Nobody inside a cradle names it. The word for the first age of print was coined 189 years after the press started work. This is an attempt to name ours while we are still in it.
LIVING DOCUMENT · VERSION 0.1 · OCTOBER 2026 · EVERY NUMBER IS SOURCED IN THE LAST TAB
00 / Start here
The reality doesn't come in the terms provided.
Ask ten people what DOGE was and you'll get ten different pictures. The stated mission. The cable-news version, in at least two flavours. The one your feed built for you. The one passed around at work or at the kitchen table. Most people hold one of those pictures with real confidence, and almost none of them saw the thing itself.
That isn't new. In 1922 the journalist Walter Lippmann argued that the world is too big for anyone to see directly. We act on the pictures in our heads, and those pictures reach us through whoever runs the press.18 Six years later Edward Bernays, who had helped sell the First World War to the American public, described the people who shape those pictures as "an invisible government which is the true ruling power of our country."18
More than five centuries before that, a Venetian scribe named Filippo de Strata asked the Doge of Venice, his city's ruler, to ban the printing press.6 The shared spelling is a coincidence. The fight isn't. It's the same one every time: over who gets to make the copies, and so who gets to paint the pictures.
Every new medium reopens that fight. Print did. Broadcast did. The feed did. What's arriving now is the biggest reopening since Gutenberg, because the pictures will no longer be printed or broadcast to everyone at once. They'll be painted in conversation, one person at a time.
00 / The concept
Knowledge that answers back.
Around 1450, a metalworker in Mainz built a way to cast letters in metal, set them in a frame and press them onto paper. Before that, every book in Europe was copied by hand. After it, the same text could be made hundreds of times over, identical, at a falling price. Over the next two centuries that one change helped break a church, build modern science, invent the newspaper and redraw the map of Europe. The people living through its first fifty years had no word for what was happening to them.
In the 1990s we were promised the world at our fingertips, and the internet delivered a faster shelf. More books, more pages, more search results, all still waiting to be fetched and read. The relationship stayed the same: you go to the source, you open it, you work out what it means.
What arrived in November 2022 is different in kind. Knowledge started answering back. You ask, it replies. You question the reply, it adjusts. Plato's complaint about writing, that a text can't answer when you question it, held for 2,400 years. It stopped holding. That is Incunabulum v2.0: the second cradle.
Why "cradle"
Incunabula is Latin for swaddling clothes, the bands wrapped around a newborn, and by extension the cradle itself: a birthplace, an infancy, a beginning. The root is cunae, cradle. When Bernhard von Mallinckrodt wrote prima typographicae incunabula in 1639, he meant the first infancy of printing, the years when the new art was still swaddled and nobody yet knew what it would grow into. Only later did the word attach to the books themselves.1
An incunabulum is something made while its medium is still in the cradle. Incunabulum v2.0 says we are back there. The infant this time is a machine that answers, and we are the ones standing over the crib, not yet sure what it will become.
1639The year someone finally named the first cradle: Bernhard von Mallinckrodt, writing in Cologne.SOURCE 1
282European towns with a printing press by the end of 1500, fifty years after Mainz.SOURCE 2
~30kDistinct editions printed before 1501 that survive and are catalogued today.SOURCE 2
1.2BWeekly users of ChatGPT alone, reported September 2026. One AI product, four years old.SOURCE 17
Why the old record matters
Print is the best-documented information shock in history. We can count its editions, its towns, its prices and its casualties. We know which cities grew because of it (cities with presses by 1500 grew about 60% faster over the next century4), what it did to the price of a book (down about 2.4% a year for more than a hundred years5), and who paid for it. We also know the order things happened in, and how long each wave took to arrive.
That order is the useful part. A new medium does not hit all at once. First the makers, then the money, then the protests, then the misuse, then the native form, then the voice that shakes an institution, then the institution's counter-move, then the long war over who decides, and only then a name.
The method, stated plainly
We read the print record as a sequence, then lay it over our own decade. The sequence is the claim. The calendar is a hypothesis.
To turn one into the other we use a single lens: our cradle runs about five times faster than theirs. A year of print history equals roughly ten weeks of ours. Five isn't a law. It's a dial, set where the early markers line up, and you can turn it yourself in the Horizons tab. Run at five, the lens puts Mallinckrodt's naming of the first cradle at around 2060. That's where this page ends.
Everything here is labelled for what it is. Where a figure is a scholarly estimate, we say so. Where a horizon is a guess, it carries a confidence mark. This is version 0.1, made to be argued with and added to.
