Hey guys, Non here, coffee in hand. Want to hear some interesting stories?
Let's start with the shelves. Did you know that in the first half of this year, people bought 14 million more secondhand books than new ones for the first time since we started counting? That is the simple fact. It matters because it is the first time we have had to ask if we have stopped believing in new things, or if we just got tired of paying for them.
First, the war in Ukraine, because that has not stopped, then we will talk about why books are the new gold, how to turn off the lights, and finally a moment for the mountains. The front line in the Donbas has shifted today, with both sides trading artillery fire across a front that has barely moved in months. The urgency is real.
Over the last 48 hours, the kinetic exchange between Russia and Ukraine has shifted into a high-volume, long-range exchange that is beginning to hit deeper into the domestic infrastructure of both countries. According to reports from NPR and local officials, Ukraine launched a strike on Saturday night involving roughly 822 drones, with 600 of those directed at Moscow. In the following 24 hours, Russian drone strikes killed two people in the Sumy region and damaged a civilian vessel in the port of Odesa. On the Russian side, local officials in the Belgorod region reported eight deaths, while a woman was killed in a drone strike in the southern Astrakhan region.
To understand why this is happening now, we have to look at the logistics of a war of attrition. When one side can no longer achieve a breakthrough on the front line, the target moves to the "rear." For Ukraine, this means targeting the Russian economy and the domestic morale of its population. Specifically, they have been systematically hitting Wildberries, which is essentially the Russian version of Amazon. Since July, Ukraine has targeted these logistics centers because, as Maj Gen Vadym Skibitski of Ukraine’s intelligence service explained, these networks are used to move military components and equipment.
For Russia, the goal is to choke off the maritime trade in Odesa and strike high-value industrial targets, like the FirePoint facility in Kyiv that produces parts for Flamingo missiles. The popular narrative in the headlines right now is that this represents a "new phase" of escalation—a move toward total, unrestricted warfare where both sides are simply trying to destroy the other’s will to live. It is framed as a desperate race to see who can break the other's spirit first. Fair enough, the scale of the attacks is certainly unprecedented for the domestic populations.
But there is a simpler, more mechanical reality at play here that the "escalation" narrative misses. This isn't just about breaking spirits; it’s about breaking supply chains. By disabling seven out of ten of the largest Wildberries centers, Ukraine isn't just trying to make people angry; they are trying to make it physically difficult for the Russian state to move goods. Conversely, by hitting Odesa, Russia is trying to make it too expensive for foreign companies to stay.
The gap here is between the tragedy of the individual deaths—like the 83-year-old man in the Moscow region or the child injured in Belgorod—and the cold math of logistics. We are seeing a war that has moved from the trenches to the delivery trucks and the warehouse doors. It is no longer just about who holds the most land, but who can keep their internal economy functioning while the other side tries to unplug the wires.
We were talking about bullets and power grids and the fragile plumbing of a country at war, and now we are in a Northumberland bookshop looking at a single, flat piece of paper. That is the distance we just crossed. Stuart Manley is not dealing with mortars or sanctions, he is dealing with a Canadian company that wants his novels. In a typical week he moves two or three thousand books.
In one recent single bulk order, the number was what he expects to sell in seven days. He has been running Barter Books for thirty years; he has never seen the like. And Barter Books is not an outlier. Around the world, booksellers are seeing these same mega orders—scores of novels shipped to far-off warehouses, often titles that have been sitting on the shelf for decades.
The suspicion is that these are not being bought by avid readers. The appetite here is for information, and the voracious thing feeding on it is an AI. The trail goes back to a US court ruling in 2025. A judge, William Alsup, decided that training AI software on books purchased in this way did not violate copyright law.
The case was brought by three authors against Anthropic, the company behind the chatbot Claude. Judge Alsup wrote that the use of the books was "exceedingly transformative," a legal phrase that effectively means the AI creates something new enough that the old rules do not apply. The more interesting part of the court documents, which were unsealed last month, is not the ruling itself but what they revealed about how Anthropic gets the data. They called the project "Project Panama." The goal was, in their own words, to "destructively scan all the books in the world." That is not a metaphor.
