{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp06_four_proposal_generalization60_20260803","cell_id":"representation_independent_interface_contract__sociology_anthropology","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_id":"attested_kin_relation_query_contract","proposal_index":2,"version":0,"title":"Attested Kin-Relation Query Contract for Re-Representable Fieldwork Corpora","problem":"A longitudinal fieldwork team records kin relations in changing technical forms: pedigree diagrams, person-by-person tables, graph databases, coded notebooks, and locally defined relationship vocabularies. Analytical scripts and researcher handoffs often rely on representation details such as row order, identifier format, a particular graph edge direction, or the convention that an empty cell means no relationship. When the corpus is migrated or a new backend is introduced, those dependencies can change who appears related, erase contradictory accounts, confuse unknown with explicitly denied, or alter the paths counted under a descent or affiliation rule even though the underlying attested claims have not intentionally changed.","actors":["Field researchers recording relation claims","People and community knowledge holders contributing accounts","Fieldwork data stewards","Kinship and social-organization analysts","Research software maintainers","Study-level ethics and data-governance representatives"],"observable_state":"For a bounded corpus, analysts can observe direct relation claims, their source and observation period, whether they remain active or have been superseded, contradictions among accounts, and paths derived under a named analytical rule. Representation coupling is observable when equivalent corpora stored as a table and a graph return different relation sets, when migration changes results because identifiers or edge directions changed, when a blank becomes a negative claim, or when an analyst must inspect backend fields to interpret a query result.","consequence":"A storage migration or implementation replacement can silently alter genealogical paths, group-affiliation counts, or case selection; contradictory testimony can be overwritten; and analysts may mistake technical conventions for ethnographic claims. Because no representation-independent oracle exists, discrepancies are difficult to classify as substantive revisions, mapping errors, or accidental backend behavior.","affected_objective":"Preserve the declared meaning, uncertainty, provenance, and derivation behavior of attested kin-relation data while allowing the corpus storage model, indexing method, and query implementation to change.","intervention":"Create an opaque Attested Kin-Relation Corpus defined by behavior rather than by a genealogy schema or database layout. Its public operations are submit_claim, supersede_claim, query_direct_claims, derive_relation_paths, create_snapshot, and export_permitted_view. The abstract value is a versioned collection of attested claims, each connecting opaque participant references through a study-approved predicate and carrying provenance, observation time, epistemic status, and visibility classification. The contract treats a claim as an attributed observation rather than as an objective kinship fact. It specifies preconditions, postconditions, transition rules, errors, and side-effect limits. Its laws require that unknown, unasked, explicitly denied, and not-applicable remain distinct; contradictory active claims may coexist; supersession preserves history; queries do not mutate the corpus; failed submissions leave it unchanged; path derivations use only claims admissible under the named rule and return a derivation trace; and a snapshot queried under the same vocabulary and rule version returns the same abstract result regardless of storage representation or enumeration order. Tables, graph orientation, internal identifiers, indexes, caches, and physical ordering remain hidden. A shared conformance oracle tests every backend against synthetic corpora before it is accepted as a substitute. Study authorities retain control of predicate meanings and derivation rules, so the contract does not impose a universal kinship ontology.","structural_mapping":[{"archetype_element":"Abstract component","domain_realization":"A corpus of attributed, temporally situated kin-relation claims and the paths derivable from them under declared analytical rules."},{"archetype_element":"Observable operation surface","domain_realization":"Submit and supersede claims, query direct claims, derive paths, create stable snapshots, and export an authorized abstract view."},{"archetype_element":"Abstract state model","domain_realization":"Versioned active and superseded claims with explicit predicates, provenance, temporal applicability, epistemic status, and visibility classification."},{"archetype_element":"Behavioral laws and invariants","domain_realization":"Contradictions coexist, missingness categories remain distinct, supersession preserves history, queries are non-mutating, and every derived path has a rule-consistent trace."},{"archetype_element":"Hidden representation","domain_realization":"Graph direction, table layout, pedigree encoding, internal identifiers, indexing, caching, and row or edge order are non-contractual."},{"archetype_element":"Abstraction function","domain_realization":"Each valid backend state maps to the same collection of abstract attributed claims and derivable results, independent of storage layout."},{"archetype_element":"Conformance oracle","domain_realization":"Synthetic claim and query sequences are run against every backend using only public operations and contract-level observations."},{"archetype_element":"Substitutability rule","domain_realization":"A new backend may replace an existing one only when it preserves abstract snapshots, transitions, errors, and derivation traces for the same vocabulary and rule versions."},{"archetype_element":"Leakage control","domain_realization":"Audits test whether identifier shape, result order, error wording, timing, or exported metadata exposes storage conventions or protected participant information."