Final Validation Report: LG21 Test-Optional Policies

Updated: 2026-09-06

1. Human Verdict

The fairness and policy results are proved on the stated policy domains.

Formalization gap: Theorem 3.1 assumes stability against profitable group entry; voluntary Lemma 4.1 assumes maximal participation. Deriving these refinements from the paper’s equilibrium definition remains unproved. Details.

2. Closeout Status

3. Source and Scope

4. Researcher Summary of Checked Results

Result Comparison with the paper
Fairness definitions and implications; score-ignoring policy Exact.
Theorem 3.1 Formalization gap: assume stability against profitable group entry with recalibrated school estimates; deriving this from source equilibrium remains unproved. Necessity is unknown.
Theorem 3.2 Restricted scope: deterministic reported output; the memo’s randomized counterexample refutes the unrestricted conclusion. Determinism is sufficient, not shown necessary.
Theorem 3.2 summary Typo fixed: “demographic” becomes “observable,” matching the argument.
Lemma 4.1, voluntary regimes Formalization gap: select self-enforcing participation that no admissible candidate can strictly enlarge. The source equilibrium definition has not been shown to imply this selection; necessity is unknown.
Proposition 4.2 Exact: the Gaussian observed-score model allows any base-only policy for students without access.
Proposition 4.3 Exact conclusion: an unfair equilibrium in each regime refutes fairness, which the paper requires in every equilibrium. The variance comparison uses a direct calculation; voluntary equilibria use the Lemma 4.1 convention.
Lemma 4.1 cutoff Typo fixed: c=(qtilde-intercept)/slope on the positive-slope domain.
Theorem 4.4 and its reporting-conditioned generalization Exact for mandatory participation; restricted voluntary equilibria: the latter use the same maximal-participation selection as Lemma 4.1. Necessity is unknown.

5. Remaining Boundaries and Gaps

Theorem 3.1’s stability against recalibrated group entry and the voluntary Lemma 4.1 maximal-participation restriction are not derived from the paper’s equilibrium definition. The corresponding conclusions for arbitrary source equilibria remain unproved. Selected equilibria do exist; the gap is extending the conclusions to the source equilibrium class.

The fixed-point argument also retains the analytic premises described in the memo; their source-model derivation remains open.

6. Additional Assumptions Beyond Paper

Theorem 3.2 restricts reported output to be deterministic; its randomized example shows that some restriction is needed, but not that determinism is necessary. The access law uses independence from the skill/noise block, beyond uncorrelated-access wording; necessity under the complete source model is unresolved. The equilibrium proof gaps have their main discussion in Section 5.

7. Proof-Strategy Deviations

The Theorem 3.1 nonreport-mixture correction and Proposition 4.3 unconditional-variance calculation are detailed in the memo.

8. Proof Tricks Worth Reusing

9. Generalizations, Conjectures, and Extensions

The positive-mass active-branch framework may be useful beyond Gaussian signals. No broader posterior calculation or unproved generalization is claimed here.

10. Source Clarifications and Exact Readings

The memo gives the nonreport-mixture and cutoff correction, the randomized-policy example and checked operational blankness conclusion, and the unconditional Gaussian variance calculation.

11. Paper Issues or Caveats

The equilibrium coverage gaps are in Section 5. Section 6 distinguishes the supported randomized-output obstruction from conditions whose necessity is unknown.

12. Detailed Formalization Evidence

PaperInterface.lean supplies the 15 selected transparent source-facing result specifications, and ProofInterface.lean supplies the corresponding checked endpoints. The selected surface covers the fairness implication, hidden- and observed-access protocols, optional and required reporting schedules, Theorems 3.1–3.2 and 4.4, and the reporting-conditioned supporting claims.

13. Paper Assumption Provenance

The fifteen paper prerequisites and one library prerequisite match their selected source connections. Result-specific assumptions and the equilibrium restrictions remain visible in the transparent specifications and Sections 4–6.

