#144: The Fine Line Between Science And Fiction — Palmer Luckey
Palmer Luckey is an American tech entrepreneur and billionaire. He has founded two companies: Oculus VR (acquired by Facebook for $2 billion in 2014), and Anduril (recently valued at $8.5 billion). He has been described as the real-life Tony Stark.
JOSEPH WALKER: Palmer Luckey, welcome to the podcast.
PALMER LUCKEY: Of course. Thank you for having me.
WALKER: The writer Neal Stephenson delineated two ways in which science fiction impacts science itself. The first is that science fiction can inspire people to pursue careers in science and engineering, and the second is that science fiction can inspire specific, tangible ideas. My first question is: do science fiction writers really inspire entirely new technologies, or do they merely amplify things that scientists and inventors are already contemplating?
LUCKEY: Oh, they definitely invent new technologies. I mean, the ideas behind virtual reality existed before computers. The idea of technology that implanted ideas and experiences in the mind goes back well over 100 years. And then, as technologists kind of created things that made it more clear how that would happen, science fiction authors, they looked and said “Ha I see there's these computational machines, they're capable of drawing graphics, I posit that's the specific mechanism by which you in the future would draw real time three dimensional graphics for VR.” So, maybe there's a little bit of guidance there, but there's definitely things that are invented from whole cloth by writers, and I'd say, like robotics, it was the same thing. The concept of automatons was very much born of science fiction. It was people who said : ”Okay, we've got these very basic mechanical machines and I posit that someday they'll be as advanced as people and they'll be able to walk around and talk around and do everything that we do.” And that wasn't something scientists were even close to doing or even really working on but the job of a science fiction author is not to work on things that are possible. It's to imagine what could be.
WALKER: I saw on Twitter that 20,000 Leagues Under the Sea was your favourite book as a kid.
LUCKEY: Oh, yeah, great book.
WALKER: Why did that inspire you?
LUCKEY: Well, I was a Jules Verne fan in general, so I love Journey to the Centre of the Earth and loved 20,000 Leagues Under the Sea. But in particular, 20,000 Leagues Under the Sea was diving into a lot of the very technical aspects of what science fiction can do. So, a lot of science fiction in the modern day is not so much hard science fiction. It's telling a story and the science is incidental. Like I love Star Wars, but it is more of a story that happens to be set in the future than what you could call science fiction. And I really love that 20,000 Leagues Under the Sea was a very early example of extremely technical fiction. I mean, it gets into… If you saw on Twitter, I was pointing out a passage where you have the captain explaining the precise chemistry of his battery packs in the Nautilus and how it works and how he gets the magnesium for the reaction and how he recharges them, and that's a type of thing that really inspired me when I was younger. And I still like hard sci-fi today where it's about the ideas and it's about the science, it's about the technology, and the rest of the story is kind of built on that scaffold but doesn't dominate.
WALKER: Have you ever written any of your own science fiction?
LUCKEY: I have. I've actually written some speculative science fiction short stories that I've never published. They're just for my own reference, but I've had some fun ideas.
WALKER: That's cool. Can you share any of the ideas?
LUCKEY: Yeah. One of them is a story called 'The Last Hot Rod'. It's about a group in the future that is the last group of people on Earth that build and repair and drive their own cars because… they're not necessarily banned but it's become so out of fashion and de facto impossible for ordinary people to own cars, instead they're all just shuttled everywhere by autonomous pods and it explores the concept of what does it mean to build your own technology, to understand your own technology, and to kind of be a master of your own destiny rather than part of a much larger machine. So that's the idea.
There's another that's a little bit more of a work in progress, but the core premise is: “What happens if perfect augmented reality really becomes part of our society, and how does that change how people relate to one another?” For example, one of the concepts — and again, it's very much not a story, so much as a kind of framework for this idea — but suppose that you end up with augmented reality glasses that do the thing that everybody wants, which is to have the ability to remember... You see somebody you haven't seen in a while and it reminds you what their name is, what you guys talked about last time, when you last met, this useful piece of information. Or if you could meet a new person and you would instantly know who they are, where they work, all this relevant information. And so what happens is, it becomes a social faux pas in the future to make small talk because why would you — you see somebody, you then see where they work and all that — why would you then say, “Oh! So you work at Microsoft?” Because, well, they know that I know that they work at Microsoft and you know that I know where I work. Why would we talk about any of the things that we instantly know about each other simply by meeting for the first time, because it's all in our HUDs or display? And so it leads to a very interesting dynamic where talking about anything that you could have already known without talking is considered just a total waste of time, very old school, old world thing that only old people do. And so you would only talk about things that are immediately productive and of a new nature that are not communicated by the AR system. So you might walk up to somebody you've never met before and say: “The best way to tackle the challenge with China is X, Y, and Z.” And I would know you're interested because you're in some part of government, and I would know that you would take me seriously because I work on some specialised area that's related to the conversation. Now, is this the future? I don't know. But it's an interesting thought that occurred to me, and I thought: “What would social norms look like if we didn't need to have small talk to get to a level of initial understanding for people? How would that change things?” And of course, there's things you gain and things you lose.
WALKER: That's fascinating.
LUCKEY: Anyway, I've written actually quite a few science fiction short stories. Just for fun.
WALKER: I'd love to read them someday. You should publish them.
LUCKEY: I've thought about it, actually. I'll say this is a little bit surprising to me that it's worked out this way, but what I've always told people is that my thought was maybe someday I'll make it into a book or stretch them out a bit, or turn it into anime series or a TV series. But I told people: “You know, I'm waiting for artificial intelligence to lower the production cost of all those forms of media so that I can do it without it being an enormous expenditure of my time and my or someone else's money.” And I always said that with a little bit of a little tongue in cheek. But now it appears that actually may be coming to pass. I actually did feed one of my stories into ChatGPT and asked it to generate a screenplay based on it, and it was able to give me a full, long form screenplay. I was like “Wow, this is actually quite good”. Better than I would have been able to do, for sure, because I'm not a creative writer, I'm a technical writer.
WALKER: Yeah. Apart from your own stories, are there any other sci-fi stories you wish were being told at the moment?
LUCKEY: Oh, you mean just entirely new ideas or other books?
WALKER: Yeah, new ideas. Perhaps, ideas that may have, like, a societal benefit.
