Disentangle: Topological Mass Consensus with Capability-Coherence Identity for Sybil-Resistant Agreement via Discrete Curvature

Abstract

Disentangle introduces Topological Mass Consensus (TMC), a permissionless consensus mechanism that derives Sybil resistance from discrete curvature on transaction DAGs rather than proof-of-work or proof-of-stake. Edges connecting attack clusters to the honest network exhibit negative Jaccard curvature due to low ancestor overlap, enabling geometric throttling without trusted seeds or economic incentives. The protocol uses exclusively post-quantum cryptography (ML-DSA, ML-KEM, SHA3-256, Plonky3 STARKs) and derives all temporal properties from topological depth. We also present the Capability-Coherence Identity Protocol (CCIP), unifying DID-based identity, object capabilities, and petname naming under the same curvature analysis. Implementation: 9 Rust crates, 349+ tests.