Slightly exploratory thread here but I wanted to highlight one thought:

The spectre of correspondence lives on even after we feel we’ve exorcised it if we still maintain a view of theories as models, i.e. as bodies of statements related to one another by a logical syntax which has some kind of isomorphism to the causal structure of the world. If you accept that relations between statements have to be partly practical rather than wholly logical, theories can’t be models. Model talk is just Rorty’s Mirror reimagined in terms of dioramas instead of reflections.