01 / Snapshot: Mainz plus twenty-five years
The world in 1475.
Pick any year in the first cradle and you'll find people certain they understood the new thing, and almost all of them wrong about what mattered. 1475 is a good year to stand in. The press is twenty-five years old. It has already had its first crash and its first protest. It has not yet had its first atrocity or its first revolution. Under the lens on this page, our own 1475 lands about a year from now.
Knowledge at the speed of a hand
Until the 1450s, a European book was a manuscript. It was copied letter by letter, first mostly in monasteries, later by professional scribes working for universities and wealthy lay buyers.3 Every copy was a fresh chance for error. A book cost weeks of a worker's wages.5 Owning one was a mark of wealth or office, and the people who held the copies held the authority to say what they meant.
The inventor already lost
Gutenberg's Bible, roughly 180 copies, went on sale around 1455.8 That November his financier, the merchant Johann Fust, sued him to recover his investment. The court sided with Fust. Gutenberg had to hand over his equipment, and Fust went on to run a successful press with Gutenberg's former assistant, Peter Schöffer.8 By 1475 the man who made the thing had spent twenty years watching other people profit from it. Printing sat outside the old guild rules, and it was one of the first industries in Europe organised as competing firms.5 The money followed the organisers.
The first crash had already happened
Printing spread fast and unevenly. Venice got its first press in 1469. Within about four years a local scribe complained the city was "stuffed with books." Printers had overestimated demand. In the crisis of 1472 and 1473, presses that had printed too much went bankrupt.6 The medium didn't die. The weaker shops did, and the presses kept spreading.
The first protest was on the desk of Venice's ruler
"This glory pertains to you, Doge: to lay low the printing-presses."Filippo de Strata, Benedictine scribe, Venice, 1473-746
Filippo de Strata, a monk and scribe at Murano, wrote that the pen is a virgin and the printing press a whore. He called printers asses and asked Nicolò Marcello, the Doge of Venice, to ban printing outright.6 His moral case was loud, but the money case was real: printers were undercutting the price of handwritten books. The Doge ignored him. The petition still tells us something useful. It shows exactly where the guarded value sat, and who felt it first.
The new thing still wore the old thing's clothes
Early printed books were made to look like manuscripts, in layout, in type, in hand-painted initials. Gutenberg first printed his red headings the way a scribe would have rubricated them.8 Books kept looking like this for years after 1500. The first truly native form, Aldus Manutius's small, portable italic octavo, didn't arrive until 1501.25 In 1475, print was still pretending to be handwriting. Our chat windows are doing the same thing now. They're dressed as search boxes and assistants and pen pals because those are the clothes we had.
The scribes were shrinking, slowly
Manuscript production in central Europe fell by about half between 1465 and 1485.3 Half is a lot, but it took twenty years, and scribes were still commissioned and still in demand decades after the first presses. One Cretan scribe, Zacharias Callierges, gave up the pen to become a printer, then picked it back up when his press failed.6 Disruption on the ground looked less like a cliff and more like a tide going out, with people wading back and forth.
What nobody in 1475 could see
Still to come
Year
Years after 1475
The Malleus Maleficarum, a printed witch-hunting manual
1486
11
Presses in 282 towns
1500
25
Luther, and the first mass pamphlet war
1517
42
Copernicus publishes a sun-centred cosmos
1543
68
The first Roman Index of forbidden books
1559
84
The first newspaper
1605
130
The Thirty Years' War begins
1618
143
Someone names the era: incunabula
1639
164
Every one of these was already implied by the press in 1475. Not one was visible from inside it.
02 / Ledger
What collapsed. What grew. Who paid. Who thrived.
The printing press is usually told as a story of progress. It was that. It was also a story of bankruptcy, persecution and war, and the two stories ran at the same time, in the same towns, out of the same shops. But before the ledger, the lesson that matters most for us: what happened to the place that said no.
