Reloading Writing Instruction: An Answer to Cognitive Offloading

Drafting a text presents a complex set of problems word by phrase as the pen moves through time. A writer works at the intersection of two planes rotating in existential space, sometimes slowly, ponderously, trepidatiously, sometimes at warp speed, subject to microcosmic or macroscopic recalibration as each added word commits the text to particular parameters while foreclosing others.

The horizontal plane is the space for the draftsperson and the text to have reciprocal influences on generation. Empty space gradually becomes filled with words and other icons or figures. The vertical plane is the space for the writer to weigh the values or impacts of particular words before the text goes into publication. A drafter working at a location in horizontal space can very quickly pause, take a position in vertical space, consider where the text has been and where it is going, and decide which word to choose.

This essay argues that development of expertise as an academic writer involves the accumulation of insights from purposeful, situated, embedded experiences in this intersection, often in collaboration with mentors and peers. As I hope to demonstrate, expert writers choose their words very, very carefully—which is not the same as slowly—and bounce around like gymnasts from a location in horizontal space to a location in vertical space and then back to a location in horizontal space which may or may not be at the cutting edge of the emerging text.

Pragmatic language production and meta-rhetorical memory intersect in the drafting mind in ways that are poorly expressed by the words ‘process’ and ‘product.’ Prewriting, outlining, drafting, reviewing, revising, editing take place not linearly but are, in the case of the expert, accomplished simultaneously at the intersection of the vertical and horizontal planes.

Commitment

But what does it mean to say that during drafting a word commits a text? Perhaps a better word is ‘promises.’ Semantically, a new word promises that it will resonate in a harmony with prior words and support upcoming choices. Icicle commits me to cold weather, not so much to banana bread. Syntactically, the new word opens certain grammatical paths while closing others. A bell can ring; it can’t automobile. Tonally, it establishes register that must be honored or deliberately violated. I’ll take a pass so as not to piss anyone off.

Logically, word upon word stakes claims that constrain future argument. I’m choosing words partly to limit opportunities for you to misconstrue my text, that is, to lead you down a channel of thought you willingly follow without raising arguments—not yet, at least. I am hyper aware—on the vertical plane—of my choice of the word “drafting” to open this text. I didn’t say “writing.”

My choice of ‘drafting’ as opposed to the more commonly chosen word ‘writing’ committed this text, at least for a period, to offer arguments about writers during acts of original text production, not revision, not editing—the moment of fusion between thought and language in Vygotskian terms, in three-cueing terms. I want to refer to this activity as loading a text with meaning; later, I want to refer to the act of reading as unloading and reloading meaning through transaction with a text.

One key difference between LLM and human writers resides in our understanding of the verb ‘commit.’ What binds the drafter is a commitment to the text. Emerging semantic circles deserve respect when choosing words; coherence and resonance must be made in and sustained by chosen words. LLMs are bound absolutely to recalibrate the text based strictly on prior text, what we might call pretext.

Human writers commit; LLMs comply. Commitment implies stakes, risk, the possibility of being wrong, misleading, vague, and bearing the consequences. The writer who commits to "drafting" a word accepts responsibility for that choice across the text, internally and externally. I bear responsibility for choosing pissed off. The LLM merely complies with statistical weights inherited from training. Compliance carries no stakes because there is no agent who could fail, no reputation risked, no relationship with readers that could be damaged by incoherence or infelicity.

Human commitment to a word or phrase in the heat of drafting extends to meanings not yet formulated. On the vertical plane choosing a word impacts arguments still taking shape in the murky prefigurative space Vygotsky described. LLMs commit only to pretext—to what has already been written. They cannot commit to what remains unwritten because they have no capacity to hold the unwritten as a constraining presence.

Human writers draft across a horizontal plane of diminishing white space. Each word narrows macroscopic possibility while recalibrating microcosmic impact. The vertical plane holds accumulated commitments—promises made to the emerging text and to readers not yet present. To commit is to stake something on coherence not yet achieved, to promise that the narrowing corridor leads somewhere worth arriving.

