Chapter 11: Vaccine Allocation, Organ Transplant, and the Geometry of Scarcity

The Problem of Population-Scale Pathfinding

Pandemic vaccine allocation differs from ICU triage in three structural ways:

1. Scale: Millions of individuals, not dozens.
2. Indirect benefit: The benefit of vaccination includes not just individual protection (d_1) but herd immunity (d_6: community impact), which accrues to unvaccinated individuals.
3. Voluntary participation: Unlike triage (which is imposed on patients), vaccination is a choice. The patient’s heuristic function h_{\text{patient}} determines uptake.

Definition 11.1 (Collective Geodesic). The collective geodesic \gamma^*_{\text{collective}} for a population S on the clinical decision complex \mathcal{C} is the allocation of a resource (vaccine doses) across S that minimizes total clinical-moral friction for the population as a whole:

\gamma^*_{\text{collective}} = \arg\min_{\text{allocation}} \sum_{s \in S} \text{CF}(\gamma_s)

This differs from the Nash equilibrium of individual geodesics, where each individual minimizes their own clinical friction without accounting for externalities:

\gamma^*_{\text{Nash}} = \{\gamma_s^* : \gamma_s^* = \arg\min_{\gamma_s} \text{CF}(\gamma_s) \text{ for each } s \in S\}

Divergence of Collective and Individual Geodesics

Theorem 11.1 (Collective-Individual Geodesic Divergence). The collective geodesic \gamma^*_{\text{collective}} diverges from the Nash equilibrium \gamma^*_{\text{Nash}} whenever individual vaccination decisions create externalities on d_6 (social and relational impact). Specifically:

1. A healthy young adult with low personal risk (d_1 benefit of vaccination is small) and high social contact (d_6 benefit of vaccination to others is large) may have an individual geodesic of non-vaccination but a collective-optimal geodesic of vaccination.

2. The magnitude of divergence is proportional to the herd immunity contribution \Delta d_6 — the externality that individual optimization ignores.

This divergence is the geometric formulation of the classic public health problem: individual rationality can produce collective suboptimality. The framework adds specificity: the divergence is not merely about “selfishness versus altruism” but about the structural omission of d_6 from individual decision-making.

Vaccine Hesitancy as Heuristic Miscalibration

Vaccine hesitancy is a specific form of heuristic corruption: the individual’s h_{\text{patient}}(n) is miscalibrated on the relevant dimensions.

Definition 11.2 (Vaccine Heuristic Miscalibration). A vaccine-hesitant individual’s heuristic function h_{\text{hesitant}}(n) is characterized by:

• Overestimation of d_1 cost: the probability and severity of adverse events is inflated relative to the evidence base.
• Underestimation of d_6 benefit: the herd immunity contribution is ignored or discounted.
• Gauge variance: the heuristic is sensitive to framing (absolute numbers of adverse events vs. base rates), violating the Bond Invariance Principle.

The gauge-variance feature is particularly diagnostic. A person who rejects a vaccine after reading “10,000 adverse events reported” but would accept it after reading “0.001% adverse event rate” has a gauge-variant heuristic — their decision depends on the framing, not the underlying clinical reality. This is a BIP violation detectable by the gauge-invariance test (Theorem 4 from the GCE paper).

Proposition 11.1 (Hesitancy Reduces to Three Distinct Failures). Vaccine hesitancy decomposes into three structurally distinct failures:

1. Epistemic failure (d_9): The individual lacks accurate information about vaccine safety and efficacy. Intervention: public health communication.

2. Trust failure (d_5): The individual does not trust the institution recommending the vaccine (pharmaceutical companies, government agencies, the medical establishment). Intervention: trust-building through community engagement, transparent reporting, and addressing historical betrayals (Tuskegee, thalidomide).

3. Autonomy assertion (d_4): The individual prioritizes bodily autonomy over collective benefit, even with full information and full trust. Intervention: this is not a failure at all — it is a legitimate manifold computation with different dimensional weights.

