What We’re Reading (Week Ending 08 May 2022) - 08 May 2022
Reading helps us learn about the world and it is a really important aspect of investing. The legendary Charlie Munger even goes so far as to say that “I don’t think you can get to be a really good investor over a broad range without doing a massive amount of reading.” We (the co-founders of Compounder Fund) read widely across a range of topics, including investing, business, technology, and the world in general. We want to regularly share the best articles we’ve come across recently. Here they are (for the week ending 08 May 2022):
1. An Interview with “Father of the iPod” Tony Fadell – Ben Thompson and Tony Fadell
Just to touch on that — I love that analogy, I’ll go back to it in a little bit — but the story of the iPod is so crazy. You weren’t even hired until April, yet you shipped in October. How was Apple able to move so quickly? Is there any company that could do that today? I doubt that even Apple could do that today with all their resources. How did that happen where you shipped this completely iconic product that didn’t even exist in the imagination of anyone, I guess in your imagination to an extent, but walk me through that process and how was that even possible?
TF: I think it was a coming together of a lot of things. The first one was experience. I and the people around me had experience for ten years. I pulled in a lot of people from either or General Magic or Philips or other people I just knew that I’d met around Silicon Valley over time. So one was having that network of being able to pull people in who knew what they were doing on this product, that was one thing.
Second thing was having a lot of failure before building these things, and they didn’t really necessarily become commercial successes, they might have been critical successes. So you had enough time doing this stuff. You’re like, “Okay, I’ve done this. I know to make boards. I know how to get software packages together, put all these things to happen.” So again, that was doing something totally new from a product perspective, but the process wasn’t necessarily new.
I think the other one was we had incredible leadership in Steve Jobs. He decreed from the minute after we gave the presentation to him in March of 2001, it was “Go!”. I had already been running it for a year before that doing MP3 players in my startup. So it was like, “Okay”, take all the latest knowledge I had gained during the contracting period, and ran with that.
And then the other one was we just cordoned off and it was, “Make it happen”. I saw so many projects that died at Philips because they didn’t happen fast enough, politics set in. So it was like, “Okay, we have to build this. We have to build it quickly. The holiday season’s coming around. This might be our one and only chance, because who knows when Sony’s going to come in and steal everything” because they were the number one in all audio categories. Every audio category Sony was number one in. So it was like, “Well they’re going to come for this”. So speed was everything.
So I had just been tempered all the time. One is technology changes, the market changes so quickly, you need to have the right experience and process, and we put it all together. Obviously, it was wonderful to have Apple in terms of the customer service angle, parts of the operations angle, but we had to do a lot of new stuff that Apple had never done before, and obviously the marketing, product marketing, pulling all that stuff together. So we got to pick the best bits of Apple and have them focused on us because of the leadership. Then we were able to build very quickly the new bits, throw them together, and just run like hell because at the end of the day, Apple isn’t the Apple you know it today. Twenty-one years ago, Apple was suffering. It had around barely 1% market share in the computer business, in just the US, that’s not worldwide, Apple wasn’t anywhere worldwide. It was only worth $4 or $5 billion, I think. maybe even $3 billion in total. Now it’s worth almost $3 trillion or $2 trillion, whatever it is this week.
So when you have leadership, when you have a competitor or at least you felt there was going to be a competitor coming very quickly, when the technology was there, right place, right time, and we had the right experience, and the company was at its wits end, because it had tried everything it could do to try to get the Mac to get back into the forefront of consumers’ minds with the iMac, whatever, and that wasn’t really going well. This was, “You’ve got to make it happen”, burn the boats, do whatever it takes to see this first product out there and even then it was a marginal success. It was a critical success. Everyone was like, “Wow!” but a lot of people were like, “I can’t buy it. It doesn’t work with my PC.” It didn’t work with Windows. We had to work really hard to make it a success.
It’s interesting that you list all those factors because in your book, when you talk about your experience at Nest, which I think the acquisition from Google maybe was a little more fraught and dramatic than you might have wished it could have been. But you had this contrast, you talk about Google antibodies resisting you and you said, “Oh, we had Apple antibodies resisting us, as well.” But the difference was Jobs protecting you and you also had this cultural bit where Apple needed a hit. To that end, I’m curious, are we reaching a point, a decade past his death, where Steve Jobs’ management abilities are actually becoming underrated? There are the scare stories that are still around. We all know he was this innovator in design, but just from being a manager and getting stuff out the door?
TF: Well, there’s one which is getting stuff out the door, and that’s a process and having good process. There’s another one, which is getting very innovative things out the door, things that are going against the grain of the internal business itself. The iPod was totally different than the computers at the time, “What? Apple’s making what? Stick to computers, Apple.” that’s what we heard from some people. So there’s leadership when you’re maintaining or when you’re operating something that’s already standing and working. And there’s another type of leadership when you’re trying to do something inside an organization that may be successful, may be not successful, but doing something very against the grain and seeing it through and saying, “We’re going to burn the boats. And this is the way it’s going to be.” That takes a different type of leadership and it takes what I always call is air cover. If we didn’t get what we needed, because we were on such a tight schedule, I could only call in the airstrike so often, but, “Steve. Need help.” and from above he would fly in and go, “Okay, what do I need to do?”
