It means: with respect to some axiomatization of arithmetic that allows Godel numbers to be constructed. Peano arithmetic, for example.
> obviously incomplete
I guess that depends on what you consider "obvious". It was far from obvious before Godel proved it. But nowadays it's not hard to understand. One generation's "obvious" is a prior generation's major breakthrough.
The big surprise in Godel's theorem is that the amount of complexity you need to introduce into a theory to render it incomplete is quite small. Peano arithmetic is sufficient, but not necessary.
That's circular logic; why would that be an interesting question to investigate absent the incompleteness theorem?
> The big surprise in Godel's theorem is that the amount of complexity you need to introduce into a theory to render it incomplete is quite small.
Agreed, but that implicitly requires understanding that small theories like Presburger Arithmetic are complete (which they are in every usual meaningful sense, but you seem to be claiming otherwise).
Because the amount of logical infrastructure that you need is surprisingly small. All you need is to be able to do basic grade-school arithmetic: add, subtract, multiple, divide, test for equality. And you don't even need all that. You can build everything you need from much simpler primitives.
The reason this is not circular is that before Goedel's theorem mathematicians were trying to construct a formal system that would allow you to derive all true statements about not just arithmetic, but all of mathematics. It turns out that this is impossible. As soon as you can do arithmetic, Goedel's theorem applies and you have lost.
> that implicitly requires understanding that small theories like Presburger Arithmetic are complete
I don't see why. Goedel's theorem matters in the context of trying to reduce all of mathematics to a formal system. Presburger arithmetic fails not because it is syntactically incomplete but because it cannot (for example) do division, so it cannot formulate (for example) the concept of a prime number and so it cannot (for example) prove the fundamental theorem of arithmetic. So yes, PB is decidable, but it is not a plausible candidate for a complete (in the informal sense) formulation of all of mathematics. Goedel's theorem shows that if you can add, subtract, multiply and divide and test for equality, then you have lost. And if you can't do those things, then you have also lost, but you don't need a theorem to tell you that.
What does "with respect to the truths of arithmetic" even mean? How is Presburger Arithmetic "incomplete" in this sense?
> And those two things would be...?
That it's so complex as to be obviously incomplete, but also so simple that anything smaller than it is obviously trivial. There are interesting systems smaller than PA, and none of this is obvious.