Skip to content

Majority-Dominated Aggregate Objective

Prime #
975
Origin domain
Cross Domain
Subdomain
aggregation failure modes → Cross Domain

Core Idea

Majority-dominated aggregate objective is the structural arrangement in which a system optimized under an aggregate objective whose mass is dominated by the majority population systematically underweights an operationally important minority, even when the system formally "counts everyone." The objective is additive, expected-value, or plurality-rule; the input distribution is skewed; and the minority is the operationally load-bearing case — the rare-but-severe event, the under-represented user, the high-impact failure. The resulting system performs well on average and fails specifically and predictably on the minority, not by oversight but because the aggregate objective never created the gradient to do otherwise. The minority neglect is built into the optimum, not introduced by a defect in implementation.

Four roles carry the structure. First, an aggregate objective function whose value derives from summed or averaged per-instance contributions. Second, a skewed prevalence distribution in the input, where one population dominates the count. Third, an operationally important minority — the cases the system exists to serve well, not merely to include. Fourth, the mass-weighted optimum that falls on the majority side and is minority-blind by construction. The shape recurs across substrates because the same mathematical structure produces the same outcome wherever it is instantiated: a low-loss strategy that is achieved precisely by ignoring the cases whose contribution to the aggregate is small but whose operational importance is large. The arrangement is the dual of a familiar intuition — the average is not a neutral summary; it is a weighting, and the weighting can be the failure.

How would you explain it like I'm…

The Few Get Forgotten

Imagine your class votes on snacks and almost everyone wants cookies, but a few kids are allergic and can only eat fruit. If you just go with what most kids want, you pick cookies and the allergic kids get nothing — even though everybody 'got a vote.' Adding up everyone's votes can quietly leave out the small group who really needed something different.

Lost in the Average

This is what happens when you try to make a system good "on average" but a small, important group keeps getting left behind. Say a town designs its bus routes to please the most riders. Most riders live downtown, so the buses serve downtown great, and the few people in far neighborhoods are ignored, even though counting total riders technically includes them. The system isn't broken; it's doing exactly what it was told, which was to make the total as good as possible. The problem is that the small group barely moves the total, so there's no reason to serve them well, even when they're the ones who needed it most.

Mass-Weighted Minority Blindness

A majority-dominated aggregate objective is an arrangement where a system optimized for one summed or averaged score systematically underweights an important minority, even though it formally 'counts everyone.' The setup has a skewed input — one population dominates the count — and the minority is the operationally load-bearing case: the rare-but-severe event, the under-represented user, the high-impact failure. The result is a system that performs well on average and fails specifically and predictably on the minority — not by oversight, but because the aggregate objective never created any gradient to do otherwise. The minority neglect is built into the optimum, not introduced by a bug. The deep point is the dual of a common intuition: an average isn't a neutral summary, it's a weighting — and that weighting can itself be the failure mode.

 

A majority-dominated aggregate objective is the structural arrangement in which a system optimized under an aggregate objective — one whose mass is dominated by the majority population — systematically underweights an operationally important minority, even when it formally counts everyone. The objective is additive, expected-value, or plurality-rule; the input distribution is skewed; and the minority is the operationally load-bearing case: the rare-but-severe event, the under-represented user, the high-impact failure. The resulting system performs well on average and fails specifically and predictably on the minority — not by oversight but because the aggregate objective never created the gradient to do otherwise. Four roles carry the structure: an aggregate objective function whose value is summed or averaged per-instance contributions; a skewed prevalence distribution where one population dominates the count; an operationally important minority — the cases the system exists to serve well, not merely to include; and a mass-weighted optimum that falls on the majority side and is minority-blind by construction. The shape recurs across substrates because the same mathematics produces the same outcome wherever instantiated: a low-loss strategy achieved precisely by ignoring cases whose contribution to the aggregate is small but whose operational importance is large. It is the dual of a familiar intuition — the average is not a neutral summary; it is a weighting, and the weighting can be the failure.

