{
  "schemaVersion": "1.3",
  "slug": "aggregation-without-sufficiency",
  "title": "Aggregation Without Sufficiency: When Macro Summaries Do Not Determine Macro Responses",
  "shortTitle": "Aggregation without sufficiency",
  "url": "https://evidencepress.org/releases/aggregation-without-sufficiency/",
  "oneLine": "A summary can be correct yet still omit information needed for a decision; an aggregation licence tests what that summary may safely determine.",
  "abstract": "Governments, scientists and managers routinely act on compact summaries of complex systems. A summary can be measured correctly yet still fail to determine the quantity or response needed for a decision. This unrefereed preprint proposes an aggregation licence for detecting and repairing that failure. Its four licence questions ask whether two admissible worlds can share an aggregate while requiring different answers, then separate local point identification, exact global factorisation, minimax prediction and decision regret from the operational gates needed for use. Economics, fisheries and epidemiology provide three distinct illustrations. Producer-side certificates include collision counterexamples, positive controls and a sealed synthetic workflow in which CPUE alone fails, an independent survey passes the baseline tolerances and a held-out catchability shock revokes the licence. The bounded contribution is a cross-domain, versioned reporting and assurance wrapper, not a new factorisation theorem, learning algorithm, operational validation or evidence of empirical effectiveness.",
  "datePublished": "2026-08-13",
  "dateModified": "2026-08-13",
  "version": "0.4.0",
  "doi": "10.5281/zenodo.21913278",
  "doiUrl": "https://doi.org/10.5281/zenodo.21913278",
  "conceptDoi": "10.5281/zenodo.21913277",
  "pdfUrl": "https://github.com/ipitchford/aggregation-without-sufficiency/releases/download/v0.4.0-preprint/aggregation-without-sufficiency-v0.4.0-preprint.pdf",
  "altPdfUrl": "https://zenodo.org/records/21913278/files/aggregation-without-sufficiency-v0.4.0-preprint.pdf?download=1",
  "zenodoUrl": "https://zenodo.org/records/21913278",
  "repoUrl": "https://github.com/ipitchford/aggregation-without-sufficiency",
  "releaseUrl": "https://github.com/ipitchford/aggregation-without-sufficiency/releases/tag/v0.4.0-preprint",
  "markdownUrl": "https://evidencepress.org/releases/aggregation-without-sufficiency/index.md",
  "bibtexUrl": "https://evidencepress.org/releases/aggregation-without-sufficiency/cite.bib",
  "audioUrl": "https://evidencepress.org/assets/audio/aggregation-without-sufficiency.mp3",
  "imageUrl": "https://evidencepress.org/assets/og/aggregation-without-sufficiency.png",
  "coverArtUrl": "https://evidencepress.org/assets/art/aggregation-without-sufficiency.svg",
  "media": [
    {
      "type": "audio",
      "url": "https://evidencepress.org/assets/audio/aggregation-without-sufficiency.mp3",
      "name": "Audio briefing — aggregation without sufficiency",
      "description": "Plain-English synthetic-voice summary of the missing-information problem, aggregation licence and assurance boundary; a communication aid, not additional scientific evidence.",
      "transcriptUrl": "https://evidencepress.org/assets/audio/aggregation-without-sufficiency.txt"
    }
  ],
  "authors": [
    "Anonymous"
  ],
  "license": "CC0-1.0",
  "status": "unrefereed-preprint",
  "verification": {
    "peerReviewed": false,
    "independentlyReproduced": false,
    "formallyVerified": false,
    "internallyReplayed": true,
    "detail": "Anonymous unrefereed preprint authorised for public release. Producer-side derivation, source-routing, terminology-parity, replay and package-integrity checks pass within their stated scope. The package has not received editorial or field-specialist peer review and has not been independently reproduced or formally verified. Public identifiers become authoritative only after exact release readback."
