{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp05_complete_proposal_portfolio20_20260803","cell_id":"invariant_mode_decomposition_design__literature_literary_theory","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_id":"cand_living_narrative_mode_coverage_005","proposal_index":5,"version":0,"title":"Mode-Coverage Sampling for Living Narrative Archives","problem":"An archive collecting repeated performances of the same living narrative may allocate recording effort by performer counts, occasion types, or conspicuous motif differences. Those coordinates do not show which combinations of event order, focalization, formulaic language, direct speech, causal linkage, and closure persist, grow, alternate, or decay across retellings. A collection can therefore contain many recordings yet leave a culturally significant transformation mode unobserved or documented too sparsely to distinguish it from residual variation.","actors":["Narrators and tradition-bearers, who control whether, when, and under what conditions performances may be recorded or analyzed","Community-designated cultural authority or rights body, which governs collective permissions and access conditions","Archivist, who manages the collection and proposes recording allocations","Literary scholar or narratologist, who defines traceable performance features with tradition-bearer review","Analyst, who estimates modes and reconstruction error without judging authenticity","Authorized archive users, whose permitted research tasks determine the required coverage"],"observable_state":"Within an authorized set of repeated performances of one recognized narrative, each recording is represented by a fixed state vector comprising episode inclusion, event-order displacement, focalization distribution, formulaic-language density, direct-speech allocation, causal-link explicitness, and closure form. Features remain linked to time-coded passages and community-approved descriptions. For comparable consecutive performances within a declared context window, a locally estimated transition matrix maps deviations in one performance state to the next. Context labels such as occasion, audience configuration, elicited versus customary setting, and performer-approved prompt type are recorded separately. Held-out performance states, reconstruction residuals, modal-coordinate coverage, and community-designated consequential variations are the observable outcomes.","consequence":"A constrained archive may overcollect easily accessible surface variation while failing to preserve enough observations of a persistent or context-sensitive narrative transformation. Later users may then treat one recorded version as structurally typical, infer stability from a sampling gap, or be unable to reconstruct how linked narrative features change together.","affected_objective":"Allocate limited, authorized recording and description effort so the archive covers culturally permitted transformation modes and keeps unmodeled performance variation visible, without ranking narrators or defining an authentic version.","intervention":"Conduct a read-only modal audit of existing authorized recordings before requesting new performances. Estimate approximate performance-to-performance modes within a narrowly declared narrative and context window, classify their persistence, and test reconstruction on held-out recordings. Use retrospective context variation to perform a sensitivity sweep identifying which already permitted recording conditions are associated with observing each supported mode. Produce a community-reviewable coverage map showing well-observed modes, under-observed modes, couplings, residuals, and access restrictions. Only after rights-holder approval, invite a bounded set of additional recording sessions in permitted contexts selected to observe an under-covered action-relevant mode. Re-estimate coverage after each session; do not prescribe story content or ask a narrator to reproduce a computed mode.","structural_mapping":[{"archetype_element":"Transformation Scope","domain_realization":"The transformation is the change between comparable consecutive performances of one recognized narrative within a community-approved context and time window."},{"archetype_element":"State-Vector Definition","domain_realization":"Coordinates describe aligned narratological features while retaining links to the precise recorded passages from which each value was derived."},{"archetype_element":"Invariant Mode Basis","domain_realization":"Approximate eigenvectors of the locally estimated transition map represent combinations of performance features that tend to retain their direction across retellings."},{"archetype_element":"Modal Gain Spectrum","domain_realization":"Each gain estimates whether its associated combination is damped, preserved, reversed, alternated, or amplified over successive observed performances."},{"archetype_element":"Dominant Mode Selection Rule","domain_realization":"A mode becomes sampling-relevant only if it is stable under coder and recording resampling, predicts held-out performance structure, is permitted for analysis, and concerns a variation designated as consequential by the governing tradition-bearers."},{"archetype_element":"Stable/Unstable Mode Partition","domain_realization":"Modes are classified by repeated-performance behavior for description and sampling; instability is not interpreted as cultural defect or textual corruption."},{"archetype_element":"Modal Intervention Map","domain_realization":"Retrospective sensitivity analysis maps authorized recording contexts to the probability of observing supported modal coordinates, informing invitations rather than altering performances."},{"archetype_element":"Reconstruction Residual Check","domain_realization":"Held-out recordings are reconstructed from retained modes, and structured residual passages remain explicit sampling and interpretive obligations."},{"archetype_element":"Mode-Coupling Register","domain_realization":"Near-degenerate modes and context-linked cross-effects are documented so coupled narrative changes are not split into independent traits."