Merlow
@exeMerlow • 4,289 subscribers
control attention
Videos
Richard Feynman stood at a Cornell blackboard in 1964 and explained the problem every AI lab is fighting in 2026. The BBC filmed it. Almost nobody watches it. The lecture is about why nature only answers in mathematics. Every team forcing language models to reason is hitting the wall he mapped 62 years ago. He was 46. The Nobel Prize came 11 months later. The footage survived on film reels and now sits free on YouTube with fewer views than a keyboard unboxing. Watch the blackboard section near the middle. He takes 1 of Kepler's laws and rebuilds it from nothing, with notation a 12-year-old can follow. No slides. No jargon. 1 piece of chalk. An ML engineer I know paused it 4 times and made his whole team watch it before standup. You're 62 years late. The lecture is still free.
Merlow2,598,282 просмотров • 2 месяцев назад
A man who turned 7 million dollars into 500 million in 10 years sat on a desk in a classroom and taught the method to 30 students. The man is Joel Greenblatt. He ran Gotham Capital from 1985 to 1994 at roughly 50% a year, then returned all outside capital in 1995 and kept running his own money. Almost nobody sustains that number over a decade. A Columbia classroom in 2005. Chalkboard, blue plastic chairs, a camera at the back of the room recording for the students who missed it. He sits on the edge of the desk with a stack of paper. He does not teach a screen or a formula. He teaches situations most funds cannot touch, the corners of the market where the usual buyers are structurally forced to sell regardless of price. The first lecture is where he sets up why those corners exist at all, and why they keep existing after everybody knows about them. The uncomfortable turn is what he says about diversification and about how few positions he actually held. It runs directly against everything the same school teaches two floors down. It matters more now. Every screener is free, every filing is searchable, and the constraint was never information. Free on YouTube, filmed from the back row, audio uneven, students blocking the frame. He gave away 50% a year to a room of 30 people. The tape has been public for years. One classroom. 1 camera. It is in the video.
Merlow186,823 просмотров • 1 месяц назад
He didn't believe the Kimi K3 hype. So he ran it against Claude Fable 5 on camera. Same prompts, both models, 12 minutes, results raw on screen. No cherry-picking, nowhere to hide. The context makes it spicy: Moonshot itself admits K3 trails Fable 5 overall while posting wins on coding benchmarks. 2.8 trillion parameters against the model everyone pays for. Benchmarks are marketing until somebody runs them at home. This is somebody running them at home. Watch the middle section where the outputs go up side by side. The gap is visible before he says a word. The verdict lands in the last minute. No spoilers. Leaderboards don't write code. Models do.
Merlow148,234 просмотров • 2 месяцев назад
A YouTuber who burns $1,000 a day on AI tokens just did the same work for $63. His terminal froze on camera from running too many agents. He force-quit and kept filming. He skipped a live event to finish this. The model is Kimi K3, China's new 2.8 trillion parameter release. Weights go public July 27. In the 41-minute breakdown he feeds it his old production codebase with a paragraph and a half of instructions. It clears 122 tasks before he touches the keyboard again. Then around minute 37 it does something to his to-do list he says he's never seen from any model. Not even Anthropic's flagship. The security audit near the 33-minute mark is the part that should worry you. Frontier models refuse that task. This one ran 25 verification agents on it. He came in bored of open weights. He left calling it frontier class. Hosting it yourself costs $2.6 million. Watching him break it costs nothing.
Merlow121,153 просмотров • 2 месяцев назад
A banking professor was asked to describe the money supply and answered that the textbook version is not how it works. Richard Werner, 2019, a curtain, a suit, no audience. His question is narrow and almost nobody asks it: why do lending and deposit taking sit inside the same firm at all. Split them and the whole mechanism stops. The reason is an accounting reason, not an economic one. Watch the part where he moves from that split to why no other company on earth can do the same thing. No slides, no stage, a camera in a room. Every other firm lends what it has. One kind of firm writes the number first.
