Counting Linux Users

Did I mention that I've been "on" the Internet since 1988? Because I have: that's when I went to work at Draper Laboratory, in Cambridge MA. Eventually, after bouncing around a bit, I wound up in the network support group where we handled all the networking issues. And at Draper, unlike most organizations at the time, "network" meant "IP network", as in, "Internet Protocol" network.

Why do I mention this? Only because whenever Linux makes news, it makes deja vu for me. This time, the article at Slashdot, Attempts to Count Linux Users Remain Pointless, reminded me of the mid-90s when everyone with something to hype about the Internet and web would go on endlessly about the seemingly endless annual doubling (or faster) rate of Internet growth.

The original article cited at Slashdot, Can Linux Adoption Ever be Accurately Gauged?, is mostly about why it's so tough to figure out the degree to which Linux is penetrating the "market". Most of the Slashdot discussion is about whether or not it's even worthwhile to count Linux users, and if so, how to actually do it with some degree of precision.

I'm a big believer in back-of-the-envelope calculation, and I also believe that, although it may be difficult, it is possible to estimate Linux market share. I'm interested in hearing your opinions on this matter--but I'm also going to get to work on an article about how you'd go about generating this information, as well as the results I get.