Sam Arbesman The World is Made of Code - Dr Brian Keating (1080p, h264, youtube).mp4

Code.

It's the closest thing humans have ever invented to magic. You write symbols on a screen and reality emerges. Money moves, doors unlock, planes land, diseases get diagnosed. And most of us have no idea how it actually works. Today's conversation is about why code feels magical and why that feeling is both powerful and also dangerous. In a moment, you'll hear why the real risk of code isn't artificial intelligence. It's scale. Tiny ideas multiplied across billions of of lives.

My guest is Sam Arbusman, complexity scientist and author of the Magic of Code, a book about how software quietly became the most influential force shaping modern civilization.

Magic and sorcery. Oftentimes in our stories, they require a great deal of effort and, like, training. And so, like, you have to go to Hogwarts for seven years to really know magic and wizardry very well. And the same kind of thing is with true with code.

So let's go cast some spells and learn the magic of code.

Let's go. Sam Armesban, welcome to the into the Impossible podcast, all the way from Cleveland, Ohio. Thanks for coming out.

Thank you. This is great.

We're here today to talk about this wonderful new book, the Magic of Code. It is a magical book, and like Arthur C. Clarke said, any sufficiently advanced technology is indistinguishable for magic. We'll get into that. We'll talk about the future of code. Is it prompting only? We'll talk about spreadsheets. We'll talk about the simulation hypothesis and whether or not the Turing Test has been passed or will it ever be passed, or is it always two years in the future? Sam, welcome.

Thank you very much. Great to be here.

And as many listeners and viewers know, I love to start by doing what you're not supposed to do, which is to judge a book by its cover. So take us through the title, the subtitle, and the COVID artwork. For those that might not be familiar with your work. Hey, book lovers, we're judging books by the covers. We know we're not supposed to do it, but I enter the impossible. There's nothing to it. Let's take a look and judge the box.

Yeah. So the title is the Magic of Code, and it's sort of. I guess it has kind of a double meaning. So there's like the Magic of Code, which is sort of like the wonders and weirdness and delight of computation. Because that's kind of the goal of the book really, is to sort of rekindle that sense of wonder that we've kind of lost in. I think, like right now when we have conversations about technology, it often feels a little bit broken, where we're just kind of constantly worried or we have this adversarial feeling towards it. And those things are fine and valid. But when I think back to my own childhood with computers, it was full of wonder and delight and excitement and Commodore VIC 20 and early Macintosh and SimCity and screensavers and all these things.

And so the book kind of looks at a lot of those different aspects alongside the fact of trying to also try to articulate the idea that coding, computation, it's not just a branch of engineering, it's also this kind of humanistic liberal art that when you think about it, it can also connect to language and philosophy and biology and art and how we think and all these different areas. And that kind of gets to certain aspects of the second meaning of the title, which is that one thing that we can kind of compare code to is aspects of like, sorcery and magic. Not in the sense of like magic, like technology just works, but more in the sense of like, we have had this desire for millennia in our stories and for being able to use text and language to coerce the world around us. And now I guess since the 75, 80 odd years since the modern digital computer, we now have that where you can actually write text and it can do things in the real world. And so I also try to explore that kind of meaning of like taking that seriously. What does that actually mean? So that's kind of the title, the subtitle is How Digital Language Created and Connects Our World and Shapes Our Future. And that one is really kind of going back to this sort of like all encompassing idea of that computation really connects to all these different kinds of things. But in addition to that, it also speaks to the fact that there's this deep history of computing and technology.

And in fact a lot of the aspects that I talk about, about how we think about simulation or certain things around artificial intelligence or connecting computers to biology and kind of thinking about how we can even model like evolution or artificial life. These things were actually present almost at the very inception of the computer. Like people were doing these things very, very early on. And. And so the book kind of looks at that historical thing. And I would say this is also kind of related to this idea that I actually think in the tech world especially there's kind of this, I guess, lack of historical knowledge of technology. I feel like there's kind of a certain amount of ignorance, almost like proud ignorance sometimes where it's like, oh, we don't care what came before us, we just want to kind of think about the new and, and sometimes ignoring what has come before you can be useful. But I do think actually being steeped in the path dependence and the history of technology can actually be very valuable to understand where we are and realize that a lot of the things that we might think are new, they actually have this, this long history.

And then. Yeah, the image is this kind of tree being turned into some sort of digital thing. It kind of speaks to the fact that. Right. That code and computation, as I kind of try to articulate it, is deeply connected to all these different areas. And there's kind of this porous boundary between the real world and the digital world. There's a whole bunch of like little, little. They look like little dots.

They're actually little zeros and ones. And so, yeah, kind of connecting to the binary nature of computers as well.

Yeah, I sort of saw, you know, projection thinking mathematically from code. But. But the tree to me, I mean, everyone can read into it what they want, but the tree to me sort of represent. Presented this, you know, transition from going right to left, you know, from paper, from scrolls, from parchment, from codex, which I never knew until I read this book. Codex, you know, was me. I knew what codexes were, the Lester Codex, but I never knew that's, you know, basically the formal name for what this is a book as opposed to a scroll, which is ironic. We're going to talk a lot about artificial intelligence. We can't avoid it.

Sure, yeah, but, but. And that harkens, of course, to the Turing test. But before we get to the Turing test and whether or not you think it's been superseded or when it will be, I joke, it's like nucle fusion that's two years away and always will be AGI and truly passing a substantive Turing test. But before we get there, I want to talk about a Turing machine. So a universal computer, which had an infinite. Was much more parchment like and scroll like than codex like, than actual computers are. I kind of see the modern computer, digital computer, at least as a codecs, you know, where it's. Can access in random order.

But a scroll, a parchment scroll like that, or the tree, et cetera, that's more, you know, kind of linear. You know, you can't random access. So walk us through, you know, just kind of the thumbnail sketch of the, of the history of computation and if you will, connect it to these statements that, you know, I've heard people say, eminent people, you know, a tree is a Computer if I try to hook up my, my tree, you know, and.

Save does not have a USB slot.

Exactly. My GPT Pro, you know, save 200 bucks a month, I won't be able to do it. So talk about what is the universal computer. What is a Turing machine? What are the kind of preliminaries that someone would need to kind of access before we can dive into the other topics like artificial intelligence.

Yeah. And so computation, Right. Is this very general property of I guess of matter where it's kind of the traditional kind of computer is it's manipulating information. And the idea behind kind of the Turing machine and kind of these like, sort of like Turing equivalents is that when a programming language or a machine can do kind of can manipulate information in a certain way where there can be branching depending on the specific situations, which allows you to kind of create if statements and loops and things like that, which we kind of know from like more traditional computing. Once you have a certain level of open endedness in terms of what you can describe, in this case describing an algorithm or a set of rules, you have this kind of universality. And Alan Turing, he was the first one to kind of develop this idea of this. And you mentioned the Turing machine is like this thing that doesn't have random access. It's a very theoretical construct.

Like it's not meant to actually be built, but someone's built. People have built these kinds of things where you imagine this little head that goes on kind of this infinite tape and then depending on whatever's on the tape, it kind of moves left or right or maybe moves, moves the tape instead of the machine. And then, and it has a certain set of rules. And the idea is that depending on the kind of rules that you have, it should be able to actually do any sort of computation that can be theoretically done. And then of course other people have actually developed other theories of what computing is. And they've all shown that they're all kind of essentially equivalent to the Turing machine. And so there's kind of once you reach this certain level, there is this, there is an equivalence between all these different things that they're all doing computing. Now of course, going back to what you're saying of like, oh maybe is a tree, is a tree, a computer is a cell computer, whatever these things are.

