More interested than interesting. — the diagnostic
A homogenizing editorial process has arrived in the dorm room. It produces structurally flawless prose with no fingerprint. This paper is a proposal — not a syllabus — for what a freshman writing course can be when the baseline of a B-level essay can now be generated in three seconds.
Paul interviewed David for forty-five minutes under one rule: no talk of AI. The point was to surface what college writing is for before naming what threatens it. What follows is the harvest of that conversation, reorganized as a working framework David can carry into the fall semester. It is also a tool: at the bottom of this paper there is a self-evaluation he can run against his own syllabus before the first class meets.
Late in the interview, David named the thing he had been circling for an hour. He said the kind of writing he is teaching against is not sloppy. It is slop. A sloppy paragraph is unfocused, ungrammatical, run-on — but it still contains the potential for an exploratory relationship to language. A piece of slop is the opposite. It is the Fresh & Co. bowl: protein in this quadrant, carbohydrate in that one, vegetable plated for legibility. It is structurally flawless and entirely homogenized.
This is the distinction the rest of the framework rests on. The threat to college writing is not that students will turn in messier essays. It is that they will turn in cleaner ones — essays that have been through a homogenizing editorial process so thorough that no fingerprint of a human consciousness remains. The pedagogical instinct to "correct sloppiness" used to be in service of pulling a student's voice into legibility. In an AI baseline, that same instinct pulls the student's voice into slop.
The course has to teach the difference between the two.
Messy, ungrammatical, run-on, unfocused. Contains the potential for exploration. Contains a fingerprint. Worth editing.
Structurally flawless. No fingerprint. Confident in its conclusions. Indistinguishable from the next one. Worth interrogating, not editing.
The student's task is no longer to finish. The baseline finish is now free. The student's task is to be responsive — to put something into the bowl that does not belong with the other ingredients.
Underneath the slop/sloppy distinction there is a more useful diagnostic — one a student can run against their own draft at two in the morning on a Sunday. Am I trying to be interesting, or am I interested? Slop is what you produce when the ratio flips. Voice emerges when a writer is more interested than interesting. A synthesis engine is the perfect engine of the inverted ratio: maximum interestingness, zero interestedness. The dangerous surplus is what is left over when the writer stops caring about appearing smart and starts caring about the thing.
The fingerprint is not style. Style is something you can describe — a sentence rhythm, a vocabulary, a tendency. The fingerprint is the specific mistake the writer made that they did not yet know was a mistake. It turns out, in retrospect, to have been them. A synthesis engine cannot make those mistakes because it cannot not-know. It is incapable of writing past the edge of its own awareness, which is precisely the place where a human writer leaves a fingerprint.
The course's job, then, is to protect the student's right to be wrong on the page in a way that retrospectively turns out to have been the most particular thing they did. An assignment that penalizes wrongness in service of cleanness is an assignment that prefers slop. The rubric in the next section has to permit a kind of error the writer hasn't named yet.
There is a sentence that names the whole frame: it is not what you look like when you are doing what you are doing — it is what you are doing when you are doing what you look like you are doing. Slop is the first one. Voice is the second.
The course can grade the second thing without conscripting the professor into AI-detection, because what is being graded is no longer a surface to be authenticated. It is the activity underneath the surface. The defense, the dialogue, the friction journal, the audit — every assignment in Section V is in service of grading what the student is doing under the performance, rather than authenticating the performance itself.
Paul asked David: if a parent paying eighty thousand dollars a year asked you in one honest sentence what their child is actually buying in your class, what is the sentence?
David's answer: critical thought and careful inquiry. The ability to organize that thought. The ability to present it in an intelligible way. The element of finishing — a public-facing version of the work, presented to peers, defended in real time. A space, in other words, in which students learn how to produce knowledge as much as a space in which they learn to express themselves.
This is the rubric that already exists in David's head. It has not changed since 2005. What has changed is that some of these criteria are now trivially gameable from outside the room — and others are emphatically not. The framework below separates them.
In every category, the question is: can a synthesis engine fake this from outside the room? If yes, the assignment around it has to be redesigned.
