The domain theorems

Chapter 7 showed how a Hollow Win can be dissolved by pointing enough independent readers at it. Some Hollow Wins do not yield to that pressure, and the reason is worth pinning down before any reform is proposed. This chapter sorts the studied domains by one question: what holds the welfare loss in place? Start from the parent result. The Missing System Theory (Axioms 1-3; see Chapters 1-3) establishes that system welfare is not a function of the two parties' payoffs, so bilateral optimization cannot recover it¹, and the Hollow Win stays invisible to any approval rule that reads only private payoffs². The sorting tool is welfare-beta (written β_W): the welfare a domain destroys per dollar of annual industry revenue, β_W = ΔW / Π, where ΔW is the yearly system-welfare loss and Π is annual revenue on the same boundary (see Chapter 5). A domain whose measured β_W is already below 1 is a control — the welfare loss is smaller than revenue, and no floor is reached. Among domains with β_W at or above 1, the class turns on why β_W cannot be pushed below 1. If a hard physical, chemical, biological, thermodynamic, geological, or informational law sets the floor — so that no institutional reform, however strong, reaches below one — the domain is an impossibility. If instead the binding constraint is institutional and a working reform model already exists in at least one jurisdiction that could be borrowed, so the right rule change could push β_W below one, the domain is an intractability. Across the 61 studied-domain records the split is 14 impossibility, 34 intractability, and 13 control. The measured β_W tells you whether a floor is reached; the binding mechanism tells you whether that floor is a law of nature you can only manage or a fixable institution you can redesign.

Consider cement. Making Portland clinker — the fused nodules made by heating limestone that are then ground into cement⁵ — releases carbon dioxide as a chemical identity of the reaction⁶: CaCO₃ → CaO + CO₂ is conservation of mass. Procurement rules, carbon pricing, substitution, and demand reduction can each lower the damage, but none of them repeals the chemistry, so cement's welfare-beta stays above one at 3.21. That is what places cement in the impossibility class. A domain is impossibility when a hard physical, chemical, biological, thermodynamic, geological, or informational law sets the welfare floor: no institutional reform, however strong, can push the measured welfare-beta below the β_W = 1 line, because a law of nature — not a rule — blocks it. Fourteen of the sixty-one records fall here. The same shape recurs. Deep-sea mining, at β_W 8.63, mines nodules that are themselves the seafloor habitat, and that habitat re-forms only on million-year timescales, so extraction and habitat loss are one act no reform can separate. Antimicrobial resistance, at β_W 1.53, rests on an evolutionary floor: the molecular event that cures an infection is the same event that selects for the resistant strain, so the damage rises with use however the market is organized. In each case the obstacle is a natural constant, and reform can only manage the trajectory. Two boundary cases show what impossibility is not. Firearms, cybercrime, human trafficking, opioids, and credit ratings all carry very large welfare-betas, but each rests on institutions rather than physics and each has a reform model proven elsewhere, so they are intractability, not impossibility. And PFAS³, long used as the flagship persistence example, measures a welfare-beta of only 0.96 (interval 0.29 to 4.05): its carbon-fluorine bond is still one of the strongest in chemistry⁴, yet the measured welfare loss does not clear one dollar per dollar of revenue, so PFAS is a control record.

Firearms sit at the top of the panel at β_W 21.98, and yet firearms are not an impossibility. In 1996, after the Port Arthur mass shooting, Australia enacted the National Firearms Agreement and a national buyback, and firearm deaths fell sharply. The floor there was never physics; it was a set of laws and incentives, and another country had already shown the redesign works. That is the signature of intractability. A domain is intractability when the binding constraint is institutional — rules, incentives, contracts, market design, enforcement — and a working reform model already exists in at least one jurisdiction that could in principle be borrowed, so the right rule change could push the measured welfare-beta below the β_W = 1 line. Thirty-four of the sixty-one records fall here, including several of the largest welfare-betas in the panel. The examples run from the extreme to the marginal. Tobacco (β_W 6.50) has the WHO Framework Convention on Tobacco Control and the plain-packaging and excise regimes that have driven smoking prevalence down across dozens of countries. Cybercrime, human trafficking, and opioids each have enforcement, disruption, and treatment models that have measurably reduced harm somewhere. Nearer the line, pharmacy benefit managers, fisheries, tax havens, and alcohol are held in place by vertically integrated intermediation, open-access subsidies, cross-border tax competition, and distribution-and-lobbying structures. None of these floors is a law of physics; each is a game that has not been redesigned at home even though a redesign has worked abroad. The reform is available in principle. The open question — the one that separates intractability from impossibility in practice — is whether the political and economic conditions for adoption can be assembled.

The control group is the panel's honesty check. A domain lands here when its measured median welfare-beta is already below 1 — the yearly welfare loss is smaller than annual industry revenue on the matched boundary, so the domain sits below the β_W = 1 line without any reform at all. Thirteen of the sixty-one records qualify, and their job is to show the method can return a below-one result when the measured loss is small, rather than stamping every studied domain a floor. A control record can still name a real physical or institutional mechanism; what it lacks is magnitude. Nuclear power measures welfare-beta 0.50¹⁶ even though its waste half-lives are fixed by nuclear physics; gig-economy platforms measure 0.39¹⁷; and PFAS — long treated as the flagship persistence example — measures 0.96 (interval 0.29 to 4.05), so it belongs here rather than in the impossibility class, despite its near-indestructible carbon-fluorine bond. Others include oil and gas (0.84), deforestation (0.88), ultra-processed food (0.83), private equity in healthcare (0.61), arms exports (0.54), aviation (0.50), defense procurement (0.47), and algorithmic pricing (0.30). That completes the three classes — impossibility, intractability, and control — and reduces classification to a short sequence of questions. Is the measured β_W already below 1? Then control. If not, does a natural law hold the floor above one? Then impossibility. Otherwise the constraint is institutional with a reform model proven in another jurisdiction, and the domain is intractability.

