Without getting into the pedantics of the specific examples, I'm sure you intuitively understand the difference between a exact value and an approximation, likewise discrete values and a fitted curve.

The argument is that models are only useful tools and should not be taken to be a full or only representation of the thing being studied.