Why people don't read anymore.

Every generation gets told it ruined reading. In the 1700s the panic was about novels themselves: too immersive, too emotional, corrupting the youth. Goethe's The Sorrows of Young Werther got blamed for a suicide wave.^[1]

Then it was comics. Then radio. Then TV. So when The Atlantic put"The Age of Reading Is Over" on its August 2026 cover, the historically literate response was an eye roll.^[2]

The uncomfortable part is that this time the numbers actually moved.

Yes, the decline is real

The best dataset on this has nothing to do with vibes or professor anecdotes. The American Time Use Survey analysis covers over 236,000 Americans across 20 years. It asks what they did yesterday.

In 2004, 28% had read for pleasure. By 2023 it was 16%.^[3]That's a 43% drop. It ground down at about 3% a year through recessions, booms, and a pandemic that supposedly made us all bakers and readers.^[4]

And it's not evenly spread. Americans 75 and older read 46 minutes a day. Teenagers read nine.^[5]

That gap matters because it tells you where this is headed. Reading isn't being abandoned all at once. It's aging out, one actuarial table at a time.

The story you've heard is shakier than the numbers

The piece that set the tone for this whole conversation was The Atlantic's 2024 story about elite college students who can't finish books. It went massively viral. The Atlantic story critique notes that it rests on interviews with 33 professors, zero students, and an explicit admission that no comprehensive data exist.^[6]One of the teachers quoted, Susan Barber, later wrote that the article left out her actual point. She cut her AP Lit list from 14 books to six on purpose, because depth beats coverage.^[7]

The newest official data complicates the collapse narrative too. NAEP scores released in June 2026 show nine-year-olds' reading scores up four points since 2022, with the biggest gains among the lowest performers.^[8] The youngest kids are recovering.

The crisis story is loudest exactly where the evidence is thinnest. That should make you cautious about every explanation that follows.

Cautious. Not dismissive. Because the time-use data doesn't care about anyone's narrative, and it keeps falling.

So what's actually causing it?

Ranked by evidence, not by how good it sounds on a podcast:

The time went somewhere else, and we know where. Reading for pleasure is discretionary leisure, and discretionary leisure now defaults to a phone. For kids, the same fight starts even earlier.

Average social media use runs over two hours a day, and TikTok users alone average around 95 minutes. Nobody scheduled a meeting to decide this. Reading lost a bidding war for the same 45 free minutes, against an opponent engineered to win it.

Attention itself changed. Not the fake goldfish statistic, which was fabricated and traced to an uncredentialed stats site. The real research is Gloria Mark's at UC Irvine.

She has logged actual screen behavior for two decades. Average unbroken attention on one screen fell from two and a half minutes in 2004 to 47 seconds now.^[9] A 2024 EEG study found that after just 30 minutes of short-form video, sustained reading attention dropped 31% and comprehension fell 22%.^[10]A novel asks for the exact thing we practice least.

Time poverty. Employed Americans read under 10 minutes on workdays, and the steepest declines are among lower-income groups.^[3]This is the least glamorous explanation and the least contested one.

Some people aren't choosing TikTok over Tolstoy. They're choosing sleep.

Schools, maybe. The "Sold a Story" argument says a generation was taught to guess at words instead of decode them. It has driven phonics laws in most states.^[11]

High school English teachers now assign about three whole books a year, down from four in 2009.^[12] Both claims are real and both are contested. Neither explains why 40-year-olds stopped reading. Grain of salt.

One cause you can mostly cross off is the idea that books are neurologically outgunned, that pages can't compete with dopamine. Anna Lembke, the Stanford psychiatrist who literally wroteDopamine Nation, has described her own two-year compulsive romance novel habit.^[13]

Books can absolutely hijack your reward system. Ask anyone 400 pages deep in a dragon-rider series at 2am. Which brings us to the strangest part of this whole story.

The book business is doing fine, actually

If nobody reads, explain romantasy. The genre did $610 million in 2024, up 34% in a year. Rebecca Yarros sold a million copies ofOnyx Storm in its first week, the fastest-selling adult novel ever tracked.^[14]

Total U.S. publishing revenue hit $32.5 billion in 2024, and print sales rose again in 2025.^[15] Audiobooks grew to $2.43 billion.^[16]

Both charts are true at once because reading polarized. Fewer Americans read at all, but the ones who stayed went from 83 minutes a day to 97.^[4]The casual middle vanished. What's left is a smaller, hungrier core that buys hardcovers, special editions, and the audiobook of a novel they already own.

