Type:

Year:

Author:

Supervisor:

Institution:

Page(s):

Website:

Abstract:

Strategy dynamics are hypothesized to be a structural factor of interactive multi-actor design problems that influence collective performance. This dissertation presents a conceptual model of collective decision-making processes in engineering systems design to understand the trade-offs and risks of strategic interaction between autonomous design actors. The approach combines value-driven design, game theory, and experimentation with cognizant and computational agents to study how technical, organizational, and social factors affect collective action in a design process. Using a bi-level decision-making model, lower-level operational decisions are mapped into upper-level binary strategy sets to identify four strategy dynamics (harmony, coexistence, bistability, and defection) characterized by low and high levels of two factors, structural fear and structural greed. A controlled two-actor human-subject study using synthetic bi-level parameter design tasks inquires the effect of both factors on collective efficiency and equality, with results pointing at greed as the most impactful of the two. A generalization of fear and greed to more than two actors and a measure of individual fear–greed resistance (the harmony index) assesses the compatibility between individual strategy profiles to achieve desired collective outcomes. Agent-based simulation of abstract and context-specific design tasks delves into the ways in which organizational-level actions mitigate or intensify unfavorable strategy dynamics. A simulation case study on the adoption of peer-to-peer and centralized data-sharing paradigms in a multi-agent surrogate aircraft design problem shows how arbitrated allocation of costs equating the actors’ harmony indices paves the roads toward more stable and efficient collective strategy results in comparison to surplus division à la Shapley value. Results of this study provide an initial general description of strategy dynamics in collective systems design and help frame future work in the structuring of incentives through social constructs or mechanism design to drive collective action towards better design solutions.

Keywords: