Under the traditional model—called here “predict-then-act”—a planner makes a prediction of what future conditions will be, and, based on that prediction, they make a decision. This can be thought of as an optimization method, creating the maximum benefit for the minimal cost. These types of methods work well when uncertainties are small.
When uncertainties are deep, “predict-then-act” methods can break down. If decision makers develop their own form of “tunnel vision,” focusing on some concerns and motivations without seeing the big picture, they may fail to consider uncertainties in the future. This can lead to decision-making gridlock or unexpected outcomes.
Rather than seeking confidence in a specific model, under DMDU a planner is seeking confidence in a decision. DMDU begins with a proposed strategy and continues with stress tests of that strategy using multiple model runs to understand how it would perform under a range of plausible future conditions. Stress tests identify conditions under which a proposed strategy performs well or poorly, which then informs revised strategies that fill gaps. The process can be used iteratively to test new proposals until decision makers are confident that they have a set of robust options. Rather than seeking confidence in a specific model, under DMDU someone is seeking confidence in a decision. In this approach, important questions about deep uncertainty are asked to bolster confidence and investment in the decision and to eliminate a narrow view of the future:
- Can a robust and flexible strategy perform well under a range of future conditions?
- What uncertainties are most important?
- What actions do we need to take now?
- What actions can we postpone to the future?