Contribution
An official accident investigation named organizational history, not a technical fault or individual error, as a cause of the accident, and showed the same decision pattern recurring seventeen years apart.
Key insight
The danger signals were seen and processed, and the processing made them normal. Each flight that survived an anomaly became evidence that the anomaly was safe, so the evidential standard moved with the experience.
Assumptions
- That the two decision histories can be compared as instances of one organizational pattern.
- That culture, structure and political environment are causes on a par with the physical mechanism.
- That the Board’s reconstruction from documents and testimony captures how decisions looked at the time.
Key results
- Identical trajectories (pp. 195–196). For O-rings and foam, a first decision to accept rather than eliminate the deviation set a precedent. Anomalies that did not cause catastrophe were treated as engineering data justifying further flight, and were classed “in-family”, within the experience base.
- Mixed and weak signals (pp. 196–197). Danger signals were followed by all-clear signals, and foam strikes became routine, not signals of danger.
- System effects (pp. 197–199). Budget limits imposed by political sponsors, the shift to a business model with schedule and cost goals raised to the level of safety, and leaders’ “doublespeak” made efficiency a strong signal and safety a weak one.
- Structure (pp. 199–201). Flight-readiness review by consensus “renders no one accountable”. Safety staff depended on the program they supervised. Engineers had to prove a flight unsafe, not safe. Dissent did not surface at higher levels.
- Persistence (pp. 198, 202). Post-Challenger changes “were undone over time by management actions”; flawed practices “continued for 20 years”.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Engineering decisions on foam and O-rings normalized the anomalies in the same way | strong | two documented decision histories, the second investigated by the Board |
| C2 | The causes of Challenger’s institutional failure had not been fixed by 2003 | moderate | comparison of the two histories; reforms shown reversed |
| C3 | Political and budget constraints, production pressure and structure sustained the pattern | moderate | the Board’s reconstruction; partly drawn from Vaughan’s published work |
| C4 | Ill-structured problems invite normalization and may be the most risky | weak | stated as a lesson, not shown |
Method
Comparative case history: the Board’s investigation of Columbia set against the Presidential Commission’s record and Vaughan’s study of Challenger, reconstructed from documents and testimony. The chapter interprets; its evidence is in the rest of the report and in the cited works.
Concepts
- normalization of deviance — operationally, reinterpreting a departure from design expectations as acceptable and non-deviant. The chapter gives no formal definition.
- in-family — “a formal classification indicating that new occurrences of an anomaly were within the engineering experience base” (p. 196).
- failure of foresight — after Turner, an accident with “a long incubation period” in which warnings were ignored or misread (n. 1).
- bureaucratic accountability — “an allegiance to hierarchy, procedure, and following the chain of command” (p. 199).
Connections
The chapter applies Vaughan’s The Challenger Launch Decision (LIT-715), which it cites for its central claim, and Turner’s failures of foresight. Rasmussen’s account of drift (THEORY-144) and Argyris’s defensive routines (THEORY-145) describe neighbouring mechanisms; see THEORY-140.
Bearing on the record
Source of THEORY-140.
For the owner’s essay: §5.4 (protective lock-in) has here a documented case of a regime that reproduces itself, since each survived anomaly becomes the justification for the next flight. §6.3’s extraction from above has a case in sponsors who accepted ambitious proposals and then imposed strict budgets, though the Board describes constraint, not extraction. §8.4 (externalized governance) has the Board’s remedy: a safety organization with independent authority and resources, because one dependent on the program could not correct it. The Board’s observation that “NASA has shown very little understanding of the inner workings of its own organization” (p. 202) bears on §8’s question whether an organization can govern a will it does not know.
Limitations
Two cases in one agency. It is an official finding, with the authority and the incentives of one. The chapter is unsigned, and Vaughan’s role in drafting it is not established by the report. The general theory of normalization is in her book, which the record has not read. The reconstruction is retrospective, which the chapter itself warns can distort how decisions looked at the time.
Open questions
Whether normalization has a signature that can be detected before an accident, rather than reconstructed after one; and whether an independent safety function stays independent once the crisis that created it has passed, which endnote 49 doubts.