View article

Deep uncertainty can be defined as the situations where the parties to a decision cannot agree on model representation, probability distributions in the model and the evaluation of the outcomes (Lempert et al, 2003). In the presence of deep uncertainty, decision making becomes a difficult task. Several method (ologie) s–mostly model-based–have been developed and are currently used for dealing with this difficulty. Models are often considered to be mathematical representations of real world systems where modelers make certain assumptions about the system of interest. When using models to predict the future, modelers assume their models to be true. However, under deep uncertainty, that claim does not hold.

In policy making, one most common way is still to seek an optimal policy which outperforms all other alternatives. However, in the presence of deep uncertainty, the assumptions under which optimal policies are chosen may not hold anymore. Adaptive and robust policies that perform well enough across an ensemble of plausible futures may well be preferable over static policies that are optimal for one or few futures (Lempert et al, 2003). Hence, there is a strong need for not static but adaptive policies which are “more robust to a range of anticipated conditions and can adapt over time”(Swanson et al, 2010).