Saying no doesn't stop it
1493A Jewish press in Istanbul prints an incunable. The technology is inside the Ottoman capital within forty years of Mainz.SOURCE 15
1727The first Muslim-run press printing Turkish in Arabic script is permitted, by sultan's decree, for non-religious works only.SOURCE 15
~280yearsFrom Mainz to the first Ottoman press in the empire's own script and majority faith.SOURCE 15
~30k vs ~2 dozenEditions printed in Europe before 1501, against the titles the Müteferrika press produced in its entire run.SOURCES 2, 15
The Ottoman Empire was the great power on Europe's doorstep, and it had the press almost from the start. Jewish printers in Istanbul produced an incunable in 1493. An Armenian press opened there in 1567 and a Greek one in 1627. What the empire didn't do, for nearly three centuries, was let the press print in Arabic script for its Muslim majority. The first such press, run by Ibrahim Müteferrika, was licensed in 1727 by decree of Sultan Ahmed III, restricted to non-religious works. Its first book, a dictionary, appeared in 1729.15
Historians still argue over why it took so long. The reasons put forward include imperial decrees, the opposition of calligraphers who saw both their sacred art and their living at stake, a thin market for printed books, and the real difficulty of setting flowing Arabic script in metal type.15 None of those reasons was foolish. The calligraphers were guarding something real: copying the word by hand was an act of devotion, and print threatened it.
Look at the scale once it finally arrived. By 1500 Europe had printed around 30,000 editions in 282 towns.2 The Müteferrika press, in its whole active life, produced around two dozen titles, some in runs of 500 copies. After 1757 it stood idle until 1784. Ottoman printing only began in earnest in the nineteenth century.15
And look at what the press carried when it came. The machines and type had to be bought from abroad. Among its most important books were a geography that drew on Western sources and Müteferrika's own treatise urging the army to reform along European lines.15 The medium arrived already shaped by the people who had adopted it first.
Holding a new medium at arm's length doesn't stop it. It only decides who else gets to shape it first, and whose ideas it carries when it finally comes through your door.
That's the lesson for Incunabulum v2.0. The bans are already being written: publications that refuse any work touched by AI, institutions that forbid it, governments weighing licences. Some of what they guard is worth guarding. But the record is clear about what refusal buys. Not safety from the new medium. Only a later, smaller, borrowed version of it, built to someone else's specifications.
The ledger
What collapsed
The scribal monopoly on copying
Before 1450, every copy of every text passed through a human hand. Monastic copyists, and later professional scribes working around the universities, were the bottleneck all written knowledge moved through. Print didn't abolish them overnight. Manuscript output in central Europe roughly halved between 1465 and 1485, a twenty-year squeeze rather than a single blow. By volume the press overtook the pen within fifty years: one estimate puts printed copies from 1454 to 1500 near 13 million, against about 5 million manuscripts across the whole fifteenth century. The figures are uncertain, but the direction isn't. What collapsed wasn't writing. It was the monopoly on making copies, and with it the scribes' power to decide which texts survived.3
The price of a book
A manuscript cost weeks of a worker's wages, so books belonged to institutions and the rich. Print brought the price down slowly and relentlessly: about 2.4% a year for more than a century, or 1.7% after adjusting for quality. Competition did most of the work. When one more printing firm opened in a city, prices there fell by about a quarter. In time a book cost less than a day's wage. This is the quiet hinge for everything else in the ledger. Once texts were cheap, they could reach people who had never owned one.5
The Church's control of interpretation
For a thousand years the Latin Church had been the main keeper of books and the final word on what scripture meant. Print didn't challenge that by itself. Luther did, using print. In 1517 he was an unknown monk and professor in a town with one poor press. Within five years he was the most published author in Europe. German printers turned out some 4 million copies of works between 1520 and 1525, most of them short pamphlets in German, and Luther and his supporters outpublished their opponents nine to one. Catholic replies were printed too. They just didn't sell. Once anyone could print an interpretation and anyone could buy one, a single authority over meaning couldn't hold.10
A shared Western Christendom
The split didn't stay on paper. For more than a century Europe fought over who had the right to decide belief, and the cost peaked in the Thirty Years' War (1618-1648): an estimated 4.5 to 8 million dead from battle, famine and disease. Print didn't cause those wars on its own. Dynasties, money and power did much of that. What print did was make the division fast, public and permanent. The Peace of Westphalia in 1648 let each ruler set the religion of his own territory. That was the end of one Christendom and the start of the modern state system.14
What grew
Cities
Economist Jeremiah Dittmar compared European cities that adopted printing before 1500 with similar cities that didn't. Between 1500 and 1600 the printing cities grew about 60% faster. They weren't already growing faster before the press arrived. The result also holds when distance from Mainz is used as the test, which rules out printers simply picking promising towns. City growth is the best measure of economic growth we have for that era, so this is as close as the record gets to proof that print paid. The edge faded after 1600, once print was everywhere.4
Commerce
Not all printing paid equally. Tracing the output of thousands of printing firms across hundreds of cities, Dittmar found the strongest link to local growth came from business education: commercial arithmetic, bookkeeping, the practical methods of trade. A merchant's clerk could learn from a cheap manual what had once taken years of apprenticeship. The practical book, not the grand one, moved the money.4
