Resolution Assurance Governance
Human Governance Override Authority
HGOA is Resolution Assurance's final human release-control layer: it can publish, hold, escalate, suppress, require manual-only release, or hide scores before public output appears.
This is the public-safe governance surface: release states, authority boundaries, domain locks, and immutable audit requirements are visible, while private thresholds, reviewer identities, and sensitive case payloads remain withheld.
Release States
Six outcomes, one final release layer
Publish
Safe for public release under the current policy posture.
Hold
Not released while freshness, evidence, or reviewer action remains incomplete.
Escalate
Requires higher-authority human review before release or suppression.
Suppress
Must not be publicly published because falsity, contradiction, policy, legal, or harm risk is active.
Manual Only
Exists internally but requires named human approval before any public release.
Score Hidden
Object may appear publicly while the public score remains withheld until stability improves.
Score Visibility
Publication is separate from score visibility
Visible
Public score is shown normally.
Hidden
Public score is withheld.
Frozen
Last approved score remains visible while live updates are paused.
Authority Levels
Who can set which governance states
| Authority | Can Set | Cannot Set |
|---|---|---|
| L1 Reviewer | publish, hold, score_hidden | manual_only, suppress, sensitive_domain_override |
| L2 Governance Reviewer | publish, hold, escalate, score_hidden, manual_only, suppress | n/a |
| L3 Executive / Legal / Senior Governance | publish, hold, escalate, score_hidden, manual_only, suppress | n/a |
Domain Locks
Sensitive domains default to manual release authority
| Class | Default Path | Examples |
|---|---|---|
| Standard release domains | auto publish allowed if thresholds pass | science, technology, climate, business, general_society |
| Sensitive domains | manual only until authorized review | conflict, terrorism, elections, public_health_emergencies, named_crime_allegations, suicide_self_harm_adjacent, civil_unrest, legal_accusations_pre_verification |
Trigger Groups
Public-safe trigger categories
Auto-publish gate
PUBLISH when governance and compliance healthy, decision readiness above threshold, contradiction drift and risk below threshold, publication gate passes existing readiness rules.
Auto-hold gate
HOLD when uncertainty above threshold, fresh contradictory evidence detected, signal coverage partial or weak, source quality below minimum posture.
Auto-escalate gate
ESCALATE when high public harm potential, political conflict health or legal sensitivity, large public reach with unstable evidence, score swing exceeds override threshold, manual domain lock active.
Auto-suppress gate
SUPPRESS when impossible or contradicted state detected, coordinated falsehood or manipulation suspected, evidence collapses below minimum support, policy or legal block active.
Audit Contract
Every HGOA action writes an immutable record
- object_id
- timestamp
- prior_state
- new_state
- trigger_reason
- reviewer_identity_if_human_action
- evidence_snapshot_id
- score_visibility_state
- domain_policy_in_effect
- revisit_time
Published here: governance structure, state model, authority tiers, and audit fields. Withheld here: private reviewer identities, exact numeric thresholds, and sensitive evidence payloads.