Reproduce without depoliticizing
There is an easy way to read Walter Benjamin, and it does not help much: turn him into a sad prophet of loss.
The work had aura.
Technique arrived.
Everything became poorer.
That reading is too comfortable. Benjamin did not write only a lament for what mechanical reproduction destroyed. He also saw that reproduction changed the political field of art. It moved works, publics, authority, distance, access and power. Technique was not neutral, but neither was the old aura innocent.
That is why Benjamin still helps when thinking about books, AI and independent publishing.
Technique does not decide by itself. But every publishing model decides which part of technique is allowed to rule.
The problem is not whether an editorial project uses technology.
It always does.
The problem is what apparatus is built around it.
Aura was not the whole point
Benjamin’s essay on technical reproducibility is often reduced to the loss of aura. But the more useful question is not only what is lost when a work is reproduced. It is what becomes possible, who gains access, what forms of control appear and how the relation between work and public changes.
When the work circulates, it loses one distance and gains another field of dispute.
The aura can be oppressive. It can protect hierarchy, distance, ritual, scarcity and the authority of those who decide what deserves reverence. Reproduction can democratize. It can bring works to other hands, other classes, other rooms, other uses.
But reproduction can also industrialize attention.
It can flatten differences.
It can turn circulation into administration.
The political question is not reproduction yes or no.
It is reproduction under what relations.
That is the part that still matters for AI. The model can reproduce language at a scale that would have looked absurd a few years ago. It can multiply variants, covers, blurbs, summaries, posts, translations and administrative language. The Benjaminian question is not whether multiplication is impure. It is whether multiplication opens a public, or whether it merely feeds a machine that needs constant material to circulate.
Reproduction can democratize access.
It can also saturate attention until access becomes meaningless.
That ambivalence is the center of the problem. The same technical capacity that lets a small publisher reach readers without a large distributor can also flood those readers with disposable material. The same tool that lowers the cost of translation can make translation invisible and badly paid. The same automation that reduces administrative burden can normalize the idea that editorial care should happen faster, cheaper and with fewer people.
Technique opens a field.
Politics decides how that field is organized.
The book was never outside technique
The book is not pure spirit in paper form.
It is a technical and social object: manuscript, editing, typesetting, design, correction, printing, binding, storage, transport, distribution, bookstore, price, review, reader, return, remainder, archive.
The book is not only text. It is circuit, support, cost, delivery, reading, memory and return.
Robert Darnton’s circuit of communication helps here because it prevents the sacred view of the book from erasing its material life. A book is never only what the author wrote. It is also the chain that allows that writing to reach someone else.
That chain can dignify the work.
It can also exploit it.
The question for an independent publisher using AI is therefore not whether the process remains untouched by technique. It cannot. The question is whether technique is used to make the chain more legible, fairer and more attentive, or whether it is used to accelerate opacity.
This is why the old distinction between “literary” and “technical” work fails. A bad contract changes literature. A late payment changes literature. A warehouse problem changes literature. A distributor’s discount policy changes literature. A metadata field can make a book disappear. A badly designed cover can misrepresent a voice. A spreadsheet can decide that a risky book will not exist.
The circuit is not outside the work.
It is one of the ways the work enters history.
This is also why an independent publisher cannot define independence only by ownership size. A small publisher can reproduce the logic of a large platform: opacity, speed, extraction, generic language, unpaid enthusiasm, symbolic prestige used as salary. Independence has to be a practice inside the apparatus.
Who understands the costs?
Who sees the sales?
Who can question the schedule?
Who decides whether a manuscript needs time rather than acceleration?
Those questions are editorial, not administrative leftovers.
The author as producer, not content supplier
Benjamin’s “The Author as Producer” remains useful because it shifts the question from opinions to apparatus. It is not enough to publish critical texts if the publishing machine treats the author as raw material.
It is not enough to publish critical texts if the apparatus still treats the author as input.
The contemporary vocabulary says content.
