{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp06_four_proposal_generalization60_20260803","cell_id":"bounded_rivalry_governance__mathematics","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_id":"math_foundation_interface_trial_v0","proposal_index":2,"version":0,"title":"Governed Foundation-Interface Trial for a Shared Mathematics Library","problem":"A consortium building a shared formal mathematics library must choose one default foundational interface and allocate its limited maintenance budget. Teams advocating incompatible foundations or encoding schemes compete for that position. Informal selection by installed corpus, persuasive demonstrations, or early migration can reward benchmark tailoring, conceal unsupported constructions, export translation costs to theorem authors, and give the selected team control over later compatibility rules. The resulting winner may become difficult to displace even if another interface later serves the library better.","actors":["Teams maintaining candidate foundational interfaces or encoding schemes","Consortium board controlling the shared library and maintenance allocation","Mathematicians contributing definitions and proofs","Independent semantic-reconstruction auditors","Translation and migration maintainers","Conflict-screened technical judges","Independent procedural appeal reviewer","Downstream readers and maintainers of the shared corpus"],"observable_state":"The consortium can observe each candidate's versioned specification, declared supported constructions, translated definitions and proofs, reconstruction results, round-trip discrepancies, unsupported cases, adapter code and documentation, judge scores and rationales, clarification contacts, maintenance hours, migration failures, benchmark revisions, governance permissions, and the selected interface's later share of consortium-controlled artifacts. Private research communications remain outside monitoring unless voluntarily submitted as evidence of a specific process violation.","consequence":"A default can be selected for its incumbent advantage or benchmark presentation rather than semantic coverage and reversible interoperability. Contributors using other foundations then bear conversion losses and maintenance work, while the winner's control of the default interface can foreclose later challengers and turn a revisable technical choice into durable institutional lock-in.","affected_objective":"Select and periodically reassess one default foundational interface for the consortium-controlled library while preserving mathematical meaning across translations, bounding migration burdens, and keeping future interface competition practically possible.","intervention":"Create a time-bounded foundation-interface trial with one default-interface designation and a fixed maintenance allocation as the scarce prize. Before identifying entrants, the consortium freezes an RFP-style rulebook covering eligibility, permissible assistance, a neutral interchange specification, semantic gates, secondary scoring, disclosure, conflicts, appeals, and transition duties. Each eligible team must provide an inspectable specification, a reversible export path, governance disclosures, and translations of the same public corpus under identical compute, documentation, and clarification allowances. Independent auditors test reconstruction on both the public corpus and a withheld corpus selected before judging. Any semantic mismatch, undeclared unsupported construction, or unreconstructable required proof fails a noncompensable gate. Passing candidates are compared on coverage, translation loss, maintainer effort, readability, extensibility, and independence of rule governance. A retained portion of the maintenance allocation is released only after the winner supplies the promised adapters, documentation, and exit package. The winner cannot control the benchmark, judges, appeal process, or challenger eligibility. A scheduled challenger window and an earlier trigger for material interoperability failure allow a qualified alternative to contest the default under the same rules.","structural_mapping":[{"archetype_element":"Rivalry purpose statement","domain_realization":"Use comparative pressure to expose how candidate foundations handle the same mathematical corpus and migration obligations, rather than treating installed base or advocacy as evidence of suitability."},{"archetype_element":"Scarce prize or selection constraint","domain_realization":"One default interface in the consortium-controlled library and one bounded maintenance allocation for the award period."},{"archetype_element":"Competitor eligibility boundary","domain_realization":"A candidate must have an inspectable specification, disclosed maintainers and conflicts, a working import-export path, and capacity to complete the common trial without receiving private judge assistance."},{"archetype_element":"Contest arena boundary","domain_realization":"Entrants may improve their interface, publish adapters, challenge corpus interpretations, collaborate transparently, or merge candidacies. They may not tamper with rival artifacts, misstate unsupported constructs, coordinate sham submissions, obtain withheld cases, contact judges privately, or condition interoperability on control of future rules."