I don't understand it either and almost ignored it because of that. I suggest changing the title on HN to something like "Can the behaviour of C code depend upon the semantics of Python?"
The conclusion of the article is that the correctness of the memory model is circular, feeding upon itself, like the Ouroboros of myth.