Imagine, for a moment, that students acquire valuable human capital during their time at university. Imagine that the grades on a student's transcript reflect his or her level of human capital. Imagine that, every term, a professor uses examinations, term papers, and other assignments, to measure how much human capital each student has acquired over the course of a term.
Conventional wisdom suggests that the best way to go about that assessment is to have clear expectations. See, for example, these helpful suggestions from University of Waterloo's Centre for Teaching Excellence:
Transparent marking criteria. Students should know what is expected of them. They should be able to identify the characteristics of a satisfactory answer and understand the relative importance of those characteristics. This can be achieved in many ways; you can provide feedback on assignments, describe your expectations in class, or post model solutions on a course website.
This sounds eminently reasonable. If professors expects answers to true/false/uncertain questions to contain a diagram, they should say so. If they expect students be familiar with IS/LM analysis, they should say so. If they expect term papers to be a certain length, or use a certain referencing format, they should say so. Clear expectations mean that students are evaluated on what they actually know, rather than on how well they anticipate the whims and idiosyncrasies of their professors' marking schemes.