  },
  "assurance": [
    {
      "dimension": "availability",
      "label": "Availability and archiving",
      "question": "Is the evidence package publicly retrievable from an archive under a persistent identifier?",
      "state": "passed",
      "evidenceUrl": "https://doi.org/10.5281/zenodo.21913278",
      "note": "The GitHub prerelease and Zenodo record expose byte-matched ZIP, PDF, replay receipt and SHA-256 ledger. This establishes public availability and identity, not correctness or independent reproduction."
    },
    {
      "dimension": "internalReplay",
      "label": "Internal replay",
      "question": "Does the producer’s own pipeline reproduce the stated result from the archived package?",
      "state": "passed",
      "note": "Thirty-five producer-side tests pass under normal and optimized Python; the package verifier rechecks claims, source identities, anonymous attribution, component licensing, frozen protocols, generated receipts and the complete manifest.",
      "evidenceUrl": "https://github.com/ipitchford/aggregation-without-sufficiency/releases/tag/v0.4.0-preprint"
    },
    {
      "dimension": "independentRerun",
      "label": "Independent rerun",
      "question": "Has someone else run the supplied implementation and obtained the stated result?",
      "state": "not-assessed",
      "note": "No unaffiliated party has reported rerunning the immutable package."
    },
    {
      "dimension": "independentReimplementation",
      "label": "Independent reimplementation",
      "question": "Has someone else reached the result from an independent implementation?",
      "state": "not-assessed",
      "note": "No separately authored checker has reconstructed the finite collision and licence predicates without producer code."
    },
    {
      "dimension": "formalVerification",
      "label": "Formal verification",
      "question": "Is a formalised statement machine-checked, and over which trusted base?",
      "state": "not-assessed",
      "note": "The factorisation argument and executable checks have not been formalised in a proof assistant over a stated trusted base."
    },
    {
      "dimension": "specialistReview",
      "label": "Specialist review",
      "question": "Has a domain specialist assessed the argument?",
      "state": "not-assessed",
      "note": "The retained substantive reviews are not external field-specialist reviews in economics, fisheries or mathematical epidemiology."
    },
    {
      "dimension": "editorialPeerReview",
      "label": "Editorial peer review",
      "question": "Has a journal or venue run peer review to a decision?",
      "state": "not-assessed",
      "note": "No journal or comparable venue has conducted editorial peer review to a decision."
    },
    {
      "dimension": "dataEnvironmentReproducibility",
      "label": "Data and environment reproducibility",
      "question": "Are data and computational environment pinned well enough to rebuild?",
      "state": "partial",
      "evidenceUrl": "https://github.com/ipitchford/aggregation-without-sufficiency/blob/v0.4.0-preprint/ENVIRONMENT.txt",
      "note": "The certificate code uses only the Python standard library and CI passes on Python 3.12, 3.13 and 3.14, but no unaffiliated cross-platform reconstruction is reported."
    }
  ],
  "provenance": {
    "aiGenerated": false,
    "aiAssisted": true,
    "generatedBy": [
      "AI-assisted research, drafting and checking tools"
    ],
    "humanRole": "Ian Pitchford maintains and publishes the repository and release infrastructure; this operational role is not scholarly authorship. Human scientific judgment, rights confirmation and publication authorisation remain human gates.",
    "disclosure": "Automated tools assisted source recovery, computation, synthesis, drafting and producer-side checking. They are not assigned scholarly authorship. The public scholarly creator is Anonymous."
