Robust decision-making

Abbreviated RDM, this decision analysis technique was developed by scientists at RAND; it is most useful in situations of deep uncertainty. Decision-making is the cognitive process of selecting a course of action from among multiple alternatives. Every decision-making process produces a final choice, including the choice to do nothing. Decision-making is a reasoning process which can be rational or irrational, and can be based on explicit or tacit assumptions. RDM is a decision theoretic framework that makes systematic use of a large number of possibly imperfect forecasts. Rather than relying on improved point forecasts or probabilistic predictions, RDM explicitly embraces a multiplicity of plausible but different futures as the best representation of the available information about an indeterminate future. RDM is an iterative, quantitative approach for: (i) Identifying decision strategies whose good performance is relatively insensitive to key uncertainties facing decision makers and (ii) Characterizing the residual vulnerabilities of these strategies. RDM allows analysts to use a computer to lay out a large number and a wide range of plausible paths into the long-term future. This exploratory modeling is used to create a large ensemble of plausible future scenarios, in which each scenario represents one guess about how the world works and one choice among many alternative strategies people might adopt to influence outcomes. RDM is consistent with Bayesian decision analysis applied to a cost-benefit framework. RDM runs the Bayes machinery numerous times in order to find robust strategies and characterize the residual deeply uncertain factors to which they might remain vulnerable. It is useful to contrast RDM with the more common “predict-then-act” approach to planning, a process that requires calculation of the probability distributions of future outcomes of interest. Four key elements or principles should govern the form and design of these interactions. Consider ensembles of large numbers of scenarios. Seek robust, rather than optimal, strategies that do “well enough” across a broad range of plausible futures and alternative ways of ranking the desirability of alternative scenarios. Employ adaptive strategies to achieve robustness. Use computer tools designed for interactive exploration of the multiplicity of plausible futures. RDM can make use of a wide variety of models, including econometric models, optimal economic growth models, game theoretic models, system dynamics models, agent-based models of technology diffusion, Bayes Nets, neural nets, and a wide range of models built on Excel spreadsheets. RDM can be used in conjunction with narrative scenario-based planning methods to help suggest and evaluate small numbers of scenarios that can be considered in greater detail by human teams. RDM can also incorporate probabilistic information and a wide variety of expert judgments developed by elicitation, Delphi, Foresight, or other methods.