Indeed, but it is really only the system response we are interested in. As you admit, a clear causal explanation for how a single input forcing is increasing doesn't automatically get us to a position where we can predict the overall system response. Or predict whether a given system response will have some positive or negative second order effect.

We have a clear causal physical explanation that eating fat should make you fatter right? Are you willing to say that? Or would you qualify yourself - 'eating more fat will make you fatter in the absence of negative feedbacks (perhaps the fat reduces your appetite more than carbs or protein?) and assuming that all other inputs (exercise level etc) remain constant'.