Follow mnm! https://twitter.com/mnmnotmail
To address some comments...
The protocol[1] has two layers, altho that's not emphasized in the draft. The outer layer covers posting messages for recipients. The inner layer describes message contents, and isn't enforced by the server (see protocol #7 Post //datahead segment). You can do special-purpose apps with the outer layer without supporting the inner if you don't send messages to normal clients.
EDIT: The protocol will see many more revisions (some already planned), and real-world use, before it's proposed to a standards body.
This isn't a drop-in/swap-out replacement for email. It's a (far) better way of doing electronic correspondence. It will take some years to build the community and infrastructure necessary for TMTP to widely supplant SMTP. It's been suggested recently that I draft an architecture doc for TMTP[2], which will also discuss infrastructure.
Part of that infrastructure involves "marketplace" sites which verify members' real-life identities and let people make contact with folks outside their normal circles (e.g. work, school, community, hobbies/interests). Most professional organizations would run marketplace sites for their members.
I gather that the IETF has (repeatedly) stated that it is not interested in adjusting SMTP to solve "the spam problem". [3]
Re end-to-end encryption, that can be considered for the inner layer protocol. I don't rule it out, but it's not a priority at present.
If successful, "mnm" won't be the only implementation of TMTP. Maybe some other client will have a better name. (OTOH, Slack did well, and that's a pretty terrible name for a workplace tool ;-)
EDIT2: SMTP and related protocols have no effective defense against phishing. A large fraction of phishing attempts originate at authenticated domains, like Gmail.
---
[1] https://github.com/networkimprov/mnm/blob/master/Protocol.md