George Hotz | Programming | what is programming? (noob lessons!) | Science & Technology

George Hotz | Programming | what is programming? (noob lessons!) | Science & Technology - https://www.youtube.com/watch?v=N2bXEUSAiTI

Hello. Cool. So I was watching Martin Shkreli's finance lectures and he gave lectures with a list of it's kind of financial advice. And he starts out by saying, what is finance? He breaks it down into, well, okay, if you want to get into it, you have a choice. You can either invest your own money or other people's money. Investing your own money probably only makes sense if you have a lot of it. You need tools to. There's no reason you should be investing your money if somebody else can invest it better. Basically, good advice. So I thought I'd do something similar for programming. I was watching. I Googled how to learn to program. Let's do this. How to learn to program also. Again, like, you guys think I'm on drugs constantly. I'm really not. I smoke weed occasionally, I drink occasionally. I don't really do harder drugs than that. I don't think I've taken Adderall or any other stimulants this year. I don't know why you guys all think that maybe there's something defective about my brain chemistry. So I'm not going to teach you Vim if you think you want to learn. Park type. Thank you for subscribing. And I'm hoping to get a lot of subscribers to this stream because I'm going to try to educate instead of just a lot of times when I talk, I'm not trying to be made sense of. I'm not. I'm trying to. I'm thinking out loud. And I'm not trying to explain things such that an audience will understand. So we could say, yeah, look, I can tell you, here's Lifehacker.com, the best ways to teach yourself to code. And really it comes down to just doing it, like almost everything. But I think a question a lot of people have is, what exactly do I do? So let's bring them up here and let's talk about what a program is. Okay, so you have a program. What's the definition of a program? Well, you can think about it like this. A program takes in some input, does some computation, and produces some output. So there's an argument to be made that we should be teaching kids how to program using functional languages, because functional languages more closely match this paradigm. But computers don't operate like functions. Computers operate a lot more like Turing machines, which we can get into too. So maybe on one side of an extreme, we have like a high level functional programming language like Haskell. And then let's not really go lower than 6C and assembly are. C and assembly are the same structurally. So you can treat C and assembly like the same thing. C is kind of assembly with a bit of syntactic sugar around it. There is nothing in C that doesn't very clearly map to something in assembly. Please subscribe. I hope you guys like this kind of content. Okay, so let's say like languages span the gamut from C over to Haskell, right? Then you can take languages and languages go in a few directions. You can think of this as like the functional spectrum, and then maybe there's C over to Python and this is kind of the ease of use spectrum. So in the 90s, it was believed that object oriented programming was going to be this revolution in programming. That's C going to C, right? Basically. And you can take it even further. Microsoft created these things called com objects, and everything was going to be an object to pass around. Smalltalk has this kind of paradigm also. But it turned out this didn't actually improve programmer productivity. The thing that improved programmer productivity was garbage collection. So when you think of a computer, first, I mean, we kind of have to ask the question, what is a computer? So you can think about it like you have a processor that runs a stream of instructions and then you have ram. Now you can break RAM into two things. There's two things stored in ram. There's instructions and there's data. So most computers really don't treat these things very differently. When I make a program like hello World and I compile this, we can take a look in this program. And that's not even all of it. Yeah, so we have a. This section here, this is actually the ASCII for hello World. So it's in the same memory space as the instructions itself, which does the printing of hello World. Here's the main function. This call to Dylad Stubbinder is actually what calls the printf. So yeah, what is a computer processor and ram? Then you can break down your program. So when you have a program, a program has text, it's called text, which is the instructions. A program has a stack, program has a heap, and a program has bss. So BSS are things like static data. There's a stack and then there's a heap. Heaps you get to through Malloc. Stacks you get to through anything like this is a variable on the stack. Local vars also control flows on the stack. If I have a main function and I call A, what this does is pushes the return location to the stack. So, you know, when I call B down here, this at the end here is returned by popping off stack. Okay, so this is a basic idea of what a computer is. Now there's a whole nother direction you can go with the idea of what is programming. Right. There is a practical, like, I want to get a job, right? Programming for work. Maybe we'll make a note too. We'll call it programming for work. So what does this mean? What does a software engineer do? Well, they don't really write algorithms. You will never in any job have to write a sort algorithm, a binary search algorithm, anything like that. So they don't write algorithms. What else? Well, very few people actually work on. There's some people who work on infrastructure, but that's not a lot of people. Most software engineers are basically translators and they translate a language, maybe we'll call the language business requirements into code. And there's a whole lot of frameworks which exist to make writing things that meet business requirements easier. If you think of a business requirement like, you know, we need a web page that's going to allow users to say, leave a phone number where we can call them back. Right. There's a lot of frameworks that are designed to do things like that. If you wanted to write something on that, you could use something like Ruby on Rails. Right. And that's why I hate when Vim does that. I don't actually know for like, you know, know, web apps. There's also this paradigm of CRUD apps. Let me get what CRUD stands for. The four basic types of SQL, command, Create, read, update, delete. So let's say we want to build Ruby on Rails or similar for web apps, and you're basically creating an app that looks like Create, read, update, delete. There's some front end, the view, there's some database which is the model, and then there's the business logic, which is a controller. So let's say the front end is. Let's go back to that. I want to build a portal where my customers can add their phone number such that we'll be able to call them. They have to be able to update their phone number due to gdpr. They have to be able to delete their profile. You can see how it translates to this, right? So this isn't. This is terrible. You're not just the same way. The word doctor means so many things. A doctor is someone who diagnoses a doctor. Someone who prescribes. A doctor is someone who has bedside manner. A software engineer, like, it's kind of misleading and it's almost nothing like what you do in school, if you're in school and you're learning how to write binary search algorithms, it's nothing like what you do in practice. What you do in practice is really, you're translating from a really shitty