Epistemic Defence-in-Depth in Mature Safety Regimes

Safety governance in high-hazard systems increasingly emphasizes explicit treatment of uncertainty and the strength of knowledge. Yet major accidents continue to reveal a persistent epistemic problem: even where uncertainty is acknowledged and safety knowledge appears mature, inquiry can become progressively constrained. This paper addresses this challenge by shifting attention from epistemic awareness to epistemic closure, a condition in which particular risk framings stabilize and alternative interpretations cease to be actively explored.To examine this problem, the paper introduces epistemic defence-in-depth (e-DiD) as a meta-level governance mechanism concerned with how attention, effort, and epistemic confidence are distributed in safety decision-making. Reinterpreting defence-in-depth as an epistemic practice, e-DiD emphasizes institutionalizing a balance between operating within established safety frames and periodically questioning those frames, irrespective of whether knowledge is judged to be weak or strong.The argument is illustrated through an analysis of the Fukushima Daiichi accident, focusing on pre-accident safety practices related to tsunami hazards and station blackout. Epistemic closure emerged under contrasting conditions: in the tsunami case, acknowledged uncertainty led to intensified but narrowly framed analytical effort, while in the station blackout case, confidence grounded in data and operational experience discouraged re-examination of foundational assumptions.Against this backdrop, the paper reinterprets safety goals not merely as benchmarks of acceptable risk but as institutional anchors for epistemic re-balancing, providing structured occasions to reflect on how confidence is formed and stabilized, thereby sustaining epistemic vigilance.