Science
A hand-copied text drifts, because every scribe adds new errors. A printed edition is the same in every copy, so scholars in different cities can look at the same table, find the same mistake and publish the same correction. Historian Elizabeth Eisenstein called this "typographical fixity" and argued it was a prerequisite for the rapid advance of learning. Her example: astronomers in Copernicus's lifetime didn't have much better observations. What changed was that they could compare, correct and build on one another's work. Maps, star charts, chronologies, law codes and bibliographies were all transformed before 1600. Copernicus published in 1543.12
Languages and news
Incunabula survive in 18 languages, and print gave everyday languages a fixed, shared written form. News became regular. In 1605 Johann Carolus, who had made his living selling handwritten newsletters to rich subscribers, bought a press because copying by hand took too long. His weekly Relation is counted as the first newspaper. Reading itself spread slowly: literacy in north-west Europe rose fastest between 1600 and 1800, a century and more after the cradle closed. The press made reading possible for everyone long before it made it normal.21316
Who it was hard on
Scribes and calligraphers
Copying was a skilled trade and, for monks, a devotional act. Print attacked both the living and the meaning. Filippo de Strata's petition and Trithemius's In Praise of Scribes make the case they put: the press corrupts, paper won't last, copying by hand disciplines the soul. In the Ottoman lands calligraphers were the main opposition to printing even after it was permitted. Many adapted. Some became printers, some kept a luxury trade in commissioned manuscripts, and some, like Zacharias Callierges, went back and forth. The ones who lost most were the ordinary copyists whose value was simply volume.6715
The inventor
Gutenberg spent years and borrowed heavily to build his press and print his Bible. In November 1455, as the Bible neared completion, his financier Johann Fust sued to recover his loans. The court found for Fust, who took the equipment and a share of the Bibles, and went on to run a successful press with Gutenberg's former assistant Peter Schöffer. Gutenberg went on printing on a smaller scale. The money went to the organisers. The inventor of a medium is rarely the one who profits from it.8
The early printers
A press needed a large investment in type, paper and labour before a single copy sold. When Venice's printers misjudged demand, the result was the crisis of 1472 and 1473: too many books, a shrinking market, bankruptcies. The survivors consolidated and the medium kept spreading. Being early meant carrying the risk that later winners didn't.6
The accused
In 1486 the inquisitor Heinrich Kramer published the Malleus Maleficarum, a handbook arguing that witchcraft was real, that women were especially prone to it, and how to try, torture and convict them. Leading theologians condemned it within a few years. The press didn't care. It ran to about 28 editions by 1600. Across Europe between 1450 and 1750 there were roughly 90,000 to 110,000 witch trials and 40,000 to 50,000 executions. About three quarters of the accused were women, many of them poor, old or isolated. A study of 553 central European cities found that each new edition of the Malleus was linked to a rise in trials in the city where it was printed. It was the first great demonstration that a confident, wrong text, copied without limit, kills people.9
Ordinary people in the long war
Most of the dead in the Thirty Years' War weren't soldiers. They were people who died of famine and disease as armies stripped the land around them. Some parts of Germany lost more than half their population. They paid for an argument carried on largely in print by people who never saw the fighting.14
Who thrived
The organisers and their backers
Printing sat outside the old guild rules. It was one of the first industries in Europe organised as competing firms. That rewarded capital, organisation and distribution more than invention. Fust and Schöffer did well where Gutenberg didn't, and the shops that could pay for type, paper and a way to sell the books outlasted the ones that couldn't.58
Clear writers in the common tongue
Luther's 1518 Sermon on Indulgence and Grace was about 1,500 words of plain German in short paragraphs. Printers loved pamphlets like it: cheap to set, quick to reprint, little money at risk. Lucas Cranach gave Luther's title pages a look people could recognise at a glance, with "Wittenberg" in bold type. In 1517 the town had one poor press. Over the century it became Germany's largest printing centre. Luther never earned a pfennig from it. He thrived because he was the first to write natively for the new medium: brief, vernacular, made to be reprinted.10
Readers
For the first time people of ordinary means could own texts. Many who had never owned a book, or could afford only one or two, could now buy dozens of pamphlets. The reader stopped being someone who was read to and became someone who could check for themselves.10
Merchants
Cheap manuals spread arithmetic, bookkeeping and commercial practice, and the cities that printed them grew fastest. The practical class, the people who ran trade rather than courts or cathedrals, gained the most from knowledge that was suddenly cheap and identical everywhere.4
How long it took
Marker
Year
Years after 1450
The first big artifact goes on sale
1455
5
The first glut and crash
1472-73
22
The first mass-printed atrocity manual
1486
36
The infrastructure is everywhere (282 towns)
1500
50
The first native-form voice shakes an institution
1517
67
The science dividend arrives in public
1543
93
The institution's formal counter-move
1559
109
Print becomes periodic: the newspaper
1605
155
The hardest wave: war over who decides
1618-48
168-198
The era gets its name
1639
189
The hardest wave came late, not early. The first fifty years brought crashes, protests and one terrible manual. The century of upheaval came after the medium had become ordinary, after everyone had stopped noticing it. That's the pattern worth remembering. The economic edge also faded with time: the growth advantage of the early printing cities weakened after 1600, once print was everywhere.4
03 / Six hundred years
The waves don't stop at 1648.
