There already is a system which could do what you propose: Just use standard SMTP via TOR. Each communication partner would run their own mailserver as a hidden service on TOR. You can then write emails to conveniently named email addresses like
But that won't ever become meainstream. Why?
The reason is that a communication medium can only replace another when it is more convenient to use. When you give users the choice between security and convenience, the majority will always choose convenience.
So when you want to replace email (please do, it's such a horribly broken protocol on so many levels), you don't just need to create a system which is better from a security perspective, but also better from a usability perspective. Unfortunately, in most cases security and convenience are mutually exclusive.
Some features of email which might be difficult (I don't claim impossible, though) with the system you propose could be:
- Easy to set up and maintain for people with zero IT skills
- Require no acquisition of new hardware
- Guarantee that a message either gets delivered or that the sender receives a useful error message when it can't be delivered.
- Checking for new messages within seconds after connecting to the network
- Delivery of new messages within seconds when both users are already connected to the network
- Downloading large attachments with your full bandwidth
- Human-readable and memorable addresses (like
- The ability to send messages from anyone to anyone without prior key exchange
- Access your existing identity on a new device using only information you can easily memorize
- Backward compatibility with standard Email to ease the transition
When your system can not do all of these, it will have a hard time replacing email. And your system doesn't just need to be just as good as email but actually better, so you better come up with one or more features which improve usability and which standard email can not do.