The "Have you ever thought about how GPS works?" question gets a scripted "No, but <blah blah blah>" answer. It's fully automatic, because there are now two possibilities:
- I have thought about GPS before. I will now fake my creative process. No hard feelings, but you are just another obstacle for me to get the job, that's it.
- I have never thought about GPS before. We will now try to design "together" a super-complicated system having so many boundaries and hidden constraints, while I will be also panicking on the inside the whole time. No wait, that would be dumb. I know, I will just refuse to discuss it further. Next question.
(Also, TIL that "flack" in addition to its primary meaning is apparently an acceptable spelling of "flak" too.)
There should be no expectation of them coming up with the “correct” answer. But they should be able come with an answer and explain it clearly, warts, holes and all.
Using something real like GPS also ensures that the candidate understands why the problem is a useful problem to solve, and what the objective of the solution is I.e. a system that lets you locate yourself on earth.
But directly talking about those topics is probably handy.
But as a one-off, and if asked in a forgiving way to explore how they think, then I'd consider this question valid.
Maybe "Have you ever thought about how GPS works?" if "yes", then let them explain, if "no", then make it easy for them to start reasoning about the system and then see how they might design it.
Seems to me as fair as asking "have you ever thought about how a double-linked list works?" or "a basic way to ensure database consistency when the DB is replicated at two different physical machines"?
It's not like GPS is something esoteric that they are unlikely to have interacted with, odds are they used it to get to the interview location.