillary products. Then it gave way to a superior program, Lotus 123, which displaced VisiCalc until it, too, was displaced when Microsoft’s program, Excel, came on the market and was hailed by reviewers and consumers alike. This is but one example of many; over time, as Stan J. Liebowitz and Stephen E. Margolis show in their recent book, Winners, Losers, and Microsoft, Microsoft’s market share of products has been strongly correlated to positive reviews of their quality.
SO, WHAT about the sheer quantity of applications? In the first place, as Richard B. McKenzie showed in a paper for the Cato Institute last year, the figure of 70,000 seems to refer not to Windows programs but to the total number of programs written for IBM-compatible machines over the last two decades; about 3,000 such programs are currently listed for sale, and only a fraction of those is in use. In the second place, beyond a certain point it does not really matter. All that most people require is a basic handful, a selection of professional software and some amusing games. That was what it took to jumpstart the Macintosh, Windows 3.0, and even IBM’s OS/2. According to the courts, OS/2, which was meant to compete with Windows, failed because it offered only 2,500 applications; but 2,500 applications is more than enough to satisfy anyone but the most obsessive user. OS/2 really failed because of poor marketing, awkward design and installation, and inadequate courting of software developers.
If “switching costs” are unimportant, and the “applications barrier to entry” is low and easily broken down, then there is nothing to prevent competition. As it happens, Microsoft has had quite a bit of it. Does the company possess the power, as the appellate court stated, simply to “stave off even superior new rivals”? Every time Microsoft has had to compete against a superior consumer product, it has not fared well, which is why it must spend so much on research and development. This has happened with competitors in specific fields like America Online (AOL) for Internet service or Quicken for financial software and there is no reason to believe
things are otherwise with operating systems for the average consumer. If varieties of the operating system Linux became easier to install and more consumer oriented, its 15 million users would multiply significantly.
As Judge Jackson himself asserted, again without acknowledging the significance of what he was saying, “Signs do not indicate large demand for a new Intel-compatible PC operating system.” Quite so: Windows stands relatively unchallenged not because of its ruthlessness but because it is both reasonably priced and, for consumer use, superior to anything else on offer.
Finally, Microsoft does not even have the ability, ascribed to it by Judge Jackson, to “set the price of Windows substantially higher than that which would be charged in a competitive market.” In the judge’s eyes, Microsoft was guilty of charging a “revenue-maximizing price” of $89 for a Windows 98 upgrade when a price of $49 would still have made money. A move like this, which failed to consider prices “charged by other firms in the market,” was itself “probative of monopoly power.”
How so? All companies are free to charge what they like for premium products, and Microsoft charged a price it believed would be paid. Besides, at the very time the trial was going on, a commercial Internet site was offering a Windows 98 upgrade for $82.95, Apple’s Mac OS 9.0 for $86.95, and a commercial version of Linux for up to $61.95. Does this suggest monopolistic pricing?
The truth is that Microsoft has had to constrain its pricing in view not just of immediate changes in the market but of future competition. If it charged $1,000 for Windows, there would be a stampede to Apple and systems like Linux. And this is a point with general significance: despite its undoubted power, Microsoft has had to improve and price its products as if it were beset by a large number of competitors; otherwise, it would be left behind just as IBM was left behind in the 1980’s. It must act, in other words, in constant awareness of its potential weakness, a weakness that is itself an inherent feature of the unpredictable fluidity of technological innovation.
III
As I have already stipulated, it is unlikely that r1. any appeal to the Supreme Court or any settlement will alter the ruling that Microsoft does indeed possess monopolistic power. Even so, however, Microsoft would be in violation of antitrust law only if it misused that power, by engaging in an illegal form of “exclusionary conduct” to protect or extend that monopoly. This, in fact, was the burden of the accusation that started the case in the first place, an accusation involving what was once called “the browser wars.”
A browser is a program used to “access” sites on the Internet. In the early 1990’s, there were as many as 50 such programs. Then, in 1994, a company called Netscape Communications developed a new browser, dubbed Navigator, that was substantially better than those previously available. Other companies: Microsoft, Apple, IBM, Sun Microsystems having already started to plan for the Internet, also soon began to package browsers with new versions of their operating systems.
Microsoft had been characteristically aggressive in developing its own browser, Internet Explorer, which it kept improving while weaving it more and more tightly into Windows. Finally, the two products (as the company itself would assert) were made inseparable. This effectively created a massive distribution network and seemed, indeed, a clear use of monopoly power to win the browser wars. In addition, Microsoft sought to make deals both with computer manufacturers and with online services like AOL to offer Internet Explorer with their products, sometimes exclusively, and it prevented manufacturers from deleting Microsoft’s browser even when they also installed Netscape’s. In short, it threw its weight around, sometimes by threatening to withhold its products altogether.
