The Only Question

"Yesterday this painting was worth millions of guilders."

— Han van Meegeren, at his 1947 trial, on the canvas a nation had just learned was a forgery ¹Han van Meegeren forged paintings attributed to Vermeer; his Supper at Emmaus was hailed as a masterpiece and sold for ~$4M in today's money. Exposed after WWII, he was convicted of forgery in 1947; the same canvases lost nearly all their value once relabeled.

I. The hidden hands

In the Age of The Machine, we expend a great deal of effort certifying how books are made. We now have a trademarked mark indicating human authorship, defended like any other trademark and searchable in a public registry.²The Authors Guild "Human Authored" certification, launched in beta Jan. 2025 and opened to all U.S. authors in March 2026, is backed by a public registry and enforced as a trademark. There is a government office ruling on how much of a manuscript a person must have controlled to own it.³The U.S. Copyright Office holds that purely machine-generated material must be disclaimed, and that prompts alone do not give a user sufficient control over expressive output to count as its author. Copyright and Artificial Intelligence, Part 2: Copyrightability (2025). A complicated taxonomy can divide a work into fourteen contributor roles, and one can only imagine this number growing over time.⁴CRediT (the Contributor Roles Taxonomy), an ANSI/NISO standard, describes fourteen contributor roles and is designed to describe contributions rather than determine authorship. credit.niso.org Detectors can scan a paragraph to tell you whether a person or a machine produced it (with mixed results, to say the least).

Theseus Press declines almost all of this. We have studied The Machinery carefully, and we will explain our findings more clinically in a companion white paper. But the founding instinct of this press is to ask a reader a few basic questions:

Is this good writing?

Does it move you?

Will you urge a friend to read it too?

Everything else — which tools touched the page, at which stage, in what proportion, whether the em dashes were hand-typed (⇧⌥-) or an artifact of The Machine who "wrote" it — belongs to the making, and the making is ours, not yours. We think a book either earns your time and attention or it does not, and that this is a judgment you are fully equipped to make without a flowchart that tracks its lifespan from design to dissemination or an inventory of its building blocks.

II. What we will not ask

No book was ever the unaided product of a single mind transcribing pure originality onto the page. Behind almost every published work, you might find a developmental editor who moved the chapters, a line editor who rescued the best sentences, a researcher, an agent, a publicist, a copyeditor, and very often a husband or wife or friends who quietly suggested countless adjustments. And, of course, some of the most successful books on the bestseller list relied on co-writers and ghostwriters whose contracts preclude their inclusion on the cover.

One of the most gripping memoirs of our time (Open, "by" Andre Agassi⁵Andre Agassi's memoir Open (2009) was written with the Pulitzer Prize–winning journalist J.R. Moehringer. On the memoir and its making, see The New Yorker.) even dedicates a highly unusual postscript to the unique roles "writer" and "ghostwriter" played in getting the stories down on paper: a rare, honest, and revealing look behind the curtain. Turns out, there is something to see there, and sometimes it's worth paying attention to both the man behind and in front of the curtain.

But, even knowing details like these exist, we rarely deconstruct the books we love into their constituent parts or the complex processes that led to their assembly. We do not pull a bestseller off the shelf and demand to see the marketing budget, the editorial memos, the fact-checking invoices, the list of darlings the author was made to kill in draft after draft (after draft). We cherish instead the quaint and largely mythical image of the writer alone at the desk, month after month, conjuring the thing from nothing.

So the question this press refuses to entertain is: why should The Machine be any different? If the workshop or editorial team never diminished a book's validity, why should one more tool in the workshop? We will return to the one honest answer to that question — there is one, but it is not the answer most people give — after laying a little more groundwork.

III. The forgery

Here is the strongest objection to everything above, and we will not duck it.

