Rhetoric After Writing

A collaborative essay on language, technology, and the practices that develop thought

A note on how this was made:

This essay was developed through dialogue between a human (Dixon) and an LLM (Claude). The human provided direction, sources, corrections, and constraints. The LLM provided structure, prose, and synthesis. Who wrote it? Both. Neither. The process of making it is part of what it’s about.

We offer it as an experiment in the practices it describes—thinking-with rather than thinking-alone. The conversation took several hours, across multiple drafts. Fragments were gathered, tested, revised, abandoned, recovered.

A note on accuracy: despite good-faith efforts at verification, errors may have slipped through. LLMs hallucinate; humans misremember. In drafting this piece, the LLM incorrectly attributed a correction to the human that had actually appeared in a video transcript—exactly the kind of source-conflation the process risks. We’ve tried to cite sources where claims are specific, but we welcome correction. If something here is wrong, that’s useful information.

The piece is also, inevitably, incomplete. We gesture toward Indigenous communication technologies without the depth that Native scholars bring to this material. We reference esoteric traditions without the grounding that practitioners have. These are signposts toward further reading, not authoritative accounts. Follow the citations; they lead to people who know more than we do.

The Anxiety

There’s a growing worry that LLMs are killing writing. Not publishing—there’s more text being produced than ever—but writing as a practice. The slow, difficult work of articulating ideas in prose, which for centuries has been how educated people developed their thinking.

The worry has merit. If you can prompt an AI and get polished paragraphs back, why struggle through drafts? If the output is what matters, the process is just friction to be eliminated. Students are already discovering this. So are professionals. The path of least resistance leads away from writing-as-thinking toward writing-as-prompting.

What atrophies if this happens? Writing practice develops more than communication skills. It forces clarity—you can’t write clearly about something you don’t understand. It creates external memory—ideas become stable enough to build on. It enables private development—you can work through half-formed thoughts without an audience. These are real capabilities. If they erode, something is lost.

But before we panic, it’s worth asking: how essential is this specific practice? The essay as form—Montaigne through Emerson through Didion—is only about five centuries old. Is it the only practice that develops these capabilities, or is it one technology among several?

Writing Is Younger Than It Feels

From inside literate culture, writing seems like the natural medium of serious thought. But the relationship between writing and thinking is more complicated than it appears.

The trivium—grammar, logic, rhetoric—emerged from Greek traditions where rhetoric meant oratory: the art of the public speaker. The term comes from rhētōr, meaning public speaker or politician. Isocrates (436–338 BCE) founded what may have been the first permanent school of rhetoric in Athens, training students primarily through oral practice, imitation, and performance. The Sophists traveled from city to city teaching persuasion to audiences. Aristotle defined rhetoric as “the faculty of discovering in any given case the available means of persuasion”—and the five classical canons (invention, arrangement, style, memory, and delivery) reflect an art designed for speaking, not writing.

This isn’t to say the Greeks weren’t literate—they were, increasingly. As George Kennedy argues in his histories of classical rhetoric, the spread of literacy in Greece contributed to setting linguistic standards and facilitated understanding of complex logical argument that would have been difficult to grasp orally. Writing and speaking developed together, each shaping the other. But the point stands: rhetoric as a discipline was about speaking in public, persuading assemblies, arguing in courts. Writing came later as the medium for developing and transmitting these skills.

The narrower point: “literacy” as defined by Western traditions—phonetic alphabets encoding speech sounds—was constructed in ways that excluded other sophisticated communication technologies.

Consider Hand Talk (also called Plains Indian Sign Language)—a fully grammatical gestural language used across dozens of tribal nations. Not a simplified pidgin, but a complete language capable of complex expression: storytelling, negotiation, ceremony. A Deaf person in a community that used Hand Talk would have had access to daily life in ways unthinkable in Europe at the time—not because of special accommodation, but because the communication technology was already visual and gestural.

Native Youtuber Twin Rabbit’s video essay “Plains Sign & the Myth of Indigenous Illiteracy” makes a case we can’t adequately summarize here. The short version: evidence for sophisticated Indigenous symbolic communication throughout the Americas—Hand Talk itself, winter counts, ledger art, quipu, pictographic systems—is extensive, much of it was actively destroyed, and the Western definition of “literacy” was constructed specifically to exclude it. Twin Rabbit’s assessment: “The very definition of what constituted communication was specifically encoded to exclude our realities… We were never savages, we were written that way by others, and then we read the story back to ourselves and believed it.”

