A thesis under active construction: as machines become able to commit actions in the world, institutions need computational infrastructure that keeps human-written meaning, warrant and authority attached to what those machines do.
Rules as Code and Policy as Code moved institutional rules closer to software. That progress matters. The open question begins one layer upstream: human-written regulation contains interpretation, ambiguity, authority, time and institutional judgment that do not disappear when the rule becomes executable.
Make parts of law and regulation machine-consumable. OpenFisca, Better Rules and legal DSL work narrowed the translation gap.
OpenFisca ↗OPA and Rego are excellent at the job they were built to do: given policy and input, compute a decision.
Open Policy Agent ↗Keep the institution attached to the executable rule: source, interpretation, warrant, authority, version, execution, evidence, consequence and correction.
package payments
default allow := false
allow if {
input.amount <= 50000
input.secondary_approval == true
}What does the human-written source mean in a form software can use?
Source anchoring → candidate meaning → ambiguity → institutional admission → machine-operational control → lineage.
Engineering infrastructure advanced materially this week; the semantic compiler gate remains deliberately blocked.What has actually been established strongly enough to rely on?
Unverified grounds remain different from earned conclusions. This week the executable/formal surface added temporal warrant, verdict receipts, ground registries and transitive invalidation.
Who is entitled to admit, rely on, revoke or correct the institutional decision?
Machine analysis remains separate from the institutional act that gives an interpretation standing. The architectural responsibility remains; public implementation maturity did not move this week.
Can another system verify what actually happened without trusting the application that produced the evidence?
The protocol validates and replays deterministic execution evidence. Institutional Continuity Infrastructure now makes its ceiling explicit: execution evidence does not by itself prove consequence, observation coverage or later examinability.
Explore the detailed protocol root mapping ↗AuthContract is the first developer-facing capability. The developer should not need a new governance vocabulary. The question is familiar: did this code change introduce behavior the source never authorized?
- required_above: 50000 + required_above: 100000 # YAML valid # Unit tests pass # Application starts
AuthContract is being built to make the relationship between institutional source and executable behavior testable in continuous integration. Its public reference implementation now demonstrates the full mechanical path in a controlled environment; automated natural-language source-to-rule comparison remains target behavior.
An institutionally grounded control still has to survive changing state, machine memory and execution. These are enabling layers around the core, not substitutes for it.
A prior approval does not establish current institutional usability after evidence, authority or operating state changes.
The reference implementation remains fast, but a public cross-project probe reported declared policy fields that may not currently participate in the hot-path evaluator. The site now treats that as an unresolved enforcement finding rather than a green maturity signal.
OPEN FINDING SOURCE ↗The public 0.1 protocol and verifier remain executable proof-of-concept infrastructure. No material implementation change was found this week; the newer continuity architecture sharpens exactly what the protocol proves and what it does not.
EXPLORE PROTOCOL ROOT MAPPING ↗The new canonical chain has nine nodes and eight mandatory joins. It asks whether warranted institutional meaning survives from its governing source all the way to a consequence that can be independently observed, reconciled and examined later.
A receipt can prove exactly what the application says it executed. It cannot by itself establish that the external consequence occurred, that all relevant consequences were observed, or that a later examiner can reconstruct the complete institutional chain.
WARRANTED CONTROL ESTABLISHED RUNTIME AUTHORITY VALID EXACT ACTION BOUND EXECUTION EVIDENCE VALID CONSEQUENCE OCCURRED? ? INDEPENDENTLY OBSERVED? ? OBSERVATION POPULATION COMPLETE? ? RECONSTRUCTABLE LATER? ? Authorization is one link. Execution evidence is one link. Neither is the whole chain.
Claim ceiling · 28 Aug 2026: the continuity architecture and test methodology are public. Full Institutional Continuity is not yet established for any public implementation, and no public implementation currently has a published Institutional Continuity Test Suite score.
This section makes one narrow infrastructure claim. A United States banking rule, a European Union regulation, a Monetary Authority of Singapore requirement, a World Bank procurement rule and a United Nations mandate can carry different legal meaning, authority structures and effective dates. Those differences must be resolved upstream. But once an institution has established the machine-operational control, the downstream evidence questions become structurally reusable.
If every jurisdiction requires a different execution-evidence engine, the industry recreates the same translation problem at the verification layer and evidence stops being portable.
This does not mean one protocol interprets every regulation. It means the protocol can carry explicit references to the institution-specific meaning that was established before execution.
This is the Aug. 28 state of the architecture already introduced above. A green week can move readiness upward, hold it despite substantial engineering work, expose a negative finding, or leave a public implementation unchanged.
NOWHarden the continuity test methodology, reference aggregation kernel and adversarial vector corpus.
NEXT EVIDENCE GATEPublish the first pinned implementation evaluations; no public score exists yet.
NOWClose source-rights/provenance prerequisites and required interface evidence while keeping semantic code-start fail-closed.
NEXT EVIDENCE GATEExecute bounded procurement semantic preflight and publish raw benchmark evidence.
NOWContinue falsifying formal/executable gaps; harden temporal warrant, grounding and lineage behavior.
NEXT EVIDENCE GATEPin externally authored evaluation corpora and obtain independent reproduction.
NOWFreeze the executable authority/adjudication model and human–machine decision boundary.
NEXT EVIDENCE GATEBuild the first reference implementation and audit case corpus.
NOWEvaluate an externally authored rule and implement multi-contract selection/conflict semantics.
NEXT EVIDENCE GATECross-platform determinism and independent non-author reproduction — required for Level 5.