Many of these kinds of dealings, involving discounts, incentives, and exclusive contracts, are common in every industry. In this indus¬try, the aggression is virulent, featuring regular commands to “kill” and “choke” competitors. But both before and during the case, Microsoft showed itself willing to compromise in its licensing and deal making and in law, as the courts themselves pointed out, these are inherently murky areas, full of counterfactual possibilities and subtle assessments. Still, that did not prevent the judges from reaching the remarkably straightforward conclusions that (a) the success of Internet Explorer had little to do with quality, and (b) Netscape was unfairly prevented from competing.
In 1999, Judge Jackson insisted that there was no “best of breed” in browsers and none was likely to emerge “at any time in the immediate future.” As for Netscape’s failures, these were laid to the machinations of Microsoft, which hampered Netscape’s most economical method of distributing its product. In the unequivocal words of the findings of fact, “Microsoft’s campaign succeeded in preventing” Netscape from developing its competitive potential “for several years and perhaps permanently.”
None of this is plausible. When Microsoft first introduced Internet Explorer, Navigator was clearly the better product, and the more successful. It was only in 1997, once its browser was widely judged by reviewers and users to equal or surpass Navigator’s in quality, that Microsoft began significantly to improve its market share. By the time the judge was presenting his “review,” the superiority of Internet Explorer was unquestioned. The design of Internet Explorer also made it easy to be customized by applications companies and by Internet providers like AOL; Navigator was unable to match that flexibility.
Nor is it easy to see how Netscape failed because of Microsoft’s actions. Microsoft never even “succeeded in preventing” the distribution of Netscape’s browser. In 1998, the year the trial began, 160 million copies of Navigator were downloaded from the Internet. The program was also available on 24 percent of all computers sold by manufacturers like Dell and Compaq the very distribution channel the courts believed Microsoft had closed off with its heavy-handed dealing. In their scrupulous study, Competing on Internet Time, Michael A. Cusumano and David B. Yoffie show that what “allowed Microsoft to catch up” in the browser wars was not its misuse of monopoly power but Netscape’s incompetence and bad decisions: the company missed deadlines, alienated customers, and failed to create a compelling business model. Netscape’s successes were largely unaffected by Microsoft’s actions, and its failures were its own to achieve.
Besides, even if Microsoft had not been involved, it was clear to executives at Netscape that they could not rely on Navigator for long-term profits, certainly not in an industry where it was traditional to give browsers away. Long before the trial began, Netscape’s business plan had shifted several times; in the end, the company was dependent not on sales of the browser itself but on a website to which each new browser was directed, a site that in mid-1997 was receiving 100 million visits a day and bringing in tens of millions of dollars a year in advertising. And while the trial was still going on, sufficient potential value remained in the company to inspire AOL to pay $4 billion for it.
BUT SUPPOSE I am wrong, and Netscape was indeed pummeled by a monopolist with no
claims on product superiority. How was antitrust law violated?
The district court held that the violation lay in Microsoft’s attempt to use one monopoly to create another er; that is, it tied Internet Explorer to its Windows operating system, and then leveraged Windows to take over the browser market. This was the conclusion the appellate court rejected as unproved. But it left standing the charge that Microsoft engaged in the browser wars in order to protect and maintain its Windows monopoly. Thus, according to the appeals court, the compulsory installation of its brows er with Windows “protect[ed] Microsoft’s monopoly from the competition.” Or again: the integration of the browser into Windows meant that Microsoft was “protecting its operating-system monopoly.”
There is an elaborate theory behind this assertion, a theory invoked by the courts to justify many of their rulings in the Microsoft case. That theory is now widely accepted, but deeply flawed. According to it, Netscape was a threat to Windows because its browser, Navigator, could have begun to function as a kind of operating system itself. Other applications would begin to be written for Netscape’s program: there might be, say, a Navigator-based word processor, a Navigator-based financial program, and Navigator-based video games. Developers of such applications, moreover, would only need to write their programs once, completely ignoring the operating system underneath; the programs would run anywhere that Netscape was in use. feet of stripping away the advantage Microsoft enjoyed by virtue of the number of applications written specifically for Windows and would thus lower the “barrier to entry.” And therein lay the peril that Microsoft had to meet and crush. Netscape’s browser called a species of “middleware” because it sat between the operating system and scores of usable applications was, in Judge Jackson’s opinion, the “Trojan horse” that “threatened to demolish Microsoft’s monopoly power.”