Add the domains up and the scale is hard to hold in the head: $69.0 trillion in system welfare destroyed each year, estimated across the 58-domain ranked revenue-ratio portfolio, and traceable to one structural mechanism — the Missing System Theory (MST) — rather than to a heap of unrelated regulatory failures¹⁸. The portfolio-level β_W is 2.61, still the average ratio ΔW/Π, with Π the annual industry revenue on the same boundary. The total is easier to read once it is split by where the damage lands. Mortality, valued at the income-adjusted value of a statistical life, is the largest channel at $15.8T/year, followed by health and disease ($12.7T/year), social and informational harm ($10.1T/year), climate and carbon ($8.4T/year, counted once), environmental and ecosystem loss ($6.5T/year), economic extraction and deadweight loss ($6.8T/year), governance and systemic failure ($5.4T/year), and lost productivity ($3.3T/year). The portfolio loss field is comparable in scale to global GDP but not identical to it.

Imagine a regulator holding every transaction record a firm produces, asked one question: can you certify that this portfolio did no harm to the wider system? Propositions 8a–8b, working-paper results under the stated MST observation boundary, say the honest answer is no — not for want of effort, but because the omitted coordinate, system welfare, is not derivable from the parties' payoffs. The consequence reaches every tool built on that data. Existing regulatory, auditing, and certification frameworks read observable private payoffs, so they are structurally incapable of detecting or preventing Hollow Wins. This is not a claim that regulation is useless; it is a claim about one specific limit — regulation built from transaction data cannot certify welfare neutrality, and closing that gap takes a new framework, the System Asset Pricing Model (SAPM).

A theory earns its keep by being tested, and this one has been, on public market data. The result is mixed but theoretically cleaner than a clean sweep would have been: two predictions hold and two fail, and the pattern of which is the interesting part. Prediction 1 — that the most system-destructive firms earn no unusual return in normal times, before markets learn to price system welfare — fits the structural claim¹⁹ that a market cannot price W before it has a theory that names and measures W. Prediction 2 — that welfare-beta predicts how sharply a firm's stock reacts when a restoration event forces the hidden system harm into view — holds, supporting restoration-event loading. Prediction 3 (the most system-destructive firms should show more downside, negative-skew returns) and Prediction 4 (firms sharing the same damaged system should have returns that move together) both fail, and they fail even after a severe delisting stress test. The honest reading is narrow and worth stating plainly: the data support the System Asset Pricing Model (SAPM) and Missing System Theory (MST) as a structural no-premium and restoration-loading story, but not yet as a broad hidden tail-risk or non-diversifiable system-covariance asset-pricing theory. The next falsifiable test is diffusion — W should become priced only when beta-W-style disclosures, activist reports, litigation models, or shareholder adoption make the coordinate legible to markets.

A reader trained in finance will want to know where the standard model went, and the answer is reassuring: nowhere. The CAPM — the standard finance model that prices an asset by the risk it adds to a market portfolio²⁰ — is recovered exactly by setting the system-welfare price to zero. SAPM does not discard it; it nests it. The nesting is built to keep two measures from blurring together. Return-risk beta stays separate from causal welfare beta, so the asset-pricing measure and the policy measure remain distinct, and the welfare aggregate is explicitly planner-relative — it depends on the prices a social planner assigns to each harm — rather than a market-identified CAPM error²¹. CAPM keeps doing its job, pricing private returns; SAPM adds the one coordinate it was never built to carry: system welfare.

For a reader, the three-way split by binding mechanism changes how a reform proposal should be judged. In impossibility domains (a natural law sets the floor), the question is not whether the damage can be eliminated but how its trajectory can be managed. In intractability domains (an institutional constraint with a reform model proven in another jurisdiction), the question is which redesign would break the Private-Systemic Tension, since a below-one outcome is demonstrably reachable. In control domains (measured β_W already below 1), the measured loss is smaller than revenue, so the domain is a check on the method rather than a target for reform. The same result reads differently from other chairs. For a regulator, Propositions 8a–8b mean current certification tools cannot detect Hollow Wins, so a new framework is required. For an executive, the CAPM nesting relation means existing risk models are not wrong but incomplete, missing the system-welfare coordinate. For a student, the Predictions 1-4 results show the theory is testable rather than philosophical, because it makes predictions that can fail. For a researcher, the next falsifiable test is diffusion — system welfare should become priced only when the coordinate becomes legible to markets.

The limits are worth stating without flinching. The empirical evidence is mixed: Prediction 3 and Prediction 4 fail, and the theory stands only as a structural no-premium and restoration-loading story, with the broad hidden-tail-risk and non-diversifiable system-covariance claims still awaiting evidence. The domain taxonomy — 14 impossibility (a natural-law floor), 34 intractability (an institutional constraint with a borrowable reform), and 13 control (measured β_W already below 1) across the 61 studied records — is set by the binding mechanism, but the underlying β_W values come from a locked reconciliation and remain working estimates that later evidence can move, and the theorem statements, expiry conditions, and proven models are not all fully specified here. The aggregate total is $69.0T per year, comparable in scale to global GDP but not identical to it. The next falsifiable test, diffusion, has not yet been run.