language into these things that are like code. And I meet these people and this is what you're going to learn, what you will learn in boot camps, right? So for example, no one who's ever gone to a boot camp would ever get a job at comma, AI. Because at comma you don't do this. Nothing at comma looks like this, this stupid paradigm. And this is the thing that's going to be replaced pretty soon by, by AIs. I think about writing this like if you want to write the low brow version of backspace, backspace is basically this. I mean, so here, step one, build a CRUD app contracting firm. Step two, record all the inputs of my contract developers. Say contract developers and make fun of this one full time developers train AI model to translate business logic into code translators. So yeah, if you guys are thinking about a job like this, you know, it's, it's, it's truly, it's really trash work. You can get paid some amount of money doing it. But your, your monks used to do this. Monks used to like spend time copying. They'd have to. Before there were printing presses. They'd have to. Well, someone's got to copy the Bible. So that's practically what you're doing. It's not very, you know, it's kind of trash, right? So notice how none of this go back to this guy here. None of this has anything to do with this. They're completely separate things. A software engineer, a lot of these people who I meet who are software engineers know basically nothing about computers. What they know about is how to translate this shitty language into code is almost giving it too much credit. And then here, so let's think of like, like Ruby react, right? Think of all the very heavy frameworks. It almost has nothing to do with programming and it's like you're memorizing weird syntax, something stupid. So let's talk about things that are more interesting. So I come from, I learned this stuff from my kind of hacking background. What is hacking? Well, so again, programs have input, computation output. So if you want to attack the system, we'll call this a system, right? If you want to say get like a, you want to get like remote code execution on the system, right? The question you ask is what input to the system achieves my desired outcome. So again, a lot of hacking is because, like the, the, the system didn't think to check for what it is I'm sending it. And you can get the system to behave in ways that aren't within the normal bounds of output. So if you think this is actually a function, right? What are the words in function? You have a domain, there's the input to a function, function, range, or like, you know, Y equals F of X, right? So this is a, this is a pure model. But most computation is impure. When it's impure, the function doesn't necessarily have to output something in the range. It can do something entirely different. So yeah, I mean, what is, what is hacking? It's figuring out how the function, you know, figuring out how to make the function behave how you want. Now you can use the same paradigm to analyze almost anything, right? Input, system output. Say you want to change your flight on United and you don't want to pay change fees. Well, you can think of. The input is, what can I do? Well, I have a. I can call, I can call the United customer support line. The system is the customer support agent, and the output is them clicking a button on their computer that says wave change fees. Probably written in some CRUD app, right? So if you think about it like that, you think, okay, how do I get the system? What inputs? Well, what do I have control over? I can make multiple calls. I can change the words that I say on each call. You can do things that are completely out of bounds of the system entirely. You can, you know, get the person's real name. You can dox them, you can threaten them. You might go to jail for these things. I'm not advising you do it. But it's all within this paradigm of input, system output. And that's what, what is programming? Input, computation, output. Thank you for subscribing. No, no, there is. It's starting to get big. Are we getting people watching this stream? Oh, yeah. Okay, now we. Let's get viewers. Let's get viewers. So when you ask the question, what is programming? How do I get good at programming? I don't understand what it is that you want. Any, any website, any boot camp will teach you functioning CRUD app programming. When it comes down to hacking, every system is different. If you want to get good at hacking, you just need to practice this basic idea. Remember that one of the reasons I don't play CTFS anymore is because it became too boxed. I feel similarly about competition programming. Competition programming is not about who can cleverly come up with New algorithms. It's just about, you know, pattern recognition. Okay. Recognize that this is actually a graph isomorphism problem. I can do graph isomorphism, that's np, but, oh, this is actually a binary search over this and I can sort the list in N log n instead of N squared. And I can solve it in time, right? Yeah. So CTFs are still fun and CTF still like. I think there's some that kind of probably break the paradigm. But a lot of it became the same tricks over and over again. And you just drilled once you can drill. It's no longer a test of intelligence. It's a test of like, you know, are you willing to practice? Are you willing to put 40 hours into a week into learning some relatively useful skills? I shouldn't hate on CTFs, and they're a great way to get into this stuff. It's just a thing like back in the day, CTFS didn't feel as much like this, but competition programming already did. And I have nothing but praise for the time that I spent doing competition programming and learning what there is to learn and then realizing once you're just going to be grinding, it's time to stop. Yeah. Hacking and coding is all about learning and applying new skills. So repetition runs counter to what it's all about. Okay, so if you watch my stream two days ago where I did the adding the for loops to clang, that's another. You can get paid a lot of money if you can do this. Right. This is kind of what my Ethereum contract was, understand. So maybe there's a like high browse software engineering. Understand a complex system, modify the system to say, add a feature, ship to new system, meaning like test well and stuff. So there's a whole lot of software engineering kind of contracts or jobs that look like this. Understand a complex system, modify the system to add a feature, ship the new system. If you make yourself challenges like that Clang for loop challenge that I showed you, you'll get good at at this. You want to be a ML engineer? Well, here's the steps to get good at that. Download paper, Implement it, implement it. Keep doing this until you have skills. No test today. Slack mage. Thank you for subscribing. When you guys say, I saw a comment on one of the, on one of the recent YouTube videos where you say, oh, if only George would do would do lessons. What is it you want to learn? What is it that, that you want to. You want me to do? What is it that you think you can't do? You want how to get A girlfriend, right? And again, let's use the same, let's use the same, the same, the same paradigm, right? Input, system, output, right? So, well, what are you going to need as an input to the system? I mean, you want a girlfriend, you want women. A system, like, I don't know what you mean. First, can you come up with a definition? You know, let's come up with an exact definition of get a girlfriend, right? Oh, not Martin Scarlet's YouTube finance lessons, right? So first, define get a girlfriend. Now the system is, you know, the female