Structural Signature

the aggregate objective summing per-instance contributions — the skewed prevalence distribution dominated by one population — the operationally important minority small in count but large in stakes — the mass-weighted optimum falling on the majority side — the minority-blind-by-construction invariant: the neglect is built into the optimum, not introduced by a defect

A system exhibits a majority-dominated aggregate objective when each of the following holds:

  • An aggregate objective. The system optimizes a function whose value derives from summed, averaged, or plurality-rule per-instance contributions — cross-entropy, a vote tally, an engagement score, expected loss.
  • A skewed prevalence distribution. The input distribution is dominated in count by one population — class imbalance, geographic concentration, a popularity long-tail, disease-burden skew.
  • An operationally important minority. A subpopulation contributes little to the aggregate but is the load-bearing case — the rare-but-severe event, the under-represented user, the high-impact failure the system exists to serve.
  • A mass-weighted optimum. The optimum falls on the majority side because the objective's mass does; a low-loss strategy is achieved precisely by ignoring the small-contribution cases.
  • A formal inclusion. The system "counts everyone" — the minority is formally present in the objective, which is why the failure is invisible to aggregate metrics.
  • The minority-blind invariant. The neglect is built into the optimum, not introduced by an implementation defect; aggregate performance under-determines per-group performance, and the minority-blind regime is the default unless the designer actively chose otherwise.

The components compose into a structural diagnosis and a four-move repair kit — reweight, retune, replace the metric, supplement with minority-specific accountability — with the caution that the kit applied naively regenerates the same mass-weighting one level up.

What It Is Not

  • Not aggregation. Aggregation is the neutral operation of combining values; this prime is the failure mode in which the aggregate's mass concentrates on a majority and the optimum becomes minority-blind by construction — the average is revealed as a weighting, and the weighting is the defect.
  • Not pareto_efficiency. A Pareto-efficient allocation cannot improve one party without harming another; the mass-weighted optimum is typically not even Pareto-improving for the minority — it serves them arbitrarily badly while the metric stays silent, which is a different (and worse) condition than efficiency.
  • Not multiobjective_optimization. Multiobjective optimisation explicitly carries several objectives and trades among them; here there is a single aggregate objective whose mass happens to dominate, and the minority is formally included yet operationally neglected — the problem is one objective's weighting, not the balancing of several.
  • Not goodharts_law. Goodhart concerns a measure ceasing to be good once targeted; this prime concerns a measure that is structurally minority-blind even when honestly optimised — though Goodhart reappears recursively when a minority-aware metric is itself targeted.
  • Not tragedy_of_the_commons. The commons tragedy is uncoordinated depletion of a shared resource by self-interested actors; here a single designed objective under a skewed distribution produces the neglect, with no rivalry or depletion — the remedy is reweighting the objective, not governing a commons.
  • Common misclassification. Treating it as a moral fairness complaint. The pattern is structural — the neglect is built into the loss-minimising optimum on a skewed distribution — and exhortatory fixes (better intentions, more diverse teams, "add more data") that never change the gradient leave the optimum, and the neglect, intact.

Broad Use

  • Machine learning — aggregate cross-entropy on an imbalanced dataset is dominated by majority-class terms; the classifier learns "predict majority" as a low-loss strategy and fails on the minority the application cares about (diagnosis, fraud, rare-defect detection).
  • Politics and voting — majoritarian or plurality rules underweight intense minority preferences; median-voter dynamics drive policy toward the centre of mass.
  • Attention and media metrics — aggregate engagement is dominated by median-click contributions; rare-but-important content is starved.
  • Public-health surveillance — aggregate cost-effectiveness measures dominated by common-condition burden divert resources from rare-but-severe events.
  • Regulatory triage — expected-value risk calculations dominated by common-risk volume under-attend to tail-risk events.
  • Education metrics — class-mean or median-pass-rate metrics are dominated by the middle of the distribution; struggling and exceptional students are equally invisible.
  • Recommendation systems — popularity-weighted ranking biases toward the head of the long tail; niche content with high per-user value is starved.
  • Insurance and actuarial pricing — pricing optimized on expected loss works in aggregate while systematically failing high-cost minority segments.

Clarity

Naming the arrangement separates two routinely conflated complaints. The fairness complaint is moral: "the system treats group X unfairly." The aggregate-objective complaint is structural: "the objective the system optimizes has its mass on majority cases; the system is behaving exactly as its objective specifies, and the specification is the problem." The structural framing relocates the intervention surface. Rather than blame the model, debate the moral weights, or retrain on the same objective, the prime points at the objective function itself as the leverage point. This does not displace the moral question — it isolates a mechanism that the moral language alone cannot diagnose.

It also clarifies why intuitive fixes fail. "Just add more minority data" can leave the failure intact: even with balanced data, a prevalence-weighted aggregate objective can still under-weight the minority if the prevalence weighting persists at deployment. The neglect sits in the objective specification, not only in the data. The clarifying move is to ask, of any system that performs well on average yet fails a population, whether the failure is in the data, in the model, or in the mass-weighting of the objective — three different diagnoses with three different repairs.

