Nostr Web Client

not meaning to say mean things but you need to define that with precision in order to propose an algorithm to do it

part of the problem is that this network depends on trusted third parties (relays) and so to prevent such access requires encryption and this complicates things because how do you give the key to unlock it only to who you want? via DMs? which sit on relays and many relays right now just hand this data out to anyone?

if we all ran our own relays this might be possible but it's still really complicated to implement on the current state of the internet, that's why i'm saying things that might sound mean

ᴛʜᴇ ᴅᴇᴀᴛʜ ᴏꜰ ᴍʟᴇᴋᴜ 1y ago

like, you could add a prefix, like PGP does, which has a series of versions of the secret used in the symmetric encryption of the message, for each recipient, this will be a fair bit of extra data but it's doable

you could also have a special message that comes through a DM that the client scans and the client would have to decrypt it automatically and it would have a header that indicates it is a cipher for decrypting private events from you

there is threat models to both of these that are relevant as well, and neither of them are as private as a live peer to peer delivery and that's impractical due to intermittent network connections (mobile devices) and general unreliability of networks

and the last piece is what is the purpose of this, exactly, because it is a lot of rigmerole to just have a private status message or something

Reply to this note

Please Login to reply.

Discussion

ᴛʜᴇ ᴅᴇᴀᴛʜ ᴏꜰ ᴍʟᴇᴋᴜ 1y ago

if everyone ran their own relays, and there was a mechanism for messages to be buffered on friends relays for you, people you trust, to bring it closer to 99.9% uptime availability, then yes, this would be practical, and in fact we could dispense with the entire third party relay model and all use outbox model on all our messages and literally not broadcast anything except deliberately, to shared outboxes used as public outboxes... and i shudder to think at the question of how to spam control that...

well, anyway, more ruminations about threat models and networks and signals intelligence, mostly very optimistic ones that presuppose that someone isn't out to get most of us, or potentially out to get most of us