Disaster 01 · Engineering disasters
The launch the engineers asked to stop
The space shuttle Challenger broke apart 73 seconds after launch because a rubber seal failed in the cold, after engineers had argued the night before that it was too cold to fly.
- NASA and Morton Thiokol
- Kennedy Space Center, Florida, United States
- 1986
- 3 min read
On this page
Before you readWhat would you do?
Your team is about to launch a new website for a big sale. A developer says the payment page sometimes fails under heavy load. "It failed twice in testing but recovered both times." The boss asks if anyone can prove it will fail tomorrow. What is the better question to ask?
Pick an answer now. The file tells you whether you were right.
01The file in one minute
1A seal that failed in the cold
O-rings in a booster joint stiffened at around 2°C and let hot gas escape. Challenger broke apart after 73 seconds.
2Engineers said no
The night before, Thiokol engineers advised against launching below about 12°C. Their managers reversed that advice.
3Damage had become normal
Seal damage on earlier safe flights had come to feel acceptable, a pattern later called normalisation of deviance.
4The lesson
With high stakes, ask for proof it is safe to go, not proof that it is dangerous.
02Timeline
1985
Thiokol engineers warn in memos about O-ring erosion on earlier flights.
27 January 1986
Engineers recommend against launch in the cold. Managers reverse the recommendation.
28 January 1986
Challenger breaks apart 73 seconds after launch. All seven crew die.
June 1986
The Rogers Commission reports on the seal failure and the flawed decision process.
03What happened
On 28 January 1986 the space shuttle Challenger lifted off on a freezing morning in Florida. Seventy-three seconds later it broke apart. All seven crew died, including Christa McAuliffe, a schoolteacher chosen to be the first teacher in space.
The cause was a joint in the right solid rocket booster. Rubber rings called O-rings were meant to seal it, and in the cold they stiffened and failed to seal in time. Hot gas escaped and burned through. The air temperature at launch was around 2°C (36°F), far colder than any earlier shuttle launch.
The night before, engineers at Morton Thiokol, the company that built the boosters, held a call with NASA and recommended not launching below about 12°C (53°F), the coldest launch so far. NASA managers pushed back hard on the data. Thiokol's managers then met privately, and one told the engineering head to take off his engineering hat and put on his management hat. The company reversed its advice and approved the launch.
The presidential commission led by William Rogers reported in June 1986. It found that O-ring damage had been seen on earlier flights and that the warnings had not reached the people making the final decision. The physicist Richard Feynman, a member of the commission, showed on live television how a piece of O-ring material lost its springiness in iced water.
04Why it happened
The sociologist Diane Vaughan later called the pattern the normalisation of deviance. Each earlier flight with some seal damage that came back safely made the damage feel acceptable. A warning sign turned into a known, tolerated quirk.
The decision on the night was also framed backwards. Instead of the managers needing proof that it was safe to fly, the engineers were asked to prove it was unsafe. With patchy data and schedule pressure, that burden was impossible to meet.
The idea behind this caseLesson: Review AI like a manager →
05Lessons for decisions
- Treat a problem that has happened before without harm as a warning sign. Past luck is weak proof of safety.
- When the stakes are high, ask for proof that it is safe to go ahead, not proof that it is dangerous.
- Make sure the person who decides hears the strongest objection directly from whoever raised it.
- Notice when a decision changes because of who is in the room or what the schedule needs, not because of new facts.
07 — Sources
- Report of the Presidential Commission on the Space Shuttle Challenger Accident · NASA History Office · 1986-06-06
- Space Shuttle Challenger disaster · Wikipedia · 2026