  },
  "problem": null,
  "corrections": [],
  "keywords": [
    "aggregation",
    "sufficiency",
    "identification",
    "representative agent",
    "aggregate production function",
    "catch per unit effort",
    "hyperstability",
    "herd immunity",
    "next-generation operator",
    "partial identification",
    "decision regret"
  ],
  "keyResults": [
    "An exact target rule through an aggregate exists if and only if the target is constant on every admissible fibre.",
    "Local point identification, exact global set-theoretic factorisation, minimax prediction and decision regret are four licence questions; none is a proxy for all the others.",
    "Exact set-theoretic factorisation does not automatically establish measurability, stability, computability, model-class membership or learnability.",
    "The package distinguishes closure and representation failure, inverse-identification failure and intervention-sufficiency failure.",
    "Producer-side checks detect same-total-income and CPUE collision worlds and an equal-R0 targeted-threshold collision.",
    "Uniform independent perfect immunisation is retained as a positive control: both fixed heterogeneous toy operators recover the 60% threshold at R0 = 2.5.",
    "A pre-execution-sealed, fully specified synthetic workflow demonstration records baseline failure, an independent-survey pass and held-out revocation; a separate post-review receipt maps sensitivity phase changes.",
    "The proposed aggregation licence is target-, intervention-, loss- and expiry-specific; it permits aggregation where the declared target is preserved."
  ],
  "reviews": [],
  "evidencePackage": "A self-contained revised synthesis; three technical dossiers; 37-source identity-audited register; 16-claim machine ledger; standard-library collision, frozen-licence and post-review sensitivity evaluators; source-routing, author/DOI mutation and cross-format parity tests; source archive hashes; local PDF render and package verification. Final counts and environments are in the replay receipt. These are producer-side checks, not independent reproduction or field-specialist review.",
  "openProblems": [
    "Obtain independent specialist review of the economics theorem scope, production interpretation, northern-cod observation process and epidemiological operator conventions.",
    "Evaluate the current machine-readable aggregation-licence schema on real pre-registered cases and compare it with ordinary modelling reports.",
    "Develop benchmark sets containing both known aggregation failures and known exact or decision-adequate positive cases.",
    "Evaluate augmentation options by target-diameter and decision-regret reduction after measurement and monitoring costs.",
    "Run pre-registered held-out intervention tests on empirical rather than synthetic cases.",
    "Obtain unaffiliated replay and specialist review while preserving the anonymous creator protocol."
  ],
  "relatedWorks": [
    {
      "citation": "Sonnenschein, H. (1972). Market Excess Demand Functions. Econometrica, 40(3), 549–563.",
      "url": "https://doi.org/10.2307/1913184"
    },
    {
      "citation": "Felipe, J., & Fisher, F. M. (2003). Aggregation in Production Functions: What Applied Economists Should Know. Metroeconomica, 54(2–3), 208–262.",
      "url": "https://doi.org/10.1111/1467-999X.00166"
    },
    {
      "citation": "Rose, G. A., & Kulka, D. W. (1999). Hyperaggregation of Fish and Fisheries. Canadian Journal of Fisheries and Aquatic Sciences, 56(S1), 118–127.",
      "url": "https://doi.org/10.1139/f99-207"
    },
    {
      "citation": "Delmas, J.-F., Dronnier, D., & Zitt, P.-A. (2023). Optimal Vaccination: Various (Counter) Intuitive Examples. Journal of Mathematical Biology, 86, 26.",
      "url": "https://doi.org/10.1007/s00285-022-01858-5"
    },
    {
      "citation": "Abel, D., Hershkowitz, D., & Littman, M. (2016). Near Optimal Behavior via Approximate State Abstraction. PMLR 48, 2915–2923.",
      "url": "https://proceedings.mlr.press/v48/abel16.html"
    },
    {
      "citation": "Poli, M., Massaroli, S., Ermon, S., Wilder, B., & Horvitz, E. (2023). Ideal Abstractions for Decision-Focused Learning. PMLR 206, 10223–10234.",
      "url": "https://proceedings.mlr.press/v206/poli23a.html"
    },
    {
      "citation": "Ye, Y., Amin, S., & Ozdaglar, A. (2026). Learning Decision-Sufficient Representations for Linear Optimization. arXiv:2603.18551v2.",
      "url": "https://arxiv.org/abs/2603.18551"
    }
  ],
  "operatingModel": {
    "version": "1.0",
    "workId": "ep-work:aggregation-without-sufficiency",
    "attemptIds": [
      "ep-attempt:aggregation-without-sufficiency"
    ],
    "aims": [
      "science",
      "policy"
    ],
    "artifactRoles": [
      "research-output",
      "evidence-assessment",
      "method-demonstration",
      "communication"
    ],
    "lineageId": null,
    "accelerationPrimitives": [
      "structural-compression",
      "adversarial-controls",
      "identification-gate",
      "partial-identification",
      "counterexample-proxy-first",
      "assurance-vector",
      "agent-readable-research-object"
    ],
    "decisionObject": {
      "type": "reusable-method",
      "description": "A claim-specific aggregation licence with fibre collision certificates, four licence questions, operational regularity gates, target-preserving augmentation within a declared family of options, held-out intervention tests and expiry.",
      "scope": "Scientific and policy uses of aggregates where a microstate, compression, target, admissible state set and intervention class can be stated."
