(final edit: okay, I posted this, had second thoughts about the prudence of posting it and hid it, but it seems like everyone saw it anyway, so what the hell, I'll unhide it again. It's maybe not the most diplomatic thing to post, but...erf.)
Very important disclaimer: this post is utterly and entirely my own personal thinking and in no way whatsoever reflective or representative of my employer.
Quick foreword, edited in later - I forgot to explain the reason for this post. When people talk about Ubuntu's shortcomings they often hedge it around by saying 'but hey, they're good at marketing', or 'hey, they make a successful desktop'. I'm arguing that this isn't so clear cut.
I'm going to try and keep this as succint as I can, as this kind of thinking has gotten me in Big Trouble before, but commenting on Greg's blog post got me thinking. Has Ubuntu, really, been successful?
Becoming the 'leading' Linux desktop distribution is a pretty paltry measure of success, after all. As I wrote on Greg's post, no-one's really taken SUSE seriously since Novell bought them (though that seems rather unfair, and probably is a problem of perception / marketing - from all I hear, it's still a very solid distro), Mandriva has been basically bankrupt since 2003 and has huge management problems (also, can't string together a press release in correct English, which makes appearing more professional than Mandriva a freaking low bar. Don't get me wrong, I love Mandriva, but the company has huge problems. I never said it didn't when I worked there. I just didn't talk about them. Publicly. :>), and Fedora and Debian aren't really Uncle Bob-type distributions. All the rest are derivatives of the others, or second-stringers. If you show up with a couple of graphic designers, anyone who's passed Media Relations 101, and a bit of cash, you can pretty much win by default, which is what Ubuntu did.
So, let's not go on the kiddy pool stuff. After all, that's not how Ubuntu says it's measuring success. Remember Bug #1, filed in August 2004? It doesn't say 'Mandrake has the biggest share of the Linux market', it says 'Microsoft has a majority market share'. So, let's look at Ubuntu's record in those terms. Let's look at the W3Schools operating system share statistics.
In July and September 2004 (so presumably also in August), Linux is at 3.1%.
In June 2010, after nearly six years of Ubuntu as the generally-perceived Linux desktop standard bearer, Linux is at...4.8%.
In March 2003, Linux was at 2.2%. So that's a rate of growth of 0.9% over 16 months to July 2004 - 0.05625 percentage points per month. The rate of growth from July 2004 to June 2010 is 1.7% over 71 months - 0.02394 percentage points per month. The margin of error in those numbers is likely huge, because we're playing with such small numbers, but even so, it sure doesn't look like Ubuntu has even managed to increase the rate of growth of Linux one iota over the 'leading desktop distributions' that preceded it (in the 2003-2004 range that was probably Mandriva; before there was Gentoo and Red Hat Linux, and SUSE was always there or thereabouts).
It's hard to find stats from the other places that track operating system usage that go back as far, but going back as far as they do - to around 2007 or so, usually - they seem to tell much the same story. I can't find any which show really significant growth in general Linux adoption, or a significant increase of the rate of growth at any point in Ubuntu's tenure.
So, is this success? Really? Discuss. :)
Given what's discussed in this post, I'd say the top 3 damage-dealers to Microsoft since Canonical started up have been Mozilla, Google, and Microsoft itself (hello, Vista). Canonical probably ain't in the top 10.
(To pre-empt the obvious point - yes, I was working for Mandriva and Red Hat during most of the period under discussion, and we haven't revolutionized the world in terms of Linux market share either. But Mandriva never exactly aimed to; it aimed to be a successful commercial Linux distributor, and was always more about effectively serving a relatively small market than trying to Take Over The World. I'd argue Mandriva's usually provided an excellent product to its mostly-loyal user base, and almost all the company's problems, which are significant, are in management and marketing. Red Hat, also, has never said we're aiming to make significant inroads into general user desktop market share. This is not the same as 'we don't care about the desktop', note, but the emphasis is slightly different. We maintain a great desktop distribution, which I work on, and do a large amount of work on the F/OSS desktop - see Greg's post - but not in exactly the way Ubuntu claims to be aiming for.)