Normalization of Deviance

When repeated departures from an established standard don’t produce obvious bad outcomes, the departures can gradually become accepted as normal, and each further departure from there looks smaller by comparison than it actually is.

3 min read

What Is It?

Sociologist Diane Vaughan coined the term studying NASA’s decision-making before the 1986 Challenger disaster. Engineers had observed damage to the shuttle’s O-rings on previous flights that fell outside original design specifications. Each time the flight nonetheless succeeded, the observed damage was reanalyzed through ordinary organizational and technical review and reclassified as an “acceptable risk,” so that engineers and managers weren’t simply ignoring a standard they still believed was being violated, they had come to understand the deviation itself as falling inside the range of acceptable performance. Over multiple flights, what began as a troubling anomaly became a documented, expected occurrence, and the standard for what counted as a normal, acceptable flight had shifted without anyone deciding, in a single moment, to lower it. The mechanism doesn’t require anyone to consciously decide safety no longer matters. Each individual step looks like a reasonable extension of what was already tolerated, and the repeated absence of failure becomes evidence, treated as stronger than it actually is, that the deviation is safer than originally believed.

Why Does It Matter?

Organizations extend the same pattern to processes with far lower stakes than spaceflight, but the same underlying logic. A control gets skipped once under time pressure, nothing bad happens, and skipping it becomes routine. A review step gets shortened, quality holds up, and the shortened version becomes the standard rather than the exception. Each individual step away from the original standard looks small relative to the step before it, precisely because the standard itself has already moved to accommodate the prior step, which is what makes the drift so hard to catch from inside the organization living through it.

The pattern is dangerous specifically because it doesn’t feel like risk-taking from the inside. The people involved usually believe they’re operating within a reasonable, tested tolerance, because as far as their recent experience shows, they are, the absence of a bad outcome so far functions as confirmation rather than as the accumulating, unexamined risk it actually represents.

What Changes Once You See It?

You start asking not just “has this caused a problem” but “how far has this drifted from the original standard,” since the two questions can have very different answers, and only the second one catches the drift while it’s still small.

You stop treating repeated success under a known deviation as sufficient evidence, on its own, that the deviation is safe, and instead ask whether the original risk model has actually been revisited, or whether experience has simply replaced analysis.

You also get more deliberate about periodically comparing current practice against the original documented standard, not against last month’s practice, since comparing only to the most recent baseline is exactly the comparison that hides cumulative drift.

Common Misunderstandings

It isn’t a claim that every deviation from a standard is dangerous. Standards sometimes genuinely deserve to be updated when they were overly conservative, the risk is specifically when a standard erodes without anyone deliberately reassessing it against the original reasoning.
It isn’t the same as the Ratchet Effect, which describes standards moving in one direction and staying there once increased. Normalization of deviance is specifically about a safety or quality tolerance eroding downward through repeated exposure without consequence.
It doesn’t require negligence or a deliberate decision to cut corners. The people involved at each step are usually acting reasonably given what they’ve directly observed, which is what makes the pattern hard to catch through individual vigilance alone.
It isn’t limited to catastrophic, headline-making failures. It operates identically, at much smaller scale, in ordinary process drift inside routine business operations, long before any incident makes the drift visible.

Diagnostic Question

How far has this practice drifted from the original documented standard, not from what we did last time, and would we approve that full distance if we were deciding it fresh today?

Explore Further

Field Notes

None yet.

Related Field Guide

Origin

Diane Vaughan, The Challenger Launch Decision: Risky Technology, Culture, and Deviance at NASA (1996), analyzing the organizational decision-making that preceded the 1986 Space Shuttle Challenger disaster.