Written 2 November 2026
Imagine a world in which everyone can write a book, nobody has the time to read one, and the only people still buying them are the people writing them.
It sounds like a joke about self-publishing. It is also a fair description of the direction of travel, and you don’t need a study to see it. You need a phone and an honest look at your own evenings.
Writing a book used to be hard. That difficulty did a job nobody noticed: it kept the supply of books roughly in line with the supply of people willing to read them. Now a passable manuscript takes an afternoon and a monthly subscription. Reading one still takes about ten hours, which is what it took in 1850.
One side of the deal got almost free. The other side didn’t move at all. Everything else here follows from that.
The cheap side
Start with the writers, because there are about to be a lot of them.
For most of history, a book was proof of effort. You couldn’t fake 80,000 words. Even a bad book told you the author had sat alone for months and finished something. That was why a book worked as a credential. The consultant with a hardback on the shelf behind them on a video call was saying, without saying it, that they had thought about this longer than you had.
That signal is gone. A book now proves that someone had an idea and access to a model. The “author of” line in a bio is drifting towards the status of “proficient in Microsoft Office”. People will keep writing it. Fewer people will be impressed.
The flood that follows is easy to predict. Every niche will fill with competent, well-formatted, forgettable books, written to rank on a marketplace search page rather than to be read. Plenty will be written by machines, recommended by machines and, increasingly, read by machines, because the next model needs something to train on. A library nobody visits, stocked by robots, for robots.
The human author isn’t replaced by this. They’re buried by it. Discovery was already the hardest part of publishing. Now the haystack grows faster than anyone can search it.
The expensive side
Now the reader.
Every generation is told its attention span is collapsing, and every generation is partly right. Socrates complained, in Plato’s Phaedrus, that writing would ruin memory, because people would lean on marks on a page instead of their own minds. In the eighteenth century, novels were blamed for rotting the brains of young women. Comics, television and video games all took their turn in the dock. The book outlived every one of them.
So be suspicious of anyone who says this time is different. That includes this essay.
Still, something has changed, and it’s less about willpower than competition. A book used to compete with other books, and with whatever was on at nine. Now it competes with every film ever made, every match being played anywhere on earth, every friend you’ve ever had, and a feed built by very clever people to make sure you never reach the bottom. All of it sits in your pocket. A book asks you to put that thing down and leave it there for an hour. Reading has quietly turned from a default into an act of refusal.
Then there’s the summary.
The summary problem
Search has already shown what happens when an AI answers the question before anyone clicks. People get what they came for and never visit the page that did the work. The page still exists. It just stops being read.
Books are next in line, and for a large slice of non-fiction the summary is good enough. Ask a chatbot what a bestselling business book argues and you’ll get the thesis in a paragraph. For many of those books, the paragraph is the book. The other 280 pages were anecdotes, padding and a chapter the publisher insisted on. The airport non-fiction shelf was always half blog posts with spines. AI has simply called the bluff.
Call that an audit. If an idea fits in a paragraph, it probably should have been one.
Fiction and serious argument are a different case. A summary of Middlemarch tells you who marries whom. It can’t give you hundreds of pages inside Dorothea’s head, which is the only reason anyone reads it. You can summarise the plot of a novel. You can’t summarise being changed by one, any more than you can summarise a holiday.
This is where the premise starts to wobble.
The case against the premise
The argument so far treats reading as information transfer, and information transfer is exactly what machines are better at. If books were only a slow way to move facts from one head to another, they’d deserve to die, and most would already be dead.
People don’t read novels for the facts. They read them for the length. The ten hours is the product. A long book is one of the few places left where one human mind holds yours for a sustained stretch, uninterrupted, with no algorithm deciding what comes next. In a world of endless, optimised, bite-sized content, that is rare, and rare things don’t usually get cheaper.
There’s an older counterargument too: most people never read books anyway. The mass reading public of the twentieth century, buying paperbacks by the million, was a historical blip, propped up by cheap printing and a shortage of alternatives. Before that, serious reading belonged to a minority. It may simply be going home.
That’s the future I find most convincing. Books won’t vanish. They’ll shrink back to the people who actually want them.
The classical music future
We already know what this looks like, because it happened to classical music.
Nobody thinks Beethoven is dead. His work is played somewhere every night. But classical music stopped being popular culture long ago and became a specialist pleasure, with its own institutions, its own devotees and a faint reputation for being good for you. It survived by becoming a minority taste, and the minority keeps it very much alive.
Reading could go the same way. A small, committed readership. Physical books as objects of care, on better paper at higher prices, bought partly to be owned. Vinyl came back once streaming made music weightless. The printed book may do the same once text becomes something a machine produces on demand. A shelf of paper becomes proof that someone chose this, specifically, out of everything.
And the reader becomes a type. Someone on a train with a novel, phone face-down beside them, will look the way someone writing a letter by hand looks now. Slightly eccentric, faintly admirable, obviously doing it on purpose.
Value from scarcity
Here’s the twist. If readers become scarce, reading becomes a signal.
When anyone can produce a book, the book stops meaning much. When almost nobody can be bothered to read one, having read it starts meaning a lot. The credential moves from the writing to the reading. The person who has read the whole thing, and can argue with it, will know what the summary crowd doesn’t. When everyone has the same AI summary, that’s one of the few edges left.
The same logic runs the other way for writers. A writer who earns ten hours of someone’s attention against that competition has done something hard. The flood of machine books will make the good ones easier to spot, the way a supermarket full of identical bread makes a decent bakery obvious. Readers will stop trusting the shelf and start trusting a name. Authorship will matter less as a badge and more as a promise: this person won’t waste your time.
That’s bad news for most authors and good news for a few. To be fair, publishing has always worked that way.
So is the premise true?
Partly. Everyone will be able to write a book. Most people won’t make time to read one. A great many books will be bought, or generated, mainly by the people who made them, and they’ll sit in an index, summarised by one machine for another.
The usual conclusion, that books are finished, doesn’t follow. The part of reading that was really information delivery is being automated, and good riddance. The part that was always about attention, patience and one mind meeting another is getting rarer, and so more valuable.
The future I’d bet on is stranger than the death of the book. Millions of books, written by people and machines, indexed and summarised for readers who will never open them. And somewhere in all that noise, a small number of people doing the least efficient thing left in the information economy: sitting still for ten hours with one other person’s mind.
They won’t be the majority. They never were.
Nobody wants to read your book. That’s exactly why the people who do will be worth writing for.