An evolutionary approach for project organization design: Producing human-competitive results using genetic programming

In the complex and rapidly changing business environment of the early 21st century, designing an effective and optimized organization for a major project is a daunting challenge. Project managers have to rely on their experience and/or trial and error to come up with organizational designs that fit their particular projects. The Virtual Design Team (VDT) simulation system, based on the information processing perspective on organizations, was a successful attempt to develop an analysis tool for project organization design (Jin and Levitt, 1996). However, VDT is an analysis tool that has no inherent ability to improve or optimize current designs automatically. A VDT user must explore different design alternatives in "What if?" mode to find better solutions.; Our research extends the capabilities of VDT and other similar organizational analysis tools by going beyond computer support for "What-if?" analysis to automated design of project organizations. Evolutionary computing methods such as genetic programming are used to design and develop a postprocessor for VDT to help project managers find near-optimal designs for their project organizations.; This dissertation describes in detail the approach I developed to represent project organization design alternatives in a genetic programming format, so that the design can effectively evolve. It also presents some of the promising human-competitive results that the postprocessor has been able to generate for two real-world project organizations. In addition, I demonstrate how my approach can create a new kind of cooperative Human-Computer Interaction (HCI) environment that can motivate humans to think "outside the box" when designing project organizations.; Using a combination of "intellective" (theorem proving) and "emulation" (natural, empirical) experiments, I validate the postprocessor's "near-optimal" solutions against findings of organizational contingency theory and human-derived solutions for a set of real test cases. By showing that "optimal" structure depends on the relative emphasis of time, cost and process quality outcome metrics, I extend contingency theory to develop a richer "micro-contingency theory" for project organizations.; Finally, the thesis concludes with a summary of the contributions of this research in the three areas of organization science, project management, and computer science.