Critical fragility in sociotechnical systems

J José Moran (Macrocosm Inc) F Frank P. Pijpers (Statistics Netherlands) U Utz Weitzel (School of Business and Economics) J Jean-Philippe Bouchaud (Capital Fund Management) D Debabrata Panja (Department of Information and Computing Sciences)

Abstract

Sociotechnical systems, where technological and human elements interact in a goal-oriented manner, provide important functional support to our societies. Here, we draw attention to the underappreciated concept of timeliness—i.e., system elements being available at the right place at the right time—that has been ubiquitously and integrally adopted as a quality standard in the modus operandi of sociotechnical systems. We point out that a variety of incentives, often reinforced by competitive pressures, prompt system operators to myopically optimize for efficiencies, running the risk of inadvertently taking timeliness to the limit of its operational performance, correspondingly making the system critically fragile to perturbations by pushing the entire system toward the proverbial “edge of a cliff.” Invoking a stylized model for operational delays, we identify the limiting operational performance of timeliness, as a true critical point, where the smallest of perturbations can lead to a systemic collapse. Specifically for firm-to-firm production networks, we suggest that the proximity to critical fragility is an important ingredient for understanding the fundamental “excess volatility puzzle” in economics. Further, in generality for optimizing sociotechnical systems, we propose that critical fragility is a crucial aspect in managing the trade-off between efficiency and robustness.

Article Details

Volume / Issue Vol. 122, Issue 9
Published March 04, 2025
ISSN 0027-8424
Publisher National Academy of Sciences

Authors (5)

J

José Moran

Macrocosm Inc

F

Frank P. Pijpers

Statistics Netherlands

U

Utz Weitzel

School of Business and Economics

J

Jean-Philippe Bouchaud

Capital Fund Management

D

Debabrata Panja

Department of Information and Computing Sciences