It says: Provenance is part of Value. We claim to judge a work on its merits, but we plainly do not. Consider Han van Meegeren, who painted Vermeers so persuasive that the most eminent Vermeer scholar of the age pronounced one of them a masterpiece, and who sold them for the equivalent of millions.¹Han van Meegeren forged paintings attributed to Vermeer; his Supper at Emmaus was hailed as a masterpiece and sold for ~$4M in today's money. Exposed after WWII, he was convicted of forgery in 1947; the same canvases lost nearly all their value once relabeled. The moment they were unmasked as forgeries, the very same canvases — not a brushstroke altered — became near-worthless curiosities. The philosopher Nelson Goodman pressed the point to its sharpest form: even a forgery you cannot yet tell from the original makes what he called an aesthetic difference, because knowing changes how you look, and because you might one day learn to see the seam.⁶Nelson Goodman, Languages of Art (1968), distinguishes autographic arts (painting), where even an exact duplicate is not genuine, from allographic arts (literature, music), where any accurate copy is fully the work. He also argues an undetected forgery still makes an "aesthetic difference." If provenance were truly nothing, the forgery (or authorized copy) and the original would be worth the same. But they never are.

We concede this objection completely. For paintings.

A book is not a painting, and this makes all the difference. Nelson Goodman drew a line between two kinds of art. Painting and sculpture are autographic: the work is a single physical object bound to the hand that made it, and even a perfect duplicate is not the genuine thing. Literature and music are allographic: the work is not any one object but a pattern — a sequence of words, a score — and every faithful copy is equally and fully the work. There is no forgery of a known text in the way there is a forgery of a Vermeer. You do not read Tolstoy's pen-strokes; you read Anna Karenina, and your paperback is as much the novel as the manuscript was. A book has no aura of the singular object to be counterfeited. The words on the page are the entire work. If Hunter S. Thompson were caught with a page of The Great Gatsby half-typed-out on his Selectric II, no one would've accused him of forgery. Perhaps especially because it was Thompson, they'd've taken the story as he told it: he just wanted to see what it feels like to write a masterpiece.

So the forgery objection, which rests entirely on the canvas's physical singularity, does not cross over to the page. For a text, "the work stands on its own" is not a brave overreach. It is very nearly a complete description of what a text is.

The honest counterargument — and we do owe it a hearing — is that art may be more than the marks: it may be the record of a human achievement, a performance. And a forgery lies about whose performance it was.⁷Denis Dutton, "Artistic Crimes: The Problem of Forgery in the Arts" (1979), argues aesthetic value includes the human achievement a work represents — which can justify revising one's judgment upon learning a work is a forgery. We take this seriously enough to build our entire foundation on it. There is a human achievement in every book we publish, and we guarantee it. What changes when a machine drafts a passage is not the words you read but the story of how they arrived. And here is our wager, stated plainly so it can be argued with: we believe the truer judgment is the one you make before you are told that story, not after. The instinct that made van Meegeren's canvas worthless the instant it was renamed — though the paint never changed — is precisely the instinct we will not feed. We will not tell you the story of the making, because we would rather you judge the story in the book than the story of the book.

IV. The death of the author, taken at its word

This press did not invent the idea that a finished work is severed from its maker. In 1946, Wimsatt and Beardsley stabbed the author's intention in the back, calling it "neither available nor desirable" as a standard for judging a work; once published, the poem belongs to the public and is to be judged as the public's object, not as a window into a private creative mind.⁸W. K. Wimsatt and Monroe Beardsley, "The Intentional Fallacy," The Sewanee Review 54 (1946): the design or intention of the author is "neither available nor desirable" as a standard for judging a literary work. Two decades later Roland Barthes twisted the knife, declaring that meaning is made not by the author but by the reader, and that to free a text we must let its author die.⁹Roland Barthes, "The Death of the Author" (1967; collected in Image–Music–Text, 1977): authorial intention and biography are irrelevant to interpretation, and meaning is constituted by the reader.

You may protest: these arguments pertain to human authors, to whether a poet's diaries or stated aims should govern how we read the poem. We simply extend these principles by a single step.

