The Field Guide › Paper
The construct-defining chapter for 'organizational errors' as distinct from individual errors, heavily cited across the subsequent error literature (including by Lei, Naveh and Novikov's 2016 Journal of Management integrative review, itself already in this corpus, which treats this piece as one of its foundational sources) precisely because it formalises a distinction practitioners often make intuitively but the field hadn't yet defined rigorously. An organizational error is the action of multiple participants, acting in their formal roles, that deviates from organisationally specified rules and can potentially produce adverse organisational-level outcomes, primarily caused by organisational conditions rather than any individual's idiosyncrasies; an individual error, by contrast, is one person's deviation, best explained by what's specific to that person. The central case, based on real events at a US hospital pseudonymised as MWH, makes the distinction vivid. In 2001, a single heparin overdose at an affiliated hospital (both infants survived) triggered a full corrective response: physically separating different drug concentrations, mandatory verification training, and departmental discussion. Despite this, five years later, five different nurses independently administered the identical thousand-fold heparin overdose to six infants over a single weekend, killing three. The paper's point is that this second event, unlike a hypothetical or comparable single-nurse incident elsewhere, is an organizational error precisely because it's the same deviation, committed independently by multiple people who share a work context, signalling shared underlying conditions rather than five unrelated personal failures, a judgement MWH's own CEO reached intuitively when he called the incident 'institutional' rather than blaming the nurses. The paper's most genuinely unsettling insight is offered as the likely explanation: procedural decay. The elaborate 2001 fix probably worked exactly as intended for years, and that very success removed the feedback that sustained vigilance, so verification behaviour quietly eroded without anyone noticing until the underlying risk resurfaced, echoing Reason's observation that 'it is easy to forget to fear things that rarely happen.' A successful safety intervention can, in other words, seed the conditions for the failure it was built to prevent. The chapter also develops a multi-level apparatus tracing how antecedents at the individual, unit, and organisational level combine via mediating mechanisms, of which three are named directly: learning, error-corrective feedback loops, and error-amplifying feedback loops. The 1995 collapse of Barings Bank is the canonical illustration of amplification: a junior trader was allowed to both initiate and settle his own trades, a basic separation-of-duties failure; losses were repeatedly covered rather than caught; and the errors compounded exponentially rather than staying contained, ultimately destroying a 233-year-old bank. On managing errors, the chapter lays out two competing philosophies: prevention (rules, SOPs, training, audits, contingency plans, which assume the environment is stable enough to be controlled in advance) and resilience (psychological safety, a high-reliability culture, improvisation skill, which assume error is inevitable and the goal is to catch and correct it fast), crossed against three temporal phases (before, during, after an error occurs) into a practical planning matrix. Crucially, the two philosophies aren't presented as rivals: Vogus and Sutcliffe's finding that nursing units combining care pathways (prevention) with mindful organising (resilience) had fewer medication errors and falls than units relying on either alone is cited as evidence the approaches are complementary rather than substitutes. A section on context proposes hazard profile, external regulation, and organisational strategy as explanations for why otherwise similar organisations differ in error rates and responses, and offers a genuinely striking natural experiment on regulation: when the US aviation near-miss reporting system was blame-free, voluntary reporting rose sharply; when the law changed to allow reports to be used punitively against pilots, reporting collapsed. Closes by contrasting a 2009 commercial airline crash caused by one pilot's individual control error (other pilots routinely handled the same situation correctly, so retraining that one pilot would have been the right response) against the MWH case (where only structural, system-wide change was appropriate) to make the point that correctly diagnosing which kind of error you're facing has direct, practical consequences for which kind of fix you reach for.