Have you ever held a battered paperback and felt, for a second, that the object itself mattered as much as the sentences inside it? I have. The cracked spine, the coffee ring, the name written in pencil on the flyleaf. Those marks are not metadata. They are evidence that a book lived with someone. That is why the latest wave of bulk buying in the used-book trade feels so strange. Warehouses are moving faster than usual. Orders arrive in odd clusters. Then the paper itself is said to disappear after it has been scanned for machine training. People in the trade are calling it AI book shredding, and the phrase is ugly on purpose.
What Booksellers Are Seeing On The Warehouse Floor
A family shop that has bought and sold used and rare titles for more than three decades is a useful place to start, because the people who work there notice patterns before headlines do. One owner grew up watching his father rescue a doomed company library from a warehouse that was about to dump everything. That story used to sound like a small victory for print. Now the same kind of volume can leave a shop and never return to another reader.
The internet already changed sourcing. Digital catalogs changed pricing. Giant online marketplaces changed who could reach a buyer in another country before lunch. I still think those shifts were messy but understandable. This new demand is different. It is not a collector hunting a first edition. It is not a student filling a reading list. It is industrial appetite dressed up as research.
A Sudden Spike That Did Not Look Like Ordinary Demand
In one warehouse holding hundreds of thousands of titles, weekly sales jumped to two or three times the usual pace. Anyone who picks orders for a living develops a feel for the mix. A run of cookbooks. A cluster of textbooks. A scatter of novels that just got a screen adaptation. This spike did not look like any of that. Digging through a stretch of recent invoices showed something colder. Almost every order in a sample of one hundred came from the same buyer.
That is the kind of detail that makes a bookseller sit down. Ninety-five out of one hundred is not a coincidence. It is a vacuum. And once you see a vacuum, you start asking what happens after the boxes leave the loading dock.
Literally no one knows; that is part of the problem. Somebody who cares about our literary heritage should be in that pipeline somewhere, but that is not what is happening.
– Used bookstore owner
I keep returning to that line because it is not romantic fluff. It is an operations complaint. If the buyer is hidden, the end use is hidden. If the end use is hidden, the trade cannot tell a common reprint from a last remaining association copy until the moment is gone.
Scanning First, Then The Paper Is Gone
Court records described a large language-model firm buying printed books in bulk, cutting off bindings, scanning page after page, and then destroying the physical copies. The stated goal sounded almost archival: build a central library, keep the text forever, gather as much printed matter as possible. The method was the opposite of an archive. An archive keeps the object. This pipeline keeps the file and discards the object.
People who work in rare books are quick to add a necessary correction. Many of the volumes in these hauls are not treasures. They can be appliance manuals, forgotten midlist novels, outdated test-prep guides, leftover trade paperbacks that already sit in a dozen other shops. That correction matters. Outrage that imagines only priceless incunabula going into a chipper is sloppy. The cultural problem is larger than rarity. It is the idea that print is raw ore.
A spokesman for one company insisted that sourcing books is common across the industry and that rare or antiquarian copies were not the target of destructive programs. Fine. I will take the claim at face value for a moment. Even if every shredded copy was a common reprint, the practice still trains the public to treat paper as disposable packaging around tokens.
Why Destruction Hits A Nerve That Digitization Alone Does Not
Digitizing a book can be a gift. Scholars know that. Travelers know that. Anyone who has searched a phrase across a corpus knows that. The trouble starts when digitization is paired with erasure of the source object as standard procedure. That pairing feels closer to a flood or a museum fire than to a library scanner humming in a quiet room. You do not have to be a collector to flinch. Plenty of readers remember a childhood copy, a college anthology, a paperback that got them through a grim winter. Watching those objects become feedstock is visceral, even when the title is ordinary.
Just the attachment to maybe something you read as a child or in college, I think it is hard for a lot of us to see that kind of destruction. It hits the same nerve.
– Rare-book trade advocate
In my experience, that nerve is not anti-technology. It is anti-waste dressed as progress. We already live with too many systems that extract value and leave a mess. Books were one of the last everyday objects that still felt stubbornly durable. Turn them into a one-way input and you change the moral temperature of the whole trade.
Fair Use In Court Is Not The Same As Comfort In The Shop
A federal judge treated destructive scanning of legally purchased copies as transformative fair use, meaning the new purpose was different enough from reading or resale. Legal readers will argue that point for years. Shop-floor readers hear something else. They hear that once you pay the sticker price, the object can be annihilated in service of a model. Ownership becomes a license to pulp.
Perhaps the most interesting aspect is how quickly law and feeling split. The ruling speaks to transformation. Booksellers speak to absence. Those are not the same conversation. One is about copyright doctrine. The other is about whether a culture still believes a printed page should outlast a training run.
I am not a lawyer, and I will not pretend the doctrine is simple. I do think public trust depends on more than a winning motion. If the only answer to a shredded novel is “we bought it first,” the trade will keep sounding the alarm, ruling or no ruling.
The Opacity Problem: Who Is Buying, And Why Won’t Anyone Say?
