{"actors":["General editor of a living digital critical edition","Section editors responsible for interpretive notes","Textual scholars and annotators who revise the edition","Independent literary scholars serving as pilot reviewers","Readers who rely on the edition's commentary"],"affected_objective":"Preserve the traceability and textual warrant of published interpretive notes as the edition, its source text, and its scholarly apparatus change.","arm":"COMMON_P1","authority_safety":{"authorized_first_step":"The general editor may authorize a read-only pilot on 20 existing interpretive claims; reviewers may record condition signals and proposed repairs but may not alter or suppress published material.","decision_authority":"Section editors decide whether to repair notes within their remit; the general editor decides whether to place a visible qualification on a published claim, defer a release, or commission deeper textual review.","excluded_actions":["Automatically retracting or rewriting interpretations from a score","Ranking or evaluating scholars by the condition of claims they authored","Treating theoretical disagreement alone as deterioration","Changing the primary text or apparatus without ordinary textual-editing review","Publishing defect flags before the responsible editor can inspect the cited evidence"],"halt_rollback":"Halt the pilot if reviewers cannot apply the condition rubric consistently, if inspection burden disrupts scheduled editorial work, or if flags are used for personnel evaluation; remove pilot-only labels and leave the edition unchanged while retaining an access-controlled audit record for rubric review."},"baseline":"At pilot start, each sampled load-bearing claim receives a versioned record linking its exact wording to primary-text passages, edition state, cited scholarship, theoretical premises, and acknowledged counterevidence. Healthy means all anchors resolve and the stated evidence remains compatible with the claim; acceptable means a minor citation or anchor defect is repairable without changing the claim; warning means unresolved anchor changes, inherited assertions without checked support, or accumulating counterevidence weaken its warrant; unacceptable means the claim's necessary textual support is missing, contradicted by the current edition, or cannot be reconstructed. Two successive inspections showing the same warning signal establish persistence; acceleration is recorded when new warnings arise faster across consecutive inspections.","candidate_id":"deterioration_monitoring__literature_literary_theory__COMMON_P1","causal_chain":["A living critical edition changes through new textual variants, corrected transcriptions, renumbered passages, revised bibliography, and inherited annotations.","Each local change can leave an interpretive note slightly less connected to the evidence or premise on which it originally depended.","Because the note remains fluent and no individual change forces reconsideration, editors and readers can normalize weakened warrant.","Claim-level records and recurring samples expose leading signals such as changed anchors, superseded premises, and unchecked inheritance, alongside lagging judgments that a claim is no longer supported.","Preserved inspection history reveals the direction, persistence, and acceleration of those signals rather than treating each defect as isolated.","Warning and unacceptable bands activate named editorial authority and a proportionate response pathway.","Editors repair citations, narrow or qualify claims, commission renewed close reading, or temporarily mark a note as under review while intervention remains feasible.","Review of false alarms and completed repairs recalibrates the indicators, cadence, and thresholds for later edition cycles."],"cell_id":"deterioration_monitoring__literature_literary_theory","consequence":"Unsupported or edition-incompatible interpretations can remain presented as warranted commentary, obscuring the boundary between a defensible contested reading and a claim whose evidentiary basis has decayed; later correction then requires reconstructing lost dependencies across multiple revisions.","diversity_from_prior_proposals":"Prior proposals were not inspected under runtime isolation. This candidate is specified independently around deterioration of claim-to-evidence links in a living critical edition, without asserting uniqueness relative to any other proposal.","experiment_id":"eoa_inverse_innovation_exp13_second_slot_policy60_20260806","intervention":"Install an Interpretive Warrant Condition Loop for a living digital critical edition. At each monthly or pre-release inspection, section editors sample versioned, load-bearing interpretive claims and record whether primary-text anchors still resolve, cited editions and theoretical premises remain current, inherited claims have been rechecked after relevant changes, material counterevidence is represented, and prior warnings were repaired. The register retains disaggregated signals and their history. A single changed anchor prompts verification; a warning persisting for two inspections requires an assigned repair plan; loss or contradiction of necessary textual support requires general-editor review and either repair, qualification, renewed close reading, or a temporary under-review notice. After each cycle, editors compare flags with adjudicated claim condition and revise noisy indicators or an ill-matched cadence.","mechanism_mapping":[{"counterfactual_removal":"Without recurring claim sampling, warrant loss can remain unnoticed between major editorial reviews.","mechanism_slug":"preventive_inspection","role":"Samples interpretive claims before readers encounter an obvious evidentiary failure."},{"counterfactual_removal":"Without a versioned condition record, editors see isolated defects but cannot distinguish fluctuation from persistent or accelerating decay.","mechanism_slug":"quality_drift_monitoring","role":"Tracks changes in the fit between interpretive claims and their textual, bibliographic, and theoretical support."},{"counterfactual_removal":"Without a visible repair queue tied to warnings, detected defects may remain passive documentation.","mechanism_slug":"maintenance_backlog_review","role":"Connects unresolved warning claims to owners, due points, and editorial responses."},{"counterfactual_removal":"Without periodic testing against adjudicated readings, easy-to-count proxies such as changed anchors could displace substantive warrant assessment.","mechanism_slug":"false_alarm_review","role":"Tests whether indicators correspond to actual deterioration and supports recalibration."},{"counterfactual_removal":"Without claim-level close reading, the system could mistake legitimate theoretical disagreement for degraded evidentiary support.","mechanism_slug":"close_reading","role":"Provides the domain-specific examination used to decide whether a flagged claim remains textually warranted."