Columbia: a known problem, landing after landing

Through “Why it happened again”: Mike’s text from the paper, pp. 18–20, with the board’s report it quotes.

A history of success #

We had successfully re-entered and landed a number of shuttles with damaged or missing heat shield tiles that had been struck by foam shedding from the external tank. We might have concluded that the damage to or loss of those tiles created an unacceptable risk on re-entry.

Instead, we continued to launch shuttles with the foam shedding from the tank and damaging the tiles, until we lost Columbia.

What the board found #

Quoting from the Report Of The Columbia Accident Investigation Board:

Paper, note 11 Harold W. Gehman Jr. et al., Report of Columbia Accident Investigation Board, National Aeronautics and Space Administration, August 26, 2003, http://www.nasa.gov/columbia/home/CAIB_Vol1.html, ch. 6.1.

The shedding of External Tank foam – the physical cause of the Columbia accident – had a long history. This raises an obvious question: Why did NASA continue flying the Shuttle with a known problem that violated design requirements? It would seem that the longer the Shuttle Program allowed debris to continue striking the Orbiters, the more opportunity existed to detect the serious threat it posed. But this is not what happened

Paper, note 12 Ibid.

With each successful landing, it appears that NASA engineers and managers increasingly regarded the foam-shedding as inevitable, and as either unlikely to jeopardize safety or simply an acceptable risk.

Why it happened again #

From the paper’s discussion of all three NASA cases:

In each of these three cases, the post-accident investigation found that a problem had existed that should have been recognized and corrected, and was not.

What may be somewhat original, or at least worthy of consideration, are some possible insights into why these problems seem so intractable. Why, after these factors have been repeatedly identified in formal accident investigation reports, do they then show up again in subsequent accidents? Why did we have Challenger after Apollo One, and why did we have Columbia after Challenger?

Paper, note 13 Ibid., ch. 11, “Recommendations.”

The Report Of The Columbia Accident Investigation Board, among other recommendations, called for a “new culture” at NASA.

I will suggest that as decision makers we can become victims not merely of a culture, but of a deeper and more persistent human psychology.

I think that as decision makers we face an unexpectedly difficult challenge arising out of a powerful combination of the inherently deceptive mathematics of probability, the potentially self-deceptive psychology involved in evaluating the quality of our own decisions, and the inevitable significant pressure we experience to make a go decision even when we have doubts about the safety of that decision.

The three cases: Apollo 1, Challenger and Columbia.

The pressure to go #

Mike’s commentary for this site, quoting the board’s report.

The board’s report examines NASA’s organizational, historical and cultural factors in its Part Two. Introducing that part, it says:

Chapter 6 documents management performance related to Columbia to establish events analyzed in later chapters. The chapter reviews the foam strikes, intense schedule pressure driven by an artificial requirement to deliver Node 2 to the International Space Station by a certain date, and NASA management’s handling of concerns regarding Columbia during the STS-107 mission.

More on this factor: Factor III · The pressure to go.

Which factors it shows #

I · Deceptive mathematics

The board observed that the longer the foam strikes went on, the more opportunity there was to detect the threat. A long string of safe landings did the opposite: it read as evidence of safety, when it is what a small risk per flight produces anyway.

II · Murphy’s Corollary

“With each successful landing,” the shedding came to be seen as inevitable and acceptable. The landings that went well were taken as proof that the foam damage could not cause a loss.

III · The pressure to go

Alongside the foam strikes and management’s handling of concerns during the flight, the board reviewed “intense schedule pressure driven by an artificial requirement to deliver Node 2 to the International Space Station by a certain date.”

Questions for discussion #

1. Which known problem in your program violates a design requirement, but has never yet caused a failure?
2. With each success, has your view of that problem moved toward “inevitable” or “acceptable”?
3. What would it take for your program to treat a long history of a problem as more warning, not less?
4. Which date in your program is fixed by a requirement rather than by readiness? How would it change your answer to a concern raised the week before?
5. The board called for a new culture at NASA. Which part of the problem would a new culture not change in your own decision making?
6. Estimate the chance that your known problem causes a loss on any one flight. Using the calculator, how likely is at least one loss over the flights you plan for the next five years?

These questions also appear, with every other case’s, in the Briefing Kit.

@misc{meier2026columbia,
  author = {Meier, Mike},
  title  = {Columbia: A Known Problem, Landing after Landing},
  year   = {2026},
  note   = {Edition 1.0},
  url    = {https://meiersafe.com/case-studies/columbia/}
}