Normalization of Deviance (Vaughan)

Diane Vaughan's concept from the Challenger disaster: when rule violations or anomalies occur repeatedly without immediate consequence, they become accepted as normal operating procedure. NASA engineers saw O-ring erosion on multiple flights before Challenger but reclassified it from 'anomaly' to 'acceptable risk' because nothing bad had happened yet. The pattern follows a predictable sequence: (1) rule or standard exists, (2) deviation occurs, (3) no immediate consequence, (4) deviation becomes the new baseline, (5) further deviation from the new baseline, (6) catastrophe. Every major industrial disaster—Deepwater Horizon, Boeing 737 MAX, Chernobyl—shows this fingerprint. The key insight is that the people involved aren't reckless cowboys; they're rational actors updating their risk models based on observed outcomes, unaware that they're sampling from a fat-tailed distribution.

When to use it

In safety-critical operations, compliance environments, or any domain where standards exist. When you hear 'we've always done it this way and nothing bad happened.' During incident investigations. When auditing high-reliability operations.

How it can help

This is one of the most practically valuable failure models because it gives you a specific thing to look for: the gap between written procedures and actual practice. Prevention requires regular 'procedure audits' where you compare what the manual says to what people actually do—not to punish, but to identify drift. When you find gaps, you must either update the procedure (if the deviation is actually safe) or enforce the original (if it's not). The worst response is to leave the gap in place. Create a 'close call' reporting system that rewards people for flagging near-misses. Every near-miss is data about how far you've drifted; every one ignored is a step toward catastrophe.