Dead Reckoning
Dead Reckoning: Air Traffic Control, System Effects, and Organizational Change documents the way air traffic controllers construct and maintain a dynamic cognitive model of the airspace — a three-dimensional, temporally evolving representation of where every aircraft is and will be — that extends beyond what the instruments display. The book traces how increasing automation, while improving the informational basis of control work, simultaneously reduced the occasions on which controllers needed to build the cognitive model independently, producing a thinning of situational awareness that was invisible under normal conditions but critical when automated systems failed.
In the AI Story
The title invokes the navigational practice of estimating position from known starting point, heading, and speed — a form of cognitive reckoning that does not depend on external instruments. Vaughan extends the term to describe the independent cognitive model that experienced practitioners maintain as a check on automated systems, and that atrophies when the occasions for building it are automated away.
The book's central finding parallels Vaughan's earlier work on normalization of deviance: automation produces a specific kind of drift in which the capacity to catch failures the automated system misses is eroded by the very reliability of the automated system under normal conditions. The most experienced controllers, Vaughan found, could articulate what was being lost because they had experienced the pre-automation practice that built the capacity; newer controllers had no experiential basis for comparison.
The relevance to AI-augmented knowledge work is direct. The mechanism by which air traffic controllers lose dead reckoning through reduced occasions to practice it is structurally identical to the mechanism by which software engineers, attorneys, and physicians lose comprehension of the systems they nominally operate when AI tools remove the friction-rich engagement through which comprehension is built.
The book extends the concept of structural secrecy by demonstrating how even well-designed information systems cannot fully transmit the embodied, practice-built knowledge that experienced practitioners rely on. The dead reckoning exists in the practitioner's nervous system; it cannot be extracted, documented, or replicated by any amount of data capture.
Origin
Vaughan began the project after the Challenger book as an extension of her work on complex technical systems. The research involved years of embedded observation at air traffic control facilities, interviews with controllers at every level of experience, and analysis of incidents in which automation and human judgment interacted under stress.
Key Ideas
Cognitive modeling beyond instruments. Experienced controllers maintain an independent representation of the airspace that extends beyond what any display captures.
Automation thins the model. Each task the automation handles removes an occasion on which the independent model would have been built or maintained.
Generational loss. Controllers trained in the automated environment cannot miss what they never had; the baseline is the post-automation environment.
Failure modes invisible under normal conditions. The thinning of dead reckoning is undetectable until the automated system fails and the independent assessment is required.
Practice builds embodied knowledge. The friction-rich engagement that produces comprehension cannot be replaced by more precise information.
Appears in the Orange Pill Cycle
Further reading
- Diane Vaughan, Dead Reckoning: Air Traffic Control, System Effects, and Organizational Change (University of Chicago Press, 2024)
- Lisanne Bainbridge, "Ironies of Automation" (Automatica, 1983)
- Nadine Sarter and David Woods, "How in the World Did We Ever Get into That Mode?" (1995)