Scientific Metadesign

Abstract

Identifiability, Intervention, Structured Nulls, and Evidential Claim Ceilings The general framework is a completed methodological construction. This manuscript develops Scientific Metadesign, a framework for specifying the scientific objects, evidential gates, result classes, and claim ceilings that govern such distinctions. The canonical architecture comprises research, claim, observation, intervention, structured-null, evidence, and governance objects. The turbulence program in Part IX is a prospective, executable scientific stress test. Except for a bounded analytic derivation of translation invariance under stated assumptions, Epistemic Status and Manuscript Scope Scientific Metadesign is a methodological framework for designing inquiries whose possible outcomes remain scientifically interpretable. It addresses a recurring problem across empirical, computational, and theoretical work: a study may generate a numerical difference, a predictive model, or an intervention effect without establishing the stronger claim suggested by ordinary scientific language. The framework therefore separates the research object, claim object, observation model, intervention, structured null, evidence object, and governance object. It also separates claim lifecycle, evaluation status, substantive result class, evidential ceiling, and confirmation status. These distinctions are designed to make invalid, unresolved, equivalent, contradictory, supportive, predictive, mechanistic, reparative, autonomous, and regenerative outcomes scientifically distinguishable rather than rhetorically interchangeable. Physical truth determines what is the case; scientific governance determines what the evidence licenses us to claim. Scientific inquiry often fails not because no observation was made, but because the relation between observation, comparison, intervention, and claim was insufficiently designed. A numerical difference may be real yet causally uninterpretable; a predictive model may be useful without explaining its target; an intervention may change an outcome without identifying a mechanism; and a rescued system may remain dependent on continuing support. Evaluation status is separated from substantive result class: only a valid evaluation may be classified as supported, practically equivalent, contradicted, or unresolved. Evidential ceilings distinguish association and discrimination, target-specific relation, positive-horizon prediction, intervention-supported mechanism, autonomous rescue or regenerative continuation, and independent confirmation. The framework formalizes structured nulls through declared transformations, preserved quantities, disrupted targets, admissibility conditions, and artifact audits. It also develops a recovery hierarchy from improvement through repair, rescue, verified support withdrawal, autonomy, and regeneration. Governance records prospective locks, adaptations, provenance, failures, salvage, and final claim control. A prospective turbulence stress test specifies a matched-amplitude phase-null benchmark, a carrier-resolved harmonic interaction-phase descriptor, positive-horizon warning, targeted intervention, and viable-successor analysis. Except for a bounded translation-invariance derivation under stated assumptions, the turbulence claims remain prospective and unevaluated. The framework’s central principle is that every possible result should have an honest scientific meaning and that no reported claim should exceed the evidence that gives it meaning. Keywords: scientific design; identifiability; structured nulls; causal inference; prediction; intervention; mechanism; repair; autonomy; regeneration; reproducibility; provenance; turbulence; Fourier phase; triadic interaction