michael-dean-k/

UNESCO says 86% of the world can read and write, but what percent can write a coherent essay that is not sand-your-eyes-out boring?

In the public cries of our literary crisis, which seem to reboot every fifteen years, we tend to reduce progress down to a single binary—literate/illiterate—and then pile on the studies to show the arrow going down and to the right. Take for example, The dawn of the post-literate society, a 2025 Substack post with +21,000 likes: it cites that reading for pleasure is down 40%, that PISA scores are down, that books are getting dumbed down, that sales are cratering too—doom after doom. Yet there’s little mention of the other form of literacy: writing.

Before we get into that, we should get concrete with a literacy framework. UNSECO has three tiers: basic (decoding text and writing a sentence), functional (operating in everyday life), and advanced (analyzing, synthesizing and evaluating). These loosely map to the more refined PIAAC levels, which map 1-5 and are measured every year across 30+ countries. The United States is in the middle of the pack, with the 2024 average score reaching 2.65 of 5. The leader, Finland, has a 3.41.

So, not only is literacy broken into multiple spheres (not just reading and writing, but also numeracy and problem solving), but each sphere has levels of competency. “Literacy is a continuum.” (UNESCO)

Now, in the 1970s, historians across Sweden, England, and America (Johansson, Cressy, Lockridge) ran experiments and noted that writing literacy was significantly behind reading—and also, that it’s been like this for centuries. So, if the average American reader is in the high 2s, we can safely assume writing is in the low 2s.

To break these scales down (in my own words, for now):

• Reading: (1) Can decode letters with correct spelling; (2) Elementary comprehension at the paragraph level; (3) Fluent comprehension of multiple themes across a text; (4) Analytical processing of subtext and intent; (5) Syntopical reading across several works for original conclusions.
• Writing: (1) Can transcribe letters with correct spelling; (2) Can form a string of sentences without grammatical errors; (3) Can organically compose a multi-paragraph essay to convey a basic idea; (4) Can weave in multiple themes and literary devices; (5) Can communicate complex ideas with original style in a way that is accessible and enjoyable to read.

As an editor that has worked with probably a thousand new writers across ages and industries, I was surprised to see how extremely competent and accomplished people can struggle past L2 writing. They can of course spell and write sentences; yet they can’t compose a string of sentences in a way that is coherent or interesting. They cower at the thought of shaping paragraphs.

If tiers T3-T5 of writing are in the sphere of “composition” then what we’re faced with is mass compositional illiteracy. But so what?

In Harvey J. Graff’s “The Myth of Literacy” (1979), he noted that we default to use a “literary crisis” to explain our social woes of our era, like crime, poverty, or moral decline, and that literacy rarely fixes these issues. And if compositional literacy is already embarrassingly low, and has always been so, then surely “better essayists” is not directly correlated to the economy and it’s thornier issues.

So the real question is what is actually gained from being compositionally literate? What happens if we don’t stop at “functional literacy,” but insist that we rise the floor and elevate the standard citizen into advanced literacy? What do we lose if AGI composes all text of a future generation, creating little incentive to advance beyond L2 writing?

A person who is extremely competent in composition gets something more important than their compositions. In the pursuit of masterpieces, one becomes a master. And while mastery can be focused into one of many specialized domains, the thing that matters is the cross-disciplinary, transferrable skills. A master of architecture can more easily become a master of music, or entrepreneurship, or technology. But more important than the mastery of specific trades are the soft-skills that carry into the sphere of everyday life. So really what we want to be promoting is general mastery, and the most accessible, specialized medium to do that through is writing—and even more specifically, not through writing novels or poems, but through writing essays.

I know it sounds like I’m saying that “writing essays will make you a better person,”—and of course history has it's share of articulate monsters—but more specifically, I’m saying that writing essays will bolster a specific set of traits. Conversely, it’s possible that someone without such traits natively might have more trouble becomes compositionally literate and well-rounded. There are some basic traits that are exercised when one spend 50-100 hours conceiving, writing, and editing a longform idea. Inherent to the process are patience, cognitive agility, detail orientation—but to go further, I wanted to transfer each Essay Architecture pattern into a corresponding skill.

(...Before I share those, I should note that Essay Architecture is really a system of “rhetoric,” and our notions of rhetoric are often all wrong. Rhetoric is not merely persuasive and manipulative speech writing. It’s about understanding the deeper motivations, the how and the why behind the what of a text. Surely, this can be applied to craft the perfect message for a specific audience, but, you can also use rhetorical tools to examine yourself. ie: The act of interrogating your first draft and probing into why you actually believe something, how you came to those conclusions, and what you really want to explore, is an act of rhetoric applied to the self. They say writing sharpens your thinking, but editing is how you change your thinking and slowly steer the evolution of your belief system...)