Print kept moving outward in waves long after the first cradle closed. Each wave widened the reach of the printed word and raised the same question again: who controls the presses, and so who paints the pictures in people's heads. Seven waves, from Mainz to now.
WAVE I
The cradle
c.1450
Movable metal type at work in Mainz
An engineering answer to a copying problem, not a philosophy. Nobody involved set out to change what a person could know.
1455
The Gutenberg Bible, and the lawsuit
About 180 copies, around 135 on paper and the rest on vellum. 49 survive. The same year, the inventor loses his press to the man who paid for it.8
1472-74
Glut, crash and petition
Venetian printers overproduce and go under. A scribe asks the Doge of Venice to ban the press. Both events change nothing about where the medium is going.6
1486
The Malleus Maleficarum
A confident, wrong, deadly manual meets a machine that can copy it without limit. About 28 editions follow by 1600.9
1492
In Praise of Scribes
Abbot Johannes Trithemius argues that parchment will outlast paper and that copying by hand is a spiritual discipline. He has the argument printed.7
1500
282 towns, about 30,000 editions
Fifty years in, the infrastructure is everywhere in Latin Europe. Mallinckrodt will later choose this year, arbitrarily, as the end of the cradle.12
WAVE II
The rupture
1501
The book you can carry
Aldus Manutius's small italic octavo. The printed book stops imitating the manuscript and becomes its own thing.25
1517-25
Luther and the pamphlet flood
Short, cheap, in German, instantly reprintable. Some 4 million copies of works are printed in German lands between 1520 and 1525. The first person to master the native form of a medium can move a continent.10
1543
Copernicus
Eisenstein's point: astronomers didn't have dramatically better observations yet. What changed in Copernicus's lifetime was that the same tables, maps and texts could now be compared, corrected and built on by everyone at once.12
1559
The Index of Forbidden Books
Pope Paul IV's Index condemns the complete works of nearly 550 writers, Erasmus among them. The institution uses print to try to control print. The Index lasts until 1966.11
WAVE III
The periodical and the war
1605
The Relation, Strasbourg
Johann Carolus had sold handwritten newsletters to rich subscribers. He switched to print because copying by hand took too much time. His petition to the city council asking for protection against reprints is the newspaper's birth certificate, and one of the earliest copyright claims.13
1618-48
The Thirty Years' War
Not caused by print alone, but fought over the split print helped make permanent. 4.5 to 8 million dead.14
1639
The cradle gets its name
For the 200th anniversary of printing, Bernhard von Mallinckrodt, dean of Münster cathedral, publishes De ortu et progressu artis typographicae in Cologne. It contains the phrase prima typographicae incunabula: the first infancy of printing. He sets the end of that infancy, arbitrarily, at 1500. The convention still holds.1
1648
Westphalia
The settlement. Rulers decide religion inside their own borders, and the modern state system takes shape.
WAVE IV
The catalogue and the late adopters
1688
The word lands on the books
Cornelius van Beughem publishes Incunabula Typographiae, a catalogue of about 3,000 fifteenth-century books. The metaphor becomes a category.1
1727
The Müteferrika press, Istanbul
Nearly three centuries after Mainz, printing in Arabic script is permitted for non-religious works.15
1600-1800
Reading becomes ordinary
Literacy in north-west Europe climbs fastest in these two centuries, long after the cradle closed.16
WAVE V
Industrial print and the crowd
1814
Steam
The Times of London is printed on a steam-powered press. Print becomes a mass daily product, and the reading public becomes a mass audience.
