Because it is the difference between protocols and APIs, and why, regardless of the status of code APIs, protocols cannot be copyrighted. This is also the difference between programs and algorithms.
But you seem to have interpreted the necessary condition that a work must be fixed in order to be eligible for copyright as "any fixed portion of a work is copyrightable." This is untrue both because being fixed is necessary but insufficient for copyright, and because tiny portions of a work aren't works in themselves. Also, I have not claimed that anything that might be called an API and is fixed is necessarily copyrightable. I only said what is definitely not copyrighted. So no goalposts have been moved, you just misunderstood the necessary condition and my point: Protocols (like algorithms), unlike code APIs (and programs), aren't fixed, and so cannot be copyrighted regardless of anything else that may or may not be. That some tiny portions of a protocol might be fixed does not change the status of the work.
But you seem to have interpreted the necessary condition that a work must be fixed in order to be eligible for copyright as "any fixed portion of a work is copyrightable." This is untrue both because being fixed is necessary but insufficient for copyright, and because tiny portions of a work aren't works in themselves. Also, I have not claimed that anything that might be called an API and is fixed is necessarily copyrightable. I only said what is definitely not copyrighted. So no goalposts have been moved, you just misunderstood the necessary condition and my point: Protocols (like algorithms), unlike code APIs (and programs), aren't fixed, and so cannot be copyrighted regardless of anything else that may or may not be. That some tiny portions of a protocol might be fixed does not change the status of the work.