The thread installs the Challenger O-ring decision as a public processing case: a temperature concern that began as a contractor no-launch was recoded, the same night, as acceptable risk, after years in which erosion had already been recoded as experience. The unresolved question is which repair would have stopped that recoding: a rule that "outside the database" is already a no, and that a launch constraint cannot be waived by the project office that owns the flight rate (the Rogers structural model); or a change in what counts as a signal, so that successful flights with unexplained damage stop counting as safety evidence (the Vaughan/Feynman model). Those are not substitutes. Rogers already asked for a policy on launch constraints and an independent safety office; CAIB Chapter 8 found foam debris following the same trajectory. The live question is which of those, if installed before 51-L, would have kept the 53°F recommendation a no.