LUCKEY: Oh, man. Well, probably just, very big picture: I think that I would love to see more science fiction stories that are optimistic. I read a lot of works from the 50’s, 60’s, 70’s, 80’s, that are unabashedly techno optimist, and of course, there's conflict in these stories. You'll talk to a lot of sci-fi authors who will admit that there's conflict in their stories because nobody wants to read a story about, in the future, VR is great, and AI is great, and everything's cheap and everything is post scarcity and everyone gets along. That's just not an interesting book for most people, so they have to create conflict. But I like that there's a lot of works where they were fundamentally optimistic about where we would be, what technology would do for humanity. And I feel like a lot of works today have fallen into more the dystopian angle, they kind of posit a world where the world is certainly more advanced but also certainly worse off for better technology. And one book I like quite a bit — I'd say right now, it's my favourite book — is The Unincorporated Man. It's a very interesting book about a tech industrialist who manages to enter cryopreservation using a machine of his own design and then wakes up hundreds of years in the future only to find that in the future, technology is extremely advanced, but society has regressed in many ways to corporatism. So for example, in the future, everyone who is born does have a universal income and their health care paid for and their housing paid for. But in exchange, from the day they're born to the day they're an adult, they give up a percentage of shares in themselves where legally they are a corporate entity to their investors, which include the government, it includes their local city, it includes their school. And then by the time you're 18, you are not majority independent. It is impossible to reach adulthood without being majority owned by other corporations, other entities. And so, nobody in the world actually has a right to do what they want without people signing off on it and agreeing to do a board, a shareholder vote to allow you to do something. And so, the titular unincorporated man is our hero. He's the only man in the world who didn't sign the contract that requires you to give away yourself.
He's the only man in the world who can do whatever he wants. And it's interesting because I see us actually maybe trending that way sometimes, where on the one hand, it's great that people have health care, on the other, that then gives other citizens more of a say for them to think. Well, I have a say in what he does every day. Because, after all, it's my taxpayer money that's paying for these people's health care, and so I should have the right to make it impossible for them to, let's say, do risky things or eat whatever they want or smoke whatever they want. And you can imagine that taken to the extreme, pretty soon society is telling you: “You can only do this small set of things that will allow you to do that we believe are productive.” And I fail to say that's one of the areas where I think it's science fiction that gets it right, where in The Unincorporated Man, the technology is actually the positive force that is actually a positive force against society doing bad things. And in most science fiction you'd have it be the opposite. Technology is the cause of the problem.
WALKER: I wonder how consequential these dystopian sci-fi narratives are. So at the end of last year, as I'm sure you would know, Peter Thiel went public with a new riff on the causes of the Great Stagnation.
WALKER: And he kind of noted that in the mid 20th century, there were all these drabbly dystopian narratives about science and technology that began seeping into the culture, and they were facilitated by the dual use nature of so many modern technologies. And in Thiel’s view, they kind of discourage progress because, understandably, people don't want to feel as if they're building the machines that will destroy the world.
WALKER: What's your take on the causal mechanism there?
LUCKEY: Oh, man. I honestly think it's dangerous to even talk about these things.
WALKER: Oh, really?
LUCKEY: I don't even particularly like to talk about stagnation. And I can talk about it because I'm not actually enough of a figure. But if I ever achieve a high enough level of prominence, I will stop talking about it.
LUCKEY: Well, let's take one example: Silicon Valley Bank. You've paid attention to what's happened there, and I would say probably about half of the money that flowed out of Silicon Valley Bank was the result of maybe five people in Silicon Valley making phone calls and text messages, and people will say: “Oh, well, Silicon Valley is a very insular community, there's only a few people who are really highly respected.”
WALKER: A dense network.
LUCKEY: It's a dense network. But the thing is, Silicon Valley Bank, their financials were… You can nitpick about them, but the reality is the entire banking system is actually similarly flawed. If you force any bank to liquidate substantially all of their assets to pay for unrealized losses at the bottom of a down market, they are absolutely going to collapse. I don't think there's a bank on the planet of any size that could survive the level of withdrawals that occurred and having to sell off assets that were at the bottom of their value and had not had time to recover. And people say: “Oh, well, Silicon Valley's insular. That could never happen to, let's say, the main financial system”. But the real world is actually also very concentrated in terms of power and attention. Imagine if one person with a great deal of significant influence across society were to come out and say “I believe that all the banks are going to be insolvent. You need to get your money out as quickly as possible.” Let's imagine it's Bill Gates. Bill Gates comes out and says this. If it was just him, people would think he's off his meds and that something's gone wrong. It's just him. But what if you also throw in Elon Musk, somebody who's very much ideologically separate from Bill Gates? And then what if you also threw in for international flavour, Richard Branson? What if you had this kind of large number of people who are well respected… Sorry, small number of well respected people who said “You need to get your money out because the entire banking system is going to collapse”. I think there's maybe five people in the real world who could also collapse our entire banking system simply by saying that you shouldn't have confidence in it. And I'd say that with stagnation, it's a similar thing. So many of the investments we make in the future of technology, so many investments we make in our university systems, they're predicated on this idea that there will be growth, that technology will continue to grow, and that there are still yet to be undiscovered massive economic reservoirs that we have yet to discover and utilise. And if people were to decide that in fact we are kind of in this stagnation period and that the next 100 years won't see significant growth, that there aren't these new reservoirs of economic value created by technology to be tapped, a lot of that would go away and it would be rational. So the government would stop investing in new technology. Why bother if there's not any big outcome on the other side? Universities would cease to be nearly as important, I think they would change the tone of what they do. And I think that you end up with a society that turns more inwardly focused. How can we simply make the best of what we have today rather than try to build a new tomorrow?
And so you're asking me what my take on it is. I'd say right now I can be candid and say I truly believe that there is room for massive growth. I actually am optimistic. But at the same time, if I ever get to the level of Elon Musk or Bill Gates or Richard Branson where I have that influence, I think I'd be very loath to give a contrary opinion, even if it was the truth, because it would be so negative for society in the short term and probably the long run.
WALKER: But isn't the beauty of Peter's cultural explanation that it's not the ‘ideas are getting harder to find’ explanation. So you don't need to be as concerned about people switching into zero sum or negative sum thinking because if it's cultural, presumably the solution is in our hands.
LUCKEY: Yes, I guess it's not so much the concern about it being zero sum as much as I think that even things that… Here's how it really is. I want things to happen that may not be economically, actually very important or socially even that important. I want humanity to colonise the stars. And I feel like it's very important for people to believe that's a very big deal for humanity. And if it weren't, I actually wouldn't want to talk about that because I simply, for my own personal reasons, I would love to see humanity colonise Mars and go out to the greater universe. Same thing with the metaverse. I want to see it happen. I'm a virtual reality nutter and I believe the metaverse will be a good thing, but I don't want to say anything that will cause it to not happen. I think Peter has a very intelligent view on it, but probably the difference between there is I think Peter's saying it how it is, where I would prefer to say it how I think will cause it to be the way that I want it to be. And so, like, I give Peter a lot of credit for saying what he thinks is true, rather than… I don't think Peter is thinking: “What can I say that will bring about the world that I personally want to see?” And you can probably tell from a lot of the more negative things he said. He's saying these things not with the hope that it fixes things, but that people are informed about the truth. I guess, in that way, I'm a techno-partisan. I'm a techno-politician. I have an agenda, and I'm going to push.
