Accurate browser numbers are always hard to come by, but for argument's sake, let's assume:
IE: 40%, Mobile: 10%, Other modern desktop browsers: 50%
(Numbers based on Wikipedia's browser market share page)
Current thinking has us building a website that works for all of those browsers, and then building an optimized version that works better for the 10% of mobile users. I've visited many sites on my mobile devices that are complete rewrites, using browser-specific features, of the regular website, for 10% of the users.
Yet, we completely refuse to use features found in 50% of the browsers because Microsoft hasn't implemented them yet. This is where the math falls apart. If resources are limited to two implementations, wouldn't it make more sense to provide a better experience for 50% of the users, and still provide the basic version for Microsoft/mobile users?
If 10% of users are enough to warrant a completely different website version as has been demonstrated with the rise of mobile users, then why not build three variants? One mobile, one old web technology version, and then one using bleeding edge technologies? What is it about Microsoft's latency in feature implementations that you speak to that prevents the latter version from even being considered?