I’m Charlie Becker, a second-generation bookseller building an AI tool for used bookstores out of my family’s store in Houston. This newsletter is a one-man “weekly magazine” about my journey building the second bookstore and all the things that come up or catch my attention in the process.
We paused sales last week when we got a single online order for 70 books. It was so unprecedented we actually had to call the smaller, slower marketplace platform it was placed on and ask, “How do we ship this?” because they only give us space for one tracking number, and all those books wouldn’t even fit into our biggest box.
I’m coming to you this week with some hot technology news from the world of used book selling—which is something I don’t get to say very often.
About two weeks ago, my family noticed something weird happening with our online sales. We sell on our own website and a handful of online platforms (think Amazon, eBay etc.). On the slowest platform we use, one where we normally make the least sales, we had been selling at 20 times our normal rate for the week. Even stranger, about 95% of those orders were going to a single buyer who was having them shipped to a different location from where they were based.
Most of our orders over the last 25 years of selling online have been one book, one buyer. You may sometimes get someone to buy a set, or two or three books. More than that is kind of crazy. In the last week, every order has been bigger than two or three.
When we got the single order for 70 books, we paused all orders because we couldn’t fulfill it and we didn’t really know what was going on. About a week later, I noticed that on bookseller subreddits, Amazon seller forums, Twitter, and other places all over the internet, other people were talking about the same thing happening. (I’m not going to comment on whether it was the same buyer, who the buyer was, or where they’re sending it, because I don’t want to break any terms of service for the websites I’m on. I can just tell you that our experience roughly resembles theirs. You can google and find the names yourself.)
What people kept noticing is that these books are going out in larger-than-normal shipments — and they’re strange books—stuff like The Insider’s Guide to Metro Denver from 1995 or How to Use Corel WordPerfect 1991—just old books people are never normally buying. Dead inventory. They’re all being bought by one company and shipped to a different location than where the company is based.
There are two possibilities. The theory that’s taken off like wildfire on Twitter is that it’s an AI company buying all these books to scan them for purposes of creating more training data for the algorithm. My theory, which is a little more mundane, is that it’s just simple financial arbitrage.
When we did the research on which books were selling and combed through all our orders, it was a lot of the aforementioned obscure books that aren’t necessarily rare—meaning these are books to which there isn’t a lot of information or surplus inventory, but people aren’t exactly collecting them or clamoring to buy them.
We discovered all the books bought from us on the slower platform were not listed by us on Amazon, for a range of reasons, but the books did have listings on Amazon. Most notably, all of the books bought from us had one of two things in their Amazon listing: either it was out of stock and said, “this book is currently unavailable,” or it had a price that was 5, 10, or 20 times what we were listing it for on the slower platform.
My theory of financial arbitrage essentially says that this buyer is exploiting this gap. Basically, if I’m that company buying all these books, it’s just a financial decision. They’re just sending the money out and purchasing these books based on some kind of algorithm that sees discrepancies in listing or availability between platforms. For them, it’s just a number.
That brings us to the other place these books are getting sent. If you’re a used book seller on Amazon, you have two options for listing. You can do fulfillment by merchant (FBM)—when someone places an order, you grab the book, put it in an envelope, and send it to the person who bought it. Or you can do fulfillment by Amazon (FBA)—you label your books, put them in a box, send the box to Amazon, who stores them on a shelf, and when someone orders, Amazon fulfills the order. Remember how I said the company is buying and having the books sent to another location? From what I’ve seen online, the company ordering these books is sending them to FBA prep companies for the most part.
I have no proof that this is happening, but undoubtedly it’s a sensible business model if it works. Raise capital, develop a way to systematically find underpriced or understocked books, buy those books and have the order fulfillment handled by Amazon. While it might be profitable, it goes against what we believe in as book sellers.
Books are not simply a commodity. Given, you do have to commodify them a little bit to make it in the used book business, unless you’re doing extremely high-end antiquarian sales. Books can’t just be some other commodity people are hedging and spot trading. But it looks like things may be going that way. My fear is that something like this was developing is a big part of why I started the AI project I’m doing. I’m building an AI tool to preserve these books’ metadata while watching another company’s algorithm vacuum them into a black hole. The spirit of the project is to create more permanent, higher-quality metadata for each of these rare books we’re selling.
If we were to continue just selling thousands of books into this black hole where someone’s buying them and sending them direct to FBA, we wouldn’t know what’s happening with those books. And we don’t know what’s going to happen if they don’t sell on FBA in a year or two. If we really do have the last copies of some of these books—as it seems when comparing the slow platform and Amazon—and we’re sending them into a portal from which they may not come back, they may be lost forever.
You might say, “Aren’t all the books you sell lost forever?” Technically, every book we sell is lost forever to a customer. That’s good. That’s why we exist, to sell books.