LLMs face no white space. They face a screen to fill with pixels. They encounter only pretext—the tokens already generated by their parameters—and predict what token likely comes next. The choice of a word has nothing to do with intention. Prediction responds to what exists. Commitment binds the drafter to what does not yet exist: the unwritten sentence, the imagined reader, the argument still forming. The difference is not performance but orientation, i.e., looking backward for patterns to follow versus looking forward for meaning to make.

The Catcher as Drafter

The catcher’s vertical plane holds the pitcher’s recent history: what’s working tonight, arm fatigue accumulating across innings, the sequence already thrown to this batter. The horizontal plane holds the batter’s tendencies in the moment, his weaknesses against breaking balls, how he reacted to the last slider. Each pitch call commits to a trajectory while foreclosing others—just as each word choice does.

The catcher cannot deliberate indefinitely on which word to choose. The pitcher awaits the sign. The batter digs in. A decision must come. So too the drafter: the cursor blinks, the sentence demands completion. For the catcher and the drafter alike, “commit” names the moment when deliberation ends and consequence begins. The sign goes down. The word enters the sentence. The planes converge in action. For the catcher there can be no revision; the drafter can take it all back—until they can’t.

Apprenticeship Learning to Write in Academic Settings

Inexperienced or lightly schooled writers are challenged just by operating on the H plane in complex academic settings. Interestingly, studies as well as practitioner reports note that students write frequently and fluently in informal digital spaces such as texts, emails, and social media, developing habits like abbreviations, emojis, and a highly conversational tone.

Although NAEP reports reading and math data every rwo years, the most recent NAEP data on writing comes from 2011. In lieu of actual empirical data, I’ll rely on what I know of the glacial speed of improvement in writing instruction over the past half century and wager that simply getting words down—drafting—takes every bit of cognitive effort such writers can muster.

Without a functioning V plane constructed from rich experiences over time doing interesting, challenging, rewarding work as a writer in a community, basic writers draft texts using their instincts in the moment. While they may have an intention, e.g., to draft an essay according to their teacher’s specifications, they have no accumulated memory of the millions of decisions they made during earlier academic writing experiences that inform options they have available at the cutting edge of the draft.

Experienced writers often have highly developed V planes from the very start of a draft. The experienced writer has a functional sense of how the text will look eventually from the very start of its construction. At first, this sense is tentative, provisional, iffy. As the draft unfolds, this sense thickens and ripens from accumulated decisions, which gradually build momentum and resolution.

LLMs as a Surrogate Tool in Horizontal and Vertical Space

Writers who perform magnificently drafting in H space on social media have developed a very specific variety of V space. They’ve had deep experience in the setting and have accumulated millions of microdecisions about tone, timing, abbreviation, emoji deployment, and audience calibration within that genre. Their V plane is robust but narrow. It does not transfer to academic writing because the accumulated decisions were made solving different problems for different readers under different constraints.

When these writers enter academic settings, they face a V plane vacuum. The instincts honed in digital spaces misfire. Conversational tone registers as inappropriate. Abbreviations read as errors. The rich vertical accumulation from social media offers little guidance when the H plane suddenly consists of a blank Word document and a prompt requiring argumentative coherence across multiple paragraphs.

LLMs offer an artificial vertical space. They can generate academic-sounding prose instantly, filling H space with words that appear to have been crafted in the interaction of the V and H spaces. The student who loads drafting onto an LLM never makes the millions of microdecisions that build a functional V plane. The white space fills with words, but the words are residue from other writers’ accumulated decisions, not commitments that bind this writer to this text for these readers.

LLMs permit students to bypass the horizontal work without building vertical capacity and yet get text produced. The assignment gets submitted. But the student was never committed to the text in any way, shape, or form. The student’s V plane remains empty. And the next academic writing task will find them just as stranded, reaching again for the tool that produces an artifact by providing a brittle, synthetic prediction of what a human text might look like in some alternate universe.

Reframing the Debate

The debate about LLMs and student writing typically rests on two assumptions that collapse under scrutiny. The first frames the problem as integrity violation. The second frames it as cognitive offloading. Each framing sheds light on the consequences of thoughtless, reflexive, routine uses of LLMs to produce synthetic essays. Both assumptions obscure as much as they reveal.

