Will the Next Version of Windows Be Worth the Wait?

The New York Times > Business > Your Money > Digital Domain: "Longhorn's gestation has already extended much longer than originally planned. Rumors of its existence surfaced in 2001, when the system was said to have been chosen as a quick 'intermediate' update of XP. Time passed, and the news media were permitted a sneak preview. But completion of even this, the interim release, came no closer. Determined to get it out the door by 2006, Microsoft decided in 2004 to remove a new file system for organizing data on the hard drive, what the company had earlier promoted as the heart of the new system. If and when this feature ever appears, it is unlikely to enhance anyone's marriage.

The professional caretakers of corporate PC's seem rather leery of Microsoft's promises these days, spurning the most recent package of security improvements and bug fixes offered for Windows XP. Last week, AssetMetrix Research Labs, a research firm based in Ottawa, released the results of a survey of 251 North American companies, measuring the adoption of Windows XP. Only 7 percent of companies had actively embraced the latest improvements, Service Pack 2, released six months ago. The improvements, it turns out, introduce software-compatibility problems. These can be overcome with tinkering but not without aggravation and additional cost for fixes that should not have been necessary in the first place.

Mark Lucovsky, a software engineer, recently described in his blog the process of writing code for a project like Longhorn and the long wait before it reaches a customer's PC. First, a bug fix or added feature is deposited in a source code control system, where it may sit for years. Eventually it is transferred into a product release and pressed into CD's. Months pass, even in the final stage, from release to manufacturing to arrival at the customer's receiving department. Slow.

By contrast, engineers who work on improvements for a newer form of operating system, the software that powers Web sites, can roll out work almost instantaneously. Mr. Lucovsky recounts how a friend at Amazon discovered a performance issue, found a fix, tested it and had it in place, all in a day. "Not a single customer had to download a bag of bits, answer any silly questions, prove that they are not software thieves, reboot their computers, etc.," he wrote. "The software was shipped to them, and they didn't have to lift a finger."

MR. LUCOVSKY'S remarks are of interest because he knows a thing or two about developing operating systems. He was a senior architect of Windows NT, was the chief keeper of the keys for the source code and was named by Microsoft in 2000 as one among its inaugural batch of distinguished engineers. Recently, after 16 years at Microsoft, however, he said he decided that he had been wrong in thinking that Microsoft knew best "how to ship software."

It was other companies, the ones who understood the potential of the Internet and software-as-a-service, that were best able to deliver benefits to customers "efficiently and quickly," he said. He resigned from Microsoft and has joined one of those other companies: Google.


Randall Stross is a historian and author based in Silicon Valley. E-mail: ddomain@nytimes.com. "