Standard public health messaging conflates all three under “hesitancy.” The geometric framework distinguishes them because they occupy different regions of the manifold and require different interventions.

This distinction has immediate practical consequences. An education campaign addresses failure 1 but not failure 2 or 3. A mandate addresses all three but at substantial d_4 cost. A community-engaged outreach program addresses failure 2. Only by distinguishing the three failures can the intervention be matched to the cause.

The Most Constrained Allocation in Medicine

Organ transplant allocation is the most resource-constrained allocation problem in clinical medicine. In the United States, approximately 100,000 people are on the organ transplant waiting list at any given time. Approximately 40,000 transplants are performed per year. The gap between supply and demand means that the allocation decision — which patient receives which organ — has life-or-death consequences for every patient on the list.

The existing allocation system (UNOS in the United States, NHS Blood and Transplant in the UK, Eurotransplant in continental Europe) uses organ-specific scoring algorithms: MELD (Model for End-Stage Liver Disease) for liver, LAS (Lung Allocation Score) for lung, and a multi-factor scoring system for kidney. These algorithms are, in effect, informal projections of the clinical tensor — they attempt to quantify “who benefits most” from a scalar scoring function.

What MELD Captures and What It Misses

The MELD score, used for liver transplant allocation since 2002, is computed from three laboratory values: bilirubin, creatinine, and INR. It predicts 90-day mortality without transplant — a pure d_1 measure.

Dimensional analysis of MELD:

MELD is an excellent d_1 predictor. It is a poor manifold representation. It assigns zero weight to five of nine dimensions — trust, autonomy, dignity, social impact, and rights. The QALY Irrecoverability Theorem predicts exactly the consequences: populations whose needs are concentrated on the zero-weighted dimensions are systematically disadvantaged.

The Trust Dimension in Transplant Allocation

Consider two patients with identical MELD scores:

• Patient A: A 55-year-old white man with alcoholic cirrhosis, strong family support, private insurance, and a history of positive healthcare experiences.
• Patient B: A 55-year-old Black woman with autoimmune hepatitis, single parent, Medicaid, and a family history of medical mistreatment (her grandmother participated in unethical medical research without consent).

MELD treats these patients identically. On the manifold, they are not identical. Patient B’s d_5 (trust) is lower — not because of any personal failing but because of historical institutional betrayal. Her post-transplant trajectory is different: she is less likely to attend follow-up appointments (trust deficit), less likely to report symptoms early (trust deficit), and less likely to adhere to immunosuppression regimens (trust deficit compounded by insurance instability). The MELD score does not see this. The manifold does.

Proposition 11.2 (Trust as Transplant Outcome Modifier). Post-transplant outcomes are modulated by d_5 (trust) through its effect on medication adherence, follow-up attendance, and symptom reporting. An allocation system that ignores d_5 will produce systematically worse outcomes for populations with lower baseline trust — not because those populations are less “deserving” but because the system’s failure to account for trust creates a post-transplant environment that undermines the transplant’s clinical benefit.

The response is not to penalize Patient B (which would compound the injustice) but to design post-transplant care that addresses the trust deficit: community health workers, culturally concordant care teams, flexible appointment schedules, transparent communication about the transplant process. This is a manifold-aware intervention: it modifies the post-transplant path to reduce total clinical friction on d_5, improving overall outcomes.

The Dignity Dimension: Waitlist Experience

The experience of being on a transplant waiting list is itself a manifold traversal. Patients on the waitlist live in a state of chronic uncertainty (d_9), progressive illness (d_1), and compromised dignity (d_7: their survival depends on someone else’s death). The waitlist experience is not morally neutral — it imposes cumulative costs on multiple dimensions.

Transplant teams experience their own form of moral injury from the allocation process. Every time an organ is allocated, the team must tell the patients who were not selected that they remain on the list — a message that becomes harder to deliver with each repetition. The accumulation of these messages is a form of low-level boundary crossing on d_5 (trust: “we’re still trying”) and d_7 (dignity: “you weren’t chosen this time”).

