{"actors":["Editors and annotators preparing scholarly editions, allusion indexes, or digital reading guides","Literary scholars and research assistants who generate candidate source-text parallels","Independent adjudicators who assess whether a parallel supports an intertextual claim","Readers, teachers, and catalog maintainers who rely on published allusion labels"],"affected_objective":"Calibration of claims that a passage deliberately invokes a particular precursor before those claims enter scholarly annotations, intertext maps, or teaching materials.","arm":"ORDINARY_DIVERSE_P2","authority_safety":{"authorized_first_step":"With the annotation team's consent, conduct a read-only shadow review of one completed batch of candidate allusions, reconstructing the candidate family and applying provisional status labels without changing any edition, database, or public annotation.","decision_authority":"Editors retain authority over annotations and edition language; scholars retain authority over interpretive arguments; repository maintainers control database publication; independent adjudicators only supply bounded assessments.","excluded_actions":["Automatically deleting an annotation because it arose from a broad search","Equating lexical similarity with authorial intention","Opening reserved corroborating materials before candidates and adjudication rules are frozen","Changing an edition, attribution, curriculum, or contributor credit during the shadow review","Exposing restricted archival materials or confidential editorial deliberations","Treating multiplicity control as proof that an allusion is aesthetically important"],"halt_rollback":"Halt if candidate provenance cannot be reconstructed, restricted archival material is exposed, adjudicators recognize reserved cases, or contributor attribution becomes disputed. Revoke pilot access and remove provisional labels while leaving source records and editorial decisions unchanged."},"baseline":"An annotation project searches many possible precursor works and target passages using phrases, motifs, names, images, translations, semantic similarities, and expert suggestions. The strongest-looking matches receive intentional-allusion annotations, while rejected candidates, altered thresholds, and alternative source pools are not retained in the published evidentiary context.","candidate_id":"multiple_testing_discipline__literature_literary_theory__ORDINARY_DIVERSE_P2","causal_chain":["A target work can be compared with numerous possible precursors at the level of wording, motif, image, plot situation, translation, or rhetorical form.","Automated searches and expert browsing generate many source-target pairings, often under several similarity thresholds or matching rules.","With enough pairings, some candidates can appear unusually close even when resemblance alone does not distinguish deliberate invocation from coincidence or shared convention.","If rejected candidates and search variations vanish, a selected parallel is presented without the opportunity set that produced it.","An editor may promote the selected resemblance directly to an intentional-allusion annotation.","The annotation can then organize readers' interpretations, citation networks, or teaching materials as though the relation had already been confirmed.","A campaign-level candidate registry, multiplicity-aware screening rule, graded status labels, and prospective blinded adjudication with search-independent corroboration make the confirmation burden reflect the full search space."],"cell_id":"multiple_testing_discipline__literature_literary_theory","consequence":"A coincidental resemblance or broadly shared convention can be stabilized as a specific intertextual relation, directing subsequent interpretation and annotation toward an unsupported precursor.","diversity_from_prior_proposals":"This opportunity concerns promotion of individual source-target resemblances into intentional-allusion annotations after mass candidate screening. Its intervention governs an annotation queue through decoy-based blinded adjudication and search-independent corroboration, rather than governing corpus-level claims about the distribution or history of literary forms.","experiment_id":"eoa_inverse_innovation_exp13_second_slot_policy60_20260806","intervention":"Establish a Source-Parallel Promotion Gate for allusion annotation. For each annotation campaign, define the family of eligible source works, target passages, match types, translations, thresholds, and expert nominations; register every generated candidate and material search change; choose an error policy that treats broad matching as exploratory screening; apply a declared multiplicity-aware selection rule to the candidate queue; and label surviving relations as generated candidate, adjusted lead, independently adjudicated parallel, corroborated allusion, or rejected. Before an intentional-allusion label is published, freeze the candidate and rationale, have independent readers assess it blindly among matched decoys under a predeclared criterion, and seek specified search-independent evidence such as drafts, notes, correspondence, documented access, or a distinctive cluster of correspondences. Preserve selected and rejected candidates in the editorial record.","mechanism_mapping":[{"counterfactual_removal":"Without the registry, rejected parallels, altered source pools, and threshold changes disappear, allowing a selected match to look like an isolated observation.","mechanism_slug":"claim_registry","role":"Records each source-target pairing, generation route, match type, search configuration, disposition, and evidentiary status."},{"counterfactual_removal":"Without a multiplicity-aware screen, a large candidate generator can promote its most striking matches under a standard suited to one comparison.","mechanism_slug":"false_discovery_rate_control","role":"Selects a bounded set of candidate parallels for follow-up while treating them as leads rather than confirmed intentional relations."},{"counterfactual_removal":"Without prospective adjudication, the people who selected a candidate can also confirm it using the same features that attracted them initially.","mechanism_slug":"confirmatory_follow_up","role":"Freezes the focal relation, decoy construction, interpretive question, and success criterion before independent readers assess the candidate."},{"counterfactual_removal":"Without independent assessment, editorial enthusiasm or knowledge of the proposed precursor can shape whether a resemblance is perceived as distinctive.","mechanism_slug":"replication_study","role":"Uses readers who did not generate or select the candidate to test whether the proposed relation remains distinguishable under the frozen protocol."},{"counterfactual_removal":"Without priority rules, a merely similar phrase can receive the same label as a parallel supported by distinctive clustering and external documentation.","mechanism_slug":"metric_hierarchy","role":"Separates lexical or thematic resemblance from stronger evidence tiers and reserves intentional-allusion language for candidates meeting the declared corroboration requirement."}],"nearest_rivals":["Source criticism can investigate a proposed precursor in depth but does not itself account for the number of alternative precursors and passages screened before that proposal was selected.","Close reading can demonstrate a parallel's interpretive richness but cannot determine how surprising its discovery was within the full candidate family.","Provenance verification can establish that an author could have encountered a source, but access alone does not control false selection among many plausible sources.","Reproducibility protocol can rerun a similarity search yet still reproduce a cherry-picked candidate without calibrating its status to all generated pairings."],"negative_tests":{"intervention_falsifier":"The gate is undermined if generated candidates receive intentional-allusion language before adjudication, if decoy identities are visible to adjudicators, or if rejected and unselected pairings remain absent from the editorial record.","problem_falsifier":"The multiplicity problem is absent when one source-target relation was specified before inspection and evaluated under a fixed rationale, or when all generated parallels remain explicitly speculative reading prompts with no attributional, annotative, ranking, or curricular consequence.","risks":["The claim family may be defined too narrowly by omitting informal browsing or unlogged expert suggestions.","Decoys may be obviously weaker than selected candidates and create an uninformative confirmation exercise.","Independent readers may share the same conventions or background assumptions as the selecting editors.","Documented access to a source may be mistaken for evidence of deliberate invocation.","Strict screening may discard interpretively productive parallels that should remain available as explicitly exploratory prompts.","Archival corroboration requirements may unevenly disadvantage authors whose drafts, correspondence, or reading records were not preserved.","The registry may expose contributor ideas or restricted source materials without appropriate consent."],"strongest_counterevidence":"Some intertextual readings are offered to disclose a productive relation between texts without asserting historical influence or authorial intention. If the annotation explicitly makes only that relational claim, requiring confirmation of deliberate invocation would test a stronger proposition than the scholarship advances."},"next_evidence_step":"For one consenting completed annotation batch, reconstruct no more than 40 generated or nominated source-target pairings and their search configurations. Have two editors independently define the candidate family and status hierarchy, then freeze one selected candidate, construct matched decoys, and obtain blinded assessments from three readers who did not participate in selection. Check a predesignated set of search-independent materials only after adjudication. Record family-definition agreement, missing candidates, blinding failures, label changes, and operational burden without altering any public annotation.","observable_state":"A scholarly edition or intertext database labels a passage as an intentional allusion to a named precursor, while working files show many discarded source matches, similarity thresholds, translations, motif queries, or expert-nominated parallels that are absent from the annotation's evidentiary record.","prior_art_status":"UNSEARCHED","problem":"Allusion annotation can involve screening many possible precursor works, passages, translations, motifs, and similarity rules. When the most striking source-target resemblance is promoted directly to an intentional intertextual claim, its evidentiary status does not reflect the many opportunities to find an attractive coincidence or shared convention.","proposal_index":2,"remaining_contrastive_claim":"A source-target resemblance selected from a large annotation search should remain an exploratory parallel until its status reflects the complete candidate family and it survives a frozen, independent assessment plus the declared corroboration requirement; rerunning or closely reading the selected match alone is insufficient for promotion to intentional allusion.","revision_record":{"claim_changes":["Initial candidate; no prior version or inherited claims."],"conceptual_changes":["Initial mapping of multiple-testing discipline to source-parallel screening and intentional-allusion annotation."],"evidence_changes":["No external or prior-art evidence consulted; all empirical implications are framed as propositions to test."],"operational_changes":["Specified a bounded read-only shadow review, candidate registry, multiplicity-aware queue, graded labels, matched decoys, blinded adjudication, and search-independent corroboration."],"parent_version":null,"progress_targets_addressed":["Defined a many-comparison problem distinct from corpus-level literary-historical inference.","Preserved the claim-family, attempted-look inventory, error policy, adjustment, status, confirmation, and discovery-record sequence.","Separated candidate generation from publication of intentional-allusion labels.","Specified decision authority, rollback conditions, negative tests, and a bounded evidence step."]},"schema_version":1,"structural_mapping":[{"archetype_element":"Define the claim family","domain_realization":"Specify all eligible precursor works, target passages, languages, translations, motif classes, match types, similarity thresholds, and expert nominations within one annotation campaign."},{"archetype_element":"Inventory attempted looks","domain_realization":"Log every generated or manually proposed source-target pairing, search query, threshold change, translation choice, rejection, and renewed candidate."},{"archetype_element":"Set the error-risk policy","domain_realization":"Allow broad candidate generation, require multiplicity-aware screening for follow-up, and reserve intentional-allusion labels for candidates meeting a stricter confirmation rule."},{"archetype_element":"Apply a multiplicity-aware rule","domain_realization":"Apply a declared discovery-rate rule to the full candidate queue and use matched-decoy adjudication rather than evaluating only the winning parallel."},{"archetype_element":"Label claim status","domain_realization":"Mark relations as generated candidate, adjusted lead, independently adjudicated parallel, corroborated allusion, or rejected in both internal and published records."},{"archetype_element":"Confirm before costly action","domain_realization":"Freeze the proposed relation and rationale, obtain prospective blinded assessments from non-selecting readers, and check predesignated search-independent corroboration before publishing intentionality language."},{"archetype_element":"Preserve the discovery record","domain_realization":"Maintain a versioned editorial record containing selected, null, rejected, and superseded pairings with their generation routes and status transitions."}],"title":"Source-Parallel Promotion Gate for Allusion Annotation","version":0}