The modern computer has a certain architecture. So it's kind of so, so von Neumann, he kind of developed the sort of modern architecture that we have for computers. And I would say the vast majority of computers that we have kind of adhere to this kind of architecture, and it's sort of this traditional thing. And then, of course, on top of that, you have many, many layers of both hardware and software. And so our computer, like, you have our chips, and they have a certain set of rules. And so each chip can kind of do certain simple operations like addition, subtraction, things like that, and moving. Moving bits of memory. But then on top of that, you have operating systems that kind of allow you to abstract away all those different details, which is kind of this exciting thing that allows you not have to worry about all things that are kind of lower down, and you can almost not think about the hardware components of it.

Now the question becomes, yeah, are other things kind of like computers? And I think, yes, in a very kind of broad information processing sort of way. And actually, one of the things I discuss in the book is around biology. When you compare biology to computation, you can say, okay, we can map on some fun things in the same way that we describe computer programs with binary, like zeros and ones. Our DNA has four base pairs, and so instead of two, we have four, but it's kind of the same. And we have maybe like code within. Within strands of DNA, and they're being kind of compiled and then eventually run, and you have proteins, things like that. And. And I think there is something to be said for having those analogies and those kind of.

They're useful metaphors. But at the same time, though, we also have to recognize that biology is deeply different. Like, it's like you zoom down to the scale of a cell, and it's not just this nice thing where kind of information is being processed from, like, one spot to another spot and kind of moving forward, it's. It's really just. It's like this weird big mess, like things just kind of all bouncing against each other and. And you get information processing, but it's very probabilistic and stochastic. And so things. Things work, but they work in a very, very different sort of way.

So in the sense, yes, maybe a cell is a machine or a computer, but for me, I think. I actually think the opposite way of thinking about it is even. Is even more useful, which is that if biology or cells are computers, they are very, very different than the traditional computers like our laptops and things like that, which means that they're actually expanding the space of what computing can actually be. And so if we kind of view computing as simply this larger set of information processing, and people have this. There's a whole space of unconventional computing which is Looking at, I don't know how slime molds do certain things, they can actually solve optimization problems or whatever it is, you can actually use all these weird things to, to do to solve computer problems, but in very, very different ways. Which shows that computing, it can be kind of this much broader kind of space and kind of traditional computer science is almost like this, like one little area and maybe biology is just another area and there should be this weird high dimensional space that we actually should be exploring more. So I think that's actually really cool. But yeah, so a tree might be doing some computation.

If so it's very, very different. And yet you cannot, you can unplug your computer.

Don't try that.

Yeah.

Sam, at the end of the book there's a chapter called the Wisdom of Computation. And I want to talk about that because the famous saying, you know, that knowledge is knowing, that a tomato is a fruit, but wisdom is knowing, not to put it in a fruit salad. So I kind of use that metaphorically to kind of highlight the fact that what we care about is not intelligence, it's actually wisdom. Wisdom is scarce. Knowledge is abundant or drowning in knowledge. And there's almost, you know, a kind of a drought of wisdom. And quote, the final chapter is called the Wisdom of Computation. And you say, obviously it's an immense one.

And you talk about. There's a beautiful turn of phrase you talk about. You say, I recoil when I think, when thinking about how much what I've written in this book will be obsolete. Large software projects evolve over time to address the inherent transience, to address this inherent transience. But I think that wisdom in the realm of computation can only come when we consciously embrace the evanescence of software. So what does that mean? Evanescence? I don't know what that means.

Yeah, Evanescence is just kind of like the transient, like things that are, that are ephemeral. Yeah. And for me, one of the ways I think about this is, and just in a very simple way, like if you look at like a website from five, ten years ago, most. Well, first of all, the website might not even work anymore, but even if it does, the links on that website are probably all out of date. And so you can see there's this clear ephemerality of like you have things out there, but then they kind of fall apart. And I think actually someone did a study where they were looking at all the, the web addresses in Supreme Court decisions because they actually have a lot of them as their citations and most of them just don't exist anymore. And so you have like the Internet Archive trying to actually record these kinds of things and preserve them. But code also is inherently ephemeral.

Like you have this giant piece of code and then it might interact with certain other bits of computing where it might, maybe it relies on certain libraries, it might rely on certain hardware frameworks. And so you can't use really old computer programs on your new, on your new machine. You might not even be able to actually access it because you really can't use like an old, old floppy disk or things like that. And so for me, when I think about software and code, I think you really have to kind of lean into the fact that, yes, this code and computation, it's very, very powerful, but it's not going to last. I mean, sometimes it does last, and oftentimes we don't necessarily anticipate that. Where you have like weird legacy code systems or so things that people might have thought were only supposed to last for a little while and are still being used like decades and decades later.

About Unix and the power that leads to its ubiquity and longevity. Right, correct.

And I think, yeah, and, and I think those kinds of exceptions are super interesting. And I think kind of the, the nature of open source software, which like Linux and kind of things like that allow for that sort of maintenance and we can kind of talk about that kind of thing more. But by and large, software isn't necessarily going to last. And oftentimes people, if they work in a large software company and they're, they're writing their code, they know within a few years that it's going to be rewritten. And, and I think just kind of having that sense that you are making an impact and you're doing something, but it's not necessarily going to last is almost kind of a microcosm of the human condition, which is recognizing like we're on this planet for a fleeting amount of time. And so kind of recognize. And so for me, the nature of software and code really just brings that home in a way that many other fields and domains do not. Like, when you're, I know you're building a bridge, and that bridge will last hopefully many, many decades, sometimes longer than even than they anticipate.

But, but yeah, software is a very different kind of substance and I think we need, we need to recognize that.

So the namesake of this podcast, of course, is Sir Arthur C. Clarke, who said, among many things, the only way of knowing the limits of the possible is to go beyond them into the impossible. One of his many sayings but he also said that any sufficiently advanced technology is indistinguishable for magic. Not magic. Is the second word of this in this title of this book. Talk about software as Matt, like, are we supposed to interpret that literally as spell casting? As sort of this, this witchcraft like behavior that is almost bordering on the occult? I mean, some of my students, you know, I couldn't see, would fit that witchcraft and warlock kind of mentality. But, but what does that mean? What is the magic in the code? Is it the code or is it the coders? Where is it coming from?

Yeah, I mean, so I think, I mean, I think there's power in taking that metaphor seriously. I, you know, I, you don't necessarily want to take it too seriously because eventually, right, you kind of go down that weird path and the analogy sort of breaks down. That being said, I do think there is something to be said for realizing, right, that magic, if in our stories, right, we want to use text to kind of understand the world, then yeah, what. When we can look at the ways in which magic and code are kind of similar and different. So like, for example, one would be the fact that unlike magic, kind of the idea is just working magic and sorcery, oftentimes in our stories, they require a great deal of effort and like training. And so like, you have to go, you have to go to Hogwarts for seven years to really know magic and, and, and wizardry very well. And the same kind of thing is with true, with code, like to really kind of understand the world of the computer, it takes a lot of work. It takes a lot of effort.

And the same kind of thing of like, in the same way that you have like grimoires that are like repositories of spells and things like that, you also have those kinds of, like, you have books about like numerical recipes and all these kinds of things in, in the realm of computing. And so, and I talk about like, like the importance of names and how that's similar and different to like variables and things like that. And so I do think there is something to be said for looking at kind of how there is, there are these similarities. I actually had one conversation where someone was saying, like, oh, maybe you should just like bite the bullet and say these things are truly the same. And I'm not quite willing to go that far. That being said, I do think there is something to be said for realizing, right, it is this, it is the culmination of these desires that we've had for, for thousands of years that now we can actually do these kinds of things. And, but, but again I actually one of the interesting things is also going back to when you're saying that like the co. The coder as the wizard and I mentioned this idea that code is right.

It requires effort to really understand it. While that is true and oftentimes it was kind of the realm of like the, the wizard or whatever, like the, the, the trained person. There was also instances where there were spells designed for kind of the everyday persons. Like if you were, I don't know, a farmer and you lost your cattle, you could utter some spell and it would kind of help you find them apparently. And we are increasingly seeing this kind of, this trend towards democratizing software creation and actually allowing everyday people, whether it's like with a vibe, coding using AI or whatever it is, to actually build software for us, like for, for each of us. And so I think there's some interesting ways in looking at the similarities there.