Stopping the practice is hard because the buying side likes silence. Bulk purchasers can hide behind intermediaries. Confidentiality can be written into deals. It is still unusual, in ordinary used-book commerce, to interrogate a customer about intended use. You sell a stack of novels. You do not demand a manifesto.
That old courtesy now looks naive. When one account can clear a warehouse aisle, courtesy becomes a blindfold. A database company even floated print-sourcing services aimed at model builders and advertised discretion around buyer identity. After public heat, the pitch was walked back. The firm said it was only testing demand and had never purchased, scanned, or destroyed a book. The episode still told on the market. If there is demand, someone will try to productize it.
- Bulk orders arriving from a single account rather than a mix of readers
- Requests for large, undifferentiated lots instead of carefully chosen titles
- Reluctance to discuss end use once the invoice is paid
- Middlemen who treat buyer identity as a feature
- Warehouses noticing speed without seeing the books come back into circulation
None of those signals proves destruction by itself. Together they make a pattern that dealers now watch the way farmers watch the sky.
Common Copies, Hidden Losses, And The Myth Of The Endless Reprint
Here is where I push back on both sides of the argument. Panic that “all rare books are being fed to machines” oversells the evidence. Complacency that “they were only common copies” undersells the loss. Out of print is not the same as rare. Common is not the same as infinite. A tired refrigerator manual can still be the last convenient copy in a region. A midlist novel can be the copy with a local inscription. A textbook can be the edition a historian needs with the original charts intact.
Print runs feel endless until they are not. Anyone who has hunted a specific paperback knows the sudden cliff. One year the title is everywhere. The next year it is a listing with a three-week delay and a price that makes you swear. Systematic destruction accelerates that cliff for whatever happens to be cheapest to acquire in bulk.
| What people assume | What dealers often see | Why it matters |
| Only treasures are at risk | Most lots are ordinary used stock | Volume, not glamour, drives the pipeline |
| A scan equals preservation | The file survives, the object may not | Binding, paper, marks, and context vanish |
| Reprints will always exist | Specific editions disappear unevenly | Research and resale both get thinner |
| Buyers are identifiable | Intermediaries and silence are common | Accountability arrives too late |
Look at that table long enough and the romance drains out. This is logistics. Logistics can still be a cultural event.
Project-Scale Ambition And The Fantasy Of Owning Every Book
One internal effort was described as a bid to collect all the books in the world and keep them forever. I understand the hunger. A complete library is an old dream. Kings wanted it. National libraries want it. Engineers want it with better uptime. The catch is the method. Forever, in this telling, means forever as data. Paper becomes a temporary scaffold.
Hiring veterans of earlier mass-scanning campaigns made the industrial logic even clearer. Teams reached out to major distributors and retailers about bulk print purchases for a so-called research library. Retail names appeared in filings as sources of thousands of copies. That is not a bookshop browser with a tote bag. That is procurement.
I’ve found that language does a lot of work in these projects. Research library sounds gentle. Central collection sounds civic. Training corpus sounds honest. The honest phrase is the one that includes the shredder.
What The Used-Book Economy Actually Does When It Works
A healthy used-book circuit is circular. A title leaves a home, sits on a shelf, finds another home, and sometimes comes back years later with a new note in the margin. Dealers are not priests, and they are not saints. They price aggressively. They discard unsellable stock. They send some paper to recyclers because storage costs money. I will not pretend every shop is a museum.
Still, the default hope is recirculation. A sale is supposed to create another reader, not a pile of confetti. When a single downstream customer converts recirculation into a dead end, the whole circuit tilts. Prices can jump in the short run. Owners notice the cash. Then the inventory mix warps. What looks like a boom can be a harvest.
One dealer even tied a sales spike to the shredding story itself. Attention created orders. Orders created empty shelves. Empty shelves created a thinner local culture of browsing. That loop should make anyone in the trade uneasy, even while the register sings.
Cultural Artifacts Are Not Only The Famous Ones
Ask a conservator what a book holds besides text and you will get a list that models do not care about. Sewing structure. Paper stock. Publisher’s cloth. A bookplate from a vanished club. A subway ticket used as a bookmark. Those details are how historians reconstruct reading lives. They are also how ordinary families recognize themselves on a shelf.
Destroy the object after capture and you keep the words in a brittle way. You lose the evidence that those words traveled. I care about that more than I expected to when I first heard the phrase book shredding. Maybe because so much of modern life already feels unlocated. A file can be copied. A life among readers cannot.
- Text can be extracted and stored as tokens for a model.
- Layout, paper, and binding can be ignored as noise.
- Provenance and owner marks can be treated as irrelevant.
- The physical copy can be destroyed to save space and speed scanning.
- The public is told the knowledge has been saved because the file remains.
Step five is the lullaby. It is not entirely false. It is incomplete in a way that should keep us awake.
Dealers Cannot Police Intent Without Breaking Their Own Trade
So what should a shop do? Refuse large orders? Demand affidavits? Ban certain zip codes? Those ideas sound tough until you try to run a business on thin margins. A dealer who interrogates every bulk buyer will lose the sale to a quieter rival. A dealer who never asks will wake up as an unwitting supplier to a shredder.
