NoConsensus.wtf 2025 — Overview

NoConsensus.wtf was a one-day research workshop run by Flashbots on August 13, 2025, co-located with the Stanford Blockchain Conference (SBC). The theme: P2P networking challenges, censorship resistance, and privacy for distributed systems — specifically applied to Ethereum and block-building infrastructure. Fourteen talks, heavy on cryptography and systems research. The unifying agenda was Flashbots’ own R&D program: building anonymous broadcast into production block-building infrastructure.

Core Research Agenda

Three interlocking problems drove nearly every talk:

1. Scaling blobs/data availability without collapsing the P2P network under bandwidth costs.
2. Censorship resistance — neither consensus liveness nor TEEs alone are sufficient; the whole network stack must be private.
3. Anonymous broadcast — as a cryptographic primitive applicable to mempool privacy, block-building coordination, and censorship-resistant gossip.

Talk Map by Theme

P2P Networking & Blob Scaling

• mempirate (Jonas, Chainbound) — blob propagation napkin math: relays face ~21 TB/month at 48 blobs post-PeerDAS. See P2P Networking in Ethereum.
• Muriel Medard (Optimum/MIT) — RLNC as a provably optimal alternative to GossipSub. See P2P Networking in Ethereum.
• Michal — PANDAS — direct UDP P2P protocol for DAS under Danksharding. 2.5s for 20k nodes. See P2P Networking in Ethereum.
• Leyton Taylor (WorldChain) — raw bandwidth numbers from a production OP Stack L2. See P2P Networking in Ethereum.
• Kartik (Sei Labs) — Autobahn multi-proposer consensus achieving 5 Ggas / 200k TPS. See Autobahn Consensus.

Anonymous Broadcast & Privacy

• Phil Daian (Flashbots) — case for DCNets as unconditional, assumption-free privacy primitive. See Anonymous Broadcast.
• Fan Zhang (Yale) — ZipNet: practical DCNet using untrusted aggregators + TEEs. See Anonymous Broadcast.
• Mark Simkin (Flashbots) — ADCNets: Flashbots’ in-progress anonymous broadcast system. See Anonymous Broadcast.
• Muriel Medard — RLNC interference alignment for anonymization. See Anonymous Broadcast and P2P Networking in Ethereum.

Censorship Resistance in Consensus

• Odysseas Sofikitis (Common Prefix) — liveness ≠ short-term censorship resistance; leaderless attestation-based primitive. See Censorship Resistance in Consensus Protocols.
• Phil Daian — consensus patches (inclusion lists) are insufficient; whole stack must go private. See Censorship Resistance in Consensus Protocols.

Distributed Block Building & TEEs

• dmarz (Flashbots) — BuilderNET production measurements: 4k TPS inbound, 2k req/s inter-node, 1k concurrent users. See Distributed Block Building.
• Quintus Kilbourn (Flashbots) — TEE fail-stop model for spam reduction and P2P attack mitigation. See Distributed Block Building.
• Patrick O’Grady (Commonware) — threshold VRF at consensus latency; timelock encryption demo. See Timelock Encryption and Threshold VRFs.

Validator Infrastructure

• Lioba Heimbach (ETH Zurich) — 16% of network validators deanonymized using subnet data. See Validator Deanonymization on Ethereum’s P2P Network.

Distributed Systems Architecture

• Louis Thibault (Wetware) — object-capability model for secure off-chain distributed apps; caretaker pattern eliminates ambient authority. See Object-Capability Security.

Cross-Cutting Signals

• TEEs as enabling layer: Five talks referenced TEEs — for disruption resistance (ZipNet), spam removal (BuilderNET), off-chain execution (Wetware), anonymous scheduling (ZipNet/ADCNets), distributed block building. The trust model is consistently “fail-stop”: TEE failures are detectable, not silent.
• The network layer is the remaining attack surface: Even inside fully compliant TEEs, metadata (IPs, packet timing, order flow source) is visible to network adversaries. Anonymous broadcast is the response.
• Production numbers as research input: dmarz and mempirate set the design requirements — any anonymous broadcast system must handle ~500 nodes and ~4,000 transactions/second to be production-relevant.
• Unconditional privacy as north star: Phil Daian’s framing — start with information-theoretic privacy (no assumptions), relax only when forced — set the methodological agenda for the ZipNet and ADCNets work.

Connections

• Anonymous Broadcast — Primary research focus of the workshop
• P2P Networking in Ethereum — Bandwidth constraints and P2P protocol work
• Distributed Block Building — Production context and requirements
• Censorship Resistance in Consensus Protocols — Consensus-layer analysis
• Validator Deanonymization on Ethereum’s P2P Network — Empirical staking network security findings

Open Questions

• Can DCNet-based anonymous broadcast reach the 4,000 TPS / 500-node operating point required for production block building?
• Does the TEE fail-stop model hold under realistic hardware attack scenarios?
• Will RLNC or PANDAS displace GossipSub in Ethereum’s P2P layer?