#4
Inspired by
The Future of Grading as Communication
Original abstract
Grades invite invalid and problematic inferences as they circulate across institutional contexts, yet proposed alternatives often address either educational quality or institutional legibility, rarely both, raising the question: what minimal information must a grade communicate to support interpretation across contexts and stakeholders?
This theoretical exploration yields a framework treating grades as structured institutional messages rather than scalar measurements. Through first-principles analytic reasoning about grading as observation and evaluation under conditions of learning, the framework derives a minimal grammar composed of Inputs (actions and activities), Outputs (performances or products), and Difficulty (learning conditions shaping interpretation). These distinctions are proposed as the smallest set required to prevent common inferential failures in grading systems.
Predominant theories serve as analytic and objective constraints. Argument-based validity and evidence-centered design frame the necessity of Inputs and Difficulty as distinct components, while opportunity-to-learn and cognitive residue motivate Difficulty's conceptual decomposition into mechanical, normative, and epistemic components. The framework is evaluated through inferential stress-testing against familiar academic practices, such as GPAs and narrative evaluation, examining which interpretations they enable, obscure, or distort.
While this framework does not attempt to improve measurement accuracy or directly resolve equity concerns, we illustrate how it can improve downstream institutional inferences and outcomes (e.g., AI-mediated vs. constrained performance contexts)--and how it enables inter-operation of disparate assessment practices.
As such, the framework offers a portable analytic lens for comparing grading paradigms, redesigning grades for clearer institutional signaling, and evaluating students under conditions of rapid technological, political, and economic change.
In simpler words, I propose a grading system, consisting of observational records, which can be seen as an evolution of ungrading or evidence-centered design. In this system, the instructor records factual observations about the Inputs, Outputs, and Circumstances of each student, and these observations are what is shared and communicated, instead of a grade.
What Inputs and Outputs entail is within our zeitgeist, so I’ll briefly expand on Circumstances, which are rapidly changing. Circumstances are facts that may shape the relationship between inputs and outputs. Think time limits, available vs. allowed tools, prior preparation, illness, school resources, and political, economic, or social conditions. They influence what actions and outputs could be possible or rational.
The framework is simple. What did students do? What did they produce? Under what circumstances? On Day 1 of the conference, Professor Wade Newhouse from William Peace University described grading as a system of expectations, and naturally, different people and organizations have different expectations, based on their needs and insights. Based on observational records, they can derive grades (and other interpretations) according to their standards.
The underlying theory is supported by a mixture of economic analysis, inversion of control, and argument-based validity and tempered against analysis of grading reform and its history. True to its conceptual roots in software development, it’s designed to be backwards and forwards compatible with any grading pedagogy.
But of course, change is unsettling. After Day 1, there remains an open question of how instructors practicing alternative grading should help students understand what their grade is. As Dr. Stommel has expressed, the goal is to preclude that need entirely, but as Dr. Streifer said in her session, that can’t happen overnight. It won’t work to lie to students that their grade doesn’t matter when, systemically, it does and the optimization incentive remains in place.
Theoretically, as I’ll outline shortly, the system of observational records can lead to an improved incentive structure, both for students and throughout the institutional system. It can improve academic portability between institutions, such as during transfers or immigration, and it can simplify operations for instructors, by removing the obligation to grade or report a grade.
Adoption could work in four phases, each adding additional stakeholders to dilute the pool of possible interpretations. The hypothesis is that there is a critical threshold of diversity at which it is suboptimal for students and their associated actors to optimize for the interpretation of an instructor or institution. I’d love help figuring that out, but I digress. Let’s talk about adoption. It should start with higher ed because no one’s going to fight a million different school boards. Consider the following heuristic plan.
Phase 1: Intra-institution.
Operate in parallel, invisible from students. There is no need to tell them; there won’t be for quite some time. Record observations, share them with colleagues, and discuss interpretations. You needn’t agree; in fact, you’re looking for the dimensions of disagreement and the justifications. Disagreement is good, compromise unnecessary.
Optionally, before the beginning of the next phase, to maintain compatibility, establish an institutional policy for synthesizing transcript grades (potentially AI-assisted). There is room for creativity. For instance, if Harvard wishes to change their grading policy every few years, their administration may prefer to maintain a distinct, very standardized interpretation instead of relying on their faculty’s judgement. I jest, but such a policy could be appropriate for some institutions. Remember, this isn’t a critical policy; its systemic weight will decline in subsequent phases, as standardized interpretation becomes less preferable.
Phase 2: Inter-institutions, for transfers and admission
Again, defer exposing the system to students. Instead, offer to share observational records with other schools instead of, or in addition to, traditional transcripts. If your counterparty is in a different phase, they’ll benefit from the optional synthesis in Phase 1.
From a myopic theoretical perspective, it could seem optimal for institutions not to share how interpretations are derived, but I have zero expectation of that happening. Sharing is natural. So is disagreeing and forming your own opinions. Why fight it?
Phase 3: Inter-system, with employers and accreditors
Now, offer the option of observational records to employers, applicant tracking system vendors, and recruiting firms. Optionally, institutions can provide a self-service application programming interface for querying observational records to facilitate adoption and software integrations (and raise revenue).
Phase 4: Students
At last, after Phase 2 and 3 succeed, we can tell students about the new paradigm, but there’s a twist: they likely already know. We didn’t have to tell them. Instructors may now freely use any pedagogical practice or grading method they wish. Experimentation and research won’t complicate operations or compliance.
The transitional generations of students might experience panoptic pressure, where they’re confused about the parameters and bounds of their evaluation. Perhaps this presents a learning opportunity and an analog about the interpretation of life and its value. There are an overwhelming number of possibilities, and one cannot pursue them all nor satisfy all judges. The simplest option is to be yourself and live according to what you value. Someone will find value in that, even in things ignored and penalized by the status quo.
If confusion remains, an illustrative train of thought could be, “what grade would you give your life so far?” Consider who would disagree.
This system will not be without issues. Privacy, eventually, will be a major concern, especially as the barrier to a comprehensive surveillance state continues to lower, and the economic incentives—the very increase in economic utility that facilitates adoption—likely deprioritize civic concerns. Cynically, and because I do not yet see a deterministic path to resolution, I think about the idea of marginal privacy. I think about the ethical responsibility of our institution and system. What is our scope? What are our priorities? The student, the society, the state, the civilization, humanity? At what point does virtue become unethical?
I do not know. I only know that the problems before us are heavy and growing. We must try to fix them. This, I offer as our first step.