Manages Complexity

The arrangement compresses a sprawling set of apparently unrelated failure modes — medical-AI bias, electoral underrepresentation, attention-economy malformation, public-health misallocation, education-metric distortion — into one diagnostic shape: the optimum lands on the majority because the objective's mass does. With the shape named, the intervention vocabulary becomes uniform across substrates. Reweight the objective (cost-sensitive learning, weighted voting). Retune the threshold (move the decision boundary toward the minority). Replace the metric (substitute per-class or per-group performance for the aggregate). Supplement with minority-specific accountability (a separate layer whose objective is not the aggregate). A practitioner who has internalized the prime carries a portable repair kit; one who has not will rediscover the same fix independently in each domain.

The leverage is that the repair surface comes into view automatically once the diagnosis names the mass-weighting. Where a failure is described only as "bias" or "unfairness," the repair surface stays contested; where it is named as majority-domination of an aggregate objective, the four moves — reweight, retune, replace, supplement — are immediately the candidate interventions.

Abstract Reasoning

Majority-dominated aggregate objective trains a reasoner to ask:

  • Is the objective additive, expected-value, or plurality-rule, and is its mass concentrated on a majority population?
  • Is there an operationally important minority whose contribution to the aggregate is small but whose failure is costly?
  • Does aggregate performance here certify per-group performance, or could any specific subgroup be arbitrarily badly served at the same optimum?
  • Is the proposed fix structural (changing what the objective measures) or merely exhortatory (better intentions, more diverse teams, more data) — and would the exhortatory fix change the optimum at all?
  • Once a minority-aware metric is introduced, does the same structural risk reappear recursively against the new metric?
  • Is the minority-blind regime here the default that obtains unless the designer actively chose otherwise?

The sharpest inferences are that aggregate performance under-determines per-group performance, that the fix is structural rather than motivational, and that the minority-blind regime is the default for any aggregate-objective system on a skewed distribution. A further Goodhart-style inference is recursive: once the system is told to do better on the minority, it optimizes the new minority-aware metric, and the same mass-weighting risk applies to that metric in turn.

Knowledge Transfer

Role mappings across domains:

  • Aggregate objective ↔ cross-entropy loss / vote tally / engagement score / QALY-cost measure / expected loss
  • Skewed prevalence ↔ class imbalance / geographic concentration / popularity long-tail / disease-burden skew
  • Operationally important minority ↔ rare disease / under-represented user / tail-risk event / niche content
  • Mass-weighted optimum ↔ majority-class predictor / median-voter policy / head-of-tail ranking
  • Minority-blind regime ↔ low recall on the minority / structural underrepresentation / starvation of the rare case
  • Structural fix ↔ reweighting / threshold tuning / per-group metric / minority-specific accountability

A machine-learning engineer fighting class imbalance, a political theorist analyzing minority underrepresentation, a health economist allocating a budget by aggregate cost-effectiveness, and a product team ranking by engagement are confronting the same structure: a mass-weighted optimum that is minority-blind by construction. The intervention vocabulary moves with the diagnosis. Cost-sensitive learning's toolkit — reweighting, threshold-tuning, focal loss — transfers into surveillance-allocation reasoning: do not let an aggregate cost-effectiveness measure starve a rare-but-severe event; build minority-specific accountability into the budget. Proportional-representation insights from voting theory port into recommendation as explicit per-cluster quotas evaluated separately from aggregate engagement. Splitting an aggregate KPI into per-segment metrics with separate accountability is the same move as splitting a class-mean grade into per-student trajectories. Recognizing that aggregate efficiency can mask systematic exclusion of expensive minorities is one insight whether you are pricing insurance premiums or rationing public services. What travels is not a metaphor but the literal recognition that the average is a weighting, plus the repair kit that follows — and the caution that the kit, applied naively, can regenerate the same failure one level up.