Interesting. So a few years ago if you wanted to put someone down who is a liberal arts major on Twitter, you would say, you know, learn to code, bro. And I do feel like that has a value in it. And you mentioned that code maybe should be taught as part of the liberal arts educational component. Maybe we should be doing better here at UCSD and doing so. But, but now I almost feel like it's reputed to be true and I'm quoting facts from the Internet that, you know, one of the reasons that, that India has been so successful is that they basically skipped landlines altogether. You mentioned the book. You know, you have a landline, you had an H vac port, you hook up your central vacuum cleaner.

I never was that posh, but I did know, you know, the ultra wealthy in my neighborhood that did have that. I'm just kidding. But a cell phone. Yeah, actual plug in landline jack. I mean houses today still have that, right? Ethernet jack, same thing. But India kind of just in Africa largely, you know, just circumvented that went straight to wireless.

Sure.

Why they've been so successful. So I'm wondering, you know, if, if it's would be, you know, sort of similarly success inducing to go straight to prompting. Learn to prompt. The magic of prompt. Is that the next book on the Arbus memo kind of oeuvre?

Yeah. I'm not sure if it's just kind of the prompting, but I do think, I definitely think there is a huge space for knowing how to code still. And oftentimes using these tools when you already actually know how to program makes you that much more Successful. So it's not kind of an either or thing. That being said, being able to prompt and just generate software for yourself really does democratize this kind of thing and opens it up in a way that people would not otherwise be able to do. And so there's the novelist Robin Sloan. He has this great essay where he talks about how an app can be a home cooked meal. And I love this idea because it shows that software doesn't need to be this massive thing that's going to scale and be used by, by everyone.

Sometimes it can be and that can be very, very useful. But in the same way that when I cook, I'm not cooking for thousands and or hundreds of thousands of people, I'm just cooking for like my family. You should be able to build software just for yourself or for your loved ones. And, and I think we need more and more of that kind of thing because like, and, but right now, until recently, it's been very, very hard for someone who has a need or a desire to build software to kind of just do that if they're not trained in this. And so for me that democratizing effect I think is really, really powerful. I don't think it's just going to be just doing that instead of actually learning certain principles of software because oftentimes the kind of code that you generate for these kinds of things, and of course it could change in 10 minutes because AI is moving very rapidly. But these things are very good for individual use cases. They don't necessarily have the security and safety and scalability of real industrial, industrial strength enterprise software.

So yes, I still think you need like good software developers that kind of help build these kinds of like build actual software but for yourself I think that's unbelievable. And, and for me it like kind of goes back to like, yeah, this is, this is one of the dreams actually of a lot of like people in computer science for many, many years who wanted this kind of thing.

Yeah, not only for, you know, for parents, you know, I kind of felt guilty as a father, you know, because some nights I would be too tired to think of a story or even read a story. So I'd say to you know, chatgpt, yeah, I'd use my kids names and say, you know, use a story with this kid and that kid and they're this age and they get into, you know, troubles and they have to solve problems and the hero's journey and, and I would tell this great story and then you know, I'd go to bed and tell my wife and she'd be like, that's like, that's horrible. You're terrible. You know, that's just like child of you, you know, how could you do that? You're like using this computer to tell stuff like what is a book? You know, if I read like Lewis and Carol to them or I read, you know, one of my sons likes to read poetry. Every poet, I'm reading someone else's stuff. So it's now it's just the compilation of every human being who's ever lived. What's wrong with that? But the opportunity I think is, is for, you know, kids as young as, you know, eight, nine, ten year old kids nowadays. One of my daughters, she's learning how to do, you know, she learned how to prompt by accident because she wanted to use, I think it's Sona, which is the music generation software.

And she wanted to do a song, you know, kind of basically her name. But, but Dua Lipa is levitating. So she put in like some of the lyrics from levitating, she said, prompted in the style of Dua Lipa and it sounds like levitating and said, you know, came back, I'm sorry darling, I cannot do that. I cannot open the pod bay doors. You know, I can't do that. You know, it's like, can you think of another way to do it? So she did it and she came up and it sounded just like Dua Lipa in the end. And so she was learning how to prompt this feedback. And so I wonder to what extent is the code training us? Like are we evolving thanks to these incredible pace of innovation? Are we being trained by the AI for example, or the code itself, the lower level code base maybe.

Although to be honest, it's always been this way. The way in which we use technology has always been this kind of co evolutionary process where I mean, on the one hand I think I discussed this in the book of like you think about typing on a keyboard, that is not a natural instinct. Like it is not something that we evolved to do. And so on the one hand you can say, okay, we have kind of trained ourselves to actually fit ourselves to these technologies. On the other hand though, it's also just like we're learning a skill. And so I think depending on kind of how you look at it, of like learning a skill versus kind of like warping ourselves to some sort of technology, that kind of feels maybe unnatural. It depends on kind of the mindset that you're looking at. Yeah, I'm, I'm of two minds of this kind of thing, I think we should try to ideally make, make these, allow these technologies to be tools that make us the best versions of ourselves as as opposed to kind of having them kind of dictate the way in which we should be.

And oftentimes and people are talking about these kinds of certainly with like social media or big tech kind of things where you feel like you have to behave in certain ways in order for the algorithm to find you or to do certain things. So there's. So There's a screen TV show about 10 years ago called Halt and Catch Fire, which is. It's about the sort of the. The early personal computer industry. So it starts in like 1980, I think, goes to like the early to mid-90s. And in one of the first episodes, possibly even the first episode, one of the characters is talking about how the computer, like they're talking about computer, they're saying like the computer is not the thing, it's the thing that gets you to the thing. And I feel like too often we've forgotten we kind of make computers as sort of an end to themselves.

Like, oh, this is really cool technology, a really cool gadget or it's doing something really interesting.

You talk about the bicycle for the mind, right?

Yeah.

The point is not just having the bike.

Right? Right. The ultimate goal. Right. It's like to get us to something and we have to figure out. Right. And yeah, the whole bicycle for the mind thing. Right. Is really like saying like we still like by analogy with a bicycle, like in the same way that a bicycle is a much more energy efficient way to get around.

Computers can be bicycle for the mind, but we still have to figure out what we want to use them for. And as long as we are actually making those choices deliberately, then I think we're in good shape. If we feel like the choices are being made for us, then that is a little bit more problematic.

So all is not rosy in the realm of code and computation, from bugs to viruses and so forth. Let's talk about the kind of unintended consequences. Could a hyper intelligent alien have predicted that we'd have computer viruses, intentional, we'd have debug code that leads to Y2K errors and things like that. In other words, what sorts of lacuna of flaws and fatals. In some cases, things could have been anticipated, maybe eliminated. And this is going to eventually tie into the simulation hypothesis. Why would a simulator make such a mistake? But tell me, let's talk about the dark side, the bugs, which literally comes from actual bugs. And and things like that to intentional viruses and so forth.

What is the meaning, purpose and how do we react and evolve with that? Or are they destined to the ash bin of history like an RJ45 jack in your home or your H Vac system?

So my sense is that, yeah, bugs and failures and glitches like these are kind of the inevitable result of building sophisticated computer systems and technology in general, which is as we build more and more sophisticated systems, we make more and more powerful technologies, which is a good thing. But they come hand in hand with sort of a loss of understanding. Like when these things are built, they, they often accrete over time. So we're, we're adding more and more functions and functionality over time. Or if you're building software, you might build it on top of other things you don't fully understand. And that can be very powerful. But that alongside with like interconnection or just dealing with kind of the complexity of the world around us means that you end up with this very complex system that is built in a way that our brains are not really, well, like, well, well evolved to understand like we are. We are not geared for dealing with things that have millions and millions of lines of computer code or tens of millions of lines of computer code or whatever it is, or in all these different interacting parts.

