At this point, it matters not what happens to IE market share. It will not ever get lower than 20% and it will not ever get higher than 90% again. As long as it stays in that range we can know with surety that a) all web sites will still have to be coded to work with IE
b) no web sites will be exclusively IE. Those days are over.
The only remaining question is how fast we can vanquish IE6 and thus rescue a small amount of sanity for web developers everywhere. The rest of the battle, in as much as it can be won, has been won.
Allow your site to gracefully degrade to a more simple version (for both mobile users and IE6) or just reject IE6 users altogether.
I plan to drop IE6 support soon. From my experience, if someone loves your site, and you make them choose between upgrading their browser and using your site, about 90% will happily upgrade their browser (most of the remaining 10% can't upgrade, but you can't please/support everyone). I know from experience; I have forced this decision on my users several years ago for mac users (dropped support for IE5/mac and older versions of Safari) and had no complaints.
Indeed, this is very much what I already do. For all non-functional aspects, IE6 is not catered for. For example, transparent pngs are assumed supported. If you visit in IE6 you will see lots of blocky icons and images for rounded corners and such. Layout is roughly right, but no effort is made in getting it pixel perfect.
Nearly any site using a web analytics package will collect browser statistics, even to the version number. Even better if you aggregate the statistics from a number of sites with different user demographics. For example, just the Apple site or the Mozilla site would give you skewed statistics. Most analytics packages will find unique users over a specified time period. Given that you use all your browsers at the participating sites, you would be counted as a unique user for all 5 browsers, distorting the statics somewhere off in the 6th or 7th decimal place.
Most of the estimation is done by log analysis, hitlinks.com is one such party, they have a panel of several 10's of thousands of sites that they get browser information for.
If you have a large enough portfolio of sites then you can do your own analysis.
It doesn't look like you are a 'typical' user, who will use a single browser only unless they run into some kind of compatibility problem.
Do you use the different browsers to test functionality or because different sites you visit function better with each browser? Just curious, as I pretty much only use FF3, but I will test my sites on FF3, Safari, Chrome, Opera, IE6 and IE7...in which case I would assume they wouldn't get counted as I'm doing internal testing.
The data in this article doesn't really tell me much. The decline is slow and is based solely on current versions. There are way too many things to factor. The increase of Firefox users, the uprise of Opera and Chrome. Then when IE6(or was it 7) came out, the different layout had a learning curve that caused users to switch. Then IE8 is supposedly the windows 7 of IE. There are really too many factors to just look at a chart with no significant data and jump to conclusions.
Until corporations upgrade from IE to something else en masse (unlikely in the near future), IE's share will remain disproportionate. I think the vast majority of people still using IE 6 are doing so because they have no choice in the matter.
This can't happen soon enough. I'm working on a web app right now that requires ie6 functionality. This browser needs to die but I think it's still got a ways to go.
Usually I'll finish an App, continue testing with FF, then cross my fingers and load in IE. Then I fix the carnage over several hours/days and test then revert back to Firefox to make sure nothing has broken. All the other browsers usually just take a couple hours of tweaks to cater to.
That browser has been a thorn in my side for too many years now.