There is a lot of talk around the interweb these days about IE6 and it's impending doom, as worldwide stats show it's decrease in usage. The trouble with the various statistics though is, which ones are accurate? What can be believed?
Browser Usage differences seen
I have seen Browser Usage Statistics from various "reliable" sources (I don't know how to tell the difference between reliable and unreliable, but pretty sure Dilbert's rubbish collector was one of the sources) that show IE6 usage as low as 1.78% worldwide amongst Internet users, and as high as 7.7%.
Both of these figures were for the month of December, 2011.
Why is there such a disparity between what appears to be a Microsoft driven site, IE6 Countdown (which uses Net Applications as it's source, showing 7.7% usage) and Stat Counter, showing 1.78% IE6 usage for the month of December?
I'm not trying to pick on IE6 in particular, but it is the one common denominator between both sites. A 6% difference between the two sites and the statistics gathered is a significant number of users. Which is the more accurate? How do these stats even get calculated?
I guess the two statistic gathering agencies use vastly different algorithms, methods, wet finger sticking in the breeze ways of coming up with these figures, and an interesting article by Sitepoint's Craig Buckler on How Browser Market Share is Calculated points to various factors (though it should be pointed out that Craig admits this is his own theoretical overview) such as time spent in one browser vs another, in the case of someone using multiple browsers.
I myself use Opera, Chrome, Safari, Firefox and Internet Explorer, in probably in about that order of popularity. If the study was just on my web usage alone, Opera would be at 85+% of usage.
Gathering such statistics over a worldwide section of users, across many, many web sites would not be a job that interests me at all, and I know much smarter people than me are working these figures out. But I still come back to how, and why, is there such a large discrepancy between two reputable stat counters?
I know that getting a hard and fast figure would be nigh on impossible, with the multitude of ways different browsers read and translate HTML for our reading or viewing pleasure (or listening pleasure, for the screen reader users) but when trying to decide whether to continue to support IE6, or maintain sites that are read well (without being exactly how I intended) then statistics such as these are all I have to go on (apart from other analytics gathered, such as Google).
It will be interesting, to me at least, how close these figures come together should Microsoft implement automatic upgrading of Internet Explorer, as mooted to happen early in 2012. There are various rules around this auto upgrade, and for many users it will mean only being upgraded to IE8 (Windows XP cannot run IE9). For many users, it will still likely be no upgrade at all.