},{"archetype_element":"Versioned stewardship","domain_realization":"Vocabulary and derivation-rule changes require substantive study governance, while storage and algorithm changes remain private when conformance is preserved."}],"mechanism_mapping":[{"mechanism_slug":"abstract_data_type_specification","role":"Defines the corpus as abstract attributed claims and operations, then requires each table, graph, or other backend to satisfy a representation invariant and map valid states to that meaning.","counterfactual_removal":"Without the abstract value model and mapping, compatibility would reduce to translating fields, leaving uncertainty, contradiction, and derivation semantics dependent on storage."},{"mechanism_slug":"design_by_contract_clause","role":"Specifies caller obligations and component promises for claim submission, supersession, querying, snapshots, errors, and failed operations.","counterfactual_removal":"Without operation-level clauses, backends could share method names while disagreeing about duplicates, malformed claims, missing values, or supersession effects."},{"mechanism_slug":"opaque_type_or_module_boundary","role":"Makes physical tables, graph edges, internal identifiers, and indexes unreachable to analysis clients, routing normal work through the declared operations.","counterfactual_removal":"Without opacity, analysts could continue joining on internal identifiers or interpreting raw edge orientation, recreating representation coupling."},{"mechanism_slug":"black_box_contract_test_suite","role":"Runs one implementation-independent battery of synthetic corpus histories and queries against every candidate backend.","counterfactual_removal":"Without a common acceptance oracle, a replacement could be judged by common examples while diverging on contradiction, missingness, supersession, or path derivation."},{"mechanism_slug":"property_based_conformance_test","role":"Generates claims, contradictions, supersessions, and operation sequences to check invariants and equivalence of abstract snapshots across backends.","counterfactual_removal":"Without generated sequences, the oracle may miss interaction failures that occur only after unusual combinations of revisions and queries."},{"mechanism_slug":"representation_leakage_probe","role":"Searches observable outputs for unpromised dependence on row order, graph traversal order, identifier encoding, cache behavior, or protected metadata.","counterfactual_removal":"Without leakage testing, hidden representation could remain observable enough for analysts to convert it into an unofficial query contract."},{"mechanism_slug":"abstraction_barrier_code_review","role":"Requires reviewers to determine whether every new analytic dependency concerns a promised claim behavior or an accidental backend property.","counterfactual_removal":"Without this review, useful implementation artifacts could become dependencies before the mechanical suite contains a test for them."},{"mechanism_slug":"semantic_versioning_and_deprecation_gate","role":"Separates compatible implementation replacement from changes to predicate meanings, missingness semantics, or derivation rules and stages substantive changes explicitly.","counterfactual_removal":"Without version governance, a backend release could change the meaning of a relation or path while presenting itself as a technical migration."}],"causal_chain":["Analytical clients are restricted to operations over attributed relation claims rather than physical genealogy records.","Explicit abstract states preserve provenance, uncertainty, contradiction, and supersession independently of table or graph layout.","Named, versioned derivation rules define which claims may contribute to an analytical path and require an observable trace.","Opaque boundaries prevent clients from treating identifier shape, edge orientation, or record order as ethnographic meaning.","The common conformance oracle compares backend behavior across claim histories and queries, rejecting replacements that change abstract results.","Leakage review and dependency-focused review identify unpromised observables before they become analytical assumptions.","A conforming backend can then be replaced without silently changing the corpus meaning exposed to researchers."],"baseline":"The fieldwork corpus exposes its current tables, graph schema, coded files, or pedigree structures directly to analysts. Migration relies on one-off conversion scripts and spot checks. Individual analyses decide independently what blanks, duplicate claims, contradictory accounts, edge directions, and revisions mean, so the current representation becomes an implicit contract.","nearest_rivals":["Mandate one normalized genealogy schema and require every field site and analysis to use it.","Preserve each source format and write a bespoke converter whenever an analysis or migration is needed.","Standardize on a graph database and let analysts depend directly on its query language and edge model.","Export periodic flat spreadsheets and have a senior researcher manually reconcile discrepancies before analysis."],"remaining_contrastive_claim":"Unlike a universal genealogy schema, the intervention standardizes observable claim behavior without requiring one physical representation or universal kinship ontology. Unlike bespoke converters, it supplies persistent invariants and a reusable acceptance oracle. Unlike direct graph access, it prevents database structure from defining ethnographic meaning. Unlike manual reconciliation, it makes transition and query obligations repeatable across implementations. The conditional structural claim is that if study-approved meanings can be stated at the abstract surface, conforming backends can change without clients depending on their representation.","authority_safety":{"decision_authority":"The study's authorized data-governance body, including community or field-partner representatives where applicable, decides which relation predicates are legitimate, how local concepts may be translated, which derivations are substantively valid, and what may be disclosed. Data stewards and software maintainers may enforce technical conformance but may not invent equivalences, resolve contested claims, or redefine relation meanings.","authorized_first_step":"An authorized