Integrity assumes understanding of norms and commitment to them. To violate academic integrity, students must first possess a functional concept of what academic integrity means within the specific conventions of academic discourse. But this understanding is not innate.

Insight into appropriate choices accumulates on the V plane through repeated, mentored experience navigating the boundaries of legitimate collaboration, proper attribution, and authentic voice. The V plane which contains robust accumulated understanding of integrity is a resource capable of navigating collaborative writing projects. In terms of drafting, integrity means in part skill in locating one’s own contribution and awareness of the provenance of other authors over words, phrases, and ideas along with technical citation skills.Accusing students of integrity violations when they lack the vertical accumulation to understand what integrity means in this context mistakes absence for transgression.

Cognitive offloading muddies the water in a different way. The term suggests students are removing cognitive work from themselves and depositing it elsewhere for someone or something to do it for them. But from the student’s perspective, the operation looks like loading, not offloading. How is it bad to offload work that seems designed without the worker in mind? They load the H space with synthetic text that appears cognitively dense, syntactically complex, and academically appropriate. The text looks loaded. It simply is not loaded with the student’s own accumulated decisions.

When the work of drafting gets offloaded—the specific cognitive work of making word-by-word commitments across white space while integrating V and H planes in real time—we face a pedagogical problem. Drafting is precisely the activity that builds V space. Each drafting session deposits memories of decisions made, problems solved, coherence achieved or abandoned. Without drafting, nothing accumulates. The V plane remains barren regardless of how many texts get submitted.

The common denominator in the learning problem is not technology, but pedagogy. Students arrive with V plane vacuums not because they are lazy or dishonest but because writing instruction has provided insufficient opportunities for genuine drafting under conditions that build vertical capacity.

Process pedagogy focused on writing prompts or detailed specifications, cookie cutter protocols in which every student is writing to the identical assignment with no choice, rubrics that are too often subjective and controlling, on whole-class staging of deliverable products—prewriting, drafting, revising, editing—without attending to what accumulates in V space across those stages and gets shared and aired in communities of learners leads to conditions where little expertise is developed, and LLMs are seductive. Students learn that writing has steps. They don’t develop the vertical accumulation that makes those steps productive.

LLMs exploit this vacuum, the toughest problem facing writing teachers. You can't curriculum-design your way out of V-plane vacuums. You have to sit with students in the slow, frustrating work of accumulation. LLMs offer instant text production to writers who have no accumulated sense of what text production should feel like, cost, or yield. The student without V space cannot recognize the LLM’s output as counterfeit because they have no genuine article for comparison. Responding to this exploitation requires instructional patience and a great deal of what DuBois called a sympathetic touch.

They have rarely experienced the experience of drafting as exploration and discovery, the satisfying weight of commitment to an intention, the prize of coherence achieved through internal clarification. The LLM’s production feels like writing to someone who has never actually written.

Writing pedagogy must refocus. Not on process as sequence, but on learning in V space—on creating conditions where students draft repeatedly, intentionally, voluntarily, reflectively, with metacognitive attention to what accumulates across drafting experiences.

The goal is not producing texts but building the vertical plane that makes authentic text production possible. Until students have robust V spaces constructed from genuine drafting experience, debates about integrity and offloading will continue to miss the point. You cannot offload what you never saw as your work to begin with. You cannot violate integrity you never had opportunity to develop.

Looking Back at My V Space

Two earlier pieces on my Substack, "Learning to Read, Reading to Learn," serve as forerunners to my present view. Revisiting them here accomplishes multiple purposes simultaneously. For readers who encountered them originally, this constitutes a reloading—a return to familiar material now illuminated by subsequent development. For readers without prior textual contact, this constitutes an unloading—a delivery of foundational ideas necessary for understanding where my argument now stands.

In June 2023, six months after CHATGPT3.5 appeared on screens everywhere, I published "Swimming in the Pacific Ocean." The piece began with a simple premise: I would ask ChatGPT to write about the direct experience of swimming in the Pacific. The bot could only offer, as I wrote then, "a second hand account from the residue of patterns in meaning-language particles it had aligned with parameters during its training." I was building V knowledge and writing about it.