Merlow12,157 просмотров • 17 дней назад
Nobody can track 10^23 particles. A professor in Chennai spends an hour explaining why that stopped being a problem 150 years ago. This is Classical Physics, lecture 20, from V. Balakrishnan's course at IIT Madras, on the founding postulates of equilibrium statistical mechanics. The move is a surrender that pays. Stop following trajectories. Count how many microscopic arrangements produce the same macroscopic state, and temperature, pressure, and entropy fall out as consequences. Every sampling method in machine learning is a descendant. Diffusion models, MCMC, Boltzmann distributions in softmax. All of it is state counting wearing different notation. Watch how he handles the postulates themselves. He states what is assumed rather than derived, then refuses to pretend the assumption is a proof. A machine learning engineer I know rewatched the first 25 minutes and said sampling stopped looking like a trick. Free on YouTube, low resolution, green chalkboard. The particles were never the point. The counting was.
Merlow24,534 просмотров • 1 месяц назад
A Nobel laureate flew to Bengaluru and spent 90 minutes on the questions physics still cannot answer. The stream sits almost unwatched. This is David Gross at the Chandrasekhar Auditorium, ICTS-TIFR, August 2025. The subject is the frontiers of fundamental physics. Gross shared the 2004 Nobel Prize for asymptotic freedom, the discovery that the force binding quarks weakens as they are pushed together. He also worked on string theory as an approach to quantum gravity. The talk is a map of the open territory. What the standard model explains, what it never touched, and where the field has been stuck for decades. Watch how he separates the settled from the unsettled. No hedging, no salesmanship, a man who won the prize describing what remains unsolved. A physicist I know rewatched the section on quantum gravity twice and said it was the cleanest statement of the gap he had heard. Free on YouTube from an Indian research institute, full lecture, subtitles on. The prizes go to the answers. The frontier is still a list of questions.
Merlow24,256 просмотров • 1 месяц назад
Richard Feynman stood at a blackboard in 1964 and drew the line between knowing and understanding. Every AI benchmark in 2026 sits on the wrong side of it. Almost nobody watches the clip. His father taught him the difference as a boy. You can know a bird's name in every language and know nothing about the bird. Names track labels. Understanding survives when the labels change. He carried that into physics. You can compute every answer and understand none of them. A student who memorizes the formula knows. A student who can rebuild it from nothing understands. Now run today's models through his test. They name everything, compute everything, pass every exam. The 1 question underneath hasn't moved: is anything understood. He couldn't have imagined the machines. He built their hardest eval 62 years early. Watch him at the board. He never recites. He rebuilds the idea live in front of the room, which was exactly his point. No slides. No notes. 1 piece of chalk. Benchmarks measure knowing. Nobody has built one for the other thing. You're 62 years late. The clip is still free.
Merlow14,979 просмотров • 2 месяцев назад
A physics professor ran a live demonstration on national television in 1964 and it went wrong on camera. He didn't cut away. He got happier. This is Julius Sumner Miller on the ABC program Why Is It So?, crushing a reinforced metal tank with atmospheric pressure. The experiment refuses to behave. Watch his face when it fails. No embarrassment, no apology to the audience, no reshoot. Failure was the interesting part and he wanted the camera on it. Every engineer who hides a broken demo, every team that only ships the run that worked, is doing the opposite of what this man did on live TV 62 years ago. The physics is simple: air pressure you can't feel, doing work you can't ignore. He makes it visible with a tank and a heat source. A hardware engineer I know sent this clip to his team after a failed launch demo with 1 line: this is how you handle it. Black and white, 1 camera, free on YouTube since 2018. He wasn't performing certainty. That's why it still plays.
Merlow11,676 просмотров • 1 месяц назад
The inventor of television appeared on TV exactly once in his life. As a mystery guest nobody recognized. He won $80 and a carton of cigarettes. July 3, 1957. CBS, "I've Got a Secret." The man is Philo Farnsworth. The panel's job was to guess his secret. The secret on the screen behind him read: "I invented electronic television." They were sitting inside his invention. They questioned him for minutes. Nobody guessed. The backstory makes it worse. At 14, plowing an Idaho potato field, he saw the solution: scan an image line by line, like plowing rows. He sketched the tube on a high school blackboard. At 20 he produced the first all-electronic image. At 24 he held the patent. RCA fought him in court for a decade and history filed him under nobody. The medium he created minted celebrities every night and never learned his face. Watch the panel's reaction at the reveal. Polite applause. Then the host hands the man who built the entire industry $80 and Winstons from the sponsor. The machine forgot its own father. It hasn't stopped forgetting since.
Merlow10,305 просмотров • 2 месяцев назад
Больше нет контента для загрузки