When I was in middle school, my homeroom teacher had us keep weekly journals. I wrote so much during one summer vacation that the entries became a book. My father printed it all out and gave it to the teacher.
I remember the feeling of that: having something to say that needed a form to hold it, and the form holding it without complaint. Writing felt like breathing then—natural, continuous, not something you could block.
I didn’t know, at that time, that I was a writer. I just wrote.
Twenty years later, I sat in an analyst’s office for the first time. The analyst asked what had brought me there. I told her I was having trouble with my indie game. That it didn’t feel “weird” enough. That I wasn’t sure I had anything genuinely original to offer.
What I couldn’t say clearly then—what I only understood much later—was that I wasn’t talking about game design. I was talking about something that had happened to the relationship between me and my own creative source. Something had gotten in the way. And because I didn’t have a framework for understanding what that was, I turned it into a question of craft. A question of skill. A question of whether I was talented enough.
This is what writing blocks do. They masquerade as technical problems: insufficient skill, missing structure, inadequate tools. But underneath, almost always, there is an emotional problem. Something that needs to be said is too difficult to approach. Something that needs to be written is connected to material the writer isn’t ready to enter.
I learned this slowly, over years of analytic work and a lot of writing. But I learned it most concretely in a writing course at my psychoanalytic training institute, where we were asked to write by free association—to let the pen follow the mind without editing or agenda. I found I could move through ideas quickly, fluently. But whenever the writing approached something particular—a specific detail, a memory with texture and temperature—I stopped. Or I generalized. Or I suddenly needed to use the bathroom.
The details were too much. Not because I couldn’t write them—I had the technical skill. But because writing them would mean feeling them. And the feelings were the hard part.
This is the insight that everything else follows from: the difficulty in writing is almost always emotional, not technical.
Better tools don’t change this. More organized systems don’t change this. Productivity optimization doesn’t change this. All of those interventions are solving the wrong problem. They’re making something easier that isn’t the hard thing.
After that workshop, I spent a year building writing methodologies. This is what I do when something is hard: I try to understand it at a structural level, find the friction points, design around them. I developed evaluation frameworks, revision protocols, prompts for different stages. My writing improved—substantially. The work I was producing became more rigorous, more argued, more like what I actually wanted to make. I even finished my very first complete fiction book.
But then I started writing my autobiography, or you can call it autofiction, autotheory, all at once.
Writing an autobiography is a particular kind of difficulty. It isn’t primarily a structural challenge, though structure matters. It isn’t primarily a research challenge, though memory is research of a kind. It is primarily an emotional challenge—you are the material, and the material resists. The project kept stopping. I had the methodology. I had the frameworks. And I still sat in front of the document feeling overwhelmed in ways that felt, somehow, before language.
That word—overwhelmed—kept appearing. Not just in the autobiography, but across all my long-form projects. By then I was working on three simultaneously: the second book of QPHU, my second long-form fiction, approaching a hundred thousand words; my dissertation, which requires holding multiple theoretical frameworks in productive tension over years; and the autobiography. Three projects, each demanding its own kind of sustained attention, each carrying its own emotional weight.
Here is something that surprised me: ten chapters of ten thousand words each is not the same kind of project as one project of a hundred thousand words. The complexity doesn’t scale linearly—it scales exponentially. A long project contains not just more words but more time, more returning, more forgetting-and-remembering, more partial versions of things that almost worked, more pieces that feel alive alongside pieces that feel dead.
After I wrote publicly about Palimpsest—what it was trying to do, why it existed—a reader sent me a message and quoted a paragraph back to me. And said: this is what writing is like for me too.
“The core premise is that serious writing—the kind of writing that involves research, drafts, revision, thinking, and emotional investment over long periods of time—happens in layers. You have fragments that might become something. You have drafts that almost work. You have pieces that feel alive and pieces that feel dead. You have notes that are trying to find each other. And you have, underneath all of it, an emotional experience of being in the middle of something that doesn’t yet know what it is.”
An emotional experience of being in the middle of something that doesn’t yet know what it is. That’s the phenomenology of serious writing. That someone recognized it—quoted it back—told me the description had landed somewhere real.
