View this post on the web at [link removed]

Nearly 12,000 book bans in one school year. When a number like that lands, the mind reaches for a supply chain: publishers must be pumping out thousands of objectionable titles a year to stock this much shelvable trouble. That’s the question worth asking — so I asked it, and ran the math. The answer is the part nobody on either side wants printed: roughly none. The supply was never the variable. The enforcement was.
Start with the ledger from PEN America’s new report, released Monday during Banned Books Week. In the 2025–26 school year, public schools logged 11,982 removals affecting 7,370 unique titles — the work of some 4,000 authors — up 74% from 6,870 the year before, and the highest total since PEN began tracking in 2021. More than 22,800 bans have accumulated across 45 states in five years. But the geography is the tell: Texas recorded over 8,100 bans and Florida over 3,200 — roughly 94% of the national total in two states. Texas took the top spot for the first time because of SB 13, which lets districts purge books by broad category rather than case by case. If publishers’ output were driving this, the bans would track publishing markets — New York, L.A., wherever books are made. Instead they track statutes.
Now the supply side. Roughly four million new titles appear in American ISBN databases each year, which sounds alarming until you see what that actually is: mostly self-published print-on-demand slush, AI-generated filler, and reprints that will never come within a mile of a school library. Traditional publishing’s frontlist is a fraction of that. And the segment that ban rhetoric is actually aimed at — children’s and young adult books — is measured in thousands. The Cooperative Children’s Book Center received 3,619 U.S. children’s and YA books in 2024; about 7% — call it 250 titles — had LGBTQ themes or characters. That is the entire annual national output of the most-challenged genre category on the most-scrutinized theme. Ban every single one the day it publishes and you haven’t filled a tenth of this year’s ban count. To hit 4,500 newly flagged titles from new supply alone, you’d have to condemn essentially the whole yearly catalogue of young-adult literature, plus change.
And the flagged list isn’t new books. This year’s haul includes Jane Eyre (1847). It includes The Shining (1977) — Stephen King, pulled 312 times, was the most-banned author in America for the second straight year, an author whose work has sat on school shelves longer than most school boards have existed. The Handmaid’s Tale (1985). Speak (1999). Crank (2004). Sold (2006) — a twenty-year-old National Book Award finalist about child trafficking, and the single most-banned title. Add Tokyo Ghoul manga and you have the full picture: this is backlist. This is the accumulated inventory of American letters, being swept off shelves not because it arrived last Tuesday but because someone finally decided to conduct an inventory review.
What actually grew between the quiet 2010s and now is the machinery of removal. State laws that purge by category. Directives that require pulling a book first and reviewing it later. Mass challenges driven by circulating watchlists. And this year’s newest tool, confirmed in public records the Texas Freedom to Read Project shared with PEN: school administrators consulting AI chatbots, which recommended removing books — not because they met any legal standard of obscenity, but, in documented cases, because another district had removed them. Sit with that. A machine in one county clearing a shelf because a machine in another county cleared a shelf. A keyword scanner cannot read Speak and understand it is a novel about the aftermath of rape that a survivor needs to find; it can only see the word. It cannot see that Sold is an argument against the thing it depicts. PEN’s Philomena Polefrone put the operating condition plainly: “Confusing and vague state book ban laws and chilling directives from state politicians have created chaos and fear.” The AI didn’t create the chaos. It industrialized it.
To be fair to the supply-side intuition: the catalogue did shift after 2020 — the CCBC found 51% of children’s books it received in 2024 had significant BIPOC content — and some genuinely explicit material did land in school libraries in that period, to parents’ entirely legitimate objection. Curation is not censorship; no student is owed any particular title. And PEN’s counting is generous — a book pulled for a week counts the same as one pulped, which is why the American Library Association’s stricter, permanent-removal-only ledger says 5,600. Halve PEN’s number. Quarter it. The arithmetic still doesn’t reach publishing output, because the books on the shelves haven’t changed much in a generation. What changed is what’s being done to the shelves.
**Does it matter?**
Reading scores are the worst since the test began — twelfth graders reading at levels not seen in the entire history of the exam, roughly a third of eighth graders below “basic,” which is the polite federal word for *can’t do it* — and the newest national mood is a shrug: does it really matter anyway? It matters, but not the way the press releases say. Not primarily because the economy needs workers, though it does. It matters because reading is the only technology we have for thinking a thought that isn’t being streamed to us. A kid who can’t read isn’t just locked out of Faulkner; he’s locked out of the fine print — the lease, the loan, the ballot measure, the label on the prescription bottle — which means somebody else will always be happy to explain the world to him, briefly, persuasively, and for a fee. A republic of non-readers becomes a republic of listeners, and listeners get governed by whoever has the loudest feed. The decline took thirty years to register in the data; the consequence doesn’t show up in the data at all. It shows up later, in the voting booth and the jury box, when a citizen who was never taught to read deeply has to decide whom to believe — and picks the confident voice over the true one. So yes. It matters. It’s the only score that ever mattered.
Reading scores are now the worst in the history of the test — twelfth graders reading at levels not seen since the exam began, roughly a third of eighth graders below basic, which is the polite federal word for can’t do it — and the newest national mood is a shrug: does it really matter anyway? It matters, but not the way the press releases say. It doesn’t matter because the economy needs workers, though it does. It matters because reading is the only technology we have for thinking a thought that isn’t being streamed to us. A kid who can’t read isn’t just locked out of Stephen King; he’s locked out of the fine print — the lease, the loan, the ballot measure, the label on the prescription bottle — which means someone else will always be happy to explain the world to him, briefly, persuasively, and for a fee. A republic of non-readers is a republic of listeners, and listeners get governed by whoever owns the loudest feed. The score decline took thirty years; the consequence doesn’t show up in the data at all. It shows up later, in the voting booth and the jury box, when a citizen who was never taught to read deeply has to decide whom to believe — and picks the confident voice instead of the true one. So yes. It matters. It may be the only score that ever did.
Here is the number that ends the argument: 11,982 bans divided by 7,370 books is 1.6 bans per book. The same titles are being removed, district after district, in bulk, by list, by category, by chatbot. That is not a community reacting to what its libraries contain. That is a process propagating itself. The objectionable-book pipeline the question assumes doesn’t exist — publishers never built anything close to it. You don’t need twelve thousand objectionable books a year to get twelve thousand bans. You just need vague laws, circulating lists, and a machine that clears a shelf without ever reading a page — and someone in the statehouse willing to call that a review.
The books weren’t the problem. The objections were manufactured at home.
This Substack is reader-supported. To receive new posts and support my work, consider becoming a free or paid subscriber.

Unsubscribe [link removed]?