That word should make us suspicious.
Content is what fills a container owned by someone else. It is adaptable, extractable, searchable, sliced, monetized, summarized and replaced. An author is not a content supplier if the editorial relation is serious. An author is a producer of form, risk, language, time and position.
An AI-assisted publisher has to be especially careful here. If AI lowers the cost of producing pages, the temptation will be to ask for more pages, more variants, more covers, more summaries, more posts, more speed. The author can become one input among many in an automated pipeline.
The opposite model is possible.
Use automation to reduce bureaucracy.
Do not use it to reduce the author.
That means the author should understand the apparatus. Not every technical detail, but the logic of the relation: how the book will be edited, priced, printed, distributed, promoted, accounted for and archived. The editorial promise cannot be “trust us, the system handles it.” That phrase is exactly where power hides.
If AI helps produce reports, budgets, schedules or revisions, the gain should be shared as clarity.
Not kept as managerial advantage.
AI as a new administration of language
AI is not only a tool that writes.
It is a new administration of language.
It can draft contracts, summarize manuscripts, generate catalog copy, compare versions, produce metadata, propose covers, translate, detect inconsistencies, write social posts, organize emails, calculate budgets, assist proofreading and build workflows.
AI can organize the table. It should not sit at the head of it.
That can help a small publisher survive. Bureaucracy consumes time that could be used for reading, editing, talking with authors, selling books, working with bookstores and building community. If AI makes the administrative layer lighter, it can humanize the editorial relation.
But the same tools can also do the opposite.
They can replace reading with scoring.
They can replace editing with smoothing.
They can replace judgment with optimized similarity.
They can produce a neutral language that removes the very strangeness for which a book deserved to exist.
The issue is not the tool.
It is governance of the tool.
Governance here does not require a bureaucratic fantasy. It can begin with simple editorial rules: no machine-generated passage enters a book without human reading; no source suggested by a model is used without verification; no authorial voice is smoothed in the name of generic clarity; no cover, copy or metadata is treated as neutral filler; no cost saving is celebrated if it merely shifts invisible labor elsewhere.
Small rules matter because they become habits.
Habits become the apparatus.
Once the apparatus is built, it starts teaching people how to behave. Authors learn what kind of text is welcome. Editors learn what kinds of delay are tolerated. Designers learn whether the cover is a conversation or a template. Booksellers learn whether the publisher sees them as partners or endpoints. Readers learn whether the project asks for trust or merely for attention.
AI will not decide those lessons alone.
But it can harden them quickly.
The bookstore is not a channel
A bookstore is often treated as a sales channel.
That phrase is technically useful and culturally poor.
A bookstore is not a sales pipe. It is a slow machine of trust.
A channel transmits.
A bookstore mediates.
It recommends, slows down, remembers, argues, rescues, refuses, contextualizes and gives books a place in a community of readers. If an AI-assisted publisher bypasses that mediation because direct sales are more efficient, it may gain margin and lose culture.
The point is not to romanticize every bookstore.
The point is to understand that distribution is never neutral. A book sold only through platforms enters a different ecology from a book held, read, discussed and recommended by a bookseller. The circuit changes the book’s public life.
Presale: community before print run
For small publishers, presale can be an ethical tool or a disguised transfer of risk.
Presale does not manufacture community. It tests it before printing.
Used well, presale allows a project to measure real support, finance part of production, avoid waste, explain costs and involve readers before the book exists as stock. Used badly, it becomes a way of asking readers and authors to absorb uncertainty without transparency.
An AI-assisted publisher can manage presales more efficiently: forms, reminders, accounting, stock estimates, communication, segmentation, shipping lists. Good. But the efficiency must serve a clearer relation, not a more opaque one.
If the community finances risk, the community deserves information.
What will be printed?
What does it cost?
When will it arrive?
What happens if the goal is not reached?
What part goes to the author?
What part sustains the bookstore and publisher?
