The Field Guide › Paper
The founding synthesis of the 'new view' of human error, built from fourteen premises that reorient how error should be studied: errors are heterogeneous rather than a single homogeneous category; an erroneous action is a starting point for investigation, not an ending point; and, critically, error is a symptom of deeper system issues, not a cause in itself. Distinguishes outcome failures from process defects and insists there is only a loose coupling between the two, so judging a decision by how it turned out is a category error, not an insight. Introduces the sharp end/blunt end framing that has since become standard vocabulary: practitioners at the sharp end act directly on the process, while the blunt end, organisational, regulatory, and technology-design decisions made elsewhere and earlier, shapes what options and information sharp-end practitioners actually have. The Local Rationality chapter is the clearest statement of the principle used across the field since: building on Simon's bounded rationality, it argues that people's decisions are locally sensible given their knowledge, attention, and the resource constraints they're actually operating under at the time, and that judging those decisions against an ideal, unconstrained rationality misdiagnoses the problem every time. The later chapters ground hindsight and outcome bias in the experimental literature (Fischhoff's original hindsight-bias studies; Baron and Hershey on outcome bias in judging identical decisions differently depending on how they turned out), and close with a case against treating 'human error' as an adequate stopping point for any investigation. This 1994 CSERIAC report is the first edition; a substantially expanded second edition (Woods, Dekker, Cook, Johannesen and Sarter, 2010) added Sidney Dekker as co-author and remains the version most commonly cited today, but the core arguments, local rationality, sharp end/blunt end, the case against hindsight-driven judgement, are already fully formed here.