},{"archetype_element":"Performance metric and scoring basis","domain_realization":"Semantic reconstruction and declared coverage form noncompensable gates; only passing candidates receive secondary scores for translation loss, maintainer burden, readability, extensibility, and governance independence."},{"archetype_element":"Fair process and due process layer","domain_realization":"The frozen corpus rules, identical resource allowances, recorded judge rationales, conflict screening, common clarification channel, and separate procedural appeal make differential treatment contestable."},{"archetype_element":"Anti-sabotage and anti-collusion guardrail","domain_realization":"Versioned artifacts, benchmark-access logs, contributor disclosures, reproducibility checks, and graduated sanctions address tampering, leaked tests, sham rivalry, and selective interoperability without treating ordinary technical similarity as proof of misconduct."},{"archetype_element":"Externality and spillover boundary","domain_realization":"Translation failures, downstream maintainer hours, unsupported constructions, and exit costs are measured as part of the trial rather than left to nonwinning foundation communities."},{"archetype_element":"Escalation and arms-race damper","domain_realization":"Identical corpus, compute, documentation, revision, and clarification allowances prevent unlimited demonstration engineering or judge access from becoming the winning strategy."},{"archetype_element":"Winner power and lock-in review","domain_realization":"The selected team maintains an interface but cannot rewrite eligibility or benchmarks, must preserve exportability, and faces a real challenger window at the end of the award period or after a material gate failure."},{"archetype_element":"Learning and recalibration loop","domain_realization":"A post-period review compares trial scores with actual migration defects, maintenance effort, governance concentration, and contributor burdens before the next rulebook is approved."}],"mechanism_mapping":[{"mechanism_slug":"tender_or_rfp_process","role":"Publishes the mathematical requirements, neutral interchange specification, eligibility conditions, evaluation criteria, judge separation, and protest path before candidate submissions are evaluated.","counterfactual_removal":"Without the tender structure, the consortium could tailor requirements to the incumbent interface, compare non-equivalent demonstrations, or select through undocumented technical preference."},{"mechanism_slug":"contest_rulebook","role":"Freezes legal moves, common resource allowances, semantic gates, secondary scoring, tie-breaks, disclosures, sanctions, and appeal deadlines for entrants and judges alike.","counterfactual_removal":"Without a binding rulebook, benchmark interpretations and permissible assistance could change after performance or entrant identity becomes visible."},{"mechanism_slug":"ranked_leaderboard_with_audit","role":"Makes comparative results inspectable while requiring independent reconstruction on withheld artifacts before a passing candidate can receive the default designation.","counterfactual_removal":"Without the audit, entrants could optimize visible examples, omit difficult constructions, or overstate reversible translation while retaining a high published score."},{"mechanism_slug":"externality_bond_or_liability_rule","role":"Retains part of the winner's consortium-funded maintenance allocation until required adapters, documentation, defect remediation, and an usable exit package are delivered.","counterfactual_removal":"Without retained security, the selected team could receive the full prize while leaving translation failures and later migration costs to contributors and consortium maintainers."},{"mechanism_slug":"challenger_access_window","role":"Reopens the default designation at a fixed cadence and after material interoperability failure, using published qualification criteria and a predeclared unseat rule.","counterfactual_removal":"Without a credible reopening path, corpus accumulation around the initial winner could make later comparison ceremonial even when an alternative becomes demonstrably more suitable."},{"mechanism_slug":"antitrust_or_competition_review","role":"Examines whether the winner's control of adapters, benchmark admission, metadata, or required tooling has foreclosed future entrants, and can require open access or separation of rulemaking from maintenance.","counterfactual_removal":"Without winner-power review, a successful entrant could convert the technical award into control over the arena needed by every future challenger."},{"mechanism_slug":"post_contest_impact_review","role":"Compares promised interoperability and maintenance properties with realized corpus migrations, contributor burdens, exit readiness, and governance concentration before another award period.","counterfactual_removal":"Without ex-post review, a metric that rewarded benchmark-specific performance or underestimated switching burdens could remain unchanged in later trials."