The detail that gets lost every time this gets retold as "ignored warnings": Morton Thiokol's engineers didn't fail to see the O-ring risk. They named it, in writing, the night before launch, and were overruled after a Thiokol vice president told the company's own engineering lead to "take off your engineering hat and put on your management hat." That's not a paraphrase. It's Jerry Mason's actual sentence to Bob Lund during the January 27, 1986 teleconference, and it is the cleanest artifact in the entire Rogers Commission record of what happens when a decision gets physically relocated from the people who bear its consequence to the people who bear its schedule.
The statistical failure underneath it is real too, and it deserves its own citation: the chart Thiokol's engineers presented that night plotted O-ring damage only against flights where damage had occurred, discarding the successful cold-weather launches that would have shown the correlation holding. Edward Tufte's later analysis of this exact chart became the canonical teaching example of selection bias distorting a life-or-death decision. But the chart is a symptom. The disease is that nobody in the room making the final call paid a personal cost for being wrong in the direction of "proceed," while delay carried an immediate, visible cost in schedule and political embarrassment that landed on identifiable people that same week.
A launch director who delays pays today. A launch director who launches and gets lucky pays nothing, ever. A launch director who launches and doesn't get lucky has, by design, diffused the decision across enough signatures that no single name attaches to it. That's exactly what happened: the final call moved from an engineer with his name on a memo to a committee vote with nobody's name on the outcome.
This was never a data problem. Mission control had a temperature chart, however flawed. What it didn't have was a structure where the person saying "don't launch" held more power than the person who needed the launch to happen on schedule. Build that asymmetry into any organization and you will keep being surprised by catastrophes your own engineers predicted, in writing, the night before.