COMMENTARY-- Windows-based PCs are too hard to use. The cure: Microsoft should let its programmers start over with a clean slate and completely revamp the OS. Want proof it'll work? Just look at Apple.

There's a major problem with Microsoft Windows that nobody seems to be doing anything about: It's not getting easier to use. And it should be.

I'm not talking about the baby steps that make Windows XP easier to use than Windows 98 and Windows 98 easier than Windows 95. I'm looking for the kind of radical change that would make tomorrow's Windows as easy to use as Mac OS X.

No, this isn't another angry screed from a Mac user excoriating Windows for not being Macintosh. I'm simply saying that the biggest barrier to making computers really easy to use and ubiquitous is Microsoft Windows.

To change that, Microsoft needs to start with a clean slate.

It bothers me when I hear Microsoft describing all the powerful new features that are supposed to show up whenever Longhorn, the next major Windows release, finally makes its appearance. Yes, a new file system should make it easier to find related pieces of information--eventually. And, of course, I'm all for changing the programming model so programmers can think more about their applications and less about low-level plumbing.

But what I'd really like to see is Microsoft designing Longhorn to make home users and workers happier and more productive. I'd like to see the new operating system do a better job of anticipating user needs and offering solutions users didn't even know existed. I'd like to see a new, simpler interface that adapts itself to the way each of us works, by watching what we do every day and adjusting itself accordingly.

No, I don't know what this magical OS will look like, but I'd know it if I saw it. And note that this OS might be designed specifically for end-user machines, never to find its way onto a server. Note also that it may be time for Microsoft to reconsider its long-standing goal of having one OS that serves both home and office markets. Perhaps it's not such a good idea for all versions of Windows to be based on a single, unified code base.

Five or six years ago, back when we used to imagine that Microsoft had some hidden vulnerability, people used to ask me what I thought would become Microsoft's greatest challenge. Back then, I imagined that Microsoft might eventually face a rival operating system that had been born in the world of entertainment and consumer electronics devices, migrating from there to desktop and portable PCs.

That may still happen--but I think Microsoft itself needs to build that radical Windows replacement.

One of the things that would require: abandoning its slavish devotion to backward compatibility. Making sure that the OS still plays nice with older apps robs developers of the chance to truly innovate.

Microsoft could learn a lesson here from Steve Jobs--that you can occasionally hit the reset button by releasing a new OS that isn't compatible with everything that came before and live to tell about it. Apple has done so twice, with the original Mac (which rendered the Apple II obsolete) and then again with OS X (which relegated OS 9 to the scrap heap).

It's interesting that MS struggles to maintain all this backwards compatibility while releasing new operating systems that pretty much require new hardware. How many people really bought shrink-wrapped copies of Windows XP to replace their old copies of Windows 98? According to people I talk to at Microsoft, not many. Unless your computer is fairly new, you've always been better off staying with your old OS on your old computer, upgrading only when you buy a newer, more powerful system.

With that new computer, you'd presumably get new applications, too. Backward compatibility is supposed to help out software vendors, who'd otherwise be forced to build new apps every time MS upgraded its OS; they'd also have to support multiple versions of their products for all the different operating systems.

But let's face it: It's not like there are all that many big software companies left to be inconvenienced. The most important apps that would have to be compatible with a new Windows would be written by...Microsoft. Somehow, I think that could be arranged.

Sure, customers might have to purchase new versions of the applications they used to run (assuming updated versions of said apps weren't included on the new PC). But we're talking about a once-a-decade phenomenon here, not something that would happen every year. There'd doubtless be point revisions in between; but those wouldn't be so drastic as to break compatibility with existing apps. One other advantage of that time frame: People could plan their transition from the old platform to the new one.

I don't have space enough here to do more than raise the issue and scratch its surface. But I believe the best way to really improve the experience of using a personal computer would be to give Microsoft's developers a clean slate to work from rather than requiring that the next release trace its ancestry all the way back to DOS and the original Windows.

I think we can do better. Certainly, Steve Jobs has shown us it's possible.

What do you think? Should Microsoft scrap backward compatibility in favor of greater simplicity? Let us know at edit@zdnet.com.au.