I think you're thinking of far more complexity than I am. I'm thinking of common applied math - accounting, weights and measures, carpentry. These things still have the reality correspondence problem, but truth isn't necessarily absolute. If I measure a cup of milk, I can safely treat it as a cup of milk regardless of the number of molecules in it. That's a kind of truth. Utility/ matching observed reality ought to be enough truth for anyone. As we get better and better at observing, the models will change, and that's good.
