Self-as-an-End
SAE Theory of the University · Paper 5

University Evaluation
大学的评价

Han Qin (秦汉) · Independent Researcher · 2026
DOI: 10.5281/zenodo.24293751 · Full PDF on Zenodo · CC BY 4.0
Abstract

Evaluation is everywhere in a university: the grade in a course, the conclusion of a degree, a recommendation letter, a disciplinary sanction, money given out on the strength of an evaluation. This paper begins with a guide the University of Melbourne gives its staff: a high score from AI writing detection does not, on its own, constitute grounds for an allegation of academic misconduct; the detector's result cannot be the only evidence an allegation relies on. Why can a number a machine reads out not simply be taken as the conclusion? This paper's answer: the judgment in an evaluation has to be completed by a person, who answers for it; to register a reading as a judgment is to turn judgment back into rules. The paper first sets down one qualification: the forms of grading are signals, not criteria. It then states seven propositions. Evaluation is power, and is subject to checks: what an advisor gives a student can be cultivation, and the evaluation in his hands is still power; conclusions such as degrees and qualifications are reached by positions separate from the advisor. Evaluation does not turn judgment back into rules. When evaluation is handed to a machine, the bearing is not handed over with it. Evaluation does not reward the step of treating people as means. A degree is a limited proof. The same holds for evaluating professors. Trust and checking can stand together, and checking does not go by identity. The paper ends with a dialogue with Campbell.

Writing Declaration: This paper was co-drafted with Claude (Anthropic). All intellectual decisions, framework design, and final editorial judgments were made by the author.

Introduction

The University of Melbourne's academic integrity website has a page for staff on how to use Turnitin's AI writing detection. It says: "A high AI detection score alone does not constitute grounds for making an allegation of academic misconduct." It also says: "To reiterate, the results produced by the AI writing detection tool cannot be the only piece of evidence relied upon in an academic misconduct allegation." Staff have to look for other evidence: "Staff may ask students to share drafts, notes or other evidence of authorship," and they may also talk with the student about his understanding of the course content.

The opening asks one question only: why can a number a machine reads out not simply be taken as the conclusion?

The guide's reason lies at the level of evidence: the reading may be wrong. What this paper has to say goes one step further. Even if the reading is accurate, the judgment in an evaluation has to be completed by a person, who answers for it; to register a reading as a judgment is to turn judgment back into rules.

The guide proves only one point. It is guidance for staff, about how staff use this tool; it is not a rule on the students' side. The page says that students cannot see the detection percentage and the highlighted report; it does not say whether students can respond to the evidence an allegation relies on. §9 returns to this.

This paper writes about evaluation: of students, of professors, of degrees. The paper has ten sections. §1 sets down a qualification; §2 says evaluation is power and is subject to checks; §3, that evaluation does not turn judgment back into rules; §4, evaluation handed to machines; §5, what evaluation does not reward; §6, how far a degree proves; §7 brings §2 down to several places of evaluation; §8, the evaluation of professors; §9, trust and checking; §10 is a dialogue. §7 states no proposition of its own; each of §2 to §9 otherwise states one at its end, and all are listed together at the end of the paper.


§1 Forms Are Signals, Not Criteria

This paper will use many examples of grading. One qualification is set down first, before the examples.

Institution Theory says: "Forms such as appraisal metrics, rankings and grades are signals, not criteria. The criterion is the boundary between limit and violation" (Institution Theory Z13). The same paper puts it more directly: "Looking at forms, one cannot judge whether the line was crossed." Abolishing a line of appraisal does not prove the line is no longer crossed; keeping it does not prove it is surely being crossed (Institution Theory, Paper 2, §6).

Applied to evaluation in universities. In MIT's first year, all subjects in the fall term and the January Independent Activities Period are graded only "pass" or "no record"; in the spring term grades are A, B, C or "no record," and the grade point average is not calculated until the end of the spring term. Brown's undergraduates can choose, for each course, one of two grading options: letter grades, or "satisfactory/no credit"; and "Brown does not compute students' grade point averages."

These are all forms. They alone cannot prove that a degree has made clear how far it proves. Conversely, a university that calculates grade point averages is not, on that ground alone, judged a misalignment. Not calculating grade point averages does not mean abolishing course evaluation, or abolishing all distinctions: Brown still gives letter grades.

What misalignment is read by, Prequel §4 has already said: what is registered. In the rule texts of the three places, degree requirements, course evaluation, and appointment and promotion, if a means is written as a sufficient condition of completion, and nowhere else are understanding, application, choice and revision asked about, that is misalignment in the text; how the rules are carried out is checked separately.

So the grading examples this paper gives are all read only as forms. They point a direction; the conclusion has to be read out of the structure.


§2 Evaluation Is Power, and Is Subject to Checks

The grade in a course, the conclusion of a degree, a recommendation letter: these evaluations, through the asymmetric effect of the position of the evaluator, can change the paths open to those affected. Disciplinary sanctions and funding decisions made on the basis of evaluation are decisions that adopt evaluations, and the same holds for them. A disciplinary sanction adopts an evaluation of a piece of conduct. Of a funding decision, the side that adopts evaluation is counted here; the side of how much, to which kind, and whether an entrance is left belongs to the division of resources, which Paper 6 writes. One and the same decision can have both sides.

"Power" here takes the broad sense of Power Theory, Paper 1: changing, limiting or directing the paths open to another; this definition does not presuppose the direction in which it is exercised. In the broad sense, to say that a position has power says nothing about what it gives. This use has to be kept apart from another. Cultivation Theory, Paper 5, said: "whether this giving relation has closed into power is seen only by whether the loop has closed, not by what anyone had in mind." There "power" is used narrowly, for what giving looks like once it has closed into a loop: once it has closed, it is no longer cultivation. This paper uses the broad sense, saying that a position can change the paths open to others, not that the loop has closed.

