GCRI · Registry
Correction and versioning
Every published record will one day need to change: a typing error, a revised figure, a withdrawn claim, a new release of a dataset. What matters is whether the change is visible, linked and propagated, or silent. This page sets out how corrections, retractions and new versions are recorded, why superseded assertions keep circulating after they are corrected, and what editors and stewards need to run a correction route that readers can follow.
How the work runs.
Interoperability console
Correction and versioning · Crosswalk · Profile v3
Field mappings
| identifier | id | copy | Pass |
| evidence_label | quality.label | enum map | Pass |
| licence | rights.licence | SPDX id | Ambiguous |
| version | version | semver | Pass |
Conformance
Fig.Interoperability console: example names and illustrative values, not a record or a live system.
Correction and versioning: how the risk develops.
Superseded assertions and correction routes. The dossier below sets out how the risk develops, what is measured, where the data come from, how it is modelled and who decides.
Silent edits
A value is overwritten in place without a version change or a notice. Anyone who downloaded or cited the earlier state now holds a record that no longer matches the source, and nothing tells them so.
Corrections stop at the source
A notice is issued where the record lives, but copies in aggregators, indexes, reference managers and derived datasets are not updated. The superseded assertion keeps circulating downstream and is cited as current.
Version forks
Preprint, accepted manuscript and version of record, or successive dataset releases, are registered without relations between them. Citations split across versions, and a correction made to one never reaches the others.
Reliance compounds
Corrected or retracted findings feed systematic reviews, models, guidelines and funding cases. Each reuse builds on the flawed assertion, so the cost of a late correction grows with every layer that came to depend on it.
How correction and versioning is modelled, measured and tested.
The model, its assumptions, its data and its uncertainty, in the form a scientist or engineer would check. Structures follow published methods. The values drawn are illustrative, not findings.
Datasets and provenance
The data classes behind correction and versioning and what is recorded about each before it is used.
| Data class | Provenance recorded | Resolution stated | Licence checked |
|---|---|---|---|
| Satellite imagery, optical and radar | sensor, processing level, acquisition date | pixel size and revisit | per scene |
| Reanalysis and climate model output | model, experiment, member, version | grid and time step | per release |
| In-situ sensors and gauges | station, instrument, calibration record | point and sampling rate | per network |
| Population and asset exposure | method, vintage, input sources | cell size and unit | per release |
| Boundaries and reference geography | producer, edition, date | scale | per edition |
- Lineage
- W3C PROV record from source to figure
- Principles
- FAIR: findable, accessible, interoperable, reusable
- Gaps
- shown as missing, never filled silently
Corrections in one domain must reach every product that used the data.
- WaterIn practiceA revised rainfall series changes drought indices.What a reader can checkProducts re-run after revision
- HealthIn practiceRevised case counts change outbreak analyses.What a reader can checkVersion history
- FoodIn practiceRevised survey weights change estimates.What a reader can checkCorrection notes
Corrections are published with every edition of Reports under COPE guidance. Versions would follow the identifier rules of NS-1 in the proposed Nexus Standards.
- Error reports with evidence.
- Source revisions that need propagating.
What is measured, and how well.
Each indicator carries its unit, its cadence, the kind of source behind it and the evidence label a value would carry. No values are shown: a programme supplies them under its own authority.
| Indicator | Unit | Cadence | Source type | Label |
|---|---|---|---|---|
| Corrections issued | count by type per period | monthly | Correction trail | OBSERVED |
| Time from report to correction notice | days | per case | Case records | OBSERVED |
| Content changes without a version increment | count | weekly | Content hash comparison | OBSERVED |
| Version chain completeness | share of versions linked to a predecessor | monthly | Relation metadata | OBSERVED |
| Retracted items linked to their notice | share of retracted items | monthly | Crossmark and retraction records | OBSERVED |
| Superseded records still cited as current | count of citing works | quarterly | Citation graph | ESTIMATED |
| Corrections reflected by downstream aggregators | share of sampled aggregators | quarterly | Aggregator sampling | ESTIMATED |
Where the evidence comes from.
Named public datasets and institutional records, each with its licence class. Sources by agreement are used only under their terms and never republished.
| Source | What it gives | Licence class |
|---|---|---|
| Retraction Watch Database | Records of retractions, corrections and expressions of concern across publishers | Open, distributed by Crossref |
| Crossmark | Publisher-deposited update status linking corrections and retractions to the original work | Open metadata via Crossref |
| Crossref and DataCite relation metadata | Relations such as is new version of, is previous version of and is obsoleted by | Open metadata; CC0 for DataCite |
| PubPeer | Post-publication comments that often surface errors before formal notices | Public comments under site terms |
| Internet Archive Wayback Machine via Memento | Time-stamped captures showing what a page said at a given moment | Free access under archive terms |
| Software Heritage | Archived source code with intrinsic identifiers for every version | Open archive, terms per content licence |
Who decides, and what the evidence serves.