mind. And let's say the output is a. Let's say, let's just say it's an imessage saying. So we are official now, right? So I mean, you can think about this, right? So, so I mean, we can use the same sort of tricks we used on the United flight attendant trying to get the change fees, right? Like, okay, one thing I've learned studying pickup is you don't get better at seducing women. You just get better at screening women. It's not about convincing women to sleep with you or date you. It's about finding the women who are already interested in sleeping, sleeping with you or dating you. So, right, same sort of approach. And you can treat this like a funnel, right? So let's talk about funnels. So this is a classic sales kind of thing. Let's say you want to sell cars, right? Same approach. Same approach. One, you have the top of the funnel, which is like getting the word out, right? You can do this with like advertising. Too many, you get some people who visit the dealership, right? And this is a funnel. Let's say 1,000 people see your ad. Let's say 10,000 people see your ad. 100 people visit the dealership. And let's say, you know, five by cars, right? So what's your ratio here? You have 1% here, and you have a 5% conversion ratio here, and you're falling off your funnel. So, you know, we can do the exact same thing here. One 100 message. This is sent to women. We can say maybe 20 replies. Let's, let's, let's let's say 30 replies of every one of those. We'll say we have five dates, two lays, one girlfriend, right? Same idea, right? It's just funnels. So this here is 30% ratio. This is a 1, 6, whatever that is in percent, like 18%. This is 40%. And this is 50%, right? Funnels, 1.2 K viewers. Let's go. No, you don't want a vim Lesson, right? That's like, oh, how do I make a product that make me a millionaire? Okay, all right, I'm getting money. Well, so the thing about getting money is there's only one way to get money. This is capitalism. Capitalism is based around consent. And this is kind of the beauty of capitalism. In capitalism, you have a buyer and you have a seller, and both parties must consent to the transaction. So if you want to make a million dollars, the only way to get money. Getting money in a capitalist system. System is one way to do it, and it's convinced others to give it to you. Right? You can't make money. You can't produce your own money. There is one way to get money, and that's to convince other people to give it to you. Yeah, welcome to NBA bullshit, right? So if you have two people and one guy's selling an apple, and it's $10, and the other guy's like, I'll give you $5 for the apple. The seller can now say, okay, I consent to your $5, but I don't consent. And there's no transfer transaction. Beauty. Beautiful, right? Getting money in a capital system. Convince others to give it to you. So to answer your question, how do you get make $1 million from product? Well, so you can kind of. All right, ready for some MBA bullshit? I'll give you some real NBA bullshit so you can think about who you want to target your product to. You can either make $1 from 1 million people, you can get $1 million from one person, and then there's a whole spectrum in the middle. Let's say $1,000 from 1,000 people, right? And you can think about basically how much effort you're willing to spend on the sale in each case. So here tons can be spent closing the deal. Here maybe one phone call, maybe, let's say a couple phone calls can be spent closing the deal. And this can be like, online only, right? So how do you make a million dollars from a product? Well, you can either convince a million people to give you one dollar, you can convince 1,000 people to give you thousand dollars, or you can convince one person to give you a million dollars. There's no other way to do it. Now. Make it look fancy, put a Lambo on the background, and post to Instagram as the new Warren Buffet. Yeah, I mean, but like, this stuff is so simple. I don't know how anybody's stupid enough to pay for MBAs that basically, like, here you go, there you go. There's basically everything you need to know, right? There's so much complexity and depth. I can sit here and type forever if you wanted me to go into, like, detail of how computers work, but nobody seems to care about stuff like that. I don't think that's what you guys care about out of a million subs on Instagram, right? All right, Right. So, yeah, look, do I know how to do shit like this? Yeah, sure, if I cared, right? So fundamentally, you have to. Again, Instagram is based around consent, right? So you have a follower and an influencer, right? It's the same basic market economics, except influencer. It's the same basic market economics, except it's not money, it's clout. It doesn't really matter. The same basic idea. So how do you, how do you get a million subs in Insta? Well, you know, convince 1 million people to subscribe to you. Now, there's a bunch of ways to do this, right? So one way to do it is like, with appealing content. Right now, there's lots of different things you can appeal based on. You can appeal to. You can appeal to novelty, you can appeal to beauty, you can appeal to sexuality, you can appeal to funny, right? So, I mean, that's kind of step one. And if you don't have appealing content, you're not going to get. Step two is be famous. So this actually works if you guys remember my. Or this is another way to do it. If you guys remember my thing yesterday. Famous works on a few different levels, right? So you can say there's, there's FOMO famous, right? Like, you want, you want people who fear missing out, right? Like, oh, man, I gotta follow this guy, right? You can use, so you can use fear basically to get people to follow you. There's like the car crash mentality, which is similar to fear of missing out, except, you know, maybe this is fear of missing something positive, Fear of missing something negative, right? So, you know, you can take like, like, like, maybe Kanye west is kind of a mix of both. Lindsay Lohan is very much in the car crash mentality. Yeah. And then there is like, you know, dark arts. All right, so you can, you can buy followers. These followers come from a few different places. There's cracked accounts, there's big Russian people who crack accounts. I think there's some new accounts you can do little psychological tricks like make your Insta private. If you make your Instagram private, you. Again, you're appealing to this. You know what's behind the curtain. Follow to find out. Bros, can you explain dynamic programming to you? So go on YouTube there's plenty of people who make great graphics for what dynamic programming is. The basic idea is say you have some function, you know, you have output F of X of Y and you can figure out how to define F of X of Y in terms of, say, you know, F of X minus 1, Y minus 1. Right. Dynamic programming is you build up like a table using the previous outputs to influence the next outputs. But again, you know, if you're looking for an exact topic like that, there's plenty of videos that'll do a better job explaining it than I can. I don't really know. And I'm going to read some comments now. And guys, if you could all. I will make it subscriber only if you guys are writing stupid crap. Is this you guys always say, oh, I want to learn, I want to learn, I want to learn. Like, what is it that you want to learn? Do some garbage collection and explain what Malloc is. Def con sold out. It was actually a really cool stream. Thank you. Here. So we have, we have a VIP what not to waste your time on. Okay, well, let's break that down using the same sort of mentality. So wasting time. So the girl last night on the