The Gold Standard · Archaeology
Make the decision path visible
Conditional approval is useful only when a breached condition changes what may happen next. If override growth, coverage loss or a provenance defect crosses its trigger and the release gate remains open, monitoring has become decoration.
The gold standard links every trigger to immediate scope control, preserves the evidence window and revalidates the redesign against the context that failed. Suspension is bounded protection, not a verdict that all prior work was wrong.
Choose the route closest to your decision
Chapter contents
Chapters 1–3
Chapters 4–6
Chapters 7–9
Chapters 10–12
CHAPTER 1 OF 12 · READ THE TRIGGER
1. Make triggers operational
Each trigger should name its metric, population, threshold, time window, evidence owner and immediate action. Distinguish warning, scope suspension and full revocation levels.
A trigger that only sends a report does not protect a release decision.
CHAPTER 2 OF 12 · READ THE TRIGGER
2. Verify the breach without delaying containment
Check source, denominator, timing and data integrity, while applying the predefined provisional control. Preserve the original observation and later corrections.
Verification prevents false alarms from becoming permanent decisions, but should not leave high-consequence cases exposed.
Evidence check
Check whether the scope, owner and decision consequence are explicit before continuing.
CHAPTER 3 OF 12 · READ THE TRIGGER
3. Bound the affected scope
Identify site contexts, release types, versions and time periods touched by the failed condition. Keep validated low-risk routes open only when their independence is evidenced.
Avoid both extremes: ignoring a material trigger and shutting every archaeological workflow without causal connection.
| Field | Purpose |
|---|---|
| Metric | Observed condition |
| Threshold | Decision boundary |
| Scope | Affected use |
| Action | Immediate control |
CHAPTER 4 OF 12 · SUSPEND SAFELY
4. Activate suspension and compensating review
Block automatic release in the affected scope, route cases to independent review and label provisional results. Set ownership and an expiry for the temporary path.
Suspension should be visible in the same systems that enforce the original conditional approval.
CHAPTER 5 OF 12 · SUSPEND SAFELY
5. Preserve the breach window
Capture cases processed since the metric first deteriorated, not only since the dashboard crossed its threshold. Retain versions, overrides, provenance records and reviewer queues.
The exposure window may start earlier than the formal alert because aggregation and reporting introduce delay.
Evidence check
Ask a second reviewer to reconstruct the example from the preserved evidence and stated assumptions.
CHAPTER 6 OF 12 · SUSPEND SAFELY
6. Worked example: coverage falls below the approval condition
The redesign was approved for ordinary trench sequences while independent review covered cross-trench joins. Monitoring shows join coverage below the agreed threshold for three weeks because the queue filter stopped recognising renamed site codes.
The gate suspends join releases, reconstructs the missed population and restores manual review. Changed-context check: if a temporary reporting error understated coverage while case-level evidence proves every join was reviewed, the trigger is corrected and suspension can end without revalidating control logic.
| Area | Response |
|---|---|
| Affected release | Block |
| Independent route | Verify before use |
| Queue | Preserve |
| Results | Label provisional |
CHAPTER 7 OF 12 · REVALIDATE
7. Find whether cause is control, data or process
Test the redesign rule, input mapping, workflow incentives, staffing and monitoring calculation. Document how the cause could create the observed breach.
A threshold breach is evidence of a problem, not proof of one specific root cause.
CHAPTER 8 OF 12 · REVALIDATE
8. Revalidate the failed obligation
Build a test set representing the affected context, known misses, ordinary cases and plausible changes. Measure necessary protection, coverage and workload.
Do not validate only the technical fix. Confirm that the complete operational route now produces review and provenance evidence.
Evidence check
Verify that the control still reaches the intended decision under a changed but plausible context.
CHAPTER 9 OF 12 · REVALIDATE
9. Reconcile releases from the exposure window
Re-review cases whose outcome could change without the breached safeguard. Preserve stable releases with the evidence that supports them.
This targeted repair respects both archaeological integrity and proportionality.
| Case | Reason |
|---|---|
| Known miss | Confirm repair |
| Ordinary case | Check regression |
| Boundary case | Test scope |
| Changed context | Test transfer |
CHAPTER 10 OF 12 · DECIDE THE FUTURE
10. Choose restore, narrow, renew or revoke
Governance compares revalidation results, residual uncertainty and operating capacity. Record the version, scope, conditions, effective time and dissent.
Restoration is a new accountable decision, not an automatic consequence of fixing the alert.
CHAPTER 11 OF 12 · DECIDE THE FUTURE
11. State limits of revalidation
A repaired test set may miss rare site types, strategic workarounds or combined failures. A quiet monitoring period does not establish universal reliability.
Keep escalation and later review routes active, especially where archaeological evidence cannot be recreated.
Evidence check
Confirm that versions, notices and unresolved limits remain visible in the final record.
CHAPTER 12 OF 12 · DECIDE THE FUTURE
12. Audit trigger response end to end
Track detection delay, suspension scope, cases in the window, cause, validation evidence, governance decision and recurrence.
The gold standard gives monitoring teeth: a trigger protects the release gate, and authority returns only through evidence.
| Choice | Meaning |
|---|---|
| Restore | Evidence sufficient |
| Narrow | Reduce scope |
| Renew conditions | More evidence needed |
| Revoke | Redesign not acceptable |
Frequently asked questions
Does every warning require suspension?
No. Use the response level defined for that trigger; necessary-control failures warrant stronger action than early warnings.
Can the team wait for root-cause proof before acting?
Apply the predefined provisional control first when consequences are material, then verify and diagnose.
Must all earlier releases be repeated?
No. Re-review only cases for which the breached safeguard could be material, preserving stable outcomes with evidence.
Who ends the suspension?
The authority named in the approval record, using current revalidation evidence and an explicit scope decision.
Continue through verified routes
Core routes
Archaeology subject root · Gold Standard of Education · Gold Standard directory · How X Works hub
Specialist routes
- Conditional approval, residual obligations and post-acceptance monitoring
- Control redesign validation disputes, competing success criteria and governance acceptance
- Regression root-cause disputes, independent review and control redesign
- Remediation regression, control-drift signals and re-escalation