If the human author's intentions were never the true measure of the work, then the question of which instrument first arranged the words simply puts us one degree further from the supposed death. The hand that made the marks was already declared irrelevant to the verdict. If the first marks were Machine-made, and subsequent changes Human, would the author be more or less relevant to the final work?

For sixty years critics have pronounced the author dead. But we are not among them. We take from Wimsatt and Beardsley the sturdier, narrower claim — that the author's intention is not the standard by which the work is judged — and we leave Barthes his funeral. The author is not dead. But neither is the author's effort the measure. The finished work is.

Where others dissolved the author into the abstraction of "language" and gave the work to the reader, this press does the opposite. We insist on a named author who governs and who answers. We hand him the keys to his own workshop. An author alive enough to sign the work and stand behind it has earned the right to keep the means of his making to himself.

V. What stays

Opacity about process is not opacity about responsibility, and the two must never be confused.

Every book Theseus publishes is governed by a human being who stands behind it. This is the one honest answer to the question of Section II — the reason The Machine genuinely is different from the ghostwriter, so we face it head-on. The hidden hands behind an ordinary book are hidden humans: people who can be named, credited, sued, and held to account; people who checked or culpably failed to check.

A machine cannot occupy those roles. It cannot be meaningfully responsible for the accuracy, integrity, or originality of a work, and it is for exactly this reason that the professional bodies of publishing refuse to list one as an author.¹⁰The International Committee of Medical Journal Editors and the Committee on Publication Ethics hold that machines cannot be listed as authors because they cannot take responsibility for a work's accuracy, integrity, and originality.

But when we choose not to disclose how a given book was made, understand what we are not doing. We are not saying no one made it. We are not saying no one answers for it. A named human governs every word and stands behind every claim. We may withhold the method; we never withhold the author. We may decline to describe the process; we never decline responsibility for the result. That is the warranty behind the silence.

Another thing we keep: Where a work trades on its own truth, honesty about that truth survives, because there the claim is the product. Theseus intends to publish fiction and poetry — forms that make no promise to have happened, that ask only to be worth your time. But the instant a book asks you to believe it is true, the contract changes, and a reader's interest in how it was made stops being nosiness and becomes the heart of the bargain. James Frey's A Million Little Pieces moved countless readers; the scandal, when it came, was not that the prose had failed but that a work sold as a true memoir had been substantially invented.¹¹James Frey, A Million Little Pieces (Doubleday, 2003), was revealed in January 2006 to contain substantial fabrications; the publisher settled a class-action refund suit. The harm lay in its having been sold as a true memoir. The betrayal was a broken promise, and the promise was authenticity. We will not make that kind of promise — to Oprah or anyone else — and we will not break the promises we do make.

So if this press delves into history through fiction — the alternate timelines, the branching paths, the what-ifs and might-have-beens — we will do it in the open, with a wink at the mechanism, because there the visible seam is part of the pleasure. We hide the making everywhere except where revealing it is the art.

VI. The Machine is not the enemy

Our enemy has a name now. In 2025 the dictionary made it official: slop, the word of the year.¹²Slop was named Merriam-Webster's Word of the Year for 2025; the term, popularized by Simon Willison in May 2024, denotes low-quality digital content mass-produced in quantity — defined by indifference and low effort rather than automation as such.

slop|noun

1 a : digital content of low quality that is produced usually in quantity by means of artificial intelligence

"An AI-enabled social media future also raises concerns around deterring AI slop — mass-produced, junky and superficial content that clogs up the web and social media accounts." — Katelyn Chedraoui "Slop can now be found anywhere, from unsettling images on Facebook of Jesus fused with prawns to poorly written Kindle books…" — Tess Bennett

Read that definition carefully and you will find it does not condemn the presence of a machine. It condemns the absence of care — content mass-produced and indifferent, optimized to be consumed and forgotten, poured out because it is cheap, plausible, and fast.¹²Slop is defined by indifference and low effort rather than by automation as such. It is this flood that forced a respected fiction magazine to close its doors to submissions.¹³The science-fiction magazine Clarkesworld closed to new submissions in 2024 amid a flood of machine-generated writing.

