Actually measuring, or even hypothesizing about, the predictive capability of different models of varying levels of abstraction is something that happens distur

Actually measuring, or even hypothesizing about, the predictive capability of different models of varying levels of abstraction is something that happens disturbingly rarely in fields that aren't the hard sciences. I suspect an important (but not primary) reason for this is that most people simply don't have an opportunity to be exposed to the type of maths and analysis required for systems theory, and thus are extremely unlikely to surmise that something like systems theory is a distinct thing that can even exist.