Systemic Uncertainty Engineering (SUE): A Quantitative Framework for Risk Reduction in Complex Socio-Technical Systems

Modern safety-critical systems fail at the boundary between engineered products and their operational environments rather than from isolated component faults. Existing domain-specific standards for functional safety, performance sufficiency, and cybersecurity each address one slice of this boundary but provide no unified method for measuring how uncertainty propagates across their combined scope, nor any instrument for identifying failure combinations that span multiple domains simultaneously. This paper proposes Systemic Uncertainty Engineering, a quantitative framework that treats uncertainty as a measurable, propagating system property and expresses residual systemic risk as expected financial loss. The framework was constructed through theoretical development and retrospective empirical validation. A four-quadrant uncertainty model decomposed uncertainty along reducibility and origin axes, establishing the measurement structure for a lifecycle-spanning propagation model with linear and nonlinear interaction terms. A dualprocess model ordered analytical activities from product-environment interface characterization through risk assessment, goal architecture, and economic translation. The framework was instantiated for autonomous vehicle development and validated against six documented failures spanning five decades of automotive engineering history. Three findings emerged that existing single-domain methods cannot produce. Cross-domain minimal cut sets spanning functional safety, performance sufficiency, cybersecurity, and organizational domains were identified before domain decomposition occurred. An 80cell risk tensor quantified residual uncertainty across all domains simultaneously and translated it into expected financial loss and return-on-investment metrics. Retrospective analysis confirmed that the constructs would have identified each failure’s dominant risk pathway before deployment in all six cases. The framework demonstrates applicability to five additional technology domains sharing the structural conditions of novelty, open-world operation, and multi-domain regulatory oversight.