Proposition 11.3 (Waitlist Moral Injury). Transplant teams experience cumulative moral injury from repeated allocation decisions, with the moral injury increment proportional to the number of viable candidates who were not selected (not the severity of any individual candidate’s illness). This prediction distinguishes the geometric framework from burnout models, which predict injury proportional to workload rather than moral cost.

When Individual Freedom Conflicts with Collective Health

Public health mandates — vaccination requirements, quarantine orders, water fluoridation, mask mandates — restrict individual autonomy (d_4) to achieve collective health benefits (d_1 + d_6). Each mandate is a forced deviation from individual geodesics in the service of the collective geodesic.

Definition 11.3 (Geometric Justification of a Public Health Mandate). A public health mandate is geometrically justified when three conditions hold simultaneously:

1. Geodesic divergence: The collective geodesic \gamma^*_{\text{collective}} differs from the Nash equilibrium \gamma^*_{\text{Nash}}. Without the mandate, individual choices produce a collectively suboptimal outcome.

2. Bounded autonomy cost: The autonomy restriction \Delta d_4 does not cross a sacred-value boundary (\beta_{\text{sacred}} < \infty). The mandate restricts autonomy but does not violate bodily integrity, religious practice, or personal identity in ways that have infinite boundary penalties.

3. Proportionate benefit: The collective benefit on d_1 + d_6 exceeds the individual cost on d_4, under the community’s metric tensor \Sigma.

\frac{(\Delta d_1 + \Delta d_6)^T \Sigma^{-1} (\Delta d_1 + \Delta d_6)}{(\Delta d_4)^2 / \sigma_4^2} > 1

All three conditions are necessary. A mandate that satisfies conditions 1 and 3 but not condition 2 (e.g., forced medical experimentation that violates bodily integrity — a sacred-value boundary) is never geometrically justified, regardless of the collective benefit.

Application: Vaccination Mandates

Childhood vaccination mandates satisfy all three conditions for most vaccines and most populations:

1. Geodesic divergence: Individual vaccine-hesitant parents underweight d_6 (herd immunity benefit to immunocompromised children who cannot be vaccinated). The collective geodesic includes vaccination; the Nash equilibrium does not.

2. Bounded autonomy cost: Vaccination is a minor physical intervention — needle, brief discomfort, small risk of adverse effects. The d_4 cost is real but finite. No sacred-value boundary is crossed (except for specific religious exemptions, which must be evaluated individually).

3. Proportionate benefit: The collective benefit of herd immunity (d_1 + d_6) vastly exceeds the individual d_4 cost of mandatory vaccination.

COVID-19 vaccine mandates are more complex:

1. Geodesic divergence: Yes — individual non-vaccination creates externalities.
2. Bounded autonomy cost: For most people, yes. But for individuals with a history of adverse vaccine reactions, the d_1 risk is elevated, potentially crossing the harm boundary \beta_{\text{harm}}. For individuals with deep religious objections, the d_7 cost (identity violation) may approach a sacred-value boundary.
3. Proportionate benefit: Depends on the population’s existing immunity, the variant’s transmissibility, and the vaccine’s effectiveness. As the pandemic evolves, the proportionality calculation changes — a mandate justified at 5% population immunity may not be justified at 85%.

Proposition 11.4 (Dynamic Mandate Justification). The geometric justification for a public health mandate is not static. As the epidemiological context changes (population immunity increases, viral transmissibility changes, treatment options improve), the proportionality condition (condition 3) may shift from satisfied to unsatisfied. A mandate that was geometrically justified in March 2020 may not be justified in March 2022. The framework provides the mathematical apparatus for this temporal evolution — the metric tensor \Sigma is context-dependent, and the justification must be re-evaluated as the context changes.

Application: Quarantine Orders

Quarantine — the restriction of movement for individuals who may have been exposed to a communicable disease — imposes higher d_4 costs than vaccination (restriction of freedom of movement for days or weeks) but also provides higher d_6 benefits (preventing transmission chains).

