Interestingly, he got the 70,000 count from an off‐hand comment made at the trial by Microsoft witness John Rose, who provided no supporting documentation. When pressed by the government’s attorney, Rose added an important clarification that has been ignored by the media: “Not all of [the applications] were written for Windows. Some of them go back to the DOS environment and carry forward into the Windows environment. But over the course of the 17 years of the PC industry, there have been approximately 70,000 applications developed.” In short, many of the 70,000 applications are outdated or no longer exist, and could not possibly be part of an entry barrier today.
A realistic count might shed light on the difficulty of entering the operating system market. For a clue, let’s examine the software listings on the Amazon.com website. While Amazon might not sell every program, it surely carries those applications commonly bought by computer users, and many others that are bought by “techies.” Surprisingly, Amazon lists only 8,301 “productivity” applications ranging from business software to computer languages. That’s just 12 percent of the supposed 70,000 entry barrier.
Moreover, many of the programs were written for non‐Windows operating systems like Linux, Apple and IBM. Many others are multiple versions of the same program — for example, the “standard” and “professional” editions of WordPerfect — or the same program packaged differently for multiple user systems. When the non‐Windows programs are deleted and the double (and triple and quadruple) counting is eliminated, no more than 1,300 unique Windows desktop applications remain — less than 2 percent of what the judge found as legal “fact.” Granted, there are 3,500 games and educational programs available from Amazon, but many of them are also non‐Windows programs, versions of the same program, or outdated.
Even more important than the inflated overall count is the judge’s presumption that huge numbers of applications are necessary to challenge Microsoft. But does a challenger really need to duplicate all 517 disk backup programs that are now available for Windows? In a June survey of executive MBA students at the University of California, Irvine, 19 percent of the respondents used only the six programs that come with Microsoft Office suite; only 16 percent used more than six additional programs. That means 84 percent of the respondents used in total a dozen or fewer applications.
Of course, the programs used by any single respondent may not have been the same programs used by the others. Allowing for that, it seems clear that a sizable segment of computer users could be adequately served with an operating system that has a few hundred well‐chosen applications. Consider that a fifth of the U.S. operating systems market covers roughly 20–25 million users. A fifth of Microsoft’s operating systems revenue is about $2 billion annually. Any number of firms would be eager to enter that lucrative market — if Microsoft were ever to behave like a monopolist.
And contrary to what Judge Jackson suggests, an “aspiring entrant” need not, on its own, develop all of the required applications. Rather, a competitor can turn to outside developers to make the investment. That shouldn’t be difficult if Microsoft were to charge high prices that left an opening for a lower‐priced rival.
In sum, the so‐called applications barrier to entry is not what the judge has made it out to be. The legal implications are important and obvious. Judge Jackson’s decision rests on a fabrication. Because his fact‐finding is clearly erroneous, it cannot support his conclusions of law.