| Traditional requirements-definition activities begin with the engineer or design team performing a needs-analysis to identify user requirements. Needs-analysis is generally subjective, and varies according to the composition and experience of the design team. Systematic procedures for defining and ranking requirements are necessary to consolidate the foundation on which the design process is predicated, and to enhance its outcome by providing the designer with a consistent, reliable approach to product development. As a first step towards developing such procedures, research was conducted at Stanford University using empirical data from a NASA spaceflight experiment that flew aboard Space Shuttle mission STS-90 (April 1998).; This research was accomplished using ex post facto data analysis. This researcher served in the central role of Experiment Manager for the spaceflight experiment, and acted as a participant-observer while conducting formal research. To better understand requirement structure and evolution, individual requirements were decomposed using AND/OR graphs. The AND/OR graph data structure illustrates requirements evolution, and reveals that the original requirement is often met by fulfilling a series of sub-requirements that are easier to implement.; Early in the product life cycle, many hundreds of potential needs were identified; however, it was a smaller subset of these initial needs that were realized in the final product. Based on analysis of a large group of requirements, it was observed that two critical components for any individual requirement were: (1) a stated need, and (2) an advocate to implement the need. Identification of need, or needs-analysis, although a necessary condition, is insufficient to ensure that a stated need evolves into a formal requirement. Equally important to the concept of requirements-definition is the notion of advocacy. Early in the product development cycle of the of the NASA experiment, many potential needs were identified; however, it was only through need-advocate pairing that a stated need became a requirement. Empirical data revealed that when needs were accompanied by strong advocates, they became clear requirements. Conversely, needs without advocates did not become requirements. Hence, need-advocate pairing is useful for predicting needs which will become requirements, and more importantly, needs at risk of not becoming requirements. |
