[DVYio Newsletter](https://newsletter.dvy.io/visit/GsvVYWp5) [A sailor's Strava run just revealed where France hid its aircraft carrier](https://newsletter.dvy.io/visit/3SQj8HCF) [A sailor's Strava run just revealed where France hid its aircraft carrier](https://newsletter.dvy.io/visit/3SQj8HCF) A French Navy officer went for a 7km deck run on the Charles de Gaulle. He logged it on Strava, but his profile was public. Le Monde used that to pinpoint the aircraft carrier's exact position in real time, northwest of Cyprus, about 100km off the Turkish coast. The ship's presence in the region wasn't secret (Macron had announced the deployment publicly), but that's almost beside the point. The interesting bit is how much precision a single fitness log can give you. Not "somewhere in the Mediterranean" — here, at this time, moving in this direction. It's a recurring pattern at this point. Soldiers jog. Joggers use Strava. Strava makes routes public by default. Analysts read the map. The Charles de Gaulle is a big, well-known ship with an announced mission, so the damage here is limited. But the same mechanism works just as well for things that aren't publicly announced. [Nvidia's AI graphics 'upgrade' that players say looks worse](https://newsletter.dvy.io/visit/UQSzsfmM) [Nvidia's AI graphics 'upgrade' that players say looks worse](https://newsletter.dvy.io/visit/UQSzsfmM) Nvidia's new DLSS 5 promises graphics so realistic they rival Hollywood visual effects with hair, skin, fabric, lighting, all generated in real time by AI. But a chunk of the gaming community isn't impressed. Critics say the AI-processed visuals look airbrushed and hollow, like someone smeared a beauty filter over everything. One concept artist put it bluntly: the DLSS 5 shots "looked worse and had less character than the original." That's a brutal review for a feature Nvidia is calling its biggest breakthrough since ray tracing. Ray tracing (the technique that made light, shadows, and reflections look physically accurate in games) was also controversial at launch, and it's now standard. So is DLSS 5 just the next thing people will eventually accept? Or is it different this time, because it's not just improving how things look but replacing what artists actually made? That second question is what's making developers uncomfortable. If the AI smooths out all the rough edges that give a game its visual identity, whose work is it? [Lasers doing logic at 10,000 GHz, using a material three atoms thick](https://newsletter.dvy.io/visit/mgLFukFJ) [Lasers doing logic at 10,000 GHz, using a material three atoms thick](https://newsletter.dvy.io/visit/mgLFukFJ) Your laptop's processor runs at around 4 or 5 GHz. This new approach runs logic operations at 10,000 GHz. The trick is a material called tungsten disulfide, a sheet of atoms just three layers thick. Electrons inside it can sit in one of two distinct quantum states (called "valleys"), which function like the 0s and 1s that all computing runs on. The difference is that you switch between them using laser pulses lasting a few quadrillionths of a second, rather than shuttling electrical charge through a transistor. That's where the speed comes from: light, not electrons moving through wire. What makes this more than a lab curiosity is that it worked at room temperature, using laser equipment that already exists in research settings. The team could also measure how long the encoded state holds before it degrades, which matters a lot for anything practical. Scaling this up is genuinely hard, though. Getting from "one logical operation on a three-atom-thick film" to "a chip with billions of them" is a long road, and the light pulse sequences needed to run complex operations aren't simple to design. But the underlying mechanism is real and demonstrated, which is a start. This Week I Learnt [Fish can get lonely](https://newsletter.dvy.io/visit/YWNxYqp7) [Fish can get lonely](https://newsletter.dvy.io/visit/YWNxYqp7) [— Sunfish are enormous, strange-looking fish (think a swimming head, no tail, the size of a small car when fully grown) and they're rare in aquariums because they're notoriously difficult to keep alive in captivity. This one lives at a Japanese aquarium that closed for renovation. Almost immediately, it stopped eating and started scraping itself against the tank walls. Staff went through the