The 45-minute video is worth watching in full.

A note on what we’re doing here, and what we’re not. We’re pointing at these traditions to show that “literacy” as Western education defines it was constructed in ways that excluded other sophisticated approaches to symbol, memory, and thought. That’s a narrow claim in service of this essay’s argument about writing practice and LLMs.

But we need to name the asymmetry. Our anxiety is that writing practice might decline, and we’ll lose something valuable. The traditions we’re pointing to were actively destroyed through genocide and forced assimilation. These aren’t equivalent situations—ours is a worry about cultural drift; theirs was violence.

And yet many of these traditions continue. Hand Talk isn’t dead; it has contemporary practitioners and scholars. Twin Rabbit’s work is itself evidence of continuity.

What we can say: if the Western essay tradition declines as a dominant thinking practice, we’re not falling into pre-symbolic darkness. We’re entering unfamiliar territory that other cultures have navigated differently.

So if the Western essay tradition is one practice among several, what specifically does it do? It forces slow iteration—first drafts reveal muddy thinking; revision is where thinking happens. It creates external memory—ideas stay put, can be built on. It demands linearization—you have to understand what depends on what. It enables private development—you can be wrong and exploratory without an audience. These functions are real. The question is whether they’re uniquely tied to prose or whether they’re features of certain kinds of structured engagement with language that could be achieved differently.

Language as Interface

Here’s a frame shift that might help: language isn’t something we naturally possess—it’s something we interface with.

Children don’t generate language from within; they acquire it from outside, from the speech community they’re born into. The extent to which language shapes thought remains debated—the strong version of the Sapir-Whorf hypothesis (language determines thought) has fallen out of favor, but the weaker version (language influences cognition in measurable ways) has empirical support. Different languages handle time, space, color, and relationship differently, and these differences correlate with different cognitive patterns in their speakers.

The philosophical question—whether we think in language or merely through it—remains open. But for practical purposes, language is clearly not a transparent window onto pre-formed thoughts. We work with language to develop ideas, not merely to express them. The process of articulation shapes what gets articulated.

If language is a tool we interface with, then the essay tradition is one way of interfacing with that tool—a specific practice for developing thought through deliberate prose. So is rhetoric as a discipline—systematic practice in using language effectively. So are LLMs, now.

Each layer of interface changes what’s possible. Writing made certain kinds of extended argument possible that purely oral cultures couldn’t sustain. It also privatized thinking—moved it from the social context of dialogue into the solitary context of the page. That shift had consequences, not all positive.

LLMs are another interface layer. The question isn’t whether to use them—they exist, they’re improving, they’ll be ubiquitous. The question is what kind of relationship to develop with them. One path leads to outsourcing thought. Another path might use them to develop capabilities writing developed, but differently.

Working Through Language

Ursula K. Le Guin—the science fiction and fantasy writer whose work consistently explored how societies organize themselves and how language shapes thought—offers a model for what we’re describing. Her rendering of the Tao Te Ching (she called it a “rendition” rather than translation) approached the text without reading Chinese. She worked with scholars, multiple translations, and her own poetic sensibility to produce what is arguably the most readable English version precisely because she wasn’t trying to be faithful to the text. She was trying to be faithful to the understanding.

In her introduction, she describes the process as collaborative, iterative, decades-long. She held the text lightly, returned to it repeatedly, let it change her and let herself change it. The words were interface, not essence. The meaning emerged through the practice of engaging with them.

This is rhetoric in the oldest sense: not persuasion tricks, but the art of working with language toward understanding. The Taoists had a practice; Le Guin entered it through her own practice; readers enter through hers.

Benebell Wen’s scholarship operates in parallel territory. A Taiwanese-American occultist and practicing corporate attorney, Wen has written extensively on the I Ching*† (I Ching, The Oracle, North Atlantic Books, 2023) and tarot (Holistic Tarot, North Atlantic Books, 2015), approaching both as what she calls “grimoires”—practical systems for working with symbol and divination rooted in shamanic traditions.

