Causal Epistemic Reconstruction: A Method for Historical Identification Audits

Historical causal analysis often proceeds from surviving observations, records and institutional structures toward claims about how the present acquired its form. That inference becomes more difficult when earlier models, classifications, measurement systems or other epistemic commitments did not merely describe their environments but became embodied in institutions and practices that altered those environments. The resulting problem is not solved by establishing performativity or feedback alone. A further question must be answered: does sufficient historical information survive to distinguish the causal contribution of epistemic intervention from competing explanations of the present? This paper develops Causal Epistemic Reconstruction, a methodological procedure for conducting historical identification audits. The method operationalizes the theory of epistemic provenance non-identifiability through eight stages: causal-query specification, epistemic-commitment identification, embodiment mapping, effect specification, construction of competing historical models, provenance audit, counterfactual specification, and identification-regime classification. Its central discipline is that incomplete history, feedback, performativity and information loss are not themselves evidence of non-identifiability. A stronger condition is required: surviving evidence must fail to discriminate among serious historical causal models that imply materially different answers to the query under investigation. Two worked cases illustrate the method. The historical sociology of option-pricing theory provides a calibration case in which epistemic intervention is causally significant but substantial provenance remains available. The historical trajectory of Multiple Personality Disorder and Dissociative Identity Disorder provides a diagnostic stress test in which classification, clinical practice and reported presentation may interact, but where looping effects alone remain insufficient to establish structural provenance underdetermination. The paper concludes by treating archives, rejected alternatives, measurement histories and comparison structures as potential identification infrastructure. Selective preservation or destruction of query-discriminating provenance can alter not merely historical memory but the causal claims available to future investigators. The method is intended not to maximize scepticism but to calibrate causal warrant to the discriminatory power of surviving historical evidence.