O-Ring Erosion
O-ring erosion was the technical name for the condition Vaughan's framework made structurally famous. The O-rings — rubber gaskets roughly a quarter-inch in diameter — were designed to maintain a perfect seal between segments of the solid rocket booster, preventing 5,000-degree combustion gases from escaping. The design specification was zero erosion. Beginning with the second shuttle flight, engineers observed that hot gases had eroded the primary O-rings in the booster joints. The erosion was reclassified across subsequent flights from design violation to acceptable operating condition. This reclassification — supported by engineering analysis, documented in official reports, and consistent with the accumulated record of successful flights despite the erosion — became the four-phase mechanism that Vaughan formalized as normalization of deviance.
In the AI Story
The erosion was never secret. It was documented in engineering reports, discussed in flight readiness reviews, and assessed through standard organizational processes. Vaughan's central finding was that the erosion's organizational meaning changed across the twenty-four pre-Challenger flights even as its technical form remained constant. Early erosion prompted investigation and active assessment; later erosion prompted documentation within established parameters.
The reclassification was not arbitrary. Engineers constructed technical justifications based on analyses of how the erosion occurred, how much was considered safe given the secondary O-ring's backup function, and how the observed erosion compared to the calculated safety margin. The justifications were internally coherent within a framework that had, across years of successful flights, expanded to accommodate the anomaly.
The structural parallel to AI-augmented work is the progressive reclassification of AI-generated output from unfamiliar artifact requiring comprehensive review to routine work product requiring selective verification. The reclassification is supported by track record, documented in institutional practice, and consistent with the competitive environment. The mechanism is identical; the material is different.
Vaughan's reconstruction demonstrated that the reclassification was invisible to participants because they experienced themselves as maintaining standards. The standards had been revised, not abandoned. The engineers who approved the January 28 launch were applying standards consistent with the framework their organization had developed over years of flights, and the framework — not the individuals applying it — had drifted into the territory where the cold-weather conditions fell just inside the expanded envelope.
Origin
The phenomenon was first documented on STS-2 (November 1981), the second Space Shuttle flight. Across the subsequent twenty-three flights through STS-51L (the Challenger mission), erosion was observed on multiple occasions, documented in institutional reports, and progressively reclassified from anomaly requiring resolution to known and managed risk.
Key Ideas
Original spec was zero. The design specification required zero erosion; any erosion was, by original standard, a failure.
Progressive reclassification. Each observation of erosion was assessed against accumulated experience rather than original specification.
Technical justification grew. Engineers constructed coherent analyses supporting the expanded envelope, not as rationalization but as genuine engineering reasoning within the evolving framework.
Invisible reframing. The reclassification was invisible to participants because it occurred through the accumulation of decisions, not through any single reframing event.
Template for AI normalization. The same mechanism operates when AI-generated output is reclassified from artifact requiring comprehensive review to routine work requiring selective verification.
Appears in the Orange Pill Cycle
Further reading
- Diane Vaughan, The Challenger Launch Decision (1996), especially chapters 4–7 on the O-ring decision history
- The Rogers Commission Report (1986), Volume 1, Chapter 4
- Roger Boisjoly, "The Challenger Disaster: Moral Responsibility and the Working Engineer" (1987)