Julian Hunt FRS (UK’s House of Lords, Earth Sciences, University College London and Lighthill Institute of Mathematical Sciences)

Increasingly all large organisations including governments and large industries are using systems modelling to analyse complex and multifarious activities, to make predictions and then make decisions for their present and future operations. These may well be iterative processes and continually adjusted as events unfold. For political organisations (which means most today) such systems-based decisions have to be explained and justified – especially when the decisions have to change quite rapidly. Previous systems-based analyses, predictions and decisions include conflict, environmental pollution, disease, organisational changes/issues; and the mathematical methods involved: deterministic coupled equations, dynamical systems analysis, statistical/probabilistic modelling; and then the new challenges of climate change: risks, changing physical and societal systems (e.g. megacities), resource issues.

Some of the big questions for system modellers working with decision makers are how to focus on key issues of optimal combination of micro and macro modelling (e.g. using data from discrete bottom up, simulation/computational methods to support or critique top-down, macro analyses of critical aspects – which are much easier to communicate and use for decision making); how systems have metastable states with sometimes rapid transitions between them – which should be explained in public decisions (highly relevant in modelling disaster/planning scenarios); and lastly how publication of predictions and future plans resulting from modelling can cause society to change its actions, and possibly invalidate the predictions! Does this imply that public targets/policies may have to be deliberately in error to obtain the desired outcome? This may require a new approach to error analysis of such systems.