Feynman

If you can't explain it simply, you don't understand it | physics enjoyer

[Akito @itsak1to]

Ron Baron put $400 million into Tesla when almost nobody on Wall Street would touch it. that position is now worth about $8 billion in profit.

he bought between 2014 and 2016, at an average split-adjusted price of $43 a share. it was 1.5% of his firm’s assets, and colleagues thought he’d lost it.

then came the hard part. over the next decade, Tesla dropped more than 30% eleven separate times. eleven chances to panic, take the gain, and look smart.

he never sold a share.

“the stock goes up and down like a yo-yo. I’m a long-term investor. near-term fluctuations aren’t something I can control.”

that single position returned roughly 20 times his money. and he’s not finished - he says he expects five times more over the next decade, and has now put $25 billion into SpaceX on the same thesis.

Baron started his firm in 1982 with $100,000. it manages around $70 billion today.

the math almost nobody can execute: he didn’t need to be right eleven times. he needed to be right once, and then survive eleven chances to be wrong.

bookmark this & comment↓

[Feynman @0xxfeynman]

@vivaanix Quant

[Feynman @0xxfeynman]

@vivaanix open

[Feynman @0xxfeynman]

@arjunquants curve!

[Feynman @0xxfeynman]

@vivaanix Trade

[Feynman @0xxfeynman]

@vivaanix SEND

[Feynman @0xxfeynman]

@0xLeohart Fable

[Akito @itsak1to]

Jordan Belfort took $199 million from 1,513 people. his only weapon was a 76-page script that taught brokers exactly what to say and when to say it.

the system was called the "Straight Line." it covered tonality, pacing, how to build rapport in seconds, and how to kill every objection a person could raise. brokers rehearsed it like actors.

he didn't hire people who understood markets. he hired kids who could read the script and never hang up first.

the stocks themselves were penny garbage. that was the point - the product never mattered, because the script did the work.

at its peak Stratton Oakmont employed over 1,000 people and moved $1 billion in shares.

"it was my desire for instant gratification. I wanted money now."

when it collapsed, Belfort pleaded guilty, wore a wire on his own partners, and served 22 months. then he did something almost nobody expects: he packaged the same technique into a course and sold it.

DiCaprio played him. Scorsese made him a legend. companies now pay him six figures to teach persuasion to their sales teams.

a former Stratton employee put it plainly: "he's just selling a different product now - and the product is Jordan."

the uncomfortable truth isn't that he sold worthless stocks. it's that the exact same words work whether what you're selling is real or not. bookmark this & comment ↓

[Feynman @0xxfeynman]

"i've quit 10x more goals than i've achieved"

that is the man writing the letter, and he thinks that should be true for most people

this is the argument that you aren't where you want to be because you aren't the person who would be there - and that every goal set on top of the old identity is a house built on a rotting foundation

the point is not that the bodybuilder grinds harder than you - it is that he would have to grind to eat badly. the ceo has to force himself to stay in bed and hates every second of it. the behaviour isn't maintained. it's downstream of who they believe they are

then it goes somewhere less comfortable: all behaviour is goal-oriented, including the behaviour you call failure

you don't procrastinate because you lack discipline - you're pursuing the goal of not being judged for finished work

you don't stay in the dead-end job because you were never a risk taker - you're pursuing safety, predictability, and an excuse that doesn't read as failure to anyone watching

when your body is threatened you go into fight or flight. when your identity is threatened the same thing fires - which is why people defend beliefs they never actually chose

the protocol at the end takes one full day: a morning of psychological excavation, timed questions through the day to break autopilot, an evening that compresses it into an anti-vision, a vision, and a one-year lens

one of the morning questions - if absolutely nothing changes for the next five years, describe an average tuesday. where do you wake up. what does your body feel like. what's the first thing you think about. how do you feel at 10pm

another - what is the most embarrassing reason you haven't changed? the one that makes you sound weak, scared or lazy rather than reasonable

he says outright it won't work for everyone, because you can't put the climax at the start of the book

pen, paper, one day. read it below ↓

[DAN KOE @thedankoe]

https://t.co/7l7Jef99QZ

[Akito @itsak1to]

Between 1977 and 1990 Peter Lynch ran the best fund on the planet. 29% a year for thirteen years. He turned $18 million into $14 billion.

