Revision Diff Review¶
Change audit — instantiates Metanarrative Coherence and Internal Consistency Check
Compares successive versions of a narrative against the repair each revision was supposed to make, catching the edit that renamed a contradiction, quietly absorbed a counterexample, or reworded a tension instead of resolving it.
A coherence audit produces a repair plan; the plan produces edits; and the edits produce a new narrative that everyone assumes is fixed. Revision Diff Review is the mechanism that checks that assumption. It sets each version of the narrative against the previous one and against the reconciliation plan the revision was meant to execute, and asks, contradiction by contradiction: was this actually repaired, or merely reworded? Its defining move — the one no other sibling makes — is that it works across time and versions, auditing the delta between drafts rather than the content of any single draft. It exists because the archetype's subtlest failure is "revision without repair": the wording changes, the contradiction is renamed or relocated, and the structural incompatibility survives under new language while everyone believes it was handled. The diff review is what makes a repair falsifiable — a claimed fix has to show up as a real structural change, not just a cosmetic one.
Example¶
A contested Wikipedia article on a disputed historical event carries a flagged internal contradiction: the introduction states the event's death toll as settled at one figure, while the "Casualties" section presents three conflicting scholarly estimates and calls the number unresolved. An editor logs the contradiction and a revision plan: reconcile the intro with the body by presenting the intro figure as one contested estimate among several. A week and forty edits later, someone runs a revision diff review before closing the dispute.
Diffing the versions against the plan, they find the repair was faked. The intro's number was changed — but only to a different single figure, still stated as settled, and the "Casualties" section's three estimates are untouched. The contradiction wasn't reconciled; it was relocated to a new number. Worse, a second diff shows an inconvenient estimate silently deleted from the body between versions, quietly shrinking the disagreement the plan was supposed to surface rather than hide. The review's output is not a verdict on the history; it is a precise finding — the plan called for reconciliation, the diff delivered a substitution and a deletion — that reopens the edit rather than letting a cosmetic change close it.
How it works¶
- Anchor to the plan. Start from the reconciliation-or-revision plan: each entry names a specific contradiction and the structural change meant to fix it, giving the review something falsifiable to check against.
- Diff version against version. Compare the before and after narratives at the claim level, not the word level, isolating exactly what changed and — just as important — what did not.
- Classify each change. For every planned repair, decide whether the diff shows a genuine structural resolution, a mere rewording, a relocation of the contradiction, or a silent deletion that hides rather than resolves.
- Flag the unaccounted deltas. Surface changes that no plan entry authorized (a claim quietly dropped, a scope silently narrowed) and repairs the plan called for that never appear in the diff at all.
Tuning parameters¶
- Diff granularity — whether changes are compared at the word, sentence, or claim level. Claim-level diffing catches rewording-that-isn't-repair but takes interpretation; word-level diffing is mechanical but drowns in cosmetic noise.
- Plan-adherence strictness — how exactly a diff must match its plan entry to count as executed. Strict matching catches faked repairs but can reject a defensible fix that took a different route; loose matching accepts "close enough" and lets substitutions pass.
- Unauthorized-change sensitivity — how aggressively edits with no plan entry are flagged. High sensitivity catches silent deletions and quiet scope-narrowing but generates noise on benign copyedits; low sensitivity lets a load-bearing deletion slip through as housekeeping.
- Version depth — whether the review compares only the latest pair or walks the full revision history. Full-history review catches a contradiction that was removed and later reintroduced; pairwise review is cheaper but has no memory.
- Repair bar — how much structural change counts as a genuine resolution versus a relabeling. A high bar prevents cosmetic fixes from closing a dispute; too high a bar demands rewrites where a real clarification would do.
When it helps, and when it misleads¶
Its strength is memory and accountability: it is the only mechanism here that can tell whether the audit's own repairs actually landed, and it makes "we fixed that" a checkable claim rather than a reassurance. By diffing against the plan it catches the specific dishonesty the archetype names — a contradiction renamed rather than resolved — and it catches the quieter one, a counterexample or an inconvenient figure deleted between versions so the story only looks consistent because the disconfirming material is gone. This is the guard against the narrative "memory hole," where the past is edited so the contradiction never officially existed.[n1]
Its failure mode is confusing change with repair: a diff shows that text moved, not that a contradiction was resolved, so a review run mechanically can bless a rewording as a fix or, conversely, flag a genuine structural repair as "just edits" because it doesn't match the plan's exact wording. The classic misuse is weaponizing the diff — treating every edit as a coverup — which paralyzes ordinary revision. The discipline that keeps it honest is to anchor every judgment to a named contradiction and its planned repair, and to require that a claimed fix show a structural change at the claim level, so "repaired," "reworded," and "deleted" stay three different verdicts.