So two sentences can hold at once. What an advisor gives a student can be cultivation; the evaluation in his hands is still power: where the position is, the pressure is. Cultivation Theory, Paper 5, says: "A person cannot, by a sincere declaration, remove all the pressure a position brings; separating assessment from those who give conditions, providing a route of appeal, and providing a way to exit belong to the institutional side." Even if the advisor means well, his evaluation still decides the student's way forward; this layer is not removed by his meaning well.

Power itself is not the problem. The founding sentence of the Prequel of Political Economy is: "power itself is not the problem; how power is used is." So the university's setting of checks on the power of evaluation does not aim to reduce evaluation. Evaluation is the university's proper work: without it, students do not know how far they have got, and a degree cannot say what it proves. What has to be asked is how evaluation is done, and who can question it afterwards.

The strength of checks follows the actual effect. Political Economy gave a method of comparison, comparing power-faces by their particular effect: "whom it converges, down to which dimension, how many must pass through it, how great the cost of walking away" (Political Economy E14). Brought down to evaluation, these are three questions: which opportunities of whom it decides, how many have to pass through it, how great the cost of exit is. A quiz in an elective course and a qualifying examination that decides whether one can get a degree do not call for equally heavy procedures. Political Economy has another sentence: "At the level of relations depth is looked at; at the level of scale, breadth; small power can run deep." (Political Economy E7) An advisor takes only a few students; few pass through him, but the pressure runs deep, and the cost of walking away is great.

Checks come in several kinds. Persons separated: those who assess are not those who give conditions. Conclusions separated: review, second marking, committees, outside opinion. Stating grounds: those affected can see them, and can respond. Channels for objection and appeal meeting three conditions, in Institution Theory's wording: "it is independent of the party questioned, internal subjects can actually initiate it, and initiating it brings no retaliation" (Institution Theory Z34). Being able to exit. Several powers of decision not bundled in one position. And one more: the power of evaluation is not exchanged for control over questioning. In the words of Political Economy: "making a record or a judgment does not acquire the capacity to monopolize its final review or to cancel outside questioning" (Political Economy E22).

For conclusions such as degrees and qualifications, the separation of assessment from those who give conditions is adopted. Of the three things Cultivation Theory, Paper 5, handed to the institutional side, the first is separating assessment; for degrees and qualifications, this paper adopts it as it stands. Other evaluations do not take separation as their only form. That a course instructor both teaches and evaluates is the normal state, and this paper does not require every course to have a separate grader. But where there is no separation, the university has to explain how the checks it provides handle the pressure of position when giving conditions and assessment sit in one person. This explanation has two layers. In the relevant rules, the university explains which checks this kind of evaluation is given and how they handle this pressure of position; when a particular evaluation is challenged, the position that made the decision still has to explain how these arrangements apply in this matter. The former layer differs from the stating of grounds described above: stating grounds is each evaluation's explanation to those it affects. Cultivation Theory, Paper 5, listed those who hold positions, and teachers are among them; that this paper does not uniformly separate persons in courses is a choice made by power-face, and it is this series' own undertaking.

The arrangements that check it are themselves open to questioning. The Prequel of Political Economy says: "the arrangements that check it are themselves open to questioning." Separated examinations, second markers, external examiners and committees are all positions, and are all questioned in the same way: who can activate it, what actual efficacy it has, whether the party questioned can cancel it on its own (Political Economy E16). The Theory of Government, Paper 6, says: "Questioning does not need an unquestionable endpoint in order to stand." Arrangements of checks do not become an end point beyond questioning.

Which kinds of evaluation use which kinds of checks, and how strong they are, is at the rule layer; this paper marks only a few floors, in §7. The evaluations in the rest of this paper are all subject to checks under this proposition; but each of the propositions that follow has its own root and is not derived from this one: §5 writes what evaluation may not reward, §6 how far a degree record proves, and §3, §4 and §9 the lines of judgment and checking.

One objection will say: evaluation is professional judgment, and calling it power reads teachers and students as adversaries. This paper's answer: Power Theory's definition does not presuppose the direction in which power is exercised; power itself is not the problem. What an advisor gives can be cultivation, and the evaluation in his hands is still power; these two sentences do not cancel each other. To say that it is power is not to say that it is bad; it is to say that it needs a place where it can be questioned.

DX22 Evaluation is power, and is subject to checks. Evaluation within the university (grades in courses, conclusions on degrees and qualifications, recommendations), together with decisions that adopt evaluation (disciplinary sanctions; funding decisions made on the basis of evaluation), can, through the asymmetric effect of the corresponding positions, change the paths open to those affected, and so shows a power-face. What an advisor gives a student can be cultivation; the evaluation in his hands is still power: where the position is, the pressure is. The university ought to set checks on these powers, not with the aim of reducing evaluation; the strength of checks follows the actual effect: which opportunities of whom it decides, how many have to pass through it, how great the cost of exit is. Checks come in several kinds: persons separated; conclusions separated (review, second marking, committees, outside opinion); stating grounds, which those affected can see and respond to; channels for objection and appeal meeting three conditions, independent of the party questioned, actually open to initiation by internal subjects, and bringing no retaliation on those who initiate them; being able to exit; several powers of decision not bundled in one position; the power of evaluation not exchanged for control over questioning, a record or a judgment not acquiring the capacity to monopolize final review or to cancel outside questioning. For conclusions such as degrees and qualifications, the separation of assessment from those who give conditions is adopted; other evaluations do not take separation as their only form, but it has to be explained how the checks provided handle the pressure of position where giving conditions and assessment sit together. The arrangements that check it are themselves open to questioning, and do not become an end point beyond questioning. Beyond the floors marked in §7, which kinds of evaluation use which kinds of checks, and how strong they are, is at the rule layer. The evaluations throughout this paper are subject to checks under it; each of the propositions that follow has its own root and is not derived from this one.