He also sold Home Depot and Toys "R" Us while they were still small. Both rose many times over without him. The best stock-picker of his generation walked away from his two biggest winners - on purpose.

One number explains why. A stock can only fall to zero. Down 100%, no further. But it has no ceiling. It can rise 1,000%, 10,000%. Your downside is capped. Your upside is not. A single winner held long enough pays for every loser you will ever take.

So when you sell the stock that's up 40% to lock in the gain, and hold the one that's down 40% because it "has to come back," you do the opposite of what the math rewards. Lynch had a name for it: cutting the flowers and watering the weeds.

Warren Buffett read that line and called Lynch to ask if he could use it. "I have to have it," Buffett said.

Lynch put the whole secret in a paperback that costs fifteen dollars: "The real key to making money in stocks is not to get scared out of them."

The book is free at any library. What's expensive is the winner you sold at 40% that ran 4,000% without you.

[Akito @itsak1to]

Anthropic's CEO says AI will delete half of all entry-level office jobs. then he went back to building the AI that does it.

Dario Amodei warned that AI could wipe out 50% of all entry-level white-collar jobs within five years, sending unemployment as high as 20%. the first to go: finance, law, consulting, and tech.

"most of them are unaware that this is about to happen. it sounds crazy, and people just don't believe it."

he accused his own industry of "sugar-coating" what's coming.

then the data backed him up. entry-level tech hiring fell 30-50% in 2025. Wall Street announced roughly 200,000 job cuts, concentrated in junior analysts. nearly 55,000 US layoffs last year were blamed on AI.

but here's the part worth sitting with: Amodei sells the exact technology he's warning about. every apocalyptic headline is also a sales pitch. one Cisco exec put it bluntly, AI leaders "have to say extreme things to instill FOMO." Nvidia's Jensen Huang disagreed entirely, arguing AI creates more jobs than it kills.

when the person predicting the disaster profits from the fear, the truth is usually in the middle, and your job is to find it before the crowd does. the entry-level jobs really are the most exposed, but "50% gone in five years" is also the best ad an AI company ever ran.

don't panic, don't dismiss it. just be the person who uses the tool, not the one it replaces ↓

[Akito @itsak1to]

Jensen Huang's company was 30 days from bankruptcy. today it's worth $4 trillion. his advice to Stanford graduates: "I wish upon you ample doses of pain and suffering."

he wasn't being cruel. he was handing them the one thing that built Nvidia, and it's the opposite of what elite schools teach.

"people with very high expectations have very low resilience. and unfortunately, resilience matters in success."

Huang knows, because Nvidia almost died. in 1996 its first two chips had failed and the company was weeks from insolvency. it survived on a $5 million lifeline from Sega for a product that didn't even work. for years, Huang said, Nvidia was always "thirty days from going out of business."

decades later, in a single day in 2025, he watched $20 billion of his own net worth vanish in a market panic. he didn't flinch. he'd been far closer to zero before.

"greatness is not intelligence," he said. "greatness comes from character. and character is not formed out of smart people, it's formed out of people who suffered."

the market doesn't reward the ones who expect to win - it rewards the ones who survive losing. high expectations make you fragile. Huang expected the struggle, and outlasted everyone who didn't.

a man worth hundreds of billions wished pain on the students he most wanted to win. he knew it was the only thing that had ever made him.

bookmark this & follow ↓

[Akito @itsak1to]

Warren Buffett has made over 450 investments in his life. he says you should be allowed only 20 - and that limit would make you richer than he is.

at a 1998 lecture, he told a room of MBA students to imagine one rule for their entire career:

"you get a punch card with 20 slots. every financial decision you make punches one hole. when they're gone, you never invest again."

think about what that does. you'd stop chasing the stock a guy bragged about at a party. you'd stop buying things you can't explain. you'd wait years for the one idea worth a punch - then bet on it like it's the last one you'll ever get. because it might be.

