Monoculture Software Cycles

One would think that at this point the picture would be complete. On one side stood Borland, the bastion of the secret world of free market software, and on the other end stood the open source movement, the antithesis of everything that Borland represented. But the really shocking thing is that Borland represented a sort of middle ground between what was happening in the open source movement, and its true doppelganger, the Microsoft software development process.

According to free market theory, monopolies are supposed to fail because their employees have no motive to work. With a market share of over 90%, both the Microsoft operating system and its office products fully qualify as monopolies. According to theory, these products should be bad by definition, since they are no longer driven by market forces. But these products were successful both in the marketplace, and in terms of technical prowess. Most versions of a Microsoft Windows or Office represented an improvement over its predecessor, despite the fact that they faced no serious competition. According to unfettered free market theory, Windows XP should have been worse that Windows ME or Windows 2000, or Windows 98. But there is no doubt that Microsoft keeps improving their products, despite the lack of true free market pressures that drives a company like Borland.

Microsoft telegraphs their every move to the world. They don’t, of course, share their source, but the whole world knows the name of most important Microsoft betas, and many of these betas are made fully public to anyone who wants to participate. The exact contents of each product — no, the wildest dreams of even the flakiest project managers — are broadcast as front page news to the world when Microsoft is in a development cycle. Borland works in secret, Microsoft in public.

If something goes wrong during a Borland development cycle, a wall of denial goes up. But at Microsoft, if innovation is dropped because of technical necessity, Microsoft shares the defeat with the entire world. And because it is a monopoly, it suffers these defeats without loosing market share, without loosing market value.

So both Microsoft and the free software movement work in the open, and ship when they want. But there are profound differences between their models. Microsoft does not churn out software five or six times a year like an open source project, nor once or twice a year like Borland. Instead, Microsoft turns out software on a glacially slow schedule. It is not at all uncommon for three to five years to pass between iterations of major Microsoft products such as Windows, Office or MS SQL Server. Nevertheless, neither of them are bound by a fixed length to their development cycle. And neither of them bothers with secrecy. They both make a point of broadcasting their plans to the world years ahead of time.