Just read an interesting post by Scoble about the London train system and how they're still running DOS apps on XP machines. Of course if you're Microsoft you have to enable your OS to continue to run those apps because there are way too many users/companies in that type of situation to change.
On the other hand you're a 272 BILLION dollar company with a stock that's been stuck at $25 forever. So where's the growth coming from? Any applications you develop need to generate tens of MILLIONS of sales to even make a dent. So Windows and Office keep you afloat but xbox, mice, all the other stuff they sell is just a pebble in the pond. I used to be a stockholder, I've read the annual reports. That other stuff accounts for almost nothing.
So how do you juice more sales out of Windows/Office when you're stuck in the mud? You can't progress fast because so many businesses won't allow it. On the other hand everyone wants better performance which is nearly impossible because if you have to keep all your old code working it's simply unrealistic to expect them to be able to deliver greatly improved software.
I think this ends up leading them into the very scary place they are where they have to move really sloooooow on everything. Sure we'll deliver a new OS every 5 years, maybe. What about an updated database every 5-6 years, nothing much going on in that space right?
Hence the companies that don't want to move to begin with are given no reason to move because nothing new happens to the OS for years at a time. There's no compelling reason to switch and Microsoft can't give them one because they must support so much old code.
They've also managed to load themselves down with thousands of extra employees. As I've written about before this is really bad for them because all these extra people do is shove in features on top of old crap. It's not their fault, you have all these really smart people and they want to work and they want to innovate.
So here's one possible radical solution. Fork Windows. Well actually not even fork it but start from the ground up. Split up that huge workforce into 2 groups. The first maintains Windows XP, providing security patches, updating code to handle new faster processors and so on. They would do this basically indefinitely. The second group now devotes all their energy to an entirely new operating system, let's call it Ralph. Ralph is 100% new, incorporating every little thing they've always wanted to do with no dependency at all on running old code.
Now for any other company this would be suicide, but not for Microsoft. They have the money and the people to support the active development of 2 operating systems. That money also affords them the ability to slowly migrate users to the new platform. At first giving it away to companies like Alienware for use in high end gaming system. Then slowly as an option with mainstream brands like Dell, making it clear that there's no backward compatibility. Sure it will take at least 10 years until Ralph becomes the primary Microsoft OS, but that's not much worse than the 6 years people are/will be waiting for Longhorn and even then all their getting is a second rate product that is extremely limited by the need to keep things backward compatible. It also puts Microsoft in a much better position for the next 30 years than yet another spruced up version of Windows 95.
side note
I hear you all out there saying in 30 years everything will be online and your desktop will only be a terminal. Not so fast I say! That makes some big assumptions about the telecommunications infrastructure. Since the .com crash spending on fiber has gone down to almost nothing. We're a long long way from having fiber right up to our houses which is basically a required prerequisite to turning your desktop into an internet appliance. Look it's already been 5 years since then and my cable modem isn't faster than it was then, I still do my word processing on my desktop, I still can barely open a 200mb photoshop file on most desktops, forget about across the internet.