§3 Evaluation Does Not Turn Judgment Back into Rules

Take a grading rubric for a writing course: clear thesis, three points; reliable sources, three points; citation according to convention, two points; complete structure, two points. A grader gives an essay eight points. The student asks: why were two points taken off under "reliable sources"? If the only answer is "graded according to the rubric," without saying which sources were the problem or on what basis it was judged so, the rubric has been used to excuse the grader from explaining the particular work. Conversely, if the grader can say how each item applied to this essay, and the student can raise an objection to it, grading according to the rubric does not in itself amount to substitution. The rubric is not wrong; it is material for evaluation; columns such as "clear thesis" and "reliable sources" themselves need judgment before they can be used.

The Theory of Disciplines says: "Standards do not judge in place of those present, and neither do seats." (Theory of Disciplines, S53) Extended to evaluation: the university can evaluate whether a particular piece of work passes; it may not register conformity with some metric as having replaced judgment of the particular work, nor use it to excuse the evaluating position from explanation and from being questioned. The grader has to be able to say why this essay got this score; if he cannot, and can only point to the rubric, he has let the rubric judge for him.

There is also a descriptive layer. When a metric becomes the only target, one that cannot be revised, it may induce behavior that conforms to the metric while departing from the work being evaluated: students learn to write the essay the rubric wants to see, teachers learn to teach the few things that will be tested. May, not surely; it has to be checked by actual effect. Political Economy said elsewhere a sentence that can be borrowed here: "A body's being responsible, in one accountability, for only one measure of performance is not the same thing as claiming that this measure is exempt from every boundary." (Political Economy E27) An examination board that deals only with marks is a division of labor; saying that once the marks are out nothing else need be asked is another matter.

This paper is not against metrics. Metrics can be used responsibly, and §8 cites normative texts of this kind.

The strongest opponent will say: metrics are objective, narrative evaluation is subjective and more prone to bias. The answer: what this proposition opposes is not metrics, but registering conformity with a metric as having replaced judgment; narrative records need standards too, and §6 has an example.

DX23 Evaluation does not turn judgment back into rules. The university may not register conformity with some metric as having replaced judgment of the particular work, nor use it to excuse the evaluating position from explanation and from being questioned. When a metric becomes the only target, one that cannot be revised, it may induce behavior that conforms to the metric while departing from the work being evaluated, and this has to be checked by actual effect. A body's being responsible, in one accountability, for only one measure of performance is not the same thing as claiming that this measure is exempt from every boundary.


§4 Evaluation Handed to Machines

The guide in the introduction comes back here. When a reading enters an evaluation, the question this paper asks is: who answers for it?

The Theory of Government says: "When judgment is handed to a machine, the bearing is not handed over with it." (Theory of Government G8) The bearing falls on the positions that hand the judgment over, decide how it is used, and act on its results. Brought down to the university: the position that adopts automated evaluation explains the basis of the evaluation and handles objections; it answers for the adoption and the conversion. The Theory of Disciplines named the points on which the adopting party answers: why it adopts, how it converts, what consequences it gives, whether there is review, whether it can withdraw (Theory of Disciplines, S34). A scoring machine reading essays, a detection tool reading out a percentage, a system tagging assignments: all are questioned on these points.

So "the machine read it out" is not taken to mean that the judgment has been completed. Paper 1 said two things (DX4): "he used AI" is not taken to mean there was no judgment, nor is "a person clicked confirm at the end" taken to mean the judgment has been completed. The latter points the same way as here: when a machine enters evaluation, the judgment still has to be completed by a person; a person pressing confirm at the end of the process cannot, on that alone, be taken to have completed the judgment. The former applies as well: even if the reading is accurate and the student really did use AI, that does not mean he exercised no judgment in producing the work.

The side of the Melbourne guide that points the same way as this proposition is that a reading is not the only evidence, and staff have to look for other evidence: they may ask students for drafts and notes, and may also talk with them. It does not carry the argument for this proposition: the root of this proposition is that sentence of the Theory of Government.

DX24 When evaluation is handed to a machine, the bearing is not handed over with it. The position that adopts automated evaluation explains the basis of the evaluation and handles objections; it answers for the adoption and the conversion. "The machine read it out" is not taken to mean that the judgment has been completed.


§5 Evaluation Does Not Reward the Step of Treating People as Means

Evaluation does not only say whether a person did well; it also tells everyone what this university rewards.

Take three situations. The record of contributions already states that the data were produced by B, and the evaluation still credits the whole of the work to A. A student carried a substantial part of the work in a project, and the evaluation, because he is a student, takes his share into his advisor's name. A group leader, threatening to hold back a member's opportunities elsewhere, forces him to take on work outside his accountability, and the evaluation records the extra numbers that result as good leadership. Each of these evaluations, in its own share, rewards the step already made clear. That step is a moment of treating people as means.

In a university whose consensus includes "a human being is an end," the consensus has to be visible in the rules, and evaluation is the place where it is most easily seen. So the university does not, through its own evaluation and allocation of resources, reward practices that treat people as means.

