A systematic approach to optimizing organizations: Models, design methodology, and applications.

摘要

The research presented in this thesis develops and empirically validates a methodology for modeling complex missions and for synthesizing the concomitant optimal organizations. This thesis presents a formal design methodology that integrates five models characterizing the organizational design challenges from different perspectives. Each model formalizes a different class of problems that range from strategic planning to controlling a stochastic environment to distributed mission execution to designing a structure and the associated processes that facilitate organizational interactions to assigning management responsibilities and balancing the overhead. As illustrated throughout the thesis, the resulting problems can be solved independently, as well as iteratively, to address a wide range of organizational design challenges. The organizational design procedure integrates a variety of optimization algorithms that optimize strategy selection, planning, work distribution, scheduling, resource allocation, information management, communication and coordination, and decision-making to optimize an organizational structure and processes to achieve superior performance in a specific mission environment. The methodology defines a theory and models of human decision-making in distributed systems---models that could ultimately contribute to design modifications that enhance the overall human-machine system performance. The presented design methodology combines intuitive visualization with the scientific rigor and formalism. Our iterative design process allows one not only to tackle complex design challenges, but also to iteratively improve the design in the order that reflects one's subjective judgment as to the relative importance and interdependence of different performance criteria. This thesis also presents the results of testing the proposed design process in experiments with human decision-makers. Presented findings from the experiments with human-in-the-loop organizations provide empirical validation of the modeling and design methodology.