date told me a very. She has a very interesting life philosophy. If you guys know the story about the fox and sour grapes. The fox is out walking, he sees a grapevine, and there's a big juicy bundle of grapes sitting up there. And he puts his paws up on the fence and he tries to reach it and he can't. He climbs the fence. He starts to climb the fence, but he cuts his paw. And then he's like, you know, those grapes were probably sour anyway, and he walks away. Right? So this is the traditional sour grape story. And usually people who have this, you know, those angry sort of people, but her take on it was when she cut her paw, it must have been a divine sign from God that the grapes were probably poisonous. It's like a positive spin to the sour grape story. So I don't know. You know, if you go through life, everything that you don't get, you just realize that you didn't want it anyway. And like, you really put a positive spin on that. This is, this is an impenetrable fortress of psychological. I don't know. But. But wasting time. I don't know. Right. So this gets into like, like existentialism, right? The basic idea of existentialism is you make your own meaning. Sorry, now you asked if this is all existentialism is. How did people write really long Books about it. And you'll have to ask those people. People. But. So I don't know. My personal philosophy is that there's no such thing as wasting time. Like, it's a stupid. A time is just wasting time is great. That's the point of time, to waste it. You know, maybe. Maybe I'll add a caveat to that. Don't fall in funnels. Right. If you are in other people's funnels, don't be in Skinner boxes. Skinner boxes, Right. Maybe a. Don't fall in funnels, don't be in Skinner boxes and don't be influenced by advertising. Can I explain algorithmic complexity? Sure. Again, let's see if there's good content on this already before I do it. What is algorithmic complexity? Algorithmic complexity is a measure of how long an algorithm would take to complete given an input of size n. If an algorithm has to scale, it should compute the result within a finite and practical time bound, even for large values of n. For this reason, complexity is calculated asymptotically as N approach approaches infinity. That's what big O notation is. Well, complexity is usually in terms of. Okay, so this is a way of thinking about it. How many operations does it take for an input of size N? Right? So you can take sorting algorithms. Sorting is usually a good way to think about complexity. So you can take like a bubble sort algorithm, right? Bubble sort in pseudocode is something along the lines of for I in 1 to n, for J in I to n. Or say I plus one to n. If j is less than I swap. So this algorithm is O of N squared. Because you see this loop start from zero. I'm not writing lua or nothing. So this loop is N. This loop you can be like, well, no, the average is actually n over two. But you know, n times n over two is, you know, it's still O of n squared, right? So this is an O of n squared sorting algorithm. So you can take other sorting algorithms, other sorting algorithms basically have this idea of we're going to build up a tree. So say I want to sort the list, you know, three, five, four, six, right? We take a tree, we put seven on the top range, right? And now we say three is three less than or equal to seven, it's less than. So we'll put it on this side of the tree. And now we do the same thing with 4. So we say, well, okay, let's take 4, 4 less than or equal to 7. It's less than 7, less than or equal to 3. It's it's greater than 3, so it goes here, right. 6. We have the same kind of. Well, this turned out to not work too well because the height of this tree is N. But let's say I added something like 10 in. You know, 10 would go here in the tree. I didn't make my tree too nice. Let's say I added a five in the five would go here in the tree. And the size of this tree, the height of this tree is log N. So that's all the sort algorithms that are in N log n time basically have some variant of this, right? So it's N inserts into a log N sized tree. So there are actually sorting algorithms that run in O of N. So we can think about. Let's say the list that I'm sorting is. Oh only is guaranteed to only be numbers between 1 and 100. So I can create a. Maybe this is something like hash sort or something. I don't know what the name of it is, but O of N sort. So I can create an array like count sub 100 for I in 0 to n count sub ARR sub I plus equals 1, right? And then I can loop through count and then I say for. Well, you know, I just like print. You want to print out basically the like. Like the number count number of times. So you see how this is O of N. I have one loop going to N and then I have one loop going to count, which can be viewed as like a K. And if K is small, but this obviously only works if your, your total number here is small because the K, if the biggest element in the array were to grow with the length of the array, then the sort would not be O of N. But assuming that it stays at 100, this sort is always O of N. But usually the smallest sort is M login. I think maybe that explains it better than this. No, we're not solving any leet codes. I'd like to see you improve an open source system that you'd like to see improve and follow along as you help progressing issues and features. People don't want to learn, they want to have your skills. Yeah, they can get fucked. Can I explain binary search time complexity? Sure. So binary search find an element in a sorted list. Something I say is hacking a lot of times is just binary search by hand. Right? So let's say you have an oracle or we'll just do it straight up. Right? So here's a sorted list, right? Let's say I want to find 67, right? So research 67. So you start out with defining a a low and a high. So we'll make the low the beginning of the array and the high the length of the array. So I now take a midpoint. Let's say I look at the array. Array submit is 22. Now I know that it can't possibly be any of these, right? So here's my pivot. I say it can't possibly be any of these because 67 is greater than 22. So we can now say low equals mid. So now I'm just looking at this array here. So that's the first run of binary search, right? So again, low zero, high is length of array mid. Low plus high equals two. Let's say mid is one. So we get 45. Low equals mid. Alright, so now, now we're doing it again for this and this time mid is 67. Run it through so we can return that. We can figure out what the index in the original array was. So notice that this took three steps, which is significantly less than here. N equals 7 because, because 7 is the length of the array. So binary search runs an O of log n time. Again, you can think about it similar to the tree where you're only walking down one path. Instead, let's say instead of this sorted list being put contiguously, it's a tree. How many jumps are you going to have to take in the tree? It's log N. What paper can implement beginner ML engineer? Well that and that's a deep. What paper can implement a beginning ML engineer? Oh, I don't know about that. Yeah, data science is largely bs. Anyone who's teaching you, if you're ever learning a like an object level skill, you're probably not learning the right thing. Object level skills will die out when the tools die out. But if you're learning kind of meta level skills, then they're never going to. It's like truth about the world, right? Are you learning something from nature or are you learning something from people? Right? If you're learning something from nature, good. If you're learning something from people, bad. Right? So you can, you