We hate slop; we welcome assistance.

We insist, against the panic of the moment, that these are two distinct issues. A machine cannot publish slop. It can only be used to do so. Published slop is not technological failure but a human one: a failure of judgment, care, taste, and responsibility — or some combination of them all.

This is why we will not pretend an AI Scanner can do our reading for us. The detectors do not work: they mislabel the innocent and wave the guilty through, and they punish hardest the writer whose English is a second language, flagging more than half of such essays as machine-made in one careful study.¹⁴Liang et al., "GPT detectors are biased against non-native English writers," Patterns (2023): seven detectors misclassified more than 60% of essays by non-native English writers as machine-generated. We decline to outsource judgment to a tool that cannot judge. Our entire wager is that care leaves deeper marks on the page than any disclosure statement can reveal, any detector can expose, or any process log can record — and that the only instrument fit to find them is a (need it be said? Human) reader.

So let us read.

VII. A personal note, from the Founding Publisher

Now let me drop the editorial we, because I owe you the truth about why this press exists.

I believe there are two kinds of writers:

1. Those who love to write,
2. Those who hate to write but love to have written,
3. Those who can't start writing, but remain convinced their first draft would have been perfect.

This third kind of writer also can't count. Until recently, I was one of them.

For over twenty years I've cheekily described myself as a "non-practicing writer."¹⁵"Non-practicing writer" was, in my mind, always a nod to a scene from Reality Bites.

I did not lack ideas, or thoughts worth sharing. I lacked a way across the single worst distance in the craft: the distance from the blank page to the first bad draft. The cursor, the desk, the nothing that would not become something. I could revise anything you put in front of me (including — as I assume we all can, as we read along — published works by some of my favorite authors). I simply could not generate the something to revise.

That is no longer true, and the reason is a sloppy first draft that I did not have to bleed for. With one in front of me — however it got there — I find I can do the thing I always could and always loved: shape it, cut it, fight it, rewrite it line by line until it sounds like no one alive but me.

And the unease — the stubborn fear that it cannot really be mine unless I suffered for those first five thousand words — I have come to recognize as the very myth this manifesto hopes to dispel. It is the romance of suffering: the belief that the worth of a thing is the price of its making. I do not believe that about anyone else's books. I refuse to believe it about my own. Ownership is not conferred by the agony of the first draft. It is earned by the judgment that governs the last. The research happens to agree — a text becomes yours in proportion to how much of yourself you put into shaping it¹⁶Fiona Draxler et al., "The AI Ghostwriter Effect," ACM Transactions on Computer-Human Interaction 31, no. 2 (2024): a person's sense of ownership over a text rises the more they shape it. — but I did not need the research. I needed permission. This is me writing it, for myself first and for anyone else who has been kept from the page by the same lie.

I keep the unease, though, as an instrument. When a line still nags, it is almost always because I have not yet made it mine — The Machine's cadence still showing through, a choice that is its and not yet ours. That nagging is the same faculty that recoils at slop.

I have learned to trust it to tell me when I am finished. I no longer let it tell me whether I am allowed to begin.

VIII. Postscript: The ship

The Ship of Theseus began as its original planks. But when all the pieces are replaced one by one, the thing that remains — the form, the seaworthiness, the care of the hands that keep it whole — is still the Ship of Theseus. It is no longer the timber it started with. It is what it has become and who keeps it true.

A book is the same. It is not the first draft, not the first hand, not the first arrangement of words to fall onto the page. It is the final form: governed by a human, authorized by a human, and handed to you to judge. We will not tell you how it was built, and we will not apologize for the silence, because the silence is a kind of respect — for the work, and for you.

We will only ask the question that has, when you strip everything else away, always been the only one.

Is it worth the voyage?

Read it, or don't. That is yours to decide.