The problem he's missing is simple math. We're a small startup making apps for social networks, and about 7% of our customers use IE6. (I keep meaning to check by revenue, I bet it's less then, since presumably a lot of them are using it at work where they can't just whip out their credit card.) Unfortunately, getting our great UI to work in IE6 takes up some percentage of our time. As time goes on, the percentage of users drops and the percentage of our time increases. I'm not sure if they've crossed paths yet, but if not, it's close.

Sure, if you have unlimited resources, develop for IE6, Opera, whatever else. But when you're on a budget, and you have to choose between making your app 20% better for 93% of your users, or making it work for the extra 7%, it's not a no-brainer.