This proposition has its edge too. What the university can govern is its own evaluation and resources. Outside fame or private advantage a person gains by stepping on others, the university does not guarantee to eliminate; it guarantees only that its own evaluation adds no further share.

DX25 Evaluation does not reward the step of treating people as means. The university does not, through its own evaluation and allocation of resources, reward practices that treat people as means; it does not guarantee to eliminate every outside or private gain.


§6 A Degree Is a Limited Proof

A degree is a record. It states what work the university verified, on what materials and by what standards it made what evaluation, and what the scope and limits of the evaluation are. The university can form a limited judgment of a person's competence at particular tasks; it does not thereby vouch for that person's judgment in all future matters.

The Theory of Disciplines gave a way of keeping the two sides' accounts apart: when a standard is adopted by others as a gate, "the construct that writes it explains the parameters, scope of application, warrants and non-warrants it has itself written in; the two explain separately and are booked separately." (Theory of Disciplines, S34) Following this: the writing party is the university, which explains what the degree wrote in, what it warrants and what it does not; "it does not vouch for this person's judgment in all matters" is written in the column of non-warrants. The using party, an employer for instance, answers for its adoption. The facts of the work, the evaluator's judgment, and the force given to it by the degree or by later adoption are registered separately. Political Economy says: "a record must have a clear scope and a reviewable provenance" (Political Economy E22).

An accurate record will also be gamed. However precisely a degree is written, someone will assemble the few things it asks for; the mechanism of §3 applies to it just the same. This is registered as remainder.

When a degree is adopted as a condition of hiring and becomes a door those concerned cannot get around, it is a chokepoint. In the words of Political Economy: "When a degree or a certification becomes a door others cannot get around on the way to sustaining their lives or unfolding their own paths, it is a chokepoint" (Political Economy E31). Those the door reaches and who cannot get around it can question the door.

The Prequel cited a view (Spence 1973): the value of a degree lies in screening, not in what was learned. This paper does not deny that degrees are used this way. A degree can be used by others as a screening signal; this neither vouches for the holder's judgment in other matters, nor relieves the university of the duty to explain what exactly it verified, evaluated and recorded.

Several practices are offered only as suggestions: narrative evaluation; ungraded periods; not calculating grade point averages; learning records, written for instance as "completed a comparison of two approaches in a given project with given materials, and revised the conclusion after questioning; this record states the grounds of that piece of work and that evaluation, and does not extend any warrant to performance in other matters." These are all forms, read by §1. Hampshire College is known for narrative evaluation, and its academic advising page says: "Criteria for evaluation are established at the beginning of any learning activity." This example serves exactly as a counter-check: replacing grades with words still needs standards; a narrative record is not a way of getting rid of the measure, but a different way of writing the measure out.

This proposition deals only with the side of a degree as a record. The public function of professional degrees and certifications belongs to a separate dedicated paper.

DX26 A degree is a limited proof. The university explains what work it verified, on what materials and by what standards it made what evaluation, and what the scope and limits of the evaluation are; it can form a limited judgment of a person's competence at particular tasks, and does not thereby vouch for that person's judgment in all future matters. The two sides keep separate accounts: the writing party (the university) explains what it wrote in, its warrants and non-warrants, with "it does not vouch for judgment in all matters" written in the column of non-warrants; the using party answers for its adoption. The facts of the work, the evaluator's judgment, and the force given to it by the degree or by later adoption are registered separately. An accurate record will also be gamed, the mechanism of DX23 applies to it just the same, and this is registered as remainder. When it is adopted as a gate for hiring, it is a chokepoint; those the door reaches and who cannot get around it can question the door. A degree can be used by others as a screening signal; this neither vouches for the holder's judgment in other matters, nor relieves the university of the duty to explain what exactly it verified, evaluated and recorded. This proposition deals only with the side of a degree as a record.


§7 Checks on Several Places of Evaluation

§2 said that the strength of checks follows the actual effect. This section brings it down to several places. Five are marked as floors. The place of the advisor and the place of the advisor who also teaches the course are the letter of §2's proposition; the three places of course teaching, disciplinary sanctions and recommendation letters are marked by this section in accordance with that proposition, and are this series' own undertaking. Those marked as suggestions stay at the rule layer. This section states no proposition of its own.

The advisor and his own students (floor). Few pass through him, the pressure runs deep, and the cost of walking away is great: this is small and deep power. Conclusions such as degrees and qualifications are reached by assessing positions separate from the advisor. The advisor can submit an opinion; the student can see it and respond to it; assessment and review are not controlled by the original advisor alone. This is how Cultivation Theory, Paper 5's "separating assessment" lands on degrees and qualifications. Paper 1's proposition on advisors (DX3) said that several powers of decision are not bundled in the advisor's single position; here it goes on to say who reaches the conclusions on degrees and qualifications. In the constructed case of Paper 2, §6, the doctoral student refuses to put his name to the paper and applies to change advisors; his degree progress is read by this item: decided by assessing positions separate from the original advisor; a committee is one arrangement, not the only one.

The course instructor and a course (course grades entered on the transcript: floor). The power-face of an instructor's evaluation is checked by the actual effect of the course: whether it is required, whether another course can be taken instead, whether it affects degrees and qualifications, and the cost to students of objecting and of leaving. Where the effect is local and can be responded to and corrected, checks can be lighter; where it actually blocks later paths, they are not lightened because it is only one course. For course grades entered on the transcript, grounds are stated, and objections and review are taken up by a position other than the instructor: this is the floor. Everyday feedback and ungraded exercises are not within this floor. If the reviewing position alters an academic conclusion, it has to be a position with the corresponding competence that bears the relevant accountability (Paper 7, DX35); a review that examines only procedure does not alter the conclusion. For the part that counts toward the conclusion of a degree, second marking or outside opinion can be set up: this is a suggestion.