And so as a result, there is a certain inevitability. That being said, there are many, many ways of reducing that kind of thing. And like there's. There are good software practices to kind of reduce that kind of thing and make it so we to don't want have too many. That being said, I like to take a more optimistic approach or a positive approach towards thinking about failure, which is, I mean, yes, we want to root out failure and we want to minimize it. And especially if it causes like some sort of catastrophic cascade, that's a bad thing. But in many situations, failure can also be a way of actually learning about a system because it often will highlight the gap between how we thought the system actually operated and how it actually does operate. And the glitches and the failures are kind of, they, they allow us to kind of bridge that, that, that divide.

And so, and not only that, they often. And we were talking about kind of computer, like computer science and code is kind of this like information stuff. It's also still like computers are deeply physical. So there was that. There was that senator a number of years ago who kind of talked about the Internet as a series of tubes and it was kind of vilified. Then there was Actually a book called Tubes, which is about the physical infrastructure of the Internet because it is a very physical kind of thing, and oftentimes we forget about that until we are confronted with a very physical error. So, like, one of the examples I give in the book is about there was some MRI machine in a hospital that when people came near to it with like, an Apple watch or an iPhone, all their. Their Apple devices all started failing.

But, like, Android devices were totally fine. And it turned out that there were a number of. There were switches that Apple used within these devices that were just the right size for helium atoms to get inside and mess them up. And it turned out the MRI machine had a helium leak. And so it was only affecting these specific types of devices. It was like this weird thing, but you would not have even thought about this kind of thing and like, the deep physicality of the computer unless this kind of error had actually happened. And so, and of course, there's many other examples. There's a.

There's a classic example of, like, someone being told in university that they could only send emails about 500 miles away. And, like, this is some weird thing. And it turned out it had to do with, like, there was some. Some older system being used. And then it, I think like the older system eventually, like, timed out, but it timed out over the course of like, some very small amount of time. But if you multiply that small amount of time times the speed of light, it was about 500 miles. And so, like, that was the one. Well, that was the kind of thing.

And so you see this kind of thing over and over. And so for me, I view perfection in computers as, like, as a goal, a very important goal. It is something that work may be asking, asymptotically approaching, but we're often only asymptotically approaching it by virtue of these kinds of bugs and glitches. And sometimes you can actually inject them into systems. So people have actually flipped bits randomly in supercomputers to see how robust they are. Netflix has a computer, a system called Chaos Monkey, where it actually periodically just take out of service various different subsystems, kind of make sure the system is robust as possible. So you can actually, in the same way that you inject error and failure and mutations into biological systems to learn about them. You can do the same kind of thing with computer systems as well.

Hey, everybody. I'm usually the one that asked my guests to judge their books by their covers, but today I'm asking myself to judge my own book by its cover. My newest Book focus, like a Nobel Prize winner, is chock full of advice, life tips and focus and productivity tips from nine of the world's greatest minds, Nobel laureates, ranging from economics to peace to physics. Of course, I will be checking out and my publisher's gotten Amazon to run a special, so go to Amazon and get the Kindle copy today.

Interesting. Yeah. Case Monkey. Title of a book by my friend and one of my very distant guests, Antonio Garcia Martinez, who was in this very office before I even had video, I think.

Oh, wow.

Half a decade ago or more and more than likely maybe eight or ten years ago when I first started up. So one of the questions that I always come back to is, where is the future? Where is the puck going to so he can skate towards it? And here, top research university. You're at my alma mater, my beloved alma mater, Case Western Reserve University, as an adjunct professor in the Weatherhead School of Management in the most abstractly designed university building, perhaps on campus, by Frank Ehry. Looks like an even more highly topologically complex version of the Disney Concert hall in la. If you've seen that. We'll have an image of what Weatherhead looks like. And I'm worried about, you know, what are we teaching our students? We're always fighting the previous war, you know, so we're teaching them Python now. You know, that's.

That's the hot language. Except, you know, almost nobody's using Python now because we're all doing, you know, react or doing other stuff. You know, JavaScript is making a resurgence and so forth. So. But, you know, of course, the ultimate evolution, perhaps the. The final evolution of computation is quantum computation. So, you know, I'm gearing up to build a quantum computer in the lab down the hall. We'll take a look at it after the.

After the podcast. But one thing we do, horrible job of at least me. I'm just speaking for myself. I don't know what my other fellow professors do or what you do at Case Western. Actually, Case Western has a quantum computing club in the physics department. I was just there last week and I got a tour along with my son. And, you know, he's thinking about college someday. And maybe his legacy may.

Maybe he'll get in easier. I don't know. I'll slip you as a $20 bill. But we saw one of my old professors who taught me numerical methods.

Oh, wow.

In. And we use Pascal in 1991. And unfortunately, he is such a pack, you know, Rat Cyrus Taylor used to be the dean there that he had the grade Book, you know, from my class from 1991.

Oh my God.

Preserved like next to his desk. Like he didn't know I was coming in. I just randomly popped in for, you know, a visit and saw his light was on. Went in with my son, he's like, oh yeah, here's what you got in 19. I was like, don't show. Finally, B, you know, case doesn't give fractional pluses and minus grades, so it's like all or nothing. That was my only B that year. But anyway, thank you, Cyrus.

You're a great professor. Taught me a lot about numerical methods. But you know, I'm thinking now we don't really teach, you know, how do you program a quantum computer? We teach it very abstractly. You know, it's quits and they can do all sorts of things. They, they can parallel process, they can perhaps decrypt using Shor's algorithm and do things much faster than a classical computer, maybe impossible for. But where's the base layer? What are you teaching your students? What are you teaching your kids? How do you actually apply and utilize kind of the expert knowledge that you have in order to maybe game the system in your favor and the students favor so that they're ahead of the curve, ahead of the puck, so to speak. Where, where are you, where do you see it going? And how do you employ these code.

In your daily life? Yeah, I mean, so first of all, kind of like caveat this with like, I'm not really sure I'm the best person at prediction. I remember when, when my first book came out now like 13 years ago or so, I was so excited that it came out in print because I was like, this is the last time books are going to be published, like published in print. They're going to be ebooks. And of course I was, I was wrong. And so I'm willing to, to, to caveat that, that being said, I mean for me, and yes, like how we program has always changed and I think this is kind of one of these things, like, I like you look at like, like very early machines. It was this very physical process of like switches or cables. And of course then it moved to like machine code with binary and then eventually to like assembly language, which was kind of like a little bit fancy, like more human readable version of, of machine code. And now we have these kind of higher level languages and now it's going to change yet again, perhaps with, with a generative AI and prompting these systems, that being said, they're all kind of on this continuum with each other of just like they are all part of the process of being able to kind of instantiate ideas that you have in your mind and actually have a computer do these kinds of things.

And so for me, I certainly think learning a certain amount of coding, even if of course it's going to be obsolete, it's like the languages I learned when I was in college or high school, like these, I don't use them anymore. That being said, the basic principles of how they work are still pretty standard. There's only a few different types, categories of programming languages that people use, and those are pretty standard. And so you kind of learn one and then through a combination of just looking things up online and kind of doing your best, you can pretty much muddle through all of them. But I would say for me, the more important things are just the fundamental features of computation that are almost upstream from, from the actual programming. Like ideas around going back to Turing machines of kind of certain things around equivalence or kind of theories of computation or kind of understanding certain basic ideas. And for me, those feel both really, they're both very, very interesting and I love them. But I actually think.

But they're also much more kind of like perennial. They're not going to be obsolete. Now in terms of how I code things nowadays, I would say my language, even though, yeah, you were saying that Python is kind of more, maybe more obsolete. That's still my language of choice. I still use that and kind of similar kinds of things. And, and that being said, like, using it in conjunction with some of these coding tools can be very useful. There was, this was not in Python, it was a different language. But I was, I had an idea several months ago for a computer game I wanted to make for, like, for myself and my kids.