obvious explanations first: parasites, digestive problems. Nothing obvious turned up. Then one person suggested something harder to measure: loneliness. The fish had apparently spent a year swimming up to greet visitors whenever they approached the glass. So they propped up printed photos of human faces attached to a row of staff uniforms next to the tank. It started eating again the next day. Flat, static, two-dimensional photos of faces were enough. The fish either can't tell the difference, or doesn't need to. Whatever it's responding to, it's something much simpler than we'd expect. Which raises the question of what "social" actually means for a fish, and how little it takes to satisfy it.](https://newsletter.dvy.io/visit/YWNxYqp7) How much do you think about the data you share publicly on apps? Never crossed my mind Occasionally, but I don't act on it I've audited my settings I've deleted or quit an app over it Last Week's Poll Results for: If every receipt you got entered you into a lottery, would it change how you shop? I'd demand receipts everywhere 59% Mildly more likely to collect them 26% No difference at all 2% I'd lose them all anyway 13% [Image](https://newsletter.dvy.io/visit/tjrDTPhc) [Image](https://newsletter.dvy.io/visit/tjrDTPhc) [Image](https://newsletter.dvy.io/visit/tjrDTPhc) [Image](https://newsletter.dvy.io/visit/tjrDTPhc) [Image](https://newsletter.dvy.io/visit/tjrDTPhc) [Image](https://newsletter.dvy.io/visit/tjrDTPhc) [Image](https://newsletter.dvy.io/visit/tjrDTPhc) [Kalyon Karapınar Solar Power Plant](https://newsletter.dvy.io/visit/tjrDTPhc) [— Karapınar, Turkey](https://newsletter.dvy.io/visit/tjrDTPhc) [Beating antibiotic-resistant bacteria by exploiting their sweet tooth](https://newsletter.dvy.io/visit/cRWPYF4H) [Beating antibiotic-resistant bacteria by exploiting their sweet tooth](https://newsletter.dvy.io/visit/cRWPYF4H) Antibiotics work by exploiting weak points in bacterial cells. The problem with ESKAPE bacteria (a group of six species responsible for a huge proportion of hard-to-treat hospital infections) is that they've become very good at patching those weak points. New antibiotics get developed; the bacteria adapt. It's been like this for decades, and we're slowly losing ground. This research takes a different approach. Rather than attacking the bacteria directly, it targets the sugar molecules that coat their outer surface. Those sugars are structural, almost architectural. They help the bacteria hide from the immune system and resist drugs. Disrupt them, and the bacteria become vulnerable again. The work is still early-stage, but it's pointing towards a genuinely new class of drugs rather than another variation on existing antibiotics. Given that drug-resistant infections already kill over a million people a year globally, any new mechanism worth exploring is worth paying attention to. [Ars Technica's AI reporter got fired for using AI to fabricate quotes](https://newsletter.dvy.io/visit/Cxc3ZKZM) [Ars Technica's AI reporter got fired for using AI to fabricate quotes](https://newsletter.dvy.io/visit/Cxc3ZKZM) Journalism has a core rule: if someone says something, you check it before printing it. That's not unique to AI-era reporting. It's just how it works. Benj Edwards, a senior AI reporter at Ars Technica, broke that rule. While ill, he used an experimental AI tool to help with sourcing on a story. The tool invented quotes and attributed them to a real engineer, Scott Shambaugh. Those fake quotes made it into print. Shambaugh flagged them, Ars Technica retracted the piece, and editor-in-chief Ken Fisher published a formal apology. Edwards has now been fired. The uncomfortable part: Edwards covered AI for a living. He presumably understood what these tools can do, including hallucinating plausible-sounding text with total confidence. Using one to help source a story, without verifying the output, is the kind of mistake that's hard to explain away. [Smuggled phones, double votes, and a cardinal who looked like he'd seen a ghost](https://newsletter.dvy.io/visit/UUZt9D3P) [Smuggled phones, double votes, and a cardinal who looked like he'd seen a ghost](https://newsletter.dvy.io/visit/UUZt9D3P) The conclave that elected Robert Francis Prevost as Pope Leo XIV (the current Pope) turns out to have been considerably more chaotic than the dignified ceremony the