The shortcut that worked ninety nine times

Normalisation of deviance is the mechanism by which a workplace drifts without noticing.

Somebody skips a step because it is inconvenient and nothing bad happens. They do it again, and again nothing happens. Others see it, nothing happens to them either. The step becomes optional, then unusual, then forgotten. Eventually a new starter is taught the shortened method as the method, and the original step exists only in a document nobody reads. At no point did anyone decide to lower the standard. Each individual instance was reasonable given the evidence available, and the evidence available was a long run of uneventful outcomes. That is the trap: absence of harm reads as proof of safety.

What makes this so difficult to counter is that the drift is invisible from inside. There is no moment of transgression to catch, only a gradual redefinition of normal, endorsed by experience. The people involved are not reckless, they are inductively reasonable. The step being skipped was usually one whose purpose was never explained, or one that protects against a condition that occurs rarely, so its value is not observable in daily work. A control that only matters in unusual circumstances will always look like waste in ordinary ones.

The counters are modest but real. Explaining why a step exists rather than only that it exists gives it a defence against erosion. Noticing the gap between how a task is described and how it is actually performed, and treating that gap as a question rather than a violation, catches drift while it is still cheap to address. And treating a near miss as the valuable signal it is, rather than as a non event, interrupts the run of uneventful outcomes that is otherwise the only evidence anyone has.