The easiest thing to look at is all the pain that people have with email as it exists today. There is the core of email, but then it is festooned with many additions, extensions, associates, and helpers. All of those things are there for a reason, because they solve a real problem with email. But, of course, those things don't have the luxury of rebuilding the whole system from the ground up, and doing so could lead to a design that doesn't have the pain points and weaknesses that need to be mended by additions. Consider a few common aspects: at a big company the MO is typically for individuals to join a large number of distribution groups and then also set up a lot of filters to move the mail they get for those groups into different folders because they have different levels of actionability (if any). Similarly, automated emails tend to require special filtering rules to make sure they don't drown out the S/N. Add to that the common difficulty of personally or publicly archiving email, because so often it tends to be used as an information storage system for key bits of data (contacts, decisions, technical details).

Personally I think the "todo list" model is good but lacking. I think ultimately what you want is something that would look to us like a mashup of issue tracking, messaging, wiki, and maybe even twitter from a certain perspective. I think switching from "private default" to "public default" for messaging is the biggest toggle, with the second one being pull vs. push (or search vs. signup).