Transaction Authorization Beyond Signing Authority

Research paper (preprint, not peer-reviewed) on digital-asset custody architecture, motivated by the Binance security incident disclosed on 30 October 2026. The paper addresses a fundamental problem in digital asset custody: an attacker may cause an authorized infrastructure to produce valid signatures without extracting its private key. A signature proves a mathematical relationship between a message and a key; it does not establish that a withdrawal reflects the asset owner's economic intention. The paper develops a custody architecture designed to prevent a compromised operational environment from unilaterally translating requests into asset movements. Its mechanisms combine independent interpretation of transaction intent, policy enforcement, distributed approval, chain-specific controls (including EIP-7702 delegation), atomic exposure limits and a physically separate hardware authorization key. The architecture is placed within a broader research programme on Mock Theta Cryptography. The guiding question is whether a carefully specified observable transcript can leave multiple admissible private structures unresolved — structural non-identifiability — as distinct from computational hardness. The paper shows why ambiguity alone does not imply unforgeability, formulates the mathematical and security obligations needed to test a transaction-bound authorization mechanism, and explains how a future validated Mock Theta layer could operate in parallel with established controls. Drawing on the multidisciplinary treatment of model error, nonlinear exposure and tail risk associated with the Real World Risk Institute and the work of Nassim Nicholas Taleb and Raphaël Douady, it also examines common-mode failures, concentrated custody and the economics of containment. Status and scope. This is a research proposal. No completed hardware product, security proof, experimental result, independent validation, or prevention of the Binance incident is claimed. Incident figures (approximately US$6.3 billion estimated impact; approximately US$3 billion frozen) are time-bound estimates reported by Binance and third-party analysts at the time of writing and may be revised. The baseline custody controls can be evaluated independently of the proposed Mock Theta primitive. Plates and figures are conceptual illustrations, not computed plots or depictions of deployed systems. Contents: incident context and limits of evidence; real-world risk and the fragility of custody; from keys to authority; EIP-7702 and the signing boundary; system model and threat boundaries; canonical transaction intent; architecture and enforcement; formal security objectives; the ontology of structural non-identifiability; mock modular structure as a candidate substrate; a Mock Theta authorization protocol as a falsifiable research object; the hardware key as an independent authorization boundary; layered comparison of design choices; a finite-field diagnostic; decision criteria; engineering the authorization boundary; threat scenarios; research prototype and evaluation; containment and recovery; economic model and exposure allocation; deployment roadmap. Version 1.0, 30 October 2026.