The Gold Standard · Archaeology
Make the decision path visible
After repeated release-gate overrides, a team may add reviewers, move validation earlier or create a bounded provisional mode. Those are sensible actions, but their completion does not prove that the cause of bypassing has changed or that archaeological outputs are better protected.
The gold standard starts with a testable causal claim, compares rate and consequence against a meaningful baseline, watches for hidden workarounds and uses predefined repeat-use triggers to move unresolved risk into governance.
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 · FRAME THE TEST
1. Turn remediation into a falsifiable claim
State what condition the repair is meant to change, by how much, over what period and for which releases. For example: an earlier validation window should reduce weather-related overrides without increasing provisional publication.
A statement such as improve compliance cannot be tested. Name the mechanism, expected signal and evidence that would count against success.
CHAPTER 2 OF 12 · FRAME THE TEST
2. Choose a baseline that matches operating conditions
Compare override events with the number and type of eligible releases, field seasons, instruments, urgency classes and review capacity. Preserve the pre-repair window and document other changes that can affect the rate.
A raw count can fall because fewer projects ran. A rate can also mislead if the remaining projects are simpler. Baselines need both denominator and context.
Evidence check
Check whether the scope, owner and decision consequence are explicit before continuing.
CHAPTER 3 OF 12 · FRAME THE TEST
3. Measure consequence as well as frequency
Track scope breaches, unreconciled outputs, delayed after-action reviews, publication exposure, control debt and severity avoided. Include time from predictable need to request.
A repair can reduce minor exceptions while leaving one high-consequence path untouched. Effectiveness should reflect the control objective, not a single attractive number.
| Part | Question |
|---|---|
| Cause | What condition changes? |
| Action | What repair acts on it? |
| Signal | What should improve? |
| Refutation | What would disprove success? |
CHAPTER 4 OF 12 · COLLECT EVIDENCE
4. Use independent and operational evidence
Combine override logs with model configuration, release-gate events, publication records, repository state, staffing schedules and reviewer observations. Ask someone outside the requesting chain to test a sample.
Self-reported improvement is useful but incomplete when the same team benefits from showing that the repair worked. Independent reconstruction tests whether recorded practice matches actual release behaviour.
CHAPTER 5 OF 12 · COLLECT EVIDENCE
5. Look for displacement into unlogged workarounds
Search for manual exports, renamed provisional products, direct file sharing, copied scripts and decisions made before the formal gate. Compare logged activity with repository and publication traces.
A lower override count is a bad signal if bypassing simply became invisible. Effectiveness includes the integrity of measurement itself.
Evidence check
Ask a second reviewer to reconstruct the example from the preserved evidence and stated assumptions.
CHAPTER 6 OF 12 · COLLECT EVIDENCE
6. Worked example: early validation lowers use but not scope breaches
A coastal programme moves bridge-model validation four weeks earlier after three recurring seasonal overrides. The next season records one override instead of three, yet two teams release maps beyond the approved provisional scope because an old template still labels them publication-ready.
The frequency measure improved, but the key consequence did not. Governance keeps the early window, withdraws the misleading template, adds a manifest check and escalates template ownership. Changed-context check: if override use falls only because the field season ends before the new process operates, the observation supplies no evidence that remediation caused the change.
| Source | Use |
|---|---|
| Override log | Rate and reason |
| Gate event | Actual release path |
| Repository trace | Hidden workaround |
| Independent sample | Practice verification |
CHAPTER 7 OF 12 · ESCALATE WISELY
7. Define repeat-use triggers in advance
Set triggers for repeated reason codes, same requester, widened scope, late review, access that survives expiry and outputs produced before reconciliation. Specify whether each trigger requires investigation, authority suspension or governance review.
Predefined triggers reduce case-by-case leniency and make escalation explainable. They should still allow documented judgment when an independent external emergency genuinely recurs.
CHAPTER 8 OF 12 · ESCALATE WISELY
8. Escalate the failed control, not merely the operator
Bring evidence about causal conditions, capacity, incentives, technical permissions and prior unfinished obligations. Identify which governance body can fund, redesign or enforce the necessary boundary.
Blaming one coordinator can hide a queue, procurement cycle or publication incentive that will recreate the bypass under a new name.
Evidence check
Verify that the control still reaches the intended decision under a changed but plausible context.
CHAPTER 9 OF 12 · ESCALATE WISELY
9. Revise the remediation and prediction
If results contradict the original claim, preserve the finding, update the causal model and select a stronger action. State what remains useful in the first repair and what must change.
Treating a partial success as total failure wastes evidence. Treating it as closure leaves the consequence exposed. Separate mechanisms and update each claim.
| Signal | Response |
|---|---|
| Same reason returns | Root-cause review |
| Scope widens | Suspend authority |
| Review debt grows | Governance escalation |
| Access persists | Technical containment |
CHAPTER 10 OF 12 · SUSTAIN THE REPAIR
10. Set decision boundaries for continuing, scaling or stopping
Continue when signals are directionally credible but observation is incomplete; scale only after the repair transfers to comparable teams; stop or replace it when consequence persists, workarounds grow or measurement cannot be trusted.
The boundary should protect archaeological claims while avoiding endless pilot status. Record who decides and what evidence they used.
CHAPTER 11 OF 12 · SUSTAIN THE REPAIR
11. Run an effectiveness transfer test
Apply the repair to a new field team, different instrument, contractor workflow and unforeseen rescue context. Keep the same core measures while recording legitimate local adaptations.
A process that succeeds only with its designer present is not yet a durable control. Transfer shows whether the causal mechanism survives changes in people and setting.
Evidence check
Confirm that versions, notices and unresolved limits remain visible in the final record.
CHAPTER 12 OF 12 · SUSTAIN THE REPAIR
12. Audit remediation evidence and governance response
Track baseline quality, observation coverage, independent checks, trigger activations, escalation decisions, time to revised action, residual consequences and relapse. Keep dissent where operational and assurance reviewers interpret the evidence differently.
The gold standard calls remediation effective only when the evidence links the repair to safer ordinary work and governance responds when that link fails.
| Measure | Caution |
|---|---|
| Count falls | Check denominator |
| Rate falls | Check consequence |
| No logs | Check displacement |
| One team succeeds | Test transfer |
Frequently asked questions
Does a lower override count prove remediation worked?
No. Check release volume, operating context, consequence and hidden workarounds before attributing the change to the repair.
When should repeat use trigger governance escalation?
When a predefined pattern shows the cause or consequence persists beyond the authority of case-level reviewers to repair.
Can a remediation be partly effective?
Yes. Preserve the mechanism that worked, identify the unchanged exposure and revise the causal claim and remaining action.
Why test a new team or instrument?
Transfer testing shows whether the repair is durable rather than dependent on one coordinator or familiar setting.
Continue through verified routes
Core routes
Archaeology subject root · Gold Standard of Education · Gold Standard directory · How X Works hub
Specialist routes
- Override recurrence, exception normalisation and restoring release-gate discipline
- Release-gate override logs, emergency exceptions and independent after-action review
- Interim-restriction drift, workarounds and release-gate enforcement
- Revalidation backlogs, interim restrictions and coverage-restoration priorities