We do that a lot today with the businesses we work with, and we have to be the air cover for that and the investments we make and what have you. We fly in and go, “Okay, can we help you in some way? Where can we go to third parties or other ones to help you get what you need to start up this startup?” Leadership is really the key difference in all of this and understanding the difference between data-driven and opinion-based decisions. Steve was really great at understanding what were opinion-based decisions, and it was his opinion at the end of the day that was going to rule, and he was going to make sure everyone understood that “We’re going to do this. And yes, we don’t know if it’s going to be a success, but this is what I want done. Get it done, please.”
You talked about this in your book, actually, the opinion-driven versus data-driven decision making, and how to build the iPod in the first place was an opinion-driven decision, but to bring the iPod to Windows ended up being a data-driven decision. And in the case of an opinion, well it was Steve’s opinion that counted, but because it was a data decision, that’s how you were able to actually change Steve’s mind about going to Windows. Did I summarize that point properly?
TF: Yeah. It’s really correct. Look, Steve’s opinion specifically was at the beginning of the iPod project was, “We are going to make this amazing thing called the iPod” — we didn’t know it was called the iPod at that time — but “We’re going to make this thing, and this is going to drive Mac sales”. So to use the iPod, you’re going to have to buy a Mac, and that was his opinion.
Two years in, the numbers were okay for the Mac fanboys who had Macs, but no one else was interested in switching to a Mac just for an iPod. So we had that data and it showed very clearly that that original opinion or that hypothesis was that more people would buy Macs because the iPod was available was not right. We had a few people, but it was not this huge mass of people switching from Windows to the Mac because the iPod existed and you had to have one. So over time, and this is the third generation, we had to have the Windows connectivity, the Windows functionality, compatibility, to make sure it worked.
And then all of a sudden people were like, “Oh, this iPod thing is really cool. I’m using it on my Windows device, on my Windows laptop, or what have you. But I wonder what the full Apple experience would be?” And then people started buying Macs after they got a taste of the Apple experience with the iPod on their Windows based computer…
…A couple other points. Another interesting episode that we talked about privately was the bake-off at Apple when it came to the iPhone, if it was going to be iPod-based or if it was going to be touchscreen-based. One of the points you made in your book is that the bake-off was very short and it was resource constrained. You needed to make a decision, and then you invested in the right one. And I think this came up in the context of Facebook changing plans on their virtual reality OS. They had the Android one and they had their internal one, and it went on for years. Why do companies fall into this? Is it just that they’re too rich? They have too much money, and so they’re just undisciplined about this?
TF: Absolutely! That’s exactly the right thing, is when there’s too much money, there’s too many people saying that they can do it better, and there’s no time limit or other constraints, money limit, market constraints, what have you, these teams go at it. If you remember, there were two different operating systems going on at the time at Apple, before Steve got back,
That’s right. Yeah.
TF: There were pink and blue and all these things, IBM had it. So there were all these different kinds of in-fighting that happens, and it’s all based on constraints. When there’s a lack of constraints, that’s where all of these things bloom. At Google, when I was there, there were at least four different competing audio projects for audio in the home for playing music. There was four of them! I’m like, four? Why is there four? Everybody had a slightly different take and nobody was willing to go and kill them and prune them and say, “No, this is the right one,” and take all the pieces together, because they were too afraid, for whatever reason, I don’t know. It’s hard enough to have one great product that’s orthogonal to what the company does and saying, “Oh, this is an all new thing,” to have four of them, and say, “We’re going to launch all of them at some point?” That just doesn’t make any sense. Constraints are really key there.
2. Going Where Few Have Gone Before – Inside All Four Rolex Manufacturing Facilities – Benjamin Clymer
While Rolex’s manufacturing and design capabilities were (and still are) the reason that this company is so respected by its peers, it was Wilsdorf’s knack for storytelling that would would elevate Rolex to become the archetype of the luxury wristwatch not only for those within Switzerland, but also all over the world.
In 1927, Wilsdorf heard of a woman British woman named Mercedes Gleitze who had successfully swum the English Channel. Wilsdorf asked Gleitze to wear a Rolex Oyster watch around her neck as she swam. It should be noted that Gleitze had attempted this feat seven times before making it successfully, and then, due an attempt by another woman to steal the spotlight, was asked to swim it again. It was this last time that Gleitze wore the Rolex Oyster, not on her wrist but around her neck.
She didn’t make it. After 10 hours in the freezing water, she was forced to abandon the attempt and be pulled into her trainer’s boat, because of numbness in her extremities. It didn’t matter, and Wilsdorf ran an ad in London’s Daily Mail citing not this most recent attempt, but Gleitze’s earlier successful attempt (which, of course, she swam without a Rolex). Still, her Oyster did withstand up to 10 hours in the bitter cold water of the English Channel, which was no small feat (you can read a detailed account here)…
…This, my friends, is where things get good. Plans-les-Ouates (an industrial park outside Geneva that’s also home to, among others, Piaget, Patek Philippe, and Vacheron Constantin) is where the Rolex of our collective imagination comes to reality – complete with robotic inventory machines straight out of Star Wars, a private gold foundry, and iris scanners. Built in 2006, Rolex Plan-les-Ouates is the largest of all Rolex facilities, comprising six different wings that are 65 meters long by 30 meters wide by 30 meters high, all linked by a central axis. I should also note that everything you can see from the outside of the building is actually less than half of what Rolex has here – the complex is 11 stories high, but you can only see five from the outside. The other six are underground and completely hidden from a casual observer’s eye, or the eyes of would-be competitors.