And I started programming myself and I was, I'm just going to prompt and kind of see what I can get. I got most of the way there and it was amazing in the sense that I quickly realized the game I made, it was actually not that fun. And so I was able to save a lot of time, be like, okay, this is something I want to do. Yeah. And so, and so I think that kind of thing is actually really, really useful. But in terms of like how to teach, I would say, I mean, I think there's something to be said for like just getting your hands dirty even if things are going to be obsolete. And I actually, in the, the software development world there, you're my senses, you're constantly on a treadmill of just like learning some new framework or learning some new language, and then of course, that one maybe falls by the wayside, then you learn something else. And the AI tools that we have lower the barrier to learning each new thing.

So I think being able to do that, kind of constantly learning, but learning in conjunction with these new tools is actually a really powerful combination.

Yeah. So before we get to educational applications of code, you bring up the humble spreadsheet. The spreadsheet is sort of ubiquitous of many different forms from hundreds of years ago, some claim even, you know, thousands of years ago, going back to pharaonic times in the Middle east and kind of its predecessor, but still today. And of course the joke is that more fiction is written in Microsoft Excel than in Microsoft Word. Talk about the humble spreadsheet and how it's evolved and what this future may be. I mean, the Linde effect suggests that it'll be around forever, right?

So, yeah, I mean, the spreadsheet as a piece of software, it's based on this physical thing, these like giant spreadsheets that were used in the accounting world. And the two creators of the, of the spreadsheet and the first spreadsheet software is called VisiCalc, they realized they wanted to kind of take this very physical thing where you have lots of different calculations and each calculation on different cell kind of leads into another and embody it in code. And, and it was. And well, first of all, it was like the first killer app for, for the personal computer. There was kind of this, I think, the, like the personal computers. And there were personal computers beginning in like the 1970s, 1977, I think was when there were like three very popular personal computers first came out, including the Apple ii. But there was a several year period where people bought them, but they weren't quite sure exactly what to use them for. Like there were games or other kinds of things.

And then the spreadsheet came out and people immediately saw the need for personal computers. And I was, oh, wow, this is the thing that we want to use. We want to put them in offices. We now recognize it. The interesting thing about spreadsheets though is that in addition to being very, very useful for lots of different things, it can, well one, it can actually like help you model the world, sometimes fictionally. And of course remodel is a massive simplification of the world around us. But they are also programming of a sort, and so many people. And so I was going back to talking about like democratizing software and kind of building software and things like that.

If you have used a spreadsheet for Anything more than just like, entering text or numbers, but actually built so some formulas or whatever you are programming in a rudimentary way. And it is also incredibly visually appealing. I mean, it could cause many unanticipated consequences, but you can kind of see how the numbers flow through and kind of do things. And the spreadsheet, in many ways, is the most popular way that people actually program. And many people who don't even realize they're programming, they are actually programming in spreadsheets. And. Yeah. And so, yeah, maybe the kind of thing it will always be around, but.

But, yeah, spreadsheets have been. Right. They're enormously valuable. And like. And of course, it's gotten to the point where we no longer talk about, like, spreadsheets as electronic spreadsheets. They're not spreadsheets because we've kind of forgotten about the. The actual physical version. But you can build so many different things in them and.

And you can kind of. You can live in them. And I think many people like to live in them for their jobs and. Yeah, and it's this. This kind of wild piece of software that allows you to kind of examine the guts of some sort of calculation and kind of build all these different things. And so, yeah, as much as generative AI has been allowing people to kind of build software for themselves, we've been doing it for many, many years since. Since Visit Calc, possibly without realizing it.

Right. And all the way up through today with, you know, AI tools in each cell.

Right. Yeah. And now. Right now it's. Yeah, it's kind of like we've overclocked the spreadsheet as well.

You can, like, insert a power. PowerPoint into a cell of an Excel spreadsheet. It's incredible. Let's talk about artificial intelligence in just a second. But before we get there, kind of maybe on the opposite end, the nuts and bolts of the Internet and a lot of code, whether you know it or not, is based upon unix, which, as you point out, was a proprietary thing that only universities had access to. I love the call out in here whenever I hear the. The cosmic microwave background referred to, which is the way I butter the bread around the Keating household, as you know, I get excited. And of course, BSD and UNIX were both invented about labs, and that's where the cosmic microwave background was discovered and many other things.

The transistor, the laser, the maser, all sorts of other things. But let's talk about that. Why has it been, again, this thing that's going to outlive civilizations and many countries have not lasted as long as Unix has. It's entering the back half, the backlap of its first century. Talk about that. What is the secret sauce, the ghost in that machine behind why it's been so successful? And what do you see as the future of just UNIX and its implications?

Yeah, no, Unix, right. It has this unbelievable story. It started at Bell Labs and eventually was kind of like rewritten number a number of different times and kind of modified. And I think there's a number of different reasons that it has such lasting power. But in one of it is kind of like you have these little primitives that can kind of be recombined and so you can really kind of make it your own. And it has this a great deal of open endedness. But in addition to it, I would say if you look at like the story of Unix and its variants like BSD and Linux and things like that, it didn't necessarily start as a great piece of software. I think that the guts were like the kind of the framework were there from the very beginning, but it eventually became a really good piece of software.

And I think that is one of the features of kind of a true lasting piece of software which is, and I talk about this in the book, when you look at open source communities and open source software or things like Unix that have had sort of this long evolutionary history that it's, it almost feels like it's a sort of like textual tradition. Like in the same way that you look at like ancient ancient mythological tales, that ancient mythological tales almost had a sort of open source kind of community around them where you might have, you have a, like, you have like the Greek gods and you have their relationships and then it's like certain people would add certain stories and then maybe those would get passed on or people would modify them and eventually the, the best ones kind of had lasting power. And then eventually you had this kind of really nice body of Greek mythology, ancient, ancient mythological tales that then were preserved by this community. And I feel like whether you're looking at Unix or other kinds of long lived software, and certainly Unix is probably one of the most long lived ones, it has these communities that kind of shape it and modify it until it becomes this really, really amazing thing. And so I think that kind of evolutionary process of modifying it, changing it, making it better, alongside this very community oriented aspect of like having people who actually have a stake in making sure this is the best version of something in the same way that you want to make sure these are the best Mythological tales. That's kind of part of the secret sauce of like making it actually so lasting. Like, what is the future? I, I think, I mean, yeah, given kind of. You're talking about like the Lindy effect.

Like, I think, like, given its lasting power, it's probably going to keep on being in everything. Like, you find it like in, in refrigerators. You find it in like weird. Like, like, I think like undersea vessels and stuff. Like it's everywhere. And given the fact that it is both cheap or in this case free and really well understood and, and it also kind of has the, the merit of being pushed and pulled and prodded so many times into, into, I wouldn't say perfection, but into a great deal of robustness is resilient. Right? It's the kind of. Right.

That's the other thing. Like, it's very, very resilient in a way that sometimes fancier things are not. And so, yeah, I feel like that's going to be. I have a good feeling about its future.

The Gilgamesh of its. Of the next half millennium. Exactly. Let's talk about artificial intelligence. And then I do want to segue into both the simulation hypothesis. We've had Rizwan Burke on not too long ago. Of course, I'm an expert in things, all things matrix and simulation. But before we get there, I want to talk about our first artificial intelligence.

And you say tools for thought. What is artificial intelligence? AI has already lurked on the edges of many topics that you've explored for years, but it's time to bring it out in the shadows from a tool or add on to a force in its own right. Simply put, artificial intelligence, a loose bundle of techniques and approaches that enable computers to do things that might be considered to involve thinking or resemble human intelligence in a range of areas. Or more cynically, it's anything that computers can't do, do yet. So that kind of brings up, you know, as I like to say, you know, aping the, the canard about fusion. It's two years AGI. What would it take to convince you that we've passed the Turing to a useful turn test? I mean, would argue we've already passed it, but, but in terms of something that could do something truly novel, humans have never been done before. You know, I've talked to a lot of people, people don't like when I name drop, but I talked to Terry Tao and many other, and he says, you know, it's not doing stuff that's truly useful yet.

