Yes, but it's done at a lower level, which enables a host of attacks, like the announcement says. What I'm talking about would just encrypt the payload, so none of the metadata would be encrypted (and thus preserved).
Although I guess you'd also need to specify a "reply" public key in the encrypted data, so this is becomes more of a protocol.
Comments
It would be interesting for services to publish a public encryption key, so the signal client could encrypt the payload with that.
However, that has very limited usefulness, so I don't see it happening soon.
That's more or less what TLS+pinning does. Also DNSSEC+DANE+TLS if you want to argue about that.
Yes, but it's done at a lower level, which enables a host of attacks, like the announcement says. What I'm talking about would just encrypt the payload, so none of the metadata would be encrypted (and thus preserved).
Although I guess you'd also need to specify a "reply" public key in the encrypted data, so this is becomes more of a protocol.