Opinion · Culture & Cognitive Psychology
In 2022, 48.5% of American adults said they had read at least one book in the past year. Ten years earlier, the figure was 54.6%. Those same adults almost certainly read thousands of words every day, in texts, captions, comments and subtitles.
So is reading really dying, or just moving to smaller screens? I don’t think the word count settles it. What fades when the book goes isn’t the habit of seeing letters. It’s the ability to follow context and see the wider picture.
What is actually being discussed?
- James Marriott’s book The New Dark Ages: The Death of Reading and the Dawn of the Post-Literate Society came out in 2026. It is now drawing reviews and opinion columns in major outlets.
- Its core claim is that as reading declines, complex knowledge weakens and public debate gets simpler.
- Only 35% of US 12th graders were proficient in reading on the 2024 national assessment, down from 40% in 1992.
- The share of US adults who read a novel or short story fell from 45.2% in 2012 to 37.6% in 2022.
- Reviewers push back on the word “death.” One argues that “evidence of decline is not proof of imminent extinction.”
Sources: Los Angeles Review of Books, Washington Monthly, National Endowment for the Arts, USAFacts
Seeing a lot of text is not the same as reading
Putting a feed full of short posts on the same line as a book makes little sense to me. The words may add up to a similar number. The work your brain does is completely different.
A book makes you hold an idea across pages. A writer sets something up in chapter two and pays it off in chapter nine. Every author also has a different way of saying things. Along the way, you pick up grammar and phrasing you were never formally taught.
Research backs up the vocabulary part. Psychologist Keith Stanovich and his colleagues found that vocabulary growth was predicted by how much print people read, even after accounting for intelligence. He points out that books use far more rare words than TV or everyday talk. He compared a novel’s word list to its film version and called the film’s list “emaciated.”
Short posts sit closer to the film than the novel. They mostly use words you already know, in sentences built to be skimmed.
A feed also has no chapter two. Every post has to make sense on its own, with no setup and no payoff. That’s efficient. It also means you never practice linking a line to something said twenty minutes earlier.
Context is a skill you build one long page at a time
There’s a phrase I like for people who can read every word and still miss the point: context-blind. They catch the sentence but not what it is doing. Sarcasm reads as sincere. A careful “maybe” reads as a “yes.”
Most of us have seen a version of this in a group chat. Someone quotes one line from a long message and reacts to that line alone. The words were read. The context wasn’t.
I think spending less time with books plays a real part in that. A book trains you to ask what a line means given everything before it. A ten-second clip rarely asks that question.
Format matters too. A 2018 meta-analysis led by Pablo Delgado pooled 54 studies with more than 171,000 readers. People understood informational texts better on paper than on screens. The gap was larger under time pressure, and it grew over the years studied.
One way to read this: screens train us to skim, and skimming is the opposite of tracking context.
“Kids these days” is a real bias, but this isn’t only about kids
The fair objection is that every generation says this. A 2019 study by John Protzko and Jonathan Schooler, with about 3,500 US adults, found a clear pattern. Well-read people were the most likely to say children no longer enjoy reading.
The cause was memory. People remembered kids from their own childhood as being more like their current selves. When researchers told some participants they were less well-read, those people became less critical of kids. It’s close to the lens I wrote about in the naive realism piece, where my own seat always looks like the neutral one.
That bias is real, and the critics of Marriott’s book are right to raise it. But my concern isn’t really about kids. It’s about me.
Finishing a book is harder for me than it used to be. There’s simply too much around me pulling the other way. One small movement of my thumb opens a stream of Reels and Shorts, the stuff people now call brain rot. A book has to win against that every few minutes, and it often loses.
Gloria Mark at UC Irvine has measured this shift. In her 2004 study, people stayed on one screen for about two and a half minutes on average. By 2020, that had dropped to about 47 seconds. A chapter takes a lot longer than 47 seconds.
When the panic was wrong, and when it came true
Worries about new media are old. Looking at three of them shows when the worry holds up.
Socrates and writing, around 370 BCE
In Plato’s Phaedrus, Socrates retells a myth about the invention of writing. An Egyptian king warns it “will produce forgetfulness in the minds of those who learn to use it.”
He was partly right. Few people today memorize long poems the way trained speakers once did. But writing let thinking stretch across centuries. It replaced one kind of memory and opened a much bigger skill.
The “reading mania” of the 1700s
In the 18th century, critics across Europe described an “epidemic of reading.” Young people devouring novels like Goethe’s The Sorrows of Young Werther were said to suffer from reading addiction, rage and lust.
That panic aged badly. The novel became the very thing people now fear losing. The worry was about content and format, not about any skill disappearing.
GPS and the memory of the road, 2020
Louisa Dahmani and Véronique Bohbot tested 50 regular drivers. People with more lifetime GPS use had worse spatial memory when navigating on their own. Among 13 who returned about three years later, heavier GPS use matched a steeper decline.
The sample was small, and the authors say so. Still, this worry seems to have come true. When a tool does the practice for you, the skill fades.
The line between the three
Put side by side, the cases point to one test. A panic tends to be wrong when the new thing only changes the container. It tends to be right when the new thing takes over the practice itself.
Novels changed the container. GPS took over the practice. Short video, I’d argue, is closer to GPS. It doesn’t just deliver words in a new shape. It removes the long stretch of attention where context gets built.
This is also why the “kids these days” bias and a real decline can both be true. The bias makes us remember the past as better than it was. It doesn’t bring back a skill that has stopped getting practiced.
The fix starts with protecting the long stretch
None of this means books are sacred or that every clip is junk. Marriott’s critics are right that reading isn’t extinct, and plenty of people still love it.
Christoph Irmscher, reviewing the book for Washington Monthly, points to signs of life: bookstores, literary festivals, and BookTube creators recommending novels to huge audiences. That fits the test above. Those are new containers pointing people back to the practice, not replacing it.
The question is whether the skill gets practiced. Context doesn’t come from reading more words. It comes from staying with one argument long enough to see how its parts connect.
If you want to check which side of the line your own reading falls on, try these four questions after your next session:
- Could I explain the main argument to someone in two or three sentences?
- Did I stay with it for twenty minutes without checking another screen?
- Did I meet a word or a sentence shape I wouldn’t have used myself?
- Did something on a later page change how I understood an earlier one?
If the answers are mostly no, the problem isn’t how much you read. It’s that nothing asked you to hold it together.
Editorial note: This is an opinion piece based on public reporting as of October 9, 2026. Most reading research shows links between habits and skills rather than proven causes, so treat these findings as patterns, not guarantees. The featured image is AI-generated.