In this paper the authors present their initial findings from a study into the contemporary use of modular synthesisers. The study works towards the development of an evaluative processmodel examining the strategies modular synthesiser composers use to manage algorithmic complexity. Including observations of users' compositional workflows, which provide insight into the process of building, exploring and rebuilding algorithmic music systems using modular synthesisers. We use these observations to consider what factors cause users to fix certain constraints and explore others. We argue that compositional practice with modular synthesisers provides the ideal testing ground for understanding how machine learning and other algorithmic composition techniques can be used in a co-creative context. Finally, we outline design principles for the design of creative music production tools that are inclusive of the unique requirements of algorithmic music composition systems.