I think of email as a schema-less, dynamically-typed system for communication. Anyone can send an email to anyone, and there's only one "type" for email (a long string of blah)... but most emails are sent to too many people, and often go unread. IT works, but it's a huge mess and it's unreliable. Also, the time behavior of email isn't right for all purposes. Some emails are generally useful 2 years later (and if you think the content is likely to be useful in 2 years, you probably shouldn't use email). The vast majority are not. Inbox search solves the time problem, but it feels "bolted on". Inbox search isn't an intrinsic fix but a late patch; it's using full-text search (which usually works, but not always, because you might not remember any key phrases) because there's no better solution right now.
So one thought is to consider something analogous to static typing, where emails have types (we'd need a different word, because it's not exactly the same concept) and these message types restrict (i.e. you might create a type T for emails sent by user U under 500 characters). Unfortunately, I can't see a way to design a system like this without making it massively complicated. The problem here is that people are already overloaded with email and adding any piece of additional complexity is going to piss a lot of people off. Google added typed edges with Circles in Google+-- clearly a good feature-- but they took a lot of flak for G+ being "too complicated" for "mere mortals" to understand because of Circles. When people are already fatigued and overloaded and starting to get annoyed with the "tragedy of the commons" (all of this being with email, and with social networks) even small bits of additional complexity turn people off.
I think the final answer is: let email be. The problem isn't email itself. There's a place for schemaless, dynamically-typed communication. It (or some form of it) will always be useful and some part of how we communicate. The problem with email is that people use it for too many things and in too many sloppy ways, so the solution might be to identify communication patterns (for example, persistent storage of knowledge) where email is not the best solution and focus on those.
For some ideas, consider one problem with forums and social networks in general: time behavior. Lots of content is generated, but quickly becomes useless. How often do you read a 5-month-old thread on a message board or social network? Occasionally, but not often. This imposes siloization by time, so the value of the content can never grow faster than linearly as a function of the amount of it. A lot of interesting discussions just won't happen if they can't occur asynchronously. They won't get the critical mass.
One forum that I think deserves a lot of props for solving the time problem is Quora. Quora has managed to structure itself in such a way that content generated 12 months ago is still extremely valuable. It doesn't fall into obscurity or become useless because it's "old", and it's pretty common that a 6-month-old answer will get a new comment actually worth reading. Quora is also great in terms of how it handles new threads. On a typical message board, an OP that doesn't get immediate response is just a failed thread. On Quora, it's an open question. A good way to set it up, because not every OP that doesn't get immediate attention deserves to die.
So one thought is to consider something analogous to static typing, where emails have types (we'd need a different word, because it's not exactly the same concept) and these message types restrict (i.e. you might create a type T for emails sent by user U under 500 characters). Unfortunately, I can't see a way to design a system like this without making it massively complicated. The problem here is that people are already overloaded with email and adding any piece of additional complexity is going to piss a lot of people off. Google added typed edges with Circles in Google+-- clearly a good feature-- but they took a lot of flak for G+ being "too complicated" for "mere mortals" to understand because of Circles. When people are already fatigued and overloaded and starting to get annoyed with the "tragedy of the commons" (all of this being with email, and with social networks) even small bits of additional complexity turn people off.
I think the final answer is: let email be. The problem isn't email itself. There's a place for schemaless, dynamically-typed communication. It (or some form of it) will always be useful and some part of how we communicate. The problem with email is that people use it for too many things and in too many sloppy ways, so the solution might be to identify communication patterns (for example, persistent storage of knowledge) where email is not the best solution and focus on those.
For some ideas, consider one problem with forums and social networks in general: time behavior. Lots of content is generated, but quickly becomes useless. How often do you read a 5-month-old thread on a message board or social network? Occasionally, but not often. This imposes siloization by time, so the value of the content can never grow faster than linearly as a function of the amount of it. A lot of interesting discussions just won't happen if they can't occur asynchronously. They won't get the critical mass.
One forum that I think deserves a lot of props for solving the time problem is Quora. Quora has managed to structure itself in such a way that content generated 12 months ago is still extremely valuable. It doesn't fall into obscurity or become useless because it's "old", and it's pretty common that a 6-month-old answer will get a new comment actually worth reading. Quora is also great in terms of how it handles new threads. On a typical message board, an OP that doesn't get immediate response is just a failed thread. On Quora, it's an open question. A good way to set it up, because not every OP that doesn't get immediate attention deserves to die.