Trade associations can issue norms. They can warn members about suspicious patterns. They cannot easily see through non-disclosure habits that the rest of commercial life treats as normal. That gap is the real governance failure. The people closest to the objects have the least reliable information about the destination.
If I were writing a code of practice on the back of a packing slip, it would be blunt. Ask. Record. Prefer buyers who will resell or house the copies. Treat secrecy as a risk factor, not a perk. Will that stop a determined purchaser? No. It would at least make complicity less casual.
Readers Have More Leverage Than They Think
Most of us are not wholesalers. We still shape the pond. Buy the used copy you actually want to keep. Donate with eyes open. Ask local shops whether they have seen odd bulk activity. Support libraries that treat print as a circulating public good rather than a scanning inconvenience. None of that is a statute. It is friction, and friction is sometimes the only brake available.
There is also a consumer habit worth breaking. The habit of assuming that if a file exists, the object is optional. Optional objects get destroyed first. That is true of furniture, of tools, of photographs, and now, more aggressively, of books bought by the pallet.
We lose something culturally when certain books are deemed commodities or expendable.
– Independent bookseller
That sentence is easy to nod at and hard to live by. Living by it means paying for storage, cataloging, and the slow work of matching a title with a person. Machines are faster. Speed is not the same as care.
What “Forever” Should Mean If We Are Being Honest
Forever as a marketing word is cheap. Forever as a conservation word is expensive. Climate control is expensive. Acid-free boxes are expensive. Catalogers are expensive. If a company wants the prestige of a complete library, it can fund the unglamorous parts. Scan without gutting. Keep samples of editions. Partner with existing repositories instead of building a secret hopper.
I do not buy the claim that destruction is the only practical way to digitize at scale. It is the fastest way to feed a scanner without managing a growing stack. Convenience got dressed up as destiny. We have seen that costume before in other industries, and it rarely photographs well in hindsight.
A rough public test for any mass-scanning program: Can an independent observer learn what was acquired? Can a copy still be handled after capture? Can a rare or locally important item be pulled out in time? Can the buyer be named without a lawsuit? If the answer is no four times, call it extraction.
That test is informal. It is also clearer than a press line about research needs.
The Quiet Market Effects Nobody Should Ignore
Even if you care nothing about sentiment, the market texture changes. Sudden demand lifts prices on the very copies students and small libraries rely on. Then destruction removes supply. The next buyer pays more for less selection. Dealers who specialized in humble working copies find their ponds fished out. That is not a morality play. That is inventory math.
There is a second-order effect too. Trust. The used-book world runs on handshake habits more than people outside it realize. Once sellers suspect that a smiling bulk account is a terminus, they get cagey. Cagey markets are slower and more expensive for everyone, including the companies that wanted cheap text in the first place.
I’ve watched other corners of collecting go through similar shocks when a new industrial use appeared. Ephemera, vinyl, even certain tools. The pattern is familiar. First comes disbelief. Then a price spike. Then a scramble for rules that arrive after the best material is gone. Books are simply the most symbolic object to go through the blender, because we were raised to treat them as more than stock.
A Better Pipeline Is Not A Fantasy
If model builders need text, they can license it. They can work with publishers. They can scan library holdings under agreements that keep the books intact. They can use already digital editions. They can pay for transcription where paper is fragile. All of those paths are slower than a knife through a binding. Slower is the point. Culture is not supposed to be a rush order.
Non-destructive scanning exists. It is fiddly. It requires people who know how to handle a tight binding without cracking it. It costs more per page. That extra cost is the admission fee for claiming you respect the record. Skip the fee and you are just another extractor with better branding.
- License text instead of quietly emptying warehouses
- Scan without separating boards from text blocks as default practice
- Publish acquisition categories so the trade can watch for vulnerable fields
- Exclude association copies, annotated copies, and true firsts by policy that can be audited
- Fund storage for a reference set of physical editions after capture
Would those steps satisfy every critic? Of course not. They would move the debate from shock to standards, which is where grown industries eventually have to live.
What I Keep Thinking About After The Outrage Cycle
Outrage has a half-life. Warehouses do not. Long after the phrase AI book shredding stops trending, pallets will still be cheap and models will still be hungry. That is why the booksellers’ alarm matters more than a single filing. They are describing a standing incentive, not a one-off stunt.
I keep picturing that Houston warehouse story from decades ago, the father and the kid loading doomed company books so they would not hit a landfill. The emotional math of that rescue was simple. Paper deserved another chance with a reader. The new math is colder. Paper is an inconvenience on the way to a dataset. If we accept that math without a fight, we should at least stop calling the result a library.
Call it what it is. A harvest. Harvests can be managed. They can also strip a field. Right now the people who love the field are the ones without a map of the trucks. That imbalance is the story. It will not end because a judge used the word transformative. It will end when buyers have to show their faces, when scanners can work without a shredder, and when a common used book is allowed to remain, stubbornly, a book.
