From Institutional Trust to Verifiable Properties

Abstract

Financial settlement systems rely heavily on institutional trust: intermediaries maintain ledgers, certifycompliance, and prevent unauthorized creation or movement of value. Zero-knowledge (ZK) techniques make itpossible to replace part of that trust with verifiable properties. This paper presents a minimal ZK settlementlayer designed around a simple principle: prove what must be true, disclose only what must be seen, anddeclare remaining trust explicitly.We describe an architecture in which transfers preserve value, spending authority is proven without sharingspending keys with the operator, double-spending is prevented, and supervisors can verify balance bands orthresholds without receiving the full ledger. We also map the residual trust surface: the operator of a singlenode can still see balances, order transactions, and censor. The contribution is not a claim of full sovereignty ordecentralization. It is a precise shift from opaque institutional faith toward a smaller, named set of trustassumptions, with cryptographic checks covering the rest.We compare this model conceptually with core banking systems and permissioned blockchains, and argue thatthe main institutional value of ZK settlement is not “trustlessness,” but trust minimization with honestresidual boundaries.This revision subjects that claim to its own standard. An audit pass against the reference implementation foundresidual dependencies the first version of this paper had not named: a confidentiality leak toward thecounterparty rather than the operator, three quantified capacity bounds, and a privilege that is counted butnever expires. We report them in §4.4 and §4.5, because a paper whose contribution is naming residual trust isfalsified by the trust it failed to name.