I love the "cost of porting to other platforms" argument. I'd like to replace it with the "cost of delivering mushy web applications to customers who expect more". That cost is infinite, and you'll never get it back.