Even though I am a junior member and I will be attempting CSS for the first time in few months but the irony you have mentioned is simple to understand actually.

Its a descriptive type of exam. Unlike many international exams in which only one answer is right and every question is designed to judge multiple aspects of candidate, CSS still falls in the category of archaic exam systems where descriptive answers are required to judge a candidate.
This poses great problem. Greater when the number of candidates is so huge. The factors which influence the final result are greatly controlled by the person who checks exam papers. His understanding, his personal approach, his idea of an ideal expression, his favorite way writing, his favorite style of referencing and so much more. So when there is no perfect answer for a question and when there is no absolute standard of awarding a candidate, how do you see the accuracy of whole system to select the best candidates?

I heard people saying CSS is all 50% about luck. I say , its the failure of a system . When a system can't give a perfect output, it relies on random selection of choices. sometimes it works, other times it fails.

Peace