WALKER: Your predictions have an element of advocacy.
LUCKEY: Oh, 100%. And I've talked about this quite a few times where I'm not a journalist, so I have no obligation of neutrality, I have no obligation of looking at the facts in an even handed way. I'm allowed to be a partisan and to focus on the things that I want to focus on and to divert people's attention from the things that I think lead to society going in the wrong direction and I'm totally happy to do that. I'm happy to be unabashedly partisan for the world that I want to see, and of course, that's not necessarily like a right wing or left wing thing. It's more this specific set of things that I want to see in the world.
And I was actually a journalism major, I don't know if you knew, before I dropped out of school to start Oculus, so it's not like I'm ideologically opposed to journalists in general. People think I am sometimes because I'm so mean to them but it's more specific journalists that I hate. It's not the concept of even being a biased journalist. I actually fully understand their point of view.
WALKER: You did journalism for three years before you dropped out.
LUCKEY: That's right.
WALKER: Why did you pick journalism? It seems a bit removed from your more technical passions.
LUCKEY: You know, it's almost embarrassing to talk about, so I rarely do. But it was a matter of arrogance. I decided, when I was 15 and going to school, that I already knew everything that I needed to know about technology and that whatever I didn't know, I was fully capable of learning on my own. I had been very successful self educating myself on electrical engineering, mechanical engineering, a lot of esoteric subjects like solid state lasers, gas lasers, high voltage power systems, and of course, virtual reality. And I said: “You know, I'm going to be able to self educate myself in basically any technical area that I need to.
I don't need to go to a school where the curriculum was designed around things that are last year's technology." And you couldn't really go to places to learn about the things I was most interested in, like cutting edge virtual reality techniques and display technology. And so I looked at that reality and said: “You know, I already know everything I need to know about technology. I'll be able to learn anything I need to know about technology. But what I do need to learn is how to better and more effectively communicate with people”, because that was not my talent. I'm not saying it's my talent today, but certainly it was not when I was a teenager. I was a very technical person. I was one of those kind of turbo nerd, super-autist internet forum megaposter types of people who got along with people who fit into that narrow window of thinking and communication. But it was clear to me that to be successful in the broader world, I needed to get good at distilling ideas that were complex into versions that I could talk about with ordinary people, to be able to persuade people of my opinions, to persuade them of why what I was doing was important. And I thought getting a journalism degree would be the most productive thing that I could learn in school for that aim.
WALKER: I see.
LUCKEY: So… I mean I could have done for example, maybe broadcast journalism, but that's its own flavour. I could have pursued… There's kind of public speaking courses and some people, because they have technical aspiration and then they get masters in business administration so they could run the business side. But for me, I felt like my biggest weakness was communicating effectively with people and I thought that a journalism degree would help me with that.
WALKER: Is there like a systematic process you follow when you're learning a new topic or do you just kind of muddle through?
LUCKEY: I don't know if it's systematic, although I'd say there's like a series of things I typically do just because they've proven to work. I think the first thing I would do is do a review of all the academic literature that people in an area consider to be of importance. So when you know nothing, that's not when you want to be the contrarian. You can't afford to be a contrarian when you know nothing about the field. And so you start by saying: “Well, what's the consensus as to the best work in this area?” Like, let's say virtual reality. First you need to say: “Well, what do the leaders in virtual reality believe is kind of… What are the top 50 papers that they think are really just seminal works that we really need to be paying attention to?” And usually by the time you get through those, you can know something well enough, and of course you check Wikipedia and you see what's going on in the commercial sector. You check what investors are investing in an area, whether it's biotechnology or DNA modification or de-extinction of species. You look at what the academics are doing, look at what investors are investing in and usually by then you know enough to start forming an opinion as to what's important or what might be done differently. And that's my favourite part of the process. Once you know enough to start to feel like you can think differently without just being an idiot. And you want to get to the point where you can talk to a person in the field intelligently enough that you may be able to persuade them of a different way of thinking. And if you show up and you don't even know their lingo, you don't even know what the latest advancements are, they're not going to take you seriously at all. And leads to the final step is actually talking to people, and I don't know if debate is the right word, but going back and forth with people and trying to sharpen your ideas against theirs is often a really good way to figure things out. And that was true even when people disagreed with me, I would talk with people in VR who disagreed with my ideas around what was important and how to accomplish better virtual reality headsets. But going back and forth with someone who… where they'd say: “Here's why your ideas are bad, here's why it doesn't make sense”, it allowed me to sharpen my own ideas until I think it was really crystal clear where I could add value.
WALKER: How do you distinguish between situations where the mainstream consensus is probably correct versus situations where you can actually make a big difference and solve a problem even as a rank amateur?
LUCKEY: Oh, man. You know, I think it's easiest when there's not much mainstream consensus in general. Like, virtual reality was great because I don't think there was a mainstream consensus on it at all. Probably because nobody had ever figured out how to make a viable virtual reality company or product outside of these small niches like data visualisation and military training. Nobody had ever figured it out. So it was much easier to come in with new ideas, and it's much harder to go in when there is consensus from people as to what the right way to do things are. You know, If I'm being honest, I think that my entire career, I've stuck to things where there's not a lot of consensus, and that's made my life a little easier. Like getting in the defence industry. I don't think there was… it seems a little more obvious now, but if you go back six years or even further, there were a lot of people who did not think that defence was important. They thought that the era of large scale conflict was past, that we live at the end of history, that economic entanglement between nations precluded any kind of real violent struggle, and the invasion of Ukraine has made very clear that's not the case. China becoming even more aggressive over the last six years has made clear that's not the case. But I would say when I started Anduril, what I was doing was very much against the public, or I guess the mainstream consensus in the tech industry. What I did with Oculus was against the consensus of the tech industry. So you asked, how do I know when the mainstream is kind of right? I actually don't know. I've never worked on anything where the mainstream really… where the mainstream audience had an opinion that was worth anything, or even more to the point, I don't think that the mainstream had an opinion. There was no mainstream opinion on VR in 2012. What's the television program? The woman says… Was it Mad Men? Where she says: “I think you're a terrible person.” He says: “I never think about you.” And I think that's how it was with virtual reality. People had this idea: “Oh, people thought virtual reality is a failure.” I don't think people thought about it at all.
WALKER: I watched an interview of you where you alluded to the fact that you keep a long list of forgotten technologies.
LUCKEY: I do, I do.
WALKER: What are some forgotten technologies that could be built which you think are very achievable?
