Extensions of stochastic multiscenario models for long-range planning under uncertainty

Abstract

In the twenty-first century, the goals of mankind are evolving from rising prosperity to attaining long-range sustainable survival in an acceptable state. Such change will lead to a paradigm shift predicted by Thomas Kuhn. This situation motivates two novel generalized systems, Risk-Constrained Optimization (RCO) and its derivative, Risk-Constrained Optimization/Decision Network (RCO/DN). RCO aims for flexible and robust strategies across large ranges of scenarios and risks. It assumes that for any “serious” (complex and long-range) planning and decision-making problem, externalities are important and uncertainty is radical. Therefore, it rejects the very concept of the correct or the best strategy, replacing it with a strategy that is the most acceptable to decision-maker(s). RCO considers decision-support systems, models, and algorithms to be tools of analysis, rather than methods of selection of the best. It replaces the artificial and risky paradigm of maximization by a cautious natural and evolutionary paradigm of catastrophe avoidance. RCO filters out the worst and riskiest candidate strategies, leaving for judgmental selection a small set of flexible, robust, and reasonably safe strategies. For that purpose, RCO employs enhanced stochastic multiscenario (ESMS) models, transforming them into optimizing filters. It is important that this operation is achieved in accordance with a new principle—changing the overall solution by affecting the values of scenario-specific, rather than general, outcome variables. Furthermore, RCO screens decisions and strategies by several synthetic criteria in a framework of novel strategic frontiers. Thus, RCO embeds ESMS into an ensemble of mutually supporting risk management techniques, legitimizing high-level analytical use of a computer plus optimization model combination.