> I'm sure that isn't always true, that if something performs at the extreme it necessarily means it performs in the usual. That’s why we test over the operati

> I'm sure that isn't always true, that if something performs at the extreme it necessarily means it performs in the usual. That’s why we test over the operating range for the different parameters, not just at the limits. > It's one thing for the purposes of science, to know, but my understanding is that engineering and material specifications for an actual thing needs to be more practical, especially for something that already costs $3.5B. You need to be pessimistic about operating conditions if you don’t want it to fail unexpectedly. That is absolutely critical in engineering. Now, you set the boundary differently depending on the field and the importance of the widget you engineer (tolerances are quite different for a phone than for an aircraft), but the concept itself is part of engineering. There are cases where practicality is less important than reliability. Things like space probes and a lot of things related to national security. That is one of the reason why they cost billions in the first place.