Aristotle’s system of rhetoric is broken into three "appeals": ethos, logos, and pathos. The Essay Architecture framework is a fractal of this system: the triad is first found in the balance of dimensions, and then in the elements within a dimension, and finally in the patterns within an element. Here are all 27 patterns, organized by appeal, and converted to the meta-skill that’s acquired in the mastery of that pattern:

• ethos: vulnerable, metaphorical, memetic, orienting, cinematic, dramatic, dynamic, emphatic, eidetic;
• logos: skeptical, erudite, discerning, lucid, disciplined, polyphonic, dimensional, rhythmic, precise;
• pathos: polymathic, catalytic, melodic, climactic, architectonic, enticing, allusive, lyrical, symbolic;

So while it’s hard to know the exact second-order effects if every citizen were to gradually inherit these traits to their most basic degrees, we can imagine how much better equipped they would be, better able to bring themselves into the public sphere, to bring order to complexity, and to resonate with others. There is much more to say to this, and I’m sure I have improperly summarized here, but given we’re in the afterglow of the 250th anniversary and the American spirit is in question, I’ll close by saying that an essay writer—as opposed to a journalist or propagandist or stylist—is one that is always drawn to think for themselves, away from polarizing mass-think, skeptical of popular dead-end beliefs, and towards their own genuine verdict on how they should live.

Another research paper on AI writing was posted on Substack, and naturally all the writers are interpreting it with a feel-good conclusion. My understanding is that they took hundreds of writing prompts, and for each they had one human and five different AIs write a 5,000 word story to it. The conclusions are what you would expect from an out-of-the-box AI: overly explicit with the themes, too linear, overdoing sensory descriptions, less intertextual than humans, and less experimental in form. And so a defensive writer will look at the graph and claim it a victory—"machines have words, but no music" (and other platitudes)—and expound on the inherent limitations of LLM-generated prose.

I often find myself arguing against the luddites, not because I think today's AI fiction is any good, and not because I particularly want a new machine class of writers, but because I think interpretation is often guided by fear instead of a basic understanding of the technical complexities. If you think AI progress in general has stalled or is reversing, and won't be unimaginably better in 5 years than it is today, then you're building yourself a cocoon that is bound to be shaken.

The common trope of AI skepticism is to point to poorly generated examples on free plans and mistake that for the state of the art. If you actually wanted to write good AI fiction, you wouldn't one-shot a 5,000 word story inside of a chatbot. Even if you prompted it to avoid the five weaknesses above, it would only marginally improve. There's only so much you can achieve with a prompt. I suppose this study might be good in determining the state of lazy generation: with zero effort or imagination or resources, if you just pitch a concept into a public chatbot, what will you get back? Useful to know, but this isn't the same thing as knowing what's possible.

Here's a $10,000 prize (with hosts like Gwern, Roon) called Unslop, focused on AI-generated fiction. Theoretically, given the incentives, you'd think this would assemble a collection of stories that display the frontier of what's possible. After reading a few though, I'm still not impressed.

Here's one of the judge's comments:  > I was a little surprised by how obviously AI these entries seemed to me, and how little distance there is from the average LLM writing that you get with a naive prompt. It’s made me think that (barring some technical breakthrough) we’re further away from LLM prose than I’d supposed we were, though I’ve been gradually moving to “pretty far, actually” over the last two years. Some of this is bound to be because of RLHF, but I really did think there was a chance the right harness would make a great short story. I’m not down on all these entries, but I did read them and think “well, better than a lot of human authors, but not good enough that I would expect most people to share them”. Also, some of the stories contain a good idea that’s executed poorly (IMO).

(I believe it was required to submit your harness with your story, but they aren't making the harnesses public. That would be interesting to see.)

It's possible to interpret this with finality: that regardless of the harness, 2026 LLMs still can't write good prose. We have abundant evidence that one-shot stories suck, and no evidence that carefully constructed AI stories are much better. But this might all come down to the truth that designing an effective harness is actually quite difficult, requiring a synthesis of skills that most people don't have.

Coding harnesses have made AI coding radically better, and maybe that's because the people building the harness also have expertise in the thing they're building the harness for. Coders can code a coding harness. But coders can't code a writing harness, and writers can't quite code. Most engineers don't have an expert-level understanding of writing. And sure, now anyone can try to build a harness (myself included), but to make something that produces human-level writing, you need an expert-level understanding of harnesses too. And so progress here might have nothing to do with size or sophistication with LLMs, and everything to do with fusing two halves of the brain. Can we encode the principles of storytelling at their fullest complexity? Possibly, but it's not an LLM limitation or an infrastructure problem. More so, it seems like a classic case of an invention having a multi-year lag behind when it's technically feasible.

