Grading, like reading, like teaching, like refereeing, is a meeting of the minds. The reader/grader brings his training, judgement, and more loosely his imagination to his engagement with the text. The writer/gradee, in turn, crafts writings that accurately reflects that he means to say on the page and does not rely on the reader to give him the benefit of the doubt and supplies whatever meanings he chooses to see.
This symmetric set of responsibilities is, of course, completely asymmetric when it comes the power the two parties exert. A grader grades; he assigns a single letter or number to the writing that encapsulates all it’s worth. A referee decides on “accept” or “reject” and may well seals the fate of the paper. A reviewer recommends or dissuades and may well seal the fate of not only the writing but the writer himself. A student may appeal the grade. A researcher may revise his work. A writer may hope to redeem himself in other venues or with other projects. All may not be lost, but the tilt of power in the relationship is clear.
This club of power should not blind the powerful to the fact that an evaluation happens on the other end as well. Students read the comments on the paper they receive and decide whether they respect the judgement of whoever wrote these comments. Researchers read the reports and form an opinion of the competency of the referees. Writers chew on reviews and considers the worth of the reviewers. They do this regardless of the grades received, the final “accept” or “reject” judgement passed, or the thumbs up or down decision (although a favorable outcomes surely color the view).
Grades may be given; publication decisions doled out; and opinions freely expressed in reviews, but respect for those grades, decisions, and opinions must be earned. If one cares nothing for the respect of the writer whose work one grades, referees, or reviews, then so be it. Let one then hope that that respect is never needed when one does not wield such ultimate power.
If one does care, then take care to be able to defend one’s judgement even if one is not immediately called on to do so. Take care that the comments are not based on a sentiment of “if I were to do this, this is what I would do differently”. Grading, refereeing, and reviewing are not co-authorship. What one would’ve done differently is irrelevant except when it adheres to some objective criteria: the requirements of the assignment, the standards of the journal for publishing, or some other agreed-upon metric adopted in advance of the production of the work. The rest is should be saved for that most powerless of forums, the blog.