|Think about Loose Coupling|
IE is broken period.
IE works just fine here already. Period.
What if thousands of users keep clicking on a section of a page they think is a link and it turns out not to be?
You must be thinking about those sites that take away the one consistent part of Web user interfaces (to replace it with something of their own imagining): Links are underlined. That's not a mistake we made here. Well, except for the two top-line buttons that some people went to quite a lot of effort to eliminate every single visual clue that they actually are buttons. I think that was a mistake. But the lack of visual clues does make it look very nice and uniform (and misleadingly non-functional).
Coding to the standards and adding hacks to get IE working is the best practice for dealing with this kind of problem.
Yes, I'm aware of lots of things that are widely practiced in "modern" web design. Unfortunately, a lot of the reason I am aware of a lot of them is that I have to deal with how f'ed up they tend to make things. Well, not at PerlMonks, thankfully.
I've repeatedly solicited actual good examples of CSS for sizing/positioning. The closest I've ever come was one example that really sucked on one of my browsers to which the submitter asserted that it was due to bugs in my browser. Yeah, cutting edge stuff is often buggy. That doesn't mean that "doesn't work" no longer matters.
So maybe you'll be the first to show me a CSS solution that actually works.
But, no, I'm not convinced that "widely accepted best practice" actually means "good idea" when I constantly see it failing without even trying to look. And when I try to look, it was always easy to find failure.
I consider myself outside the asylum of modern web design. If you actually come up with something that actually works very well, then I'll be overjoyed but very surprised. But all I care about is how well it works and at what cost.
Coding to the standards [....]More progress is being made.
There are plenty of standards to choose from. When you are done with "progress" enough that your newer standards actually "work", then I think it would be a fine time to adopt them. When it comes to choosing standards, I much prefer "it works" over "it is new". Especially when "doesn't work on" includes some of the most prevalent browsers. New standards are nice and can be important in setting direction, etc. And that includes early adopters finding and working out kinks. PerlMonks is not in the business of early adopting new web standards and working out kinks and worrying about market share of versions of browsers.
I'm sure that annoys several people. But it is also appreciated by quite a few people. And it is actually working pretty well, IMO.