can, you can break pretty much everything down into, into this paradigm. If you're learning like physics comes from nature, learning physics is good. If you're learning celebrity gossip, this is just arbitrary bullshit about people, it's bad. So one of the founding themes for the college. So we reject power over people and this will actually make us ignored by the system. A system cares deeply about power over people, embrace power over nature. So yeah, and then like Like, Like. Okay, so like, answering that question, what is data science? Right. If data science. If data science means, like, statistics. Statistics, sure. If data science means, like tooling used at company X. No, let's define F by. For complex S with a positive real part. How do I find all zeros of the function that looks oddly like the Riemann Hypothesis? Most of my math knowledge is in discrete math. I don't know that much calculus. I wouldn't really know how to do that. I mean, I could write a simulator for that. And then we can binary search for zeros. Assuming the function is continuous. We can't exactly binary search. Well, those things are going to get smaller, too. Complex S with positive real particles. Oh, God, I gotta do complex numbers too. I don't really know. We want to learn how to learn efficiently learning strategies. I don't know. Helm AI is four years behind. Come on. Some of these questions, like, what books do you recommend? How to stay hungry. Elon talks about this a lot, right? So you want to build a knowledge tree. You build a knowledge tree, and that means that whenever you learn new information fits into the tree tray. People who I've met who know a lot of things, it's not even always that they know, maybe interpolation is possible. Even if you don't know exactly what's a node of a tree, if you have some nodes nearby, you can kind of interpolate to it, right? Yeah. So it's kind of a framework for understanding. You can. You can root this knowledge tree. So Elon talks about it. Elon roots knowledge tree in physics. Iroot, knowledge tree in information. So my whole input, computation, output is this idea of information. This is a modernist perspective, postmodernist perspective, but both of those things are very, very good ways of understanding. So. So, I mean, how to explain anything. I mean, you know, some of these were like, how to get a girlfriend. Right. Input, system output. This is my core paradigm for understanding any. This is what. Richard Hamming was promoted a lot. That's like a Hamming tree guy. Coding theory was one of the classes I took that I couldn't do. It was too hard for me. I haven't seen. So Bezos. Bezos basically uses this one trick. Bezos operations mastermind. Right. So, for example, Amazon ships many packages. When Amazon negotiates with ups, Bezos has his guys draw up a Amazon shipping plan. Right? Plan is, let's say, sales pitch to ups. Ups, you know, gives real rate, right? Right. Bezos wins. I mean, that's, that's. That is the Amazon trick. Over and over and over again. Again. Oh, post structuralism. Oh. See, I like weird things like this. I really, I really want to meet more. I really want to meet more women and, and go on more dates. Like women who like, studied this stuff, you know, like, like, like I, I met this, I have this. Dated this one girl at Berkeley and she like, knew about all this stuff, man. It's, It's a whole world. It's a whole world and I try my best to understand, Understand it. I feel like I'll always kind of be a tourist there. Structuralism proposes that one may understand human culture by means of a structure modeled on language. This understanding differs from concrete reality and abstraction ideas instead as a third order that mediates between the two. Yeah. And then what's downstream of all this is the, is the, is the modern trash. Postmodernists were trolls, man. That's. I was, I was on the phone with my friend last night and I was like, you know what? I'm a post modernist. Like, I really, I embrace these, these, I hold these truths to be self evident. Uh, yeah, yeah, Bordj, Derrida, Foucault. And they're trolls, but they're trolls who've shaped the modern world. One thing that I wish I'd done differently when I was younger was actually, I didn't. I rejected almost outright the humanities as being any. Oh, they were trolls. People just took them seriously. I mean, they admitted as such, I rejected the humanities as any sort of, any sort of source of relevance in my life. When, as I've grown up, I've come to realize that the college and the singularity, the science doesn't matter. I think one of the greatest tragedies in history was that these incredibly intelligent scientists built a nuclear bomb and then handed that bomb off to politicians like they're subservient to them. What. Why do politicians or those kind of people have this power? So something that is changing throughout history is power over people. Used to be all, Used to be everything, right? If you, if you, you didn't. If you lived in a, a tribe, right? If you lived in a, like a hunter gatherer tribe, the, the only thing. What, what made you alpha chimp, what made you alpha chimp was, was having power over the other, the other chimps not having power over nature. You weren't a 10x better hunter there. There were no 10x better hunters. But power over people is waning and power over nature is waxing. Waxing is the right word. No, I don't want top 10 best waxing salons. What I got waxing like a moon. Yeah. Having. Increasing its apparent size. I dropped out of philosophy undergraduate because I realized how much you were just learning to tell logically sound stories to each other about how the world is. I wish. I wish there were a place where I could go to learn real humanities. It seems like being self taught in the humanities is more difficult. I found a few. You come upon a few texts in there and they're. And they're kind of mind blowing. Sartre. What is literature? Yeah. You should read this. I'm reading. That was kind of the first when I started to realize that. That there was a huge path to what is what is what is running a company. What is what is being. What is being CEO. It's just telling stories. All you're doing is telling stories. And we compete. The world is a big competition of these stories. What is the Wittgenstein verse? Tiring debate. I have so much to read. I don't know and I guess this is a lot of what I've been doing off stream, especially now is. Is just. It's just reading. And I'm not. There's no, there's no future in hacking. There's practically no future in programming. It's not a question of if the singularity is going to happen. It's a question of when and what will it look like. And it seems like what it will look like will be a lot less shaped by programmers doing the grunt work and a lot more shaped by who's telling the best stories. So, you know, if you want to save the world, better learn how to tell stories. Please teach me on becoming an expert in programming. So wait, hang on. What's the central kind of thesis of. By the way, everybody should read. Should read Turing's paper. Ramsey. Wow. It's all the same people, man. These people are all biologists want to be chemists, all chemists want to be physicists, all physicists want to be mathematicians, all mathematicians want to be philosophers and all philosophers want to be employed. Programmers are translators. Yeah. This is what most software engineers do all day, Go learn your stupid frameworks. Right. But again, that's my same thing about data science. Right? You know, tooling that will go out of vogue. No. Out of vogue. Maybe that's right, maybe that's wrong. Yeah. Memetic theory, these basic