Dear Christine,
your notion of 'metadynamics' is an intriguing one. The correspondence between logic, physical systems, and computer programs has been remarked upon before, perhaps most well-worked out in Baez' 'Rosetta Stone'-paper (https://arxiv.org/abs/0903.03409), where he proposes a categorical equivalence between Feynman diagrams, cobordisms, proofs in logic, and computer programs. However, I do not recall any attempt, in that paper, to fold the correspondence back in on itself---to find an equivalent to logic talking about itself within Feynman diagrams, for instance.
I also like your idea that one might get physical laws out of metadynamics---as fixed points, perhaps, of dynamics applied to dynamics, maybe in a way similar to how Löb's theorem (https://en.wikipedia.org/wiki/L%C3%B6b%27s_theorem) can be used to construct modal fixed points, the most famous of which is the Gödel statement itself, equivalent (in this fixed-point sense) to the inconsistency of arithmetic.
However, I believe the correspondence is usually drawn a bit differently from the way you frame it---one takes the axioms to be something like the initial state of the system (sort of like the input into a program), then proofs as parallel to the dynamics (the computation), with theorems (truths) coming out at the end (final states). This then leads to something like my proposal, in which what 'comes out at the end' may be subject to undecidability---thus leading to the unpredictability of measurement outcomes, as in quantum mechanics.
Anyway, thanks for an enjoyable essay. I wish you the best of luck in this contest!
Cheers
Jochen