OSNews

I've had my share of problems with Not "you are pathetic cultists" Parker. But I think .4% is probably a fairly good starting point for usage, at least in the US. Probably Europe is higher than that. I don't know about you, but I don't know anyone who uses Linux except other techies like me, and we are not a huge group compared to everyone else. Counting hits on a website isn't a great technique, but it is at least a starting point. Maybe archisteel can enlighten me as to why he thinks there must be a lot more Linux users - I'd guess around 1% myself.

I also use 2003 on my workstation at work.

Notwithstanding the fact that NotParker very likely cherry-picked the webstat sites to only show the lowest scores for Linux (after all, he's done that with the Server Growth figures from Gartner), there are many reasons that make webstats an inaccurate tool for establishing market share:

a) Actual web site contents: this one is obvious. Certain web sites will appeal more to certain demographics, where OS distribution may be skewed. Language and local can also skew the results dramatically. In the case of the examples given by NotParker, these certainly overrepresent English-speaking countries.

b) The results are not independently verified and can easily be falsified or even fabricated. "Page hit" counters can also be easily rigged with a few simple scripts.

c) Most Linux users I know use User Agents to masquerade their browser as something else; in the case of Konqueror, the masqueraded browser most often does not mention Linux at all, but rather Windows or some other OS.

d) I think we can all agree that "page hit" counters are *completely* useless to figure out market share, and that the only real way to make web stats more accurate is to use cookies (which not everyone accepts) or record IP addresses. The problem with IP addresses is that they over-represent narrowband (i.e. dial-up) users, because these users routinely obtain a new IP address everytime they login, whereas broadband users usually keep the same IP for long periods of time. Therefore, narrowband users are more likely to be counted more than once, and therefore will be over-represented. This wouldn't be an issue if narrowband users were evenly distributed among Linux and Windows users, but in fact very few Linux users use modems (for two reasons: they are usually more tech-savvy, and winmodems are notoriously hard to run under Linux).

I do agree that Linux usage is higher in Europe and Asia than it is in the US, however. But that's besides the point. The point is that, for all the reasons outlined above, it is *clear* that webstats cannot give an accurate figure, and therefore using them to further an anti-Linux (or even a pro-Linux) agenda is nothing more than propaganda.

That's only part of the problem, of course. The other part is that NotParker brings this up *every* chance he gets, no matter how off-topic it may be (what the heck does Linux market share have to do in an article about Microsoft's "Flash killer"?)