The geometric analysis:

1. Geodesic divergence: Yes — an exposed individual’s individual geodesic (go about normal life) diverges from the collective geodesic (quarantine to prevent transmission).
2. Bounded autonomy cost: Quarantine restricts freedom of movement (d_4) and imposes social isolation (d_6: loss of social contact, economic harm). The d_4 cost is higher than vaccination but still finite — no sacred-value boundary is crossed by temporary restriction of movement.
3. Proportionate benefit: Depends on the probability that the individual is actually infected, the transmissibility of the disease, and the severity of outcomes for contacts.

The proportionality condition is most clearly satisfied for highly transmissible, highly lethal diseases (Ebola, SARS). It is more contested for diseases with lower case fatality rates (influenza, COVID-19 in vaccinated populations), where the d_1 + d_6 benefit may not clearly exceed the d_4 cost.

Application: Water Fluoridation

Water fluoridation presents a different manifold structure. The intervention is universal (everyone who drinks tap water receives fluoride), involuntary (no individual opt-out short of avoiding tap water), and long-term (continuous exposure over a lifetime).

1. Geodesic divergence: Modest — dental caries prevention is primarily a d_1 benefit to the individual, with small d_6 externalities (reduced community dental disease burden).
2. Bounded autonomy cost: Minimal — fluoridated water has no meaningful d_4 cost for most individuals.
3. Proportionate benefit: For populations with limited access to dental care, the d_1 benefit is large and the d_4 cost is negligible. Proportionality is clearly satisfied.

The primary objection to fluoridation is not on d_4 (the autonomy cost is too small to be meaningful) but on d_2 (the right to choose what chemicals enter one’s body) and d_5 (trust in the government’s judgment about what is safe). These objections activate different dimensions than the autonomy argument — and the geometric framework can distinguish them.

The Special Status of Irreversibility

Some allocation decisions are irreversible: once made, they cannot be undone. Organ transplant allocation is the paradigm case — once an organ is transplanted into Patient A, it cannot be recovered for Patient B. But irreversibility extends beyond transplant:

• Germline gene editing (CRISPR): Changes propagate to all future descendants. The allocation of genetic modification technology is an irreversible allocation across generations.
• End-of-life treatment withdrawal: Once life support is withdrawn and the patient dies, the decision cannot be reversed.
• Pandemic vaccine allocation (partially): First doses allocated to one population cannot be retroactively reallocated to another if the epidemiological situation changes.

Definition 11.4 (Irreversibility Penalty). An irreversible allocation decision carries an additional boundary penalty \beta_{\text{irrev}} proportional to the forecasting uncertainty:

\beta_{\text{irrev}} = \kappa \cdot \text{Var}[\text{future outcome}]

where \kappa is a constant reflecting the manifold weight on precaution and \text{Var}[\text{future outcome}] is the variance of the outcome under the irreversible decision.

The irreversibility penalty is not conservatism for its own sake. It is the manifold cost of foreclosing future options. A reversible decision (prescribing a medication that can be discontinued) has low \beta_{\text{irrev}} because the path can be corrected. An irreversible decision (removing a patient from the transplant list) has high \beta_{\text{irrev}} because the path cannot be corrected if the forecast is wrong.

Proposition 11.5 (Irreversibility Favors Caution Under Uncertainty). For allocation decisions with high \beta_{\text{irrev}} and high d_9 uncertainty (poor prognostic accuracy), the manifold-optimal strategy is more cautious than the utilitarian-optimal strategy. This is the geometric formulation of the precautionary principle — not as a blanket prohibition on action, but as a specific cost that scales with irreversibility and uncertainty.

Information and Dimensionality

A counterintuitive consequence of the manifold framework: acquiring more information about patients can make allocation decisions harder, not easier.

Consider a simple allocation scenario: two patients, one ventilator. If the clinician knows only d_1 (clinical prognosis), the decision is straightforward — give the ventilator to the patient with the better prognosis. The decision space is one-dimensional.

