Microsoft's problem is that it can't answer the question "If I am switching to a totally new OS, why does it have to be from Microsoft? Why not [any current non-Windows OS]?"
I do not mean this as a quality slam, it's a business observation. Give up windows and they're suddenly just another entrant, and what business sense does that make? Other OSes already do virtualization with software, and it wouldn't take that much to put together a %buntu that natively supported VirtualBox and Windows, where you stick the XP/Vista/7 install CD in at a defined point in the installation.
Better to just start dumping their biggest bits of baggage over time and slowly migrate to better tech than have a big bang "everybody dies" big OS. They can get there just by lowering the priority on backwards compatibility a little, rather that slamming it to 0.
This bears a certain striking resemblance to what they are actually doing, oddly enough. Either way, it can't be a fast process; a brand new OS would take years for the ecosystem to spin up, and these changes will take years. There's no fast solution.
I think that you're certainly right. The biggest problem is outlined by your initial question "If I am switching to a totally new OS, why does it have to be from Microsoft". I think, that such a question shouldn't, ever, be guiding a companies strategies. But for Microsoft, it is. Their products legitimate their existence through their Windows-Brand heritage.
The actual question that they _should_ ask themselves should rather be "What innovative features could we create, that would make customers crave for our products".
Sadly, their whole corporate culture has stunning difficulties coming to terms with this line of thinking.
A new Non-Windows Microsoft-OS might actually work, if it would be crammed full of breathtaking, innovative, extraordinary great features (whereby features I mean useful, life-enhancing ideas, not gimmicks). And it's very, very difficult to start with a blank slate and come up with something distinctly different but still incredibly awesome and useful.
Windows can already be virtualized. The problem is accessibility and cost.
We've seen that most users can't be bothered to upgrade machines. You have to give them an old-car, new-car model: copy all their precious data, then throw their entire box away and give them something new that happens to be running Windows virtually.
A virtual PC must be as easy to buy as an iPod.
It must be cheap (free of any nonsense about not being able to reuse old upgrade-only copies or otherwise paying again for Windows). A program that would let people trade in old PC hardware for fast new virtual boxes would be ideal: something that copies their entire Windows image onto it right at the store, preserving their files, much like you'd trade in your car and keep whatever stuff you had in the trunk.
And, it must start looking "uncool" or almost unheard-of to use anything else as a Windows PC, similar to the effect the iPod has had on other music players. It has to feel "ridiculous" to the average person to keep using some old clunker PC, it can't just be something that tech people would think.
Microsoft have had a "WIN" project one the go in one form or another for years. "Cairo" anyone? For years, Longhorn was going to be an almost complete replacement of the platform. They've never got it to work though, perhaps because it's always been an engineer driven revamp rather than a UI driven one.
To get something like this done they'd probably be better off abstracting the OS and making it insignificant in the way as web browsers do. Imagine the Windows Phone 7 OS running on laptops and tablets, all silverlight driven and no sign of a filesystem anywhere. Then they could release a version that runs on classic Windows, OSX and linux.
From what I understand of the (very preliminary) plans for 8 is to integrate some of the elements from Surface, finally do away with the registry, and hide much of the filesystem from the end user among other things that may be more or less surprising to some. Backwards compatibility will still hang around in the form of virtualization.
Microsoft is still an innovative company that needs to do a better job fast tracking R&D to real world applications.
I do not mean this as a quality slam, it's a business observation. Give up windows and they're suddenly just another entrant, and what business sense does that make? Other OSes already do virtualization with software, and it wouldn't take that much to put together a %buntu that natively supported VirtualBox and Windows, where you stick the XP/Vista/7 install CD in at a defined point in the installation.
Better to just start dumping their biggest bits of baggage over time and slowly migrate to better tech than have a big bang "everybody dies" big OS. They can get there just by lowering the priority on backwards compatibility a little, rather that slamming it to 0.
This bears a certain striking resemblance to what they are actually doing, oddly enough. Either way, it can't be a fast process; a brand new OS would take years for the ecosystem to spin up, and these changes will take years. There's no fast solution.