Destructive scanning means shipping books to a facility where they are processed at industrial scale. In some cases, the spine is removed so the pages can be fed into a high-speed scanner, digitised, and then pulped. The book is read, then physically destroyed. Stuart Manley says he is not sure his books are going to that specific project, but he acknowledges the name is "new to me, but the reality of the project is not." He knows the books are being used to train a Large Language Model.
To put it simply, a Large Language Model is a prediction machine. It does not know facts; it knows patterns of language. If you ask it to write a paragraph about the French Revolution, it predicts which words are likely to follow one another based on what it has seen before. If that training data is only news articles and modern novels, the model will be excellent at writing news and modern novels, but it will be bad at history.
It will lack nuance, it will lack the weird specific details that only come from older texts. Booksellers have noticed that the books being bought are random with "no rhyme or reason." They are not buying the bestsellers; they are buying the obscure, the academic, the specialized. Professor Emily Hudson at Oxford explains that this makes sense from a technical standpoint. AI needs diversity.
The more different kinds of text it sees—the more it sees how a 19th-century sermon sounds compared to a 21st-century legal brief—the better it becomes at parsing human language in general. There is, of course, an immediate objection. This is theft. If I buy a book and read it in my own home, the copyright holder has been paid.
If a company buys that same book, scans it in seconds, and then destroys it so no one else can read it, the author has not been paid. The author has been bypassed. Derek Walker, who runs Edinburgh bookshop McNaughtan’s, puts it bluntly: the starting point in the UK is that copying a book requires the permission of the copyright owner. "It would be sad if they are ultimately destroyed," Walker says.
And it is sad. There are books he has sold that are the only known surviving example of an 18th-century edition. If the company buys it and burns it, that version of history disappears. It is not just that the author loses money; it is that a piece of human knowledge is erased from circulation.
Someone will say that this is the cost of progress. We are building tools that will help us write code, draft contracts, and diagnose diseases. To do that, the tools need data. The US legal system has decided that using secondhand books is fine.
The law has changed to accommodate the new technology. The books are not rare or antiquarian in the way that first editions are; they are just old books, and old books are not unique. The objection is that by destroying them, we are losing the very thing the AI is pretending to learn from—human creativity, context, and the messy reality of the physical world. But here is where the picture gets sharper.
We tend to think of AI as a cloud in the sky, something happening on servers far away. In reality, the physical world is crashing into the cloud through books. We are pulping the physical books to build a digital ghost. The irony is that the booksellers are celebrating the sales.
"In some ways, it's very nice to sell some of these titles which haven't sold for many years," Manley says. He gets the cash, he clears the shelf, and he has no idea what is happening to the book once it leaves his shop. He is happy, the company is happy, and the AI is happy. The author is the one who has been cut out of the deal entirely.
The more important question is not whether the law allows it, but whether we are comfortable with a system where knowledge is treated as a commodity that can be consumed and discarded. If the next generation of AI is trained on a library that is permanently being deleted, it will be built on a foundation of things that no one can read anymore. The books are not just fuel; they are the record. And right now, the record is being burned to run a machine that is supposed to help us understand it.
If we are burning records to run a machine that helps us understand them, we eventually have to decide which parts of our lives we want that machine to understand first. We are moving from the physical archives of books to the digital architecture of our daily tools, where the goal is no longer to preserve information, but to automate our interaction with it. According to reports on Hacker News, users are increasingly seeking out manual ways to strip AI features out of their basic software. This is becoming a significant friction point because many developers are moving away from "fallback states." To put it simply, if you turn off the AI, the app might simply stop working.
A recent example is Apple CarPlay, where you currently cannot use the maps or play music without having Siri enabled. The developers haven't built a version of the car interface that functions without the AI assistant, even though listening to a podcast shouldn't require a neural network. The logic here is about the cost of complexity. Companies are forcing these features into products like Microsoft Word, Gmail, and even the Windows 11 File Explorer because the market currently rewards the "AI-first" branding.
However, these features are expensive to run. They require constant server pings and data processing. For a user, this creates a situation where you are forced to pay—either in subscription fees or in the "cost" of your privacy—for a feature you didn't ask for and might not even want to use. Someone might argue that these features are just the natural evolution of the user interface, making things faster and more accessible for the average person.
