My title, above, comes from a Shadetree comment from last week:
A minority opinion ....
... not based on facts. The best you can say is one platform is better for some things then the other. It cuts both ways. You would see that if you weren't such a Unix biggot!
The minority opinion in the case was, course, my claim that Unix is generally better than Windows - something IMS_scott also picked up on, labeling it "factually incorrect."
There's no doubt that Windows has the larger market share on desktop and small server computers, but to see just how little this says about its quality relative to Unix consider that the flat earth hypothesis held an overwhelming market share among western scientists well into the 17th century - despite the fact that almost all of them had been forced to read Eratosthenes of Alexandria's 230 B.C. demonstration to the contrary as part of their education in classical Greek.
The underlying issue with market share, of course, is that "better" does not mean "preferable" - Galileo's family was quite sure (as was he) that the earth centric view of the solar system was socially preferable, but it wasn't better. Thus when I say that "Unix is better than Windows" I'm not saying it's preferable for people who just want to get along by going along - I'm saying it's objectively better technology.
So here's the challenge: define a metric not directly or indirectly a consequence of market share which, when fairly applied to both Windows and Unix, shows Windows to be better.
I'll bet nobody can do it - and remember, please: Linux, MacOS X, Solaris, and OpenBSD - are all examples of Unix.
I couldn't do this for Windows - but maybe I'm just biased, so let me offer some notes on some of the metrics I'd use to make the case for Unix over Windows:
Basically, pick any user process - and you'll find that MacOS X wins with the KDE/Linux combination either tying Windows or beating it.
Notice that the availability of device drivers for XP says nothing about XP - it's purely a market share artifact. Basically, if you want to cite device drivers you'll have to argue that they're easier to write or more reliable under Windows than say, Linux - and, you know, good luck with that.
Want to compare Windows Server 2008 to Solaris? or BSD, or Linux? Really? You're kidding, right?
Windows licenses cost money - Solaris, Linux, and most BSDs are free.
As far as I know Windows holds no non-clustered performance records for which Unix companies have filed results.
Thousands of functionally identical VBA scripts do not diversity make; so go ahead: make my day - name a real software innovation that wasn't available on Unix before being released on Windows.
On Windows, ya, right.
The PC vision of security does not, of course, even apply; but go beyond that to the real thing and you'll see why organizations like NSA build on Linux, Solaris, and OpenBSD - not Windows.
So, start your metrics: "Unix is better than Windows", minority opinion or plain fact?