I agree that there are tons of improvements to be made to effective e-mail use in organizations. But many of the behaviors we exhibit are completely natural to us and difficult to change. For example, we check our e-mail on what psychologists refer to as a variable interval reinforcement schedule; we know that at approximately some interval we get an e-mail that is helpful or enjoyable to us, and we waste our time checking for that e-mail.
It's hard to change the e-mail we receive on a global scale in one instant. It's quite a lot easier to develop tools that change how the e-mail we receive is displayed to us. One such concept was the Gmail priority inbox: what if you only got notifications for the e-mail that was important and could buffer up the rest for later?
A complete theory of how e-mail could work might easily involve a hybrid approach. Why aren't people suggesting solutions that could work instead of debating approaches? Does it have to be one way or the other?