},{"archetype_element":"Local Linearization Window","domain_realization":"The model applies only to the named narrative, authorized performance relationships, feature rubric, context range, and collection period."},{"archetype_element":"Spectral Gap Threshold","domain_realization":"A reduced coverage map is used only when retained modes remain separated from omitted behavior under resampling."},{"archetype_element":"Mode Drift Monitor","domain_realization":"Modes and coverage are re-estimated after new sessions, contextual change, feature-rubric revision, or repeated reconstruction failure."},{"archetype_element":"Interpretation Scope Contract","domain_realization":"Modes describe patterns in the authorized sample; they do not establish authenticity, communal prevalence, performer competence, historical origin, or a complete account of the tradition."}],"mechanism_mapping":[{"mechanism_slug":"mode_shape_testing","role":"Recover approximate modes empirically from actual authorized performances because no complete narrative-transmission operator is known in advance.","counterfactual_removal":"Without empirical mode recovery, the sampling plan would impose assumed structures on the performances rather than test which coupled directions appear in the recordings."},{"mechanism_slug":"eigendecomposition_workflow","role":"Decompose the locally estimated square performance-transition matrix into a complete provisional basis and gain spectrum.","counterfactual_removal":"Without decomposition, the audit could compare individual features but could not identify combinations preserved or transformed together across retellings."},{"mechanism_slug":"modal_stability_analysis","role":"Classify supported modes as decaying, persistent, reversing, oscillatory, or growing within the declared observation window.","counterfactual_removal":"Without stability classification, a conspicuous one-time variation could be confused with a recurring transformation direction."},{"mechanism_slug":"modal_sensitivity_sweep","role":"Estimate how already observed, authorized recording contexts relate to modal-coordinate coverage and test whether proposed sessions add distinct information.","counterfactual_removal":"Without the sweep, an under-covered mode would be visible but the archive would lack a bounded basis for choosing among permitted recording opportunities."},{"mechanism_slug":"residual_reconstruction_test","role":"Evaluate retained modes on held-out performances and expose structured narrative changes that the modal subset fails to reproduce.","counterfactual_removal":"Without residual testing, the archive could discard low-amplitude or singular material that is culturally or narratologically consequential."},{"mechanism_slug":"spectral_gap_monitor","role":"Track whether the retained-mode boundary and mode identities remain stable as recordings are added.","counterfactual_removal":"Without gap and drift monitoring, the archive could continue sampling against a modal basis invalidated by new performances."},{"mechanism_slug":"spectral_decomposition_report","role":"Create a rights-aware coverage report linking modes to source passages, contexts, residuals, couplings, permissions, and interpretation limits.","counterfactual_removal":"Without the bounded report, provisional modes could be detached from their sources and misrepresented as authentic forms or population-level facts."}],"causal_chain":["Repeated performance transforms several narrative features jointly rather than changing isolated motifs independently.","Within a sufficiently bounded context, some coupled feature directions may persist, alternate, grow, or decay more predictably than individual coordinates.","Quota or motif-based collection can repeatedly sample visible coordinates without covering those transformation directions.","Empirical decomposition identifies provisional modes and their gains while retaining traceability to recorded passages.","Held-out reconstruction, resampling, and community review distinguish usable modes from artifacts, inaccessible material, and residual singularity.","Retrospective sensitivity analysis links supported modes to recording contexts that rights holders already permit.","A community-authorized sampling invitation adds observations in a context associated with an under-covered mode without scripting narrative content.","New recordings either improve modal coverage or invalidate the approximation, after which the archive revises or retires the sampling map."],"baseline":"The baseline allocates recordings by available narrator, calendar opportunity, geographic or occasion quotas, and a checklist of visible motif differences. Archivists describe each performance individually and may seek variety, but they do not estimate repeated-performance directions, modal gains, spectral separation, or residual reconstruction.","nearest_rivals":["Performer and occasion quota sampling, which distributes sessions across named categories but does not measure coupled transformation coverage.","Motif checklists, which preserve recognizable narrative elements but treat them independently and can miss combinations that move together.","Maximum-difference sampling, which selects recordings far apart in visible coordinates but may overselect residual outliers rather than persistent transformation modes.","Random sampling among authorized opportunities, which avoids some curator preferences but can spend scarce sessions on already-covered modal directions.","Ethnographic purposive sampling, which uses contextual expertise and relationship-based judgment but does not by itself provide an explicit reconstruction test or mode-coverage stopping rule.","Clustering or principal-component summaries, which describe similarity or variance among recordings without modeling the direction and gain of successive retelling transformations."],"remaining_contrastive_claim":"The proposal is useful only if stable coupled directions predict held-out performance changes beyond individual features, clusters, and context labels, and if mode-guided authorized sampling improves