Flashnet in the Tradeoff Space
Mark Simkin, Mateusz Morusiewicz
Session summary
The presenters describe Flashnet as an anonymous broadcast protocol where many clients send encrypted messages via multiple servers, and only a shuffled set of messages appears on a public bulletin board. Their anonymity goal is that, even if many clients or servers are compromised, each honest user’s message remains indistinguishable among the honest users. Clients run protocol logic in trusted execution environments (TEEs), which are used to enforce message structure and correctness. The presenters emphasize that, in their design, breaking a client TEE should at most enable denial-of-service, not reveal honest users’ messages.
They position Flashnet in a tradeoff space relative to Tor, mixnets, and mailbox/PIR-style approaches. The authors aim for strong anonymity and low latency per broadcast round, while accepting higher bandwidth costs and a round-based model. They discuss how choices such as tolerated numbers of corrupt servers, desired anonymity set size, message size, and sending frequency determine the system’s parameters. They also note that zero-knowledge proofs are being explored as an alternative to TEEs, and that cover traffic from non-sending participants can expand the effective anonymity set without linearly increasing communication for senders.
Key takeaways
- The authors aim for an anonymous broadcast where any honest message remains indistinguishable among the other honest users, even if many clients are corrupt.
- Client logic runs in TEEs to ensure messages are well formed; the presenters state that a broken TEE mainly risks denial-of-service rather than exposing other users’ anonymity.
- Flashnet trades higher bandwidth and a round-based interface for strong anonymity and low latency, with parameters tuned by desired anonymity set size, message characteristics, and server corruption thresholds.