The Field Guide › Paper
On the night of 6 July 1988 the offshore platform Piper Alpha caught fire in the North Sea and 167 people died, in what remains the worst disaster in the history of the offshore oil industry. Paté-Cornell's postmortem is the definitive organisational analysis, and its opening claim sets the tone: the fire was not an unpredictable act of God but an accumulation of errors and questionable decisions, most of them rooted in the organisation, in its structure, its procedures and its culture. Working backwards from the technical accident scenario through the human decisions that produced it and on to the organisational conditions that produced those, she identifies causes that generalise well beyond oil rigs: design practices with tight physical couplings and insufficient redundancy; misguided priorities in the trade-off between productivity and safety; failures in the management of personnel on board; and errors in how financial pressure was transmitted through the company's structure of profit centres, which degraded inspection and maintenance at the sharp end. Two threads matter especially for this map. The first is the permit-to-work system, the paper procedure meant to guarantee that a machine taken out of service is not restarted: it failed at handover, the incoming night shift did not know that a pump's pressure safety valve had been removed, and they started the pump. The second is more damning still. As Piper Alpha burned, the neighbouring platforms went on pumping oil and gas into it, feeding the fire, because the men in charge did not believe they had the authority to shut down production without permission from onshore management. The gradient of authority, the very thing psychological safety exists to flatten, kept a fire supplied with fuel while people died. For a corpus about speaking up this is the starkest case in the literature: it is not that nobody knew, but that knowing was not enough, because the organisation had not made it safe or even permissible to act. Its limits are those of a single retrospective case analysis conducted through a probabilistic risk-analysis frame, so its causal chains are reconstructed after the fact and its organisational claims are argued from the inquiry record rather than measured. (Text drawn from the 1993 Risk Analysis paper, 13(2), pp. 215-232.)