Here there are not only no cameras allowed inside, but we are also asked to surrender our mobile phones. This facility is, in my opinion, the core of Rolex’s competitive advantage and unlike any other Swiss (German, or Japanese) watchmaking facility on the planet. It may actually be completely unique in other industries too. I’ll explain why below.
Upon entry (and surrender of all digital device), we take a small elevator a few floors underground. The doors open to reveal what looks to be something akin to Dr. Evil’s underground lair, in the best possible way. The floor is cement, the hallways are wide. Access control points are everywhere – if someone doesn’t absolutely need to be in a particular room, then they simply do not have access to it. We immediately notice a gigantic elevator door – and when I say gigantic, I mean an elevator at a scale that I’ve never seen. I inquire about it – it can hold a load of up to five tons.
We are shuffled into a secure room – we are about to see the legendary Rolex automated stock system. Our guide places his eyes to the iris scanner (no lie) the doors slide open, and what we see is downright startling…
…Sorry guys. No photos allowed, nor provided. So what I will do is give you my best written description of what this absolutely extraordinary automated system looks like. There are two 12,000 cubic meter vaults, spliced by a network of rails totaling 1.5 kilometers, transporting over 2,800 trays of components per hour between the 60,000 storage compartments and the workshops upstairs. The view is straight out of Star Wars, minus the 1970s camp. This is efficiency defined.
Once someone within the workshops above requests a component, this incredible system takes just 6-8 minutes to retrieve it and deliver it to their work station. I remember when I was in undergraduate business school, our supply chain professional proclaimed Wal-Mart to be the model of professional logistics. I would almost guarantee you he said that because he’d never been to Rolex Plans-les-Ouates…
…Rolex owns its own foundry, where it creates its very own formulas for three different kinds of gold, and its own formulation of 904L stainless steel. Every single alloy used by Rolex is produced entirely in-house because, as they are quick to point out, the composition of the metal is the most important factor in determining a watch’s aesthetic, mechanical, and dimensional properties.
Rolex is able to make these special compounds because they have invested in something that few other watch companies would even dream of: a central laboratory with world-class experts in not only materials, but also tribology – the science of friction, lubrication, and wear – chemistry, and materials physics. This laboratory was truly extraordinary to see, and what was perhaps most impressive about the lab was not only the incredible testing going on, and the machines they’ve developed themselves (for example, Rolex invented a machine to open and close an Oyster bracelet clasp 1,000 times in a matter of minutes), but also the people who work there. I was asked not to mention from where Rolex has retained many of its top-tier scientists, but you can guess, and they are 100 percent not from the watch industry…
…I think what was perhaps most surprising about my visit to Chene-Bourg was the quality of gemstone and setting work Rolex does. I don’t really think of Rolex producing many watches with diamonds and stones, and they admit they don’t. But, this is Rolex and if they are going to do something, they are going to do it the Rolex way. This means 20 in-house gem setters, some of whom have names like Bulgari and Cartier on their resume. The stones they use? Only IF quality – otherwise known as “internally flawless” for those not familiar with jewelry-speak.
One of the coolest things I saw here was a machine that Rolex uses to filter the stones they receive for fakes, or anything that might not be what it’s supposed to be. One assumes that any supplier of Rolex understands just how big a business it is and might be tempted to take advantage of this, perhaps by including fake diamonds in with the real stones. Yes, well, Rolex has a machine in-house that can filter stones in mass to cull out anything that isn’t a real diamond. The machine costs tens of thousands of dollars so I asked how frequently they received a stone from a supplier that wasn’t an actual diamond. The answer? About one out of 10 million. They do it anyway, because this is Rolex.
3. Where Do Space, Time and Gravity Come From? – Steven Strogatz and Sean Carroll
Strogatz (02:56): It’s very exciting to me to be talking with the master of emergent space-time. Really mind-boggling stuff, I enjoyed your book very much. I hope you can help us make some sense of these really thorny and fascinating issues in, I’d say, at the frontiers of physics today.
Why are you guys, you physicists, worrying so much about space and time again? I thought Einstein took care of that for us a long time ago. What’s really missing?
Carroll (03:21): Yeah, you know, we think of relativity, the birth of relativity in the early 20th century, as a giant revolution in physics. But it was nothing compared to the quantum revolution that happened a few years later. Einstein helped the beginning of special relativity, which is the theory that says you can’t move faster than the speed of light, everything is measured relative to everything else in terms of velocities and positions and so forth. But still, there was no gravity in special relativity. That was 1905. And then 10 years later, after a lot of skull sweat and heavy lifting, Einstein came up with general relativity, where, he had been trying to put in gravity to special relativity, and he realized he needed a whole new approach, which was to let space-time be curved, to have a geometry, to be dynamical. It’s the fabric of space-time itself that responds to energy and mass, and that’s what we perceive as gravity.
