Both of these solutions are distributed, encrypted entirely (even metadata) and solve email's privacy problems.
The Tor Anonymity Router mentions the types of monitoring they attempt to subvert, and mention that full internet monitoring is not one of their design goals. This is simply because with any low-latency system, correlation between the entry and exit can be obtained. Similar attacks exist for networks such as Bitcoin for similar reasons.
> But what journalists who need to communicate more anonymously that that used pseudonymous PGP keys - one per story
In this case you run into troubles with authenticity. Without persistent keys being authenticated in some way (see "web of trust"), there's no way in telling if the person is communicating with a government proxy or the real target journalist.
That's the issue with the "encrypt everything" mantra, authenticity is stupid difficult to verify.
I think something sufficiently secure is possible for email. I think it would require a whole new set of protocols and approaches, and for some things (like voice) it is not.
There seems to be research under way to get a messaging protocol built into GNUnet. Google turns up this pdf [2] for example.
Freenet is a global distributed p2p file database. Email does not fit that medium.
Why not?
Email can be implemented in a global p2p file database. Every "email" simply adds another file. Privacy can be ensured with encryption (as long as the encryption is not broken, at least). See existing Freenet message boards for an example.
It may not be Internet RFC821/822 (and subsequent standards) Email, but it is email.
What part of this will not fit the Freenet model?
See https://freenetproject.org/understand.html.
Something like:
Publish a site using a USK (perhaps one per recipient), containing all encrypted messages sent recently. Update the site with an additional message when you need to send one. The recipient checks it for updates periodically.
I'm not sure if my understanding is inaccurate, but since Frost already achieves something like this, I don't see that it's automatically impossible such as what you seem to be inferring.
In fact, now that I look, there's: https://freenetproject.org/freemail.html
I'm dubious about the safety of using non-Freenet clients, but surely this demonstrates that it's possible to build a messaging system built over Freenet?