So how does one avoid this basic error while formulating a good theory of nature ?
Optimistically, and very speculatively, I would like a good theory to, at least, formally suggest how a bunch of particles (or whatever concept we replace them with in the future) can come up with a good theory of themselves in the first place.
Or why not make this the starting ansatz such that one builds upon this very requirement in a way similar to how one builds a quantum field in a manifestly covariant way. Since infinite recursion seems to get in the way, maybe these good theories should incorporate a fundamental "unit of approximation" related to the maximal recursion depth or complexity cutoff.
Excuse my rambling.