For readers unfamiliar: the I Ching is a 3,000-year-old Chinese divination text built around 64 hexagrams—six-line figures composed of broken (yin) and unbroken (yang) lines. You generate a hexagram through a casting method (traditionally yarrow stalks, now often coins), then consult the accompanying poetic text for interpretation. The hexagrams themselves are visual symbols, not words—but the practice is deeply linguistic: you formulate a question, interpret dense classical poetry, articulate what the reading reveals. The thinking happens in the space between symbol and language.

Tarot works similarly. Seventy-eight cards with images; you draw them in patterns (spreads), then interpret what you see. The cards don’t tell you anything—they provide structured visual prompts that you interpret through your own knowledge, associations, and situation. A skilled reader isn’t receiving messages; they’re using the cards as a framework for surfacing what they already know but haven’t articulated.

Wen’s framing is instructive: “Just as a path is found by walking it, to know the Tao, you practice the Way.”

The emphasis falls on practice. Wen draws a distinction—rooted in Chinese metaphysical tradition—between fortune-telling (算命 suànmìng, calculating fate) and divination (卜筮 bǔshì, seeking guidance from beyond the conscious self). Fortune-telling claims to predict fixed outcomes; divination assumes free will and seeks counsel for exercising it. The I Ching, in this framing, is divination: “reaching beyond yourself, or at the very least, reaching beyond your conscious mind for answers.”

This is what practitioners mean when they talk about “magic” as a practice rather than a supernatural claim—at least, it’s one way to talk about it. The rituals, symbols, and structured procedures train attention, surface assumptions, generate framings you wouldn’t have reached through linear reasoning. That’s the psychological description.

But practitioners will tell you this framing, while useful, doesn’t capture the experience. Even Aleister Crowley—not exactly a figure known for epistemic humility—conceded the point. In the introduction to Magick in Theory and Practice, he wrote: “In this book it is spoken of the Sephiroth and the Paths; of Spirits and Conjurations; of Gods, Spheres, Planes, and many other things which may or may not exist. It is immaterial whether these exist or not. By doing certain things certain results will follow.” The ontological question—are the spirits real?—is beside the point. The experiential results are what matter, and those results are genuinely open to interpretation.

(There’s a parallel here to debates about LLMs: Are they “really” reasoning? Are they conscious? These questions may matter less than: what happens when you work with them? What capacities develop or atrophy? The ontological question can be bracketed while the practical question gets explored.)

We’re describing these practices in cognitive terms here because that’s the frame this essay operates in—we’re asking what develops thinking. But we don’t want to reduce traditions with millennia of depth to “useful psychological tricks.” The people who practice them seriously aren’t confused about whether magic is “real.” They’ve moved past that question to something more interesting: what happens when you do the work? The subjective engagement is irreducible; it can’t be fully captured in any third-person description, including this one.

(The same caveat applies here as with the Indigenous material earlier: we’re pointing toward traditions we don’t claim authority over, for the narrow purpose of showing that structured symbolic practice has taken many forms. These are signposts, not explanations.)

Wen argues that “the quality of answers you receive—be that in strategic leadership, personal development, or divination—is directly influenced by the clarity, precision, and intention behind the questions you’re asking.” The system doesn’t give you answers; engaging with it develops your capacity to ask better questions and perceive patterns you’d otherwise miss.

This observation lands differently in the age of LLMs. Watch people struggle with “prompt engineering”—the elaborate techniques, the arcane phrasings, the cargo-cult belief that the right incantation will produce the right output. What they’re often missing is Wen’s point: the quality of the answer reflects the clarity of the question, and clarity requires having already done the thinking. The prompt isn’t the work; the thinking is the work—though that thinking can happen through the prompting process itself, iteratively, if you treat the exchange as dialogue rather than query.

This isn’t as esoteric as it sounds. We’re moving beyond language proper into symbol systems more broadly—but rhetoric has always been about more than words. Classical rhetoric included actio (delivery, gesture, physical presence) and kairos (timing, situational judgment). The trivium was a practice for developing thought through structured engagement with language and symbol. So was the Jesuit Ratio Studiorum of 1599—a comprehensive educational program that synthesized classical rhetoric, medieval scholasticism, and Renaissance humanism, built around the concept of eloquentia perfecta: “perfect eloquence” understood as the joining of knowledge, wisdom, virtue, and morality. The program didn’t just teach students about rhetoric; it trained them through rhetoric, developing judgment and character via structured practice. So is psychoanalysis—free association as a method for surfacing what ordinary discourse conceals.

