A Modest Construct

Show Stopper!
by G. Pascal Zachary

Publisher: Free Press

Year: 1994

Pages: 312

See the rest of this year's listings

What is 52 Books in 52 Weeks?

№46

While perusing the (absolutely wonderful: you should be reading it every day) blog of Jeff Atwood, I came across his review of Show Stopper!, and was intrigued enough to check it out myself.

I’m going to immediately confess my ignorance here: this heyday of Microsoft was long before I was ever interested in computers at such a level. In fact, I had no idea that NT was around so early. My thoughts while reading Show Stopper! were that Windows 95 (codenamed “Chicago” and mentioned only once in the book) made a much bigger splash than Windows NT did, at the time. Of course, Windows 95, being based on the ancient DOS kernel, eventually died out. It wasn’t until almost ten years after NT 1.0′s debut that the kernel was used in Microsoft’s latest consumer desktop offering, Windows XP.

It was difficult, too, reading something written in 1994, about 1994-era technology, as a well-informed software enthusiast in the latter half of 2007. I’m not sure whether Pascal’s writing seemed condescending because he was writing for people who might know what an operating system is, or because that’s simply the way technology books were written a decade ago. But the technical side of the book was lacking, focusing rather on the soap opera of NT’s development, headed by the notoriously gruff Dave Cutler.

If this book is about anything, it’s not so much NT, but rather a look (a) at a time in computing history where writing a new operating system from scratch was not only likely, but necessary, (b) at the group dynamics of building an extraordinarily complex piece of software with over 250 code writers, and (c) at the sort of environment that Microsoft cultivated in the mid-1990s. Now that Google is king of the world, it’s easy to forget that for a long time, Microsoft was basically The place to go if you were a bright programmer. Their salaries were below the industry standard, but their stock options made millionaires out of a goodly portion of its employees, but this came at a price. At least for those working on NT, Microsoft became their lives, and it destroyed a lot of relationships that way.

I could talk at length about the books foibles—e.g. Pascal’s insistence on giving clichéd descriptions of each character as he introduced them—but I suppose what’s really disappointing to me as a modern reader is that its scope is so limited: it’s not able to talk at all about NT’s eventual success, but merely make prescient statements about its revolutionary nature. Then, too, while the book reads like a traditional plotline, it never really climaxes: by the time NT is finished and released, everyone is exhausted (reader included) and the moment comes and goes with little fanfare. The development team sort of dissolves, and then Pascal waxes philosophical about the project for a while.

At the risk of going on a tangent, I want to make a few technical points before the feeling leaves me. It’s important to note that NT (and by extension, Windows 2000, XP, and Vista) are all essentially conceptual children of the Mach kernel, which is of the microkernel variety. Linux, by comparison, is monolithic, just like the DOS timeline of Windows systems (which ended ignobly with Windows ME). Microkernels are supposed to be safer at the expense of performance, but strangely enough, “safe” hasn’t really been the case for Windows.

Which brings me to another point that I think Show Stopper! underscores, and that is the heavy cost of legacy in the computing world, and the strain that business requirements put on technical innovation. The project scope of NT was redefined so many times that the end result was just about unrecognizable compared to the initial vision. The necessity of supporting, for instance, OS/2, Windows, and DOS code, bloated NT and significantly extended its development time. While a purely academic project may have delivered much better performance and the promised security, Microsoft’s real-world business requirements turned NT into something that, while still successful, would eventually draw as much criticism as praise.

Finally, it’s interesting to note that Dave Cutler, the computer genius, lead developer, love-to-hate-him antihero of the book, is still working at Microsoft. The 64-bit operating systems you’ve been hearing about since 2005 is largely the result of his work.

Swinging drastically back on topic, I can only recommend this book as an object lesson in real-world software development, and as a hugely interesting piece of software history. If you’re looking for technical details, stay away.