Rectification, retraction, and doctrinal supersession
As a doctrinal corpus begins publishing not only principles, but also limit cases, comparative dossiers, formalized test cases, and published probative surfaces, a new requirement appears: it is no longer enough to publish. One must also know how to modify a published object without falsifying the memory of its publication.
A doctrinal site is not a social feed. Every correction acts on a regime of authority. A silent change in a doctrine page, a baseline, a probative annex, or a comparative dossier does not merely alter wording. It alters what may later be cited, enforced, compared, archived, or reused as a reference state.
This page extends version power, memory governance, public benchmarks, and applied observability. It states a simple discipline: in a reference corpus, one must distinguish what is corrected, what is withdrawn, what is replaced, and what must remain visible as an earlier state.
1. Why revision discipline becomes doctrinal
When a corpus publishes only general theses, its revisions remain relatively discreet. But as soon as it publishes protocols, snapshots, comparative cases, or test objects, every revision acts on a larger chain: citation, proof, comparison, memory, and third-party interpretation.
Without revision discipline, three drifts appear quickly.
The first is rewriting the past as if the earlier state had never existed. The second is leaving online an object that can no longer be sustained while still treating it as reference. The third is replacing an object without declaring the relation that binds it to its successor.
In all three cases, the problem is not merely editorial. It is doctrinal: the chain linking a statement to its state, date, perimeter, and authority level becomes blurred. A doctrine that accepts such opacity ends up asking for trust where it should provide an interpretation trace.
2. Four gestures that should no longer be conflated
a) Correction
Correction repairs a local error without changing the main regime of the object. It may concern a typo, a broken reference, a misreported date, a mislabeled example, or a minor precision. The thesis, protocol, and hierarchy remain the same.
b) Rectification
Rectification occurs when a published object remains broadly valid, but its wording, perimeter, protocol, or justificatory chain had to be made more exact. It is already more than a simple editorial adjustment. It declares that an earlier state was insufficiently accurate in doctrinal terms.
c) Retraction
Retraction removes an object from reference status. It becomes necessary when the object is no longer doctrinally sustainable: corpus too weak, source hierarchy poorly established, a major contradiction ignored, cherry-picking, archive misread, evidence not reconstructible, or observation confused with attestation.
d) Supersession
Supersession replaces an object with another that is more recent, more bounded, more complete, or more authoritative, while keeping the succession relation visible. It does not deny the existence of the earlier state. It requalifies it historically.
To conflate these gestures is to conflate a local error, an amendment, a withdrawal of authority, and a replacement of version.
3. Why silent rewriting is doctrinally weak
Silent rewriting often feels attractive. It gives the impression of a cleaner, tighter, more coherent site. In reality, it destroys precisely what makes a corpus auditable.
When a benchmark changes without saying so, comparisons become dubious. When a comparative dossier is cleaned up without declaration, contradictions seem to have vanished on their own. When a doctrine page is quietly narrowed, the actual scope of its first formulation becomes impossible to reconstruct.
The damage is not merely historical. It is interpretive. A third party can no longer tell which formulation governed when. A reader can no longer tell what was assumed, corrected, or abandoned. A machine encountering several traces of the same object may then reconstruct a composite authority, blending the old state, the new state, and their external reprises.
In other words, silent rewriting does not protect doctrine. It exposes it to a harder-to-govern interpretive fossilization, because the past continues to circulate while the site has failed to maintain an explicit relation to it.
4. Minimum conditions for a publishable rectification
A doctrinally clean rectification does not require infinite bureaucracy. It requires at least that the modified object make five things visible.
- What changed: wording, perimeter, source, protocol, example, or status.
- Why it changed: factual error, clarified boundary, revealed contradiction, improved hierarchy, or corrected protocol.
- What remains valid: so that rectification is not read as total cancellation.
- From when the new version prevails: date, state, or version relation.
- What effect it has on derivative objects: benchmark, annex, comparative dossier, baseline, translation, or third-party citation.
Those conditions are not ornamental. They preserve the relation between the present object and its previous state. Without them, rectification remains a surface improvement. With them, it becomes a governed operation of memory and authority.
5. When retraction becomes necessary
Retraction should not be understood as a spectacular admission. In a doctrinal corpus, it is often the most sober gesture.
It becomes necessary when an object can no longer be maintained without producing more confusion than clarity. For example:
- the corpus mobilized was too weak to support the published generalization;
- a source presented as authoritative was not;
- a case was not reconstructible from the pieces made public;
- a decisive contradiction had been omitted;
- the object mixed historical state and current state;
- the publication turned indicative evidence into strong proof.
In such cases, continuing to “correct a little” an unsustainable object often worsens the problem. It is better to remove its reference status, keep its existence as a trace, and possibly reconnect it to a more sober formulation or to a successor object.
6. Supersession: replacing without erasing
Supersession becomes indispensable as soon as a corpus works with versions, protocols, and observable objects. A comparative dossier may be replaced by a better-bounded version. A benchmark may be overtaken by a tighter protocol. A doctrine page may be restated in a more stable form. In all those cases, the new document should not magically absorb the old one.
Supersession exists precisely to keep three things together:
- the present validity of the new object;
- the historical existence of the old one;
- the legibility of the passage from one to the other.
It therefore presupposes explicit links, dating, justification, and, where necessary, archive continuity. Without that, the successor appears more stable than it really is, because the cost of its replacement has been erased.
This is where the page meets archives, residual temporalities, and surviving authority. A superseded object does not disappear from the world. It often continues to act through reprises, screenshots, citations, or delayed indexing. Declaring it superseded does not suppress it; it only makes its survival more governable.
7. Scope and limit
This page defines neither a legal correction policy, nor a universal withdrawal procedure, nor an industrial versioning system. It fixes a narrower and more important requirement: a doctrinal corpus that publishes cases, comparisons, and probative surfaces must know how to correct without masking, withdraw without denying, and replace without erasing.
A doctrine that cannot govern the post-publication life of its own objects eventually confuses present coherence with organized amnesia.