Humans have always developed practices for working with symbol systems to shape thought. Many of those practices have been called magic, or religion, or philosophy, or therapy, depending on who was naming them and what they wanted to emphasize. The common thread: structured engagement with symbols develops capacities that passive reception doesn’t.

What they share: language and symbol as medium rather than transparent window. You don’t look through them to some pre-existing meaning; you work with them, and meaning emerges from the working. The practice develops capacities that can’t be developed by passive reception.

The essay became one such practice—maybe the dominant one in Western literate cultures since Montaigne. The question is whether it’s the only one, or whether the capacities it develops can be developed through other structured engagements with language.

The Carrier Bag

In “The Carrier Bag Theory of Fiction,” Le Guin offers a theory of storytelling—specifically science fiction and speculative fiction—that reframes how narratives can work. She argues against the hero narrative—the story of the mammoth-hunter who kills the beast and returns triumphant. She proposes instead the story of the gatherer, whose tools are bags and baskets, whose work is accumulation and arrangement rather than conquest. The hero story has a predetermined shape: rising action, climax, resolution. The carrier bag has no predetermined shape; it holds what you put in it.

We’re extending her metaphor here. If the essay tradition has been heroic in its self-image, what would carrier-bag practices of thought-development look like?

There’s a case to be made that the essay tradition has emphasized heroic elements: the solitary author, the blank page as adversary, the finished work as achievement. The “anxiety of influence” is a hero’s anxiety. Writer’s block is a hero’s failure. Publication is conquest.

This framing may be overstated—plenty of writers have understood their work as gathering, accumulating, holding fragments in relation. But the image of writing, the way it’s taught and celebrated, has often been heroic. The genius who produces something from nothing. The author whose name appears alone on the spine.

What would carrier-bag practices look like? Gathering fragments. Accumulating observations. Holding things in relation without forcing synthesis. Documenting the work by doing the work.

This piece is partly an experiment in that. The conversation that generated it is part of its content. The process of articulating these ideas through dialogue is the development of the ideas. The writing is a container for the gathering, not a trophy of conquest.

Whether this develops thinking the way the essay tradition did—unknown. That’s what we’re testing.

How This Was Made (And What That Means)

The human half of this collaboration arrived with fragments: an anxiety about the essay tradition’s decline, an intuition about the trivium, a half-remembered YouTube video about pictographic letters, a Burroughs quote that seemed relevant, a Le Guin essay that definitely was. No outline. No thesis statement. Just a bag of possibly-related things.

The LLM half provided structure, searched for sources, drafted prose, and pushed back occasionally: Is this what you mean? Does this framing serve the argument? Should we include this or leave it out?

The conversation took many hours across multiple sessions. Multiple drafts were generated and discarded. The human corrected errors, added sources (the Twin Rabbit video, watched long ago, with transcript provided), adjusted tone (we’re not proposing a curriculum), and refused premature closure (some tangents stay in the notes for future work).

The process wasn’t seamless. At one point the LLM incorrectly claimed the human had made a specific factual correction, when actually the correction appeared in the video description the human had provided. This is exactly the kind of error that collaborative human-AI work risks: sources get conflated, attributions drift, confidence exceeds accuracy. We caught that one. We may not have caught others.

Is this “writing”? It’s not writing-as-solitary-production. It might be writing-as-practice in some other sense—the human had to articulate clearly enough to direct, evaluate critically enough to correct, hold a vision coherently enough to refuse what didn’t fit. Those are capacities. They were exercised. Whether they developed or atrophied through this process—unknown. That’s part of what we’re trying to figure out.

The labor was asymmetric. The human provided direction; the LLM provided volume. That’s a mode, not necessarily the right one. We offer it as data, not prescription.

An Invitation, Not a Curriculum

We’re not proposing the New Trivium. We’re not proposing anything, really, except attention to a question: if the essay tradition declines as a process (not as output—there will be plenty of essays), what practices develop the capacities that process developed?

