The practical modelling of context-dependent causal processes--a recasting of Robert Rosen's thought.

This paper is a critical recasting of some of Robert Rosen's thought. It is argued that a lot of the thrust of Rosen's work can be better understood when recast in terms of the context dependency of causal models. When recast in this way, I seek to highlight how his thought does not lead to the abandonment of formal modelling and a descent into relativism, but a more careful and rigours science of complex systems. This also sheds light on several aspects of modelling, including the need for multiple models, the nature of modelling noise, and why adaptive systems cause particular problems to modellers. In this way, I hope to decrease researchers fear that, by taking Rosen's criticisms seriously, they would have to abandon the realm of acceptable science.