OneAndOneIs2

So, we all know Vista & Office 12 are both delayed. Again. Apparently, after working on Vista for several years, MS believe that just a couple more months will allow them to remove the last few flaws.

Who knows, maybe they're right. But it's made it abundantly, panifully clear that MS cannot go on the way they have been. The system is broken, in fact it's FUBAR. They need a new way to go.

What's the cause of the problem? Backward compatibility. In two ways:

Firstly, you've got MS's fondness for taking the current project, be it the NT kernel or the Office suite, and just adding new stuff to it. This gaurantees that you've got code years out of date running stuff it was never intended to cope with. Ugly hacks abound. This was a big cause of delays in Vista: The problem got so bad, they literally had to throw away a whole load of work and start from scratch.

Secondly, you've got the insistence on being able to run the old stuff. Don't laugh. It's true that software for a Windows more than a couple of versions old tends not to run (My original Lesiure Suit Larry disks merely laugh at XP, for instance), but you'll mostly be able to upgrade from current Windows to new Windows and still use your existing software.

We all remember Apple's OS X approach: They turned their backs on everything that had come before, and created a whole new OS. OS 9 didn't get a look-in. Result: A very well-designed, modern OS.

Microsoft feels they can't do this. They may be right. After all, if you have a load of stuff that only runs on Windows, and then the new upgrade won't run it, your choices are:

Don't buy the new upgrade - result: MS makes no money

Get hold of Linux & Wine and hope it runs your stuff - result: MS makes no money

Use something like Citrix to use old versions of Windows on a new desktop - result: Well, it makes *some* money, but it makes the new OS a really hard sell.

So MS has a dilemma: It can't turn away from backward compatibility, but it can't go on trying to make a new OS with old OS parts.

My predicition, therefore, is that Vista will be the last of its line. Microsoft will do an Apple, and start from scratch. In fact, I'd be unsurprised if they copied Apple so far as to build on a BSD kernel. Given MS's lousy reputation for security & code cleanliness, I'm sure the benefits of starting out with a kernel renowned for being well-written are obvious.

Or maybe they'll just stick with NT (again) but give it a thorough rewrite to take out all the old cruft. Either way, they'll build a totally new version of Windows, and eliminate all the tired old mistakes that have hobbled them for so long. Like being a true multi-user OS instead of one-user OS that's been hacked to allow multiple users. You know the kind of thing.

All their familiar old problems will finally go away with the new start. The only issue will be the lack of vital backwards compatibility.

You've heard of VMware, I'm sure. You may even have heard of Qemu - If not, take a look at DamnSmallLinux and its ability to be run from within Windows.

I doubt Microsoft will actually go so far as to recommend either of these products: Going by their track history, they'll make their own & sell it as an add-on: Buy the new version of Windows, beautifully designed but hopeless incompatible; But if you need old stuff to run, buy the "Vista compatibility layer" product as well.

This would appeal to MS for two reasons: They get shot of all the problems that have caused all the delays and bugs and security holes; and they get to charge you for something they currently supply free.

They couldn't really have done it before: The techn wasn't there. The hardware was too slow & the virtual-machine stuff wasn't all that hot. But with the performance savings they'll make with a sleek new OS, the overhead of running an 'emulator' of their old OS will be minimal enough to escape notice.

I'm fairly certain they're going to do the start-from-scratch thing. They may be able to do something very clever & make a completely new OS that's still compatible with the current one, but I have my doubts. I reckon they'll do it via a Wine/VMware style additional product that lets old-OS software run on the new OS.

Because if they try and get yet another version out of their current codebase, they're going to collapse.