Right now I'm thinking that at least for read messages, archives of revisions would need to be kept. If a message was never read I'm not sure it's necessary to keep a revision of it though.
When I'm checking my mail on my mobile phone, I'm sure as hell do NOT want to make extra network roundtrips just to be able to see my mail.
It's also far harder to index, sort and search in the proposed scenario: while I can quickly file an email under any label or folder I choose, and be sure that it will stay there, have the same contents forever, and I can search it easily, the proposed scenario does not allow for that. While I can still file and label the emails that have the links, I cannot easily search in the contents, unless I archived that separately, or I go online.
Again, a lot of indirection for no real gain. I can already use something like http://gist.github.com/ or the various pastebins, and a multitude of other solutions if I have a particular problem where they are appropriate.
But for everyday mail? Why would I want to do that? It just makes things more complex, for no purpose or gain at all in the vast majority of cases.
You've given me a lot to think about, thank you very much for taking the time to respond.
I'm still convinced I can turn this into a marketable product, but the challenges will be bigger than I first thought and it definitely will not be for everyone.
Maybe I should approach it more of being a collaboration platform that can use email as a form of initial discovery for participants rather than as an enhanced email platform. Then I could also expand it to include other delivery methods as well. Want to collaborate with someone on Twitter and don't know their email, send it to them via Twitter.
Thanks!