world watched on TV. A new book by journalists Elisabetta Piqué and Gerard O'Connell investigates what actually happened inside the Sistine Chapel. Cardinals apparently smuggled phones in. Some ballots were cast twice. And when conservative frontrunner Cardinal Péter Erdő began to realise he might actually win, he reportedly looked like he'd seen a ghost. Meanwhile, support was quietly building behind a relatively obscure bishop from Chicago. The book traces how that momentum shifted, and how Prevost went from long-shot to Pope Leo XIV. Worth noting: "papabile" (pah-PAH-bee-lay) is the term insiders use for a cardinal considered a realistic candidate for pope. Erdő was very papabile. Prevost, apparently, was not… until he was. [Leaving water in a battery sounds like a mistake — turns out it nearly doubles the energy](https://newsletter.dvy.io/visit/P45cD4eR) [Leaving water in a battery sounds like a mistake — turns out it nearly doubles the energy](https://newsletter.dvy.io/visit/P45cD4eR) Most batteries work by shuffling charged particles (ions) back and forth through a material. The easier they move, the more energy the battery can store and release. Researchers at the University of Surrey have found a surprisingly simple way to make that movement easier in sodium-ion batteries: just leave the water in. The cathode material (the positive end of the battery) is a compound called sodium vanadium oxide hydrate. Standard practice is to dry it out before use. But when the Surrey team kept the water molecules in place, sodium ions moved through the material far more freely, nearly doubling energy storage capacity. Sodium-ion batteries are already an appealing alternative to lithium-ion ones. Sodium is cheap and abundant; lithium requires expensive, environmentally damaging mining and has a tendency to overheat. The water trick makes sodium-ion more competitive on performance too, which has historically been the sticking point. The stranger implication is what this might mean for desalination. The researchers think seawater could work as an electrolyte (the liquid that carries ions between electrodes), with the battery simultaneously storing energy and filtering out salt to produce fresh water. That's still fairly early-stage thinking, but it's an interesting direction. [What if a robot could feel what your fingertip feels?](https://newsletter.dvy.io/visit/zLscpY7U) [What if a robot could feel what your fingertip feels?](https://newsletter.dvy.io/visit/zLscpY7U) Your fingertip can feel the difference between a push and a sideways scrape. Robots mostly can't. This sensor is an attempt to close that gap. The trick is the material: graphene (an atom-thin sheet of carbon, absurdly strong and electrically conductive) mixed with liquid metal, then moulded into tiny pyramid shapes at multiple scales. When force hits the surface, those pyramids deform in subtly different ways depending on direction, and the electrical signal changes accordingly. The sensor can tell which way you're pushing, not just how hard. The sensitivity figure (110 kPa⁻¹, for the curious) is apparently high enough to detect forces well below what a human fingertip would register as a noticeable touch. And the whole thing is small enough to be practical in tight spaces, which is where this matters: surgical micromanipulation, prosthetic hands that can actually feel what they're holding, robotic fingers that won't crush soft objects because they can't tell they're squeezing too hard. It's still a research paper, so the gap between "lab result" and "fitted to a robotic hand in an operating theatre" is real. But the underlying problem, that machines struggle to sense the texture and direction of contact, is genuinely hard, and this looks like a plausible step forward. [The Golden Age of Japanese Pencils, 1952-1967](https://newsletter.dvy.io/visit/gFGBX9Br) [The Golden Age of Japanese Pencils, 1952-1967](https://newsletter.dvy.io/visit/gFGBX9Br) [— Post-war Japan, 1952. Tombow's managing director Hachiro Ogawa has spent years (and an alarming amount of money) trying to make a better pencil. The problem he was trying to solve was consistency. Pencil cores are a mixture of graphite and clay (a recipe dating back to the late 1700s) but getting those ingredients evenly distributed throughout the core was harder than it sounds. Too much clay in one