Software comes and software goes. You try things, they work or they don't and you move on. Personally, I don't think the story here is Longhorn's (or more accurately WINFS) ultimate failure. That's a symptom. The story is the icepick through the heart of Microsoft's revenue stream that results from their next generation "Must Have" becoming an "already have".

Microsoft is in a position similar to the big Automobile companies a few years back. Remember when cars had a lifespan of about 100,000 miles -- after which the car was effectively useless? That changed. Precision manufacture, higher grade materials, computer modeling, and most of all better quality lubricants doubled that number for most cars and suddely the car companies couldn't count on new cars flying out. They had to compete by offering more and more features. The results are visible in what we're driving today -- almost anything you buy in any showroom is truly a precision instrument that would leave jaws hanging open if shown to a dealer as little as 15 years ago. The lack of major new features in the next rev of windows leaves users in the same position as a car owner today with 75,000 miles on the car. NO REASON TO UPGRADE. If you sell the upgrades, that hurts. It hurts alot.

At the same time, Microsoft goes into the next five years facing an epic battle for the their prime customer base.

IBM seems finally to have gotten its act together (IMO they have Dr. Goyal to thank for that) and begun executing a mail and collaboration strategy that should satisfy its devoted existing customer base and at the same time play to its long time strength selling to the I.T. people behind the glass wall. The people who won't like the new strategy much are the departmental project teams and the end users. IBM is really going to have to prove itself to those people. At the same time, the free love -- er, software -- linux and open source crowd is really nibbling away the lead Microsoft has long held on users at the desktop. Keep in mind that those users were only reluctantly dragged away from their Apple computers in the last big battle (IBM XT vs. Apple Mac in the 80s) which IBM won by pushing network requirements down from central I.T. that the Mac crowd couldn't handle well at that time (LanMan, Netware, and 3270 emulation to name three).

Microsoft's hope for a renewed strength in this battle comes down to executing on two key tasks. First, they have to keep innovating on the desktop in ways that users really like. This has always been their strength and a strict comparison of IBM's new rich client should be within their realm of capability. For them, that's Sharepoint Portal Server and (quite by accident) OneNote.

The second is harder. They must base those innovations on technology that stays way ahead of the Linux crowd, forcing them to always play catch up. WINFS was a big part of that strategy. If they'd been able to push it out the door in a fast, secure, and stable manner they would have put the Linux folks at least a year and half behind -- a period during which anyone who wanted to work at home the way they work at the office would have had to buy a Windows machine. That's the game they've always been good at.

Sadly for Redmond, in something like 3 years they've been unable to build WINFS in a secure, stable, and fast manner that they could roll out the door. In software terms that points to major architectural problems. Software is complicated, but its not THAT complicated. If the design is a good one, there is no valid reason why it couldn't be functional at this point.

Microsoft going back to the drawing board on WINFS is metaphorically dropping its drawers for the Linux/Open Source crowd to finish their work making Linux a user-maintainable operating system. They may need to take a lesson from Apple on that score, but have no doubt they're doing just that.