AirRabbit, assuming that I have interpreted your commentary correctly (#1101), then there may not be any significant difference in our views. However, there are some interesting points within the differences and in the viewpoint.
The training views appear to follow a linear cause-effect-fix path, identifying and improving an aspect which has an acceptable, but not guaranteed probability of success (the human).
The alternative is an attempt to understand a dynamic interaction of factors, like a flywheel which cannot be totally fixed (stopped) because then nothing would be achieved, thus the task is to apply a brake to maintain an acceptable speed range (influence human behaviour with small changes in the overall operating system – man, machine, and environment).
Rather than go into detail, there may be better explanation in the documents:
‘How to be safe.’
‘Myths in safety’.
‘A Tale of Two Safeties.’