A complexity measure for new product development projects

A novel complexity measure for new product development projects is introduced and its validity is evaluated. Therefore, a stochastic model of project dynamics based on previous work of Smith and Eppinger is developed. The model is able to represent the concurrent task processing of a large number of involved individuals who make at least partially autonomous decisions but also strongly interact. Because the model is stochastic in nature it can account for random and unpredictable performance fluctuations. This dynamic model is the mathematical foundation to compute the novel complexity measure-named the Effective Measure Complexity (EMC)-in a simple explicit form. The complexity theory behind the novel measure goes back to seminal work of the theoretical physicist Grassberger. His forecast complexity is an information-theoretic concept that was derived from first principles and is highly satisfactory in many respects. In the validation study a NPD project on sensor design in a small-sized company is considered and the external validity of the stochastic project model as well as the structural validity of the novel measure are analyzed. Due to the labor time system of the company, very accurate and fine-grained data about the task processing were acquired. If stopping criteria are used the results of the external validation of the dynamic model show a sufficient goodness-of-fit between the real and simulated project. The structural validation of EMC demonstrates that small complexity values are assigned to projects with uncoupled tasks, as it is intuitively required. The stronger the task couplings are, the more the complexity values grow until the stability bound of the project is reached. Beyond the stability bound the work remaining grows over all limits and the project diverges. This divergent behavior is indicated by infinite complexity values.