Decisions when the evidence changes

Boeing’s crewed Starliner test brought Barry E. Wilmore and Sunita L. Williams to the ISS in June 2024. NASA’s August 24 decision account describes helium leaks, reaction-control thruster concerns, testing and disagreement about the remaining uncertainty. The agency decided that Starliner would return without the crew.

That was a decision about a particular flight and its available evidence. It should not be rewritten as a claim that the vehicle had already completed regular certification, or as a prediction of every future program outcome. Test evidence can support a changed plan rather than the originally expected one.

NASA’s August announcement described a planned crew return using Crew-9. Plans can change, so a history should also consult a completed-event account. NASA’s March 18, 2025 release records the return of Wilmore and Williams with Nick Hague and Aleksandr Gorbunov aboard a Dragon spacecraft.

The two sources serve different purposes. One explains the decision with information available at that time; the other establishes the later return. Using only the forecast date from an earlier announcement would turn an expectation into an inaccurate outcome.

Keep three kinds of statement apart

A later successful uncrewed landing does not prove that a prior crew-risk concern was irrational. A decision must consider possible outcomes and uncertainty before the result is known. Learning from the outcome can improve later models without pretending that earlier teams possessed later information.

For a review exercise, make three columns: observation, uncertainty and decision. Put a measured anomaly in the first, what is not yet explained in the second, and the chosen response in the third. This separates evidence from interpretation and helps explain why additional data, an alternative vehicle or a delayed mission can matter.