This also changes the reader’s position. The reader is not only a buyer waiting for a product. In a transparent presale, the reader becomes part of the conditions that allow a book to exist. That should not be abused with sentimental pressure. It should be honored with precision.
Community is not a marketing asset.
It is a relation that can be damaged.
A transparent economy for a cultural commons
Culture often hides money under noble words.
That usually means someone pays invisibly.
The author.
The proofreader.
The designer.
The bookseller.
The reader who waits.
The publisher who romanticizes debt.
A cultural commons is not sustained by economic fog. It is sustained by visible rules.
A small editorial project should make the economy more visible, not less. Costs, print runs, margins, discounts, royalties, returns, reserves, reprints, direct sales and bookstore sales should be understandable to the people involved.
This is not because literature should become accounting.
It is because hidden accounting often becomes hidden exploitation.
AI can help make that economy legible: dashboards, reports, stock control, royalty statements, budgets, reminders. But again, the question is political. Does the tool clarify the chain or merely make extraction more efficient?
Polanyi helps here because he reminds us that treating social relations as if they were only market commodities can destroy the fabric that makes exchange possible. In publishing, that fabric includes trust, time, symbolic value, patience, the slow formation of readers, the dignity of labor and the right of a book not to be reduced to immediate performance.
Illich helps from another angle: a tool should expand human autonomy, not create dependence on a system whose logic only specialists can understand. A convivial editorial technology would make authors, booksellers and readers less helpless before the apparatus.
That is the standard.
A transparent economy also protects desire. Many cultural projects fear that talking about money will kill the magic. Usually the opposite happens. What kills desire is suspicion: not knowing who is paid, what is owed, why a book costs what it costs, whether the author will ever see a statement, whether the bookstore discount was considered, whether the presale is financing a real project or a foggy promise.
Visibility does not remove conflict.
It makes conflict adult.
Reproduce without surrendering intelligence
To reproduce without depoliticizing means accepting technique without surrendering judgment.
It means using AI to automate what makes the editorial relation heavier, not what makes it human.
Automate repetitive administration.
Do not automate taste.
Automate reminders.
Do not automate responsibility.
Automate metadata drafts.
Do not automate the decision that a book deserves to exist.
Automate accounting where it improves transparency.
Do not automate the conflict over how value is distributed.
Benjamin helps because he prevents two lazy answers. The first is nostalgia: technique corrupts everything, so purity lies in refusal. The second is technocratic enthusiasm: technique improves everything, so politics can rest.
Both are false.
Technique reorganizes the field.
The task is to politicize that reorganization.
An independent publisher using AI should be judged not by whether it uses tools, but by what apparatus it builds: for authors, readers, bookstores, costs, rights, time, attention and responsibility.
Reproduction is inevitable.
Depoliticization is not.
The point, then, is not to build a handmade myth against the machine. It is to build a publishing practice where the machine does not govern the meaning of the work. A printer, a spreadsheet, a model, a design program, a payment processor, a distributor platform: all of them can be part of the table. None of them should become the hidden sovereign of the table.
Reproduce, yes.
Automate, where it frees time for judgment.
Calculate, where calculation prevents abuse.
Circulate, where circulation creates readers rather than only metrics.
But keep the politics visible.
That visibility is the real promise of a small editorial model. It cannot compete with the scale of large platforms. It can compete in the quality of its relations: with authors who know the terms, with readers who know what they are supporting, with bookstores treated as mediators, with tools governed by people rather than the other way around.
If AI helps that, use it.
If it hides that, stop.
The criterion is not novelty.
The criterion is whether reproduction leaves the common world more legible than before.
If it does, technique has been put to work for culture. If it does not, the publisher has only replaced one opacity with a faster one. That is not modernization. It is administration with better tools.
The difference will be visible in the chain: who understands more, who decides more, who depends less, who can answer when the book enters the world.
A publishing technology that cannot answer those questions may be efficient, but it has not become politically serious.
That is the editorial test.
Martín Álvarez
@unfalsoguru