1922
Lippmann: the pictures in our heads
In Public Opinion, Walter Lippmann argues that the world is too big for anyone to see directly. People act on the pictures in their heads, and those pictures arrive through the press. He coins the phrase "the manufacture of consent."18
1928
Bernays: the invisible government
Edward Bernays, a veteran of the US government's First World War propaganda committee, publishes Propaganda: "Those who manipulate this unseen mechanism of society constitute an invisible government which is the true ruling power of our country."18 Five centuries after Mainz, the question the scribes raised in 1474 comes back in its final print-era form. Not who can copy, but who can shape what the crowd believes.
WAVE VI
Broadcast, then the faster shelf
1966
The Index is abolished
The longest-running attempt to control print ends after four centuries.11
1979
Eisenstein names the unacknowledged revolution
The Printing Press as an Agent of Change argues that historians had missed how much print shaped the Renaissance, the Reformation and modern science.12
1990s-2022
The faster shelf
The web and then the feed. Unlimited pages, all of them still waiting to be fetched. The feed learns to flatter the reader instead of showing the world, and Bernays's invisible government gets an algorithm.
WAVE VII
The second cradle
Nov 2022
Knowledge answers back
ChatGPT goes public. For the first time, anyone can ask a machine a question and argue with the reply. The pictures in our heads start being painted in conversation, one person at a time, by systems a handful of companies build. Lippmann's question and Bernays's answer come back sharper than ever: who holds the press now?
04 / Incunabulum v2.0 in numbers
Four years in.
We can't measure our cradle as precisely as scholars have measured the first one. We're inside it, the numbers move every quarter, and a lot of them come from the companies doing the building. Here's what can be counted as of October 2026, and what each figure echoes from the print record.
1.2B/weekChatGPT weekly users, reported by OpenAI on 29 September 2026. Up from 400M in February 2025 and 900M in February 2026. One product among several.SOURCE 17 · ECHO: EDITIONS IN CIRCULATION
52%Of US employees use AI in their role at least a few times a year. 30% use it a few times a week or more, 15% daily. Gallup, May 2026, 22,573 workers.SOURCE 19 · ECHO: READERS
−13%Relative drop in employment for 22 to 25 year olds in the most AI-exposed jobs since generative AI spread. Experienced workers in the same jobs held steady or grew.SOURCE 20 · ECHO: THE SCRIBES
~4.2monthsHow often the length of task an AI agent can finish on its own has doubled since 2023, per METR. The best measured result passed 16 hours of skilled human work in March 2026.SOURCE 21 · ECHO: PAGES PER DAY
~13×/yearHow fast the price of a fixed level of AI performance has been falling since 2023, per Epoch AI. Print's price fell about 2.4% a year.SOURCE 22 · ECHO: THE PRICE OF A BOOK
~$700B2026 capital spending guided by Amazon, Alphabet, Microsoft and Meta, up from about $410B in 2025. Most of it goes to AI data centres.SOURCE 23 · ECHO: PRESSES IN TOWNS
945TWhProjected data-centre electricity use by 2030, about what Japan uses today. AI is the main driver.SOURCE 24 · ECHO: PAPER MILLS
2047The year 2,778 AI researchers gave even odds for machines beating humans at every task, in a 2023 survey. They put full automation of every occupation at 2116.SOURCE 26 · ECHO: NONE. THEY COULDN'T ASK
Reading the numbers against the record
Speed. Print needed fifty years to reach 282 towns. A single AI chatbot reached a seventh of humanity in under four. Whatever the right compression is, it isn't small.
Price. Print made books cheaper by a couple of percent a year, and that was enough to remake a continent. The cost of getting the same quality of answer from an AI is falling by more than an order of magnitude a year.22 If the press's slow price fall was a tide, this is a flood.
The scribes. Our first visible casualty sits right where the record says it should: not the masters, but the apprentices. Young workers in jobs where AI automates rather than helps are being hired less, while experienced workers in the same jobs are fine.20 The scribes' halving took twenty years. Ours has started.
The glut. In 1472 Venice's printers built ahead of demand and paid for it. In 2026 four companies are spending around $700 billion in a year,23 while the price of what they sell drops more than tenfold a year. Both can be true: the medium is real, and plenty of the people building it will lose money.
The old clothes. The chat window is our hand-painted initial. The native form hasn't arrived yet.