ideas. Peter Thiel, I read some new Peter Thiel thing. He really went hard on that. Do I think today we have less revolutionary, innovative minds in general than before? No. How many people can you name from the 20th century, like a hundred. There's definitely revolutionary people around today. I think that. Who's the most revolutionary person, I think, of the current millennium, Satoshi Nakamoto. If you really want to do things in some way, you got to do them anonymously. I mean, that. That's just. I don't know. Maybe that's not that true, but. Yeah. I mean, look. Look at. Look at what Elon Musk is doing right there. There are definitely people who are alive today who are going to be remembered in. I read the annotated Turing book. Yeah. Yeah. That's why I read the paper. I'll show for the book. Yeah. Yeah. I recommend that book. What is a good work ethic to have? So one test that I like to use often to ask a question, if somebody's saying something, say. Would anybody ever say the opposite? Right. So if you have a politician who stands up on stage and says, the children are our future. I care about the children. They're not really saying anything because nobody would ever say the opposite. So what is a good work ethic to have? I mean, it's a meaningless question because the. Like, you should slack off a lot and, you know, not really pay attention, do things sloppily. Yeah, Said nobody. Ever. Yeah. I'm chilling for myself, man. I'm just. Hoshi, what's from less wrong. I'm not saying you should do anything. It's just a framework. It's a framework for thinking. No, I see. I'm not. I'm really not interested in answering questions about myself. Why do you care? I'm. I'm nobody. Who are you? You're nobody, too. Oh, they're coming after me, man. Oh, no. Lock your doors, man. I don't know. Like, it's. Again, people are boring. Nature's interesting. Interesting. I'm a people, too. I'm pretty boring. How do you think about recursion? This is a stack. All right. Swat on my house. Oh, no. This guy's gonna call a swat on my house. No, he's not. He's banned. CRUD development will go away. Yeah, but it'll just become. CRUD development will become the most lowest common denominator. AI. You know, who's that guy in office space you like? Well, I take the thing from the engineers and I give it to the. I talk to the customer, and I give it to the engineers. Right. That's all it's going to be. Constructivist mathematics. Yeah, of course. So my understanding of constructivism says that you can't. The Law of the excluded middle doesn't apply. So you can, you can't just make an assumption, find a contradiction and say therefore the assumption is true. That's why I really like things like cock, which like, so they have a constructivist cock which says if you want to say something's true, you can't just say, you can't just like, yeah, find a contradiction. So I think that like finitism, I believe is true in a deep philosophical sense, whereas constructivism is just another lens. Where are the differences? What can you not. What can you say in non constructivist mathematics that you can't say in constructivist or just proofs? Harder. Taking the principle of the excluded middle from the mathematician would be the same, say as prescribing the telescope to an astronomer or the box or the use of his fists. Yeah, so. So yeah, construct. Is Google going to actually give me anything for this? I want non constructivist truths. You can't derive in constructivism. Oh God, yeah. Aaron Schwartz, take down J Store. J store Some like learning theory. I don't know about that. Yeah, What I want to know is like, like, yeah, what are the truths? Right. And then do they feel, I don't know. I mean, I like, like intuitionism as well. It's one thing to innovate in the computer science world, but it's another thing to utilize it to make money, which is where you need translators to deliver. So what is money? I mean, okay, it gets. I'm not going to go into this, but if your goal in life is to make money, all I will say is that is never for anybody a terminal goal. Making money cannot be a terminal goal. Right. Money. Money itself does not have any intrinsic value. Money only has. Always think about what your terminal values are. Right. Oh, this is. Let's find the slate star codec society shocks. Ah, yes. Bostra makes an offhanded reference a the possibility of a dictatorless dystopia, one that every single citizen, including the leadership, hates, but which nevertheless endures unconquered. It's easy enough to imagine such a state. Imagine a country with two rules. First, every person must spend eight hours a day giving themselves strong electric shocks. Second, if anyone fails to follow a rule, including this one, or speaks out against it, or fails to enforce it, all citizens must unite to kill that person. Suppose these rules were well enough established by tradition that everyone expected them to be enforced. So you shock yourself for eight hours a day because you know if you don't, everyone else will kill you because if they don't Everyone else will kill them, and so on. I heard the Russians used a similar tactic in order to prevent defection in World War II. You know, a soldier on the front line who's being told to charge a machine gun turret might want to run the other way. But, you know, it is the job of everybody behind them to shoot all the cowards. And if you do not shoot a cow coward, you're a coward yourself and you'll be shot by somebody else. Dictatorless dystopia. So think about what you really want money for. And are we really just in one of these dictatorless dystopias? Where's my viewer count? It's gone. And coward is whatever is defined by the Russian system. Right? Are we running out of innovative ideas? No. Oh, here we go. Easy example of a non constructivist proof without an obvious fix. Some digit occurs infinitely often in the decimal expansion of PI. Have we proven that? Oh, some digit occurs. Okay. Perhaps you think of the statement that every digit occurs infinitely often, which is probably open. Yeah, but some digit occurs in infinitely often. I mean. Yeah, okay, you're right. So the obvious proof there is assume for the sake of contradiction, that no digits occur infinitely often. Then given that there's a finite number of digits, by the pigeonhole principle, PI is not a transcendent number. But we know PI is a transcendent number, therefore contradiction. Qed. This is another classic one. Yeah, I mean, but I'm sure you can prove that the square root of two is irrational. I'm sure you can prove that in a constraint. Constructivist proof. Square root 2. Irrational. Direct proof of irrationality. Simple direct proof that I found as a teenager. Theorem did a conductor ideal. What? That's confusing. Okay. Theorem is that it's integral irrational. So, meaning it's an integer. If it's. If it's rational, the proof. Okay, so it's a reduced fraction. So the greatest common divisor is one that's using the extended Euclidean algorithm. Yeah, that works. Sure. Yeah. There you go. That's so much better. It's so much better than the other proof. Oh, Brower and can are examples. Let's say I think that all of higher education today is whatever the opposite of based is. You know, maybe cupped. That's kind of the problem. Like there's probably small numbers of holdouts in these tentaculous institutions. Rise of administration costs in colleges. Here, it's there. Put it this way. Administrative spending comprise just 26% of total educational spending by American colleges. Three decades later. Administrative wait. Huh? Where'd the Other spending go. I don't know, maybe this isn't true. I'm not. I'm not that interested in