To be honest, I am a bit Microsofted-out after the last few weeks, but I have one last thing to mention before I go on a Microsoft-moratorium (which will only last as long as they fail to make some newly egregious newsworthy move in the marketplace, I'm sure). That thing is the new operating system, singularity, released recently to Codeplex by Microsoft Research.

Singularity is not intended as a Windows 7 precursor or in fact as a "true" Microsoft operating system at all; instead, it's a think-piece, an academic OS offering proof-of-concept design characteristics which may eventually find their way, in bits and pieces, out of rarified academia and onto our desktops. Researchers stress the dependability and security offered by the system concepts, surely emphases born out of recurring frustrations the company has had with Windows on those fronts.

Larry Dignan reads this as a subliminal cry for help from the company, an expression of a desire to ditch the crippling legacy code base of Windows and start over fresh with something better. The obstacle, as he sees it, is the ever-present crux of backward compatibility... the next Windows, or whatever the company calls any OS, needs to run the same applications as the old one. Should the company bite the bullet and cut Windows off for something new and better, he wonders?

I think it's a false choice, though. The answer, absolutely, is yes; Microsoft needs something new, dependable, and lightweight to offer the market. This need will only increase as SaaS continues to grow and the desktop operating system as we know it becomes more a liability than an asset. In order to continue to sell operating systems, Microsoft needs to both make them more secure and less costly to produce... both which needs seem to be fulfilled by Singularity.

As far as backward compatibility goes, I'm surprised it's even a question anymore, frankly. Virtualization is the other big story of 2008 and it's hard to imagine a new OS that wouldn't have robust virtualization features built in from the outset. Singularity's SIP isolation concept seems ready-made for it, in fact. Considering that, it seems a short step to building in solid Windows virtual machine support to allow legacy applications to run seamlessly within the framework of the new OS, and far more securely than they can on either XP or Vista today.

The real question is whether or not Microsoft can capitalize on the concepts being developed in Microsoft Research, which absolutely includes some cool stuff, or whether it will be another Xerox, forever consigning the visionaries from PARC to a niche outside the mainstream of the business.