"you're not going to get 20 great ideas in your lifetime. you'll get five, or three, or seven. and you can get rich off five, or three, or seven."

this is the opposite of everything the market sells you. Wall Street profits when you trade constantly. Buffett got rich holding Coca-Cola and American Express for 30+ years - doing, most days, absolutely nothing.

Seth Klarman, who manages $30 billion at Baupost, credits this single rule for his entire track record. "invest like you have a single lifetime punch card with 20 punches. make each one count."

in investing, activity feels like progress and is almost always the opposite. amateurs lose by doing too much - too many trades, too many average bets, too much reacting to noise. the punch card forces the one skill nobody has: sitting still until something is truly worth swinging at.

the greatest investor alive didn't get rich from the decisions he made. he got rich from the thousands he refused to.

bookmark this & follow↓

[Feynman @0xxfeynman]

Elon Musk says don't go to medical school because Optimus robots will be better surgeons than any human within 3 years. The prediction deserves the same scrutiny as every timeline he's given.

The claim is specific and confident. 2029, robot surgeons better than any human, at scale. By 4-5 years, "not even close." The best medicine in the world, free.

Here's the pattern worth remembering before anyone drops their pre-med plans.

Musk said full self-driving would arrive in 2017. Then 2018. Then every year since. He said humans would be on Mars by the mid-2020s. His timelines are directionally interesting and almost never accurate on the date. The technology often arrives - years later than promised, and harder than described.

Surgery is a specific problem. The manual dexterity part is genuinely advancing - robotic surgical assistance is real and improving. But "better than any human surgeon at scale" isn't just a dexterity problem. It's judgment under uncertainty, responding to the unexpected mid-operation, ethical decisions about what to do when the plan fails. Those are exactly the capabilities AI is furthest from.

And telling an 18-year-old not to train for a profession based on a 3-year robot timeline from a man whose 3-year timelines have a long history of becoming 10-year timelines is genuinely bad advice.

The direction may be right. Medicine will be transformed by AI. Diagnosis, imaging, drug discovery - already happening.

But "don't become a doctor because robots take over in 2029" is a prediction with a track record attached to the person making it.

Build for the transformation. Don't bet your career on the date.

[Feynman @0xxfeynman]

Elon Musk says don't go to medical school because Optimus robots will be better surgeons than any human within 3 years. The prediction deserves the same scrutiny as every timeline he's given.

The claim is specific and confident. 2029, robot surgeons better than any human, at scale. By 4-5 years, "not even close." The best medicine in the world, free.

Here's the pattern worth remembering before anyone drops their pre-med plans.

Musk said full self-driving would arrive in 2017. Then 2018. Then every year since. He said humans would be on Mars by the mid-2020s. His timelines are directionally interesting and almost never accurate on the date. The technology often arrives - years later than promised, and harder than described.

Surgery is a specific problem. The manual dexterity part is genuinely advancing - robotic surgical assistance is real and improving. But "better than any human surgeon at scale" isn't just a dexterity problem. It's judgment under uncertainty, responding to the unexpected mid-operation, ethical decisions about what to do when the plan fails. Those are exactly the capabilities AI is furthest from.

And telling an 18-year-old not to train for a profession based on a 3-year robot timeline from a man whose 3-year timelines have a long history of becoming 10-year timelines is genuinely bad advice.

The direction may be right. Medicine will be transformed by AI. Diagnosis, imaging, drug discovery - already happening.

But "don't become a doctor because robots take over in 2029" is a prediction with a track record attached to the person making it.

Build for the transformation. Don't bet your career on the date.

[Feynman @0xxfeynman]

Ken Griffin offered a Harvard senior a job at Citadel, then asked what he'd do with $10 million. The kid answered honestly and Griffin told him to turn the offer down.

The question was a filter, and the answer failed it.

"If you made $10 million, what would you do?" The kid said he'd quit and climb the highest peaks in the world. Griffin pulled the offer on the spot - even after making it - because that answer revealed exactly what he was screening for.