LUCKEY: Oh, man. Well, I'm not sure if I want to give away my whole secret list, but some of them are forgotten, others are underinvested. So I'll say one example is I think steam engines have an enormous amount of unexplored potential in the modern day, particularly closed cycle steam engines, where you're condensing your working fluid and you're not… And when I say steam, I'm talking about all supercritical fluids. So carbon dioxide or nitrogen, water, it doesn't really matter. But I think that, for example, there's a very reasonable likelihood that the future of ground vehicles in particular is not actually battery packs. It's actually going to be steam. Because you can recharge a boiler, as it were, with electric power, very quickly, very efficiently. You can fast charge a boiler at rates of many megawatts without any problems, and it doesn't degrade over time. You don't have these chemicals that are wearing out. You don't have these systems that are wearing out. So imagine a closed cycle steam engine car where you've got a closed loop steam engine hooked up to a boiler that you can plug into any fast charger, it dumps electrical energy into it. You have an insulated tank that can store that energy for days or weeks with only minimal losses. People will point out, oh, but you'll lose 5% a day of the heat to losses. Well, guess what? You're going to lose 5% of your battery on a cold day in an electric car, that already exists. But the coolest thing about this is it could be much cheaper to manufacture, actually, than battery packs. It could be, certainly using a lot less rare metals. You can build these systems out of the dirtiest cheapest, lowest purity metals that you want. You don't have to have all this high end lithium, you don't need cobalt, you don't need nickel. So I'm a big fan of steam engines, particularly for cars and for home power storage applications. I think people kind of started ignoring steam because there were cooler things like the internal combustion engine. And then the internal combustion engine got billions of dollars, hundreds of billions of dollars in research and development that has made it dominant as long as we had cheap sources of fuel. And I'm not sure that's going to last.
And then the other side, I'd say, is fuel cells. Are you familiar with fuel cells in general? So it's a very old idea, but this idea of solid oxide fuel cells that consume hydrocarbon fuels even is very old. And I actually think that there's an alternate universe where instead of internal combustion engines, we instead invested in solid oxide fuel cells and basically just got really good at coating and stacking sheets of thin metal to make electricity come out when you flow gas in between them. And that, unfortunately, is not what happened. And so instead, we've seen hundreds of billions go into internal combustion engines. People will point out that solid oxide fuel cells are more expensive and harder to make and don't last as long as normal engines. But I think if they were to receive a similar level investment, you'd actually see them become very viable.
So these are old ideas. Some of them actually predate the internal combustion engine. And I think there's people who are going to go back to the future on these and a lot of other ideas and really make a lot of money and also solve some big problems. But you're right, I have a long list of about 50 things that we can go back to the future on.
WALKER: That's so cool. To return briefly to sci-fi, several years ago, Magic Leap hired Neal Stephenson to be their in-house futurist.
WALKER: Should more tech companies hire sci-fi writers to just sit and speculate on possible technological futures in their space?
LUCKEY: I haven't thought about that. Just off the cuff, I'd say that they'd probably get better returns on making sure that the executives running these companies have read all of the relevant science fiction. For example, I think… You look at Meta, I think that they would get much more value out of forcing all of their product managers to read Snow Crash than they would hiring Neil Stevenson to kind of be this independent advocate of the future inside of the company. And I'm also not sure… You'd say: “Oh well, what if you made Neil Stevenson the head of the VR division?” Well, I think that doesn't necessarily work out either because running these companies requires a different skill set and a different set of knowledge than the ability to ideate on science fiction, which is important, but not the skill set you need to, let's say, manage a team of thousands of people to orient themselves towards actually productive work. So that would be my initial guesses. You can actually get most of the value… If you really believe in the value of science fiction writers, which I do, you should try to get that value without putting them nestled in the company. And I'd say also, there's a lot of product managers at Facebook who definitely have not read Neil Stevenson's works or a lot of other VR works, and that is unfortunate.
WALKER: You used to give all new starters at Oculus a copy of Ready Player One, right?
LUCKEY: Yeah, we did. People point out and say: “Oh, that's so dystopian, because in Ready Player One, things aren't so good in the real world.” To which I say: “You know what? Things aren't so good in our real world either.” And I've talked with Ernie Klein about this, we're friends. He doesn't think that Ready Player One is predicting the future of society per se. He's not saying: “Oh, look, society is the future is terrible and everything's going to be bad.” It's what you need to do when you tell a story. I mentioned this earlier, but you need to have conflict. Nobody wants to read a story about how everything's great. Imagine if Ready Player One was just VR is everywhere and the world is so much better for it. That's an interesting kind of techno utopian vision that people like me would find interesting. But it's not going to become a New York Times bestselling novel. It's not going to get a Steven Spielberg film adaptation. So the things that we wanted people to pay attention to in Ready Player One were these ideas that virtual reality was going to be a huge part of the way people not just played, it wasn't just a game console. It's the way they live, it's the way they work. You had huge portions of the economy taking place in virtual spaces. You had this kind of seamless merging of the real world and the virtual world with people seamlessly going back and forth. And then, of course, a recognition that is a staple of science fiction about sci-fi, which is that the virtual world is just as real as the real world, if you accept that human experience is what defines something being real or not. There are experiences that people had in VR that were as impactful and as important as things that happened in the real world, if not more so. And that's where I see things going. There's another good quote that.. I actually post it every April, so it's coming up again. For the last decade, every April, I post a quote from the light novel series and anime Sword Art Online. And one of the characters asks the main character, Kirito: “What's the difference between the real world and the virtual world anyway?” And he says: “The quantity of data. That's all.” And I've always loved that because it's such a cool idea that boils it down. Yes, obviously, virtual reality as it exists today, it's hard to argue that it's as impactful as the real world. But I hope for a world where all of our communication with people and all of our interactions with people are equally meaningful, whether they happen to take place in a digital space or a real space, whether the photons coming off of their eyes are real or synthetic.
WALKER: How long until we get VR that's virtually indistinguishable from reality?
LUCKEY: First you have to pick the particular… The way you've put it, I would say maybe a year, but I'm being obtuse here. You ask until we get VR that's indistinguishable from reality? What I mean is there will be virtual reality that can simulate certain slices of real world experiences perfectly in the next year or so. So if you were to constrain it to, let's say, an indoor room where the lighting is artificial, not too bright, you don't have too much dynamic range, and the things that you're looking at are all synthetic, highly polished geometric shapes. I'd say we're actually just about there, not in a mainstream mass market product, but in terms of the technology existing and people making these prototypes, we're actually very close.