Microsoft did not completely discourage the court’s theory. In May 1995, Bill Gates himself had described Netscape as a “competitor” pursuing a strategy that would weaken Windows, and in the trial Microsoft referred to that threat in order to demonstrate that it had significant competition. And there were some at Netscape who felt they had indeed zeroed in on Windows itself as a target. But in actuality, the chances of Navigator’s making good on this threat were minuscule.
There is still not a single example of an important application written for Navigator, and no evidence of one’s having been on the horizon. Even Judge Jackson said “it remain[ed] to be seen” whether major applications would develop. It is highly unlikely that any browser, even one significantly more sophisticated than Navigator, could easily support so many different applications and take on the many functions of an operating system, a significantly more complicated and demanding piece of software.
By focusing on this theory, in any case, both courts ignored the most important aspect of the browser wars and the real reason why Microsoft competed with Navigator so urgently and aggressively. That reason was the Internet. Although Judge Jackson did mention the “exponential growth of the Internet,” he asserted in 1999 that it would have no impact on consumers or operating systems for “the next few years,” and went so far as to argue that many consumers would not even want a browser on their computer. These statements were absurd in 1999 and are more absurd now.
Indeed, given the presence of the Internet, it would have been suicidal for Microsoft not to have been at work on a browser, and not to market it aggressively as part of its operating system. In 1995, Gates sent a memo to his managers: “In the next twenty years, the improvement in computer power will be outpaced by exponential improvements in communications networks.” In the future, as he and every other manufacturer of computers and operating systems recognized, locating a document stored on the Internet, or on a server in Sri Lanka, would be no different from locating a document on one’s own hard drive. The computer would no longer stand alone; it would be a networked product, working with other PC’s and interacting with them on the Internet. This meant that the operating system would also have to evolve to be able to work within those networks.
All this is already taking place. Companies that dominate the market for computer networks like Sun Microsystems and Oracle want to extend their dominance into the workings of the PC. So did Netscape, in its case by hoping (according to the courts) to combine a browser with the functionality of an operating system. Microsoft, moving from the other direction, wants to do the same thing, and will do it more effectively in its forthcoming new operating system, Windows XP. If anything, and despite Judge Jackson’s odd assertion that there are “no plausible benefits” to such changes, this conjoining of the browser and the opera ting system will become more complete over time, thereby providing a benefit to consumers that the courts were unable to see and tried to obstruct.
IV
OF COURSE, the courts could not help acknowledging that the technological marketplace is subject to constant disruption. Judge Jackson, referring to the “dynamic vigorous competition” in the software industry, admitted that “nascent paradigms could oust the PC operating system from its position.” The appellate court noted that “In technologically dynamic markets . . . entrenchment may be temporary.” Judge Richard A. Posner, who attempted to mediate a settlement of the Microsoft case before a final judgment was reached, went further, raising questions about the very applicability of antitrust law in an economy characterized by such continuous disruption. These observations follow the thinking of the economist Joseph Schumpeter, who in his theory of “creative destruction” argued that displacements of monopoly power are common in innovative industries.
But even as they recognized these key qualities of the technological marketplace, the courts declined to regard them as consequential or even relevant. Such considerations, they said, were for the long term, whereas their own focus was on price constraints in the “near future.” In so saying—and even putting aside the question of how they could arrive at the far-reaching decisions they did while admitting that the justifications for them were temporary and limited-the judges revealed, again and again, their obtuseness about technological reality.
The Internet, whose power is here and now, they put off as an issue for the future. The near-future prospect of merging handheld computers, game systems, and network computers they put off into the distant future. But the idea that Navigator could threaten Windows as “middleware” a threat for the distant future if ever there was one, they treated as so imminent and perilous that Microsoft, in their reading, violated the law in order to prevent it. Setting the real technological marketplace in amber, the courts conjured up a false one in urgent need of their supervision and control.
Regardless of their decisions, the Internet will soon become even faster and more commonplace than it is now and will be transformed into a medium for television broadcasts, film, and telephone communications. AOL, after its merger with Time Warner, is already a cable and communications giant; telephone companies and cable television companies are becoming Internet providers; game systems are acting like computers and browsers. In ten years, both browsers and operating systems will be nothing like what they are today. But thanks to the grievously anticompetitive actions of the U.S. Department of Justice and two federal courts, it will take us longer to get there, the way will be strewn with more obstacles, and the cost to consumers and taxpayers will be much, much higher than was ever necessary.