soverby

Sorry but your argument doesn't work because your key assumption for scoring 33% (i.e. random guessing) can not be used for a candidate who, instead of randomly guessing, is trying to select the correct answer.

That is, if you randomly guess, your expected return is 33%. However, if you instead attempt to answer multiple choice questions based on a combination of the provided information and your knowledge, this 33% is no longer relevant and your 'starting point' is back at zero.