research-methods and data-governance group may test a draft contract using entirely fictional people, claims, vocabularies, and corpus histories. It may compare prototype backends without ingesting or transforming live fieldwork records.","excluded_actions":["Treating an attributed claim as an adjudicated fact solely because it conforms to the interface","Inventing cross-cultural equivalences between relation predicates without substantive authorization","Automatically merging or deleting contradictory accounts","Inferring unrecorded sensitive relationships from identifiers or graph structure","Exposing participant identities or protected provenance in test fixtures or public outputs","Migrating live records or replacing a production backend during the first evidence step","Changing vocabulary or derivation semantics under the label of an implementation update","Using conformance results as ethical approval for a research analysis"],"halt_rollback":"Stop if authorized reviewers find that the abstract model collapses a locally meaningful distinction, converts uncertainty into absence, or requires disclosure of protected provenance. Discard the synthetic mappings, leave existing records and analysis paths unchanged, and return the disputed vocabulary or rule to substantive governance before further prototyping."},"negative_tests":{"strongest_counterevidence":"The strongest counterevidence would show that representation changes already preserve all contract-relevant claim and path results, while observed analytical disagreements arise from irreducibly contested ethnographic interpretation rather than storage coupling. In that case, an interface contract would formalize the wrong problem.","problem_falsifier":"For the examined corpus workflow, the problem is falsified if independently implemented migrations between materially different storage forms preserve direct claims, missingness distinctions, contradictions, supersession histories, and named-rule derivations without clients inspecting backend details or requiring representation-specific corrections.","intervention_falsifier":"The intervention is falsified if two backends can pass the stated oracle yet disagree on held-out contract-level queries; if a valid local distinction cannot map to the abstract model without semantic loss; or if routine analysis still requires access to hidden layout, identifiers, or traversal behavior.","risks":["The abstract vocabulary may flatten locally specific or contested relationship concepts.","Formal derivation rules may create unwarranted confidence in analytical paths assembled from uncertain claims.","Opaque storage may make data-quality defects harder to diagnose unless authorized inspection remains available.","A weak suite may certify two backends that share the same semantic defect.","Generated test cases may omit culturally important structures and produce misleading coverage.","Stable derivation traces may reveal sensitive relationships or contributor identities.","Version proliferation may make longitudinal comparison harder rather than safer.","Contract maintenance may shift uncompensated interpretive work onto field partners or community representatives."]},"next_evidence_step":"Run a bounded two-week synthetic comparison overseen by one fieldwork methodologist, one data steward, and authorized field-partner or community representation. Before building anything, define a fictional relation vocabulary, missingness distinctions, visibility classes, and two named path rules. Construct 15 fictional corpus histories covering duplicate claims, contradictory sources, adoption or caregiving predicates, multiple simultaneous affiliations, unknown values, explicit denial, temporal change, supersession, invalid references, vocabulary revision, snapshot replay, and restricted provenance. Implement one relational-table backend and one graph backend independently. Apply the same example-based and generated black-box tests, then test held-out histories and inspect outputs for ordering, identifier, provenance, and timing leakage. Record every divergent abstract result, unmappable distinction, test-suite gap, and requested reach into hidden state. Do not use real participants or treat completion as authorization for migration.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"Proposal 1 addresses normative control over whether archived ethnographic material may be used: its abstract object is a community authorization, its principal actors are community decision authorities and archive access stewards, and its causal path prevents local governance records from leaking into access decisions. This proposal addresses epistemic and analytical continuity when a kin-relation corpus changes representation: its abstract object is a collection of attributed claims, its principal operational actors are field researchers, data stewards, and analysts, and its causal path preserves claim, uncertainty, revision, and derivation semantics across storage backends. It neither grants nor enforces access permission, and it remains independently adoptable for a fieldwork corpus even if archive authorization continues under the baseline process. The shared representation-independent contract structure is the archetype, not a shared intervention or problem.","revision_record":{"parent_version":null,"progress_targets_addressed":["Second complete proposal","Materially different domain problem","Distinct intervention and causal path","Explicit contrast with proposal 1","Bounded first evidence"],"conceptual_changes":["Initial formulation of attributed kin-relation claims as a representation-independent behavioral object.","Separated claim conformance from factual adjudication and cross-cultural semantic authority."],"operational_changes":["Specified corpus operations, invariants, hidden representations, substitution criteria, governance limits, falsifiers, and a synthetic dual-backend trial."],"evidence_changes":["Kept prior art unsearched and restricted initial evidence to fictional corpora and non-live implementations."],"claim_changes":["Limited the claim to conditional preservation of observable corpus semantics and made no claim about novelty, prevalence, demand, or effect size."]}}