The bot tokenized my prompt in two dimensions: the Pacific Ocean and feelings humans report while swimming in it. It then encoded patterns of meaning along these parameters, juxtaposing immensity with awe, temperature variation with invigoration, powerful waves with thrills. The output possessed grammatical coherence and topical relevance. What it lacked was any evocation of the cold shock of salt water, the pull of undertow against ankles—what I knew it was really like to spend time on a beach in Catalina Island off of Southern California. I chose this topic because of my store of raw experience with it.

The essay diagnosed a fundamental LLM limitation. The bot's wonky logic—describing swimming as both "diverse and exhilarating" in grammatical coordination that was not logically coordinate—revealed something about how statistical patterns in language can mimic coherence without achieving it. The bot had access to how humans write about swimming in oceans. It had no access to what swimming in Catalina water felt like and meant to me. How could it?

A year later, in July 2024, I published "A No Brainer," which deepened the analysis through engagement with Vygotsky's distinction between thought and language. Vygotsky argued that thought and word are not connected by a primary, stable bond but that connections originate, change, and grow through development.

In Vygotskian theory, human infants think without words for the earliest months of their lives. Thought precedes language and remains partially independent of it throughout life. The implications for language bots are profound. If verbal thought requires the spontaneous emergence of meaning apart from words, followed by the fluid, murky process of formulating that meaning in language, then bots cannot engage in verbal thought at all. There is no prior thinking from which language emerges. They are thoughtless.

I drew on Vygotsky's insistence that word meanings are not fixed and stable. Meanings evolve and change with individual interactions and experiences. Therefore, verbal thought must always begin from scratch. Words are not ready-made garments. The ready-made garment metaphor illuminated precisely what bots offer, i.e., residue, prefabricated linguistic patterns that can be assembled according to statistical weights but can never represent raw experience. Bots offer ready-made garments because that’s all they have.

The 2024 piece also introduced the concept of unified consciousness to my emerging perspective, a concept rooted in theoretical models of reading comprehension. Humans bring together mental consciousness (thought) and perceptual consciousness (sensations) through verbal cognition. We attend to meanings while simultaneously inhabiting bodies that perceive, feel, and act.

Bots can mimic mental consciousness through pattern manipulation. They cannot participate in perceptual consciousness in a human sense because they have no bodies, no sensations, no reproductive organs, no skin in the game. This unified consciousness renders the meanings of human existence different in origin and character from the residue of mechanical language generation.

These two pieces—the 2023 meditation on residue and the 2024 on unified consciousness —represent stages of my vertical space development. Each builds on what came before, though the connections are not always explicit. The residue concept required philosophical grounding that Vygotsky provided. Words are what is left when thought ends. The Vygotsky analysis required empirical specification that subsequent engagement with neuroscience would supply.

My core claim remains constant: LLMs operate on the residue of human language without access to the embodied grounding that makes language meaningful. Bots do not make nor communicate meaning. They assemble words into residual patterns from past uses of language. The supporting argument has become more nuanced and more firmly anchored in empirical research.

Takeaways for Writing Pedagogy

Horizontal Space: Drafting as Foundation

Drafting should have a secure, protected place in the writing classroom. The horizontal plane—white space gradually filling with words—is where writers make the millions of microdecisions that eventually constitute vertical capacity. Without extensive drafting experience, students have nothing to accumulate.

Building Community Through Drafting

Activities that generate low-stakes drafting build both horizontal fluency and collaborative community. Think-pair-share with writing components. Anticipatory free writing before discussion. Academic diaries, logs, and journals maintained across a term. These practices normalize drafting as a regular cognitive activity rather than a high-stakes performance. Writers drafting alongside other writers develop shared understanding of what the work feels like, what it costs, what it yields.

Sequencing Human and Machine Involvement

Students should draft without an LLM in the loop until they have developed a solid sense of what they intend to say in a formal piece of text—what I’ve called a commitment. This sense cannot be borrowed. It must emerge from drafting itself, from natural mental work involved in finding words adequate to intention, from the recognition that intention clarifies through the act of articulation. Premature LLM involvement short-circuits this clarifying thinking.

