WHY MODELING? ON THE SOCIETAL VALUE OF PREDICTIONS WITH GREAT UNCERTAINTY

Quantitative models of physical and chemical processes are used routinely to make predictions across the suite of human-induced environmental and natural hazards problems. These models, which generally require intensive computation, are an attempt to provide society with "the best" predictions possible to make political and economic decisions. Due to data and computational limitations and the inherent complexity of environmental systems, these models typically are not able to provide predictions with a reasonable degree of uncertainty. The tendency is to ignore this limitation and provide "bald forecasts" or underestimate the magnitude of uncertainty present. This tendency reflects a societal need to have predictions regardless of their quality. Predictive models serve a function that is both useful to society in a sense of providing a tool, albeit flawed, for decision-making, and can provide a calming influence in a time of environmental crisis. The scientific basis for using these models in real-world applications may often be questionable, but their application likely is necessary.