Appendix F. Communications Resilience Mechanisms

Jason St George. "Appendix F. Communications Resilience Mechanisms" in Next Generation Stores of Value: Privacy, Proofs, Compute. Version v3.1. /v/3.1/read/appendix/f-comms/

Communications Resilience Mechanisms

P2P transport set:

• v2 Encrypted P2P (BIP-324 [Mehta et al. 2023]) as default for all full nodes and relays; expose QUIC/TLS fallbacks.

• Onion/I2P modes first-class in clients; auto-failover if clearnet handshakes exhibit DPI resets.

• Handshake camouflage for relays in high-interdiction ASNs — with the caveat that fully encrypted, uniform-random-bytes transports (Noise, obfs4-class) are themselves a fingerprint the Great Firewall has blocked heuristically since 2021 [Wu et al. 2023], so camouflage must include look-like-something transports, not only look-like-nothing ones.

What a lab would need to specify: the transport list.

The transport set is stated as classes so that VerifyReach can publish $p_{95}$ per class (Appendix A: Formal Model of Verification Asymmetry & VerifyPrice, Definition 4), and each class names a reference implementation a lab would integrate rather than write:

• Direct encrypted: BIP-324 [Mehta et al. 2023] v2 transport (reference: Bitcoin Core [Bitcoin Core 2024]); QUIC/TLS 1.3 with Encrypted Client Hello pinned.

• Fully encrypted obfuscation: obfs4-class pluggable transports (reference: Tor pluggable transports). Retained for jurisdictions that do not run entropy-based blocking; published as its own class because it fails first where they do.

• Mimicry (look-like-TLS): transports whose wire image is a valid TLS session to a plausible host — uTLS-style fingerprint mimicry and WebTunnel-class HTTPS tunnels (reference: Tor WebTunnel) — so the flow is indistinguishable from ordinary web traffic to a passive observer.

• Rendezvous: Snowflake-class WebRTC through ephemeral volunteer proxies, with a domain-fronted or AMP-cache broker (reference: Tor Snowflake), for the case where every fixed endpoint is blocked.

• Refraction: Conjure-class refraction networking, in which the client connects to an unused address in a partner ISP’s space and a station in the path redirects the flow (reference: Refraction Networking / Conjure [Frolov et al. 2019]); the highest-latency and highest-partnership-cost class, and the fallback of last resort.

• Onion routing: Tor and I2P as first-class transports, not fallbacks, for receiver privacy as well as reachability [Dingledine et al. 2004].

The client’s failover order is a published policy, not a heuristic, and every fallback taken is recorded in $\text{fail}_{\text{mix}}$ so that a region quietly living on refraction is visible on the board.

Receiver-private routing for payments:

• BOLT12 Offers [NoBS Bitcoin 2024b] (receiver-private, reusable invoices) in wallets and merchant stacks.

• Path-blinding where supported.

• Federated ecash [Fedimint Project 2024–25] and shielded pools as optional “last-mile” sinks/sources.

Settlement survivability:

• Adaptor-signature atomic swaps [Gugger 2020] for BTC$\leftrightarrow$XMR/ZEC corridors with typed failure & refund flows.

• Wallet UX exposes “abort & refund” as a first-class path.

Topology & anti-eclipse hardening:

• Peer-set diversity targets (geo/ASN spread), randomized peer rotation, inbound slot quotas.

• Gossip-path multiplicity to reduce single-jurisdiction capture.

• Light-client modes that verify receipts over any reachable path (browser, mobile, enclave).

New here? Start with the one-minute version.

Tip: hover a heading to reveal its permalink symbol for copying.