Ground Work

In West Africa it is not uncommon for people to build walls of a new home season by season, as resources permit. Many such half-built homes can be seen, with trees over-growing their interiors. Eventually the walls may be finished, and their vegetation cut down, but it may take years. A new generation may build upon the work of the last.

Although the stakes are less elemental in software development, the vicissitudes of economics determine how many castles are completed, and how many are left hanging in the air. The case to be made for open standards and open source building blocks is simply this: that a wall once erected should be of some use even if the author moves on, or the economy turns sour. More often than not, standards have been lacking, and walls have crumbled. Which brings us to Windows.

Whereas clay walls will stand for a lifetime, digital structures and systems mutate much more quickly. In the case of Microsoft's system software, the name convention "98", or "95", implies this planned obsolescence. When compared with other operating systems, it is clear that the arcane, opaque and unmanageable nature of Windows' guts compels users to "upgrade" regularly. All manner of obscure incentives drive their marketing policies. Programs like Office 98 bear the "98" tag as if to warn users that the Damoclean sword of incompatibility hangs over their heads if they fail to purchase ALL of the upgrades.

Rapacious as it is, Microsoft does not control all the mutations of the marketplace, however, as the ongoing Department of Justice lawsuit suggests. In the whirlwind dynamics of the computing industry, there are factors that escape prediction, much as there are unforeseen contingencies in satellite launch failures.

Rich Murphy (NASA engineer) explaining why the Delta III exploded: "When designing the control system, we had to model the vehicle.... So one can theorize that the model was not correct. It did not model the vehicle properly, and therefore the control system, which was designed to the model, was inaccurate. Or the vehicle model was correct and the control system's design was wrong or in error somehow. Those are the two cases I can think of, short of having something that came loose or broke in the hardware and therefore the control system was unable to compensate for that."
T his multiple choice depiction of the problem scarcely conceals the underlying fact that they @!#*&ed up (to the tune of $250 million). The salient "it" at the heart of the Delta III explosion, may also have caused the system crash that greeted Bill Gates as he presented Windows 98 at its gala spring unveiling. Which model was to blame? Perhaps it was simply a matter of hardware failure. However, massively contingent development projects, whether to build operating systems or rockets, are fraught with potential problems; and it may be that the model one ought to examine is of a higher order: the coordination and cooperation of labor. As multi-national corporations now eclipse nation states with respect to their size, complexity, and budgets, the Cold War may have lessons to impart to industrial giants like Microsoft. For if it is the case that technology and industry evolved more slowly within the centrally organized Soviet block, then it may follow that distributed software development poses a similar threat to Microsoft. Could the Gates of Redmond, Washington fall as unexpectedly as the Berlin Wall?

In spite of its domineering stature, Microsoft has good reason to fear attempts to implement standards outside of its control. Java is a case in point. The Java language has been adopted more quickly than any previous computer language, and it has done so in spite of Microsoft's wishes to the contrary. Unfortunately, in spite of its many virtues, the Java language is chiefly a technology directed by one company. Seeing this problem, Sun applied to the International Standards Organization for official recognition of the Java language as a standard. This could open the door to public debate over the merits of specific features. It is clear that the pace of development that has driven Sun, as it expands the functionality of Java, threatens to undermine the whole project. It remains unclear whether the people at Sun will be able to apply their corporate motto, "the network is the computer", to the evolution of their own brain child. Increasingly it appears that Java will not be widely adopted as a standard.

By contrast, the relative smoothness of the growth of the free operating system, Linux, stands out. Fully a decade younger than the Microsoft and Apple operating systems, Linux is by far the most stable. Although its market share remains small (less than %5), the open source code movement that it has successfully launched, has many wondering how far it will go. Impressed by Linux's example, Netscape Communications decided to release the source code for its browser Navigator, (then) the most widely used World Wide Web browser. Against claims that disorganization will ensue, Linux stands as a paragon of cooperative software development. Similar "cottage industry" projects have sprung up all over, and are literally "staffed" by residents from all around the world. Linux technical support abounds on the internet, because many people want to help the project forward.

Against this array of networked workers, the traditional corporate monopoly appears somewhat inert. Distributed production, with open source code licensing agreements cannot be repressed through acquisition. The sheer numbers of individuals contributing, around the globe, around the clock, make this a movement to contend with. The lumbering, inefficient Microsoft "vehicle" charts its course of acquisitions, industrial sabotage, and infantilization of consumers, all the while fearing a miscalculation, a bad model. And in fact, theirs is exactly the model they should fear.

© Andy Deck 1998