Once students have intentions taking shape and acknowledge publicly what they plan to try, they can experiment with drafting in a chat window—not to have the bot write for them but to externalize their thinking in a space where they can then interrogate what they have produced. The bot becomes an analytical interlocutor: What did I just do? What assumptions am I making? Where does this seem to be heading? The student remains the committed agent; the bot provides a surface for reflection.

As students approach a fully human sense that a draft is ready for formal revision, they should seek human feedback first. Peers, tutors, instructors—readers who bring their own V planes to the encounter, who can respond as genuine audiences with genuine preparations. Only after processing human feedback should students seek targeted bot feedback on specific features: clarity of a particular passage, consistency of tone, adequacy of evidence for a specific claim. The bot serves as a tool for refinement, not a surrogate for drafting.

Critically, writing teachers undercut their own practice when they use LLMs routinely to evaluate student texts on human grounds. If it is true that LLMs produce residue, not texts in the sense we use in writing classrooms, it’s logically untenable to expect human feedback. This is not to say that LLMs are useless as assessment tools. It is to say that teachers need to read student writing themselves and respond in human terms to both H space and V space issues and questions. Formal writing products for submission to a teacher should include reflective commentary from students on V space issues which teachers respond to in depth either in conferences or in writing.

Vertical Space: Accumulating Capacity Over Time

The vertical plane does not build itself. It requires deliberate pedagogical attention to what accumulates across drafting experiences and how that accumulation becomes available for future work.

Revisiting Earlier Drafts

Students should regularly log and reflect on their drafting decisions—not just what they wrote but why they wrote it that way, what alternatives they considered, what they foreclosed. This reflection deposits metacognitive sediment on the V plane. Over time, students develop awareness of their own patterns as thinkers and problem-solvers under the pressure of composition.

They learn what works when they get stuck: walking away, skipping ahead, talking the problem through aloud, dropping back to reread the last solid paragraph. They accumulate strategies for pressure-testing a draft: reading against their own claims, asking where a skeptical reader would push back, identifying the paragraph that felt easy to write and therefore probably needs the hardest scrutiny. They find productive ways to chat with a bot to gain V insights from problems they encounter

They learn to recognize the difference between productive struggle and unproductive spinning, between a direction worth pursuing and a dead end demanding abandonment. They develop instincts about when to trust their own judgment and when feedback from another human is essential. This awareness constitutes vertical knowledge that informs future drafting—not knowledge of surface features but knowledge of how to navigate the cognitive and rhetorical challenges that drafting inevitably presents.

Making V Space Explicit

Instructors should name the vertical plane and discuss it directly. Students benefit from understanding that experienced writers carry accumulated knowledge into every drafting session, that this knowledge took years to build, that they are in the early stages of building their own.

Framing writing development as V plane construction gives students a conceptual handle for understanding why drafting matters, why it cannot be offloaded, why drafting isn’t the obstacle but the substance of growth.

Integrity as Vertical Knowledge

Academic integrity should be taught as V plane content, not assumed as prior knowledge. What counts as legitimate collaboration? When does influence become plagiarism? How do conventions differ across disciplines and genres? These understandings accumulate through explicit instruction and repeated practice navigating boundaries. Students with robust vertical knowledge of integrity can make informed decisions about LLM use. Students without it cannot violate what they do not yet possess.

The Pedagogical Reorientation

Writing pedagogy must shift from process as sequence to learning in V space. The goal is not moving students through stages—prewriting, drafting, revising, editing—but creating conditions where vertical capacity accumulates across every stage. Each drafting session should deposit sediment. Each revision should deepen awareness. Each completed text should leave the writer better prepared for the next.

LLMs are not the culprit. Pedagogical neglect of V space is the culprit. Students who arrive with robust vertical planes can use LLMs as tools without being captured by them. Students with V plane vacuums reach for LLMs because they have nothing else. Fill the vacuum through sustained, reflective, communal writing experiences where everyone actually cares what others have to say, and the LLM problem resolves into appropriate tool use by writers who know what writing is because they have done it, repeatedly, with accumulating awareness of what they are doing and why.