So a HUGE part of the problem was NASA all along as usual.
I am going to disagree with that!
That is the same type of logic as a drunk driver telling the police officer that he has always gotten home alive in the past, even if he had some new scratched on the car in the morning.
The issue is what level of
certainty can you get that the craft is going to work sufficiently well to provide a safe trip.
"Modeling" risk is not a prefect science. The models "fail" as well as the hardware and the human operators, in that there are problems that are not modeled correctly or completely and others that are just not known so are not in the models at all.
So, at best, the results of a risk model need to be considered in the context of saying that the risk is at least as bad as the results. If there is a lot of uncertainty in how to model something, sometimes all you can really do is look at the "importance" of the system failing, to see what the
conditional probability is for unacceptable results.
In this case, with so many thrusters needing to fire in so many directions so many times, the logic model for what can go wrong enough to cause a fatal situation is quite complicated. And, with a "common cause" for failure affecting of all of the thrusters, the model has to consider if one thruster failure will create higher demand on other thrusters and cause them to fail too , in a cascading process. So the end result of the calculation becomes extremely sensitive to the input probability that a thruster will fail, which is not even a single number. For a single thruster, the probability is going to be a function of the firing sequence - involving the time between firings, their duration and total times fired - for each possible scenario of multiple thruster failures at various times.
I doubt anybody on this forum, including myself, is in a position to realistically second-guess NASA on this decision.
I fully support their decision.