But he's, you know, everyone's convinced it will. So tell me, what is your, what is your philosophy about AGI and artificial superintelligence? Maybe, and what criteria you would see personally? Arbusman criteria. The Arborsman test, Arbusman, you know, exam. What would you take? What would it take for you to be convinced it's here?

Yeah, I mean, certainly going back to like the Turing Test, I feel like in kind of like a low level version of that, like we definitely have passed that. And I even mentioned the book. I'm like, I want, I wanted there to be like a ticker T tape parade for that moment. We just kind of blue path it and never really grappled with this moment of like, we are generating like really impressive conversational abilities. Now in terms of those other kinds of things, like generating true novelty or scientific creativity or those kinds of things, I feel like we're not quite there yet. I think right now the AI tools that we have are much more kind of like in partnership with, with, with people. And, and for me, I mean, I actually like that kind of thing. I like the sort of the human machine partnership where we're kind of working together.

I'm not entirely in a rush to become obsolete. And so that's okay if we have that for a little while. And so like, yeah, they're kind of like human machine partnership and kind of like augmenting our abilities I think is very useful now in terms of like what I would need to see. Yeah, maybe some really novel creative jump, like potentially some like technological advance that was actually discovered by. By AI or some like really scientific advance. So really impressive scientific advance. I think there's a number of people who I needs to win some sort of like scientific prize or whatever it is, potentially. Yeah, people have actually, I think there are companies that are kind of like aiming towards that kind of thing.

And so, yeah, I'm not sure it needs to be that. But I, but I do wonder. I mean, going back to the way I was kind of describing it, like kind of the cynical version of like, what is AI? It's like, oh, it's the things that we have like, that we haven't quite solved yet. Because you see this especially earlier on in the world of AI, it was like anytime an advance was made, it stopped being considered AI. It was just like, oh, now it's just like another feature of what computers can kind of can do. And you're like, oh, it can't quite do these other things. And, and so, yeah, I definitely think we're not quite there yet. I wonder if the current frameworks and structure of large language models are going to get us to that kind of thing, there's a number of people who've talked about whether or not we're plateauing in their abilities.

And so I think we still have potentially a number of different advances that we need. For me, though, one of the other things I also think about is I don't want to focus as much on whether or not computers are able to do the things that humans can do. Although that is interesting. I'm much more focused going back to the human stuff. And the wisdom is around just what is our quintessential humanity, what are the things that I want to keep on doing? And so how can I use computers to make those kinds of things better? And so you talking about telling stories to your children, if they allow you to spend more time in a way that, that you find meaningful, then that can be a good thing, right? If you feel like, oh, maybe I should be spending more time making up the stories, that's a whole separate thing. But there's always. We're always going to feel like we are kind of getting to where we're going to say, okay, here are the things that humans can't do. And then of course, AI 30 minutes later can do a whole bunch of those different kinds of things.

And that's okay. It's always going to be a moving target. I do think, though, I do wonder actually if when some, like an advance, like an actual scientific discovery is made, some people will be like, oh, that's actually kind of trivial. But I, I feel like that would be a. Probably pretty good.

Yeah, I've proposed that too. The Keating test. Well, the Keating test, there's the strong Keating test and the weak Keating test. The weak Keating test is, yeah, generate new scientific knowledge based purely on empiricism. You know, some, some fact that we plug in and it explains it or successfully even retrodict a known theory that considered novel for it. Like the perihelion advance of Mercury wasn't well understood. And it took this novel insight, this brilliant brilliancy of Einstein to suggest it's due to spacetime curvature, not some unseen action at distance. And that even when we plug in perihelion data for Mercury, we've studied this with my student Evan Watson here.

The data going back a thousand years, it still can't predict unless you clue John, and put on like basically like a spreadsheet, you know, basically add a spreadsheet onto it.

There have been situations and I've seen This where, like where you can, yeah, you give data from a scientific experiment and it kind of develops some sort of equation to explain these kinds of things. And sometimes it can be kind of a messy equation, sometimes it can be quite novel and non trivial for the scientists to actually understand the thing that the system like popped out. And so in some ways we might already be there. And that's not even using a lot of these like AI advances, just using like evolutionary computation or things like that. So it's always going to be a moving target. It's going to be probably kind of fuzzy as to where that boundary is. And I think I'm okay with that.

Me too. So the front cover, you know, you mentioned the, the tree and stuff, but the kind of dividing line or the plane rather between it, it sort of looks like a matrix to me. Which you know, brings up the inevitable, the conversation about the matrix and, and the simulation hypothesis in general. What, what are some of the attitudes, modern day attitudes towards it? As I said, Rizwan Vir has rewritten his book the Simulation Hypothesis with an Update in the second edition. Basically going from 50, 50 chance that we'll get to that we're simulated to greater than 50, 50 chance that we are simulated creatures. And that brings up a whole universe of other questions and interesting topics. We could go down, we'll run out of time before we do so. But talk about that.

The attitude towards simulation is this always been there, this kind of ghost in the machine and you know, Frankenstein and the anxiety and the nervousness that that humans have always had about being superseded by technology kind of reaching its ultimate, you know, kind of admonition for humanity which is that we may be simulated entities. So what are your attitudes towards that? And describe how in the book you, you really comment on this as this sort of like a recurring theme and throughout humanity's relationship with technology.

Yeah, I mean I, I, I'm pretty, I guess agnostic as the likelihood of the simulation hypothesis. I, for me, I kind of view this as like one instance of almost this like cry for mythos or, or meaning like there's, there's a number of ideas that we have in science and technology. Whether it's like ideas around the singularity or like things like superintelligence or like trying to understand like ideas around like the Fermi paradox, like Fermi paradox and I feel like simulation hypothesis kind of is in sort of that same kind of category of like, yeah, this very interesting grand idea that kind of fits in this space where the tech world might need Kind of like a larger kind of grander narrative. This almost kind of like mythological or like mythos kind of tale. But for me, I'm less concerned with the meaning. Like, the meaning of its grand importance, like, whether or not we are truly in a simulation. I didn't. Could be interesting in the absence of, like, very clear evidence.

I think it's like a fun thought experiment. And so for me, I find it much more interesting as a window into a whole bunch of different ways of thinking about computing we might not otherwise think about. So whether or not it's like the deep physicality of computing I mentioned before of, like, the helium atoms doing weird things with iPhones. But, like, of course, I mean, you can think about computation in many, many different ways and kind of understanding the true. Like, like the. The limits of computing of, like, how much computing computational power can you actually have? And I don't know, like, one. One, like one cubic foot or whatever, a kilogram or whatever it is, or thinking about glitches and failures and how those can be windows into better understanding or how do you actually inject code that's unexpected? And, and, and for me, like, so there's this great example where I think it's. Was it Super Mario World? I don't remember.

I think it may be from, like, the Super Nintendo, where people figure it out that if you kind of do certain moves within the game and, like, jump in a certain way, you can actually inject code into the game and suddenly you're no longer playing, like, Mario Super Mario World. You're now playing like, Flappy Bird or some other kind of thing. And, and so for me, like, that ability to almost, like, play inside a computational world and then modify it, that's the kind of thing that I find much more interesting. Like, oh, like, then maybe that has implications for how we think about reality. Whether or not the simulation hypothesis is real. I just, I love using it as less as a question of cosmic importance and more as a window into thinking about all these weird kind of like, little edge cases and things like that. And so there was a novel that came out, I think, maybe like the past six months called When We Were Real. And in the book, I think like, seven years before the book takes place, the world has been revealed to be a simulation.

And so then in the book, it's a bunch of people going on a. On a bus tour across the United States just visiting all the weird glitches in reality as, like, these are just kind of like roadside attractions. And. But you also then see, like, how people would come to grips with this and how they would think about this and, like, thinking about computing in reality. And. And for me, like, that kind of. That kind of approach, or even just like, the approach around, like, code injection and unexpected consequences or thinking about the limits of physics or the limits of mathematics and what that might mean for how we think about these kinds of things. That feels much more exciting than, like, saying, like, oh, my God, like, what does this mean? And, like, how does this kind of change my own version, like, vision of myself.