Asked for the smallest version of this course he could run starting day one — no committee, no budget, no permission — David said the strictest analog form. Here are your notebooks. You can only write in these notebooks. All your writing will be evaluated in these notebooks, and you will only write when you are in this class. Reading quizzes to force intimacy with the text. Printed handouts. Phones away.
And then he said, in the same breath, that this would be a travesty.
It is a travesty because the most important hours in college writing are not in the room. They are the all-nighters in the library, the dorm-room solitude, the smoking-Marlboro-lights-out-the-window hours of intense, dialogical struggle with a text. The proctored notebook preserves the writing process by throwing away its most important element — the solitary one.
It is also a travesty in a second, quieter way. It conscripts the professor into the role of policeman. It burns the three-hour class block on quizzes and supervision that David already names as a compromise. It reduces the time available for the only thing he loves teaching, which is the live "swept-up" moment where students adapt a framework and produce knowledge in front of each other.
The retreat works. It also costs the course its reason for existing.
"It almost feels like preserving the writing process, but throwing away one of its most important elements — this solitary experience."
"As soon as my own internal AI-detection device goes off, there's a part of me that also loses attention."
"It's clear that there's a lot of thought that goes into it. I still don't end up giving A's — because if it sounds like AI, it's still not of the quality we're looking for."
"The most exciting moments are when students take a concept, adapt it in the context of the way they're seeing the world, and do it with some eloquence."
The pivot is one move, stated once, in one sentence: stop treating AI as contraband and start treating it as a text.
The course David already teaches has a unit on the Scholes power-pleasure matrix. It is the framework for analyzing how a Budweiser commercial can affirm a meritocratic vision of America while everyone in the room knows the underlying material reality is more complicated. The Nike commercial of 2019, deploying LeBron and Serena and Sam Cooke, is the perfect demonstration: you cannot resist its pleasure even when you can name its ideology.
An AI essay is the same artifact. It is a corporate, mass-produced cultural text that arrives with a homogenizing editorial process baked in. Students are getting swept up in the frictionless ease and magic of it, ignoring the material reality of where it came from — the scraped training data, the energy cost, the homogenization of thought. The course already has the tools to teach students to resist exactly this kind of artifact. It just has to point those tools at a new object.
This reframe does three things at once. It validates the analog classroom as the sacred space where the "dangerous surplus" is built. It eliminates the policing posture, because the AI is no longer hiding under the desk — it is on the projector. And it converts the threat into the most important text the syllabus will study.
Redline note · the strategic moveThe professor does not have to defend writing from AI. The professor has to do what the professor already does: teach students to read a culturally dominant artifact against its own grain. AI is the next assigned text. It is not the enemy of the syllabus; it is on the syllabus.
Each of the assignments below is designed to be assigned unmonitored — the student can use any tool, any time, anywhere — because the artifact the student is producing is one a synthesis engine cannot produce. The course reclaims the dorm-room hours by giving students something to do in them that the machine cannot pre-empt.
Students are given the assigned text — a Scholes essay, the David Foster Wallace piece on Lynch, a chapter of Debord — and told to prompt an AI to summarize it. They submit the AI's summary, marked up by hand.
The graded deliverable is the markup, not the summary. They are looking for the ideological subtexts, the nuances, the textures of argument that the machine flattened. Intimacy with the source is achieved by being made to correct the machine.
Close reading. Granularity. The "attunement and responsiveness" David wants. Returns the three-hour block to live discussion because intimacy was built at home.
UnmonitoredSubmit on paperWeeklyEvery essay is submitted with a journal alongside it. The journal documents the moments the student fed their working thesis into an AI and asked it to argue the counterpoint — and then specifies, in prose, exactly where the machine's response collapsed into expected, homogenized argument and what the student had to assert against it.
The solitary, late-night struggle. Reframes the struggle as against the algorithm's tendency toward B-level argument rather than against the blank page. Produces visible evidence of the fingerprint.
UnmonitoredSubmit with essayEvery essayThe essay is the entry ticket. The grade lives in a fifteen-minute peer-defense session. The student is asked, in front of their peers, to adapt the framework they used in the essay to a new, on-the-spot example chosen from contemporary media. If they cannot adapt it live, the framework is not theirs — regardless of the prose on the page.
