TFA is ridiculous with its stance. Yeah, there's this aspect of the design they can't test in their labs. It might even be an important aspect. But safety isn't a boolean "safe"/"not safe", it's a risk assessment, and I'm quite sure there are 100 (or 10000) other things they didn't test for. As long as they're taking all of this into their risk calculations… it's fine.

And if it doesn't blow up due to heat shield failure, TFA (and its references) will be forgotten.

And if it does blow up due to heat shield failure, TFA (and its references) can suddenly claim prescience, all the while this is one of thousands of factors that went into the risk assessment. If one really wanted to claim prescience, it'd need to be a ranking of a sufficient number of failure modes.

To illustrate the problem: I hereby claim they will have a "toilet failure". Now if they actually have one, I'll claim `m4d ch0pz` in rocket engineering.

(P.S.: it's a joke but toilet failures on spacecraft are actually a serious problem, if it really happens… shit needs to go somewhere…)