looking into the case. I mean, I witnessed this firsthand. I've heard from speaking. Speaking with a lot of people just how. What does this word mean? Yeah, that's. That's not what I was thinking of, but yeah, I'm not. I'm not. I'm not hopeful for the future of the higher education system. Curtis describes it as a state church. Where do I expect newer generations to learn things like lower level programming or theory? The same thing place everyone else did. What do you mean? People always say this. People always say this about computer science, man, the kids aren't really gonna learn how computers work. Why? Because our math majors don't know how arithmetic works? Like that's just a sign that someone was poorly educated. It's not gonna like fade from the world. People say this, but, you know, there's been so many other disciplines and I have no idea why they think computer computer science is going to be different. Look up the journal Constructive Approximation. There's so much to learn. Guys know what approximations and expansions are, by the way? Oh, I mean, unfortunately, it's a web archive, but can we look at how nice this web page is? Someone criticized my use of like looking at the form and not the content. But the form is an incredibly good predictor of content. A website like this contains truths because there's no other reason for someone to build it. Unfortunately, it's been taken offline. Let's see. A website like this probably contains way less truths, right? Compare the websites. So now we see what's wrong with colleges. I promise our university will have a website that looks much more like this and much less like this. Computer architecture A quantitative approach. Let's see what we got. College is a funnel. Society's a funnel, man. Why would it. Show me the. Show me a table of contents. Is the president of Stanford. I tried. Yeah. I click off the Kindle version. I think. I don't think I need to worry. Oh, take a look inside. Oh, cool. See what we got. Yeah, I mean, this is. Yeah, if you want to study computer architecture, that's like a decent. This is. I mean, this is from Google. Takes like 10Ns to get to L1 cache. 100Ns to get to L2 takes like a microsecond to get to memory. And it takes like a millisecond to get to disk or whatever those numbers are. You should know that. Understand all of that memory hierarchy. There's an old saying like there's two hard problems in computer science. It's like search and caching or something. That's the quote Cache invalidation and naming things. The only two hard problems in computer science. Cache invalidation, naming things and off by one errors. Cool. How do you get over a thousand people to watch you rant about how you. They shouldn't need to watch people like you rant? Oh no, we're just hanging out and like thinking, I'll go back to reading in a bit. I try my ideas out on Twitch, right? I tried Hammer out and decided there's never gonna be a hammer. Fuck hammer. I tried backspace out and it got a pretty good response on Twitch. The problems with backspace are more technical and actually comes down to more the question of what is programming? I don't want to make if you want to do programming for work. If you want to make CRUD apps like that's not really what I want to do in backspace. I mean, the problem I ran into with backspace is like, is our modern programming languages already the best way to describe what you want? How. Yeah, like, yeah. What is is programming really just describing? Is. Is there a more concise language? That's probably a better way to. Is. Does there exist a more concise programming language? AKA what is the next gc? Object oriented programming didn't deliver the improvements Garbage collection did not having to deal with freeing your memory. I mean, so my basic thesis is like compilers should do more search. GC is garbage collection. So if I write some. If I write some C code, let's make a function. So that's all well and good, except for this fact. See, look at A out's memory usage, right? Its memory usage is growing and we're going to have to kill it before it keeps using more. Because there's a memory leak. Because what happens in this program is every time A is called called, I'm mallocing six bytes of memory, I'm taking six bytes off the heap. And even though once this while loop comes around again, those bytes are never accessible. It doesn't matter. The program's already allocated them. So to fix this bug, I can do that. And now we'll see that A out does not increase memory usage. So a GC would automatically find unreachable pointers and free them. Yeah. So Rust has a separate idea. Rust has this idea of ownership, but makes Rust very, very hard to code. The beauty of gcs is that they operate completely in the background and it would make this a valid program to run. I could write this in go and go as a GC and it's fine if you could find the shortest program that outputs the data. Yeah, sure. I shill for the hotter prize a lot. Make it subscribers only. You never see my chat. I see your chat and you're a subscriber. All right, be right back. And then we'll do subscriber only. Q and and then going back to reading. What problems are we solving today? No real problems. Shining like a diamond. What you reading? I'm almost done with my. My book from yesterday. White noise. What is this Dud baiting? Twitch prime dud. Can we do a meetup before the singular? What happens if you pass string copy a character array without a null terminator? I mean it'll just keep copying to its ever on the stack till it finds a null terminator. Is it any indication that we're not close to an efficient to write language when we have functional programming that's so fundamentally different Languages are getting better. I'm very. Oh man, you know what I give to code in the language 20 years from now I'd be so happy. Cameo.com oh this is like I can book like famous people to come to my party. Okay. Personalized messages. Wow. Oh, I gotta think about my brand. You know how to solve constant growth of misinformation. I mean has that ratio actually gotten worse? Has the ratio of information to misinformation changed? I'm not sure it has. How does one be motivated? Bro, this is dumb questions. Like think of why that's a dumb question. What do you one one way to ask if if something's a dumb question or not is like think about what the structure of the answer you want is what kind of structural like like like what what does the space what does the range look like of the what does the output space look like? You are the kind of person who deserves Google because this one weird trick to stay motivated is what you're looking for there and that's the only answer you're going to get. And it's not a real answer. So your curiosity is never going to be satiated and you're going to keep googling that same crap over and over and over again. There's not an answer are there's less hierarchy and information distribution. I'm not even sure about that. So think about in some ways there's more hierarchy, right? Oh, that's all right. Let me see if I can find. I wish I had a Nice way to index this book. Better about families. Find the exact quote, they put it in the book a lot better than I'm going to say it. But families are held together based on untruths and, and commonly shared lies. I think that despite people talking a lot about misinformation in some ways, from my perspective, when people are screaming about fake news, what they're actually mad about when they keep talking about misinformation is actually that they don't control the narrative anymore. It's not that the new narratives are any less true than the old narratives. The old narratives