A note on these sources: usage and spending figures come partly from the companies involved and from trade press summarising them. The METR and Epoch measurements are independent but fast-moving. The expert survey was taken in late 2023, before the latest jumps in capability. Treat each number as a reading on a gauge, not a fixed fact.
05 / If history rhymes
Six months to thirty-two years.
History doesn't repeat exactly. It does tend to run in the same order. So here's the experiment: take the print sequence, zero it at Mainz around 1450 and ChatGPT's release in November 2022, and compress it. At a factor of five, one year of print history equals about ten weeks of ours.
Why five? Because the early markers line up there. At five, the scribes' halving (1465 to 1485) maps to late 2025 through 2029, which is right where the first entry-level employment data has started to bend. Venice's printing crash of 1472 maps to about April 2027. That's a falsifiable call, and it's six months away. If it holds, the dial is close. If it misses, we turn the dial and say so.
Turn the dial and the echo table recalculates. A higher number means our cradle runs faster than theirs. The horizon cards below are written for the default of 5.
#
In the first cradle
Print year
Lens date
+6 monthsAPR 2027
Echo: Venice, 1472-73
The glut
Venice's printers built ahead of demand, the market contracted, and many went under.6Today: about $700 billion of capital spending in a single year from four companies,23 against prices for fixed performance falling roughly thirteenfold a year.22
Watch for: write-downs, a lab or cloud provider failing or being absorbed, price wars, and headlines calling the whole thing a bubble. The record says the medium survives its first crash. The weaker shops don't.
Confidence
+8 monthsJUN 2027
Echo: Filippo de Strata, 1473-74
The petition
A scribe asks Venice's ruler, the Doge, to ban the press outright. Underneath the moral language, printers were undercutting his price.6Today: publishers, guilds, schools and magazines are already writing rules that bar anything made with AI.
Watch for: formal bans, licensing bills, "human-made" labels and lawsuits framed as morality. Most will fail the way de Strata did. Read them anyway. Each petition marks exactly where the guarded value sits, and some of those values, like first-hand human testimony, are worth guarding.
Confidence
+12 monthsOCT 2027
Echo: 1475, the scribes' tide
Our 1475
Manuscript output in central Europe was halfway through a twenty-year halving, yet scribes were still in demand and some went back and forth between pen and press.36Today: a 13% relative drop in employment for the youngest workers in the most exposed jobs, while their seniors hold.20
Watch for: the entry-level squeeze widening to more fields, and new hybrid roles for people who move between doing the work and directing the machine that does it. The bottom rung of the ladder is where this shows first.
Confidence
+2 yearsOCT 2028
Echo: c.1480, the press goes everywhere
Into the vernacular, into everything
By the 1480s presses were spreading town by town and printing in local languages. Caxton had brought the press to England in 1476. Today: half of US workers already use AI in their role.19 If METR's doubling holds, agents will be finishing tasks that take a person weeks.21
Watch for: AI stopping being a place you go and becoming a layer inside every tool you already use, in every language. Also watch the drafts. The Malleus was being written in these years. The record says the first mass-amplified, confident, wrong manual is close.
Confidence
+4 yearsOCT 2030
Echo: 1486-92, Malleus and Trithemius
The manual and the praise of scribes
A printed witch-hunting manual spreads persecution city by city.9 Six years later an abbot publishes his defence of handwriting, in print.7Today: data centres are projected to draw 945 TWh a year by 2030, Japan's whole appetite.24 The machine becomes infrastructure.
Watch for: a synthetic text or system that drives real harm to real people at scale, and keeps spreading after the experts condemn it. Watch, too, for the most eloquent critics making their case with the very tools they oppose, and for "handmade" becoming a luxury label. Manuscripts were still commissioned long after print won.
Confidence
+8 yearsOCT 2034
Echo: 1500-1510, Aldus's octavo
The native form
By 1500 the press was in 282 towns.2 In 1501 Aldus made the book small and portable and stopped pretending it was a manuscript.25Today: the chat window still imitates search and correspondence. Prices falling tenfold a year put today's best AI on cheap personal hardware within a few years.22
Watch for: the form that isn't a chat box. Personal, portable, local, carried rather than visited. And watch for the first people who write natively for it. Luther is seven print-years out from here.
Confidence
+16 yearsOCT 2042
Echo: 1543-59, Copernicus and the Index
The dividend and the counter-move
Shared, fixed, correctable knowledge lets a generation of scholars outrun the ancients, and Copernicus publishes.12 Sixteen print-years later the Church publishes its list of forbidden authors.11Today: in 2023, AI researchers put even odds on machines beating humans at every task by 2047.26
Watch for: discoveries in science and medicine that only happen because knowledge can now question itself back. And, close behind, the formal index: registries of forbidden models, licensed AI providers and approved outputs. Both arrive in the same era. The record says the index loses, slowly.
