I don't know, I see people beginning to use linux in my social circle, even non-technical users. Would have been unthinkable a couple of years ago. Nothing scientific here but I get the definite impression linux is gaining momentum.

MacOSX user here btw, not a desktop linux zealot at all, just reporting what I see anecdotally.

The best thing MS could do for linux is perfect its anti-piracy strategy. If it was impossible to pirate Windows, Linux would explode I reckon.

I guess MS knows this. They use (tolerated) piracy as one tool to gain and secure market share.

You're both over-estimating the amount people really choose their OS vs use what has shipped with every computer they've ever owned.

Is the worldwide deployment of Vista comparable to the aggregate of the top 10 desktop Linux distributions?

I don't know those numbers, but I think that's the important question (rather than just Windows vs. Ubuntu).

If you accept browser usage as a proxy for desktop deployments, you can find statistics pretty easily. http://www.w3counter.com/globalstats.php and http://www.w3schools.com/browsers/browsers_os.asp both show similar spreads between Linux, OS X, and Vista (eg, 2:5:15 and 4:6:17, as percent of total).

So web usage is comparable, but Vista seems to be well ahead. At the same time I suspect that deployments are quite a bit different than web usage, so take that with a grain of salt.

I would imagine that the total worldwide number of desktop Linux installations is statistically insignificant. Vista has never really taken hold, and XP is still the dominant OS by some way, but I've never seen a study that put desktop Linux even on the same order of magnitude as OS X.

If you look at the links posted above, the w3schools.com numbers, 4% and 6%, are not even a binary order of magnitude apart, never mind the usually meant decimal (10x spread).

The w3counter.com numbers are 2% and 5%. Both single digits (hint).

There are whole companies out there doing engineering work on Linux, except for a few odd laptops. Actually, site http proxies may cause Unix web undercount (i.e. stats are dominated by home access, presumably with lower Unix use).

Do you seriously believe that the visitors to a site like w3schools.com are even close to representative of the population as a whole?

There are indeed whole companies out there doing engineering work on Linux; I've worked with some of them. But they are a tiny fraction of all companies, and the desktops in most companies run Windows. Of those that don't, I'm sure Linux has had a few big successes, but Apple sells all those computers to someone. As I said, I've never seen a study that puts Linux even close. If you've got more up-to-date information that is actually representative, go ahead and share it, but please don't pretend that the server log from a random web site is a real study.