> Obviously, explanations can be more or less fundamental.
That's not obvious, in fact, it's not even true. Feynman's point is that people bring context to a question. People will accept different things as 'fundamental'; different assumptions and axioms. Can you sometimes explain things using fewer axioms? Sometimes; but is it more fundamental? Is it better to build a theory on fewer axioms, even if they can't be directly validated? Would you prefer a theory based on 4 axioms, none of which can be directly validated, over one with 6, but you can measure all six directly? What if the four can't be re-derived from the 6, but all observable statements can? What if either set can be re-derived from the other?
Feynman isn't saying that "physicists don't really know on a fundamental level why magnets repel each other". He's alluding to the idea that that question isn't well defined. He's encouraging the questioner to figure out what sort of answer would satisfy them, and why.
I think it's obviously true that some explanations are more fundamental than others. We have, e.g. a more fundamental explanation of tides than anyone did in 500AD. That is not to deny that it is difficult to make precise what exactly it is that makes one explanation more 'fundamental' than another. As always, individual cases are clear; the general principle is elusive.
>He's alluding to the idea that that question isn't well defined.
It's well-defined enough to answer. Hundreds of millions of school children learn a perfectly sensible answer to the question every year. (Would this answer satisfy someone with a PhD in physics? Obviously not. But that's not the point.)
It's in any case bizarre to insist that a layperson ask a question that's well-defined according to the standards of a particular field.
That's not obvious, in fact, it's not even true. Feynman's point is that people bring context to a question. People will accept different things as 'fundamental'; different assumptions and axioms. Can you sometimes explain things using fewer axioms? Sometimes; but is it more fundamental? Is it better to build a theory on fewer axioms, even if they can't be directly validated? Would you prefer a theory based on 4 axioms, none of which can be directly validated, over one with 6, but you can measure all six directly? What if the four can't be re-derived from the 6, but all observable statements can? What if either set can be re-derived from the other?
Feynman isn't saying that "physicists don't really know on a fundamental level why magnets repel each other". He's alluding to the idea that that question isn't well defined. He's encouraging the questioner to figure out what sort of answer would satisfy them, and why.