Jupiter Research, on its Microsoft Monitor Weblog, revealed ignorance that plagues most technical reporters that are not real developers.  The post, Simple Security, is good overall, but hints that Microsoft should rewrite Windows from scratch in this quote:

“At some point Microsoft has to consider whether taking Apple’s bite-the-bullet approach isn’t the best route.”

The author backs up the claim with this:

“Windows XP builds on 10-year-old, pre-populist-Internet code.”

So what’s wrong with this argument?  Developers know, but apparently technical writers do not.  All these self-proclaimed “industry pundits” or “technical editors” keep calling for Windows to be rewritten from the ground-up.  My advice to you, if you see this claim made, discount the source heavily.

First off, how much of Windows needs to be “post-populist-Internet code”?  Well, with the newer, more modular architecture of Windows XP and Window Server 2003, only the networking and some integration with a few apps.  Does the bulk of the OS, which interacts with hardware, the CPU, memory scheduling, and all that stuff, really need to be rewritten?  No.  In fact, doing so will reintroduce a ton of bugs that have been fixed over the past 10+ years.  All of the fallacies in this argument are well covered by Joel Spolsky here.  Ironically in another post the author notes that Windows Server 2003 is more modular and this allows new functionality to be released without complete rewrites of the OS.

Second, most of the problems with running as a non-admin have to do with the applications, not the OS.  Applications can and are (occasionally) written to be run as non-admin (by good companies, at least).  If the author’s little girl can’t play her educational games, is that Microsoft’s fault?  No, it is the developer of the games that is at fault.  Microsoft even provides verification tools which include testing for running in a normal user account. But the author succumbs to blaming Microsoft because it sells.

Third, when the hell did populism become a software quality?  Oh right, never.  It is just more analysis blather.

Come on Jupiter, if you’re going to sell research, your analysts better know what the hell they are talking about.  With more and more senior executives having a technology background combined with writing like this, is it any wonder research companies are getting hammered?