On 8/26/13, rysiek
It's a seemingly unsolvable conundrum: - start with a clean slate and create a new, "perfect" solution; - start with something people already use and improve it as much as possible.
The former approach has the risk of the solution not being adopted; the latter -- of the new solution not being good enough due to technicalities of the solution it is based on.
But maybe it could be possible to get the best of both worlds?
Had a new solution been created in a way that is usable via existing mail clients (e.g. implementing IMAP for message retrieval and SMTP for message submission) while designing and implementing a completely new way of comminicating server-to-server -- it might achieve just that.
When you cleanslate things you are not held back by the past. Napster, bittorrent, icq... all completely new and instant mass appeal, because they did something that people wanted. That overrode every other consideration the user had, including... now hang on, I know this will be hard for some to believe.... downloading and installing the app and actually learning how to use it. These days, there is a want for secure messaging. Thanks to recent news... I'm actually taking questions from morons off the street that now want to 'send secret messages'. I don't have time so the standard answer is usually 'read and pick something from prism-break.org.' There are already cleanslate systems cryptographically secure against content and internal addressing snooping. The resistance to global passive/active adversaries from connecting the realworld IP talkers needs work. And to truly replace traditional email (especially in the business world) you will need a community of powerful messaging clients... lots of spec decisions to be made there... lots of traction with existing thunderbird, mutt, etc UI/handling frontends. But can you save that and swap out their legacy smtp/imap network semantics. Then you have the userspace key management for end2end message security to work on as well. All of this is interconnected too... once you start designing a strong messaging system, it begins to look like a general purpose transport, possibly one with parameterized options. Then you start thinking how to plug other apps into it. So you're back to looking like another I2P, Tor, GnuNet, Phantom, Freenet, AnoNet, etc... lots of the same stuff under the hood. Will there ever be a generic, possibly modular, overlay to come along and handle it all? Or are we just running in overlay circles?