Examples

Formal/abstract

A binary classifier trained on an imbalanced dataset is the cleanest formal instance, because the failure can be read straight off the objective. The aggregate objective is cross-entropy summed over training instances; the skewed prevalence distribution is the class imbalance — suppose 99% of instances are the negative class (no disease) and 1% positive (disease present). The operationally important minority is the positive class: it contributes 1% of the loss mass but is the entire reason the system exists. The mass-weighted optimum falls on the majority side because the objective's mass does: a classifier that predicts "negative" for every input achieves 99% accuracy and very low cross-entropy, so "predict majority" is a genuine low-loss strategy the gradient discovers and rewards. The formal inclusion is what makes the failure invisible — the minority is present in the objective, every positive instance contributes its term, which is why aggregate metrics (accuracy, mean loss) look excellent while recall on the disease class is near zero. The minority-blind invariant is exact: the neglect is not an implementation bug but a property of the optimum itself, and aggregate performance under-determines per-group performance. The four-move repair kit reads off the structure: reweight (cost-sensitive loss, focal loss, up-weighting positive terms), retune (move the decision threshold toward the minority), replace the metric (optimise per-class recall, not aggregate accuracy), or supplement (a separate minority-specific accountability layer). The recursive Goodhart caution applies: once the new minority-aware metric is the target, the same mass-weighting risk can reappear against it.

Mapped back: Class-imbalanced classification instantiates every role — cross-entropy as the aggregate objective, the 99/1 split as the skewed prevalence, the disease class as the operationally important minority, the majority-predictor as the mass-weighted optimum — and the minority-blind invariant is shown directly: the neglect is built into the loss-minimising optimum, not introduced by a defect.

Applied/industry

Public-health budget allocation and recommendation ranking are two applied instances sharing the identical structure across designed-objective substrates. In health allocation, the aggregate objective is a cost-effectiveness measure (cost per QALY summed across the population); the skewed prevalence is disease-burden skew, where common conditions dominate the total burden; the operationally important minority is the rare-but-severe disease whose sufferers are few but whose untreated outcomes are catastrophic. The mass-weighted optimum directs the budget to common-condition interventions, starving the rare disease — and the failure is structural, not a failure of compassion, because the aggregate measure never created the gradient to fund the minority. The repair ports directly from cost-sensitive learning: do not let an aggregate cost-effectiveness measure starve a rare-but-severe event; build minority-specific accountability (a ring-fenced orphan-disease budget) into the allocation — the supplement move. Recommendation ranking runs parallel: the aggregate objective is total engagement, the skewed prevalence is the popularity long-tail, the operationally important minority is niche content with high per-user value, and the mass-weighted optimum is head-of-tail ranking that starves the niche. The voting-theory insight ports here as explicit per-cluster quotas evaluated separately from aggregate engagement — the replace-the-metric move — and the warning against intuitive fixes bites: "just add more minority data" or "show more niche content occasionally" leaves the failure intact if the prevalence-weighted objective persists at deployment.

Mapped back: Health-budget allocation and engagement ranking are the same mass-weighted optimum as the imbalanced classifier, with QALY-cost and engagement as aggregate objectives and the rare disease and niche content as operationally important minorities — repaired by the same reweight/retune/replace/supplement kit, with the caution that naive application regenerates the weighting one level up.

Structural Tensions

T1 — Aggregate optimum versus per-group performance (scalar, local vs global). The core claim is that aggregate performance under-determines per-group performance — a 99%-accurate optimum can serve a subgroup arbitrarily badly. The failure mode is aggregate-certifies-subgroup fallacy: reading excellent mean accuracy or low expected loss as evidence the minority is well served, when the metric is structurally silent on it. Diagnostic: never certify subgroup performance from an aggregate number — disaggregate and measure per-group, and treat any system reporting only aggregate metrics on a skewed distribution as having undisclosed minority performance.

T2 — Structural fix versus motivational fix (sign). The prime relocates the repair to the objective function, distinguishing structural fixes (reweight, retune, replace, supplement) from exhortatory ones (better intentions, more diverse teams, "add more minority data"). The failure mode is exhortatory non-fix: pursuing remedies that never change the optimum, so a prevalence-weighted objective keeps producing the same neglect after the intervention. Diagnostic: ask whether the proposed fix changes the gradient — if the same objective is optimised on the same distribution, the optimum (and the neglect) is unchanged no matter how the data is augmented or the team composed.

T3 — Repair kit versus recursive Goodhart (temporal). Introducing a minority-aware metric repairs the first-order neglect, but the system then optimises that metric, and the same mass-weighting risk reappears against it one level up. The failure mode is one-level-up regeneration: declaring the minority protected because a minority metric exists, while the new metric's own aggregate structure quietly re-creates a sub-minority blindness. Diagnostic: after introducing a minority-aware objective, re-run the whole diagnosis against the new metric — check whether its mass concentrates and whether a sub-population is now the operationally important minority it under-weights.

