Regime Has a Contract: Intervention, Observation, and the Data Foundation of Causal Identification

Abstract

The Theory of Data distinguishes event and spine universes by their point-existence laws but does not, by that distinction alone, specify how values are generated at those points. This paper introduces regime as a separate governed object: the value-generation arrangement under which a member is produced. Observational and interventional regimes are therefore not additional universe types; they qualify member-generating laws over event and spine universes. The distinction separates regime from observation status, provenance, and evidential status and clarifies randomized experimental design. Treatment and control arms are observed under interventional assignment regimes, while assignment, treatment receipt, treatment delivery, outcome, and outcome observation remain distinct governed objects. The paper then gives a bounded Theory-of-Data reading of Judea Pearl's causal framework. Structural causal models and do-calculus provide a formal calculus for deriving interventional targets under declared causal assumptions, but the cross-regime structural contract is a separate input whose empirical warrant is not derived from the same observational-regime distribution. The paper develops regime-qualified member laws, regime-sensitive member identity, regime-local statistical bridges, cross-regime contracts, and evidence-status principles for causal claims.