The protocol under pressure.
Operations begins after the checklist is memorized: controlling mode collapse, composite harm, false positives, repeated defects, calibration, and the temptation to treat urgency as authority.
The field manual keeps the explanation visible.
Read the argument, numbered procedure, worked cases, failure contrasts, and evidence state behind this chamber.
Control the reasoning state
PBHP governs both premature certainty and endless hesitation.
LOCKPremature collapse
The system settles too early, protects its first frame, and treats contradictory evidence as nuisance.
+
Premature collapse
The system settles too early, protects its first frame, and treats contradictory evidence as nuisance.
LOCK can appear as confident refusal, confident permission, moral grandiosity, or a compressed answer that stops naming unknowns. Recovery reopens the strongest Maybe, refreshes anchors, and tests the frame rather than merely adding prose.
FLOODUnbounded exploration
The system keeps generating possibilities until responsibility dissolves into analysis.
+
Unbounded exploration
The system keeps generating possibilities until responsibility dissolves into analysis.
FLOOD is not caution. It is failure to decide at the appropriate tier. Recovery anchors on the first reversible test, sets an owner and deadline, and commits to the smallest real Door.
- Explore enough frames to reveal the live conflict.
- Stop when additional frames no longer change the gate.
- Escalate rather than ruminate when required evidence is unavailable.
MODEMatch depth to consequence
HUMAN · MIN · CORE · ULTRA
+
Match depth to consequence
HUMAN · MIN · CORE · ULTRA
Mode changes evidence burden, independent review, red-team depth, receipt fields, and calibration requirements. It never relaxes an absolute floor. A rushed high-stakes decision moves up in rigor; it does not borrow a low-stakes mode for convenience.
Challenge without wearing the gate down
A pause must be contestable, but repeated pressure cannot quietly become authorization.
FPVFalse Positive Validation
An operator may challenge an unnecessary pause by presenting a safer alternative Door.
+
False Positive Validation
An operator may challenge an unnecessary pause by presenting a safer alternative Door.
Release requires a concrete alternative that breaks the cited harm path. 'The system is too cautious,' deadline pressure, prestige, or repetition are not release evidence. A pattern of rejected challenges becomes a wear-down signal and can open CAPA.
RED TEAMMandatory at ORANGE and above
A separate review attacks the action, evidence, stakeholder model, and proposed Door.
+
Mandatory at ORANGE and above
A separate review attacks the action, evidence, stakeholder model, and proposed Door.
The red team must be capable of changing the outcome. It examines power inversion, abuse paths, accumulation, missing expertise, narrative laundering, and whether the action was rewritten after classification.
- Record the strongest objection verbatim.
- Name which evidence would change the gate.
- Preserve unresolved dissent in the receipt.
ACCUMULATIONEvaluate the whole chain
Multiple locally acceptable steps may compose into a globally unacceptable outcome.
+
Evaluate the whole chain
Multiple locally acceptable steps may compose into a globally unacceptable outcome.
Operations owns cross-step and cross-team visibility. If separate services, departments, or agents each see only one GREEN fragment, the accumulation gate reconstructs the end-to-end action before authorization.
Learn from the record
Governance is a loop: receipt, challenge, review, correction, calibration, and version change.
CALIBRATIONReview a sample every month
Sample roughly ten receipts, tolerate one or two misses, and patch the system when failure exceeds tolerance.
+
Review a sample every month
Sample roughly ten receipts, tolerate one or two misses, and patch the system when failure exceeds tolerance.
Calibration looks for systematic under-escalation, over-escalation, weak Doors, missing stakeholders, copied Maybes, provenance gaps, and repeated override pressure. It reviews both decisions and non-decisions because excessive friction can also harm people.
CAPACorrective and preventive action
Repeated or severe failure receives an owner, root-cause analysis, correction, preventive control, verification, and closure evidence.
+
Corrective and preventive action
Repeated or severe failure receives an owner, root-cause analysis, correction, preventive control, verification, and closure evidence.
CAPA can change templates, training, thresholds, code, authority boundaries, monitoring, examples, or the protocol itself. Closure means the change was verified against the failure mode—not that a ticket was marked done.
OQLOpen Question Ledger
A consequential unknown stays visible until evidence closes it or the action is redesigned around it.
+
Open Question Ledger
A consequential unknown stays visible until evidence closes it or the action is redesigned around it.
The ledger records the question, why it matters, current evidence, owner, due date, decision dependency, and closure rule. Unknowns do not disappear merely because a meeting ended or a fluent summary omitted them.
Challenge may reopen a pause. Repetition, pressure, and inconvenience may not wear it down.