I have always claimed that the widely used Operating System market share statistics from Net applications were not really accurate when it came to MacOS and Linux market share. In my opinion there were two factors that prevented an accurate worldwide market share to be produced: Linux browsers potentially ignoring the counter and improper geographical distribution. The second problem has been fixed and it does impact the market share numbers significantly.
The problem was that a large portion of the website visitors that are counted come from the USA and other English speaking countries. This means that worldwide data was more representative of the USA than of the rest of the world. The problem is that the OS market share is currently very different in the USA than the rest of the world: the Mac OSX market share is much higher, but the Windows and Linux market share are lower. Now that the data has been adjusted MacOSX market share has fallen from 10% to around 5%, and Linux market share has increased slightly to 1.05% (with a peak at 1.17% in may). This is much more consistent with the data provided by other firms such as XiTi Monitor which recently placed the Linux market share at 1.2% and MacOSX at 4.6%
Astute readers will notice that the Linux market share numbers are still much lower than the ones provided by W3Counter which place Linux around the 2% mark. The explanation advanced by some would be that it is in fact Firefox's market share that is not correct. It is feared that some Firefox plug-ins like adblock and noscript would prevent a visitor from being counted. Since these are widely used by Firefox and Linux users their market share would be underestimated, and this would account for the difference. Personally I would trust the numbers from W3counter the most, but if you don't know what to do you can always try to count the Linux market share by Twibbons.
Read more in the Linux category
2 Responses to "Net Applications Changes Methodology: Windows & Linux Market Share Rises":
I've always found the stats available at seti@home (BOINC)
http://boincstats.com/stats/host_os_stats.php?pr=sah&st=0
and Folding@home (Stanford)
http://fah-web.stanford.edu/cgi-bin/main.py?qtype=osstats
to be "interesting", though it's hard to say that the figures are truely representative than, say, W3Counter.
The overall BOINC stats are here
http://boincstats.com/stats/host_os_stats.php?pr=bo&st=0
And checking out the OS share stats for various other BOINC projects gives some idea how much variability can creep into the results from various unstated biases (the most obvious being whether the project has clients, or easily installed clients, available for each relevant operating system.
I suspect that Seti@home is the most "neutral" (certainly the best known and most accessible to a "typical user), but I couldn't prove it.
Bernard Swiss
And Microsoft estimates Linux Desktop usage to be the same as MacOSx usage around 10 to 15%.
I have a sneaking suspicion the Microsoft estimate is the more accurate.
Post a Comment