An important task of fisheries management is deciding amongst alternative policy options. In doing this, policymakers must anticipate, typically using models, how key elements and dynamics of the system are likely to change in the future, and evaluate how the outcomes of management policies might be affected by this change. However, the future is loaded with uncertainty and surprise, and generating accurate, long-range biological, economic or political forecasts is a major challenge. In some regions, improved understanding of system dynamics and breakthroughs in computing power have led to the development of whole-of-system models (e.g. Atlantis, [22]), which has gone some way to improving the accuracy of forecasts. However, this depth of understanding and complexity of modelling is still beyond reach in most fishery systems, and in complex and uncertain systems the usefulness of modelled predictions of the future is limited