Windows as a standalone product is essentially dead, but selling it as part of a system for running Outlook, Exchange and Excel is not and won't be for a long while.
True, the OS kernel is an enabler not a feature.
> Windows as a standalone product is essentially dead
And on the server world it's a platform for IIS servers, SQL servers, and azure clouds.
Yes, a company that does this could decide to use someone else's OS kernel to enable all that. The original article that I linked to in grandparent comment explains why that would be a dangerous decision and is not going to happen.
Look at your examples for server stuff: it would make no difference to the end user if those were running on Windows or not. IIS gets value from ease of .net application deployment, SQL server from playing well with .net, and Azure can deploy Linux because that department can't afford not to be in the embrace/extending mode.
From an MS point of view making the OS might make these things easier to deploy, but just because you're good at doing something doesn't mean other people are going to keep paying you to do it, and so it will be with Windows. MS are going to end up disadvantaging themselves given they have to duplicate a lot of work going on in the open source community.
If you use Chrome in Metro mode on a Win8 machine you'll see the extent to which Google are outplaying MS, as it apes the Chrome OS experience completely. Google are leveraging the Windows user base to reduce friction to them eventually moving to Chrome OS.
IIS also gets value from Windows account permissions integration, from Computer Manager integration and MS Best practises Analyser, from PowerShell and VBScript and Visual studio integration, from Windows certificate services integration, blah blah.
The "Windows" benefit is integration, Microsoft products aren't standalone.
What benefit rewriting everything for Linux? Except alienating a huge chunk of Linux-ambivalent administrators.
I agree that it is MS's point of view on this topic. We're going to have to disagree that it's "misguided". IMHO Jeff Atwood was right on the mark in that essay, and it applies to this case.
Will it be enough for MS and for Windows Phone in particular? I don't claim to know. Would swapping over to android help? Unlikely, once you factor in the loss of face involved. Are others like Google doing it better? Probably.
That's your opinion. I doubt that it would get any traction inside MS.
It just doesn't make sense - are MS more likely to take the android kernel and reskin it to look like WinPhone, or continue to believe that writing OS kernels is a core company competence ... which it is, really. If one day MS could not write OS kernels any more they would be in serious trouble. It's a core business function, so they do it themselves.
How would re-skinning android help MS get "in front" anyway? They have technically solid product in Windows Phone, they need other things to make it win, and they know it. Hence, the whole Nokia deal.