MODEL VALIDATION IN SOFT SYSTEMS PRACTICE

The concept of 'a model' usually evokes the connotation 'model of part of the real world'. That is an almost automatic response. It makes sense especially in relation to the way the concept has been developed and used in natural science. Classical operational research (OR), with its scientific aspirations, and systems engineering, use the concept in the same way and in addition use models as surrogates for the real world, on which experimentation is cheap. In these fields the key feature of a model is representativeness. In soft systems methodology (SSM) models are not of part of the world; they are only relevant to debate about the real world and are used in a cyclic learning process. The paper shows how the different concepts of validation in classical OR and SSM lead to a way of sharply defining the nature of 'soft OR'.