(04:14) And as revolutionary as all that was, sort of replacing fundamental ideas that had come from Isaac Newton, both special relativity and general relativity were still fundamentally classical theories. You know, we sometimes prevaricate about the word “classical,” but usually what physicists mean is, the basic framework set down by Isaac Newton in which you have stuff, whether it’s particles or fields, or whatever. And that stuff is characterized by what it is, where it is, and then how it’s moving. So for a particle, that would be its position and its velocity, right? And then, from that, you can predict everything, and you can observe everything and it’s precise and it’s deterministic, and this gives us what we call the clockwork universe, right? You can predict everything. If you knew perfect information about the whole world, you would be what we call “Laplace’s demon,” and you’d be able to precisely predict the future and the past.
(05:08) But even general relativity, which says that space-time is curved, that still falls into that framework. It’s still a classical theory. And we all knew, once quantum mechanics came along, circa 1927, let’s say. It was bubbling up from 1900, and then sort of — it triumphed in 1927, at a famous conference, the fifth Solvay Conference, where Einstein and Bohr argued about what it all meant.
(05:32) But since then, we’ve accepted that quantum mechanics is a more fundamental version of how nature works. I know — you said this for all the right reasons, but it’s not that quantum mechanics happens at small scales. Quantum mechanics is the theory of how the world works. What happens at small scales is that classical mechanics fails. So you need quantum mechanics. Classical mechanics turns out to be a limit, an approximation, a little tiny baby version of quantum mechanics, but it’s not the fundamental one.
And since we discovered that, we have to take all of what we know about nature and fit it into this quantum mechanical framework. And we have been able to do that for literally everything we know about nature, except for gravity and curved space-time. We do not yet have a full, 100% reliable way of thinking about gravity from a quantum point of view…
…Strogatz (11:32): So I think that segues very nicely into the next thing I was going to ask you. We’re hoping, by the end of this episode, to give people a feeling of what it means for space-time to be emergent. But what would it mean for you, or anybody studying space and time, for them to be emergent?
Carroll (12:05): So I don’t think that there is any such thing as a position or a velocity of a particle. I think those are things you observe, when you measure it, they’re possible observational outcomes, but they’re not what is — okay, they’re not what truly exists. And if you extend that to gravity, you’re saying that what we call the geometry of space-time, or things like location in space, they don’t exist. They are some approximation that you get at the classical level in the right circumstances. And that’s a very deep conceptual shift that people kind of lose their way in very quickly.
(12:58) It’s a tricky word. We have to think about it. Emergence is kind of like morality. Sometimes we agree on it when we see it. But other times, we don’t even agree on what the word is supposed to mean. So, the physicists, and mathematicians, and other natural scientists tend to — but not always — rely on what a philosopher would call weak emergence. And weak emergence is basically a convenience, in some sense. The idea is that you have a comprehensive theory, you have a theory that works at some deep level. Let’s say, the standard example is gas in a box, okay? You have a box full of some gaseous substance, and it’s made of atoms and molecules, right? And that’s the microscopic theory. And you say that, okay, I could — in principle, I could be Laplace’s demon, I could predict whatever I want, I know exactly what’s going on.
(13:47) But, we human beings, when we look at the gas in the box with our eyeballs, or our thermometers, or whatever, we don’t see each individual atom or molecule, and its position and its velocity, we see what we call coarse-grained features of the system. So we see its temperature, its density, its velocity, its pressure, things like that. And the happy news — which is not at all obvious or necessary, it’s kind of mysterious when it happens and when it doesn’t — but the happy news is that we can invent a predictive theory of what the gas is going to do just based on those coarse-grained macroscopic observables. We have fluid mechanics, right? We can model things without knowing what every atom is doing. That’s emergence, when you have a set of properties that are only approximate and coarse-grained, that you can observe at the macroscopic level, and yet you can predict with them. And weak emergence just means, there’s nothing new that happened along the way. You didn’t say that, oh, when you go to the larger scales and you zoom out, fundamentally new essences or dynamics are coming in. It’s just sort of the collective behavior of the microscopic stuff. That’s weak emergence.
(15:01) There’s also strong emergence where spooky new stuff does come in. And people talk about the necessity of that when they think about consciousness or something like that. I’m not a believer in strong emergence at the fundamental level. So, to me, what the emergence of space-time means is that space-time itself is like, the fluid mechanics. It’s like gas temperature and pressure and things like that. It’s just a coarse-grained, high-level way of thinking about something more fundamental, which we’re trying to put our finger on.
Strogatz (15:34): Wow, as you’re describing the gas in a box, I happen to be sitting in a box. I’m in a studio that is kind of box-shaped. There is a gas in here, which is the air that I’m breathing.
So anyway, yeah, very vivid to me, the example you’re talking about. And it is amazing, isn’t it? That there are laws at that collective or emergent scale that work, that don’t — you know, like thermodynamics was oblivious to statistical physics. In fact, was discovered first, and only later, the microscopic picture came out. And so, I guess you’re saying something like that would be happening now with space and time and gravity, that we have the macroscopic theory that’s Einstein’s.
Carroll (16:14): When I’m not spending my research time studying quantum mechanics and gravity, I’m studying emergence. I think that there’s a lot to be done here, to be sort of cleaned up and better understood, in a set of questions that spans from philosophy to physics to politics and economics, not to mention biology and the origin of life. So, I think that these are deep questions that we’ve been kind of messy and sloppy about addressing, but I don’t think that the emergence of space-time is difficult for that reason.