There is a real case that "no-AI" versions of software are a luxury for the tech-savvy, and that eventually, the "smart" features will become so deeply integrated that opting out will be as difficult as choosing to use a rotary phone today. But for now, there are specific ways to push back. You can go into Chrome flags and search for "GLIC" to filter out Google’s live features, or use browsers like LibreWolf which strip these out by default. On Windows 11, you can sometimes find the Copilot toggle under Personalization and Taskbar settings, or even right-click the icon in the Start menu to uninstall it.
If you have an iPhone 14 or older, you are generally safe from the newest integrated AI models. It is a manual, one-by-one process of reclaiming your own screen.
You just spent the last few minutes learning how to keep a person’s hand off your screen. Now, picture the opposite problem: a room full of bright, expensive screens that nobody is allowed to touch. Japan is in that room. According to the OECD, just 8.4 per cent of Japanese workers use AI as part of their job, at the end of last year.
That puts the country behind the United States and the United Kingdom. Japan has an ageing workforce and a chronic shortage of people, which usually creates a perfect market for machines to step in and do the work. But the machines are sitting on the shelf. Austin Xu, the co-founder of a US start-up called Kuse AI, opened an office in Tokyo specifically to sell his software.
He says the hesitation comes down to a specific cultural habit called process and consensus. In the United States, an AI agent is often given a task, it tries it, and the boss corrects it later. In Japan, the tolerance for mistakes is close to zero, especially in anything client-facing. It is easier to leave a vacancy unfilled than to let a machine make a public error.
Parrisa Haghirian, a professor at Kyoto University, argues that Japanese firms are hyper-sensitive to reputational risk. Because generative AI is not fully reliable, she says it is mostly being used for low-risk tasks like writing or summarising. It is not being used to improve core processes or make decisions. Even the healthcare sector is lagging; one hospital employee described paper documents accumulating at a staggering scale, saying it is “like the Stone Age.” The government acknowledges the problem.
Parliament passed an AI Promotion Act last year with the stated goal of making Japan “the world’s most friendly country for developing and utilizing AI.” The Ministry of Finance claims that 75 per cent of companies now use AI, up from 11 per cent five years ago. The critics say that number is misleading. They argue that in many of those businesses, only a tiny proportion of staff actually use the technology, and they use it very lightly. Yasushi Ogasawara, a professor at Meiji University, points out that the workforce does not have enough people who know how to use it.
He notes that despite liking gadgets like smartphones, Japan has a low level of digital literacy. So Japan has the screens and the machines, but the room stays cold.
Before the last story: if you are getting something out of this, subscribe. The World in Twenty Minutes is on Spotify, on Apple Podcasts, on iHeartRadio, and anywhere else you already listen. It is free, it lands every morning, and subscribing is the whole reason it keeps finding people.
So Japan has the screens and the machines, but the room stays cold. We are jumping continents and altitudes now. The cold on the screen was a failure of habit; the cold up there was a failure of physics. A Gurkha mountaineer and Netflix star died in an avalanche on July 30th, aged 43, according to The Economist.
That is all I have. The line is thin. I do not know what he climbed, how he climbed it, or what he wanted to prove by doing it. I do not know if he was the first, if he was fast, or if he was the last.
I only know that he performed momentous feats in the Death Zone. That is the limit of my knowledge, and I have to live with it. The Death Zone is the part of a mountain above eight thousand metres. Up there, the air is so thin that the human body starts to dissolve.
Every breath is a fight to keep the heart beating, every step is a negotiation with the air. You do not go there for leisure, or even for glory. You go there because you have decided that the danger is acceptable. You decide that your body is worth the risk.
Perhaps you go because you are bored with living a life where nothing is hard enough to hurt you. Perhaps you go because you were told that your people do not belong in the heights, that you are built for the plains. Perhaps you go because the world has taught you that climbing is the only way to be seen. Someone will say he was just rich enough to pay for the risk, that the mountain killed him because the mountain always wins.
That is fair enough. The mountain is a better actor than any man. There is a real case that he was doing what he had to do to belong. He was a Gurkha, and Gurkhas have been told to be useful for two hundred years.
He climbed because the system told him to climb, and he kept climbing because the system never asked him to stop. Here is the thing about the Death Zone. You do not go there to find peace. You go there to disappear.
It is not about the view. It is about the silence. He is gone. We do not know what he looked like on the top of the world.
We do not know what he saw when the snow buried him. All we know is that he went there anyway. That’s the news for today. Take care.