His reasoning: "I don't want to hear from somebody who's 22 that there's some number that represents their magic number." He doesn't want people for whom the money is the finish line. He wants people who'd climb the next mountain at Citadel because the work itself is the mountain.

It's a sharp filter and it clearly works. $70 billion, 35 years, a firm built on people who don't have an exit number.

But the filter has a blind spot worth naming.

The kid who answered honestly got punished for honesty. The kid who understood the game would have said "I'd reinvest and build something bigger" - the answer Griffin wanted to hear. Griffin's filter doesn't select for people without an exit number. It selects for people who know not to admit they have one.

Everyone has a number. The billionaires included. The difference is that some people have learned that saying so, out loud, in an interview, is disqualifying.

Griffin's genuine point stands: obsession beats transaction. People who love the work outperform people counting toward an escape.

But the test doesn't measure who lacks an exit.

It measures who's savvy enough to hide it.

[Feynman @0xxfeynman]

Ken Griffin offered a Harvard senior a job at Citadel, then asked what he'd do with $10 million. The kid answered honestly and Griffin told him to turn the offer down.

The question was a filter, and the answer failed it.

"If you made $10 million, what would you do?" The kid said he'd quit and climb the highest peaks in the world. Griffin pulled the offer on the spot - even after making it - because that answer revealed exactly what he was screening for.

His reasoning: "I don't want to hear from somebody who's 22 that there's some number that represents their magic number." He doesn't want people for whom the money is the finish line. He wants people who'd climb the next mountain at Citadel because the work itself is the mountain.

It's a sharp filter and it clearly works. $70 billion, 35 years, a firm built on people who don't have an exit number.

But the filter has a blind spot worth naming.

The kid who answered honestly got punished for honesty. The kid who understood the game would have said "I'd reinvest and build something bigger" - the answer Griffin wanted to hear. Griffin's filter doesn't select for people without an exit number. It selects for people who know not to admit they have one.

Everyone has a number. The billionaires included. The difference is that some people have learned that saying so, out loud, in an interview, is disqualifying.

Griffin's genuine point stands: obsession beats transaction. People who love the work outperform people counting toward an escape.

But the test doesn't measure who lacks an exit.

It measures who's savvy enough to hide it.

[Akito @itsak1to]

Ken Griffin just bought an entire hedge fund's book overnight. Situational Awareness turned a 1,000% run into zero in a single week - and Citadel was waiting to buy the wreckage.

the fund had built its gains on 4x leverage, betting everything on AI and semiconductor stocks. when those dropped 40% this month, the borrowed money vanished. all three of its banks issued margin calls at once. by Monday it had to dump its entire book at any price.

the buyer was Griffin's Citadel. same as it's always been.

1998: Long-Term Capital Management - two Nobel laureates - blows up. Griffin buys the wreckage while everyone else is frozen.

2006: Amaranth loses $6 billion in a week on natural gas. Citadel and JPMorgan take the whole energy book at a discount.

2007: Sowood Capital implodes. Citadel closes the deal at 3:30 in the morning, while rival traders had gone home.

the pattern never changes. someone over-leverages, the market turns, and the moment they're forced to sell at any price - Griffin already has the capital waiting to take it.

the money in a crisis doesn't go to whoever predicted it. it goes to whoever kept cash and nerve while everyone else was fully invested. forced sellers hand you their best assets at the worst possible price - for them.

Citadel now runs $65 billion and made $39.6 billion in trading revenue last year - more than Goldman Sachs.

everyone else fears the crash. Griffin waits for it.bookmark this & follow↓

[Akito @itsak1to]

Ken Griffin just bought an entire hedge fund's book overnight. Situational Awareness turned a 1,000% run into zero in a single week - and Citadel was waiting to buy the wreckage.

the fund had built its gains on 4x leverage, betting everything on AI and semiconductor stocks. when those dropped 40% this month, the borrowed money vanished. all three of its banks issued margin calls at once. by Monday it had to dump its entire book at any price.

the buyer was Griffin's Citadel. same as it's always been.