Registry prepares evidence. The decisions below are taken by the authorities named, under their own rules, and never by the platform.
| Who decides | The question | Evidence this page serves |
|---|---|---|
| Journal editors and publishers | Whether an error warrants a correction, an expression of concern or a retraction | COPE guidance, the nature of the error, and the citing and derived works that relied on it |
| Data stewards and repository managers | Whether a change is a correction to the same version or a new version | The versioning policy, the extent of the change and the identifiers already cited |
| Research integrity offices | Whether a pattern of corrections signals possible misconduct needing investigation | Correction history across records by the same contributors, with dates and stated reasons |
| Systematic reviewers and guideline panels | Whether included evidence must be re-examined after a correction | Retraction and correction status of every included item at the review date |
Outputs this topic produces
- Correction trails showing each change, its type, author, date and reason
- Version chains linking every release to its predecessor and successor
- Notices for superseded assertions kept visible beside the corrected record
- A published correction route stating whom to contact and how a report is handled
What experts ask about corrections.
- When is a change a correction, an update or a replacement, and how should each be labelled?
- How can a correction reach citations, copies and derived records?
- What must be kept of a superseded version so that past decisions can be understood?
- Who may initiate, review and approve a correction?
- How is a correction itself corrected?
Why uncorrected errors persist.
Studies of retracted research have long shown that flawed papers continue to be cited after retraction, largely because readers meet them through copies, summaries and reference lists that never learn of the change. The same happens with datasets, dashboards and programme descriptions. Silent edits make it worse: when a page changes without notice, readers cannot know that what they cited is no longer what the page says.
Good practice therefore treats records as append-only. A correction creates a new version that points to the one it supersedes, states what changed and why, and is announced through a notice that downstream systems can detect. Publishing communities have developed shared guidance for retractions and corrections, and versioned identifiers let citations point to a specific state. The Registry’s procedure is described on corrections and version history.
How versioning works in practice.
- 01
Version chains
Each version records its predecessor and successor, so the full history can be followed.
- 02
Typed changes
A change is labelled as a correction, an update, a supersession or a withdrawal, each with a reason.
- 03
Change notices
Machine-readable notices let copies elsewhere update, a feature discussed under registry interoperability.
- 04
Limits
No registry can recall copies it does not control. Notices help only systems that read them, and some errors are found too late to change decisions already taken. Correcting a record does not correct the reasoning others built on it.
How it is studied, and what the methods cannot say.
- 01Data engineering, interoperability and stewardshipSchema-validated ingestion · Persistent identifier reconciliation · Lineage capture in pipelinesApplied here
- 02Cybersecurity and cryptographyTamper-evident record history · Signed exports and releases · Threat modelling of edit pathsApplied here
- 03Digital identity and public digital systemsIssuer authentication and authority · Contributor disambiguation by the personApplied here
Where this page’s references lead.
The bodies and documents this page is read against. Status and editions change; each link goes to the issuing body.
Questions and answers.
What is the difference between a correction and a new version?
Are retracted records deleted?
How can I tell if a record I cited has changed?
Why do retracted findings keep being cited?
Will an earlier version disappear after a correction?
Can anyone request a correction?
Is a withdrawal the same as a correction?
Contribute evidence from this page.
Reports is the Consortium’s quarterly publication. What you know about this subject can become evidence, a case or a correction in the next edition, credited to you and published under CC BY 4.0 where it is your own work.
- EmailGateway GW-GCRISend it from this pageEvidence, a correction, a case, a method, a review, a translation or profile input. It arrives tagged with this page, so the editor who owns it receives it.
- ZenodoOpen depositDeposit it and submit to the communityPublish your own record on Zenodo with your ORCID iD, then submit it to the Reports community there. Curators review it; the DOI stays yours.
- ReportsGCRISee every way to contributeThe contribution types, how each is reviewed, what credit you receive, the licence and what is never accepted.
Never send personal data, sensitive infrastructure detail or anything you have no right to share. A contribution is reviewed before use; sending it confers no authorship or editorial rights, and sponsorship never buys a finding.
Bring a method, a dataset or a question.
Information and enquiries are open today. An enquiry receives a reference and names the next step; applications, payments, compute allocations and registrations open only when their team, terms and review capacity exist.
Legal documents.
Maps, data and figures on this site are shown for context and general information. They are not advice, forecasts or warnings; the limits of the information are set out in the Disclaimers and Data Notices.