you’re not wrong, but that’s just the trade off that has to be made, i think. it’s the only way i can think to do it, at least. need -some- authentication for practical usability.

your gpg example removes the deniability since it proves who wrote the message.

interesting point! i chose symmetric shared key because it means you can’t prove who sent what message.

the shared secret does add some authentication, which i think is necessary. the goal is it only creates enough to be practical (a random person can’t eavesdrop), but not enough to prove things. messages themselves still aren’t authenticated by any one person.