An advisor who also teaches the course (floor). The effects of the two places are read together. Within his accountability as instructor he can evaluate his own students; the pressure formed by the advising relationship is not shed by moving to the position of instructor. The power-face of his evaluation of his own students in this course is checked as at the place of the advisor, and review is not controlled by him alone.

Disciplinary sanctions (floor). In disciplinary procedures, the position that brings the allegation does not also hold the position that reaches the disciplinary conclusion. The position that brings the allegation has already formed a judgment to be submitted for handling; separating this step leaves room for independent checking and response. This is one case of "persons separated" in disciplinary matters, and it is this series' own undertaking. A zero an instructor gives for misconduct in a piece of coursework adopts an evaluation of conduct, and counts as a disciplinary sanction, not as grading: the instructor can bring it forward, and the conclusion is reached or confirmed by another position. The position after a sanction follows Paper 2's proposition (DX12): handled by event, with no identity fixed on the person.

Funding decisions made on the basis of evaluation (suggestion). Not through one place only; pointing the same way as Paper 2's suggestions (transitional funding not depending on the original advisor, recommendation letters from more than one source).

Recommendation letters (the writer's side and the university's own use: floor). A recommendation letter is a record about a student. When a hiring or admitting position uses it as a condition of a chokepoint decision, what shows a power-face, in Political Economy's wording, is "the position that makes this decision, and the positions that can use it to vouch or to seal off a path" (Political Economy E5): the hiring or admitting party that makes the decision, and the letter writer, who can use this record to vouch for the student or to seal off his path. The writer's own standing, credentials and fame are another thing, read separately; the Prequel of Political Economy says that fame and reputation are to be used apart: that the writer is famous does not mean the evaluation in the letter is reliable, and the weight of the letter lies in the grounds within it. The two sides, writing and using, are each questioned by their own effect. Whether to recommend, and how to evaluate, remains the writer's judgment by his accountability, and can be questioned; this does not give the applicant the power to rewrite the evaluation. The writing side answers for its grounds, for any vouching or sealing off formed through position, and for whether obedience elsewhere was made a condition of recommendation (Paper 2, DX9, DX10; Paper 3, DX18). The using side answers for its adoption, the weight given and the sources of alternatives. The academic evaluation in the letter, by the line of Paper 7's proposition (DX35), is not replaced. When the using side is outside the university, another university or an employer for instance, this paper questions it by the same reading, but it is not within this item: this item is addressed to the university, and what the university can govern is the writing side and the side of its own use of letters.

What falls outside this. Setting rules, dividing resources and checking are also power: by Power Theory, Paper 1's definition, they change the paths open to others. They are subject to checks on the grounds of the two sentences of the Prequel of Political Economy, power itself is not the problem and the arrangements that check it are themselves open to questioning; not through §2's proposition. This paper does not rename them evaluation. Setting rules belongs to Paper 7; dividing resources to Paper 6; checking is in §9.

For outside opinion at the level of degrees, there is a text to read. The External Examining Principles of the UK Standing Committee for Quality Assessment, August 2022, published by QAA: external examiners take part "as independent experts"; "They advise the institution on whether assessment regulations are followed in determining student marks." The same text also says these principles are "not mandatory under any of the UK regulatory systems." Both faces are there: at the level of degrees there is a place of outside independent opinion; it is not a mandatory requirement, and its particular powers depend on the institution's rules and the terms of appointment.

The strongest opponents come in two kinds. One says: with checks everywhere, evaluation will slow down, and teachers will not dare to say "no." The answer: checks follow the actual effect, not the length of a sentence. Everyday feedback that can be responded to and corrected and that does not tie into other decisions need not carry heavy procedures; if one sentence can block a degree, funding or exit, checks are provided by the effect it actually has. Political Economy says: "scale and power-face affect the form and strength of checks; they do not decide from what moment a construct must guard the line and be questioned." (Political Economy E14) The other says: a recommendation letter is a form of trust to begin with, and setting checks on it harms the trust. The answer: the item above does not require everyone to write letters for others, nor to write favorable evaluations; it requires only that the two sides, writing and using, can each be questioned on their own effect.


§8 The Same Holds for Evaluating Professors

The mechanisms of evaluating students and evaluating professors can be read side by side; the accountability and consequences involved in each remain separate.

Using journal metrics in place of judgment of a piece of research itself is the kind §3 spoke of. The general recommendation of the San Francisco Declaration on Research Assessment (DORA, 2012) is: "Do not use journal-based metrics, such as Journal Impact Factors, as a surrogate measure of the quality"; this sentence is not quoted in full, and its second half says that journal metrics are not to be used as a surrogate measure of the quality of individual research articles, to assess an individual scientist's contributions, or in hiring, promotion or funding decisions. The 2022 Agreement of the European Coalition for Advancing Research Assessment (CoARA) has as its second commitment: "Base research assessment primarily on qualitative evaluation for which peer review is central, supported by responsible use of quantitative indicators"; its third commitment speaks of abandoning inappropriate uses of journal- and publication-based metrics, and its fourth of avoiding the use of rankings of research organizations in research assessment.

These two texts prove only that such normative texts exist, not that any university's mechanisms run according to them. CoARA's second commitment also says that quantitative indicators are to be used responsibly: this paper is not against metrics, and neither are they.