And if you talk to people who were excited about Magic Leap, pre Magic Leap collapse, it's worth noting that all of them saw this huge, enormous rig that was built, that was meant to show the limits of what augmented reality technology could do to build kind of a perfect synthetic light field of virtual objects overlaid with the real world. Those people were very impressed by it. And it's because when you didn't have these practical limits, the technology already exists to do a more or less perfect job of a limited set of the real world. So put another way to simulate sitting in a room full of geometric cubes. We're right about there. In five or six years, we'll be able to simulate the experience of looking at a person on the other side of the room, and it will be at human visual fidelity. The audio won't be perfect. You won't have touch, you won't have taste, you won't have scent. You won't have the ability to feel like you're moving through the world and feel the wind on your face. But for the experience of sitting in a sealed, fluorescent lit room and looking at things, that's going to be perfect. And so I guess what I'm really getting at here is, virtual reality is just about capable of simulating some set of things perfectly today. And what's going to happen is there's going to be more and more things that it's able to simulate perfectly as time goes on. One of the hardest things is… Imagine the experience of surfing, where you've got a high thermal delta between you and the water. It's pulling heat out of you at a rapid rate. You have wind in your hair and salt in your eyes, and you taste the salt in your mouth, and you also have motion… you have a sense of locomotion borne by your eyes, your skin, your ears, and your inner ear, feeling all of that locomotion. That experience is going to be extremely hard. Getting to the point where you can mimic the sun reflecting at extremely high brightness off of little specular points on the water, that's going to be very hard. That might not even happen in the next 50 years, maybe longer. Maybe we're past the entire concept of visual displays before we actually figure that one out perfectly. But that's the way that I look at it. The good news is the roadmap is clear. I'd say with AI, it was not clear what the breakthrough would be and when it would happen. There were kind of a dozen different paths that were all relatively promising. And then all of a sudden, one of them has started to work recently. With VR, everyone knows what the path is. We just need extraordinarily high resolution optics with very high resolving power. And all you have to do is take the pixels off of that display, collimate them, shoot them on your retina, and we know what we need to build.
WALKER: So, is the path, so to speak, lit only by synthetic photons?
LUCKEY: So that's my belief. You've probably heard me talk about this at some point. I'm not a believer in optically transparent augmented reality systems. I think that they are an interesting thing to do in the short to medium term, while our ability to build perfect VR systems is limited. But at the end of the day, trying to build a system that simultaneously allows for the unhindered passage of real world reflected photons into your eyes while also adding synthetic imagery on top of that is almost an impossible compromise. The things that make it a better digital display make it a worse transparent lens and vice versa. And there's certain parts of this you'll never be able to overcome. You'll always be limited to the dynamic range that you can perceive in the real world. It becomes almost impossible. Augmented reality systems that are optically transparent can only add light to your view of the real world. They cannot subtract light. They cannot erase things. They cannot erase bright objects. They cannot truly put a hard line in front of something else. All they can do is kind of add glowy things in front of the world, which is, I think, going to be very limiting if you think about the metaverse as a place where you're merging the real world and the virtual world, where it's not just adding directions to Starbucks on top of your view. I'm talking about, what if you're merging a real world office space with another real world office space on the other side of the world, also simultaneously merged with a digital space that is only going to exist in this instance for another five minutes and then disappears forever? For that, you have to have fully reprojected augmented in virtual reality where you are using sensors to perceive the real world, building an image and model out of that and then merging it with digital images that you project onto the eyes. Every photon that hits your eyes, in my opinion, should be coming off of a display that's creating a synthetic image for you to view. Now, in the short term, it's very hard to build displays of the kind I'm talking about that are high enough resolution to be something you would want to wear them all day and use them for everyday general purposes. But we're going to get there. It's a matter of single digit years before we do so, I'm very happy that Apple's moving in this direction for example. They were building an optically transparent headset. Originally, this product they were building was more of a developer kit, like, oh, it's a way to get started, it's not optically transparent, but it can do things well enough. Now, I think that Apple, whether it's on purpose or by accident, is going to end up stuck to this path. I don't think it's actually good in the long run for them to try to pivot back to optical transparency. I hope they just stay all in on reprojected AR, or as we used to call it at Oculus, Hard AR.
WALKER: So what's your concrete vision for a human future that includes the metaverse? Are we all just plugged into some kind of Robert Nozick hedonistic experience machine that distracts us from reality? Something like in Ready Player One. Concretely, what does your future look like?
LUCKEY: I think that — People will point out that virtual reality could bring about a future where people don't work and they are just entertaining themselves, and that's all that they do — I think that we're actually already sufficiently technologically advanced to do that. If such a world is going to come to pass, I don't think that higher levels of fidelity and more fidelity and human to human communication… which I think is what VR is. We've been moving away from face to face communication as technologies advanced. I feel like VR actually brings it back to the fundamentals, which is good. It's actually more human than basically any form of communication except actual face to face when implemented to its limit. I think we already live in a world where you can have people who sit in the pod and they smoke their weed and they eat their fatty foods and they watch their Netflix. If society moves in that direction, then that's fully possible today. And so I think that virtual reality is not going to be the thing that causes that to pass. I don't think virtual reality is going to make people less ambitious in and of itself. I don't think it's going to make people want human connection and human productivity less than anything else. I'm concerned about what you're talking about, but I don't think virtual reality is the thing that's going to bring it to pass.
So what's my concrete vision? I talked about it a little bit already, but I hope that the virtual world is a place that is seen as equally important to the physical world. I don't say the real world anymore because I'm getting ahead of it, that's going to end up being the new bias. It's going to be the new microaggression to call it the real world. You heard it here first. But the virtual world versus the physical world, I want it to be as real, I want it to be seen as equivalent or better in certain ways, just like the real world probably will be in our lifetimes. And I'd love to be a place where we can do things that would be too dangerous, not cost effective, or too otherwise destructive to do in the real world.
I think that there's a lot of things that people want to experience that they will never be able to experience simply because it's too expensive for every person in the world to experience it. And I think virtual reality is going to be a biggest deal, not to people like you or me who live in kind of first world democracies where everything's pretty great. I think people who live in places where their country doesn't have the natural resources or the economic resources to, let's say, send every kid on a field trip to visit all of these historical sites, or a place where every person can afford to, on the weekend, go and do whatever hobby of choice they want. Those are the people who are probably going to see the biggest increase in quality of life. It's not you or me. And people will say: “Well, I'll never go to Paris in VR. I'd rather just go for real.” I’d say: “Okay, that's great for you, but there's another 7 billion people in the world for whom that's not a practical reality.” And so we're not trying to necessarily make things perfect, but we can make them better.
WALKER: In a post AGI world, does VR become how we spend most of our time? Or do you expect an AGI to invent new and better mediums of communication and experience?