(16:45) So, when you talk about, is the United States emergent from its citizens? Or is Apple Computer Company emergent from something? Those are hard questions. Those are like, tricky, like “where do you draw the boundary?”, etc. But for space-time, I think it’s actually pretty straightforward. The lesson, the important take-home point for the podcast is, you don’t start with space-time and quantize it, okay? Just like when you have the gas in the box, you’re trying to get a better and better theory of the gas in the box, but you realize that it’s made of something fundamentally different. And I think that’s what I’m proposing, and other people are proposing for space-time as well, that the whole thing that used to work for electromagnetism and particles and the Higgs boson and the Standard Model, where you started with some stuff and quantized it, that’s not going to be the way it’s going to happen for gravity and space-time. You’re going to have something fundamentally different at the deep micro-level, and then you’re going to emerge into what we know of as space-time.
Strogatz (17:46): Shouldn’t we start talking about entanglement, at this point, maybe?
Carroll (17:49): Never too early to start talking about entanglement.
Strogatz (17:51): Let’s talk about it. What is it? I hear it a lot. I hear quantum people talking about it. Nowadays, especially, with quantum computing, we keep hearing about entanglement. Why don’t you just start with telling us what it means, where the idea came from?
Carroll (18:04): Yeah, I mean, let’s think about the Higgs boson. We discovered it a few years ago, it’s a real particle, and I wrote a book about it, The Particle at the End of the Universe. The Higgs boson — one of the reasons why it’s hard to detect is that it decays. It has a very, very short lifetime, right? So, you can imagine if someone put a Higgs boson right in front of you, it would generally decay into other particles in about one zeptosecond. That’s 10-21 seconds. Very, very quickly.
(18:31) One thing it can do, it can decay into an electron and a positron, an antielectron. So it can decay into two particles, electron and positron. Now remember quantum mechanics. So, you can predict roughly how long it will take the Higgs boson to decay, but when it spits out that electron and positron, you can’t predict the direction in which they’re going to move.
(18:54) I mean, that makes perfect sense because the Higgs boson itself is just a point. It has no directionality in space. So there’s some probability of seeing the electron, in a cloud chamber or whatever, moving in whatever direction you want. Likewise, for the positron, there’s some probability, seeing it moving in whatever direction you want. But you want momentum to be conserved. So you don’t want the Higgs boson sitting there, stationary, to decay into an electron and a positron both moving rapidly in the same direction. That would be a shift in the momentum, right?
(19:26) So, even though you don’t know what direction the electron is going to move in, and you don’t know what direction the positron is going to move in — sorry, I’m already, I’m being, I’m being the person who I make fun of, I’m speaking as if these are real. Even though you don’t know what direction you will measure the electron to be moving in, and you don’t know what direction you will measure the positron to be moving in, you know that if you measure them both, they will be back to back. Because they need to have equal and opposite momentum, for those to cancel out.
(19:54) So what that means is, if you believe all those things, right away, this is why we believe there’s only one wavefunction for the combined system of the electron and the positron. It’s not an independent question, what direction are you going to measure the electron in? What direction are you going to measure the positron in? It’s a statement you need to ask at the same time. That’s entanglement, right there. Entanglement is the fact that you cannot separately and independently predict what the observational outcome is going to be for the electron and the positron.
(20:26) And this is completely generic and everywhere in quantum mechanics. It’s not a rare, special thing. Many things are entangled with many other things. It’s the unique and fun and very useful time when things are not entangled with each other. It took a long time — like, Einstein and his friends — Einstein, Podolsky and Rosen, EPR — published a paper in 1935 that really pointed out the significance of entanglement. Because it was sort of there, already, implicit in the equations, but no one had really shone a flashlight on it, and that’s what Einstein did. And the reason why it bothered him is because when that Higgs boson decays and the positron and the electron move off in opposite directions, you can wait a long time, let’s say you wait a few years before you measure what direction the electron is moving in.
(21:14) So, both particles are very, very far away from each other. And now when you measure the location of one, supposedly the location of the other one is instantly determined. And there’s no limit of the speed of light or anything like that. So for obvious reasons, Einstein, very fond at the speed of light as a limit on things, he didn’t like that. He never really quite thought that that was the final answer, he was always searching for something better.
Strogatz (21:39): And the argument goes nowadays that it’s okay, it’s no violation of special relativity, because you can’t use this to transfer any information or something? Is that the statement?
Carroll (21:39): Yeah, well, you know, there’s, there’s a whole bunch of statements that one can make. But the one that we absolutely think is true, is the one that you just made. If you imagine these two particles moving back-to-back, and one person detects one, and there’s another one, you know, a light-year away, who’s going to detect the other one, the point is that they don’t know what your measurement outcome is, you would have to tell them.
So even though in the global point of view, now, the location where the other particle is going to be detected is known to God, or to the universe, it is not known to any particular person sitting at any location within the universe. It takes the speed of light time to take a signal that would let you know that there is some now new fact about the matter, where you’re going to observe the positron. So, you cannot actually use this for signaling, you just don’t know what has happened when your other observer has measured something. And you can actually prove that, under reasonable assumptions, in the theory as we know it.
