commons-sentience-sandbox

Claim Evidence Ledger

Evidence classes

Ledger

Claim Current class What exists Required next evidence
Scoped, expiring, revocable authority can be represented in software E3 Reference implementation and automated tests Independent security review and integration test
Audit events can be linked so alteration is detectable E3 Hash-chain implementation and tamper test External implementation cross-check and durable witness storage
A failed integrity or authority check can force de-elevation E3 Fail-closed state machine and tests Adversarial concurrency and crash-recovery testing
Prior incidents can be retained as bounded moral-memory records E2 Structured append-only model and consent boundary Retention, correction, deletion, privacy, and bias evaluation
A multimodal interface can classify volunteered affective signals E1 Plausible conventional sensor architecture Ethics review, protocol, dataset, baseline, uncertainty, and replication
Distributed agents can measure policy agreement without centralized command E1 Proposed scoring models Operational definitions, benchmark tasks, and comparison baselines
Ethics can be encoded as an intrinsic state of a physical qubit E0 Speculative prose and notation Physical derivation consistent with quantum information theory and experiment
Entanglement enables instantaneous ethical coordination E0 rejected formulation No supporting evidence Reformulate because entanglement alone cannot transmit information faster than light
Temporal causality can be controlled E0 Toy equations only A physically defined observable and model consistent with relativity and quantum theory
A planetary biosphere forms an addressable consciousness network E0 Speculative hypothesis Operational definition, falsifiable prediction, controls, and empirical protocol

Rule

No claim advances because it sounds compelling. It advances only when the next evidence class is satisfied and the underlying artifacts are public or available for qualified review.