Confidence
+32 yearsOCT 2058
Echo: 1605-1648, the long settlement
The hardest wave, and the name
At a factor of five, 2058 is our 1629. The newspaper is a generation old. Europe is in the middle of thirty years of war over who gets to decide what is true.14 Ten print-years later, in 1639, a cathedral dean looks back and calls the beginning a cradle.1 On the lens, that lands around 2060.
Watch for: this is a pattern, not a prophecy. The record's warning is that the deepest conflict came after the medium was ordinary and everyone had stopped looking at it. The record's promise is that a settlement followed, and the modern world was built on it. The people who did best were the ones who saw early what they were living through.
Confidence Speculative
06 / Don't wait for 1639
Not the end. The next shape of us.
The scribes weren't wrong that something was ending. They were wrong that it was everything. Copying by hand did end as the way knowledge moved. Reading didn't. Writing didn't. Scholarship, prayer, argument and art didn't. They grew, in forms the scribes couldn't have pictured from the scriptorium.
We've always been made partly of our inventions. Writing was a prosthesis for memory, and Plato worried it would rot our minds. Spectacles extended the working life of every scholar who wore them. Print turned memory into a library anyone could join. Each time, the tool didn't replace the person. It changed what a person was.
AI that talks back is the next one. It won't stay on a screen. Our own inventions are going to work their way into how we learn, how we heal and how we think, and the line between the person and the tool will keep moving the way it always has. That's not the end of being human. It's the next version of it.
Mallinckrodt only named the cradle once he was looking back at it from two centuries off. We don't have to wait that long.
So why worry about it? Worry doesn't tell you anything the record can't. The record says there'll be a glut, and petitions, and a terrible manual, and a native form, and a voice nobody expected, and a long fight over who decides. It also says that people who could see what they were living through did better than people who couldn't.
See it. Understand it. Be present in it. You are not approaching the cusp. You're standing on it.
You're standing in it.
Open threads / Help build version 0.2
Where the record runs thin
Calibrate the dial. Track the April 2027 glut call and the 2027 petition wave. If they land, five holds. If they don't, re-fit the compression and publish the change.
Literacy by country, 1475 to 1800. Pull the Buringh and van Zanden estimates per country so the "readers" line has real numbers.
Scribes, by the head. Find counts and wages for scribes and stationers before and after 1470, especially Paris, Florence and Venice.
1524-25. The German Peasants' War and the pamphlets behind it: the first time the native form reached people who couldn't read well, through those who could read aloud.
The press without a revolution. East Asia had movable type centuries earlier. Why did it change different things there? A key control case for the whole argument.
Our Malleus watch. A running log of synthetic texts and systems that cause documented harm, with dates, so the four-year horizon can be checked rather than asserted.
07 / Sources
Check the work.
Numbers in the text link here. Where a source is a summary of a scholarly work rather than the work itself, it says so. The Gutenberg-AI parallel and the compression lens are this project's own creative premise, not findings of any source below.
[3]Buringh, E. and van Zanden, J.L. (2009). Charting the "Rise of the West": Manuscripts and Printed Books in Europe. Journal of Economic History 69(2). Abstract. Summary of print vs. manuscript totals and the central European decline: Purple Motes
[4]Dittmar, J. (2011). Information Technology and Economic Change: The Impact of the Printing Press. Quarterly Journal of Economics 126(3). LSE Research Online. Business education and city growth: CEP DP1365. Weaker advantage after 1600: summary
[19]Gallup, Q2 2026 workplace AI use (May 6-20, 2026, n=22,573). Gallup
[20]Brynjolfsson, E., Chandar, B. and Chen, R. (2025). Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence. Stanford SIEPR
[22]Epoch AI on inference price trends: 9x to 900x a year by benchmark (2025); about 47% a quarter since 2023 (September 2026, as reported). Epoch AI · report of the Sept 2026 figure
[26]Grace, K. et al. (2024). Thousands of AI Authors on the Future of AI (2023 Expert Survey on Progress in AI, n=2,778). arXiv 2401.02843
Dates given without a source number (Copernicus 1543, Caxton at Westminster 1476, the steam-printed Times 1814, Westphalia 1648, ChatGPT's release in November 2022) are standard reference dates.
