Rather than waste precious brain cells explaining why Randall Stross’s shrilly anti-Windows op-ed in the New York Times (Windows could use rush of fresh air) is full of crap, allow me to outsource the job:
Anonymous Microsoft blogger Shipping Seven, New York Times gets it wrong:
Trying to future proof software over more than a decade is like trying to predict the weather at 2pm, on the 4th of July 2045. Windows is pretty modern, just like everything else out there. I can’t really think of any ancient technology in Windows that is beyond salvage; everything important that needs to be overhauled can be overhauled, and has been…
Robert McLaws, I Can’t Believe New York Times Let This Get Published:
I believe that Mr. Stross has allowed his prejudices to cloud his search for the real facts. He uses Singularity as the impetus for his overall argument (hardly a novel strategy, as others with similarly nonexistent experience with this research OS have also taken this tack) without ever actually using the OS that he suggests should replace Windows on the at least 800M PCs it is currently running on. Singularity is designed to help people re-think application isolation for robust security, not for getting your 9-year-old printer to work.
Paul Thurrott, Randall Stross jumps the shark:
I am freaked to be saying this, but you, sir, know absolutely nothing about either Windows or Mac OS X and shouldn’t be giving this kind of advice. Shame on you for publishing such a story. Microsoft is right now working on further componentization of Windows ("MinWin"), a project that could very well result in the type of "just-enough" OS that, no, Apple doesn’t have today either. But even today’s Windows versions (Vista and Server 2008) are architecturally and factually quite different–i.e. "superior"–to what you’ve described.
Brandon Paddock, NYT article says we should throw away Windows:
Stross and others seem to be under the mistaken impression that Microsoft is somehow unable to change the existing Windows codebase. These guys present two options:
1) Build stuff on top of the last version of Windows
2) Start over.
Why pretend that these are the only two options? Especially when historically Microsoft has always chosen door number 3:
Take what you have and make it better.
Replace the parts that need replacing.
Don’t break something without a good reason.
For a professor of business, Stross sure is fast and loose with his facts and sloppy with his research. I lost track of the number of gross factual errors I read in Stross’s original article. Here’s my favorite part:
Microsoft … should take heart from Apple’s willingness to brave the wrath of its users when, in 2001, it introduced Mac OS X. It was based on a modern microkernel design, which runs a very small set of essential services that make the system less vulnerable to crashes.
Quick quiz: When was the Windows NT kernel (which is at the heart of all current Windows versions) written? When was the Mach microkernel (which is at the heart of all current OS X versions) written? Answer: Both date back to the early 1990s. So why is one more “modern” than the other?
Ed Bott 
“Quick quiz:”
Bonus quiz: Who led Mach development and where does he work now?
“Bonus quiz: Who led Mach development and where does he work now?”
I believe Mr. Rashid works for a company that rhymes with “Tycrosoft”… 😉
Mr. Bott,
I really appreciate your well thought-out and intelligently-presented blogs. Spot on.
On the micro kernel bit, he does realize that NT is just a step away from being micro, and OS X isn’t there either, bastardizing the Mach kernel in order to try to do what Microsoft did with NT.
1975-1978 Aleph microkernel development (3yrs)
1981-1985 Accent microkernel, based on Aleph (4yrs)
1985-1994 Mach 3 microkernel used in NeXTStep->OpenStep, based on Aleph (9yrs)
1985-1989 NeXTStep 1.0 launched; Objective-C frameworks concurrently developed. (4yrs)
1989-1995 OpenStep 3.3 final version, made available on four processors: 68K, PA-RISC, SPARC, x86 (6yrs)
1997-2001 Rhapsody+XNU microkernel->OS X 10.0 PPC release (4yrs)
2001-2007 5 versions of OS X developed, transition from PPC to x86, full concurrent 32/64-bit compatibility modes, MacOS Classic transition layer, etc. (6yrs)
All told, to develop what became OSX from the original microkernel (Mach 1, 1985), we’re looking at 27 years. Even if Microsoft used available kernels and took half the remaining time to develop frameworks, a gui, programming API’s, etc. adopted from Windows, it might still take them 10 years… and that doesn’t include time to get it right on the second or third version.
The New York Times doesn’t have fact checkers (and neither does this Web site, obviously).
Apparently, Joe, your sense of humor took the day off.
Well i have to agree with Ed Bott and the many Commenters when i say “Good Journalism starts with Real Facts, not BullS*it.”
Just like Randall Stross to go Berserk and Make out a Hefty Big Kahuna like that. I understand his point, When up to XP, you were Dealing with a NT floor plan on a Supposedly new Product. but still…
When it comes down to it, Security ends up being the hot-head Issue with Every OS, From Windows, OSx, BSD, Unix, Sparc, Linux, OS2, ETC. but i still hold the same idea that Kevin Mitnic Swore by in 1989, when ever there’s a new OS, A security flaw will be discovered. because there’s always some one out there that wants to poke holes in your products.
So i say, Work on the GUI, Make a Far Faster more Stable Kernel if you want, but what ever you do, dont lose Application and hardware support. because thats where all the real money is with, when Corporations update rather than upgrade.