Sounds fun.

Yeah.

I talked with Rizwan when he was here about, you know, my favorite thing was the Easter egg in Atari 2600's adventure. So there's a game called Adventure, not to come with the Atari 2600. The original one, I, like you had a Commodore product for my first computer, but. But the. The first game console, of course, 2600, and there's a special room you go into, and there's a little pixel, and you pick it up and then you transport it throughout the kingdom and you get into this other one, left, left, right, right, up, down, sideways, whatever, and you place it in a place, and it starts flashing rainbow colors and it says, created by Warren Robinette. And one of the fun things that I pointed out is that Warren Robinette went on to develop a lot of virtual reality stuff and study it. You know, University of North Carolina at Chapel Hill, where there was this famous Gravity conference that Ed Whitten's father was a. Was one of the organizers of.

Feynman was there. All these greats were there, and they were working on, you know, quantum gravity and string theory and alternate realities and. And, you know, just kind of wondering, you know, maybe if that fell through in the wormhole and. Oh, interesting.

Yeah.

Warren picked that up as part of his work. He later went out to work at NASA Ames, which is. Oh, that's so interesting. Yeah.

Yeah. But it's.

Right.

Yeah. Thinking about. Yeah, like, Easter eggs. Yeah. Easter eggs in reality. What does that mean? How. Like, how can we think about that? Like. Yeah, that kind of stuff is.

Yeah, it's just a lot of fun to. It's very provocative, I feel like.

So, you know, I love the book the Magic of Code, and can't wait to see the magic of prompt when it comes out. I wanted to talk about, I think. Was it your first book, the Half Life of Facts, or was that your second book?

The Half Life Facts was my first one.

First one. Okay. So I haven't read it. Disclaimer. So, you know, unlike this one, which I, which I loved. And I actually converted this to an audiobook. I only do audiobooks thanks to Speechify, which is not a sponsor, but I'm trying to get them as a sponsor. But they take this.

They have this great AI software which you can clone your own voice. So I can have it read my voice or you can have Snoop Dogg read it to you, which I've done in the past. I have Snoop Dogg, Gwyneth Paltrow, you know, Kim Kardashian, you can do whatever you want. But I did convert it to my voice. So I listened to this book as well as read it, but I didn't read the magic. And that's fine. You know, sometimes I don't like to think of myself as, you know, as a podcast host, you know, it's kind of a short form, you know, which is a sponsor. Thank you.

Short form. Not sponsoring this video, but other videos, you know, kind of a book summary, one page summary, audio summary. Because, you know, it's nice to encounter it the way the audience will, which is they get to read it for the first time. So I did hear about it from our mutual friend or contact Jesse Michaels, who runs a wonderful podcast called American Alchemies. A little bit out there, Jesse, you know, call me brother, you know, before you start ragging ranking on physics too negatively as you did with Chris Williamson. But you mentioned the Half Life of Facts on Chris Williamson show. So I thought, you know, let's introduce my audience to it. You know, why like Chris and Jesse have all the fun.

So talk about that book. What is the core thesis of that book for someone who hasn't read it, I. E. Me.

Yeah, yeah. So yes, the basic idea of the Half Life of Facts is, I mean, obviously what we know now has changed over time. And so like things that you might have learned when you're young or in your things in your textbooks, they change over time. Because I mean science, science changes. We learn new things. This is not surprising. I mean like when I was, and I guess when I was young, I already kind of began what dinosaurs were. Kind of changed.

But it used to, used to be we thought they were like, they were kind of like slow gray, green reptilian monsters. And now we think that they're like these almost like warm blooded feathered chickens. They're kind of much more colorful. It's a very different kind of thing. And so knowledge has always been changing. And actually it's my, my grandfather, he was, he was a dentist and when he was in, when he Was in dental school, he actually learned the wrong number of human chromosomes in a cell. There was a. He learned 48 instead of 46 because there was this period of several decades where I guess imaging techniques were good enough to count, but not necessarily count correctly.

And so in textbooks, they like, for 20, 20, 30 years or so, they learned the wrong number. And you can see this example after example of things we thought were true or are no longer true. And so the book looks at the ways in which kind of the regularities in how what we know changes, how knowledge grows over time, how errors appear, how they're rooted out, how things become obsolete. Kind of like what are the patterns? And so, and it's kind of by analogy with like the half life of radioactivity, where. And you. Can you give me an atom of radioactive, radioactive isotope? And I can't really tell you when that thing is going to, to decay. It might decay in the next fraction of a second, might take a very, very long time. But things change when you give me a whole chunk and then you can actually statistically understand, you can actually chart out the half life.

You have this kind of very clear decaying relationship. And the same kind of thing is true with knowledge where I can't necessarily tell you when some new discovery is going to occur or some fact is going to be overturned. But overall, there is this clear shape to how what we know changes over time and how we select slowly but surely, kind of like asymptotically approach what I hope is the truth. And it also kind of goes to this idea that ultimately science is not just a body of knowledge. It is a rigorous means of querying the world. And science is necessarily in draft form at all times, and we're learning new things. Actually, I remember a professor of mine from grad school, he told me this story where he. He went, he went, he went to lecture and gave some lecture on it Tuesday, taught about some topic.

The very next day, he read a paper that actually invalidated everything he taught. So he came in like the following day, like on Thursday into class, and he said, remember what I taught you, it's wrong, and if that bothers you, you need to get out of science. And so being able to kind of actually embrace this idea that everything is tentative, and of course we are getting closer and closer to truly understanding the world. But there is this sense that, right, things are in draft form and we're constantly learning new things. That is a very scientific mindset. And I think by and large, we need this whether or not we Are scientists or not to really just kind of actually have a handle on all the knowledge around us that is changing.

So on the periodic table, which is off camera here, but I have to keep near me at all times is a variety, a spectrum of different half lives ranging from element, you know, 115 with a half life of, you know, a trillionth of a pico second, all the way up to, you know, uranium, plutonium, stable, you know, it's more stable isotopes or, or tritium, things like that. So what are sort of analogously to facts? What sorts of facts are likely to have longer half lives versus other?

Yeah, I mean, I would say, I mean and we're kind of using fact as kind of like a, in sort of a hand wavy kind of way. But I would say certain bits of knowledge in mathematics, those are the kind of things like if you prove it correctly, that thing's not going to change. And so I would say they have a very long half life or effectively you don't have to worry about them changing. On the other extreme, you have certain things like, like certain, like medical knowledge or like what are kind of like the current medical techniques. Like those kinds of things actually are changing much more rapidly. And so I mentioned before, like my grandfather learning the wrong number of human chromosomes, that's kind of an exception. That being said to my father, who's a retired dermatologist, I think one of his professors told him that he, that his professor used some multiple choice exam, like multiple years in a row and with the exact same questions, exact same answers, like multiple choices. And one year one of the answers was correct and the next year a different answer was correct.

And in medicine they've kind of internalized this kind of thing where they're told that a large fraction of what they learn is going to become obsolete within a few years of graduation. And there's obviously the need for continuing medical education. So there are fields that change very, very rapidly. And I think I'd like to think they internalize it for me. I think the, and I talk about this in the book where I kind of talk about the rates at which knowledge changes. There's one category that I call like a mesofax or kind of in that middle, middle space where they're not changing really rapidly, they're not changing very slowly or effectively, never like, I don't know, learning the number of fingers on a human hand, but they're kind of changing on the order of decades or a human lifetime. Those are sometimes the most dangerous areas of knowledge because you often learn them in the same way you learn things that never change and then you forget to mentally update it. So whether or not it's like, things about the current state of like, scientific understanding, or it could even be certain bits of knowledge about the state of the world, like how many billions of people there are on the planet, if you are.

