I am not sure what you mean -- measuring impulse response in cars good but measuring impulse response of datacenters bad (or impossible). Measuring the weight o

I am not sure what you mean -- measuring impulse response in cars good but measuring impulse response of datacenters bad (or impossible). Measuring the weight of a car is an empirical observation as well. I also don't see how it follows that PID will necessarily overfit second and 3rd order effects whereas bang-bang wouldnt