}],"causal_chain":["A single default interface and bounded maintenance allocation create strategic dependence among incompatible foundation teams.","A frozen common trial converts informal installed-base competition into a bounded comparison over the same mathematical artifacts and obligations.","Noncompensable semantic and reconstruction gates prevent presentation quality, incumbent corpus size, or low apparent cost from offsetting loss of mathematical meaning.","Withheld audit cases reduce the value of tailoring an interface only to published examples.","Equal trial-resource allowances redirect competitive effort from demonstration scale and private judge access toward coverage, reversible translation, and maintainable specification.","The retained maintenance allocation makes promised adapters and exit readiness part of the winner's payoff rather than an unfunded burden on outsiders.","Separation of interface maintenance from benchmark and eligibility control limits the winner's ability to govern future rivals.","The challenger window preserves a practical route to replacement after network effects begin accumulating around the default.","Post-period evidence on defects, burdens, and concentration informs revision or retirement of the trial."],"baseline":"The consortium informally adopts the interface with the largest immediately usable corpus or the maintainers most able to support an early migration. Demonstrations use different examples, interoperability promises are not secured, selection rationales are not appealable, and the chosen maintainers participate in later compatibility decisions.","nearest_rivals":["Neutral architecture-board selection: independent experts choose a default through holistic deliberation without organizing candidate teams into a contest. This can accommodate foundational differences that resist common scoring but supplies less direct comparative evidence and no inherent challenger discipline.","Permanent pluralism with no default: the library supports every qualifying foundation through adapters. This avoids crowning a winner but may exceed the bounded maintenance budget and leave responsibility for cross-foundation coherence diffuse.","A common interchange layer with no foundation selection: all systems translate through a jointly governed neutral representation. This cooperative alternative is preferable if the interchange layer can carry the required mathematical meaning without privileging a foundation or recreating a default indirectly.","One-time technical benchmark: candidates translate a shared corpus and the top score becomes permanent. This is simpler than the proposed arena but omits secured exit duties, procedural appeal, winner-power review, and reopening after lock-in begins."],"remaining_contrastive_claim":"The candidate's testable structural distinction is the combination of a scarce default designation, semantic gates, common trial constraints, secured interoperability duties, separation of winner maintenance from arena rulemaking, and a credible reopening path. It does not assert that competitive selection is preferable to pluralism, cooperative interchange, or expert deliberation in any particular library.","authority_safety":{"decision_authority":"The consortium board may choose the default interface, allocate its own maintenance funds, and govern artifacts and services it controls. Auditors may determine trial compliance and reconstruction results; an independent reviewer may remedy procedural error. No actor in the trial may declare one foundation universally correct, compel external mathematicians to migrate, restrict publication, or control attribution for imported mathematical work.","authorized_first_step":"Authorize only a sandboxed shadow trial on copied, nonproduction artifacts using anonymized candidate labels. Do not migrate the live corpus, publish reputational rankings, award maintenance control, or alter external contributors' access.","excluded_actions":["Forcing authors or external repositories to adopt the selected foundation","Representing the winner as mathematically superior outside the trial's stated corpus and criteria","Allowing the winning team to select judges, reveal withheld cases, determine appeals, or set challenger eligibility","Using private-message surveillance or speculative similarity as proof of collusion","Withholding attribution, publication permission, or existing library access from a losing entrant","Accepting semantic loss in exchange for a higher secondary score","Releasing the retained maintenance allocation before independently verifying adapters and exit materials","Changing corpus inclusion, scoring, or disqualification rules after candidate identities or results are known"],"halt_rollback":"Pause the trial after benchmark leakage, judge conflict, unequal resource access, ambiguous semantic criteria, or a reconstruction discrepancy capable of changing eligibility. Preserve all versioned artifacts, replace conflicted reviewers, and rerun every affected candidate under the same corrected rule or void the shadow round. In a later live trial, a failed semantic gate suspends new default migrations and restores the last verified exportable state; it does not automatically transfer control to another entrant."