This is classic HN pedantry. You really think these people would allow people to die to prove the point in the article? That they couldn't exercise sensible discretion on when failure might be the best course of action?

You mean like airplanes where they expect them to fail and so every vehicle has a nearly indestructible telemetry box with a homing beacon?

Failure is not the problem. Failing the same way and doing nothing about it is the problem.

Failing in predictable ways is always a problem. Even if it's a novel way of failing. Those boxes are there as last resort, not as an engineering strategy.

Having a last resort is an engineering strategy (and not a very popular one outside of safety critical industries).

I've seen plenty of projects that choose not to design or think about handling "unknown unknown" failure vectors because YAGNI

What does that have to do with anything?

I don't want swiss cheese code written by an under-resourced operation in my airplane. If anything, the more critical the application, the more this message applies.

I suspect if people were going to or actually died the author would have mentioned it.

Did you even read the article? It was a managed failure, nothing major.