The Rule Change

Bilateral decisions become Hollow Wins when the welfare object is outside the rule

The paper defines the Rule Change as a move from G to G1: the old game approves private bilateral gain while W is absent from the operative decision record; the new game requires a reconstructable Field 17 record and an enforcement lane that can act on it.

• Hollow Win: A and B gain while system welfare falls, coded as (0,1,1).
• Aggregate calibration: betaW = 3.45 from 100,000 Monte Carlo draws with seed=42.
• Headline estimates: system-adjusted payoff = -61.3T and aggregate reform dividend =$73.8T per year across covered decision classes.

The paper’s failure mode is welfare exclusion, not bad intent

The Missing System Theorem says A and B can select an action xH because private payoff P is counted while the welfare loss W is outside the decision rule. SAPM converts that excluded loss into a measurable system cost.

• Formal condition: P_a(xH) and P_b(xH) rise relative to baseline x0 while W(xH) falls.
• The paper links the loss term to Ls, where systemic loss can exceed the private gain through risk multipliers and system friction.
• The Satterthwaite bridge is specific: even truthful bilateral revelation can still produce a destructive optimum when C is absent.

A rule change works only when the record is reconstructable and the lane changes payoff

Proposition 1 is deliberately a weak-reduction claim. A decision-reconstructable and conflictoring-sufficient rule change reduces the feasible set of Hollow Win actions; if it makes one formerly profitable Hollow Win privately dominated, G1 admits a higher-welfare equilibrium.

• Decision-reconstructable means the actor must quantify, justify, mitigate, or falsely deny the effect on W.
• Conflictoring-sufficient means the record can trigger denial, liability, contract breach, capital penalty, procurement exclusion, insurance repricing, enforcement, or reversal.
• Disclosure alone fails when it does not change the approval predicate, expected sanction, or feasible action set.
• A voluntary safety or ethics framework remains in G when the actor keeps unilateral discretion.

Field 17 turns the welfare object into evidence before execution

Decision Accounting is a 17-field structured decision record. Fields 1-15 capture ordinary decision metadata; Field 17 is the Systemic Externality Accountability record that makes W part of the approval path.

• Fields 1-15 include identity, timestamp, inputs, and algorithmic weights; the paper ties them to regulatory convergence at alpha=0.91 across 16 regulatory regimes.
• Field 17 must name the system-affected entity, direction, magnitude, probability, time horizon, method, confidence interval, responsible party, pricing statement, and signature.
• Boilerplate fails the theorem if it does not name a welfare object, evidence, alternative, and reversal trigger.

The rebate-for-access game rewarded high list prices and shifted cost to patients

In the PBM case, Caremark Rx, Express Scripts, and OptumRx control roughly 80% of U.S. prescriptions. The paper treats the rebate-for-access model as a Hollow Win: manufacturers raised list prices to fund larger rebates, PBMs granted formulary access, and patients with deductibles or coinsurance paid against the inflated list price.

• Case betaW = 3.30 with illustrative system cost of $52.5B per year.
• Novo Nordisk’s Novolog U-100 list price rose from 122.59 in 2012 to 289.36 in 2018.
• Cheaper alternatives were excluded when they produced lower PBM rebates.
• Rule-change target: delink PBM compensation from list price and require formulary Decision Accounting records.

The Express Scripts settlement changes the insulin approval path, not merely the disclosure level

The paper uses the FTC’s PBM action as a regulatory lane example. The settlement’s importance is structural: it attacks the list-price compensation link and the formulary preference rule that made the old game profitable.

• FTC lawsuits against the Big Three PBMs are dated 2024 and 2026 in the paper.
• The February 2026 Express Scripts settlement bars preferencing high-list-price drugs over identical low-list-price versions.
• Express Scripts must offer a standard model basing patient out-of-pocket expense on net cost, not inflated list price.
• Estimated patient savings: up to $7B over 10 years.

Issuer-pays ratings made inflated risk certification privately valuable

The CRA case maps the Hollow Win to ratings inflation. Banks wanted AAA labels for RMBS and CDO products; S&P, Moody’s, and Fitch were paid by issuers; the system absorbed the downstream market fragility.

• Case betaW = 5.50 with illustrative system cost of $57.5B per year.
• The paper identifies issuer-pays as the old-game payment incentive that rewards inflated certification.
• Rule-change options: reassign payment incentives or require auditable model override records.
• A Field 17-style record would log manual overrides, client pressure, model exceptions, and the attesting person.

Post-crisis settlements left the issuer-pays game mostly intact

The paper’s lesson from the CRA case is that fines after failure do not necessarily alter G. The relevant question is whether a record and lane can change ratings approval before inflated certification reaches the market.

• DOJ settlements named in the deck: S&P paid 1.375B in 2015 and Moody’s paid 864M in 2017.
• The issuer-pays incentive remains the mechanism to attack, because the buyer of the rating is also the seller of the rated product.
• A subscriber-pays model would alter payment incentives; override logging would alter reconstructability.
• Without a live lane, disclosure can document the old game without changing it.

Delegated certification failed at the safety edge where MCAS and training exceptions mattered

The Boeing case applies the theorem to safety certification. Under FAA ODA delegation, Boeing held certification authority for key work; MCAS risk and simulator-training implications were not independently reconstructed before two crashes killed 346 people.

• Case betaW = 7.15 with illustrative system cost of $50.0B.
• The paper’s rule-change target is Field 17 safety records for delegated sign-offs and training exceptions.
• The record must attach the welfare object to the safety decision: passenger life, crash probability, training burden, evidence, alternative, and reversal trigger.
• A whistleblower, prosecutor, regulator, insurer, or procurement buyer can be the lane that changes the expected payoff of deficient certification.

The summary table applies the same rule to release gates for frontier AI

The paper’s summary table extends the framework beyond the three main cases. Frontier AI is treated as a covered decision class where capability competition can reward fast deployment under unresolved risk.

• Table betaW = 4.60 with illustrative system cost of $275.0B.
• Old-game failure: capability competition rewards fast deployment under unresolved risk.
• Rule-change implication: require release-gate Decision Accounting records tied to evaluations, rollback triggers, and external lanes.
• This is an extension case in the findings table, not one of the three detailed empirical case sections.

The record changes behavior only when someone can use it

Conflictoring is the enforcement side of the Rule Change. The paper names six agents that can move G to G1 by making the Field 17 treatment costly to omit, falsify, or ignore.

• Whistleblower: exposes pressure, omissions, or false certification in the record.
• Plaintiff or prosecutor: converts deficient records into liability, fraud, or criminal exposure.
• Regulator or legislator: changes the approval predicate through denial, rulemaking, settlement, or statutory mandate.
• Investor, insurer, or supranational actor: reprices capital, coverage, procurement, or market access.

Mandate Field 17 where betaW is high and the decision predicate is observable

The paper’s policy claim is institutional: high-betaW domains need Decision Accounting, a standard betaW metric, and enforcement lanes strong enough to change private payoff before execution.

• Systemically important institutions named in the paper include PBMs and sole-source aerospace contexts.
• For PBMs: delink compensation from list price and require formulary DA records.
• For CRAs: reassign payment incentives or require auditable model override records.
• For Boeing-style delegation: require explicit system-safety accountability for delegated sign-offs and training exceptions.