Mathematical modelling is the dominant linguistic instrument in contemporary economics, and the canon with which to assess the scientific character of theories; but it has forever lost the pretence of truth. In current research practice in the discipline, mathematical models are conceived as born from the economist’s imagination, and flexibly adaptable for reasons of interpretation or technical convenience. A model is the formalization of some fictional world. The criteria used to associate models with the explanation of economic events are loose and controversial. No shared, structured canon exists to validate or reject the significance of a model, and even the criterion of empirical validation has lost appeal. New currents in methodology reflect the practices in advanced research, suggesting that models in economic theory should be conceived as fables, fictions, or thought experiments on fictional cases. Academic communities validate models according to their presumed technical novelty, or because of fashion, ideology, and power. Sophistication in mathematical language is the primary canon conventionally adopted to validate models as acceptable within the mainstream. The paper questions the paradoxical coexistence in economics of the ambitious pretensions to hard science, and the ideas on mathematical models as fictional narratives about economic Wonderlands, which dominate in research practice and are reflected in studies on methodology.