The Evolution of Writing Rubrics: Good, Bad, and Ugly

Jan 21, 2025

If I had to write the rubric of all classroom writing rubrics, the first feature I would consider is ‘visibility.’ There are always rubrics in every classroom. The most impactful ones aren’t written down anywhere, what is good work, what is bad work, what is ugly work. Dragging the bad and the ugly out from academic shadow spaces—spotlighting what to avoid—are these judgments visible for all stakeholders to see? There is a matter of fairness.

Next, I would consider ‘validity.’ Much more challenging than visibility, validity cuts through any hemming and hawing about the worth of any particular rubric as an assessment tool. Does this rubric pertain to the real thing it purports to evaluate? Or does it support someone’s ideology regarding what should be? Color, clarity, cut, and carats—the four Cs expertly define a diamond in terms of its rank in the field of jewelry. Diamonds scoring “off topic” end up in the industrial grade program. It helps to know how to tell the good from the bad.

Third, I would look at ‘utility.’ Simple enough on the surface, but once you open the package, does it work? For whom? Often enough, they come with no instructions. And, again, for whom does it work? Where does it come from? What will be done with the scores? They may have great utility for those in positions of authority, who must write reports for a living, but what about the students, who must write reports for college and career? Does anyone know about the rubric’s uptake? Does anyone care?

Finally, I would consider ‘beauty.’ Elegance, clarity, depth, resonance—how does this rubric look in the eyes of a student? Is its simplicity in its treatment of complexity? Or is it just oversimplified? What does it invite the student to think about during an opportunity to improve their writing skills? Is it aesthetically captivating? A rubric to assess writing should itself exemplify beauty.

THE WRITING RUBRIC OF RUBRICS FRAMEWORK

VISIBILITY

I: The Good: Assessment criteria are fully transparent and accessible to all stakeholders. The class has discussed the meanings of each level. Both exemplary practices and potential pitfalls are clearly documented in the Bad and the Ugly. Hidden or implicit standards are brought fully into the light and made explicit.

II: The Bad: Some assessment criteria remain obscured or unstated. Expectations are partially documented but leave important nuances implicit. Not all stakeholders have equal access to understanding evaluation criteria.

III: The Ugly: Assessment criteria operate largely in shadow spaces. Expectations remain hidden and unspoken. Students must guess at unstated standards or rely on privileged knowledge to succeed.

VALIDITY

I: The Good: Criteria directly measure genuine qualities of effective writing in the given territory of the universe of discourse. Like the four Cs of diamond assessment, cues are concrete, relevant, and observable. Criteria align with intentional writing purposes.

II: The Bad: Criteria sometimes refer to superficial or vague aspects of writing (e.e., voice). Connection between the criteria and the writer’s intention is unclear. Some criteria lack concrete, actionable specificity (e.g., voice).

III. The Ugly: Criteria fail to describe meaningful aspects of writing quality (e.g., interesting, absorbing). Criteria are unrelated to the writer’s intentions. Criteria are arbitrary or trivial.

UTILITY

I: The Good: Rubric is readily usable by both students and teachers to improve students’ ability to use the tools of writing to craft effective texts. Implementation guidelines are clear for students. Assessment process serves learning goals first and administrative needs. Students can use the rubric effectively to self-assess. Students can modify the criteria as needed to craft adjacent genres.

II: The Bad: Rubric presents implementation challenges. Guidelines leave room for confusion as students work to construe them and use them. Assessment process serves some stakeholder needs but may neglect others. Student self-assessment is an afterthought and difficult.

III: The Ugly: Rubric is impractical or impossible to implement effectively. Guidelines or explanations are absent or unclear, often externally mandated. Assessment process fails to serve key stakeholder needs. Students cannot use criteria for self-assessment.

BEAUTY

I: The Good: Design demonstrates exceptional elegance and clarity. Visual presentation enhances understanding. Language inspires deeper engagement with writing craft. Format invites thoughtful reflection and discussion on quality. Implementation practices are flexible in that students are invited to rephrase and rewrite criteria.

II: The Bad: Design is functional but uninspiring, sometimes paternalistic. Visual presentation is basic. Language focuses on ideological mechanics over sociocultural craft. Format treats assessment as routine checklist rather than opportunity for growth.

III: The Ugly: Design is cluttered and confusing, often designed for scoring or reporting purposes. Visual presentation impedes understanding. Language is mechanical and reductive. Format reduces writing to shallow technical exercise.

Why Do We Have Rubrics?

