The paper by Hannah et al. (this volume) invokes foodweb theory and the ideas of complexity theory to guide the construction of models of intermediate complexity, which sacrifice explicit process detail to increase the number of interacting components of the system and simulate the web of feedback loops. This approach has its merits, if best practice modelling guidelines are followed and the method is used well. However, if this is not the case then the fundamental weakness of the intermediate model approach is that it may end up producing models that are over general and therefore not useful. To make the most of the intermediate complexity approach it is essential to keep the tenets of the middle-out approach in mind. Under the middle-out approach computational models are constructed and tested at the levels where we have the most detailed information, building on our existing knowledge and data and coupling a hierarchy of models rather than being swamped in a morass of detail or missing key feedbacks through over-generality.