(22:43) So it seems as if this is the tension, that the way the universe works involves correlations that travel faster than the speed of light, but in some well-defined sense, information does not travel faster than the speed of light. That should worry you, that we didn’t define any of these words. So you know, what does that mean? You’re not going to build a transporter beam or anything like that out of this stuff.
(23:09) But — but let me just add one other thought, which I think, again, is a result of my quirky way of thinking about these things, which is not entirely standard, which is, people really like locality. Like, locality is a central thing. Locality is just the idea that if I poke the universe at one point in space-time, the effects of that poke will happen at that point, and then they will ripple out. But they will ripple out to other points no faster than the speed of light, okay? There’s nothing I can do to poke the universe here that will change the state of the universe in a tangible way very, very far away. And you can see how this entanglement thing is kind of on the boundary of that, like, the description of the universe changes instantly far away, but no information is traveling.
(23:51) So then, if you believe that locality is fundamental like that, then you’re sort of asking this question, why does the universe almost violate that but seem to not quite? That’s the puzzle that we have. And this is — a lot of ink has been spilled in the foundations of quantum mechanics.
(24:06) I think about it entirely the other way around, because I think of the wavefunction as the fundamental thing, right? I think that’s what exists in reality. And the wavefunction, like the wavefunction of this positron and electron is utterly nonlocal. It just exists all — it’s a, it’s a feature of the universe as a whole right from the start. So, I also have a mystery to be explained, but my mystery is the opposite way. It’s not “why is locality approximately or, you know, seemingly violated by entanglement?” It’s “why is there locality at all?” Like, that’s the puzzle to me.
4. Nvidia: The Machine Learning Company (2006-2022) – Benjamin Gilbert and David Rosenthal
Ben: This was occurring to me as I was watching Jensen ensuring the omniverse vision for NVIDIA and realizing NVIDIA has really built all the building blocks—the hardware, the software for developers to use that hardware, all the user-facing software now, and services to simulate everything in our physical world with an unbelievably efficient and powerful GPU architecture.
These building blocks, listeners, aren’t just for gamers anymore. They are making it possible to recreate the real world in a digital twin to do things like predict airflow over a wing, simulate cell interaction to quickly discover new drugs without ever once touching a petri dish, or even model and predict how climate change will play out precisely.
There is so much to unpack here, especially in how NVIDIA went from making commodity graphics cards to now owning the whole stack in industries from gaming, to enterprise data centers, to scientific computing, and now even basically off-the-shelf self-driving car architecture for manufacturers.
At the scale that they’re operating at, these improvements that they are making are literally unfathomable to the human mind. Just to illustrate, if you are training one single speech recognition machine learning model these days—just one model—the number of math operations like adding or multiplying to accomplish it is actually greater than the number of grains of sand on the earth.
David: I know exactly what part of the research you got that from because I read the same thing and I was like, you got to be freaking kidding me.
Ben: Isn’t that nuts? There’s nothing better in all of the research that you and I both did to better illustrate just the unbelievable scale of data and computing required to accomplish the stuff that they’re accomplishing and how unfathomably small all of these are the fact that this happens on one graphics card.
David: Yeah, so great…
…Ben: It’s funny because that feels like that’s the killer use case, but that’s just the easiest use case. That’s the most obvious, well-labeled data set that these models don’t have to be amazingly good because they’re not generating unique output. They’re just assisting and making something more efficient.
Then flash forward 10 more years and now we’re in these crazy transform models with, I don’t know if it’s hundreds of millions or billions of parameters. Things that we thought only humans could do are now being done by machines and it’s happening faster than ever. I think to your point, David, it’s like, oh, there was this big cash cow enabled by neural networks and deep learning in advertising. Sure, but that was just the easy stuff.
David: Right. That was necessary though. This was finally the market that enabled the building of scale in the building of technology to do this. In the Ben Thompson, Jensen interview, Ben actually says this, when he realizing this talking to Jensen says, this is Ben talking, “The way value accrues on the internet in a world of zero marginal costs where there’s just an explosion in abundance of content, that value accrues to those who help you navigate the content.” He’s talking about aggregation theory.
Then he says, “What I’m hearing from you, Jensen, is that, yes, the value accrues to people to help you navigate that content, but someone has to make the chips and the software so that they can do that effectively. It used to be that Windows was the consumer-facing layer and Intel was the other piece of the Wintel monopoly. This is Google, and Facebook, and a whole list of other companies on the consumer side, and they’re all dependent on NVIDIA. That sounds like a pretty good place to be.” And indeed, it was a pretty good place to be.
Ben: Amazing place to be.
David: Oh my gosh. The thing is, the market did not realize this for years. I didn’t realize this and you probably didn’t realize this. We were the class of people working in tech as venture capitalists that should have.
Ben: Do you know the Marc Andreessen quote?
David: Oh, no.
Ben: Oh, this is awesome. Okay, it’s a couple years later, so it’s getting more obvious, but it’s 2016. Marc Andreessen gave an interview. He said, “We’ve been investing in a lot of companies applying deep learning to many areas, and every single one effectively comes in building on NVIDIA’s platforms. It’s like when people were all building on Windows in the ’90s we’re all building on the iPhone in the late 2000s.” Then he says, “For fun, our firm has an internal game of what public companies we’d invest in if we were a hedge fund. We’d put in all of our money to NVIDIA.”
