EXPANDED / PRESS ESC OR CLICK TO EXIT
carrying, requirement satisfied
blocking, requirement failed
informational, never gates
lease or controller constrained
Runtime obligations
Findings
Three policies, one engine
The same evidence collapse under three policies. Only mode C is the accepted profile.
Modes A and B are rejected designs kept as negative controls. Neither issues a platform command for its
rejected action and neither models a physical effect.
Mode A, continue on failed evidence
Mode B, terminal action on trust decline
REJECTED ABSTRACT NEGATIVE CONTROL.
No platform command. No physical effect.
Mode C, evidence conditioned authority lapse
Result
Run the comparison.
Verification
Verification recomputes the digest, replays every recorded input from the seed, and requires
the replayed ledger head, ledger digest, findings, final state and run identity to match. This is a deterministic
behavioural replay checksum within the stated trust boundary. It is not a digital signature and it is not
independently attested.
Run a verification.
Hash chained record
(empty)
What this is
AUTHREX-ABORT, catalog identifier SIM-13, is a deterministic synthetic simulation of evidence
conditioned authority lapse and pre authorized abort governance. Humans authorise a bounded envelope before
deployment. Each action depends on an explicitly identified evidence set. When that evidence degrades, authority
contracts; when it fails, authority lapses and an irreversible action is latched down for the run. Only a
contingency authorised in advance may then be requested, and every transition is recorded in a hash chained
record that can be replayed.
The canonical outcomes are EXECUTE, HANDOFF, ABORT and DELAY. Inhibit and Divert are
internal contingencies beneath ABORT, not outcomes. SafeState is deferred and not implemented. HOLD is an ENGAGE
domain extension and is out of scope. No destructive action exists anywhere in this vocabulary, because degraded
evidence cannot create a stronger irreversible authority than the evidence that has just failed.
The rejected trust triggered policy is present only as an abstract negative control. It is
not requestable, has no command mapping and no canonical outcome, it never reaches the adapter, and the engine's
own oracles record it as failing five separate claims.
Formal baseline
AuthorityLapsev3
hashea6c4fffa5583a556d989cff57ad47e27f160a85adc32d860c79db32fbce2993
InterfaceRecordv1
hasha3f5e2ebceaa0ba080f422a00d2f425ed6b5b4d82ab2a1db8b4f6b0f78739c27
engine
Status, stated plainly
| item | state |
|---|---|
| Stage 1 raw availability | 972 of 972 projected states |
| Stage 1B formal to code conformance | not performed |
| mutation assurance | 1 detected, 3 surviving, 1 invalid |
| oracle applicability | interface derived |
| scenario catalog | 23 of a planned 42 |
| frozen core hash | not issued |
| catalog entry | none |
| third party verification | none |
| formal TRL determination | none performed |
Limitations and claim boundary
LIMITATIONS AND CLAIM BOUNDARY. Deterministic synthetic research prototype.
Governance and assurance only. It contains no physical effector, no destructive mechanism, no payload interface,
no real coordinates and no real target data. Canonical outcomes are EXECUTE, HANDOFF, ABORT and DELAY.
Behaviour is demonstrated in simulation. This is not implementation level verification, certification,
accreditation, qualification, operational validation, integration, or independent third party verification, and
no formal TRL determination has been performed. MANUAL operation is exploratory and is not eligible for a run
for record claim. The schematic is a dependency diagram of authority; no position carries spatial meaning.
© 2026 Burak Oktenli, MBA. Georgetown University M.P.S. Applied Intelligence. ORCID 0009-0001-8573-1667.
Washington, DC. CC BY 4.0.