First off thanks for all the thoughtful replies, I figured a post on forking Windows would get some people riled up!
This post is in response to some of the items mentioned by Martin Weber and others in the comments
First let me clarify my position. I don’t really care if they actually start from nothing. I would imagine in fact that they wouldn’t. They would use .Net and it’s common runtime as the basis of apps for the new OS and perhaps reuse parts of existing code that made sense. All I think is that they should try and remove themselves from maintaining tons of legacy code which runs apps from 1988 in the new OS. I agree this is suicide for most software companies, but they have the resources to pull it off.
It would also reinvigorate the workforce at MS, which is an aspect I think most peole overlook. Maintaining code which runs apps from 87 isn’t fun! How do you get the best minds to work for you if what you’re doing is boring work? See this post for confirmation from the inside on this.
Now Martin said, “This must be the single worst business decision Microsoft ever could make” but you can’t have it both ways because you go on to say that Microsoft has already done what I’m suggesting when they created NT, Tablet, Media Center, and Mobile. So if they’ve already done it why is it a bad idea and why don’t you think they could pull it off again?
Another item is the distractions of the other divisions. Martin says “Even if businesses like their Hardware division just are a “pebble in the pond” it definitely would be a business other companies would like to have just on their own”
Again I disagree. I’ve read the financial statements. Last year the business solutions division (CRM,etc) lost $255 million, mobile lost $224 million, and home and entertainment lost a whopping $1.2 billion. Those don’t sound like businesses which other companies would want to me.
All I’m saying is Microsoft needs to free it’s mind and find a way to inspire it’s customer base again. If there was a compelling reason to upgrade they would.