I didn’t need a more powerful system. I needed a container—something that could hold the complexity, hold the unfinishedness, hold the emotional texture of being in the middle of something that hasn’t found its shape yet—without demanding that I resolve it prematurely.
When people talk about what writing tools fail to do, the criticisms tend to be functional: Scrivener is too complex, Word is too linear, Obsidian doesn’t integrate research well enough. These are real problems. But I think they’re downstream of a more fundamental misunderstanding.
Most writing software assumes writing is production. You move in one direction—from blank page to finished draft, from less to more, from idea to output. The interface reinforces this: word counts, chapter lists that imply a destination, the implicit message that you should be moving forward. The feedback loop the tool provides is a production feedback loop.
But this is not what writing is, for most people who write seriously. Writing is return. You go back to what you wrote yesterday and it looks different. You find a note from three months ago that suddenly belongs somewhere. You write something and don’t know what it is yet, put it aside, and six weeks later you know. The project is not moving toward a destination—it is circling, layering, accreting.
Linear tools handle one kind of return—revision of a linear draft. They don’t handle the other kinds: the associative return, the temporal return, the emotional return to material you weren’t ready for before.
The interface is not neutral. Every interface encodes a theory of the user—what they need, how they think, what their relationship to the work is. Word assumes you are producing a linear document. Notion assumes you are managing information. Obsidian assumes you are building a knowledge graph. These are useful assumptions for their intended purposes. But they all assume the primary challenge of writing is organizational or structural.
I’ve spent years with the spatial tools—Heptabase and its kin—the ones that lay cards out on a canvas so you can see everything at once, move things around, find connections spatially. I know what they’re trying to do. But I kept encountering a problem: when everything is visible at once, organizing becomes a second job. The canvas is already overwhelming before you’ve added anything new. The tool designed to help you think has become something you have to manage. You go to write and instead you find yourself tidying.
What I wanted was something closer to how Gingko Writer works: a structured outliner where each card exists in relation—you can see its parent and its siblings, sense where you are in the whole—without the full weight of the project pressing in from all sides. Enough context to think. Not so much that thinking becomes managing.
This led to the stage-specific views in Palimpsest. Each stage of writing has a different emotional problem, and each needs a different relationship to the material on screen.
Ideation—the stage before coherence—needs permission. You need to write without pressure, without the sense that anything you put down is committing you to something. The outliner view gives each fragment its own space while letting you feel the emerging structure, without demanding that structure be finished.
Writing—the stage of deep focus—needs protection. The awareness of everything you haven’t written yet, everything you’ve already written that isn’t quite right, sits on the writing session and slows it down. A single card removes that weight deliberately. One piece of attention. The rest is outside the frame.
Organization—the spatial stage—needs room to listen. Writers have a tendency to impose structure too early, to fit pieces into a shape before the pieces have indicated where they want to go. The card view is designed for discovering rather than deciding: moving things, noticing what wants to be near what, finding shape rather than forcing it.
Evaluation—the stage of reading the whole—needs distance. When you’ve been living inside a piece of writing, you can’t read it. The chapter view is designed to restore that distance—to make the writing visible as a reader would encounter it, not as you remember writing it.
Each stage has a different emotional register. A container that understands this doesn’t try to make all four stages feel the same. It holds the differences.
This is where AI becomes genuinely interesting to me—and the question I keep returning to is one I think is most underasked:
If AI doesn’t write for you, what can it offer for a writer?
I’ve watched the discourse around AI writing tools settle into a binary that misses the most interesting territory. On one side: AI writes your content, faster, cheaper, at scale. On the other: AI is contamination, writing should be human, resist. Both positions accept the same premise—that AI’s primary value, or threat, is its capacity to generate text.
But I’m not primarily interested in text generation. I’m interested in thinking.
Writing is a technology for thinking. It’s how you find out what you know, what you believe, what you’re trying to say. The hardest parts of writing—the parts that send you to the bathroom, the parts that make you suddenly need coffee—are not the parts that require producing more words. They’re the parts that require encountering difficult material, staying with contradiction, saying something you’re not sure you’re allowed to say.
These are not problems AI can solve by generating more text.