Publishing learned to run a profitable business on a shrinking congregation. That works until the congregation stops replacing itself, and the age chart above shows the replacement rate.

Who counts as a reader?

Every chart in this essay sits on a definition, and the definition has been quietly moving. The American Time Use Survey only started counting audiobook listening as reading for personal interest in 2021.^[3]For 17 of the 20 years on that first chart, a person who spent an hour a day with novels in their ears got logged as not reading.

The study's authors say the coding change doesn't rescue the trend, since most audiobook listeners also read print. Fair. But it confirms the category was always narrower than the activity. The official measure of reading spent two decades deciding one kind of reader didn't count.

That exclusion has history, and so does the workaround. Audiobooks were an accessibility technology before they were a commuter convenience. In 1931, Congress passed the Pratt-Smoot Act to provide books for blind adults, pushed along by veterans blinded in World War I. Two years later the program added recorded "talking books" on vinyl discs.^[20] Audio has been how shut-out readers get in the door for nearly a century.

The purist objection, that listening is lesser, keeps failing its tests. A randomized study assigned 91 adults to an audiobook, an e-text, or both at once, then tested comprehension right away and again two weeks later. No significant difference, in any direction, at either point.^[21] The one caveat: the print condition was an e-reader, not paper.

And the stamina question matters more than book people like to admit. Decoding print for 300 pages is a trained physical skill, and it is not evenly distributed. The International Dyslexia Association estimates 15 to 20% of the population shows some symptoms of dyslexia.^[22]For a lot of brains the ear is simply the wider door. Mine included: audio is how I actually get through most books.

What nobody has measured is how many audiobook listeners are neurodivergent, or came to audio because print never fit. The industry's own 2026 survey found 58% of American adults have tried an audiobook, and the top reasons offered are multitasking and commutes.^[23] Nobody asked the question behind the question. Add it to the pile of studies someone should run.

What about AI?

The honest answer for 2026: nobody has measured it yet. Student AI use jumped from 48% to 62% in a single year, and teachers report kids summarizing novels with chatbots instead of reading them.^[17]

But there is no study showing whether AI summaries substitute for reading or sit alongside it. Anyone telling you otherwise is guessing. I work in AI. My bet is that summarization eats assigned reading first, the reading people already didn't want to do, and leaves pleasure reading mostly alone.

That's a bet, not a finding. Check back when someone runs the study.

One AI effect on reading is already measurable, though: access. In January 2023, Apple started selling AI-narrated audiobooks, aimed at backlist titles that could never pay for a studio and a human narrator.^[24] Audible's Virtual Voice catalog went from roughly 1,600 titles in 2023 to tens of thousands by 2025; no two counts agree, which mostly tells you how fast it's moving.^[25]Spotify and ElevenLabs now let self-published authors generate narration in 29 languages.^[26]

So the first confirmed thing AI did to reading was widen the door. A midlist novel from 2009 was never getting a studio recording. Now it has one, and the reader who needs it in audio, whether dyslexic, ADHD, low-vision, or just wired for the ear, finally gets the same shelf as everyone else. Whether that grows the 16% is the study I'm waiting on.

What actually brings people back

Twenty years of "reading is good for you" campaigns moved nothing. You know what moved the numbers? A TikTok trend. 48% of BookTok users say they read more because of it, and titles the community picks up see sales jump 113% on average.^[18]

E-reader programs took the share of boys who called reading cool from 34% to 67% in one study. Audiobook-plus-text reading measurably improves comprehension, whatever the purists say.^[19]

The pattern across all of it: nobody was guilted back to books. They were invited by people they liked, toward books that felt like pleasure instead of homework, in formats that fit the life they actually have.

Reading survived the novel panic, the comics panic, and television. It'll probably survive the phone too. But "probably" is doing real work in that sentence, and the time-use line only bends one direction so far.

Anyway. If you made it this far, congratulations: you're in the 16%.