According to several researchers, rubrics specify criteria and quality levels for evaluating student work (Andrade, 2000; Bruning et al., 2013), and as such, are expected to reduce students’ guesswork about grades. Empirical research has demonstrated benefits from using them in this regard, but in the age of AI, it is essential to reflect on the limitations of rubric use in writing classrooms. For one thing, students can now prompt a bot to write a product based on a topic and the specifications in a rubric. In this instance, the rubric becomes an evaluative tool that makes it easier for students to plagiarize. The “good” rubrics are better tools for plagiarism than the “ugly” rubrics.

Studies suggest that rubrics can enhance assessment accuracy and provide clearer expectations for students. According to Andrade et al. (2009), rubrics help students understand assessment criteria and improve their ability to self-evaluate, though we understand there is a difference between self-assessment and self-evaluation. Additionally, research by Panadero and Jonsson (2013) found that rubrics can increase transparency and help generate detailed feedback to guide student improvement, although we have no knowledge of the quality of the rubrics under study in these analyses.

Emerging research raises concerns about potential drawbacks. Bender and Koller (2020) found that introducing rubrics too early in the writing process may place limits on student creativity and metacognition. Similarly, Wilson (2017) argued that using rubrics solely for summative assessment, rather than formative learning, fails to capture student growth as writers. In other words, rubrics are designed to assess products; what transpires along the way to making a textual object is beyond the reach of the criteria.

According to the evidence, the most effective approach appears to be involving students in rubric creation and using rubrics throughout the learning process. As demonstrated by Andrade (2001), when students participate in developing assessment criteria, they gain deeper understanding of quality standards while maintaining agency in their learning. Implementation anchored in agency transforms rubrics from simple, taken-for-granted tools of compliance into tools of craft that expert writers make for themselves as they set out on a productive path toward publication.

This evidence-based perspective suggests that thoughtful rubric implementation, focused on student engagement and ongoing learning rather than final assessment, offers the greatest benefit to teaching and learning outcomes.

AI Rubric Generators

Educators are adopting generative AI platforms like ChatGPT, Microsoft Copilot, and others to design rubrics tailored to specific assignments and learning objectives. AI simplifies the process by generating detailed evaluation criteria, scoring scales, and descriptors, which can be customized further by educators. This approach reduces the time spent on rubric creation, work that in earlier years required expert human activity. But as bit of an expert on rubrics, I find these rubrics flat, often vague, encouraging subjectivity. In this matter, I can’t see much of a role for AI. Humans are my choice of agent to write writing rubrics.

The amount of human labor expended to write formal rubrics for large-scale evaluation was massive, and important, before AI. Having participated in a variety of rubric development projects, I witnessed the emergence of a suite of a dozen genre-specific writing rubrics implemented across the state of California as teaching and accountability tools. These rubrics were written empirically through statewide field tests. Hundreds of student essays written in response to genre-based writing prompts were analyzed by teams of teachers, searching for examples to serve as dead center anchors for each one of six score points. Then the work of writing concrete descriptors for each scorepoint began. The entire process took half of a decade.

The creation of national portfolio rubrics for assessing writing in the English-Language Arts in the 1990s lasted six years. Building a rubric to score student portfolios created in 19 differing states by middle and high school students involved 1) constructing a strand course-embedded writing tasks lasting several days to serve as data anchors, 2) field testing these mini-portfolios created by students in response to those tasks, 3) sorting student responses into the remarkable, the good, the ok, and the ugly, and 4) doing the painstaking work of analyzing student writing falling dead center in the score points, classic exemplars, to 5) deduce concrete criteria recognizable as the good, the bad, and the ugly in each of those 19 states.

Educators now can use AI tools to design rubrics tailored to specific assignments and learning objectives in the blink of an eye. These tools simplify the process by generating detailed evaluation criteria, scoring scales, and descriptors, which can be customized further by educators. Platforms such as Canvas LMS are incorporating AI-based rubric generators to automate parts of the rubric development process. These tools analyze assignment descriptions to produce relevant criteria and logical scoring scales, attempting to assure consistency and fairness in grading.

We hear much gnashing of teeth rooted in traditional grading about students’ prompting bots to write their assignments for them rightfully so. But where is the sturm and drang surrounding this usurping of the traditional teacher’s job in creating rubrics based on real student work? Ceding the work of prompt writing and rubric creation to AI gives me pause for concerns about pedagogy.

Just as students learn more about writing when they are engaged in the rubric-writing process, teachers gain expertise when they participate collaboratively in examining student work products and portfolios to construct their own rubrics. Using AI to create writing assignments and rubrics is the first wobbly step down the slippery slope to further dehumanization of writing instruction.

Composing a Negotiated Rubric

In a reflective essay, Furman (2024) explicated the complex metacognitive processes involved in creating rubrics. He argued that rubric development demands intense metacognition as teachers "grasp for rationale, rigor, depth, and real value" (p. 221). While this process traditionally fell solely to educators, as noted earlier, Furman wrote from the vantage point of involving students in rubric creation even in the presence of AI.