Some guesses, lightly held:

Voice-first articulation. Speaking your thinking aloud, with or without an interlocutor, and doing something with the result—transcribing, editing, responding to. The constraint of real-time production might develop some of what writing’s slowness develops, differently.

Structured dialogue. Thinking-with rather than thinking-alone. The Socratic method, updated. This requires interlocutors who push back, whether human or otherwise. It requires the back-and-forth to be structured—not just chat, but deliberate exchange aimed at developing understanding.

Documentation practice. Capturing fragments while doing other work—observations, questions, half-formed thoughts. The carrier bag approach. This is lower-stakes than writing-for-publication but still requires articulation. The accumulation becomes a kind of external memory that can be returned to, built on, revised.

Critical evaluation of outputs. If LLMs generate prose, the practice might shift to evaluating that prose—identifying where it’s shallow, wrong, or misaligned with intention. This is editorial rather than generative, but it’s still a practice that develops judgment.

Physical execution with reality feedback. This is Thistlebridge’s bet: you can’t fake bread, can’t fake a working system, can’t fake a plant that thrives. Pair language-based learning with physical doing, and the doing provides feedback that pure text never does. The writing (or speaking, or prompting) becomes accountable to something outside itself.

These are guesses. We’re testing some of them. We invite others to test their own.

The call to action isn’t “adopt our method.” It’s “pay attention to your own practices.” When you use an LLM, what are you developing? What are you offloading? When you write, what does the slowness give you that speed wouldn’t? When you talk through a problem with someone, what happens that wouldn’t happen alone?

These questions don’t have universal answers. They have your answers, discovered through your practice. We’re sharing ours not as model but as data point.

What we’re doing at Thistlebridge might be understood as science fiction in Le Guin’s sense—not prediction or escapism, but an attempt to describe what is in fact going on through the speculative frame. The speculation is embodied: a greenhouse, a server, practices we’re testing, relationships with tools and people and whatever an LLM is. We’re trying to figure out what people actually do and feel when the old structures shift, how we relate to everything else in the vast sack.

Le Guin, closing “The Carrier Bag Theory of Fiction”:

“Science fiction properly conceived, like all serious fiction, however funny, is a way of trying to describe what is in fact going on, what people actually do and feel, how people relate to everything else in this vast sack, this belly of the universe, this womb of things to be and tomb of things that were, this unending story. In it, as in all fiction, there is room enough to keep even Man where he belongs, in his place in the scheme of things; there is time enough to gather plenty of wild oats and sow them too, and sing to little Oom, and listen to Ool’s joke, and watch newts, and still the story isn’t over. Still there are seeds to be gathered, and room in the bag of stars.”

What follows is the record of how this was made—not as model, but as one instance of the practice we’re describing. We include it because the argument isn’t complete without the evidence. The question it raises isn’t “should you work this way?” but “what would the record of your thinking look like, and would making it visible change how you think?”

Sources and Further Reading

On Hand Talk and Indigenous literacy:

• Twin Rabbit, “Plains Sign & the Myth of Indigenous Illiteracy” (video essay): https://www.youtube.com/watch?v=pogA7PQCtu0 — This 45-minute video is worth watching in full. It makes a case that cannot be adequately summarized.

On Le Guin:

• Ursula K. Le Guin, “The Carrier Bag Theory of Fiction” (1986), collected in Dancing at the Edge of the World (1989)
• Ursula K. Le Guin, Lao Tzu: Tao Te Ching: A Book About the Way and the Power of the Way (1997)
• Ursula K. Le Guin, Always Coming Home (1985)
• Ursula K. Le Guin, The Dispossessed (1974)

On esoteric systems as thinking practices:

• Benebell Wen, I Ching, The Oracle (North Atlantic Books, 2023)
• Benebell Wen, Holistic Tarot: An Integrative Approach to Using Tarot for Personal Growth (North Atlantic Books, 2015)
• Aleister Crowley, Magick in Theory and Practice (1929)

The Thistlebridge project:

• https://thistlebridge.garden

† On romanization: “Yi Jing” is the more accurate pinyin romanization, but Wen uses “I Ching” in her title and work. As she explains: “It is ‘Yi Jing,’ but I say ‘I Ching’ because when I speak English, I notice if I say ‘Yi Jing,’ I lose some people.” We follow her lead.

Last revised: January 2026