Now the clinician learns that Patient A is a single mother of three children (d_6: high social impact of death) while Patient B is an elderly widower with no dependents (d_6: low social impact). The decision space has expanded to two dimensions. On d_1 alone, Patient B may have the better prognosis; on d_1 + d_6, the answer is ambiguous.

Now the clinician learns that Patient B is a Holocaust survivor who specifically chose this hospital because of its commitment to caring for elderly patients (d_5: high trust). The decision space expands to three dimensions.

Each piece of information activates a new dimension of the manifold. The manifold-optimal allocation changes with each activation. The clinician’s distress increases — not because the information is distressing but because the dimensionality of the decision space has grown, making the geodesic computation harder.

Proposition 11.7 (The Allocation Paradox). Acquiring more morally relevant information about patients in an allocation scenario increases the dimensional activation of the decision, which in general increases the clinician’s moral distress and the computational difficulty of finding the manifold-optimal allocation. Scalar triage systems are simpler precisely because they discard this information.

The paradox explains why many triage systems deliberately restrict the information available to the triage decision-maker. ESI protocols explicitly exclude social history, personal narrative, and family context from the triage assessment. This restriction is not callousness — it is a deliberate dimensional reduction that makes the triage computation tractable. The manifold framework makes the trade-off explicit: dimensional reduction makes decisions easier but sacrifices moral information that, if included, would change the decision.

When to Restrict Information

The paradox raises a genuine design question: how much information should a triage decision-maker receive?

The scalar answer: As little as possible. The more information, the more dimensions are activated, the harder the decision, and the higher the moral injury risk. Restrict to d_1 alone.

The manifold answer: Enough to compute a reasonable approximation of the manifold-optimal allocation, but not so much that the computation becomes intractable. Specifically:

1. Always include d_1 (clinical prognosis) — this is the core of triage.
2. Include d_3 (justice considerations) — to detect and prevent systematic disparities.
3. Include d_9 (epistemic uncertainty) — to identify cases where the d_1 estimate is too uncertain to drive allocation.
4. Exclude d_6 (social value) in most cases — social value judgments (“this patient contributes more to society”) are both gauge-variant (they depend on the evaluator’s perspective) and historically discriminatory.
5. Monitor d_5 (trust) at the population level — not for individual allocation but for Bond Index computation.

This selective inclusion balances dimensional richness against computational tractability and moral injury risk.

Worked Example 1: Childhood Vaccination Mandate

Consider a state legislature considering a childhood vaccination mandate that would eliminate non-medical exemptions for school enrollment. The geometric analysis:

Condition 1 (Geodesic divergence): The individual geodesic for a vaccine-hesitant parent includes non-vaccination (the parent’s h_{\text{patient}} overestimates d_1 cost of vaccination and underestimates d_6 benefit of herd immunity). The collective geodesic includes vaccination for all children capable of being vaccinated. The geodesics diverge. Condition satisfied.

Condition 2 (Bounded autonomy cost): The mandate restricts d_4 (parental autonomy over the child’s medical care) but does not cross a sacred-value boundary. The parent retains medical decision-making authority for all other medical decisions; the mandate restricts one specific decision. The d_4 cost is finite: \Delta d_4 < \infty. Condition satisfied.

For a specific subset of parents — those with sincere religious objections where vaccination conflicts with core religious identity (d_7) — the autonomy cost may approach a sacred-value boundary. The geometric analysis for this subset: if \beta_{\text{religious}} = \infty for these parents, the mandate’s d_4 cost is infinite on their manifold, and Condition 2 is not satisfied. This is the geometric justification for religious exemptions — not as a blanket permission but as a recognition that the autonomy cost is infinite for a specific subpopulation.

Condition 3 (Proportionate benefit): The d_1 + d_6 benefit of universal vaccination is substantial: prevention of serious illness in the vaccinated child (d_1) plus herd immunity protection for immunocompromised children who cannot be vaccinated (d_6). The benefit exceeds the d_4 cost by a large margin for most populations. Condition satisfied.

