The Gold Standard · Archaeology
Make the decision path visible
A repository can publish a corrected version while the scholarly route still leads readers backwards. Search indexes may retain the recalled title, reference managers may resolve a generic identifier to an older landing page, and reports may repeat claims that were valid only under the superseded dataset. The correction exists, but its meaning has not propagated.
The gold standard follows correction signals from the versioned object into catalogues, indexes, citations and claims. It measures lag without confusing delay with refusal, preserves transparent lineage and reviews residual claims according to their real dependence on the recalled evidence.
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 · DEFINE PROPAGATION AND RESIDUE
1. Define correction propagation
Specify the corrected object, superseded versions, related identifiers, notices and the catalogues, indexes, reports and datasets expected to carry the correction.
Propagation is a chain of visible relationships, not a promise that every copy disappears.
CHAPTER 2 OF 12 · DEFINE PROPAGATION AND RESIDUE
2. Separate indexing lag from citation residue
Indexing lag is a delayed metadata update; citation residue is a scholarly route or claim that still presents the superseded evidence as current.
The distinction determines whether to wait, re-submit metadata, correct a citation or reopen interpretation.
Evidence check
Check whether the scope, owner and decision consequence are explicit before continuing.
CHAPTER 3 OF 12 · DEFINE PROPAGATION AND RESIDUE
3. Set residual-claim review criteria
Prioritise claims whose sample, chronology, provenance, classification or uncertainty materially depends on the recalled version.
A citation can be stale while its claim remains valid; another can look current while still reproducing the withdrawn inference.
| Route | Evidence |
|---|---|
| Repository | Version relation |
| Resolver | Target tested |
| Index | Refresh date |
| Publication | Correction status |
CHAPTER 4 OF 12 · TRACE LAG IN CONTEXT
4. Map resolvers, mirrors and discovery systems
Test version-specific identifiers, canonical records, repository mirrors, library catalogues, search indexes and common citation exports. Record what a reader reaches and when it was last refreshed.
The practical question is which object ordinary discovery routes present, not only what the source repository stores.
CHAPTER 5 OF 12 · TRACE LAG IN CONTEXT
5. Measure propagation with time-stamped evidence
Capture metadata snapshots, HTTP targets, version labels, relation types and correction notices at defined intervals.
A time series distinguishes expected refresh delay from a broken or unowned correction route.
Evidence check
Ask a second reviewer to reconstruct the example from the preserved evidence and stated assumptions.
CHAPTER 6 OF 12 · TRACE LAG IN CONTEXT
6. Worked example: a stale index keeps a recalled claim visible
A corrected faunal dataset receives a new version and an obsolescence link. The repository landing page is accurate, but a disciplinary index still exposes the earlier abstract, and a regional synthesis repeats the recalled abundance estimate.
The team re-submits metadata, links the synthesis to the correction and reviews whether its conclusion survives the revised counts. Changed-context check: an older citation may remain historically correct when it explicitly identifies the superseded version and discusses how the interpretation changed.
| Finding | Meaning |
|---|---|
| Stale snippet | Possible lag |
| Old default download | Unsafe route |
| Explicit historical citation | Transparent lineage |
| Repeated old claim | Review needed |
CHAPTER 7 OF 12 · REPAIR CITATIONS AND CLAIMS
7. Contain misleading resolution without hiding lineage
Add visible correction notices, disable unsafe default downloads where governance permits, and point generic routes to a version-aware landing page. Preserve the earlier record and reason for correction.
Readers need both a safe current route and an intelligible history of change.
CHAPTER 8 OF 12 · REPAIR CITATIONS AND CLAIMS
8. Repair citations at the right level
Update repository metadata and documentation, notify maintainers of consequential indexes, and issue corrections or addenda for publications whose claims materially change.
Not every mention needs the same intervention; the repair should match the claim consequence.
Evidence check
Verify that the control still reaches the intended decision under a changed but plausible context.
CHAPTER 9 OF 12 · REPAIR CITATIONS AND CLAIMS
9. Review residual claims from evidence to inference
Reconstruct the claim using the cited version, corrected version and stated analytical steps. Classify it as unchanged, qualified, reversed or unresolved.
The review should make reasoning visible instead of treating a corrected identifier as automatic proof that every conclusion failed.
| Question | Record |
|---|---|
| Evidence version | Exact identifier |
| Dependency | Material or incidental |
| Reanalysis | Changed result |
| Decision | Retain, qualify or correct |
CHAPTER 10 OF 12 · CLOSE WITH LIVING EVIDENCE
10. Verify correction reach and closure
Define critical discovery routes and consequential claims, then confirm that each carries the correction or a documented unresolved status.
Closure can be bounded even when low-consequence caches remain, provided residual risk and owners are explicit.
CHAPTER 11 OF 12 · CLOSE WITH LIVING EVIDENCE
11. State limits and common misconceptions
Indexes refresh on different schedules, private libraries are not observable and persistent identifiers do not themselves force readers to use the current version. A removed file may also survive in lawful archives.
Record unverified routes and avoid claiming universal erasure or universal uptake.
Evidence check
Confirm that versions, notices and unresolved limits remain visible in the final record.
CHAPTER 12 OF 12 · CLOSE WITH LIVING EVIDENCE
12. Monitor versions, citations and claims
Watch relation metadata, resolver targets, index refresh, new citations and reuse requests. Reopen review when a residual claim becomes consequential or a new mirror appears.
The gold standard keeps correction propagation alive for as long as the recalled evidence can still shape interpretation.
| Critical route | Pass condition |
|---|---|
| Canonical page | Correction visible |
| Version DOI | Relations reciprocal |
| Major index | Current metadata |
| Material claim | Disposition recorded |
Frequently asked questions
What is citation-correction propagation drift?
It is divergence between the authoritative corrected record and the versions, metadata or claims still presented through downstream scholarly routes.
Does a stale index prove a correction failed?
No. It may reflect normal refresh delay, but the lag should be measured and escalated when it keeps consequential evidence misleading.
Must every old citation be replaced?
No. Historical citations can remain when they identify the version accurately and do not present superseded evidence as current.
When is residual-claim review complete?
When material claims have been reconstructed against the corrected evidence and each has a recorded, visible disposition.
Continue through verified routes
Core routes
Archaeology subject root · Gold Standard of Education · Gold Standard directory · How X Works hub
Specialist routes
- Derived-dataset recall closure, citation residue and repository-version reconciliation
- Export provenance loss, derived-dataset recall and reuse remediation
- Evidence-only archive governance, unapproved reuse and redesign retirement
- Re-revocation appeal, shadow-evidence preservation and redesign retirement