Understanding Design Decisions Under Competition Using Games With Information Acquisition and a Behavioral Experiment

Abstract

The primary motivation in this paper is to understand decision-making in design under competition from both prescriptive and descriptive perspectives. Engineering design is often carried out under competition from other designers or firms, where each competitor invests effort with the hope of getting a contract, attracting customers, or winning a prize. One such scenario of design under competition is crowdsourcing where designers compete for monetary prizes. Within existing literature, such competitive scenarios have been studied using models from contest theory, which are based on assumptions of rationality and equilibrium. Although these models are general enough for different types of contests, they do not address the unique characteristics of design decision-making, e.g., strategies related to the design process, the sequential nature of design decisions, the evolution of strategies, and heterogeneity among designers. In this paper, we address these gaps by developing an analytical model for design under competition, and using it in conjunction with a behavioral experiment to gain insights about how individuals actually make decisions in such scenarios. The contributions of the paper are two-fold. First, a game-theoretic model is presented for sequential design decisions considering the decisions made by other players. Second, an approach for synergistic integration of analytical models with data from behavioral experiments is presented. The proposed approach provides insights such as shift in participants' strategies from exploration to exploitation as they acquire more information, and how they develop beliefs about the quality of their opponents' solutions.

Understanding the Effect of Task Complexity and Problem-Solving Skills on the Design Performance of Agents in Systems Engineering

“…. Following the experimental results in [10] and [14], we assume that a "skillful" agent selects X i+1 by maximizing the expected improvement in the attribute function. Suppose that the agent made a hypothetical query at x ∈ X and they observed the attribute value a ∈ R. The improvement they would have gotten over the observed attributes A 1:i is…”

How Do Designers Choose Among Multiple Noisy Information Sources in Engineering Design Optimization? An Experimental Study

“…Prior work on human strategies in design has compared human subjects' decisions against proposed rational strategies based on the expected utility theory and game theory [14,15]. Studies have estimated human strategies in terms of the parameters of Bayesian optimization (BO).…”

“…Therefore, the optimal strategy is to only select cheap simulations, and perform one expensive prototype at the end when the EI is small. For the decision to stop, the optimal strategy is to stop when the value of information from an additional evaluation is negative for either of the information sources [4,14].…”

Wisdom of Microcrowds in Evaluating Solutions to Esoteric Engineering Problems

“…Example methods include the use of explicitly verifiable questions to identify malicious users and to encourage honest responses, and task fingerprinting to monitor completion time, mouse movements, key presses, and scroll movements, which can all be used as indicator attributes for detecting suspect responses [20]. Effect of incentives and competiton in crowdsourcing data quality has been investigated in [21]. The presented work uses response speed as one such indicator to vet data quality and monetary compensation as incentive.…”