T4 — Minority importance versus its prevalence weight (measurement). The pattern presumes a minority that is small in count but large in stakes — but operational importance is a judgment the objective does not contain, and over-applying the prime can elevate a genuinely unimportant minority to load-bearing status. The failure mode is minority inflation: ring-fencing budget or recall for a rare case whose true operational stakes do not justify the mass diverted from the majority, degrading aggregate service for a minority that was correctly down-weighted. Diagnostic: separately establish the minority's operational importance (the cost of failing it) from its prevalence — only a genuine importance/prevalence mismatch justifies overriding the mass-weighted optimum.

T5 — Reweighting the minority versus majority cost (coupling). The repair moves are not free: reweighting, threshold-tuning, and quotas shift performance toward the minority by trading away majority performance, and the trade is coupled, not a pure gain. The failure mode is uncosted reweighting: cranking minority weight until the majority — still the bulk of real cases — is materially under-served, swapping one mass-weighting failure for its mirror image. Diagnostic: price the majority performance surrendered per unit of minority gain, and confirm the reallocation reflects the actual relative stakes rather than an over-correction that simply inverts the original neglect.

T6 — Designed objective versus emergent aggregation (scopal). The prime's framing and repair kit presume a designer who owns the objective function and can reweight it — but some majority-domination is emergent, arising from decentralised dynamics (median-voter drift, popularity feedback) with no single objective to edit. Here the boundary is with emergent_dynamics and collective-choice mechanism design. The failure mode is phantom-objective intervention: reaching for "reweight the loss" where there is no central objective, only an aggregation rule produced by many independent actors. Diagnostic: locate whether an editable objective function actually exists — if the mass-weighting emerges from a process rather than a specified function, the repair is mechanism redesign (voting rule, ranking dynamics), not loss reweighting.

Structural–Framed Character

Majority-dominated aggregate objective sits on the framed side of the structural–framed spectrum, at aggregate 0.5 with all five criteria at 0.5 — the balanced midpoint. There is a genuine mathematical skeleton underneath: an additive or expected-value objective whose mass is dominated by a skewed prevalence distribution, so the loss-minimising optimum is minority-blind by construction. That structure is real and abstract — the average is a weighting, and the weighting is the failure. But the prime's framing tips it onto the framed side, and the prose must own the even spread.

vocab_travels reads 0.5 because the core terms — aggregate objective, optimum, cross-entropy, prevalence — travel mathematically across ML, voting, and allocation, yet the load-bearing framing ("operationally important minority," "minority-blind") carries a designed-objective flavour that rides along. evaluative_weight is 0.5: the prime is partly value-neutral mathematics and partly a normative complaint — the neglect of the minority is framed as a failure the system "exists to serve," a real but bounded normative load the prime is careful to separate from a moral fairness claim. institutional_origin is 0.5: most instances are designed-objective substrates (loss functions, voting rules, KPIs, actuarial pricing) rooted in human institutions of optimisation, even though the mass-weighting mathematics is medium-neutral. human_practice_bound is 0.5: the typical case presupposes a designer who owns and can reweight an objective function, yet the prime's own T6 tension concedes a non-deliberative emergent form (median-voter drift, popularity feedback) with no editable objective, so it does not strictly require a human practice. import_vs_recognize is 0.5: invoking the prime imports a structural-diagnosis-and-repair frame (reweight, retune, replace, supplement) as much as it recognises a mass-weighted optimum already present. The relational core is genuine and the criteria sit evenly, correctly placing the prime at the framed midpoint rather than the structural pole.

Substrate Independence

Majority-dominated aggregate objective is a strongly substrate-independent prime — composite 4 / 5 on the substrate-independence scale. Its domain breadth (4 / 5) is wide: the mass-weighted-optimum-with-minority-blindness pattern recurs across machine learning (a loss averaged over a skewed dataset optimizing for the majority class), electoral and majoritarian politics, attention economics, public health (interventions tuned to the modal patient), regulation, education (curricula pitched at the median student), recommendation systems, and insurance. The structural abstraction (4 / 5) reflects that the core insight — an average is not a neutral summary but a weighting that places the optimum on the majority side and renders an operationally-important minority blind by construction — is genuinely abstract and relational, applicable wherever per-instance values are combined into one objective over a skewed distribution. What keeps it below the structural pole is that the instances are predominantly designed-objective substrates: the pattern bites where someone has chosen an aggregate objective to optimize, so it carries a mild design-and-fairness framing rather than being a pure fact about distributions. The transfer evidence (4 / 5) is concrete: the same diagnosable pathology and the same repair kit (reweight the loss, retune the threshold per group, replace the aggregate with a min-over-groups objective, supplement with subgroup monitoring) appear across ML fairness work, public-health equity analysis, and electoral theory, with aggregate performance shown to under-determine per-group performance in each. The pattern is recognized rather than imported wherever a skewed distribution is collapsed into one optimized figure, but its tie to chosen objectives holds the composite at the strong-but-not-maximal band.

  • Composite substrate independence — 4 / 5
  • Domain breadth — 4 / 5
  • Structural abstraction — 4 / 5
  • Transfer evidence — 4 / 5

Relationships to Other Primes

One-hop neighborhood: parents above, mutual partners to the right, children below.Majority-DominatedAggregate Objectivecomposition: AggregationAggregation

Parents (1) — more general patterns this builds on

  • Majority-Dominated Aggregate Objective presupposes Aggregation

    The file: this prime IS 'a specific, diagnosable pathology of aggregation' — an additive/expected-value objective whose mass concentrates on a skewed majority so the optimum is minority-blind by construction. It presupposes the aggregation operation (the average revealed as a weighting) and names its failure mode.

Path to root: Majority-Dominated Aggregate Objective → Aggregation → Micro Macro Linkage

Neighborhood in Abstraction Space

Majority-Dominated Aggregate Objective sits in a sparse region of abstraction space (77th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely rather than landing on a neighbor.

Family — Aggregation & Scale Artifacts (16 primes)

Nearest neighbors

Computed from structural-signature embeddings · 2026-06-14

Not to Be Confused With

The nearest neighbour is aggregation (similarity ~0.90), and the boundary is the prime's whole point: aggregation is the neutral operation of combining per-instance values into one figure, while this prime is the failure mode that operation exhibits when the input distribution is skewed. The decisive insight — "the average is not a neutral summary; it is a weighting, and the weighting can be the failure" — is precisely what distinguishes them. Aggregation tells you how values combine; the prime tells you that, on a skewed distribution, the combination places the optimum on the majority side and renders an operationally-important minority blind by construction. A practitioner who reasons only about aggregation will compute the summary and trust it; the prime insists that aggregate performance under-determines per-group performance, so a 99%-accurate optimum can serve a subgroup arbitrarily badly. The prime is thus a specific, diagnosable pathology of aggregation, carrying its own repair kit (reweight, retune, replace, supplement), not the aggregation operation itself.

The prime is also confusable with multiobjective_optimization, because both concern a tension between serving different populations. But multiobjective optimisation explicitly carries multiple objectives and trades among them along a frontier — the conflict is represented in the problem. This prime has a single aggregate objective in which the minority is formally included (every minority instance contributes its term) yet operationally neglected because the mass sits elsewhere. The failure is invisible because there is only one objective and the minority is nominally inside it; there is no second objective whose trade-off makes the neglect visible. The remedy reflects the difference: multiobjective optimisation negotiates weights between declared objectives, whereas this prime must change the single objective's mass-weighting (or split it into per-group metrics). A practitioner who frames the problem as multiobjective will look for a frontier that the single-objective formulation never exposed, missing that the fix is to restructure the one objective.

A subtler confusion is with goodharts_law, and the two genuinely interlock. Goodhart's law says a measure, once made a target, ceases to be a good measure. This prime's first-order claim is different: the aggregate objective is minority-blind even when honestly optimised, before any gaming — the neglect is a property of the optimum on a skewed distribution, not of strategic target-chasing. But Goodhart reappears recursively inside the prime: once a minority-aware metric is introduced to repair the neglect, the system optimises that metric, and the same mass-weighting risk can re-create a sub-minority blindness one level up (the prime's T3 tension). So Goodhart is not the prime but a recurring hazard within its repair loop. A practitioner who frames the original neglect as Goodhart will hunt for gaming where there was none, and may miss the genuinely recursive Goodhart that the minority-aware fix introduces.

These distinctions are load-bearing because each mis-frame mis-targets the repair. Treating the problem as neutral aggregation trusts a summary that is structurally minority-blind; treating it as multiobjective_optimization seeks a frontier the single objective never exposed; treating it as goodharts_law hunts gaming in an honestly-optimised optimum. The prime's contribution is to relocate the intervention to the objective function's mass-weighting and to supply the reweight/retune/replace/supplement kit — while warning that the kit, applied naively, regenerates the same weighting one level up.

Solution Archetypes

No catalogued solution archetypes reference this prime yet.