are all lies as well. It's just there's a different kind of group providing the new narratives. The new narratives are coming from a different place. And I think there is slightly more true than the old narratives, but only slightly. Any advice to crack fang interviews? Honestly, if they sit you down and they're like, due to leetcode hard problems, this is why I'm not doing leetcode on stream for you guys. It's like, why waste your time memorizing that? They're. No, it's just insulting. I mean, I think it's insulting to everybody who's being interviewed. Like, they're not checking to see whether you're a good programmer or not. They're checking to see whether you're willing to grind for hours and hours and hours. Right. The leetcode grind is real. Like, is that, is that the kind of organization you want to go work at? An organization that values you just grinding on something stupid and irrelevant for hours. That said, I think that programming challenges are an incredible way to judge candidates. But not timed ones. Right. We have a. Like, or, or at least not time crunch in front of a person. We, we have a, we have a, we have a hackering challenge for, for comma. But we give you four hours and if you're good, you can do it in under an hour. It's, it's not designed to be, it's designed to be a freshman weeder test, not some insane. This is how we're truly going to judge you. The speed challenge. Yeah, like, I like the speed challenge. It's kind of, it's more open ended, but it has, it's, it's, it's more open ended but it has a clear, rigid evaluation criteria. It's an important, it's important to have interviews that are, have clear evaluation criteria to avoid bias. Don't take amphetamines. So my experience with amphetamines, Adderall stimulants is they do not make you better at things. They make you think you're better at things. They trigger this delusion. If I am good at something, I want to believe I am good at something. If I am bad at something, I want to believe I am bad at something. Nootropics are mostly a scam. Do you think humanities will ever be able to escape structures and hierarchies within nature that combat transhumanism? I'm not sure exactly what that means. LSD doesn't give deep thoughts. It makes you think. All your thoughts are deep. I have more positive things to say about LSD than I do about amphetamines. I think LSD shows you something. And I mean, like with all drugs, you know, understand what it has to teach you and then stop. I think I want to go to one of those ten day meditation retreats. But I don't know. I don't really want to talk about drugs. They're not drugs. Aren't that interesting. Josh, about the new God of transhumanism. What's the happiest memory? I don't really like talking about myself either. Most of these things here, I don't really talk about myself. I do give anecdotes for my wife. They're more to illustrate a point. Because anything that I say about myself is personal and useless to you. Don't be like me. It's dumb. You know, try to give you information so you can think. Don DeLillo. White noise. Some of the stories are made up. Yeah, doesn't matter. I haven't been drinking. Recess. Yeah, it's kind of nice. I never read Euclid's Elements. Right. Cool. It's about Pascal's Wager. Pascal's mugging? Yeah. It's a funny idea. And this is why I'm not a consequentialist. Many mangoes. That is kind of funny things. By oh, oh, you mean just things in general. Oh, I felt like things was like a. Yeah, of course. Web app frameworks like Electron. Why don't we ship the operating system in? Every Electron is just Docker man. It's the same idea. No, I don't know. Anything that makes deployment development easier is probably good. Stannis La Limb, can you think of any fun problems to tackle in your repos if you come up with one? No, because it's in local folder. I'm probably much more likely to lose it than the cloud. The is a tc. If you want a high tc, you gotta play the game. You know what? There is a Star Trek Voyager episode. Tc. Yes. Treatment Coefficient. Yes. Yes. Critical care. Yes. Just as the doctor asks what TC stands for. Oh, Total compensation. Oh, you mean the team blinds. Oh, my God. Oh, my God. It's Pascal's mugging man. Yeah, I'd like to go to Twente Persona. The problem is I have to like, schedule it months in advance. World Signia YT thank you for subscribing describing turning into a lizard. Man, I love Voyager growing up, you know. You want to know where I learned morality from? Star Trek. Voyager. So whatever Janeway believe there. I believe. All right, guys, thank you for watching the stream. If you enjoyed today's stream, I'm gonna look at my revenues and if I got a lot of revenue for this stream, we're gonna do more streams like this one. But I don't know, I'm probably not going to stream for a bit. It's the weekend. I got things to do that aren't streaming. Please subscribe if you haven't subscribed it. We like subscribers. Somebody asked me earlier, do I believe that there are some people are better than others? Yes, I believe that subscribers. Subscribers are better than non subscribers. Rex Doors, thank you for subscribing. Can you do an IRL stream? Should I like, rent out a. You know, like they're hyped on the. On the COVID Feels good, man. Subs rise up. Yeah, there's a class system. All twitchers aren't equal. There's subscribers and non subscribers. I don't. I don't tell lies. You know, I'm a flex. Thank you for gifting subs. Today's stream is top tier. I'm glad you liked it. Yeah, just get a mask or something. Where's my mask? That's right. That's right, boys. Today's flow is better. Thank you. What's ocsp? Oh, just some like, certification thing. Vic, Jack, thank you for subscribing. FaceTime is so funny. Subscriber open pilot full lidar. You take care to real port your subscriber. You're a good person. Yeah. None of any certifications. Don't. Don't take me over security itself is all. It's all racket. First we got to scare them about hackers. Then we got to sell them the anecdote to hackers. Antidote. Yeah, yeah. Thank you, Ranch. Leo, thank you for subscribing this. A take on morality of piracy. I don't believe in universal morality or ethics. Make up your own mind, man. Anybody who's trying to. But I do dislike anybody who tries to preach to you morality or ethics. She tries to tell you how you should act, man. Because this is just powerful people. Moral relativist. I have personal morals. I don't talk about. Like I said, I don't really like talking about my on these streams. I don't really think I talk about, like myself. Even this. This is not really a comment about myself. This is a comment about the content. The streams. Post more Instagram pics. Gula bulash. All right, well, you know what? If I find out about this whenever streaming again, you're banned. And you know what? People DM me all the time. My DMs are full of. Can you please unban me? Yo, dude. Can you unban me, sir? Yo, can you unban me, bro? You're banned for life. I don't think J was a moral relativist either. It's true. Can I ban you? I'm not going to ban you. You're a subscriber. Subscriber. Good person. Good. I'm glad. If Ro Jogan was like a mockery of the podcast, I would totally go on Road Jogan. Like a mockery of that whole format. Rojogen. Yo, is there really a Road Jogan? Let's check it out. Rojan Laser Cat boss, you're close to getting banned. You don't care if you're a subscriber, bro. Jogan, there's nothing wrong with the format. It has more to do with the reach is why I wouldn't go on Joe Rogan. I just imagine. Saris. Wasn't that the bad guy in Galaxy Quest? All right, guys. Thank you all for watching. Good afternoon. Go learn real things.