ONE OF Catherine the Great’s closest advisers was Prince Potemkin, who famously tried to persuade her that his policies were improving the life of the Russian people by constructing fake prosperous villages for Catherine to see as she sailed by on the Volga River. Today, another powerful woman, Judge Colleen Kollar-Kotelly, the judge in charge of the Microsoft case, is looking at what can only be described as the Potemkin Village of antitrust settlements. On its face, the proposed settlement between Microsoft and the Justice Department looks fairly impressive, as befits the most consequential antitrust case since the government broke apart the Bell Telephone monopoly two decades ago. Over 90 percent of personal computers use Microsoft’s Windows operating system, which runs a variety of software by Microsoft and rivals. The government litigation proved that Microsoft had taken various illegal steps to bar competition from rival software makers, including: (1) mixing together its operating system and software code in order to exclude rival software, even though this conferred no technological benefit and indeed degraded performance; (2) making or encouraging design decisions that made it more difficult to run rival software on its operating system–again, with no technological benefit to the user; and (3) making agreements or threats that barred other firms from using rival software. These are serious violations that demand serious remedies. The proposed settlement purports to require Microsoft to do the following: (1) Disclose to rival software makers the code they need to run their products on Microsoft operating systems. (2) Allow others to customize Microsoft’s operating system in order to substitute rival software for Microsoft software. (3) Stop excluding rivals through agreements or threats. The basic idea is that with these restrictions in place, rival firms could compete on an even playing field in developing the software that runs on operating systems. It all sounds pretty sensible. Unfortunately, there is remarkably little of substance behind this impressive looking fa ade. The most fundamental problem is that, even if the settlement works as advertised, it will do nothing to prevent Microsoft from using technological means to prevent competition on the merits. All Microsoft has to do is commingle its software code with its operating system code, or design its operating system to disrupt rival software. If it does the former, buyers will not want the now redundant rival software since they will have already paid for an operating system that includes the Microsoft version. If it does the latter, no one will want the rival software because, through no fault of its maker, it won’t work as well. It would be one thing if Microsoft’s rivals were subdued by genuine technological innovation that happened to make their products less desirable. But in fact the settlement leaves Microsoft free to exclude rivals through product bundling or design decisions that confer no technological benefit on users and even degrade performance. Given that Microsoft was found guilty of doing precisely that repeatedly, it is hard to see how such a settlement could be in the public interest. Thus, even construing it in the most favorable light, the proposed settlement would be like trying to stop traffic on a five-lane highway by closing one lane. But in fact even that one lane remains open because, after two rounds of revisions, a series of sneaky definitions make the settlement’s supposed restrictions on Microsoft speed bumps at best. For example, the proposed settlement would not oblige Microsoft to disclose the code rivals need to run on Windows, unless those rivals use it for the sole purpose of interoperating with a “Windows Operating System Product.” Sounds okay, but this term is then defined to include only Windows XP or Windows 2000 Professional running on personal computers. Microsoft would remain free to withhold the code necessary to run software on other Windows operating systems that are in much wider use (like Windows 95, 98, and 2000) or on any hand-held devices. Relegated to use on only a fraction of Windows-run devices, rival software would already be at a huge disadvantage. Consider next the obligation to allow computer makers (like Dell and Gateway) to customize the Microsoft operating system to substitute rival software. It turns out this only applies to rival software that qualifies as a “Non-Microsoft Middleware Product”–a term of art defined in a narrow way that limits it to products of which “at least one million copies were distributed” last year. The Catch-22 is clear. Microsoft can bar computer makers from customizing the operating system to substitute any new software until it reaches the one million-user threshold. And no new software is likely to reach that threshold as long as Microsoft can bar such substitution. What about the obligation to refrain from threats or agreements that induce firms not to use software that competes with “Microsoft Platform Software”? This term is defined to mean either a “Windows Operating System Product” or a “Microsoft Middleware Product.” The former leaves Microsoft free to use threats to exclude rival software that competes with all of its Windows products except XP and 2000 Professional. The latter means Microsoft can exclude rivals until they distribute one million copies. Again, one wonders how many rivals will get to the one million threshold in the face of such threats and agreements. Finally, the proposed settlement does not really protect all rival software that might run on Windows, but a certain set of software it calls either “Middleware” or “Middleware Products,” whose definition is narrowed in additional odd ways which raise problems that cut across the settlement provisions on disclosure, customization, and agreement or threats. Consider new software products that are not listed in the proposed settlement. Under the “middleware” definitions, Microsoft can avoid any restrictions against excluding such new products by the simple expedient of not trademarking its own competing product or by not distributing its product separately from its operating system. Another odd limitation on the “Middleware” definition means the settlement does not protect rival software that competes with updated versions of Microsoft’s products unless those updates are “major”–a term defined to include only those updates that Microsoft chooses to identify “by a whole number or by a number with just a single digit to the right of the decimal point.” So, even if Microsoft did want to trademark or separately distribute a software product, all it would have to do to avoid its settlement obligations is create an update and assign it a version number with two decimal places. Thus, as long as updates to Futureware 1.0 are simply relabeled 1.01, 1.02, and 1.03, and not Futureware 1.1, 1.2, and 1.3, Microsoft can exclude competitors with impunity. In short, as applied to new products, these definitions effectively leave it to Microsoft’s discretion whether it has any obligation to disclose its code, allow customization, or refrain from exclusionary threats and agreements. The company is effectively free to resume all the practices that were held illegal and that the agreement ostensibly forecloses. Prince Potemkin would have been proud. Einer Elhauge is a professor of antitrust law at Harvard Law School.
