{"actors":["Literary scholars responsible for period-level claims","Corpus curators who define the included works and metadata","Close readers who annotate passages","Anthology and syllabus editors who use genre or period descriptions","Students and readers affected by those descriptions"],"affected_objective":"Keep a valid corpus-level account of a literary period usable while preventing it from being treated as a uniform description of every work, chapter, or author subgroup.","arm":"COMMON_P1","authority_safety":{"authorized_first_step":"A corpus curator and designated literary scholar may run a reversible, read-only pilot on 24 public-domain novels from one predefined period, using existing metadata and passage annotations created for the pilot.","decision_authority":"The responsible scholar retains authority over interpretive claims; anthology or syllabus editors retain authority over selection decisions, with affected claims reviewed by at least one close reader who did not produce the distributional summary.","excluded_actions":["Automatically relabeling works, authors, genres, or periods","Changing an anthology, syllabus, catalog, or public-facing interpretation from pilot results alone","Inferring protected or sensitive author attributes from textual features","Treating computational passage labels as ground truth","Publishing identifiable annotator-performance rankings"],"halt_rollback":"Stop the pilot if annotation disagreement remains unresolved, subgroup cells become too small for non-stigmatizing interpretation, or the corpus boundary changes mid-analysis; retain the original period description, withdraw pilot-derived flags, and archive only de-identified audit records."},"baseline":"The period is represented by a corpus-wide average or dominant-mode label, illustrated with a few canonical passages. Individual works are then discussed as instances of that label, while divergent passages or work-level trajectories are handled as exceptions without a systematic crosswalk to the aggregate claim.","candidate_id":"ensemble_and_population_level_equilibrium_versus_individual_level_heterogeneity__literature_literary_theory__COMMON_P1","causal_chain":["A curator fixes a corpus of works, passage units, publication window, and inclusion rules.","Annotators classify sampled passages by a predeclared set of narrative modes and record uncertainty and disagreement.","Passage labels are aggregated with equal work weighting into a corpus-level mode profile for each publication cohort.","The corpus profile is tested for stability across adjacent cohorts while work-level distributions and within-work trajectories remain visible.","A crosswalk states that stability of the corpus profile supports a period-level claim but does not imply that any particular work or author conforms to it.","Excursions are checked for clustering by work, publication cohort, attributed genre, or narrative position rather than being dismissed as anecdotes.","A stratified close-reading panel examines passages from the center, tails, and identified clusters to determine whether variation is interpretively meaningful, annotation error, or functional formal diversity.","Only corroborated local patterns receive qualified work-level notes or targeted teaching prompts; the corpus-level description is preserved unless the macro stability test itself fails.","Subsequent reviews display both corpus stability and work-level distributions so correction of local overgeneralization cannot silently replace one totalizing label with another."],"cell_id":"ensemble_and_population_level_equilibrium_versus_individual_level_heterogeneity__literature_literary_theory","consequence":"A defensible statement about a period's aggregate narrative profile is overextended into claims about particular works, causing formally hybrid texts to be misdescribed and giving anthology or teaching decisions an unsupported appearance of uniformity.","diversity_from_prior_proposals":"This is the sole proposal in the requested output; no comparison with prior proposals was performed under runtime isolation.","experiment_id":"eoa_inverse_innovation_exp13_second_slot_policy60_20260806","intervention":"Create a Period–Work Narrative Mode Crosswalk that pairs the stable corpus-level distribution with per-work and within-work distributions. Use equal-work aggregation, predeclared subgroup cuts, uncertainty-bearing passage annotations, excursion flags, and a stratified close-reading panel. Preserve the period-level claim when its aggregate indicator remains stable, but qualify work-level descriptions when corroborated local patterns depart from it.","mechanism_mapping":[{"counterfactual_removal":"Without the dashboard, reviewers see the period profile or selected examples but cannot inspect the work-level distribution hidden beneath it.","mechanism_slug":"distributional_dashboard","role":"Display the corpus indicator beside work quantiles, within-work ranges, subgroup cuts, and annotation uncertainty."},{"counterfactual_removal":"Without the crosswalk, passage evidence can be cited without an explicit rule connecting it to, or limiting, the period-level claim.","mechanism_slug":"micro_macro_crosswalk","role":"Document how weighted passage labels produce the corpus profile and which interpretations are licensed at corpus, subgroup, work, and passage levels."},{"counterfactual_removal":"Without stratified review, convenient canonical passages can substitute for evidence spanning the observed distribution.","mechanism_slug":"stratified_sampling_review","role":"Select close-reading cases from central, tail, subgroup, and uncertain regions of the profile."},{"counterfactual_removal":"Without excursion alerts, repeated local departures remain anecdotal while the stable headline indicator continues to govern interpretation.","mechanism_slug":"subgroup_excursion_alert","role":"Flag sustained work-level or subgroup departures for human review without declaring them errors or invalidating the macro profile."