},"negative_tests":{"strongest_counterevidence":"The strongest counterevidence would be that a neutral interchange layer or independent architecture board achieves equivalent semantic coverage and exit readiness with less governance and migration work, while the trial induces benchmark tailoring, discourages cross-foundation cooperation, or entrenches the candidate already best resourced to satisfy the submission format.","problem_falsifier":"The inferred problem is falsified if the consortium has no need for a default, can maintain all qualifying foundations without material tradeoffs, can switch interfaces without semantic or maintenance cost, or already separates default maintenance from benchmark and challenger governance under an effective recurrent review.","intervention_falsifier":"The intervention is falsified for its intended mechanism if reconstruction gates cannot compare candidates without assuming the disputed foundation, withheld-corpus results reverse under reasonable corpus substitutions, equal trial allowances are not enforceable, the retained allocation does not secure usable exit artifacts, or a qualified challenger could not practically replace the selected interface despite satisfying the unseat rule.","risks":["The neutral interchange specification may encode one contender's foundational assumptions and make the contest circular.","A withheld corpus may weaken transparency or leak through repeated trials.","Secondary metrics may reward benchmark-specific engineering rather than general mathematical expressiveness.","Resource caps and retained funding may exclude small teams even when their interfaces are suitable.","The contest may redirect effort from cooperative translation infrastructure into positional advocacy.","Foundational differences may be partly incomparable, making a single ranking misleading.","Required exportability may be technically satisfied while remaining too burdensome for actual maintainers.","Frequent challenger windows may create migration churn, while infrequent windows may permit lock-in.","Competition review may be applied selectively against an unpopular winner rather than tied to observable foreclosure.","Anonymized candidate labels may fail to conceal recognizable formal styles."]},"next_evidence_step":"Pre-register a shadow trial using a copied corpus of 24 nonproduction artifacts spanning definitions, theorem statements, proof dependencies, notation overloading, and foundation-sensitive constructions. Freeze a public subset, a withheld audit subset, the neutral interchange rules, a reviewer-hour ceiling, and a zero-tolerance semantic-mismatch gate before anonymized candidate packages are loaded. Have separate teams perform translation, semantic reconstruction, and maintenance handoff, then rotate a portion of the withheld corpus to test ranking stability. Stop without a live pilot if any passing candidate has an unreconciled semantic discrepancy, results depend materially on one reasonable corpus rotation, auditors cannot apply the gates independently, or the shadow exercise exceeds its preregistered governance budget.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"Proposal 1 governs rival research teams seeking one proof-verification slot for competing routes to the same theorem; its central hazards are premature completeness claims, concealed proof dependencies, and reviewer-facing escalation, and its causal path runs through a correctness gate followed by route selection. This proposal instead governs rival foundational interfaces seeking default status in a shared library; its central hazards are semantic translation loss, migration externalities, rule control, network-effect lock-in, and foreclosure of later interfaces. Its intervention is an interoperability trial with secured exit duties, separation of winner maintenance from arena governance, competition review, and recurrent challenge. It can be adopted without running a proof-route challenge, and proposal 1 can be adopted without selecting a library foundation.","revision_record":{"parent_version":null,"progress_targets_addressed":["Initial complete formulation for proposal index 2","Material separation from sealed proposal index 1"],"conceptual_changes":["Instantiated bounded rivalry around selection of a shared library's default foundational interface rather than selection of a proof route.","Made semantic interoperability, reversible migration, and future contestability the governing objectives.","Located the principal externality in conversion and exit burdens imposed on nonwinning foundation communities."],"operational_changes":["Specified a common-corpus RFP, noncompensable reconstruction gates, withheld audits, equal trial resources, retained maintenance funding, rulemaking separation, and a challenger window.","Restricted the first evidence step to an anonymized nonproduction shadow trial with rollback conditions."],"evidence_changes":["Defined observable semantic, migration, maintenance, and governance records.","Specified bounded tests for corpus dependence, reconstruction consistency, exit readiness, and governance overhead.","Kept prior art unsearched and made no novelty, prevalence, demand, or effect-size claim."],"claim_changes":["Limited the contrastive claim to the proposal's governance structure and falsifiable mechanism.","Explicitly treated expert selection, permanent pluralism, and cooperative interchange as potentially preferable alternatives."]}}