The paper addresses some misconceptions of Software Engineering, requirements analysis and modelling in particular, due to underlying epistemological flaws., e.g. the believe that the system analyst's task be simlar to that of a natural scientist's. The fundamental issues, constitution of objects and signs, conceptualization and definability, are discussed. It comes out that the paradoxical situation of software engineering is having to formalize what cannot be formalized. This is reflected in the fuzzy notion of 'model' in general as well as in the epistemological presumptions of 'object oriented modelling' in particular. The paradigm of 'objective modelling' has to be replaced by a paradigm of 'purposive description' shifting the focus of Software Engineering research to non-formal methodologies.