The division of labor with Paper 3: there, how the university adopts and converts research standards (DX14); here, the mechanisms of evaluation. The interdisciplinary work Paper 4 handed over has one sentence here: the metrics of one discipline are not used in place of judgment of a piece of interdisciplinary work itself; who evaluates is written into the rules by Paper 4's proposition (DX20). The evaluation of professors is also power, and is subject to checks by §2. How tenure is evaluated is here; who sets it belongs to Paper 7.

DX27 The same holds for evaluating professors. Using journal metrics in place of judgment of a piece of research itself is the kind DX23 speaks of.


§9 Trust and Checking

Princeton set up its honor code in 1893, and in-class examinations had always been unproctored. On 11 May 2026 the faculty voted almost unanimously that from then on all in-class examinations would be proctored, effective 1 July.1 In August the university issued its "Requirements and Recommendations for Proctoring": "Proctoring is only observational"; even if a violation is suspected, the proctor "must not interrupt the exam"; details observed are put into a written record; "proctors must inform the instructor of record of suspected violations prior to contacting the Honor Committee"; examinations "must not be videorecorded or photographed"; and proctors are also to remind students to "write out and sign the Honor Pledge." The proctoring page on the university's scholarly integrity website says that the honor code itself has not changed, undergraduates still have to report violations they observe, and suspected violations in in-class examinations remain under the jurisdiction of the student-run Honor Committee.

What this case reads is: what was withdrawn, and what remains. What was withdrawn is one kind of trust, unproctored examination. By these two texts, what remains is the Honor Pledge and the student-run Honor Committee's jurisdiction over suspected violations; the proctor only observes and does not decide. How students defend themselves in the Honor Committee's procedures, these two texts do not say, and this paper has not checked.

The normative claim is here. A university whose consensus includes "a human being is an end" can entrust its examination rooms to an honor code; in mixed reality, adding checking, such as proctoring, does not thereby depart from the direction. Institution Theory says: "An institution does not presuppose that its internal subjects are all acting by their own law; its presupposition is mixed reality." (Institution Theory Z3) The same proposition also says: "An institution's identification of a person goes no further than the event and does not become an identity." So checking does not go by identity: the same for everyone, or by event. Princeton's proctoring is the same for all in-class examinations.

Checking departs from the direction in several situations: checking by identity, for example checking only a certain kind of student; the results of checking becoming an identity, so that one who has been checked once is treated as that kind of person; the rules of checking being beyond questioning; the reading of a tool or a proctor being registered as a ruling, with the student unable to respond to the evidence. The last brings back the page of the introduction. The Melbourne guide and this Princeton case are set side by side: one speaks of the evidentiary limits of a tool's reading, a reading not being the only evidence; the other of a change in the manner of checking, proctors only observing and the record being handed to another position to decide. The Melbourne page does not say that students have a right to respond to the evidence; the Princeton text says that proctors do not decide, and does not say how students respond when a decision is made, which belongs to the Honor Committee's own procedures. Both faces have to be read.

Checking is also power: by Power Theory, Paper 1's definition, it changes the paths open to those checked. It is subject to checks on the grounds of the two sentences of the Prequel of Political Economy, power itself is not the problem and the arrangements that check it are themselves open to questioning; not through §2's proposition. This paper does not rename it evaluation.

This proposition has its limits. Of this change at Princeton, this paper matches only two points: the same for all in-class examinations; proctors only observe and do not decide. Whether the results of checking become an identity, whether the rules can be questioned, and whether students can respond to the evidence have not been checked. So the phrase "the direction has not changed" reads only this one change of rules, and does not certify the whole university. The Princeton case serves here as an illustration, not as a ground: the root of this proposition is that proposition of Institution Theory.

The strongest opponent will say: the honor code is too naive. The answer: this proposition allows checking, and asks only that checking not go by identity and that readings not be taken as rulings. Others say from the other side: once proctoring is added, the honor code survives in name only. The answer: what was withdrawn is one kind of trust; what this proposition keeps is the student's standing as a subject, his defense, and his standing to question the rules. Whether it survives only in name depends on where the ruling falls, and on whether students can still respond.

DX28 Trust and checking. A university whose consensus includes "a human being is an end" can entrust its examination rooms to an honor code; in mixed reality, adding checking (such as proctoring) does not thereby depart from the direction. Checking does not go by identity: the same for everyone, or by event. What is withdrawn is one kind of trust, unproctored examination; what remains is the student's standing as a subject, his defense, and his standing to question the rules. Situations in which checking departs from the direction: checking by identity; the results of checking becoming an identity; the rules of checking being beyond questioning; the reading of a tool or a proctor being registered as a ruling, with the student unable to respond to the evidence. "The direction has not changed" reads only this one change of rules, and does not certify the whole university.


§10 Dialogue: Campbell

Campbell wrote an essay in 1976 on assessing the impact of planned social change, reprinted in 2011. Two sentences in it are often quoted: "The more any quantitative social indicator is used for social decisionmaking, the more subject it will be to corruption pressures" (p. 34). And: "When test scores become the goal of the teaching process, they both lose their value as indicators of educational status" (p. 35).2

Where they point the same way: an indicator used as a target loses part of its value as an indicator. §3's saying that a metric that becomes the only target, one that cannot be revised, may induce behavior that conforms to the metric while departing from the work being evaluated, points the same way.