There’s a technique programmers use called rubber duck debugging: when you’re stuck on a problem, you explain it—out loud, in detail—to a rubber duck on your desk. The duck doesn’t respond. But in the act of explaining, you often find the answer yourself. The duck’s value is not its expertise. It’s its presence: having something to address forces you to articulate what you couldn’t articulate before.
This is closer to what the AI in Palimpsest does. Not answering for you. Not suggesting what comes next. Asking: what are you trying to say here? What’s the feeling underneath this sentence? Is there something you’re avoiding? And sometimes, slowing you down when you’re rushing past something important—because I do this, write quickly through a passage that needed more time, because staying with it would mean feeling something I wasn’t ready for.
The AI also works with methodology. Every serious writer has a theory of what good writing is—even if they’ve never articulated it. Palimpsest includes built-in methodologies for different kinds of writing: fiction, creative nonfiction, academic work. Not prescriptions; frameworks that define what the writing is trying to do. With a methodology, the AI can notice gaps that are invisible without it. Not I’ll write this better for you, but: you said you wanted to ground each scene in a specific sensory detail—this one doesn’t have that yet. Do you want to stay here? The methodology defines what “done” means. The AI holds you to what you decided you cared about.
In Palimpsest, the AI never touches your text. Every word in the document is yours. Palimpsest works entirely without AI—every feature, every view, every stage functions without it. But the AI, when it’s present, offers something I haven’t found elsewhere: the experience of not being entirely alone in the middle of something difficult.
I want to say something that is, at this moment, slightly out of fashion.
Does anyone still write?
I mean: does anyone still do the thing where you sit with material you don’t yet understand, return to it over weeks or months, let it teach you something you couldn’t have known at the beginning? The thing where the difficulty is the point—where the friction between you and the material is where the thinking happens?
This is the counter-current Palimpsest is swimming in. The prevailing offer, right now, is to remove that friction. AI will draft it, AI will revise it, AI will fill in the parts you don’t know how to say. The output is faster. Whether it’s writing—in the sense I mean—is a different question.
I have recently started attending to what I call the narcissistic predicament of AI-assisted creation: the way that AI can produce outputs that look like creative work while bypassing the psychic work that gives creative work its meaning—for the writer. What gets lost isn’t legibility. It’s the confrontation with yourself that serious writing requires.
I’m not interested in speed. I’m interested in that confrontation. Palimpsest is built for people who are, too.
The name comes from the medieval writing surface: a palimpsest is a manuscript that has been written on, scraped clean, and written on again—but where traces of earlier writing remain visible beneath the new. A palimpsest holds its history. It shows its layers.
Writing is like that. The things you wrote last week and deleted are still present in what you’re writing today, as influence or avoidance or something working below the surface. A tool that understood this wouldn’t show you a clean current state. It would help you hold the accumulated layers of a project in relation to each other.
I started writing when I was young enough to do it without wondering whether I was good at it. The summer vacation book my father typed out—I wasn’t performing for anyone. I had things to say, and writing was the place they could go.
Something happened between then and the analyst’s office. I won’t trace the whole arc here. What I can say is that the work of the past several years—the analysis, the methodology development, the long-form projects, the building—has been, in different registers, the same project: finding my way back to the kind of writing that is return, not production.
Palimpsest is a tool I needed before I could make it. An attempt to build, in software, a space where that return becomes possible: unhurried, layered, held.
Not faster. Better contained.
Palimpsest is in Early Access, which means the people using it now are not just users—they’re thinking partners in an ongoing question: what does a better container for writing look like? What would it need to hold that it doesn’t yet? I don’t think I can answer this alone. The container I needed was specific to my projects, my patterns, my particular ways of getting stuck. Yours will be different.
If you’re someone who writes seriously—who has felt the texture of being inside something unfinished—I’d like to think about this with you.
Palimpsest is in Early Access for macOS and Windows. Available on Gumroad.
As someone with so many notes across documents and platforms, this was the first writing tool that finally spoke to me. I'm excited to start reorganizing my drafts and notes
Your article is inspiring as always (which is not an empty praise – it means I will add some comments in the previous texts, when my thoughts take shape, as I am thinking now how cultural difference and similarity has been influencing my work and writing). I am in for Palimpsest :) Will speak back in several days, when I have more time to dive into it.