Mailet's (2,2)-threshold credential scheme distributes a user's social media credential as Cred1 ⊕ Cred2 across two randomly chosen servers; an adversary corrupting fraction ρ of the server pool has at most probability ρ² of compromising both servers for a given user, and under standard AES assumptions a single compromised server leaks no information about the credential beyond its length.
From 2016-li-mailet — Mailet: Instant Social Networking under Censorship
· §4.2, §6.2
· 2016
· Privacy Enhancing Technologies
Implications
XOR-based (n,2)-threshold secret sharing provides information-theoretic credential privacy against any single compromised server at near-zero computational cost; apply this pattern wherever a proxy must authenticate on behalf of a censored user without trusting either relay.
Pair threshold secret sharing with a Checking-by-Sampling integrity mechanism — Initiator opens n parallel TLS sessions; Interceptor opens n−1 to verify non-credential fields — to deter malicious server tampering without requiring full mutual trust.