When I was an undergrad, we had a PhD student running one of our sections of a large course. The course had a number of essay assignments which were marked by the section teacher. After working really hard on the first essay, I got a bad mark with kind of cryptic comments on the paper. For the second one, I did a sloppy last minute job and got a good mark. For the rest of the term marks varied randomly with respect to the quality of the work, so there was no sense that the teacher was in any consistent or repeatable way distinguishing good work from bad work. This very effectively demotivated the whole class.
I think rubrics are really great systematic tool to avoid this type of thing. It gives a standard assessment tool to both the marker and the students, so the goals and the methods of measuring progress are clear.