To the non-programmers in my flist, it may not be obvious why this matters -- after all, IE6 works okay, right? The problem is, while it looks okay to the end user, it is *horrible* to the programmers. It's wildly incompatible with the rest of the Web, in a myriad of subtle ways. Seriously: making a sophisticated webpage work on IE6 can take as long as writing it in the first place. (There's a standard joke in the business that you spend 50% of all web-development time wishing a horrible death on Bill Gates for inflicting this thing on us; it wouldn't be funny if it wasn't so true.)
Historically, though, the pain of IE6 couldn't be avoided: too many people are using it, so you have to support it if you're going to have commercial success. It's okay for the end users, who get the support, and it's okay for Microsoft -- since the programmers are supporting their wretched code, Microsoft doesn't have any real problem. But for the programmers in the middle, it's a miserable experience.
I have to think about this pretty seriously for CommYou. I'm almost up to that point in the system, where I need to go waste a couple of weeks making the thing work on IE6 while *not* breaking it for everybody else. I'm tempted to simply not. Facebook already discourages you from using IE6 (because little bits and pieces of FB break if you use it), so my user base is already being pushed in the direction of either upgrading to the not-quite-as-bad IE7 or switching to a better browser like Firefox or the suddenly-more-relevant Safari. (Or, for the true geekerati, Opera.) Spending those weeks on actually *deploying* CommYou would be much more productive.
So here's a question (I'm not going to bother formalizing it into a poll): how many of you are using IE6 regularly, and actually care about that enough that it would be a problem for you to upgrade? I'm guessing that most users of IE6 are, at this point, just using it out of inertia more than anything else. Given that it's finally down to a minority of Web users (albeit a fairly large minority -- still 30+%), I'm thinking of just putting in a SaveTheDevelopers banner and instead working on making the product better...
ETA: Good food for thought -- thanks to everyone. It sounds like the main argument for IE6 is that a moderate number of people are forced to use it at work. That may well be enough to convince me to provide some minimal support -- enough to make it functional -- but not worry about making it work as *well* as with the more modern browsers. We'll see, but thanks for all the input!