14. Displayed Formula Provenance

The statement map routes the paper’s access, signal, reporting, fairness, cutoff, and equilibrium formulas, including 19 formula presentations and two equation presentations. The four clarified Theorem 3.2 targets are identified in the source-to-Spec ledger and explained in the source clarification memo.

15. Library Lift Pass

The reusable Gaussian signal-family primitive is source-screened in the library ledger. Paper-specific equilibrium, policy, access, reporting, and fairness statements remain in the paper layer.

16. DAG Audit

DependencyDAG.pdf, generated from DependencyDAG.tex, distinguishes the source model, hidden-access and observed-access result families, the reporting-conditioned generalization, and open directions. The rendered PDF was visually inspected for readability and arrow/box overlap.

17. Validation Checks

The focused-build receipt records the paper build, and the import-closure receipt records its Lean dependencies. The closeout record records acceptance evidence for its pinned inputs.

18. Paper Definitions Checked

The checked definitions cover access actions and observation regimes, the hidden-access optional protocol, latent-skill, observable, demographic, and test-blank fairness, the access estimator, distributional equality, requirement policies, school information, the Gaussian student signal model, and the cutoff and local-candidate models used by Theorems 3.1–3.2. Exact routes are in the statement map.

19. Named Theorem Statements Checked

Source presentation Transparent specification Proof endpoint
Fairness implication chain fairness_implication_chainSpec fairness_implication_chain_proof
Lemma 4.1 lemma4_1_observed_access_strategy_proofness_source_coreSpec lemma4_1_observed_access_strategy_proofness_source_core_proof
Proposition 4.2 proposition4_2_all_observed_access_requirement_protocols_source_coreSpec proposition4_2_all_observed_access_requirement_protocols_source_core_proof
Proposition 4.3 proposition4_3_each_requirement_protocol_has_unfair_pbo_equilibriumSpec proposition4_3_each_requirement_protocol_has_unfair_pbo_equilibrium_proof
Reporting-conditioned generalization reporting_conditioned_generalizationSpec reporting_conditioned_generalization_proof
Theorem 3.1: fairness failure theorem3_1_hidden_access_pbo_fails_all_fairness_definitionsSpec theorem3_1_hidden_access_pbo_fails_all_fairness_definitions_proof
Theorem 3.1: optional reporting theorem3_1_optional_reporting_source_timedSpec theorem3_1_optional_reporting_source_timed_proof
Theorem 3.1: reporting required theorem3_1_report_required_source_timedSpec theorem3_1_report_required_source_timed_proof
Theorem 3.2: optional schedule theorem3_2_optional_reporting_clarified_modelSpec theorem3_2_optional_reporting_clarified_model_proof
Theorem 3.2: report-required schedule theorem3_2_report_required_clarified_modelSpec theorem3_2_report_required_clarified_model_proof
Theorem 4.4 theorem4_4_all_observed_access_requirement_protocolsSpec theorem4_4_all_observed_access_requirement_protocols
Threshold acceptance interpretation thompson_acceptance_interpretationSpec thompson_acceptance_interpretation_proof
Report-rule skill independence report_decision_skill_independenceSpec report_decision_skill_independence_proof
Ignoring-test-scores witness ignoring_test_scores_achieves_fairnessSpec ignoring_test_scores_achieves_fairness_proof
Theorem 3.2 observable summary theorem3_2_observable_summary_clarifiedSpec theorem3_2_observable_summary_clarified_proof

20. Paper-Facing Statement Validator Ledger

The source-to-Spec ledger contains 15 current judgments: 12 matches and three corrected-target matches for the clarified Theorem 3.2 rows. The human review packet presents the same selected claim surface.

21. Source-Coverage Audit Ledger

The statement map retains 81 source items, including supporting model and formula presentations. Fifteen result claims, fifteen paper prerequisites, and one library prerequisite form the current selected semantic comparison; the same source scope appears in the packet.