How it implements the components¶
versioned_narrative_record— its working substrate: the ordered record of narrative versions whose deltas it audits, and which it keeps so a later reviewer can see what changed when.reconciliation_or_revision_plan— it operationalizes the plan, turning each planned repair into a checkable diff expectation and verifying the version-to-version change actually discharged it.
It does not detect contradictions within a single version — the pairwise contradiction_and_tension_log those repairs answer to is Contradiction Scan's — and it does not track a single term's meaning across the text, which (whether within or across versions) is the terminology_glossary of Term-Stability Review.
Related¶
- Instantiates: Metanarrative Coherence and Internal Consistency Check — it verifies that the audit's repairs actually changed the narrative's structure, not just its wording.
- Consumes: Contradiction Scan supplies the logged contradictions whose planned repairs this review checks were carried out.
- Sibling mechanisms: Causal-Temporal Trace · Claim-Lattice Mapping · Contradiction Scan · Evidence-to-Claim Traceability · Exception Classification · Red-Team Coherence Review · Term-Stability Review · Counter-Narrative Probe · Scope-Boundary Stress Test
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Revision Diff Review operates as a bounded evaluation of existing evidence or work that produces a finding or disposition because it compares successive versions of a narrative against the repair each revision was supposed to make, catching the edit that renamed a contradiction, quietly absorbed a counterexample, or reworded a tension instead of resolving it.
Independent corroboration: The frozen evidence defines Revision Diff Review as 'Compares successive versions of a narrative against the repair each revision was supposed to make, catching the edit that renamed a contradiction, quietly absorbed a counterexample, or reworded a tension instead of resolving it', so its operative form is Assessment, Review & Assurance.
Nearest alternative: Analysis, Modeling & Optimization — Revision Diff Review includes features of an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution, but its defining operation is a bounded evaluation of existing evidence or work that produces a finding or disposition.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Literature & Literary Theory
Origin pattern: Historically ambiguous
Present-day reach: Multi-domain
Rationale: Comparing narrative revisions for whether they resolve contradiction is rooted in textual criticism and literary editing.
Related originating lineages:
- Computer Science & Software Engineering — Computer science and software-engineering practice supplies a parallel or contributing lineage for the mechanism's defining operation: compares successive versions of a narrative against the repair each revision was supposed to make, catching the edit that renamed a contradiction, quietly absorbed a counterexample,….
- History & Historiography — Version and source criticism independently trace narrative change over time.
- Rhetoric — Argument revision materially tests whether counterexamples and tensions are substantively answered.
Review resolution: Both blind reviewers agree that literature_literary_theory is the primary historical origin. Explicit reconciliation of reported ambiguity, alternate origin disagreement, origin mode disagreement starts from reviewer_a’s mechanism-specific evidence: Comparing narrative revisions for whether they resolve contradiction is rooted in textual criticism and literary editing. Reviewer A proposed alternates=history_historiography, rhetoric, origin_mode=cross_disciplinary_synthesis, domain_reach=multi_domain, and encyclopedia_synthesis=true; reviewer B proposed alternates=computer_science, history_historiography, origin_mode=historically_ambiguous, domain_reach=multi_domain, and encyclopedia_synthesis=true. The final record retains every independently supported alternate from either review (history_historiography, rhetoric, computer_science) without an arbitrary cap, selects origin_mode=historically_ambiguous to represent the combined lineage evidence, and keeps domain_reach=multi_domain and encyclopedia_synthesis=true from the more mechanism-specific assessment. Present-day transfer is recorded as reach and is not treated as proof of historical origin.
Attribution caveat: Version comparison comes from software practice, while narrative contradiction review belongs to literary and historiographic criticism.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; medium confidence.
Notes¶
[n1] In Orwell's Nineteen Eighty-Four, the "memory hole" is the chute into which records of the inconvenient past are dropped and destroyed so the official narrative can claim it was always consistent. Revision Diff Review is the structural defense against the milder, everyday version of that move: keeping the versioned record means a deletion or a relocation between drafts leaves a trace that can be found and questioned. ↩