Definition Determines Mechanism Why the Way We Define Disease Shapes — and Sometimes Misdirects — the Search for a Cure

Medicine often treats disease definitions as descriptive labels rather than as decisions with causal consequences. This paper argues the opposite: a disease's definition fixes the conceptual frame within which mechanistic investigation, target identification, and therapeutic search proceed. The central claim — the Visibility Principle — is falsifiable: mechanistic detail accumulated downstream of a misidentified primary cause converts into validated therapeutic targets at a lower rate than detail accumulated under a causally faithful definition. Formally, the conversion rate C = V₂/V₁ falls as mechanistic output M grows when the elasticity of validated targets with respect to M is lower than the elasticity of nominated targets — a condition predicted to hold when causal fidelity D is below a threshold D* = −β_M/β_DM. The empirical programme uses a bounded binomial model for V₂ given V₁ (log link; V₂ ≤ V₁), checked against a negative binomial count model, with disease-level bootstrap inference. A pre-registered simulation study (Stage A) has confirmed that the procedure satisfies its null-scenario calibration gates across all 16 cells (8 scenarios × 2 sample sizes, 5,000 reps each). Stage B power evaluation has not been run. The paper also develops a state-space view of illness as a context- and tissue-specific deviation from a dynamic healthy reference, a flow-based taxonomy for classifying candidate mechanisms as causal, consequential, or compensatory, and an analysis of four historical cases. Counterexamples the principle must survive are stated explicitly. Version 4.5 incorporates a flow-informed causal audit, an abstention rule for binding datasets, and the Stage A simulation calibration result. keywordsphilosophy of medicine, disease definition, causal fidelity, drug discovery, therapeutic target validation, Visibility Principle, biostatistics, simulation calibration, preregistered study