NOWMap the bounded state-transition predicates into the continuity chain.
NEXT EVIDENCE GATEReuse the experiment as an integration/conformance vector rather than expand it as a standalone product.
NOWReproduce and adjudicate the public hot-path policy-field finding; repair semantics or narrow claims.
NEXT EVIDENCE GATERe-run full correctness/performance evidence and obtain independent relevant-environment reproduction.
NOWMap the public verifier precisely onto continuity joins and preserve its execution-evidence claim ceiling.
NEXT EVIDENCE GATEVersioned profiles plus independent implementation/reproduction.
* ASTERISK = PROVISIONAL VERAXIS.IO EVIDENCE-DERIVED READINESS ASSESSMENT. CATEGORY ARCHITECTURES ARE NOT GIVEN IMPLEMENTATION READINESS LEVELS.
Every previously stated claim now carries an Aug. 28 maturity movement, current public evidence, current work and the next evidence gate. Negative findings and unchanged surfaces remain visible.
Can institutional meaning, authority and evidence survive from governing source through real-world consequence and later examination?
Public nine-node/eight-join architecture, category constitution, test methodology, release identity and research paper. No public implementation has a published continuity score.
Harden test methodology, reference kernel candidate and adversarial corpus.
First pinned implementation evaluations and compatibility profiles.
What does the institutional source mean in a form software can use?
Non-semantic infrastructure materially advanced: repository evidence reports 9/9 schema validation, 52/52 bootstrap integrity, 4/4 infrastructure falsification outcomes and 1,232 passing tests plus one intentional skip. Semantic code-start remains blocked.
Close source-rights/provenance and required interface prerequisites without weakening the semantic gate.
Bounded procurement semantic preflight and raw benchmark evidence.
What has actually been established strongly enough to rely on?
Current public head reports run_all ALL GREEN with 128 stands plus Lean. This week added typed preservation, temporal/expiry semantics, verdict receipts, ground registries and transitive invalidation.
Keep source authority and external truth separate from logical warrant while falsifying executable/formal gaps.
Externally authored evaluation corpora and independent reproduction.
Who is entitled to make, admit, rely on, revoke or correct the institutional decision?
The architectural authority/adjudication responsibility remains part of the category; no material public implementation evidence moved this week.
Freeze the executable authority/adjudication model and human–machine decision boundary.
First reference implementation and audit case corpus.
Can a developer prove that the machine-control object, runtime facts and exact action remain mechanically bound?
7/7 end-to-end, 38/38 adversarial, 342 regression tests, 5/5 public falsification outcomes; ~603 μs mean end-to-end and ~1,697.5 observed single-threaded operations/second median in the current controlled environment.
Use externally authored rules and implement multi-contract selection/conflict semantics.
Cross-platform determinism and independent non-author reproduction.
May a previously valid authorization still be institutionally usable now?
Bounded deterministic experiment with seven predicates and seven scenario types remains the public evidence. Its architectural role is now mapped more clearly into the continuity chain.
Preserve as a reference experiment; map predicates into integration/conformance vectors.
Use state-transition cases across maintained continuity integrations.
Which prior machine state is still entitled to influence the next action?
Reference benchmark remains ~33 ns/op, 0 B/op, 0 allocations/op on the hot path. A public cross-project probe reported declared policy fields that may not be read by that evaluator; no corrective code commit is visible in the current snapshot.
Reproduce/adjudicate the finding; make the semantics executable or narrow unsupported claims.
Re-run correctness/performance after repair and obtain independent relevant-environment reproduction.
Can another system verify execution evidence without trusting the application that emitted it?
Public 0.1 specification and reference verifier provide schema validation, canonicalization, tamper checking and deterministic replay. No material public code movement this week.
Map exact coverage to continuity joins and keep the execution-evidence claim ceiling explicit.
Versioned profiles and independent implementation/reproduction.
Open infrastructure becomes credible when it is forced into real institutional problems. These proposal tracks are not endorsements; they are places where the architecture encounters procurement, public authority, jurisdiction, evidence and consequence.
Public-sector procurement / audit application of open institutional-control and execution-evidence infrastructure.
Application / submission context only. No partnership, sponsorship or endorsement implied.Apply the open architecture to public-sector machine action that must remain attributable to institutional authority and reconstructable later.
Application / submission context only. No partnership, sponsorship or endorsement implied.Apply portable institutional controls and evidence continuity to cross-organizational or cross-jurisdictional workflows.
Application / submission context only. No partnership, sponsorship or endorsement implied.This week’s log records maturation, held gates, negative findings and unchanged public implementations. Every public claim links directly to the supporting artifact; private evidence is labeled private.
The Aug. 28 continuity paper is now the leading research artifact, followed by the broader published research and protocol family. Download is the primary action. Zenodo remains the authoritative record and supplies live usage statistics after deployment.
The new end-to-end continuity model: source → admitted meaning → warranted control → runtime authority → exact action → consequence → independent observation → reconciliation → examination.
Consequence, downstream reliance and correction.
The field boundary and public challenge.
Whether an institution may legitimately rely on an agentic output at the moment of operation.
Published formal account of the Zero-Trust Logic core.
Computable classification of classical paradoxes built on the Zero-Trust Logic core.
Knowledge treated as an earned verdict using the Zero-Trust Logic core.
Persistent concept record for the formal logic project.
Public working-draft specification for execution integrity and evidence verification.
Reference developer surface for the public protocol.
Support can mean funding a week of open research, contributing code, introducing an institutional problem, challenging a claim, or putting the architecture into a real operating environment.