The Trusted Chip Is Not the Trusted Signal: How Signal Veracity Changes Hardware Partitioning in Consequential Systems
Dana Xiadani · Zenodo (CERN European Organization for Nuclear Research) · 2026
Hardware roots of trust, secure boot, measured state, attestation, and protected communication can establish strong evidence about the identity and operating state of a computing platform. They do not, by themselves, establish that a physical observation entering that platform remains coupled to the claimed source or event, is fresh enough for the intended decision, or satisfies application-specific admission requirements. In sensor-driven medical, defense, industrial, and autonomous systems, that distinction changes architecture. Observation-evidence requirements can determine which properties must be captured or bound near acquisition, which functions require protected execution or non-bypassable enforcement, which evidence may be transported downstream, which logic should remain programmable while the evidence model evolves, and which stable trust invariants may eventually deserve dedicated silicon. This paper introduces evidentiary locality as a hardware-partitioning pressure and proposes a signal-to-substrate framework in which the consequential claim defines the required evidence; the evidence requirements define the trust boundary; the trust boundary constrains partitioning; and partitioning, together with economics, determines substrate. The result is a broader rule for consequential compute: performance determines where computation is efficient; trust determines where computation remains valid. Living Cipher Analysis 002. This publication presents an architectural framework for treating observation-evidence requirements as a first-order hardware-partitioning variable. It does not claim empirical validation of any particular source-veracity mechanism.