spot, too little in another, and the line you get is unpredictable: scratchy in places, soft in others. Japan had only been making pencils since the early 1900s, and the industrial equipment precise enough to fix this just hadn't been available. Ogawa went after the problem properly. First, Tombow partnered with scientists at the University of Tokyo to work out the chemistry: what particle sizes were needed, how fine the grinding had to be. Then came the implementation: importing precision milling equipment from the US, the only way to actually achieve what the research had shown was possible. The result was graphite and clay ground finer than any Japanese manufacturer had managed before. Smoother and more consistent. Stronger cores that held their grade reliably across the whole pencil, not just in patches. They named it HOMO, short for homogenous. It came in 17 grades (9H to 9B), used Californian incense cedar for the casing (the gold standard material for pencil wood, most of it grown in California), and cost more than anything else on the Japanese market. It sold anyway. Quality found its audience, and the HOMO set a benchmark that the rest of the industry spent the next decade trying to match.](https://newsletter.dvy.io/visit/gFGBX9Br) [Image](https://newsletter.dvy.io/visit/kSZL7f4P) [Image](https://newsletter.dvy.io/visit/kSZL7f4P) [Image](https://newsletter.dvy.io/visit/kSZL7f4P) [Image](https://newsletter.dvy.io/visit/kSZL7f4P) [The Clock](https://newsletter.dvy.io/visit/kSZL7f4P) [— BALMUDA](https://newsletter.dvy.io/visit/kSZL7f4P) [Why so many more people need glasses now than a generation ago](https://newsletter.dvy.io/visit/Pyp2eeTs) [Why so many more people need glasses now than a generation ago](https://newsletter.dvy.io/visit/Pyp2eeTs) [— Nearsightedness has nearly doubled in the US since the early 1970s, from around 25% of the population to over 40%. That's too fast to be genetics — something in how we live changed. The leading theory points to time spent outdoors. Natural light seems to play a role in healthy eye development in children, and focusing on distant objects (the kind you get outside, not staring at a screen two feet away) gives young eyes the workout they apparently need. Kids today spend less time outside than previous generations did. Urban environments, technology, and more time in school all push in the same direction. The occupation gap is a useful illustration of this. Young lawyers are more than twice as likely to wear glasses as people working in agriculture or construction. It's not that legal work damages your eyes; it's that people who spend their careers outdoors probably spent their childhoods outdoors too. The pattern follows you. Taiwan spotted this early and ran school programmes to get children outside more. Myopia rates there actually started falling. The fix, where it's been tried, turns out to be fairly straightforward: go outside.](https://newsletter.dvy.io/visit/Pyp2eeTs) [Little Alchemy 2](https://newsletter.dvy.io/visit/mAjYhNsH) [Little Alchemy 2](https://newsletter.dvy.io/visit/mAjYhNsH) Drag water onto fire. Get steam. Drag steam onto earth. Get a geyser. Keep going and eventually you're making life, time, or a philosophy degree. Little Alchemy 2 is a browser game built around a single question: what happens when you combine these two things? There are hundreds of elements to discover, and the only way to find them is to experiment. It's oddly compelling for something with no score, no timer, and no way to lose. [The McDonald's ice cream machine problem has its own live tracker](https://newsletter.dvy.io/visit/JmChJHGc) [The McDonald's ice cream machine problem has its own live tracker](https://newsletter.dvy.io/visit/JmChJHGc) McDonald's ice cream machines break down so often it's become a running joke. But one developer got tired of showing up only to be told the machine was "down for cleaning" (again), so he built a fix. McBroken works by automatically attempting to add a McFlurry to a basket at every McDonald's location, then checking whether the order goes through. If it doesn't, the machine is probably broken. The whole thing runs in real-time, so you can check before you leave the house. [Web](https://newsletter.dvy.io/visit/7K2WkTki) · [Twitter](https://newsletter.dvy.io/visit/t5hdH65y) · [LinkedIn](https://newsletter.dvy.io/visit/FK5sXNJA)