On January 28, 1986, Challenger exploded 73 seconds after liftoff. The O-ring seals had never been tested at a temperature as low as that of the launch morning. The accident was not a failure, it was a prediction inscribed in the limits of test conditions.
Each critical component accepts a calculated failure rate. Risk calculation transforms statistical accident into determined event. This acceptance programs the probability that the system will collapse.
Morton Thiokol had recommended postponing the launch. NASA asked for proof that the seals would fail, not proof that they would work. NASA asked to be convinced of failure, not reassured of success.
The accident does not close the negotiation between optimization and collapse, it documents it. The report that follows programs the tolerances for the next cycle.
Doctrine
The accident is not proof of system failure. It is proof that the system functioned to its contractual limit. What was tolerated becomes prescribed.
Vecteur ouvert
In a dead language, who bore the burden of proving that a syntactic compromise was unacceptable, and at what moment did this reversal of the burden of proof sign the trajectory of the language?