1998: Long-Term Capital Management - two Nobel laureates - blows up. Griffin buys the wreckage while everyone else is frozen.

2006: Amaranth loses $6 billion in a week on natural gas. Citadel and JPMorgan take the whole energy book at a discount.

2007: Sowood Capital implodes. Citadel closes the deal at 3:30 in the morning, while rival traders had gone home.

the pattern never changes. someone over-leverages, the market turns, and the moment they're forced to sell at any price - Griffin already has the capital waiting to take it.

the money in a crisis doesn't go to whoever predicted it. it goes to whoever kept cash and nerve while everyone else was fully invested. forced sellers hand you their best assets at the worst possible price - for them.

Citadel now runs $65 billion and made $39.6 billion in trading revenue last year - more than Goldman Sachs.

everyone else fears the crash. Griffin waits for it.bookmark this & follow↓

[Feynman @0xxfeynman]

A 91-year-old professor is why Nvidia is worth $4 trillion. His name is Gilbert Strang, and every AI model on Earth runs on the course he put on YouTube for free in 2005.

MIT 18.06 opens with "The Geometry of Linear Equations." No advanced math. Strang takes two equations, draws them two ways, and shows that a matrix is a picture, not an abstraction.

The row picture is two lines that cross. The column picture is two arrows that sum to a target. Every neural network on Earth operates on the column picture.

His central insight is the thing most people are never taught. A matrix is not bookkeeping. It's a linear transformation - a way of moving space. Once you see the space actually move, the math stops being algebra and becomes geometry.

That reframe is everything. GPT is a stack of matrix-vector products. Each one is a scene from 18.06 running on a Blackwell GPU. PCA, the Kalman filter, every gradient step in training - all linear algebra, all the geometry Strang has been drawing on a chalkboard since 1962.

He wrote the textbook in 1976. It's on every serious engineer's shelf. Every quant fund, every ML lab, every rendering engine at Pixar runs his math.

The chip costs $40,000. The course that explains what the chip is doing costs nothing. He never asked for a royalty.

The most valuable idea in the AI economy - that a matrix moves space - has been free for twenty years.

The GPU is the expensive part.

Understanding what it's actually computing was always available at zero cost.

[Feynman @0xxfeynman]

A 91-year-old professor is why Nvidia is worth $4 trillion. His name is Gilbert Strang, and every AI model on Earth runs on the course he put on YouTube for free in 2005.

MIT 18.06 opens with "The Geometry of Linear Equations." No advanced math. Strang takes two equations, draws them two ways, and shows that a matrix is a picture, not an abstraction.

The row picture is two lines that cross. The column picture is two arrows that sum to a target. Every neural network on Earth operates on the column picture.

His central insight is the thing most people are never taught. A matrix is not bookkeeping. It's a linear transformation - a way of moving space. Once you see the space actually move, the math stops being algebra and becomes geometry.

That reframe is everything. GPT is a stack of matrix-vector products. Each one is a scene from 18.06 running on a Blackwell GPU. PCA, the Kalman filter, every gradient step in training - all linear algebra, all the geometry Strang has been drawing on a chalkboard since 1962.

He wrote the textbook in 1976. It's on every serious engineer's shelf. Every quant fund, every ML lab, every rendering engine at Pixar runs his math.

The chip costs $40,000. The course that explains what the chip is doing costs nothing. He never asked for a royalty.

The most valuable idea in the AI economy - that a matrix moves space - has been free for twenty years.

The GPU is the expensive part.

Understanding what it's actually computing was always available at zero cost.

[Kevin Carpenter @kejca]

Charlie Munger: "In terms of humility, I've frequently said that when they passed that out I didn't get my full share. That was a serious problem when I was young — and I only cured it, partly, by becoming very rich and generous." 🤣

"If humility means that you know the edge of your own competency and you aren't arrogantly stepping over the boundary, I'm very good at that. But within my area of competency, my best friend would not adore me for my humility."

(Caltech || 2008)