According to Furman (2024) and aligned with my experience, the rubric development process turns on several critical questions teachers need to answer for themselves: 1) determining expected local student knowledge and skills, 2) establishing local evaluation methods, 3) identifying local transfer goals, and 4) clarifying local instructional objectives. When students participate in addressing these questions with their teachers, they gain deeper insight into assessment criteria while developing metacognitive skills crucial for academic success. This collaborative approach transforms rubric development from a unidirectional teacher task into a dynamic learning opportunity.

One approach to negotiated rubric building I call “Looking back” means collecting examples of student-written products in response to a prompt and then tasking students working together to rank the products and find specific criteria that describe what they see as good, bad, or ugly (I don’t advise using these terms in class—perhaps excellent, good, needs work, needs a resurrection).

Another approach I call “Looking ahead” is to task students with writing rubrics for well-established genres they enjoy or for their immediate writing assignment or project even before they begin to draft. If you were going to write a book or a movie review, what might you see in a good one, a bad one, an ugly one? If you were going to write a memoir about your earliest memory, what’s a good memoir going to read like?

Traditionally, I can affirm that educators have received minimal training in rubric design during their credentialing programs, if any, leading them to rely heavily on their own educational experiences and conventional grading practices (Furman, 2024). I don’t advise looking to AI to generate situated, embedded, community cultural materials of the significance of rubrics.

However, when teachers partner with students to create rubrics, both parties benefit from fresh perspectives and increased engagement in the assessment process. AI exchanges may be a reasonable use to insert into student-teacher negotiations, but student and teacher control must permeate. Students bring valuable insights about their learning needs and challenges, while teachers guide the development of meaningful criteria aligned with learning objectives they have in mind for this particular group of writers at this particular time..

The collaborative development process requires careful consideration of multiple elements, including who sets assignments parameters, specific assignment criteria, quality level gradations, clear expectations, and feedback mechanisms. When students help craft these elements, they develop a deeper understanding of assessment and self-assessment and take greater ownership of their learning. This shared responsibility for assessment criteria can increase student motivation and self-regulation.

Despite the additional time and effort required for collaborative rubric development, the benefits justify the investment. Students gain metacognitive skills, deeper understanding of assessment as a practical matter of construction, and increased agency in their learning. Teachers benefit from student insights and increased student buy-in to assessment criteria. Together, they create assessment tools that provide "real value to the student and instructor" (Furman, 2024, p. 221) while fostering a more equitable and engaging learning environment. Fortunately, I may have the opportunity to test this hypothesis in a Spring term English class for seniors where we are viewing the students as stakeholders in the evaluation process much like their teachers are stakeholder.s

So What?

Professional development sessions, such as "Lunch and Learn" events, are being conducted to train educators on leveraging AI for rubric design. These sessions highlight the benefits of AI-generated rubrics, including improved grading transparency and reduced workload, while also addressing challenges like ensuring alignment with external standards. Conspicuously absent are discussions of how these rubrics land with the students who are expected to use them and get grades based on them.

Some institutions are concerned about the ethical use of generative AI in rubric creation and recommend aligning rubrics with higher-order cognitive skills (e.g., Bloom's Taxonomy), although it remains unclear to me precisely how Bloom’s Taxonomy from the 1950s interfaces with AI web scraping from 2025. Perhaps the issue is equity? Perhaps the issue is privileging higher-order thinking in rubrics? Frankly, I don’t understand this recommendation from the language published online and bring it up as a serious question. Higher-order thinking is good. Rubrics are good. Something seems interesting, but I need help. The writing online would not fall in he “good” category, but who am I? The following caveat from the authors online may help:

“We would however encourage you to cross reference your Gen-AI agent with other outputs from publicly available Gen-AI tools, knowing well that, as an educator, you would still need to cast your critical eye over the final product. A peer review of your final product would also enhance the rigor in this process.”

Of course, peer reviews of rubrics are good. I would encourage this practice with or without AI involvement, but even more powerful would be teacher’ peer reviewing rubrics negotiated with students.

The stakes are high. Magic School, for example, an online rubric generator that aligns with extant standards, argues for mass use of its service because it produces standard-aligned, no fuss, no muss tools for ‘objective’ scoring. I leave you with the Magic School argument for teaching efficiency and learner effectiveness and invite you to draw your own conclusions.

“The Rubric Generator is a valuable resource for educators seeking to create well-organized, table-format rubrics that clarify assignment expectations and facilitate objective assessment. By enhancing assignment clarity, assessment objectivity, and feedback for improvement, this tool promotes effective teaching and learning practices. It is an indispensable companion for educators dedicated to providing transparent and fair assessment tools for their students.” (Magic School, 2025)

Learning to Read, Reading to Learn

Discussion about this post