LUCKEY: You know, I have disagreements with people on this. I think I used to believe and still believe that virtual reality in one form or another is not just the next form of computing and communication, but the final form. It's the ability to convey to a person any experience, anything that they could see or experience themselves, or that any other person could conceive, or that any AI could conceive. Perfect VR system should be able to present that to them. The only thing that I think goes further are things that fundamentally change our human nature. For example, what if you could build a hive mind? What if you could have telepathy where you have people who are literally sharing thoughts and operating as a Borg like collective. That is certainly a potential thing that could happen in the future. I'm not super excited about that. I think it's dangerous to meddle with human nature like that. Humans are not perfect, but we've gotten along for thousands of years in the way that we've gotten along and to really tamper with that at a fundamental level is maybe playing with things I'm uncomfortable with. But who knows? Maybe in 100 years they're going to be lamenting how Grandpa won't join the Borg and he needs to just get with the times and join one of the three global collectives. And there's plenty of choice for everybody between the three. But I think VR is the final platform, absent things that change the nature of humanity, and I also include in that, like, uploading to a machine consciousness, I'd say that's also fundamentally, probably changing the nature of humanity. But you could also argue that uploading into a simulation really is VR too. So even that is VR, I guess.
WALKER: Okay, I have some quick-fire questions about Oculus.
LUCKEY: Let's do it.
WALKER: And then I want to move on to general lessons that you've learned from starting companies. So can you describe what the actual breakthrough was with Oculus in terms of using software to solve the predistortion of an image?
LUCKEY: Sure. I mean, we had quite a few really good ideas, but I'd say one of the key ones is the one you touched on, which was deciding that instead of building a complex optical system that was expensive and heavy, that presented a perfectly crystal clear image from the display that it was working with, which is the way most VR headsets work. They took a nice rectangular display, they collimated the light coming off of it, reprojected it to a wider field of view, and did so with minimal distortion so that all the lines are straight. Your view of the display was unimpeded, but it cost a lot of money and it weighed a lot. What I decided to do was instead design a lens that only optimised for the things that could not be done in software. So wide field of view, low cost, lightweight, those are things that are physical properties that cannot be corrected for. And then all of the things that I sacrificed to get there. So my lens had high amounts of geometric distortion, high amounts of chromatic aberration, so colour separation between different channels. I realised those were things you could simply pre distort in real time on a graphics card using a shader. In other words, you could understand the distortion transform created by your lens on the image, and then you could apply an inverse transform to the image so that you would basically render a very distorted image that looked terrible. But then when you viewed it through my equally terrible lens, it would all revert back to being straight lines, correct colours, et cetera. And that was not a brand new idea. People had come up with it as early as the 1980’s. NASA had a program where they were working with leap optics, the Jet Propulsion Laboratories. And they actually, in their paper, specifically noted: “These lenses have quite a bit of distortion and we did write a piece of software that's able to correct for this distortion by pre distorting. However, you actually spend more resources on your computer rendering that transform than you do on the rest of the simulation. Therefore, it's an impractical technique, cannot be run in real time, and therefore we just run with the distortion and we accept it.”
My lens had even more distortion than those optics, so I didn't have the option of not correcting it. But the thing that had changed is graphics cards had become extremely performant, very cheap. And so maybe in the 1980’s, it required most of the computational power of a high end computer to perform this inverse transform. By the time I was there, it was a fraction of a percent of the computational resources of a graphics card. So the graphics cards getting much better was really important.
I'd say the other really big thing that we did right was… We did a lot of things that were just basic. We did good sensor fusion algorithms, we did good motion sensing in general. But we also did something that was probably even more key than the straight up tech advances. We built a software development kit that made it very easy for any game developer to work with a virtual reality headset. Prior to Oculus, if you wanted to work in VR, you had to understand VR very deeply. You had to understand how to render stereoscopic wide field of view images. You had to understand how to minimise latency in a rendering pipeline. You had to understand how to integrate with a motion tracker and how to correct for drift in that motion tracker. And you basically had to be a virtual reality expert. And that was what a lot of these academics were and military researchers were. But if you were a normal person, you would never have any hope of affording or building a virtual reality game. What we did is built an SDK that took all the complexity and buried it in a few lines of simple code that anybody could put into their game. And then we also integrated with the major game engines, which is a time where Unity and Unreal Engine. And we made it so that in fact, someone could have literally zero coding experience and if they were able to make a world in unity, they could click a button, click export to VR, and they were now viewing it in their Oculus Rift. And that's what made Oculus successful. It's that we sold tens of thousands of developer kits that allowed tens of thousands of people to experiment with building VR games that never could have built VR games before and honestly wouldn't have been able to get a development kit even for the next Xbox or the next PlayStation. We sold dev kits to anyone and everyone, and that's why we saw such massive creativity in our content space that you weren't seeing in the console space, where dev kits only went out to developers at major companies.
WALKER: In the summer of 2011, as an 18 year old, you landed a part time gig with Mark Bolas lab in the university of Southern California. How did you convince Mark to hire you? Were your projects just so impressive that they spoke for themselves, or did you pitch yourself in a specific way?
LUCKEY: Well, I got to talk myself down a little bit. It's true that I did get myself hired, but I was working part time as a lab technician. So I want to be clear, it's not like I convinced Mark or anyone else to make this crazy bet and put me in charge of a major program. I was a lab technician, and I was a cable monkey, and I was working on the low cost VR design team. And look, I don't want to downplay the importance of that work, but it's worth noting I was low man on the totem pole. That said, the first time I reached out to Mark was actually with two things I wanted to talk to him about. One was he had previously, in the 1990’s, run a high end virtual reality device company called Fakespace labs. And I had managed to obtain a virtual reality device that cost about $90,000 back in the 90’s, a very high end virtual reality display. And I had purchased one for less than $100 at a hospital equipment auction. It had been used for some kind of high end data visualisation, and they'd had gone into a closet for 20 years, and then they were just selling off all their old, obsolete equipment. I managed to buy it for $100. There was a particular part in it. It was a colour display, but the field sequential colour generation module in it that basically allowed it to generate colour signal from the input that drove the LCD shutter that basically gave it colour, was fried or otherwise not working. Maybe that's why they took it out of commission. And so I reached out to Mark asking if he had either replacement parts or any information on how I could re implement the field sequential colour generation unit hooked up to the LCD colour filters.
And then my second bit was I had read some of the academic papers that he had been one of the authors on, and I actually had some critiques for an implementation of a VR system that they had done with two iPhones that were rendering right eye and left eye for a mobile virtuality system. And I laid out to him how I would have built such a system and explained that I would have used one iPhone and then they would have been able to use one motion tracker. They wouldn't have had the mismatch between tracking on both phones, they wouldn't have had mismatch in rendering times, they wouldn't have different screen tearing attributes. I said you should have used one phone and then used the video output and then put it to this particular display that I've been using in my virtual reality headsets, a BOE Hidist display that was higher resolution than an iPhone as well. And you should have just output to that and you could have built a self contained unit with one phone that would have been higher resolution, wider field of view, and not had some of these synchronisation problems. And that was what got me in the door. So I ended up meeting with Mark and some of the people there and they were also impressed with the fact that I knew about all their old VR equipment.
