Abstract
The predominant strategy of climate modelling is to continually increase resolution and complexity of general circulation models (GCMs). At present, there are calls to double down on this strategy and invest a lot more financial and computational resource into GCM resolution and complexity, with the assumption that this will improve the usefulness of climate predictions to support climate adaptation decision making. We argue that this is not the best use of scientific effort. Because there are many different kinds of questions encompassed within climate decision making - involving different individuals, communities and organisations with plural value systems - many different climate modelling strategies are needed which have different methodological aims and do not necessarily form a simple linear “hierarchy”, but can still learn from and complement each other. We contrast the strengths and weaknesses of approaches such as GCMs, machine learning methods, EMICs, toy models, and narrative or storyline approaches as well as physics-informed models such as IAMs, ecosystem models and climate fiction. We outline some ideas for what a (more) pluralist ecosystem of climate modelling strategies would look like, and how it could more effectively answer adaptation decision questions.