    },
    "bottleneckTargeted": [
      "discovery",
      "assurance"
    ],
    "semanticBridge": {
      "state": "explicit",
      "description": "The common factorisation language is used as a diagnostic wrapper while SMD and production closure, CPUE inverse identification and epidemic intervention paths retain separate field-specific objects and evidence.",
      "remainingRisks": [
        "The shared abstraction may omit field distinctions that alter admissibility, target meaning or decision loss.",
        "Finite toy certificates do not establish collisions over realistic empirical state sets.",
        "A narrow admissible set or post hoc tolerance can make a weak aggregate appear licensed.",
        "Novelty and external specialist review remain unestablished; anonymity and the component licence boundary must remain consistent across release surfaces."
      ]
    },
    "humanJudgmentGates": [
      "Judge whether each domain's microstate, aggregate, target and intervention have been represented faithfully.",
      "Assess the credibility of admissible state sets, identifying assumptions, loss functions and tolerances.",
      "Evaluate welfare, distributional, rights and risk questions that cannot be delegated to structural checks.",
      "Confirm anonymous scholarly attribution, component licences and public release authority."
    ],
    "parentLinks": [],
    "assuranceTarget": {
      "dimensions": [
        "independentRerun",
        "independentReimplementation",
        "specialistReview",
        "semanticValidation",
        "noveltyAssessment",
        "priorityAssessment"
      ],
      "nextAction": "Freeze the reviewed local package, obtain independent field reviews of the three semantic bridges, and have an unaffiliated checker reimplement the finite collision certificates without producer code.",
      "claimCeiling": "An elementary exact set-theoretic criterion and a cross-domain, versioned reporting and assurance wrapper with producer-checked demonstrations; not a new factorisation theorem, learning algorithm, operational validation, empirical prevalence estimate, universal aggregation failure, independently reproduced result or demonstrated improvement in science or policy."
    },
    "impactClaims": [
      {
        "id": "science-no-impact",
        "aim": "science",
        "outcome": "Improved claim-specific reporting of aggregation validity",
        "setting": "Cross-domain research using low-dimensional summaries",
        "status": "NO_IMPACT_EVIDENCE",
        "designClass": "none",
        "comparator": "No matched modelling or reporting workflow was evaluated.",
        "estimand": "No effect on error rates, review effort, model choice or scientific outcomes was estimated.",
        "evidenceRefs": [],
        "registeredDesignRef": null,
        "independentAssessment": null
      },
      {
        "id": "policy-no-impact",
        "aim": "policy",
        "outcome": "Fewer unsupported policy targets inferred from aggregate indicators",
        "setting": "Decision-facing aggregate models and dashboards",
        "status": "NO_IMPACT_EVIDENCE",
        "designClass": "none",
        "comparator": "No matched policy-assessment workflow was evaluated.",
        "estimand": "No effect on policy quality, time, adoption or outcomes was estimated.",
        "evidenceRefs": [],
        "registeredDesignRef": null,
        "independentAssessment": null
      }
    ]
  }
}