This project aims at moving forward the discussion about how to improve the use of mathematical models in policymaking. It takes on from the critiques and challenges associated with integrated assessment models of global change and raised in different disciplinary communities, to then gather as many diverse disciplinary positions and rethink the construction and use of models. This initial phase of the study aims to construct a platform of interdisciplinary collaboration. Given the immense potentialities but also risks involved in the use of this mathematical modelling and given the absence of a scientific community on the matter, it is urgent to establish a coherent discussion about how models can be at the same time scientifically sound and policy-relevant. A technical fix alone would not solve the problem: mathematical modelling cannot be fixed in the sense of getting the model right, nor getting the model right is the way to affect the preferences of its users. Therefore, this project proposes to debunk the valueleaden character of models by undertaking a more ‘poetic’ approach to science: the goal is to install a kind of ‘poietic’ process into modelling, so as to extract the hopes, values and meaning that the modeller engages when facing the inescapable limitations to her knowledge in modelling the future. To make poetry out of science, mathematical models will be interpreted as ‘stories’ of the future, and accordingly analysed as narratives. The reason behind this methodological choice lies in its contribution to knowledge. Building on Hayden White’s theory on history and narrativity, narrative analysis allows revealing the consciousness of the storyteller as well as the moral of her story. This project suggests that making the moral character of mathematical models explicit can greatly improve public discussion about the meaning and the urgency of policy action.