Building the criteria x performance grid
A rubric takes a fuzzy concept like 'good writing' and breaks it into specific criteria (organization, clarity, mechanics, voice) and then defines what each looks like at different performance levels. A novice paper might have scattered ideas; a proficient one has a clear introduction, body, and conclusion; an advanced one has sophisticated structure that serves the argument.
The power is in the specificity. Instead of a teacher's gut feeling, a student can read the rubric and understand exactly what 'advanced' means for 'voice'. Two teachers using the same rubric will grade more similarly than if they each had their own mental standard.
Anchors and consistency across raters
Good rubrics use work samples (student papers, projects, performances) as anchors at each level. A real example of advanced work is better than any written description. When multiple teachers grade using the same rubric and the same anchor pieces, inter-rater reliability climbs dramatically.
The trade-off is labor upfront: writing clear criteria takes iteration, and collecting anchor pieces takes time. But once a rubric is solid, it scales. One well-designed rubric can guide grading for hundreds of students and be reused across years.