How Many People Really Use Linux -- And Stick With It?

I am growing infernally curious about what the end-of-the-year sales figures for Dell's Ubuntu machines will be. Not just how many bought those machines or in what proportion to Windows users, but how their long-term experiences shape up against others (as well as whether or not they elected to buy support). What if Linux has its big day in the sun, and simply doesn't achieve more than a small percentage of the market?Is that necessarily a bad thing, or just a reflection of what percentage of people want something like Linux? And how many Linux users are there out there? I've read about a number of different ways to tabulate this sort of thing, and they all suffer from different methodological issues; Web server statistics on visitors, for instance, aren't very reliable because the client can claim it's any kind of OS or browser it wants.

With Windows, it's at least possible to guess based on the number of licenses sold, or the number of activated machines, etc. With Linux, it's far harder to guess because of things like biased self-reporting, redundancy, and the lack of a really reliable set of metrics. If I have a Linux box and a Windows box, does that count in one, the other, or both, or neither? And if I download ten distributions but don't actually use any one of them in a fulltime production system, does that count?

This is why the Dell / Ubuntu sales figures are going to be immensely interesting, because they'll provide at least some indication of how things are shaping up for Linux -- at least within Dell's ecosystem. It's probably not going to be a reliable guide as to how Linux is faring overall on the desktop, since that's not something you could gauge from any one vendor's sales -- Dell's sales probably do not extrapolate smoothly into desktop sales in general. It's also not going to tell us how many people obtain such machines but don't actually run Linux on it in the long run -- some folks might buy such a box to get a taste of Linux, then wipe it and install Windows (or even BSD) after a few months.

That's something else that is going to be tough to measure, even for a company like Dell: Long-term commitment to Linux, especially by people who are not interested one way or another in its technical merits. I've received many comments from people who have picked up Ubuntu or another distribution and sworn off ever going back to Windows. That's great, but I always wonder: Will they still say the same thing in a year? I don't ask these questions to be snide -- I just think that if Linux wants to work well on the desktop (notice I didn't say "replace Windows," just "work well"), there needs to be at least some attention paid to why a user might defect, or how long they stick around before doing so (and what pushes them off).