[bell-cot]

> But I don't think you can characterize a burned-through flap as minor...

In prod, a localized flap burn-through would be a Major Incident.

Vs. in dev...what competent manager would be bothered if the some bleeding-edge beta code dropped 10% of packets the first time that it faced a full-load test?

[ajross]

That sort of depends on context. If the failure was reasonably expected, like for example your router is running on prototype underclocked hardware or is a debug build that is known not to have the final performance characteristics, sure. But if it fails and you don't immediately suspect why it failed, then that's a sign of a genuine design flaw.

And sure, it's good to find the design flaws in "dev" vs. "prod", but it's still bad to have them.

In this case, we don't really know what the expected performance of that flap shielding was. Maybe it was a kludge and they were just hoping the glue held together, maybe it was a finished design they just thought they were validating. The latter is a much (!) bigger problem than the former.

[bell-cot]

Per Wikipedia, SpaceX's 3 "Get out of Mishap Investigation, Free" cards from the FAA for this test flight were -

Starship burns up during reentry

Flaps lack sufficient control authority

Raptor engines fail to relight

- suggesting that both the flaps and heat shielding were regarded as highly experimental, and failure(s) in them very likely to occur.