David: It was a paradigm that called all of their capital in one of their funds and put it into Bitcoin when it was like $3000 a coin or something like that. We also have been doing this. Literally, NVIDIA stock—this is now 2012, 2013, 2014, 2015—doesn’t trade above $5 a share. NVIDIA today as we record this is I think about $220 a share. The high in the past year has been well over $300. If you realized what was going on, and again, in a lot of those years, it was not that hard to realize what was going on, wow, it was huge.
Ben: It’s funny. We’ll get to what happened in 2017 and 2018 with crypto and a little bit, but there was a massive stock run up to like $65 a share in 2018. Even as late as I think the very beginning of 2019, you could have gotten it. I tweeted this, and we’ll put the graph on the screen in the YouTube version here. You could have gotten it in that crash for $34 a share in 2019. If you zoom out on that graph, which is the next tweet here, you can see that in retrospect, that little crash just looks like nothing. You don’t even pay attention to it in the crazy run up that they had to $350 or whatever their all time high was.
David: Yeah. It’s wild. A few more wild things about this. AlexNet happened in 2012. It’s not until 2016 that NVIDIA gets back to the $20 billion market cap peak that they were in 2007, when they were just a gaming company. That’s almost 10 years.
Ben: I really hadn’t thought about it the way that you’re describing it. The breakthrough happened in 2010, 2011, 2012. Lots of people had the opportunity, especially because freaking Jensen is talking about it on stage. He’s talking about our earnings calls at this point.
David: He’s not keeping this a secret.
Ben: No, he’s trying to tell us all that this is the future. People are still skeptical. Everyone’s not rushing to buy the stock. We’re watching this freaking magic happen using their hardware, using their software on top of it. Even semiconductor analysts who are like students of listening to Jensen talk and following the space very closely think he sounds like a crazy person when he’s up there espousing that the future is neural networks, and we’re going to go all in. We’re not pivoting the business, but from the amount of attention that he’s giving in earnings calls to this versus the gaming. I mean, everyone’s just like, are you off your rocker?
David: I think people have just lost trust and interest. There were so many years, they were so early with CUDA and early takeout. They didn’t even know that AlexNet was going to happen. Jensen felt like the GPU platform could enable things that the CPU paradigm could not, and he really had this faith that something would happen. He didn’t know this was going to happen. For years, he was just saying that we’re building it, they will come.
Ben: To be more specific, it was that, well, look, the GPU has accelerated the graphics workload. We’ve taken the graphics workload off of the CPU. The CPU is great. It’s your primary workhorse for all sorts of flexible stuff. But we know graphics need to happen in its own separate environment, have all these fancy fans on it, and get super cooled. It needs these matrix transforms. The math that needs to be done is matrix multiplication.
There was starting to be this belief that like, oh, well, because the apocryphal professor told me that he was able to use this program that matrix transforms to work for him, baybe this matrix math is really useful for other stuff. Sure, it was for scientific computing. Then, honestly, it fell so hard into NVIDIA’s lap that the thing that made deep learning work was massively parallelized matrix math. NVIDIA is just staring down their GPUs like, I think we have exactly what you are looking for.
David: Yes. There’s that same interview with Bryan Catanzaro. When all this happened, he says, “Deep learning happened to be the most important of all applications that need high throughput computation.” Understatement of the century. Once NVIDIA saw that, it was basically instant. The whole company just latched on to it.
There are so many things to laud Jensen for. He was painting a vision for the future, but he was paying very close attention, and the company was paying very close attention to anything that was happening. Then when they saw that this was happening, they were not asleep at the switch.
Ben: Yeah, 100%. It’s interesting thinking about the fact that in some ways, it feels like an accident of history. In some ways, it feels so intentional, that graphics are an embarrassingly parallel problem because every pixel on a screen is unique. You don’t have a core to drive every pixel on the screen. There are only 10,000 cores on the most recent NVIDIA graphics cards, but there’s not, which is crazy, but there are way more pixels on the screen.
They’re not all doing every single pixel at the same time every clock iteration. But it worked out so well that neural networks also can be done entirely in parallel like that where every single computation that is done is independent of all the other computations that need to be done, so they also can be done on this super parallel set of cores.
You got to wonder, when you kind of reduce all this stuff to just math, it is interesting that these are two very large applications of the same type of math in the search space of the world of what other problems can we solve with parallel matrix multiplication? There may be more, there may even be bigger markets out there.
5. Twitter thread on an interview of Ted Weschler – Thomas Chua
1. Who is Ted Weschler? He was the founder and managing partner of Peninsula Capital Advisors. Between 1999 and 2011, the $2B fund returned 1,236% to its investors. He wanted to meet his hero and so he bided on the annual auction lunch with Buffett.
2. One fateful Tuesday morning, he received a phone call that changed his life. It was Buffett on the other end. He had won the annual charity auction lunch. Ted flew out to Omaha two days later to meet his hero. Everything clicked. Ted bid again the following year and won!
3. This time, Warren asked him: “I think you’d be a pretty good fit out here. Would you have any interest in working at Berkshire?” He panicked. On one hand, he was running a successful fund and his family was in Charlottesville. But on the other, this is Warren Buffett!
