The night before the Challenger launch, engineer Roger Boisjoly and his colleagues argued that the shuttle should not fly in the forecast cold.
The rubber O-rings sealing its rocket boosters had no flight or test experience anywhere near the forecast temperature. The coldest previous launch had been 53°F, and it had produced the worst seal damage on record. Morton Thiokol, the contractor responsible for the boosters, initially recommended against launching below that temperature.
NASA challenged the recommendation.
After NASA's pushback, Thiokol's Joe Kilminster requested a private management caucus. During it, Senior Vice President Jerald Mason asked engineering executive Bob Lund to "take off his engineering hat and put on his management hat." Management reversed the recommendation. The engineers' opposition continued, but it did not travel upward with the decision.
The next morning, Challenger broke apart 73 seconds after launch.
The Rogers Commission later concluded that the launch decision was flawed. The senior people who approved it had not been told about Thiokol's initial written recommendation or the engineers' continuing opposition. The Commission wrote that, had they known all the facts, it was "highly unlikely" they would have launched.
But the simple story - managers ignored engineers who had proved the danger - is too neat.
Sociologist Diane Vaughan's later analysis argued that the decision conformed to established organizational norms: what she called the normalization of deviance. The engineers could not demonstrate a conclusive relationship between temperature and seal damage that night. The chart they presented omitted the launches with no blow-by, weakening the counter-case just when it needed to withstand institutional pressure.
That complication sharpens the point.
The problem was not that nobody raised an objection. They did. Nor was it that the objection arrived as incontrovertible proof. Refutation rarely does.
The problem was that an incomplete counter-case had to compete with deadlines, precedent, hierarchy, and the institutional momentum behind launching. The challenge was voiced, but it lacked the resources, protection, and authority needed to be developed and carried through the decision process.
I call this the refutation deficit.
A refutation deficit is not simply an absence of disagreement. It is the gap between the resources we give a proposal and the resources we give the attempt to prove it wrong.
At my previous company, we chose to concentrate on the largest enterprise customers. Each reason was defensible — credibility, revenue, scale — but we never put the whole wager on one page: longer sales cycles would slow learning, demand a heavier product and a larger team, and require more capital before we had proved the product was essential. By the time we tried to move downmarket, the missing branch was visible: we had built for a buying motion whose costs our contracts could not support, without the self-serve path the alternative required.
Research suggests that this gap has both cognitive and social causes.
Paul Nutt's 1993 analysis of 168 organizational decisions found that only 29% considered more than one alternative. In exploratory interviews with 40 full-time employees across diverse industries, Frances Milliken and colleagues found that 85% had withheld concerns from a supervisor at least once. The most common reason was fear of being labeled negatively and damaging valued relationships.
And the resistance is not only organizational. Research on motivated reasoning shows that once we want a conclusion to be true, we become selective about which evidence we seek, how critically we examine it, and which explanations feel reasonable.
Refutation is therefore unusually expensive.
It costs attention because we must search beyond the evidence that supports us. It costs time because we must keep a decision open. It costs ego because our idea may not survive. And inside an organization, it can cost status, relationships, and sometimes a career.
Supporting the favored plan demonstrates commitment. Challenging it can look obstructive. The proposal has an owner. The strongest counter-case often does not.
Organizations do not necessarily lack dissent. They lack reliable machinery for turning dissent into a serious test of a decision.
This is one problem a deliberation calculator would have to solve.
It would not simply produce more objections or recommend an answer. It would give the case against a decision an institutional existence: explicit assumptions, developed alternatives, disconfirming evidence, failure conditions, and a record of which challenges were answered - or ignored.
Its product would be a tested map of the decision: what was considered, what was challenged, what evidence ruled something out, what survived, and what remains unexamined.
That would not eliminate bad judgment. Sometimes the reasoning is sound and the world changes anyway. Sometimes incentives or power override the analysis.
But now that we have AI, could we build an instrument that makes it cheaper - and safer - to discover that a decision is wrong before the decision meets reality?
References
- Presidential Commission on the Space Shuttle Challenger Accident, Chapter V: The Contributing Cause of the Accident — NASA History
- Diane Vaughan, The Challenger Launch Decision: Risky Technology, Culture, and Deviance at NASA — University of Chicago Press
- Paul C. Nutt, 1993 analysis of alternatives in 168 organizational decisions — INFORMS / Management Science
- Frances J. Milliken, Elizabeth W. Morrison, and Patricia F. Hewlin, An Exploratory Study of Employee Silence (2003) — Wiley Online Library
- James R. Detert and Amy C. Edmondson, Implicit Voice Theories: Taken-for-Granted Rules of Self-Censorship at Work (2011) — Academy of Management Journal
- Ziva Kunda, The Case for Motivated Reasoning (1990) — PubMed