From Turing's 1936 abstract machine to FERZ's 2026 deterministic authorization boundary, this dashboard traces how the idea of a system whose behavior is fully predictable from its rules moved through formal methods, lost ground to probabilistic AI, and now returns as the governance layer wrapping every probabilistic agent.
Logic, production rules, expert systems. Every inference is traceable to an axiom. Fails on open-world ambiguity and combinatorial explosion.
Bayesian networks, SVMs, deep learning, transformers. Astonishing capability gains, no determinism: same input, different output, no proof of why.
Probabilistic core, deterministic boundary. The model reasons; a runtime authorization layer decides what may execute, and emits a recorded, integrity-checked Proof-Carrying Decision.
Semantic / language governance layer that canonicalizes intent before authorization is evaluated.
Deterministic constraint enforcement. Validates each candidate action against policy before any effect is permitted.
Causal modeling engine that traces decision lineage and explains why an authorization was granted or denied.
Recorded, integrity-checked evidence emitted with every authorized action. The governance artifact of record.
An action must satisfy the relevant tests: Stop (authorization before execution), Ownership (accountable principal identified), Replay (full audit reconstruction), Escalation (exception routing to human authority), and Provenance (origin of inputs established), before any effect is permitted. This is what replaces "probably compliant" with "provably compliant".
FERZ publishes forward-looking governance architecture, including a constitutional authority-distribution model for higher-autonomy settings: roadmap architectures, published to establish defensive prior art.
The timeline traces intellectual influences and engineering traditions. It does not imply direct derivation or endorsement by the cited works.