David's "finishing" criterion. The socialization element. The "swept-up" moment, but inverted: now the student has to produce it, not just recognize it. Forces the room to do what only the room can do.
In-classReplaces 30% of gradePer essayEach student writes an exploratory paragraph by hand about a piece of media — embracing genuine sloppiness, not aiming at a finished thought. They then feed that paragraph into an AI with one prompt: "Make this sound professional and academic." They read both versions out loud.
The class names, line by line, what was lost in the homogenization. They are reading the difference between sloppy and slop directly off the page.
It is a single in-class demonstration that does more pedagogical work than a semester of AI policy. The students see, in their own voice, what the homogenizing editorial process strips out. They name the surplus themselves.
In-classWeek 220 minTwice a month, the class is given a prompt. They have ten minutes to write a human response by hand. The same prompt is run, live, through an LLM on the projector. The final ten minutes are spent ripping both apart against whatever framework is current in the syllabus — Scholes, Debord, the surface-symptomatic split.
Gives David a structured way to introduce AI to the analog classroom on his own terms, as an object of study rather than a tool of evasion. The students get the analog free-write he already loves; they then get the thrill of dismantling the machine version against it.
In-classBiweekly15 minThe syllabus does not have to ban AI. It does not have to require AI. It states: any tool, any time. The artifact you owe me is one the tool cannot produce. The grading rubric carries the policy, not a paragraph of warnings.
David — this is for you. Eight questions about the syllabus as it currently stands. The point is not to score yourself well; it is to surface which corners of the course are still being defended against AI rather than re-built around the surplus. Answer honestly and the diagnosis at the bottom will tell you, plainly, which of the five assignments above is the highest-leverage one to add first.
The closing exercise of Asking Better Questions Session 01 asks every participant to leave with one reframed question — the assumption version of something they have been avoiding. This is yours. The information question is what does my AI policy say. The meaning question is what is my course actually for now. The assumption question is below. Write a draft of it here. We'll talk through it before the first class.
— Paul & Jesse
"Let me show you a Nike commercial from 2019 — LeBron James, Serena Williams, Alicia Keys, callbacks to Sam Cooke, shot in black and white on the streets of Chicago. And all of a sudden they're like, oh wow, yeah, I am totally swept up in this. I cannot resist the pleasure of this text — even if I'm well aware that Nike has a history of using child laborers to sew together their soccer balls in Indonesia."David Markus, in the interview, on the Scholes power-pleasure matrix in action
This is the moment to hold onto. It is the entire course distilled into one classroom anecdote: a student knows the ideological situation, watches the artifact, gets swept up anyway, and only then has the framework to name what is happening to them.
The proposal in this paper is that the next artifact in that pedagogical sequence is no longer a Nike commercial. It is the AI essay. The student is already getting swept up in it. They are already producing it. They have not yet been given the framework to name what is happening to them when they do.
That is the framework. That is the course.
At the end of the interview, asked one more time what the class is actually for, David arrived at the sentence that does what the eighty-thousand-dollar-a-year question was trying to get at. It is the sentence to put at the top of the syllabus. It is the sentence to put at the top of the rubric. It is the sentence every assignment in this paper is in service of.
David went back to the sentence and added two things that change everything. First, peer-to-peer dialogue — the room itself is now named as a graded activity, not a venue for it. Second, the verbal and written expression of the singular manner in which each of you engage. Singular is the word doing the work. He did not say "your expression of the engagement"; he said the singular manner, which is the fingerprint stated as a course objective. Together the two revisions move the rubric into the room and explicitly into the body of each student. They also reject, in their grammar, the kind of enforcement mechanism his colleagues have started using — oral exams to verify that a writer is the author of their own thinking. David is naming that he doesn't want that. He wants the singularity to be expressed both ways, verbally and in writing, because the course is teaching them to do both — not building a tribunal to catch them when one fails.
This is now the sentence at the top of the syllabus. The rubric measures four things: close reading, critical thinking, peer-to-peer dialogue, and the singular expression — verbal and written — of how the student does the first three. Every assignment in Section V can be checked against this rubric. Each one preserves at least three of the four.