Thermostat controller: flawed logic behind control offset?

I have asked this question a couple of times over the years but never got a clear answer: why does the control offset get applied to the setpoint of the controller whereas you would think it should be applied to the current temperature of the slave physical thermostat that is being servo'ed? This leads to the situations when set target does not kick off any action if the offset is not unnecessarily large. If controller reads temperature off a sensor that differs from the controlled physical thermostat by more than the control offset , no action occurs. One has to set offset high to make things work. If you have an efficient two stage or a variable rate AC system, this slams it to high output when it could have just nudged it gently, just enough to turn on. The slave should be controlled dynamically by reading its temperature and updating target, wouldn't you agree?