Is the Normal Accidents Perspective Falsifiable?

Antti Silvast, Ilan Kelman

Research output: Contribution to journalArticlepeer-review


Purpose – This article is motivated by debates regarding Charles Perrow’s Normal Accidents perspective which describes how technological systems are prone to failure if they have complexity and tight coupling. The purpose of this paper is to explore Normal Accidents conceptually to understand whether or not it might be feasible to disprove it or to find counterexamples.

Design/methodology/approach – The approach taken by this article is to identify and explore assumptions inherent in Normal Accidents which might make the perspective non-falsifiable.

Findings – The findings and discussion cover two principal assumptions inherent in Normal Accidents. First, no past record of the absence of a Normal Accident excludes the possibility of a future Normal Accident. Second, analysis of a Normal Accident is always relative to the selected definition of the system, but a system can potentially be defined so that there was or was not a Normal Accident.

Practical implications – Although the Normal Accidents perspective does not appear to be falsifiable, the perspective should still be taught and considered when designing and operating technological systems. The reason is that, even if Normal Accidents is a truism, it is not accepted as such, meaning that society is setting itself up for continual catastrophic failures of technology.

Originality/value – An exploration is provided of digging beneath the “Normal Accidents” perspective and its discussants to explore why the perspective is so important yet so rarely implemented in practice.
Original languageEnglish
Pages (from-to)7-16
JournalDisaster Prevention and Management
Issue number1
Publication statusPublished - 2013


  • Complexity
  • Coupling
  • Normal accidents
  • System accidents
  • Technology
  • Systems analysis
  • Disasters

Fingerprint Dive into the research topics of 'Is the Normal Accidents Perspective Falsifiable?'. Together they form a unique fingerprint.

Cite this