It was never that. What happens is that the exam is set by a panel (think Maths, cos that can be seen as 'right/wrong'), and a mark scheme is developed to match it so that the army of markers are consistent. So far so good. Let's assume a paper is out of 100. In your memory, 50% is 'a pass', less is a fail. Not how it works.
All the papers are marked, and let's say raw scores are low across the country. That says, 'Exam too hard', not 'kids didn't learn enough'. From there grade boundaries are set, and since the paper was too hard, an A grade might need 65 out of a hundred. If it appeared from the raw results that the exam was 'too easy', then an A grade might need 85 out of a hundred.
Grade boundaries are set AFTER the exams are marked. Not to decide what grades 'to give out'; but to decide what grades 'have been earned'.
This method is called 'norm referencing'. When I was training in the RAF, many of my flying training tests were pass/fail questions with a fixed pass mark. That's called 'criterion referencing'.
O level, A level and GCSE have never been criterion referenced.