2021 article

Toward Quantifiable Evidence of Excess' Value Using Personal Gaming Desktops

Long, D., Morkos, B., & Ferguson, S. (2021, March 1). JOURNAL OF MECHANICAL DESIGN, Vol. 143.

co-author countries: United States of America 🇺🇸
author keywords: excess; value; requirements modeling; uncertainty; technology forecasting; design evaluation; systems design
Source: Web Of Science
Added: March 22, 2021

Abstract Complex systems may operate in scenarios where the current requirements were “unknown” at the time of their original design. Such “unknown” requirements might be outside the probability distribution expected during the design phase or, more drastically, might not have been predicted. Yet, not meeting these “unknown” requirements can significantly reduce system value. Engineering design researchers have begun addressing this challenge by exploring how incorporating margins when the system is being designed—a form of strategic inefficiency—might increase a system’s total lifetime value by reducing sensitivity to requirement changes and truncating change propagation. Quantitatively studying excess margin beyond what is required for known uncertainties has been particularly challenging as information is needed about how requirements change, how system performance is impacted by requirement changes, and how components are priced. A quantitative study around excess using 20 years of data for desktop computers, video game consoles, and video games is developed. Evidence is provided that excess can improve end-user system value when future requirements are unknown. This paper also advances the notion of strategic excess (excess incorporated in a single component), showing as one example that excess RAM would have improved system performance by 14% (on average) for 7% of total system cost. In demonstrating the value of excess, we strengthen the argument that engineers (and end-users) should embrace strategic inefficiencies—even though they might never be used—and further study the implications of system architecture and module interfaces decisions.