No Private Vistas

Computerworld - Think Microsoft Vista's latest schedule slip doesn't matter? Think again. Sure, last week's announcement -- that the next version of Windows won't be out in time for holiday sales -- is more an industry sideshow than a big deal for corporate IT. It'll stunt PC sales at the end of this year and give Microsoft-haters something more to bleat about. But for most IT shops, that delay just means we'll start testing Vista a little later.

Forget those details. Look at the big picture, and you'll see a nasty object lesson in what happens when you (and others) put too much trust in your ability to deliver software.

That's what Microsoft executives did. That's also what vendors and IT industry pundits did, which is why words like "bombshell" and "unthinkable" are being tossed around to describe the "Vista slips" announcement. They believed. They trusted.

They shouldn't have.

Vista is the most complicated software product in Microsoft's history, incorporating higher- quality standards and a new approach to software projects -- and facing the same old "keep piling on the new features" culture. That's nothing to inspire confidence in a firm delivery date.

But Vista boss Jim Allchin promised that Vista would be delivered on time. Allchin is an executive who's well liked by his developers, and this product is his last hurrah. He staked his reputation on getting Vista out the door before he retired at the end of this year. And he believed, probably rightly, that his team would do everything in its power to make that happen.

Yes, they're all doing their best. But they can't work a miracle. Vista is irretrievably late. Microsoft is embarrassed again. PC makers insist that they won't take a significant hit in holiday sales, but Wall Street analysts are already recalculating their Q4 sales estimates.

Lost revenue, lost trust, a lost reputation. That's Vista's legacy, even before it's finished.

Ugly, yes? And it probably sounds familiar. Your IT shop likely has huge, challenging (and challenged) projects too. How do you avoid watching the next one become a Vista of your very own?

Start with the first rule of posterior protection: Underpromise and overdeliver. Or better yet: Make no promises at all.

If -- pardon me, when -- your boss, users or business partners demand a guaranteed deadline, give them a percentage chance that you can deliver the goods. That percentage goes up if they can accept functionality that doesn't work in the first version. But promise nothing.