Modelling Uncertainty

Abstract

Simulation models are increasingly used for exploring the consequences of deep uncertainty in complex societal issues. The complexity of societal grand challenges, often characterised by the interrelatedness of different elements in the systems underlying these challenges, often renders mental simulation impossible, necessitating the use of simulation models to assist human reasoning. In addition, these grand challenges are typically also subject to deep uncertainty, making it, for example, impossible to come to a shared understanding of parts of the system and exogenous inputs to it, or even a shared problem definition.
Under deep uncertainty, simulation models can be used to explore the consequences of different combinations of assumptions about uncertain factors or attributes of the problem situation and the underlying system. This type of simulation model use was introduced in 1993 as Exploratory Modelling and Analysis (EMA). In more recent years, this approach has become a major underpinning of the Decision Making under Deep Uncertainty (DMDU) field.
The treatment of deep uncertainty in much DMDU research can be improved, however. In most DMDU research to date, pre-existing models are used. These models were generally developed for ‘consolidative’ use: the modellers tried to unify existing knowledge to come a single, ‘best’ model. While most modellers will agree that these models are not perfect representations of reality, and often agree that they as such cannot be validated in the strict sense of the word, these modellers and their models do not acknowledge deep uncertainty. The use of consolidative models is arguably problematic if one agrees that the issue at hand is characterized by deep uncertainty. Therefore, models are needed that are explicitly developed for ‘exploratory’ use: models that explicitly incorporate deep uncertainty potentially relevant for the research question or questions at hand. However, little experience and guidance exists regarding development and use of specifically exploratory models.
In this dissertation, a first attempt is made to identify, and provide guidance for, the critical choices made during the development and use of exploratory models.