The contractual requirement was for max 1-in-1000 probability of loss of crew on re-entry.

Let's suppose, for the sake of the argument, that due to these thruster issues, the actual probability had increased to 1-in-100.

In that scenario, it is totally expected that Starliner made it back to earth in one piece – there was a 99% chance it would happen. Yet simultaneously, 1% is 10 times riskier than 0.1%, and NASA absolutely made the right call in not putting astronauts on it when they had the choice.

The real issue, apparently, is NASA wasn't even confident in calculating that probability, since there were aspects of how the thrusters were behaving which nobody could explain. Although ground-based testing replicated some similar issues, the issues reproduced on the ground had some key differences from those occurring in space, and nobody had a convincing explanation for the differences.

Due to the flawed design, Boeing was pushing components past their certified thermal limits. And how they behave when you do that isn't well understood, because the industry standard approach is to not do that. Due to safety margins, you can get away with it to an extent – but exactly how far you can get away with it, and how exactly they will fail when they eventually do – those aren't questions people have a lot of experience in answering.