If you learned it as one thing and forget. Forgot to realize. Oh, wait, actually, yeah, yeah. And the truth is that that's not even the first time. It was like in the 1800s, they thought some of the first, like, like asteroids. The first asteroids, they were actually planets. And, And I think there was like, maybe a generation of children that learned though, like Ceres and the other ones were. Were part of that.

And so we kind of constantly have to update our information. Oftentimes we don't realize it, though, until our kids come home and say, guess what? Dinosaurs look completely different. And then you're kind of blindsided by the slow, steady pace of knowledge change, which of course is like you. You perceive it as in this kind of very stepwise fashion because you're not confronted until the next generation tells you.

Something new and unlike, you know, interesting. Just kind of encountering and having skimmed, you know, through the book, just offline before this interview, but not read in. I intend to read it. You know, the interplay between the. The epistemological fact or knowledge itself and the knowledge, you know, gatherer. You know, for example, Aristotle has had an influence, you know, to this day in most everything, but most every single fact that he had had a half life of zero because it was all wrong. You know, he thought, you know, basically the heavy things fell faster than light things. He felt there were four elements.

He felt that there were, you know, that, that women had fewer teeth than men. You know.

You couldn't even be bothered to count.

Yeah, honey, can you open your mouth? And so I can count it. It's all sorts of things, except for the fact that he was the one that really first provided some evidence that the whales were mammals or dolphins, you know, or mammals by, you know, Aristotle's lagoon is sort of lore. So there's basically like one fact, and yet he's got this tremendous influence, obviously the laws of rhetoric and, and, and poetics and all sorts of other things and Nicomachian ethics and so forth. But, you know, it seems like the. Just his reputation allowed the half life to be extended. It's like this, this concept in knowledge theory I talk about in my latest book Focus, like a Nobel Prize winner, where I Look at the habits of Nobel prize winners and how they would, would do what's called spaced repetition. So there's something called the forgetting curve, which is a half life exponential decay. But if you inject, you know, repetition and kind of bring things back up.

Another potential sponsor, Readwise, does this for me every day and it sends me highlights for my Kindle collection which now has this in it. And so you know, but, and you get a boost in knowledge and then if you do that frequently enough, you can kind of get a, get what's called a rectifier. You get this, this conversion from an alternating current to a direct current. Right. So I wonder, yeah. If this, if the interplay between the knowledge gatherer, the first person to gather the knowledge, his or her reputation. Reputation, if that can artificially kind of forestall the forgetting curve and, or the half life of a particular fact. What do you make of that?

Yeah, I mean, so certainly there are situations, right, where a certain bit of knowledge, whether it's because of the reputation of the person or just because most people are not bothering to actually verify it themselves, it might be incorrect and then kind of just gets propagated. And so errors can actually take far longer than you would expect to actually root. Like to root out or to uproot. Um, and so yeah, you, you can definitely see that kind of thing where it's, it's, yeah, there's, there is knowledge that is not accurate, that takes, that has a much longer lifespan. Um, for that being said, I think part of it is also like, even if the knowledge has been overturned, many people don't necessarily bother to update it. And so I remember a number of years ago there was a lot of concern around like, oh, like, like is the Internet kind of ruining our memories or making us dumber and we have to constantly look things up. I think the flip side of that though is that constantly actually looking things up can mean that you are more likely to be, to have access to the most up to date knowledge as opposed to things that were kind of half remembered that you learned that are not actually accurate anymore. So it's kind of a trade off.

But yeah, rooting out error. And it is a much harder thing to do than we realize.

And the last topic before we wrap up, another one of Sir Arthur C. Clarke's famous phrases, in addition to into the impossible and any sufficiently advanced technology is indistinguishable from the magic of code, is that when an elderly but distinguished scientist says something is possible, he or she is very much likely to be correct. But when he or she says something is impossible, they're most likely to be wrong. I wonder about that. What does that say about first of all, do you agree with that? And second animal like what would that say if true about again this kind of the imprimatur of the fact gatherer on the epistemological quest to actually obtain truth about the physical universe that surrounds us.

Yeah. I mean this is also kind of related to what I think was like Max Planck kind of talks about that like science proceeds like one, one funeral at a time which. And that being said there. And I think I discussed this in the half life facts. It's been a while but there are people have actually studied whether or not this kind of is true is like whether or not older scientists are more amenable to or less amenable to newer ideas or how are they going to be kind of stuck in their kind of older ideas. And I think one of the studies was actually looking at around after Darwin developed his idea of evolution by natural selection to see whether or not which scientists were actually more likely to kind of move into the evolutionary camp. And it looked like I think there was actually no difference in age. And so, so that does give me a little bit of hope.

And I think I would say there's other kinds of evidence that shows it's not necessarily an age thing. I think it's more of like a mindset kind of thing.

I think if you're reputational.

Yeah, reputation. I would say if your reputation is dependent on not keeping an open mind or mentally updating things. Yeah. Then maybe it's going to be very hard to kind of overturn those kinds of things. That being said if you're constantly curious and being willing to, to, to be wrong and, and say oh wow, these are the things that I thought were correct or this idea that I kind of developed is, is no longer, is no longer accurate. I think that, I mean it's often easy to speak about that kind of in the abstract. It's often a lot tougher when it's kind of like your own theory that's kind of on the line. That being said, I think that is ultimately like that is.

That is the goal for science is like, that is like, like the scientific mindset. Not only is it recognizing that these things are draft form that even includes your own ideas, like you have to constantly recognize that but almost be excited by this idea that knowledge is going to be overturned. But like, but that is part of the process of learning more and more about Our world.

That's amazing. I went to Glasgow this summer after meeting at Manchester for the Simons Observatory and, and I did a tour and we went to, you know, the Hunterian Museum and which is, you know, like five different museums and all of them are related in some way to Lord Kelvin, who was there. William Thompson gave us the Kelvin temperature scale. I have a video kind of vlog on the channel about my exploits there. But one of the things that was so startling to me was that Darwin delayed the publication of the Origin of Species because Kelvin had just a few years earlier provided what he thought, and many claimed was basically incontrovertible evidence that the earth was 20 million years old. And Darwin was smart enough to know that evolution could not take place on those. So he, he was kind of, you know, in. Had physics envy or what you might call, you know, just extreme deference to another great scientist.

But he was often wrong, but as they say, never in doubt. And, you know, Calvin made a lot of mistakes and many other people did as well, including, you know, reputedly reportedly, you know, supporting this idea that, you know, after, you know, before 1901 or something like that, he'd said something or supported the notion that, you know, physics, the future of physics, was in the sixth decimal place. You know, he didn't say that, but. But it's sort of misattributed to him.

Yeah.

So it's funny to think, yeah, you're. A later plan comes up with the quantum theory and everything goes to. Goes to hell. Sam Ombudsman, thank you so much for coming all the way from my alma mater in Cleveland, Ohio. Congratulations on this wonderful book, the Magic of Code. I really had such, such a good time just reading it. It's so well written. It's, it's really.

It reads like, it reads like a novel. It's, it's, it's such a. You're such a good, great writer and I did really look forward to reading and listening. I'll make my own Snoop Dog version of the Half Life of Facts. And. And then your third book, what's the name of your third?

It's called Over Complicated.

Over Complicated. Okay, great. I look forward to having. I'll have Kim Kardashian read that one. Sam Arman, thank you for coming all the way. Ghost, Spartans. And we hope to have you back again again for your next great, great triumph.

Well, thank you so much. Yeah, this is a lot of fun. I really appreciate it.

Thanks, Sam.

I hope you enjoyed this video featuring the remarkable Sam Arbusman. I know you're going to love this video featuring Terence Tao, the Mozart of math and certainly a magician himself, describing not just how AI does what it does, but why. Click here for that. Don't forget to, like, comment and subscribe. It really helps me in the algorithm, which, as you know, is just another form of magical code. See you next time on into the Impossible.