Article — "The New Dark Ages Thesis: Whether Civilizational Decline in Reading Can Be Proven or Only Felt": Article — "The New Dark Ages Thesis: Whether Civilizational Decline in Reading Can Be Proven or Only Felt": A professor assigns a story he is sure will land — spooky, chilling, a little over the top — and the students are left cold. He'd hoped Washington Irving's The Legend of Sleepy Hollow would win them over, since Henry James had not. When he asked them to dig into James's 1891 story Brooksmith, he recalled, "My students could barely muster the energy for a thumbs-up or -down. It was as though some pneumatic pump had sucked out the last dregs of their spirits."
They had been "entirely defeated by James's prose," he wrote — not just the story but "the medium of it." And he felt the ground shift: "a break in historical continuity," a corroboration of something he'd been "on the verge of grasping for years."
You have heard this scene before. My Students Can't Read. The Elite College Students Who Can't Read Books. There's a Very Good Reason College Students Don't Read Anymore. It is the defining dispatch of our moment.
Except that professor is Sven Birkerts, the story is from The Gutenberg Elegies, and the book came out in 1994 — when the students being mourned were reading paper and the smartphone did not exist.
James Marriott himself tells the Birkerts story in The New Dark Ages, his 2026 lament for "the end of reading and the dawn of a post-literate society." He reportedly knows the report is old. He files it anyway, because it is exactly his report.
Last week, in "Post-Literate but Still Reading," this column left a question open: is comprehension decline actually measurable, or is it assumed? This week I want to press on the seam that runs straight through Marriott's book — the seam between what the data can prove and what the dread only lets you feel.
Because there are two claims here, welded together, and they are not the same kind of claim.
The first is behavioral, and it is real. In the first two decades of this century, reading for pleasure in the US fell by more than 40 percent. A study of 236,000 responses to the American Time Use Survey found the share of Americans who read for pleasure on any given day dropped from 28 percent in 2004 to 16 percent in 2023.
Fewer than half of American adults reported reading a book of any kind in 2022. Only 38 percent read a novel or short story. And the line that lands hardest: last year 57 percent of Americans placed a bet — gambling is now a more common leisure activity than reading a book.
More than a third of UK adults say they have given up leisure reading. The National Literacy Trust reports "shocking and dispiriting" falls in children's reading. The OECD found literacy "declining or stagnating" across most developed countries.
These are measurements. They count acts — books opened, minutes logged, bets placed. You can argue about instruments and definitions, but they describe observable behavior, and the scale is genuinely new. Something happened in the mid-2010s, and Marriott names it plainly: the smartphone, which, unlike television, follows you into every room.
That is the honest, load-bearing part of the book. Hold onto it.
The second claim is different in kind. It is that these numbers mean "the golden chain of knowledge linking reader to reader through the centuries is breaking for the first time since the collapse of the Western Roman Empire." That print is "almost certainly ... a necessary precondition of democracy," and as it dies "we risk returning to the chaos, tribalism, and rage of a pre-literate society."
Marina Hyde calls the book "a flashing emergency light for our culture." That is the register. Not measurement — alarm.
And here is the trouble: that alarm has been sounding, in nearly identical words, at every media transition on record. It is not a symptom of our crisis. It is a genre.
In March, The Imaginative Conservative reportedly announced that the Encyclopedia Britannica going out of print was, in the essay's own escalating words, the most significant cultural event of the last fifty years — no, make that a hundred. Its verdict: "The New Dark Ages are upon us." "Will the last person who even knows what Western civilization is please turn out the lights?"
Same title as Marriott's book. Same eschatology. Different casualty — an encyclopedia, not a species of attention.
And that essay does the honest thing Marriott's blurbs don't: it keeps going back. To Gutenberg, where "for every work of Shakespeare put between covers, there are twenty cheap romance novels." Further back still, to Plato's Phaedrus, where the god-king Thamus warns Theuth that writing itself "will only create forgetfulness in the learner's soul," producing "hearers of many things" who "will learn nothing."
Read that again. The foundational text of the reading-decline panic is an attack on reading. The technology being mourned today was the technology being mourned then. The chain Marriott says is breaking for the first time was declared broken at its first forging.
Even Marriott's own evidence contains its own rebuttal. The professor in the Chronicle who assigned a 20-page article and had not one student finish it also writes: "Every generation of professors has complained that their students cannot read. The lament is usually overblown."
Then: "but data have caught up to anecdote." That is the exact move I want to isolate — the pivot from felt continuity to claimed proof.
Because the data have caught up to one anecdote and not the other. We can prove that fewer people read for pleasure. We cannot prove, in the same way, that a load-bearing beam of civilization is being pulled out from under us.
Literacy statistics measure behavior. Deep reading as a civilizational precondition is an inference — a plausible one, drawn from Postman and Ong, that print culture built the classifying, inference-making mind democracy runs on. But an inference dressed as a measurement is still an inference.
Notice where the certainty peaks. Not at the numbers, which are hedged and surveyed and footnoted. It peaks at the conclusion — the dark ages, the emergency light, the lights being turned out. The panic runs hottest precisely where the evidence runs thinnest.
This is why the Birkerts reveal matters so much. His 1994 students, "entirely defeated by James's prose," had no smartphones. Whatever defeated them, it was not the device Marriott blames. And yet the sentence is interchangeable with 2026. Either the cause keeps changing while the lament stays fixed, or the lament was never really about the cause.
Either the cause keeps changing while the lament stays fixed, or the lament was never really about the cause.
None of this makes the numbers fake. Fewer people read; more people gamble; children's reading is at record lows. Take it seriously. But taking behavior seriously is different from ratifying an apocalypse, and the discipline is to keep those two acts apart.
There is a model for that discipline, and it is quiet. Anne Enright's essays, reportedly collected this autumn under the title Attention, come from someone who has spent thirty years, as the copy puts it, "paying attention" — reading Munro, Morrison, Joyce, closely, without emergency. That is what deep reading actually looks like from the inside. Not a flashing light. A steady gaze held on one thing long enough to see it.
The honest position holds both truths at once. The behavioral decline is real, measurable, and new in scale. The civilizational verdict is old, recurrent, and unprovable — a dread that has been felt, and filed, at every threshold since Thebes.
So prove what you can prove. Count the readers, count the bets, fund the libraries, pilot the curriculum. That is real work, and last week's open thread on it stands.
But when the sentence arrives that says the golden chain is breaking for the first time in fifteen hundred years — the one that feels the most true — notice that it is the one you cannot check. That is not proof. That is the oldest feeling we have about the newest thing in the room.