Last September the Wall Street Journal ran a fascinating lead article about Microsoft's Vista development effort. Robert Guth chronicled how the Vista project had initially ballooned as Bill Gates and others piled on their dream features, like the advanced, metadata-rich WinFS file system. When Vista hit trouble, Windows czar Jim Allchin brought in two software development experts, Brian Valentine and Amitabh Srivastava, to whip the project into shape by introducing rigorous new testing methodologies.
Still, by mid-2004 the whole project was in danger of collapsing. Microsoft decided to postpone Vista till "the second half of 2006" and cut back lots of promised features (including WinFS).
As Guth's article had it, the result, finally, was a development process Microsoft could begin to be proud of:
| |
On July 27 [2005], Microsoft shipped the beta of Longhorn -- now named Windows Vista -- to 500,000 customers for testing. Experience had told the Windows team to expect tens of thousands of reported problems from customers. Instead, there were a couple thousand problem reports, says Mr. Rana, the team member. |
When I read the article at the time, I took it as a kind of victory-lap valedictory for Allchin, who'd announced he was retiring once Vista was done. Unless you're certain of prevailing, though, victory laps are dangerous (just think of the phrase "Mission Accomplished"). With this week's news of a another slip in the Vista schedule -- the software won't be out until January 2007, after the crucial holiday buying season -- we're left wondering, what happened to that vaunted new process?
Certainly, this widely linked story that claims Microsoft is now going to rewrite 60 percent of the operating system between now and release seems hard to credit (something tells me rewriting that much code would take a lot more than 8 months). But between this embarrassing delay and the recently announced "reorg" of Windows leadership, it's clear that this turn of the Windows cycle is going to be no smoother or predictable than any of its predecessors.
My book, Dreaming in Code, is all about what I call "software time" -- the peculiar spell that software projects so often cast on the people involved, turning schedules into Mobius strips and stretching time like taffy. I imagine that, as Valentine and Srivastava described the beauty of their testing systems to Guth last year, they honestly believed that they'd meet their deadlines. They thought they'd cheated software time. That confidence doesn't look too smart today.
UPDATE: Steve Gillmor wonders whether maybe there really is 60 percent of the Vista code that needs a rewrite -- and much more. Adam Barr, on the other hand, offers some reasons why that notion might be far off-base.
|
|