Where they part is in the standing of the sentence. What Campbell put forward is that intensity of use and corruption pressure go together: an empirical tendency. This paper does not take on the general strength of this tendency. The normative sentence of §3 is not derived from it; its root is that sentence of the Theory of Disciplines, standards do not judge in place of those present. As for how much corruption follows from how intense a use, this paper judges only by particular effect, checking case by case.3


Where This Paper Stops

What it has not done: it has not said which grading system is best; it has not written about entrance examinations, which belong to a separate dedicated paper; it has not written about whether tenure ought to be kept or abolished; it has not said that metrics can never be used; it has not judged whether any university's grading system is right, nor whether any student cheated; it has not said that setting checks makes evaluation more accurate.


Notes

  1. The date of the vote, the count and the effective date follow the Princeton Alumni Weekly report of 12 May 2026; the "Requirements and Recommendations for Proctoring" and the proctoring page of the scholarly integrity website are the university's own texts. How these texts are carried out, this paper has not checked. ↩︎
  2. Neither sentence is quoted in full. The first goes on to say that the indicator will also be more apt to distort and corrupt the social processes it was meant to monitor; the second goes on to say that the scores also distort the educational process in undesirable ways. ↩︎
  3. Goodhart's law, often mentioned alongside Campbell, this paper has seen only in secondhand accounts, and does not cite. ↩︎

The Seven Propositions of This Paper

  • DX22 Evaluation is power, and is subject to checks. Evaluation within the university (grades in courses, conclusions on degrees and qualifications, recommendations), together with decisions that adopt evaluation (disciplinary sanctions; funding decisions made on the basis of evaluation), can, through the asymmetric effect of the corresponding positions, change the paths open to those affected, and so shows a power-face. What an advisor gives a student can be cultivation; the evaluation in his hands is still power: where the position is, the pressure is. The university ought to set checks on these powers, not with the aim of reducing evaluation; their strength follows the actual effect: which opportunities of whom it decides, how many have to pass through it, how great the cost of exit is. Checks come in several kinds: persons separated; conclusions separated; stating grounds, which those affected can see and respond to; channels for objection and appeal meeting three conditions; being able to exit; several powers of decision not bundled in one position; the power of evaluation not exchanged for control over questioning. For conclusions such as degrees and qualifications, the separation of assessment from those who give conditions is adopted; other evaluations do not take separation as their only form, but it has to be explained how the checks provided handle the pressure of position. The arrangements that check it are themselves open to questioning. Beyond the floors marked in §7, it is at the rule layer. The evaluations throughout this paper are subject to checks under it; each of the propositions that follow has its own root and is not derived from this one. (First sentence: conceptual distinction, taking over the broad sense of T1 in Power Theory, Paper 1, which speaks of power in general, extended to evaluation, this series' own undertaking; used apart from the "power loop" of F28 and F29 in the Special Study on Rights, quoted via Cultivation Theory, Paper 5, §1. Second sentence: taking over Cultivation Theory, Paper 5, HY29. From the third sentence on: normative, newly set down by this series; taking over the founding sentence of the Prequel of Political Economy and "the arrangements that check it are themselves open to questioning," together with E14, E7, E16 and E22, which speak of economic positions and accumulation, extended to evaluation, this series' own undertaking; the three conditions follow Institution Theory Z34; not becoming an end point beyond questioning follows Theory of Government, Paper 6. Separation for conclusions such as degrees and qualifications takes over HY29; that other evaluations do not take separation as their only form is this series' choice by power-face, this series' own undertaking)
  • DX23 Evaluation does not turn judgment back into rules. The university may not register conformity with some metric as having replaced judgment of the particular work, nor use it to excuse the evaluating position from explanation and from being questioned. When a metric becomes the only target, one that cannot be revised, it may induce behavior that conforms to the metric while departing from the work being evaluated, and this has to be checked by actual effect. A body's being responsible, in one accountability, for only one measure of performance is not the same thing as claiming that this measure is exempt from every boundary. (The normative sentence: extending Theory of Disciplines S53 to evaluation, this series' own undertaking. The descriptive sentence: pointing the same way as Campbell, without taking on the general strength of his tendency; checking by actual effect takes over the reading of Political Economy E27)
  • DX24 When evaluation is handed to a machine, the bearing is not handed over with it. The position that adopts automated evaluation explains the basis of the evaluation and handles objections; it answers for the adoption and the conversion. "The machine read it out" is not taken to mean that the judgment has been completed. (Normative; the first two sentences extend Theory of Government G8, which speaks of government's judgment, and Theory of Disciplines S34 to evaluation in the university, this series' own undertaking; the last sentence follows DX23 and Theory of Disciplines S53, consistent with Paper 1, DX4)
  • DX25 Evaluation does not reward the step of treating people as means. The university does not, through its own evaluation and allocation of resources, reward practices that treat people as means; it does not guarantee to eliminate every outside or private gain. (Normative; this series' own undertaking)
  • DX26 A degree is a limited proof. The university explains what work it verified, on what materials and by what standards it made what evaluation, and the scope and limits of the evaluation; it can form a limited judgment of a person's competence at particular tasks, and does not thereby vouch for that person's judgment in all future matters. The two sides keep separate accounts; the facts of the work, the evaluator's judgment, and the force given to it by the degree or by later adoption are registered separately. An accurate record will also be gamed, and this is registered as remainder. When it is adopted as a gate for hiring it is a chokepoint, and those the door reaches and who cannot get around it can question the door. Serving as a screening signal does not relieve the university of the duty to explain. This proposition deals only with the side of a degree as a record. (Normative; taking over Theory of Disciplines S34 and Political Economy E31, E22)
  • DX27 The same holds for evaluating professors. Using journal metrics in place of judgment of a piece of research itself is the kind DX23 speaks of. (Normative: DX23 applied to the evaluation of professors. DORA and CoARA prove only that such normative texts exist, not that the mechanisms are running; CoARA's second commitment also requires the responsible use of quantitative indicators)
  • DX28 Trust and checking. A university whose consensus includes "a human being is an end" can entrust its examination rooms to an honor code; in mixed reality, adding checking does not thereby depart from the direction. Checking does not go by identity: the same for everyone, or by event. What is withdrawn is one kind of trust, unproctored examination; what remains is the student's standing as a subject, his defense, and his standing to question the rules. Situations in which checking departs from the direction: checking by identity; the results of checking becoming an identity; the rules of checking being beyond questioning; a reading being registered as a ruling, with the student unable to respond to the evidence. "The direction has not changed" reads only this one change of rules, and does not certify the whole university. (Normative; rooted in Institution Theory Z3 and Z34; Theory of Disciplines S53 and Theory of Government G8 point the same way; the Princeton case serves as an illustration, not as a ground)

