1.1The question
Every organisation that runs a hazard keeps two kinds of record. The first says what happened: alarms, breaches, losses, near things. The second says what is allowed: thresholds, limits, procedures. Between them sits a third thing that is rarely written at all, the belief of the people in the room about what the first record means for the second. When a pressure alarm has sounded many times and nothing has followed, the operators come to believe that the next alarm will be followed by nothing as well. When a trader has broken a leverage limit and made money, the desk comes to believe that the limit was set too low. Nobody has to decide to believe either thing. It arrives.
This article asks whether that arrival is what David Hume, in Sections IV and V of An Enquiry Concerning Human Understanding (1748), called custom, and whether two decision records written for the RISCON platform set down what would break it. The question has two halves. The first is a question about ideas: does Hume's account of how the mind passes from past to future describe the mechanism that organisational sociologists later named the normalisation of deviance? The second is a question about documents: when a decision record is written before its outcome is known, does it say what observation would show that the quiet has ended?
1.2The contribution
The contribution is narrow. The organisational literature on accidents is divided between those who think the drift towards accepting warnings is built into complex organisations and those who think some organisations can hold it off. Both camps argue from cases: launches, nuclear alerts, carrier decks, air traffic. Both assume something about what a long run of uneventful experience licenses, and neither says what. This article places Hume's argument underneath the debate, as the claim both sides presuppose. Section IV of the Enquiry shows that a run of uneventful experience gives no reason, considered as reasoning, to expect the run to continue. Section V says that the mind expects it to continue anyway, by custom, and that custom is not something reasoning can switch off. If that is right, the two camps disagree less about the mind than about institutions: about whether an organisation can build practices that keep alive what Hume called "any suspicion that the course of nature may change" when custom is pressing the other way.
The second, smaller contribution is a reading of two documents. WSTiger, a studio in the RISCON family that connects economic history and market signals to the question of what risk to take now, publishes synthetic decision records for study. Two of them, belonging to a thread the studio calls "Repeated Violations Become the Rule", set the problem of the quiet warning in a 1984 chemical plant and in a 2026 trading desk. Both records are modern, synthetic and written for the platform. They are not archival documents and report no real plant, desk or event. What they offer is the decision written before its outcome, which is the moment Hume's argument concerns. This article reads them with the Enquiry open beside them and asks what they write about the end of the calm.
1.3The debate joined
The sides are named in the next section and set out at length. In brief: Barry Turner (1978), Charles Perrow (1984), Diane Vaughan (1996) and Scott Sagan (1993) stand on one side, holding in different ways that organisations produce their own accidents and that a long quiet is part of how they do it. Todd La Porte and Paula Consolini (1991) and Karl Weick (2001) stand on the other. La Porte and Consolini observe that some organisations run hazardous work with very few failures; Weick describes what high-reliability organisations do: they attend to failure, refuse to simplify, stay close to operations, keep the capacity to recover, and defer to expertise where it is. Sagan tested the two against each other and found his evidence closer to the first. This article adds Hume to the bottom of that exchange and asks what each side must say about custom.