Does climate adaptation policy need probabilities?

Estimating the likelihood of future climate change has become a priority objective within the research community. This is the case because of the advancement of science, because of user demand and because of the central role played by climate prediction in guiding adaptation policy. But are probabilities what climate policy really needs? This article reviews three key questions: (1) Why might we (not) need probabilities of climate change? (2) What are the problems in estimating probabilities? (3) How are researchers estimating probabilities? These questions are analysed within the context of adaptation to climate change. Overall, we conclude that the jury is still out on whether probabilities are useful for climate adaptation policy. The answer is highly context dependent and thus is a function of the goals and motivation of the policy analysis, the unit of analysis, timescale and the training of the analyst. Probability assessment in the context of climate change is always subjective, conditional and provisional. There are various problems in estimating the probability of future climate change, but reflexive human behaviour (i.e. actions explicitly influenced by information) is largely intractable in the context of prediction. Nonetheless, there is considerable scope to develop novel methodologies that combine conditional probabilities with scenarios and which are relevant for climate decision-making.