What is automatic writing? There is allure, appeal, mythology, and history behind this term, which involves surrealists, beats, and sometimes random non-writers getting possessed at their keyboard, whether by Muse, by Enochian, by Lumerian, or by an unspecified spirit. Certainty, there are some techniques to get you there.

Last night, after a many-month drought, I broke out the typewriter along with an exercise book, "The Practice of Poetry," (Behn and Twichell). The second exercise spanned a 10-day stretched: each day I'd write whatever came to my mind, and, most importantly, I would not read it. Instead I'd stash it away in a drawer, only to come back to it after ten days, and only then would I read analytically, extracting accidental, emergent, resonant phrases that I could recombine into a poem. (Every time I finished, I am extremely tempted to read, because that's what I've always done after created anything, but denying yourself of that any editing or judgment forces you to ask yourself if the process itself was enjoyable.)

The details are slightly more specific than that, but there was no guidance on how to enter the forbidden realm of the Muse (or whatever you call it). No technique for the imagination. And so last night, on Page 1, I was quite in my head, quite analytical, quite my normal self, and felt quite uninspired. Afterwards, I reflected why this happened: I wasn't touching the "imagination," which I should probably define.

"Imagination" might not be the word here, but to me there is a dream-like mode of internal seeing, where I don't just see imagery, and it's not just hyper-realistic, but it's "self-evolving." Whether it's a scene filled with bustler's moving or a grotesque face, whether I am stationary or in motion, it takes on a life of it's own. It's also hard to pin down: it shapeshifts and flows into the next image, and the next and next and next, so afterward I'm left with a weird sense of what-in-the-world-was-that?, and I can loosely remember some scattered images, but it's mostly gone. I'm left with the post-trip spins of a rollercoaster or moving boat. I've always wondered, how might I capture the raw logs of these experiences? (They usually happen before bed or while meditating, and so I don't enter the realm with any intentions or techniques of rememberance.)

Whatever that is, I want to access it at the keyboard. That was my only goal this morning, and I got close. I figure, in addition to those streams of consciousness (which I will eventually read, and maybe share), I should write about the technique that seems to unleash it.

The main insight I'm left with is that you have to see, not read. Reading is very naturally part of writing. As I'm writing now on my laptop, I see each letter clearly formed—f-o-r-m-e-d—and I'm fully aware of where the sentence started, as well as the grammatical affordances ahead of me. Often I pause and re-read the last sentence or paragraph, and sometimes the whole piece, so I can get a sense of what's been said and what's left to say. This kind of analytical literacy is on the opposite end of the spectrum from poetic sight. I don't know if I can simultaneously read and hallucinate—this makes me curious to look up trip reports of people attempting to read/write on mushrooms/acid.

And so to really see as you write, a few things help.

First, my typewriter is slightly broken. After pressing a key, the carriage (?) lifts, and the metal prong pushes the ribbon onto the page, but the ribbon stays up after I unpress they key, blocking my view of the words being formed. This annoyance proved to be quite helpful. I should not be conscious of letters, words, or misspellings in this process.

Additionally, watching the back of your hands to make sure you're anchored over the right keys takes up bandwidth. When you watch yourself type, you expend attention that should be completely focused on conjuring sight, and so I just looked up. There is some NLP lore about how looking up triggers visions. Up and to the right is remembering past scenes. Up and to the left is inventing visual scenes. Or maybe it's reverse; I still cannot spot a liar... But once I did see a hippie in Mount Shasta meditate with his eyelids open and his eyes rolled completely up into his heads, so I could only see two white balls; it was creepy, but afterwards he explained in some not-entirely-convincing-but-still-plausible way how his specific patterns of eye-fluttering release DMT from his third eye that helps him see things... All that aside, you can hopefully type without looking. This is slightly harder on a typewriter, because when I pull the metal bar to jump to the next line, I lose my position. It's possible that Page 2 and all future pages will be completely illegible.

If I'm not reading words or watching my hands, then why have my eyes open? I did not have a blindfold nearby—I assume I have one in my apartment, and will try this tomorrow—but I did close my eyes. This lets me attempt to focus my attention purely on the "inner light" (inner sight), and my hands translate.

Halfway through, I started speaking out loud—in a slow, semi-dramatic British accent—to prosify what I saw. I don't know if this is for affect or actually useful. One theory is that spoken word is a bridge between sight and typing: I can easily narrate what I see, and I can easily type what I say. I can't doubt the tongue's ease of turning images into a prose; compared to the prose of fingers, the tongue feels one layer closer to the source, the original way we learned to speak, the way people spoke before writing was invented. And so this means the eyes are "seeing," the tongue is "prosing," and at this point, the hands have been demoted to a mere stenographer at the mind circus.