}],"nearest_rivals":["Ordinary peer review assesses a submission at a decision point but does not maintain recurring trend memory after the edition changes.","Citation checking verifies bibliographic correctness but need not test whether an interpretation still follows from its cited passage and premises.","Version control records what changed but does not define healthy condition, warning bands, or editorial response thresholds.","Close reading evaluates textual details but, by itself, does not create a recurring portfolio-level deterioration and escalation loop.","A scheduled wholesale commentary revision performs maintenance by calendar rather than using observed claim condition to prioritize response."],"negative_tests":{"intervention_falsifier":"The intervention is undermined if pilot flags do not precede independently adjudicated unacceptable warrant, if reviewers cannot distinguish deterioration from legitimate interpretive disagreement using the rubric, or if threshold crossings repeatedly fail to produce a feasible editorial response.","problem_falsifier":"The inferred problem is unsupported if version-history review finds that sampled claims retain resolvable evidence and premises across revisions, apparent defects are random one-off errors rather than accumulating condition loss, or ordinary editorial processes already identify and repair such changes before warrant weakens.","risks":["False alarms may recast plural interpretation as objective defect.","Reviewers may privilege theories familiar to them when judging whether premises remain viable.","A composite score could hide severe deterioration in one claim, so signals must remain disaggregated.","Frequent inspection may divert effort from substantive scholarship.","Visible warnings may reduce reader confidence without sufficient context.","Editors may game indicators by adding superficial citations rather than repairing reasoning.","Repair demand may exceed available editorial capacity."],"strongest_counterevidence":"Interpretive claims are contestable and can remain defensible under changing editions; a changed anchor, new counterreading, or superseded theoretical vocabulary may indicate scholarly development rather than deterioration."},"next_evidence_step":"Run one read-only, eight-week pilot on 20 load-bearing notes from a single living digital critical edition. Two independent literary scholars first apply the condition rubric to archived versions without seeing the monitoring flags; editors then reconstruct monthly signal histories from two prior release cycles and test whether flagged changes occurred before the adjudicated warning or unacceptable state. Record reviewer agreement, false-alarm reasons, time required, available repair options, and whether each proposed threshold maps to existing editorial authority. Do not modify published notes during the pilot.","observable_state":"A claim-level, versioned register shows exact text anchors, cited edition and scholarship, theoretical dependencies, counterevidence notes, anchor changes, unresolved citations, unreviewed inheritance after relevant revisions, reviewer condition judgments, signal direction and persistence, assigned owner, repair status, and elapsed time since a threshold crossing.","prior_art_status":"UNSEARCHED","problem":"In a living digital critical edition, interpretive notes can gradually lose their textual warrant as source readings, passage identifiers, bibliographies, and surrounding annotations change. Each mismatch is locally small, and fluent inherited prose can continue to appear authoritative even when its primary-text anchor or necessary premise has weakened. The edition has change records and editorial review, but no recurring condition model that trends claim-level warrant and activates repair before unsupported commentary accumulates.","proposal_index":1,"remaining_contrastive_claim":"The defining claim is that a versioned, thresholded trend loop for claim-to-evidence condition can identify repairable deterioration in interpretive warrant that one-time peer review, citation checking, version control, or close reading alone does not operationally connect to recurring editorial response.","revision_record":{"claim_changes":["Initial formulation; no parent-version claims were revised."],"conceptual_changes":["Initial transfer of deterioration monitoring to the condition of interpretive warrant in a living critical edition."],"evidence_changes":["No external evidence added; the candidate remains closed-book and prior art is unsearched."],"operational_changes":["Initial specification of condition bands, monthly or pre-release inspection, persistence thresholds, decision authority, safeguards, and an eight-week read-only pilot."],"parent_version":null,"progress_targets_addressed":["Concrete domain problem","Observable deterioration signals","Trend memory and thresholds","Action-linked editorial response","Authority boundaries and rollback","Bounded falsifiable evidence step"]},"schema_version":1,"structural_mapping":[{"archetype_element":"Condition to preserve","domain_realization":"The traceable textual and scholarly warrant of each load-bearing interpretive note."},{"archetype_element":"Baseline and condition bands","domain_realization":"Healthy, acceptable, warning, and unacceptable states defined by resolvable anchors, current premises, represented counterevidence, and compatibility between claim and evidence."},{"archetype_element":"Leading and lagging deterioration indicators","domain_realization":"Leading signals are changed text anchors, superseded dependencies, unresolved citations, and unchecked inheritance; lagging signals are independent judgments that necessary support is absent or contradictory."},{"archetype_element":"Inspection cadence","domain_realization":"A monthly inspection or review before each edition release, with cadence reconsidered if observed change rates or repair lead times differ."},{"archetype_element":"Trend interpretation","domain_realization":"The register preserves direction, persistence across two inspections, acceleration across cycles, and localized claim histories instead of relying on a single aggregate score."},{"archetype_element":"Thresholds and escalation","domain_realization":"Anchor change triggers verification, persistent warning triggers an owned repair plan, and missing or contradictory necessary support triggers general-editor review."},{"archetype_element":"Response pathway","domain_realization":"Repair the reference, revise or narrow the interpretation, add a qualification, commission renewed close reading, or place a temporary under-review notice."},{"archetype_element":"Recalibration","domain_realization":"Compare flags with independent adjudication and completed repairs, then adjust indicators, thresholds, sampling depth, or cadence when false alarms or missed deterioration appear."}],"title":"Interpretive Warrant Condition Loop for Living Critical Editions","version":0}