Expert opinion is a common resource for identifying potential failure scenarios, but experts can miss novel failure combinations that have no historical precedent. A variety of computational techniques offer their own strengths for failure scenario identification, but can be limited by historical data availability or prescribed models. This paper examines the question of whether a distributed group of non-expert humans can outperform a brute force algorithm in a failure scenario prediction task. This approach uses human intuition to guide solution space exploration, and provides feedback through a system simulation. The results of the paper show that human non-experts outperformed a Monte Carlo simulation, but converge to relatively few critical failure scenarios. These results indicate that while non-expert reasoning may not be directly applicable to effective exploration of many possible failure modes, human computation has the potential to augment or compete with stochastic algorithms in a complex systems failure analysis context.

This content is only available via PDF.
You do not currently have access to this content.