}],"nearest_rivals":["Close reading alone: it can establish passage-level complexity but does not define or test the relationship between a stable corpus profile and the distribution of works.","Variability characterization: it maps differences among works but does not require a valid aggregate period claim whose scope must be bounded.","Aggregation-bias correction: it challenges a distorted corpus statistic, whereas this candidate allows the equal-work aggregate to remain valid and addresses its overextension to members.","Genre reclassification: it replaces labels at the work level, whereas this candidate can preserve existing labels while qualifying what they imply."],"negative_tests":{"intervention_falsifier":"The intervention is falsified for this decision if masked reviewers given the crosswalk make no fewer unsupported work-level generalizations than reviewers given the aggregate profile and canonical examples, using a predeclared rubric, while requiring comparable source evidence.","problem_falsifier":"The problem is falsified in the pilot if reviewers using the baseline period description already keep corpus-level and work-level claims separate, or if work-level narrative-mode profiles show no consequential departures from the aggregate under the predeclared relevance criteria.","risks":["Narrative-mode categories may impose false precision on ambiguous passages.","Small subgroup cuts may encourage interpretive fishing or stigmatizing author classifications.","Equal-work weighting may answer a different scholarly question from weighting by length, readership, or publication volume.","Excursion flags may privilege measurable formal features over historically situated meanings.","Reviewers may treat annotations or computational assistance as authoritative despite disagreement.","Added distributional detail may obscure rather than clarify the limited period-level claim."],"strongest_counterevidence":"If independent close readers consistently judge the flagged tails to be ordinary within-work modulation with no bearing on work-level interpretation or selection decisions, then the visible heterogeneity exists but is not action-relevant."},"next_evidence_step":"On 24 public-domain novels from one predefined period, sample 20 passages per work across narrative position. Have two blinded annotators apply a fixed narrative-mode codebook, adjudicate disagreements without altering the codebook, compute an equal-work cohort profile and per-work distributions, and ask a separate pair of reviewers to assess 12 baseline summaries and 12 crosswalk summaries using a predeclared level-of-analysis rubric. Record disagreement, unsupported inference counts, review time, and all halt conditions; do not change external descriptions or selections.","observable_state":"Across adjacent publication cohorts in the bounded corpus, the equal-work proportion of predeclared narrative modes remains within a fixed stability band, while individual works show different mode mixtures and within-work trajectories. Current period summaries report the stable corpus profile without showing those distributions or specifying that the claim is corpus-level.","prior_art_status":"UNSEARCHED","problem":"A literary scholar or anthology editor uses a stable corpus-level narrative-mode profile for a period as though it described each included novel. The aggregate description may be valid, yet formally hybrid works and patterned work-level departures are erased because the translation from passages to corpus and the boundary between period and work claims are not reported.","proposal_index":1,"remaining_contrastive_claim":"Unlike close reading, variance mapping, aggregation correction, or genre relabeling alone, the candidate is specifically warranted when a valid and stable corpus-level literary claim must be retained while its implications for individual works are explicitly limited by distributional evidence.","revision_record":{"claim_changes":[],"conceptual_changes":[],"evidence_changes":[],"operational_changes":[],"parent_version":null,"progress_targets_addressed":["Defined a concrete literary-theory problem and affected decision.","Specified an observable macro indicator, microstate profile, and aggregation rule.","Bounded interpretive authority and added reversible safeguards.","Provided problem and intervention falsifiers plus a finite first evidence step."]},"schema_version":1,"structural_mapping":[{"archetype_element":"Ensemble Frame","domain_realization":"A fixed corpus of 24 public-domain novels, divided into adjacent publication cohorts; passages are micro-units and novels are the equal-weight ensemble members."},{"archetype_element":"Macro Equilibrium Indicator","domain_realization":"The equal-work mean proportion of passages assigned to each predeclared narrative mode, considered stable when adjacent cohort profiles remain inside the pilot's fixed band."},{"archetype_element":"Microstate Variability Profile","domain_realization":"Per-work narrative-mode mixtures, within-work trajectories by narrative position, annotation uncertainty, and distribution tails."},{"archetype_element":"Aggregation Translation Rule","domain_realization":"Passage-mode proportions are calculated within each novel and then averaged equally across novels, preserving each work's contribution while discarding passage sequence and much local semantic context from the macro indicator."},{"archetype_element":"Level-of-Analysis Boundary","domain_realization":"The aggregate supports a claim about the bounded corpus and period window; claims about a genre subgroup, novel, chapter, or passage require evidence at that level."},{"archetype_element":"Heterogeneity Relevance Test","domain_realization":"A departure matters only when it is sustained within a work or clusters under a predeclared cut, survives uncertainty review, and changes a close reader's warranted description or a pending selection rationale."},{"archetype_element":"Subgroup and Locality Map","domain_realization":"Departures are mapped by novel, publication cohort, attributed genre, and narrative position without inferring new author identities."},{"archetype_element":"Multi-Level Feedback Design","domain_realization":"The review surface retains the period profile while linking corroborated excursions to qualified work-level notes and monitoring both levels over later corpus updates."},{"archetype_element":"Representative Case Guardrail","domain_realization":"Close-reading cases are sampled from central, tail, subgroup, and uncertain regions; no single canonical or extreme passage may stand in for the corpus."}],"title":"Period–Work Narrative Mode Crosswalk","version":0}