I knew as much about a lot of it as they did, and that was pretty weird because I was a teenager and these were older gentlemen who had kind of worked in the VR industry in the 80’s and 90’s. We got along really well, so I think that was how I managed to get my foot in the door. Also, not a lot of VR jobs available back in 2010, 2011, so it was lucky that I did.
WALKER: What were the big lessons you learned working with John Carmack?
LUCKEY: Oh, man. Well, John Carmack, I have to start by saying people are not familiar with him. He's probably one of the best five programmers in the world. A giant of the computer science field. He basically invented 3D game engines. He invented the first person shooter. He made Wolfenstein, Doom, Quake, and many other titles. We managed to steal him away from his own company that he had been at for 30 years and hire him away to be the CTO of Oculus, which was one of my best hiring achievements ever. I learned quite a bit from John, although I'd say a lot of the things that people would talk about learning, like: “Oh, he focuses on what really matters, and he makes sure that he's obsessed with delivering customer value rather than corporate value.” A lot of those are the reasons that I wanted to hire him in the first place. So I don't know if I learned them from him so much as just always respected it. One of the things that John has always done that I've really appreciated is open sourcing his old work. So when they would release a new version of a game, they would open source the previous version and allow everyone to learn from the code and build on top of it. And that was really a great thing for the games industry, and it was something that we copied at Oculus. So we actually open sourced the Oculus Rift DK1. We open sourced the Oculus Rift DK2. Every time we released a new headset, we would open source the software and the hardware for the old headset to allow people to learn from it and build on top of it, which I think was good for the VR industry. Unfortunately, Facebook stopped doing that after they fired me, but in that time it was a really good thing for the industry. And I think I learned a lot about the importance of that from John. But I could go on forever about the things I learned from John.
WALKER: Some questions about starting companies generally. What's the best, or most interesting lesson you've received from Peter Thiel, either explicitly or just from observing him operate, that we can't find in Zero To One? And it could be really specific and tactical. In fact, even better if it is.
LUCKEY: I think one of the tricky things is here, Peter's done a pretty good job of distilling the things that are broadly useful into his external communications, which is rare. There's a lot of people out there who are as smart as Peter who do not see fit to share their insights. That's probably the reason Peter is such a force in the startup world, is his willingness to kind of throw all of his secrets for success out there. I'm trying to think of something that's not in Zero to One or some of the associated stuff. I would say probably one of the things that I've learned from and really internalised from Peter is, and this unfortunately, is somewhat in Zero to One. He's talked about how companies should try to build monopolies. And I'd say the way that I've looked at some of this is to look at industries where… I take it further than just “you should try to build a very strong business and you should be able to build a monopoly.” But I've looked at areas where you could build a monopoly and where you kind of punch above your weight. One of the reasons I did so well with Oculus is that I was working in an area where honestly, there weren't all that many smart people looking at it. It was kind of a backwater. And I don't think I would have done as well if I had, let's say — people love to compliment me on my technical prowess and that's very kind of them. — But if I had jumped into, let's say, the enterprise SaaS world, which is highly competitive and a lot of money, I don't think I would have done very well because it was already so hyper competitive. I think even in the defence space, if every tech company in the West was working with the defence apparatus, I don't think that Anduril or myself would have been able to be as successful as we have. I don't think we would have been able to kind of punch above our weight, as it were. And so I'd say a really tactical thing that people should take away that I've kind of learned from Peter, is to look for areas where it is plausible that you could have a monopoly in the first place, and that is where you're going to be able to do your best work. It's not to follow the things you're passionate about. People say that: “Oh, do what you're passionate about.” Well, what happens if you're passionate about something that everyone else is passionate about? What you're doing is dooming yourself to a lifetime of working extremely hard to try and be on the good side of a curve that you probably just aren't on the good side of. I encourage people to look for areas where it's feasible that they could have a monopoly, even if they aren't the one in 1000 genius that we all wish we were.
WALKER: If starting a second company is so much easier than the first time, given that you have access to a bigger network of talent and capital — and you can reject that premise if you want — why don't more one-time founders start two or more companies?
LUCKEY: This is a good question. When I was starting, Anduril… I'd say that the premise is generally true, it's easier to start a second company. I'm not sure that it's easier to get a second company to the same level of success as the first. It's actually very rare for ex-unicorn founders to start a second unicorn. It almost never happens. And in fact, even the examples people will point to, like Elon Musk, it's worth noting, most of his companies, he didn't found them, he actually bought them, or acquired them, or took them over. And I know that I'm not downplaying his contributions, but it's not like… People say: “Oh well, Elon founded SpaceX and Tesla.” I say: “Well, he founded SpaceX, but actually he basically took over Tesla.” It had already existed, and they already had a roadmap and existing prototypes and everything. So it's very rare for people to start a second company that does as well as their first. And there's all these theories for why. I think when I was raising for Anduril, there were a lot of investors who are sceptical and pointed out: “Palmer, there's very few second time founders who have it in them to really grind it out and really just fight and put all the hours in to do what they did the first time because they've already got the money and they've already got the recognition. They just don't have the innate motivation that they did the first time around.”What I pointed out to them is that might be true for many ex unicorn founders, but most of them retired by choice, decided to take time off, and then they got bored and wanted to get back into the game. In my case, I was fired. And so I pointed out that I was kind of ripped out of my element at the pinnacle of my career, and I was nowhere close to being bored or done with building businesses. And that was a pretty convincing argument. If you look at it this way, basically every billionaire who has been fired from the unicorn they started has gone on to do great things. Steve Jobs did pretty well, and they said: “Oh, yeah, that's a really good point.”
What I've seen is there's a lot of people who start a company, it does really well, and they eventually move on from that. And they end up having so much fun cruising the Mediterranean in their yachts and racing vintage air cooled Porsches and spending time with their families, which are important to them, that they don't necessarily want to go through the grind that is a successful company. I don't want to make it sound like it's hard or bad to run a successful company. It's obviously much more demotivating to run a company that's not going anywhere, but a company that's in hyper growth mode, it's very stressful to run. It's probably one of the most stressful things you can voluntarily sign up to do. And I think a lot of people, they just don't have it in them to do it again. And I don't blame them, I think that's actually probably the smart move. What I'm doing is extremely unhealthy. I am destroying myself, body, mind and spirit. And I will probably destroy myself in the long run because of what I'm doing, but that's okay for now.
WALKER: Your story has clear echoes of Byrne Hobart's theory. Byrne Hobart, the finance blogger. Have you heard it?
LUCKEY: I've heard the name, but I'm not very familiar.
