Retrenching the Purse: The Balance Enquiry Quandary, and Generalised and (1, 1) Forward Refinements

We argue that formal modeling should be the starting point for any serious development of computer systems. This claim poses a challenge for modeling: at first it must cope with the constraints and scale of serious developments. Only then it is a suitable starting point. We present three techniques, refinement, decomposition, and instantiation, that we consider indispensable for modeling large and complex systems. The vehicle of our presentation is Event-B, but the techniques themselves do not depend on it.