Robust Derivation of Risk Reduction Strategies

Effective risk reduction strategies can be derived mechanically given sufficient characterization of the risks present in the system and the effectiveness of available risk reduction techniques. Quantitative assessments of risks and risk reduction techniques are likely to be inaccurate. In this paper we describe sensitivity analysis experiments which we carried out to evaluate how inaccurate quantification of risk and risk reduction techniques affect the performance of mechanically derived risk reduction strategies. Our experiments show that mechanically derived risk reduction strategies are likely to produce significant improvements in risk reduction compared to a alternative risk reduction strategies, and arguably should be used as a matter of course, even when knowledge of risk and risk reduction techniques is very inaccurate.