Almost everybody trying to do standards compliant stuff feels the annoyance of IE's old versions out there hurting any attempt at making something of the new features we could use were it not for "that" browser and its former dominance.
Yet with some efforts of the industry as a whole IE6, IE7 are clearly slowing down.
I quit testing IE6 when my virtual image updated itself unasked for to IE8 (it seems it only had the IE7 blocker on it and someday it decided IE8 would be better for me - not).
I check stat of browsers regularly and see a declining trend. I'm quite aware some of my audiences are different in demographics (work-home young-old nerd-technophobic.)
I'm also very aware that how you measure (and deal with bots in there) makes a big difference.
So on my biggest site (which still works quite well in IE6 and IE7), which attracts a bit an older crowd, mostly for leisure, and are not specifically technical in nature, as measured by google analytics:
- IE: 33.55 % of the visits are:
----- IE 6: 0.83 % = 0.28% overall
----- IE 7: 7.18 % = 2.41% overall
----- IE 8: 38.98 % = 13,08% overall
----- IE 9: 52.84 % = 17,73% overall
----- IE10: 0.15 % = 0,05% overall
Which clearly makes IE6 a non-issue on this site (less than 0.3% of visits overall). IE7 at a total of 2.41% overall is still quite a few users, but I might well let them go on the next update. Esp. since IE7 really is retarded and causing me a lot of stress.
That just over 13% in total share of IE8 is the worst of it all: it'll block much for a long time to come (at least some of those users need to be weened off of IE and/or XP) - and it has plenty bugs that annoy me a lot.
I think for established sites we know the effort obsolete version of IE cause us , and we can predict the loss if we stop supporting one of them.
For new sites it's all much less easy to predict. I recently did a few pages on a new site and I had them made in polyglot xhtml5, CSS3 features, even using svg graphics as stretching backgrounds etc. I had anticipated some fall-back, and I was surprised at not having to do anything for any version of IE I tested (did not test IE6, for me that one is dead and buried) - they used the fallback I already had in the the CSS.
What continues to rise is safari on iOS: serious second place behind IE on windows, I think I'll put some more effort in the iPad.
So the question then:
- your demographics ?
- how are your statistics ?
- which IE versions do you test for ?