Linux Today: Linux News On Internet Time.

More on LinuxToday

Why We Don't Want the Windows Source Code

Apr 15, 1999, 08:14 (41 Talkback[s])
(Other stories by Nathan Myers)


Desktop-as-a-Service Designed for Any Cloud ? Nutanix Frame

Copyright 1999 by Nathan Myers
All Rights Reserved.

Microsoft is widely reported to have failed to present a convincing case against charges of using their (otherwise-legal) monopoly position illegally, to stifle competition. Microsoft apparently agrees, despite its protestations to the contrary, as they are reported to be involved in settlement talks with the Department of Justice and the states' Attorneys General. The news media are already speculating on what remedies might be appropriate.

We are starting to see articles (and even petitions) calling for Microsoft to be forced to "open" the Windows source code. While superficially appealing, I argue here that this, like the 1985 consent agreement, would be another slap on the wrist for Microsoft. I believe that, despite what they might argue behind closed doors, a source-code publication order is really in Microsoft's own estimation the best-case outcome of the trial.

"freeing the Windows code would be little help to us in the Free Software community, or to anyone else."

Why would opening the Windows source code fail to benefit us? First, as in the case of Netscape Communicator, a large fraction of the code is bound under non-disclosure agreements, and would not be opened under any settlement or court order. Furthermore, the remaining secret code would be the most _interesting_ fraction: device drivers, codecs, and patented compression code. It is primarily the widespread use of incompatible devices that interferes with easy installation of alternative operating systems, so freeing the Windows code would be little help to us in the Free Software community, or to anyone else.

Second, with such a large fraction of the code remaining under wraps, users would be unable to build their own releases of the system. This would leave Microsoft in the same monopoly position, but, having complied with their new consent agreement, immune to prosecution. At best, the code might help provide guidance to such projects as WINE, but only if the code were not encumbered in such a way as to add a new danger of contamination.

Finally, going "open-source" might ultimately benefit Microsoft as companies which have committed to use of their software would end up sending them free fixes for their numerous bugs.

What do I suggest instead? Microsoft is able to attack new markets in part through using, or threatening to use, their huge war-chest account built up by monopoly control of their existing markets. Our best-case scenario for the outcome of the trial would be the simplest: start with a _fine_.

Microsoft is reported to have something like $20 billion dollars in the bank. If in fact the court determines that this money was obtained by illegally exploiting their (otherwise-legal) monopoly position, then the court may simply and legally confiscate it. That money, placed in grants foundations, could fund or educate quite a lot of Free Software developers, even after the lawyers took their cut.

I see no reason to stop with (say) a $20 billion fine. In addition, since Microsoft still holds a monopoly in their traditional business areas, operating systems and office applications, and it would be demonstrated that they cannot be trusted not to abuse that monopoly, the court should break up the company into at least three parts: e.g., operating systems, office applications, and "other". Each new company's products would be allowed to use only publicly-documented properties of published interfaces of components it uses that are supplied by other companies.

Since a requirement to publish detailed specifications allows far too much room for outright fraud -- how accurately have they documented the functions they _meant_ to be public? -- the form of publication should be source code. In other words, they should be forced to release the source code not because that source code is intrinsically valuable, or because they don't (or pretend not to) want to release it, but because that is the only form in which we can trust they have actually fulfilled their documentation obligations connected with the breakup of the company.

Nathan Myers is best known for his work on the ISO/ANSI C++ Language standard. He has used C++ since 1986, and GNU/Linux since 1994. His previous publications have appeared in Dr. Dobb's Journal and C++ Report. Contact him via his web page, http://www.cantrip.org/.