Very neat! Great job. I am always interested in these sorts of projects that rethink how we communicate and aren't built on the assumption of always-on, always connected to global web.
Few questions:
With the nodes registration being centralized to a single server, doesn't that make that server the main target for attacks? As if an attacker could get into that, they could add themselves, drop people, etc.
Rather than just using something like a single shared secret, have you considered something like Shamir's Secret Sharing where you need the network consensus / multiple people to make modifications to the list. This would distribute the risk, making the central server less 'tasty'.
With this being targeted towards privacy-conscious / high-risk targets such as journalists, what are your thoughts on the privacy being breached if one person gets co-opted / snatched and the other party/parties being "the bad guys". Is that outside the scope or unreasonable? I was trying to think of how you could avoid that, but all of them get killed by thinking if one of the secrets/parties gets compromised, the network itself is compromised, so no amount of keys, checks, consensus, etc would fix it.
I think that last point _does_ break your model of the network being inherent trust vs zero trust.
Few questions:
With the nodes registration being centralized to a single server, doesn't that make that server the main target for attacks? As if an attacker could get into that, they could add themselves, drop people, etc.
Rather than just using something like a single shared secret, have you considered something like Shamir's Secret Sharing where you need the network consensus / multiple people to make modifications to the list. This would distribute the risk, making the central server less 'tasty'.
With this being targeted towards privacy-conscious / high-risk targets such as journalists, what are your thoughts on the privacy being breached if one person gets co-opted / snatched and the other party/parties being "the bad guys". Is that outside the scope or unreasonable? I was trying to think of how you could avoid that, but all of them get killed by thinking if one of the secrets/parties gets compromised, the network itself is compromised, so no amount of keys, checks, consensus, etc would fix it.
I think that last point _does_ break your model of the network being inherent trust vs zero trust.