Tired of fanboys telling you how many millions of users run the OS you know perfectly well is lucky to have a few hundred thousand users at best (or often only a few hundred or thousand in terms of the more niche linux distributions)?

Get the real stats from gs.statcounter.com.

You can adjust the time frames, stat types (os, browser, and so on) and find the sad truth that in fact, Linux market share has if anything dropped over the past year...

By the way, bridgeman of Linux ATI programming group at phoronix also once confirmed the exact 0.8% number, ATI had seen that as well. What many fan boys don't realize re web stats, once you get above a certain number, the law of averages kicks in, and unless you are comparing huge areas, everything settles to what the stuff actually is.

One nice thing about that site though is you can compare regions, and there are differences between regions.

Here's a sample from Feb 28 2009 to Feb 28 2010

I can by the way confirm these stats from other sites I've had full log access, that's large scale, non geek, standard user, sites. 0.8 to 0.7% is exactly in line with the numbers I saw, and also by the way exactly in line with the real stats I've seen from my own work doing Free Software for Linux.

So if you don't like these numbers, here's what to do about it: start yelling at the kernel developers and make them stop breaking the kernel apis every 6 months, and get someone in charge of that project who has the technical skill and vision to create long term stable APIs that won't break people's drivers and hardware (free or non-free, doesn't matter).

Then get to the Gnome/KDE guys and convince them to create stable desktop apis. Problem is, to do that, takes skill, commitment, and probably what is most lacking, leadership to force people to not just break things whenever they feel like.

So again, if you don't like these stats, you can complain and whine and say it's not true or that MS is evil, or you can start applying real pressure on the core geek groups to stop their constant breaks of the core subsystems, and to accept the challenge of creating long term stable apis.

I know they are harder to make, that's given, they are. You need to be a good planner, a good long term architect, and you need to be able to think through problems in advance. Then you need to be able to support it.

So let's see if free software can achieve this goal. My guess is it can't, and will remain fringe except on embedded devices and servers, embedded it largely doesn't matter what kernel runs it (and remember, what android is is exactly what I'm talking about: a controlled, api set, desktop application programming interface, Java set over a linux kernel).

Personally I don't think any of the core linux hackers are good enough to do this, sorry to say.

Oh, and by the way, by long term I don't mean the asberger's style 12 to 18 months, with 3 years being an eternity, I mean something in line with what most real world users want and expect, 5 years or more.

A stable desktop API, for example, means that you can write a NEW program using those older apis, and run it on a fresh install of that desktop without having to totally rewrite it. A stable kernel API means you can write a driver, free or non free, and have that driver work for the people using that hardware through kernel after kernel. Don't get me wrong, this is hard to achieve. It takes skill. A lot of skill. Anyone can hack out a quick fix that has nothing behind it, which breaks things, again, in the name of 'improving matters', that's trivial and simple.
Back to top