4. He wrote Buffett a letter when he got back to Charlottesville explaining that it was difficult because his family was rooted here. Buffett replied “You can manage money from the moon as far as I am concerned.” Buffett was a real pioneer in the work from home trend 😂…
…7. Investing is a game of connecting the dots. We want to build up a lot of data in our minds and understand why the business will be vastly different five years from now than what the market perceives. He reads trade journals regularly to understand businesses…
…9. Why he always feel positive? United States has a system that works. There’s will be negativity every now and then. But if you take a long-term view, there’s innovation coming out every day and it keeps getting better. It’s hard not to be optimistic.
6. Sources of Enduring Business Success – John Huber
I recently read through the letters of Nick Sleep, who ran a very successful investment fund in the United Kingdom before closing it last decade. Sleep is a great thinker and I highly recommend his work. One thing Sleep wrote a lot about is how the average holding time period for many of the stocks he owned was around 50 days, whereas he planned to hold these stocks for more than 250 weeks (5 years). I think his key observation is important: The marginal buyer who is holding a stock for 2 months is not placing much emphasis on that company’s competitive advantage because that advantage won’t matter much at all over the next few months; what matters over that period of time are things like market perception, news flow, sentiment, and perhaps short-term business momentum…
…So what Sleep did is he decided to compete in a different game. Instead of attempting to determine how the crowd will react this quarter or how the trajectory of the business will fare this year, he wanted to focus on the factors that contributed to a business’s ultimate potential. What attributes give this company an advantage? What will lead this company to success through both good times and bad times (because if you’re a long-term shareholder, all companies face headwinds at some point).
…Sleep used the example of Walmart’s cost advantage. Walmart’s business model was to offer the lowest prices on everyday merchandise, and steadily gain scale advantages through larger and larger bulk purchases from suppliers at lower and lower unit prices, which meant further savings to customers, which led to more growth and more scale advantages. Sleep coined a term for this business model: “scaled economies shared”, meaning the business gained scale, but instead of keeping the excess profits for itself, it gave these scale advantages to the customer in the form of lower prices. This sacrificed near term profits but led to far greater future profits, which of course is where value comes from.
Walmart, Costco, and Amazon all exhibit this basic business model, and all have achieved great success. But what Sleep noticed is that investors — even when they understood this business model — still undervalued all of these companies because they placed too much emphasis on shorter term factors such as seasonal same-store sales trends, quarterly margins, or the business cycle. All of this focus came at the expense of what really mattered, which was the cost advantage that was so hard for competitors to replicate….
…Last summer, investors sold Amazon after its Q2 earnings report because the next few quarters would face tough comps from the gangbuster 2020; but Amazon’s value in 2032 has little to do with the comps it faces in 2022. It has a lot to do with the durability of its network, the economies of scale, the distribution advantages, the culture of operational excellence; none of that will likely drive the stock this quarter, but it’s what matters most to the stock over the next decade.
A mismatch of time horizons lead some investors to more heavily weight the short-term and deemphasize these sources of “enduring business success”.
7. Twitter thread on how company leaders handle crises – Dan Rose
I was at Amzn early ’00s when we lost 95% of our market cap. Later at FB I negotiated a down-round in ’09, and then in ’12 our stock dropped 50% post-IPO. I was on the board of a public company that went bankrupt (Borders) and a start-up that went under (Hello). Some lessons:
1/Raise capital when you can, not when you need it. Amzn tapped convert debt in Feb ’00 – if we had waited another month we would not have survived. 9 years later at FB we raised a 33% down-round despite having plenty of runway. Don’t wait until your back is up against the wall
2/Cash is king. Forget about valuation, dilution, etc – if you run out of cash, none of it matters. Borders used its free cash flow to buy back stock for yrs, ignoring the internet. By the time a PE firm fired the board and asked me to join in ‘09, we had no runway for turnaround…
…4/Change the tone. Amzn did a small but symbolic RIF in 2000. Around that time, Jeff was presented with a team t-shirt – he threw the team out of his office and banned all company swag. We even removed aspirin from the break rooms, served coffee and water. Small acts set the tone
5/It starts from the top. Zuck showed up to work in Jan ’09 wearing a tie, and he wore it every day for an entire year. His message to the company: “it’s time to get serious about our business.” Every time we walked into a meeting with Mark, we were reminded things had changed
6/Reset the team. In the middle of covid I addressed the exec team of a travel start-up whose revenue dropped to zero overnight. I encouraged them to re-evaluate their team. Some people step up in a crisis – they are your future leaders. Others will jump ship – good riddance…
…9/Communicate, a lot. When FB’s stock plummeted after our IPO, I addressed the issue with employees rather than pretending stock price didn’t matter. It’s tempting to go into a foxhole when times are tough. Don’t do that, your team needs you more than ever
10/Keep telling your story. I stayed at Amzn during this time because Jeff sold me on his vision. When GFC postponed FB’s IPO by 4 years, Zuck never stopped talking about the mission. Churchill taught the world the power of storytelling in a crisis
Disclaimer: None of the information or analysis presented is intended to form the basis for any offer or recommendation. Of all the companies mentioned, we currently have a vested interest in Alphabet (parent of Google), Amazon, Apple, Costco, and Meta Platforms (parent of Facebook). Holdings are subject to change at any time.