Geometric verdict: The childhood vaccination mandate is geometrically justified for the general population. Religious exemptions are geometrically justified for the subpopulation where \beta_{\text{religious}} = \infty.

Worked Example 2: COVID-19 Workplace Vaccine Mandate (2021)

The geometric analysis of a COVID-19 workplace vaccine mandate in 2021:

Condition 1: Geodesic divergence exists — individual non-vaccination creates transmission externalities (d_6). Satisfied.

Condition 2: The autonomy cost is moderate. The mandate restricts employment, not bodily autonomy directly (the worker can choose not to be vaccinated and seek employment elsewhere — though this creates a d_6 economic cost). For workers who have had prior COVID-19 infection and may have natural immunity, the d_1 benefit of additional vaccination is smaller, potentially shifting the proportionality calculation. The sacred-value boundary is not crossed for most workers — but the mandate is experienced as closer to the boundary than childhood vaccination, because it conditions employment on a medical decision.

Condition 3: In 2021, with low population immunity and high transmission, the proportionate benefit was substantial. By late 2022, with high population immunity (through vaccination and infection), the proportionate benefit was smaller. The geometric justification was time-dependent — satisfied in 2021, possibly unsatisfied in 2023.

Geometric verdict: The COVID-19 workplace vaccine mandate was geometrically justified in early-to-mid 2021 for most populations. The justification weakened over time as population immunity increased. The framework provides the mathematical apparatus for this temporal evolution — something that the binary mandate/no-mandate policy framework does not.

Extending the Clinical Manifold to Population Scale

The clinical decision complex \mathcal{C} was constructed for individual clinical decisions. Population-level allocation decisions require an extension: the bioethical decision complex \mathcal{B}.

Definition 11.5 (Bioethical Decision Complex). The bioethical decision complex \mathcal{B} is a weighted simplicial complex with the same nine-dimensional structure as \mathcal{C} but operating at population scale:

• Vertices: Population health states — configurations of disease prevalence, resource availability, population immunity, institutional trust, and equity.
• Edges: Policy actions — vaccination campaigns, allocation protocols, mandate implementations, infrastructure investments.
• Edge weights: Aggregate clinical-moral cost across the population, including distributional effects:

w_{\mathcal{B}}(v_i, v_j) = \sum_{s \in S} w_{\mathcal{C}}(v_i^s, v_j^s) + \lambda \cdot \text{Gini}(\{w_{\mathcal{C}}(v_i^s, v_j^s)\}_{s \in S})

where the Gini term penalizes inequality in the distribution of costs across the population.

The Gini penalty \lambda \cdot \text{Gini}(\cdot) is the mathematical expression of distributive justice at population scale. A policy that imposes equal cost on all individuals has \text{Gini} = 0; a policy that concentrates cost on a minority has \text{Gini} \to 1. The parameter \lambda controls how strongly the population values equity over efficiency.

The Equity-Efficiency Frontier

Every allocation policy sits on a frontier between equity (d_3) and efficiency (d_1). The utilitarian policy maximizes d_1 without regard to distribution. The egalitarian policy equalizes cost distribution without regard to d_1. The manifold-optimal policy finds the point on the frontier that minimizes total clinical friction including the Gini penalty.

[Figure 11.1: The equity-efficiency frontier for vaccine allocation, showing utilitarian, egalitarian, and manifold-optimal allocations, with specific COVID-19 prioritization schemes plotted on the frontier]

Proposition 11.6 (No Free Lunch on the Equity-Efficiency Frontier). There exists no allocation policy that simultaneously maximizes d_1 and minimizes the Gini coefficient of cost distribution, except in the degenerate case where the population is homogeneous on all nine dimensions. Every allocation involves a trade-off between equity and efficiency — and the trade-off’s optimal point depends on the community’s metric tensor \Sigma, which encodes how the community weighs the nine dimensions against each other.