System Reliability: A Cold War Lesson

Abstract

AbstractDefence technologies, such as early-warning systems, are subject to exogenous and endogenous threats. The former may issue from jamming or, in a combat situation, anti-radiation missiles. The latter may issue from latent errors (Reason in Human Error. Cambridge University Press, Cambridge, 1990) introduced into the system at the initial design stage or during an upgrade, that is, through reactive patching (Weir in Debates in Risk Management. UCL Press, London, pp 114–126, 1996). It is easier to defend against exogenous than endogenous threats. Nevertheless, mindfulness when designing or upgrading a defence system reduces the risk of latent or embedded errors compromising reliability. This chapter will argue that systems that permit manual intervention, that is, manual override, are more reliable than systems that provide little or no opportunity for intervention. Referencing a Cold War near-miss, the chapter posits a negative relationship between coupling and reliability. That is, the more tightly coupled—that is, automated and linear—a system’s architecture, the less reliable it will be (other things remaining equal). It has become fashionable to characterise the human component as a liability—a latent error. The manner in which the Cold War crisis described below was resolved demonstrates the unfairness, indeed, recklessness of this characterisation.KeywordsDefenceSocio-technical systemsCouplingReliabilityHuman componentAsset