But (as I said in my O’Shaughnessy Ventures application) we are experiencing a second burning of the Library of Alexandria. Everyone thinks everything is really well preserved and organized on the internet. If you want a classic book—just the text, anyway—there are great resources like Project Gutenberg, and a lot of librarians really working hard across the world to preserve books. But our digital infrastructure is not as sophisticated as we think it is for preserving the metadata of these books, and we’re losing a lot of those details. A lot of ebook platforms from 10 or 20 years ago were shuttered, and a lot of the exclusive books on those platforms disappeared and were never heard from again. And obscure books that don’t sell via FBA after a certain amount of time often get liquidated.
It’s also happening to mid-list and antique books at a much higher rate than people realize, because they assume that once something is on the internet, it’s preserved. They don’t think about the fact that somebody has to pay to host that data and preserve that book.
While we’re not scanning full books for preservation, the tool I’m developing will at least give us a map of what’s out there, what we might be missing, and what we stand to lose.
Do you think it matters what happens to the books after someone buys them?
Note from Charlie
We have officially started reading The Count of Monte Cristo for the Six in ‘26 Book Club! Join us for the discussion at the end of June. Find all the details here.
This week’s newsletter was based on a video I made that you can watch on Substack below, or check out on Instagram, Twitter, TikTok, or YouTube.
Stuff Worth Sharing
The End of Elsewhere was an incredibly thought-provoking essay about the decrease of travel (or even the decrease of the desire of travel) among young Americans, and why it’s happening and what it means. I disagree with a lot in this essay but it is profoundly interesting and has a lot of things worth thinking about.
I spent a lot of time looking up the authors who had written the obscure books we were selling in the above scenario because I wanted to interview them. One of them is Peter Rinearson, who helped co-write Bill Gates’ book The Road Ahead, as well as many many other things. He has been a prolific writer and entrepreneur, winning a Pulitzer and being an early adopter to Microsoft Word and creating such good tools that eventually his company’s products were integrated into Microsoft Word itself. It was actually pretty hard to find him writing or talking about himself, but I did find this brief interview, and I liked two sections so much that I want to paste them here in their entirety. I find his words very inspiring as a great example of what it can look like to reinvent yourself but be thematically consistent:
“At the Seattle Times, I was the first reporter to use a computer terminal exclusively to write, and this lead to an interest in word processing and my eventual departure from daily journalism to the worlds of software and the Web. . . My career has been one of deliberate change. Every few years I do something new, where I can be a beginner again. I started as a newspaper reporter. I wrote the leading books of their time on how to use Microsoft Word. I did foundational work for Word’s built-in document designs. I started the first software company devoted to enhancing Microsoft Word, and among other things designed and developed the toolbar that Microsoft eventually licensed for use in Microsoft Word for the Mac. For a decade, we manufactured the “Foreign Proofing Tools” that people use to work with Microsoft Office in various languages. I wrote a book (“The Road Ahead”) with Bill Gates, and collaborated with him for four years on a newspaper column carried by the New York Times Syndicate. I ran a design studio that did projects as varied as special effects for a television series (Bill Nye the Science Guy), logo design for Real Networks, and an 85-foot mural for Disney’s Epcot Center. In the second half of the 1990’s I built an Internet company that tried to interest the newspaper industry in Web technology that allowed a blend of professional and community content. When I got no traction with newspapers I sold it to a cable television network (Oxygen) that ultimately didn’t understand community content either. I was a vice president at Microsoft, where I ran a product incubator, developed products intended to drive enterprise adoption of Microsoft Office, and oversaw the company’s internal intranet, electronic and physical libraries, and archives. Now I’ve started a new business, which will launch a service at Intersect.com later this year. Unless I stay with Intersect indefinitely, in a few years I’ll probably try something entirely different, my most radical departure from my newspaper roots. So if I have a value at this meeting, it may be as somebody who has seen change as opportunity more than threat. I’ve had successes and failures, and have appreciated both. Change fosters anxiety in me too, but I embrace it when it isn’t for its own sake.”
Finally, I’ve been reading a lot of Leszek Kolakowski, and I wanted to share this brief essay he wrote called “How to Be a Conservative-Liberal-Socialist.” I found Kolakowski because I was looking for philosophers who had tackled modern problems in a generative and rigorous way in the second half of the 20th Century. Kolakowski stood out for his rigor and approachability. Here is a link that downloads a PDF to your computer (this essay is pretty hard to find).
Urgh. Sounds like a version of redlining against small, independent booksellers and their legitimate customers. Yes, I'm biased toward those who value books for their essence, sell them for a fair price, and don't view them as objects for the most exorbitant profit.
Is it Anthropic?
https://arstechnica.com/ai/2025/06/anthropic-destroyed-millions-of-print-books-to-build-its-ai-models/