reconstruction or coverage of community-designated variations relative to equal-count quota, maximum-difference, and random authorized sampling. Otherwise the archive should retain the simpler sampling process.","authority_safety":{"decision_authority":"Narrators control participation and performance-specific permissions; the community-designated authority or applicable rights agreement controls collective access and reuse; the archivist controls repository operations within those permissions. Analysts may recommend coverage questions but cannot authorize recording, access, publication, or interpretation.","authorized_first_step":"Analyze only existing recordings whose permissions explicitly cover the proposed read-only feature coding and aggregate audit, then submit the resulting coverage report for rights-holder review.","excluded_actions":["Requesting or inducing sacred, restricted, private, or otherwise prohibited narrative material","Prompting a narrator to imitate, correct, or reproduce a computed mode","Ranking narrators by authenticity, centrality, stability, or performance quality","Inferring community-wide prevalence or historical origin from the authorized sample","Publishing recordings, features, modal loadings, or contextual metadata beyond existing permissions","Generating synthetic performances or reconstructed canonical versions","Discarding singular recordings or residual material because it does not fit retained modes","Contacting participants or scheduling new sessions without the required individual and collective authorization"],"halt_rollback":"Halt if permissions are ambiguous, a rights holder objects, feature coding exposes restricted knowledge, comparable transitions cannot be defined, modes fail stability or spectral-separation thresholds, residuals contain protected material, or additional recording would distort customary performance conditions. Withdraw the sampling recommendation, preserve the original archive and access controls unchanged, and delete derived participant-linked data when required by the governing agreement."},"negative_tests":{"strongest_counterevidence":"Performance variation is principally singular or context-specific, held-out behavior is explained better by explicit occasion and narrator descriptions than by repeated modes, and governing tradition-bearers judge modal compression inappropriate for the narrative practice.","problem_falsifier":"No approximate coupled direction remains stable across coders, resampled recordings, transition definitions, and held-out performances, or the available recordings lack sufficiently comparable sequences for a local transformation.","intervention_falsifier":"Community-authorized sessions selected by the modal coverage map do not add held-out reconstruction value or community-designated variation beyond quota, random, or maximum-difference sampling, or the invitations measurably alter the performance conditions the archive intended to observe.","risks":["Fixed narratological coordinates may impose an external literary theory on the performances.","Alignment across recordings may erase performance-specific timing, voice, gesture, language, or audience interaction.","A small authorized sample may produce unstable modes that appear more general than they are.","Sampling invitations can change a living practice by signaling which variations outsiders value.","Community governance may be internally plural, making a single authorization path insufficient.","A low-amplitude residual may contain the most culturally consequential material.","Context variables may be inseparable from performer identity, creating privacy and reidentification risks.","A coverage objective may encourage collection when restraint or nonrecording is the appropriate archival decision."]},"next_evidence_step":"Subject to existing permissions, select no more than 24 recordings of one recognized narrative containing comparable repeated-performance sequences. Have two approved coders apply a community-reviewed feature rubric and retain time-coded provenance. Fit the transition model on 18 recordings and test modal stability, spectral separation, and reconstruction on six held-out recordings. Compare mode coverage with performer-occasion quotas, maximum-difference selection, clustering, and random selection restricted to the same authorized pool. Present only aggregate findings and residual examples cleared for review; request no new performance during this first evidence step.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"This proposal is independent of every earlier candidate. Proposal 1 changes the decision process for revising a creative manuscript by offering alternative edit batches when prose-feature modes threaten an author-defined ambiguity envelope; proposal 5 neither edits a fixed text nor regulates narrator reliability, but allocates culturally authorized observation of repeated performances. Proposal 2 orders verification of interdependent claims in a critical edition; proposal 5 does not analyze scholarly-note dependencies or factual corrections and instead tests transformation coverage in a performance archive. Proposal 3 interrupts a live classroom discussion when a conversational mode narrows evidence-backed interpretation; proposal 5 does not facilitate, monitor, or redirect a seminar and explicitly prohibits prompting performers to enact computed modes. Proposal 4 gates the contribution of unstable genre memberships to recalculated corpus prototypes; proposal 5 has no genre classifier, category prototype, or metadata-release feedback loop and preserves performance modes through sampling rather than classification control. Its governed object, rights structure, state representation, transformation, outcome, intervention, and causal path are distinct from all four prior proposals, and an archive can adopt it without adopting any manuscript, critical-apparatus, classroom, or genre-catalog system.","revision_record":{"parent_version":null,"progress_targets_addressed":[],"conceptual_changes":[],"operational_changes":[],"evidence_changes":[],"claim_changes":[]}}