In the past I've defended Microsoft's operating systems, particularly NT, as
being a lot more stable than people generally credit them.

I wish to confess some exceptions to this defense.  I've got some IBM
ThinkPads that are definitely not running well under NT.  Several have
exhibited severe problems running Office 97 apps, and have had to have '97
reinstalled.  Several have just been plain weird.  One can't see it's modem,
and we've replaced the modem and motherboard and reinstalled the software
twice.

And I have one here on my desk that fails to boot about
one-out-of-five-times.  It looks like the "Battery Meter" is colliding with
other items in the startup sequence, and causing the whole interface to
hang.

And today I lost three days of work, when a system that I'd loaded a ghost
image to booted to the BSOD.  I went back to the source system, and it too
gave me the blue, so it's back to square one: format, install, prototype,
test, image, copy.

Now I've got to say that when it comes to the ThinkPads, I suspect that the
problem is to be found in a combination of hardware and driver faults, and
perhaps some poor software.  And the BSOD is probably the result of me
trying to guess my way around a registry change that I should have used the
Policy Editor for.  Others who have analyzed the alleged instabilities in MS
software have often concluded that MS is not the weak link (We're talking
only Win95 and NT4 here).

I once read an opinion that said that if it weren't for our expectation that
MS software would crash, we would have realized that a lot of the PC
hardware we buy is absolute CRAP, and we would never have tolerated such
crap in any other guise.  Misdirecting the blame to Microsoft has saved the
keyster of many a shoddy manufacturer.

But it happens.  They crash.

And we do our best to catch and cure the crash before deployment, and
therefore like plumbers we are appreciated mostly in our neglect, and
vilified in our moments of glory.

But hey, we get the cool toys first...