'

Microsoft: The comeback tour

Microsoft's extended upgrade cycle seems as unfashionable as a gatefold LP. The market may be moving too fast for yesterday's supergroup

When Wired magazine was at its peak, it pulled off the impossible. "Make these nerds look like rock stars" was the call to arms, and for a while it worked better than anyone could have hoped. Perhaps too well: it seems that the crueller side of the music biz may be taking over at the heart of the soft machine.

Take the greatest code-rock dinosaur of all time: Microsoft. There are signs of creative differences at the heart of this supergroup -- for starters, Microsoft has just reorganised its Windows core code division, an odd thing to do half-way through recording the massive Longhorn project.

Then there are the worrying album sales: a recent survey of nearly 400,000 PCs in 670 companies by US consultants AssetMetrix showed that Windows XP has totally failed to impress -- at a piddling 6.6 percent penetration, it's by some way the least popular version. Even R&B classics NT4 and Windows 95 manage more than twice that, at 13.3 and 14.7 percent apiece. The most popular? The Dark Side Of The Moon of the OS world, Windows 2000, running on over half the computers checked.

Even in the home market, where nobody's had much of a choice of operating system for a while, there's been no thundering rush to upgrade. Google reports that while Windows XP is the most popular operating system used by searchers at 38 percent, Windows 98 is not far behind at 29. That's a five-year-old operating system holding its end up most persuasively in an Internetted, digital-media-rich environment very different from that of 1998.

You can if you wish tip your hat to Microsoft for producing such a flexible, extensible and useful OS: I'm not sure you'll be thanked. The company finds itself in the role of ageing rockers previewing difficult new material to an audience who just want the old favourites, the lead singer snarling in frustration when the calls come up from the crowd.

But there are good reasons why the golden oldies keep striking the right chord. Software doesn't wear out the more you use it. Quite the opposite: once you've got it doing what you want, it'll do it until the end of time -- or until something else comes along and spoils the song. Everything's hunky-dory until there are changes.

New operating systems are the worst, as update after bugfix force mutations throughout: old operating systems are known, stable, safe. If you don't change anything over a period of time, even those parts of the computer you rarely visit should stay lemon fresh.

In such circumstances, the news that Microsoft is to start ending support for Windows 98 early next year is a positive boon. It's not as if we need that sort of help. The world is full of experts keeping old Windows alive and productive: after all this time battling with the beast without too much help from above, we've domesticated it and can milk the old cow without divine intervention. Critical security patches? Microsoft can no more stop issuing those than Boeing can ignore rudders jamming on old 737s.

The message to Microsoft is clear: we don't come to you for innovation and the latest, greatest thing on our desktop. Once we've got what we want -- even if it is some time after we thought we'd paid for it -- we'll keep it, thanks.

Has this got through? Something has. You don't reorganise your most important division in the middle of what you've called your most important software project on a whim. It might be a sudden focus on deadlines due to impending release -- but with Longhorn still three years away, that's a very atypical depth of vision.

It could be the realisation that those upstart young punks called Linux have something new and better, a model of loosely coordinated specialist groups clustered around a tightly controlled core that manages to produce good, stable code and be flexible at it. That would mesh well with a loss of faith in the model of periodically rolling out a massive upgrade with all the ponderous implausibility of a Yes triple album.

When your main product never wears out -- indeed, improves with age and use -- you can't go on pretending you're eighteen forever with something fresh and important to say with each outing. There will always be a need for new products, just not the whole shebang -- and anyway, the market decides what it wants and when it wants it. In a world where a new idea can spread across the Internet in a month, why wait for 2006 to get them all at once?

Microsoft has known this for a long time, in much the same way that yesterday's rock gods, awash with fame, drugs and groupies, know that sooner or later the inexorable laws of fashion, toxicology and microbiology will stop the show. It's tried to reinvent itself as a service company, as king of the Internet, and even dabbled with street cred youth culture in Xbox and mobiles. Lots of flash, but not much cash.

The reorganisation may be the first signs that it's resigning itself to the populist option: a few new songs, a lot of remixes and milking the back catalogue for all its worth. Let's just hope that we never have to suffer Bill Gates - The Christmas Single.