In the best of all worlds, we would all benefit from the so-called network effects that result from most people using the same software: everyone could easily communicate with each other and teach each other how to use the software efficiently. Unfortunately, since Microsoft uses network effects to maximize its profits rather than to benefit users, the world it delivers is far from the best.
Consider Vista, yet another great new operating system that Microsoft rolled out this year, together with Office 2007. The first person at my company to use Vista was our Executive Vice-President. He was furious. Vista and Office 2007 came with his new Dell computer by default. Dell didnt ask: Would you prefer the old versions of the operating system and MS Office that you know how to use? So our VP got a shiny new computer that he didnt know how to use: functions were rearranged, and keyboard shortcuts were different.
Think of the productivity cost of millions like him having to adjust to a new system. Moreover, his coworkers couldnt read the Microsoft Word files that he sent them in the new .docx format. . . . To be sure, Microsoft does provide a patch that allows old versions of Office to read the new .docx format. But Microsoft doesnt publicize it or warn you if your Office 2007 file is about to become incompatible with older versions.
While Microsoft could have kept the traditional .doc as its default format for MS Word, this would not have served its purpose: eventually, after enough of the world pays for Office 2007, holdouts will be dragged along, kicking and screaming. Then, in four or five years, Microsoft will begin our agony all over again. . . . Microsoft can control expectations, which turn out to be self-fulfilling when network economies are substantial, as they are in software markets. . . .
Whenever Microsoft rolls out a new operating system, the question is not whether you should switch, but when. Adding new features can speed the transition, but what is necessary is only that the new system be incompatible with existing systems in certain respects, and that a sufficient number of people expect that it will become the new standard.
Of course, creating new software is costly. So why should Microsoft bother?
The Nobel laureate Ronald Coase answered that question long ago. According to the Coase Conjecture, a monopolist selling a durable good must sell it at marginal cost. For Microsoft, the problem is that the marginal cost of software is zero. As a result, Microsoft cannot extract anything close to its full monopoly rents unless it sells upgrades. After all, even though Microsoft has a monopoly of primary sales of its software, copies sold in 2007 compete with those sold in 2006.
So, by creating incompatibilities, some subtle and some obvious, that make its old software obsolete, Microsoft can sell its operating systems at high profit margins without fear that people will wait until the price drops. The price will never drop, because Microsoft will just roll out a new system, again at high profit margins.
Microsoft has been in antitrust trouble for 15 years, and, despite the companys recent agreement with the European Union to license its source code, it will probably be in trouble again. When that happens, I hope the antitrust authorities will consider a remedy that Ian Ayres, Hal Varian, and I devised.
Suppose Microsoft had to license its old software freely whenever it brings out a new version. This would give the company an incentive to ensure that new versions are compatible with and significantly better than old versions otherwise the new versions wouldnt sell, or at least not easily. If Microsofts new software had to compete successfully at least against its old software, we would know the world was improving.
In the meantime, I recommend installing the Microsoft patch to your old computer (http://office.microsoft.com/en-us/word/HA100444731033.aspx) and just suffering the devil you know.