Caltech tests were not based on memorization, as they were "open book open note". You had to reason your way to a solution.

But I do agree that real world physics, like designing an actual electronic circuit, have behaviors that are not modeled by the usual mathematical models. For example, resistors vary widely from their marked resistance. And I was told, when building digital circuits, to make sure it worked with chips faster than the spec, as replacement chips are always faster, never slower.