When prevention cannot learn: evaluability failures in risk-control regimes

Governments and regulators often respond to corruption, accidents, administrative error, and regulatory failure by adding preventive controls. Some controls improve safety and integrity. Others remain in place even when their contribution and organizational costs are uncertain. This article develops the Evaluability Loop of Risk-Control Regimes, a process model explaining how evidence about a control becomes, or fails to become, an authorized decision about its future. A control is evaluable when a regime can explain how it should affect risk, observe how it operates, interpret the resulting signals, assess both performance and burden, and decide whether it should be retained, recalibrated, replaced, suspended, or retired. The model rejects the assumption that preventive instruments are inherently difficult to evaluate or that ex post accountability naturally produces learning. It identifies six failure modes and explains how their interaction with political and institutional asymmetries can produce a procedural ratchet. Illustrations from safety management, regulatory enforcement, and crisis accountability show that preventive controls can support learning, while observable ex post events can still produce weak evidence or blame displacement. The article’s central contribution is to specify the institutional passage from control experience to control revision.