References

All SAE works below are cited by concept DOI, which always resolves to the latest version.

I · This system

  • Qin, H. (2026). What a University Is Not (SAE Theory of the University, Prequel) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24293733
  • Qin, H. (2026). University Teaching (SAE Theory of the University, Paper 1) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24293737
  • Qin, H. (2026). University Students (SAE Theory of the University, Paper 2) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24293740
  • Qin, H. (2026). University Research (SAE Theory of the University, Paper 3) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24293742
  • Qin, H. (2026). University Disciplines (SAE Theory of the University, Paper 4) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24293746
  • Qin, H. (2026). University Governance (SAE Theory of the University, Paper 7) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24293757
  • Qin, H. (2026). On the Origin of Power · Shi (SAE Power Theory, Paper 1) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.20370224
  • Qin, H. (2026). When One Holds Power (SAE Cultivation Theory, Paper 5) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24154515
  • Qin, H. (2026). To Whom the Ought Is Addressed (SAE Institution Theory, Paper 1) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23269635
  • Qin, H. (2026). Not Crossing the Line (SAE Institution Theory, Paper 2) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23270384
  • Qin, H. (2026). Open to Questioning (SAE Institution Theory, Paper 6) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23271324
  • Qin, H. (2026). Judgment (SAE Theory of Government, Paper 1) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23412130
  • Qin, H. (2026). Questioning and Bearing (SAE Theory of Government, Paper 6) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23413511
  • Qin, H. (2026). What an Economy Is Not (SAE Political Economy, Prequel) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23874133
  • Qin, H. (2026). Accumulation (SAE Political Economy, Paper 1) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23874138
  • Qin, H. (2026). Chokepoints (SAE Political Economy, Paper 2) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23874140
  • Qin, H. (2026). Checks (SAE Political Economy, Paper 3) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23874142
  • Qin, H. (2026). Conversion (SAE Political Economy, Paper 4) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23874146
  • Qin, H. (2026). Growth (SAE Political Economy, Paper 5) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23874150
  • Qin, H. (2026). Withdrawal (SAE Theory of Disciplines, Paper 6) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24177783
  • Qin, H. (2026). Non-Substitution (SAE Theory of Disciplines, Paper 9) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24177799

II · Rule texts and institutional documents (original language)

  • University of Melbourne, Academic Integrity. "Turnitin's AI writing detection tool." https://academicintegrity.unimelb.edu.au/staff-resources/turnitins-ai-writing-detection-tool
  • Massachusetts Institute of Technology, Registrar's Office. "First-Year Grading." https://registrar.mit.edu/classes-grades-evaluations/grades/grading-policies/first-year-grading
  • Brown University, The College. "Grade Options." https://college.brown.edu/design-your-education/explore-open-curriculum/course-selection/grade-options
  • Hampshire College, Center for Academic Support and Advising. https://www.hampshire.edu/academics/advising-and-academic-resources/center-academic-support-and-advising-casa/advising-0
  • UK Standing Committee for Quality Assessment. External Examining Principles. Published by QAA, August 2022. https://www.qaa.ac.uk/docs/qaa/quality-code/external-examining-principles.pdf
  • San Francisco Declaration on Research Assessment (DORA). 2012. https://sfdora.org/read/
  • Coalition for Advancing Research Assessment (CoARA). Agreement on Reforming Research Assessment. 2022. https://coara.org/agreement/the-commitments/
  • Princeton University, Scholarly Integrity. "Requirements and Recommendations for Proctoring." August 2026. https://scholarlyintegrity.princeton.edu/document/186
  • Princeton University, Scholarly Integrity. "Proctoring." https://scholarlyintegrity.princeton.edu/proctoring

III · External literature

  • Campbell, D. T. (1976). Assessing the impact of planned social change. Occasional Paper, Public Affairs Center, Dartmouth College. Reprinted in Journal of MultiDisciplinary Evaluation, 7(15), February 2011. https://jmde.com/index.php/jmde_1/article/download/297/292/988
  • After 133 years, Princeton is going back to proctoring exams. (2026, May 12). Princeton Alumni Weekly. https://paw.princeton.edu/article/after-133-years-princeton-going-back-proctoring-exams
  • Spence, M. (1973). Job market signaling. Quarterly Journal of Economics, 87(3), 355–374.

Interfaces

This series: DX22 to DX28.

Handed to Paper 6: all these arrangements need money; dividing resources is also power, and equally subject to checks.

Left for Paper 7: who sets the rules of evaluation; the accountability of academic judgment and its being questioned; the position of appeal; the division of labor between outside opinion and internal rulings.