JUNE 2000 IN

Microsoft Faces the Blue Screen of Death

When people start to use a descriptive phrase to refer to your computer product, it is usually not a good sign. Some years ago, you might have been tipped off that the then-dominant word processing program was about to become a has-been when users, frustrated at the program’s inability to do a whole range of normal computer things, started to call it “Manual Typewriter.” When a file server software vendor’s name became synonymous with The Server Is Down, it was a sign that customers were all too ready to jump to the more-stable products of its competitors. And when the users of one brand of removable-media storage products couldn’t seem to remember the manufacturer’s name as easily as the phrase Disk Error, you could see a hint of the grudging quality of the company’s “market support.”

I think it may be a similar story with Blue Screen of Death, the unofficial name of the Microsoft Windows NT operating system, at least among some of its users. I should explain that the blue screen of death is a text display of a hexadecimal diagnostic message that may appear in unrecoverable system failures. It always comes as a nasty surprise, because in the instant that it flashes on your screen, you know that whatever you were working on before that moment is gone forever. Windows users have to expect frequent system crashes, but when new users of Windows NT compare notes, the blue screen of death seems to be all they talk about.

The hourly system crashes become less frequent as users adapt, but the blue screen of death remains the most memorable feature of the Windows NT user interface, a monster always lurking in the shadows, ready to strike at any moment if you do the wrong thing. The same way you learn how to stay alive in a shoot-em-up video game, you can unconsciously adapt to an unstable operating system, avoiding actions that resemble the things you were doing when the system crashed in the past.

I experienced this process, not with an operating system, but with Microsoft’s word processing program, Word. The current version of that program trashes large sections of documents at random times. I’ve managed to minimize that problem by avoiding certain actions. Making sure my documents are no more than 5 pages long has helped too, although, since I’m a book author, that creates a large amount of extra work.

You might think I would have given up Word by now, but I haven’t. I have the idea in the back of my mind that there should be something better, but I haven’t gone to the trouble of finding it yet. Similarly, the experience of ten crashes a day hasn’t been enough to make any Blue Screen of Death user go back to Windows 98 — at least not among the people I’ve talked to — but they must have the idea, in the back of their minds, that they deserve something better than an operating system that is so unstable.

To be fair, such system crashes are rarely created by an operating system acting on its own. They result, usually, when an application program does something wrong — when the application asks the operating system to do something that is incorrect or unreasonable. But the operating could still handle those situations gracefully. That is, even when an application requests something incorrect, that doesn’t mean the operating system has to crash.

And there is another reason that Microsoft must take the blame for the instability of its operating systems, an issue that contributed to the recent court decision to break up Microsoft into two companies. Microsoft has been purposefully obscure in telling developers how to develop Windows applications. The court found that Microsoft withheld information in order to give its own applications developers an advantage. By withholding some information and providing other information that was confusing and contradictory, Microsoft contributed to billions of Windows NT system crashes — crashes that were much more frequent than they had to be because Windows applications developers quite honestly did not know how to develop applications that would work consistently with Windows NT.

Microsoft’s strategy of giving an unfair advantage to its own applications developers was a key factor that led the court to decide that Microsoft’s operating systems developers and applications developers must be put into separate companies. This court decision is a more serious blow to Microsoft than it might seem on the surface. At the time that Windows 3.1 became the leading operating system, it was laughably clunky, and it was only the Microsoft Office application suite that made it look good. Microsoft now needs to defend its position against the seemingly unlimited momentum of the Internet, Linux, and other current trends. The success of the Internet has demonstrated the power of open standards to create stable, reliable software. Internet standards such as the Web and Java have already made operating systems less critical. Software built around new standards such as XML may soon offer easier, more reliable, and less expensive ways to do all the key things that Microsoft products currently do. Microsoft will have to do something new to stay competitive and to overcome the hits its image has taken in recent years — the antitrust case and the blue screen of death are just two of many problems Microsoft has created for itself that it now has to face. And, unless it can overturn the court decision on appeal, it will have to do so without using one part of the company to cover over the weaknesses of the other part.

Microsoft founder Bill Gates is genuinely worried. He looks worried, he sounds worried, and he has talked openly about the threats that new technologies such as Linux and Java pose to Microsoft. As for splitting Microsoft into two companies, that would be unimaginably bad. So Microsoft has to do something.

Microsoft customers probably wish the company would just deliver reliable versions of its current software. But, like it or not, Microsoft has to think bigger than that — it doesn’t make much money on maintenance upgrades — and the strategy it recently unveiled continues to show much more emphasis on high concept than on practical considerations.

Suspiciously dubbed “Microsoft.net,” the plan is actually something more than an effort to pass Microsoft off as an Internet startup. Microsoft hopes it can undercut the Internet’s momentum by turning Microsoft’s current line of products into a sort of Web operating system. Microsoft plans to build around Internet standards, especially XML. At the same time, it plans to keep its processes proprietary, integrated, and obscure enough that it can freeze out competitors and continue to charge premium prices for its software and its new Web-based services. Gates said Microsoft was ready to “bet the company” on this strategy.

I realize I’m missing a lot of the details, but from what I’ve read about it, I don’t see how it could possibly work. Microsoft’s only major experience with open standards has been its web browser, Internet Explorer — and that was a product that became a success only after Microsoft lowered the price to zero. The management style and cost structure of Microsoft won’t allow it to continue to operate if the same thing happens to the prices of its other key products. But there is no hint that Microsoft has given any thought to how it can resist the market pressure that pulls prices of standards-oriented software down toward zero. That part of the plan may be nothing more than wishful thinking.

There are also serious questions about Microsoft’s ability to deliver. Microsoft’s early experience with OS/2 and its more recent experience with Windows NT 5 — major projects that failed outright after long delays — suggest that it is not a company that is capable of quick, large-scale software development. Microsoft CEO Steve Ballmer admits that Microsoft.net will take two years or longer to deliver. Realistically, it might take much longer than that. Considering the turmoil that is likely to result from the process of breaking up Microsoft, the other current problems it faces, its track record of long delays in big projects, and its head-in-the-sand response to the countless bugs in its current shipping products, the core elements of Microsoft.net could easily take ten years to develop.

Or they might never happen at all. When your name has become synonymous with Blue Screen of Death, you don’t have ten years to take the next step forward. You can’t be absolutely certain that next week is soon enough. When your customers are already looking for something better, your most dangerous competition is the company or product you don’t hear about until it’s too late.

Manual Typewriter’s fall was surprisingly quick — and surprisingly quiet. If the further details of Bill Gates’ new software plan turn out to be no more impressive than Gates’ testimony in the antitrust trial, Blue Screen of Death may be headed — quietly — down that same road.

Fish Nation Information Station | Rick Aster’s World | Rick Aster