A mathematical understanding of the success of multidisciplinary development teams

Today a product development problem has all the characteristics of a complex multi-criteria optimisation problem, with constraints that are often interdependent and sometimes conflicting. Until quite recently, a functionally organised development organisation would take 3-7 years to solve such a problem. Each step in the development process is doing a sub-optimisation over a limited set of constraints and criteria. Thus many iterations were necessary to find a solution that satisfied all constraints and criteria. A “modern” multi-disciplinary, project-oriented development team can do the job much quicker. Depending on the type of product, development lead-times have come down to less than a year. In this paper an attempt is made to explain this phenomenon by showing the similarity between a traditional development organisation and “conventional” optimisation algorithms such as linear programming. In a multi-disciplinary development team the interaction between criteria and constraints is more immediate. This resembles the behaviour of optimisation algorithms such as genetic algorithms. Genetic algorithms applied as a search technique are known to find good solutions in a limited amount of time. However these algorithms cannot be guaranteed to find the optimum