My understanding of ETTR must be fundamentally different from other folks. I have always practiced it by placing the brightest significant highlight as close to clipping as is possible without actually clipping it. This would make it scene referred and dependent on the tonal scale of the subject and would preclude a simple and fixed, one-size-fits-all EC adjustment.
You are absolutely right in theory.
The difficulty with an
in camera histogram is that it doesn't necessarily tell the truth with respect to the RAW information.
It varies sensor to sensor but it is likely that the RAW file does not clip where the histogram says it does.
This is because the the histogram is based on JPEG representation and not the RAW file.
(Many in the industry are calling for camera manufacturers to at least give the user the option of using a RAW based histogram.)
So, in some cameras almost two stops of exposure headroom may be available despite the histogram warning of clipping.
Thus, exactly how to expose with an individual camera is a dynamic combination of the principle you have elucidated along with knowledge of exactly how much exposure headroom that camera possesses. One usually works this out by trial and error.
Considering that half the information that a sensel can gather is found in the brightest stop and a further quarter in the second brightest stop a lot of tonal information is lost if one's exposure decisions are wrong.
So it is well worthwhile working out exactly how the sensor in one's camera behaves to take full advantage of what the sensor can capture.
One post in this thread has made mention that Canon cameras seem to chronically underexpose anyway and in my experience this is correct.
However with a little experimentation I have worked out exactly what the exposure headroom really is and thus expose accordingly.
Regards
Tony Jay