WALKER: So he has this essay called ‘Where Do Business Mafias Come From?’ and he considers why was the PayPal mafia so prolific and successful? The same for Tiger. And his answer is, well, that's…
LUCKEY: Well [Tiger]’s a group of people who did have the wherewithal to go out and do a lot of new things too.
WALKER: Exactly. And his answer is that they exited but too early. So I'll quote him: “Both PayPal and Tiger exited with around 200 employees, most of whom were young and rich enough to start their own companies, but not rich enough to retire. They clearly weren't losers, but they arguably weren't yet winners either. That's a good combination. The means, motive, and opportunity to come back for round two.”
LUCKEY: Yeah, I totally agree with that. And I think especially the sense of also feeling like you have something to prove is very powerful. I think if I had been fired today from Oculus, where they've sold 20 million Oculus Quests, I think I might have not felt such a strong motivation. But six years ago it was six plus years ago, actually. Six years and two days now. Six years and two days ago, I very much felt like I needed to prove everybody wrong. I was going to show them that I was somebody, that I wasn't a one hit wonder, that they were all wrong for writing me off and that I was going to prove that was the worst mistake that they had ever made. And that's my own personal motivation outside of why I really started Anduril but it's been a powerful one. It's good to wake up in the morning for a few years and say: “They'll see. They'll all see.” There's that scene from The Office which maybe you've seen… Actually in Australia, do you guys watch the British version of The Office or the American version?
WALKER: A little bit of both. Mainly the British.
LUCKEY: Okay, that just occurred to me. Well, in the American one, there's a great scene where Ryan says: “Every day I'm writing my list of people who have wronged me so that someday they'll recognise that they were wrong.” And I'm definitely Ryan in many cases, I've got that list of people who have wronged me. They'll see. They'll all see.
WALKER: Right beside the long list of forgotten technologies.
LUCKEY: I've actually got quite a few lists of various things.
WALKER: Yeah. Have you seen Extras?
LUCKEY: No, I haven't.
WALKER: There's a good scene about writing lists.
LUCKEY: You'll have to send it to me.
WALKER: I will. As someone who straddled both the world of atoms and the world of bits to invoke Peter's distinction, do you see any fundamental differences between atoms-founders and bits-founders? Are there different archetypes?
LUCKEY: It's hard because I guess initially I would pick things that I think are more typical of a bits-founder than an atoms-founder. But at the same time, I've seen people of both archetypes do either side. I mean, I consider myself more of a hardware guy, but if you look at my past two companies, Oculus and Anduril, both of them, interestingly, had about twice as many people working on software as hardware. And that's because a lot of the value of what we built, while it kind of manifests in a hardware product, was actually created on the software side. So there's a very strong argument that I'm a software founder, that I'm primarily a guy who runs software teams and software companies. I wouldn't agree with that argument, mostly because my sense of self worth is tied to me being an atoms-founder and building real physical things. But that's my own issue. I'd say probably the characteristic that I hope to see in people is to not be too religious about whether they're hardware or software companies. If you think of yourself as a hardware person but you can build better value for your customer by focusing on software, you have to get over your ego and start hiring more software people instead of trying to win on the hardware side. And vice versa if you're a software person but you can do better for your customer by getting into hardware and by investing more in hardware, you have to do that as well. You can't afford to do the thing that you like or that you're good at doing.
WALKER: Yeah. Let's talk about defence.
LUCKEY: Let's do it.
WALKER: To the extent that the UFO sightings the Pentagon has disclosed are sightings of vehicles with capability beyond our current arsenal, how likely is it that the US Government is trying to reverse engineer that technology? Because it would strike me as kind of baffling that there's not some kind of Manhattan Project going on around it.
LUCKEY: I'd say to the extent that they're reverse engineering it and who knows what the extent of that is?
WALKER: How likely is that?
LUCKEY: Well, it's 100% that they're reverse engineering it to the extent that it's possible. For example, even some of these characteristics that we've observed, like the fact that you have movement through space without creating typical effects like sonic booms or massive distortion in the air, or plasma sheaths where you're superheating the air on the front of your vehicle. These were kind of considered just the inescapable side effects of something that moves that fast in atmosphere. And so I think even just the fact that something of some kind appears to be doing that has opened a lot of minds in terms of thinking, well, what would it mean to build a craft that doesn't have those? How could you do such a thing? And if there's no examples of it, then it's kind of hard to go down that rabbit hole fully. But if you have an example to look at, it kind of forces you to think about it differently. And I'd say, like, VR was kind of similar, where it's not even that people necessarily were copying all of our techniques on the rendering or the latency side, but people saw that you could get latency as low that we did. And literally just the fact that somebody had done it somehow was proof that they could get there through some means of their own. And so you had Oculus and several other companies building totally different tracking systems, totally different rendering pipelines, but people were inspired by the fact that we had proved it was even possible to get there. I'd say that to that extent, there is already reverse engineering going on, trying to figure out what can we learn from the fact that these things exist at all? How could that guide our research and development? Why put it into paths that don't lead to potentially those outcomes? Now is there a crashed spacecraft somewhere being worked on? I don't know. My personal theory is that it's less likely that this is extraterrestrial in origin and more likely that it's actually from Earth in some form. Now, that doesn't necessarily mean, I think, it's a Chinese weapon or a Russian weapon. I'm going to sound a little crazy here, but you always sound crazy talking about unidentified aerial phenomenon. But for example, I think that it is actually more likely that there is a remnant human or adjacent to human supercivilization remaining in hiding from hundreds of thousands of years ago than that they came from another star. I think it's more likely even that perhaps it's people who have moved from an adjacent dimension or who have come from our own future than necessarily someone who has come all the way across the stars specifically to us. I don't know if it's likely to be any of those things, but it could also be something along those lines equally outrageous, but so abstract that I can't even conceive of what it is because we haven't even thought of what it is yet. But I suspect it's something like that and not actual aliens.
WALKER: Yeah. The universe is stranger than we can think.
LUCKEY: Have you read Michael Crichton’s book Sphere?
LUCKEY: Oh, man. Well, all right. If anyone hasn't read Sphere — or is it The Sphere — Sphere or The Sphere, I don't want to ruin it for you, so if you want to read it, just stop listening for the next 30 seconds. In it, there is an enormous huge sphere shaped… no, it's not... Well, there's a huge craft discovered in deep ocean waters, and the United States Navy launches a saturation diving team to very secretly figure out what this thing is, and it appears to be a spacecraft. And they get to it, and it is very obviously a spacecraft that's been there for thousands of years. They say: “Oh, my God, this is incredible. Aliens have visited Earth. This absolutely changes everything.” And then the divers at one point, months into the operation, are scraping off marine growth from the hull. And then they discover, etched into the hull: “United States Enterprise”. And they realise that in fact, this ship is from the distant future of Earth. The United States is still the dominant superpower, and they must have accidentally gone back thousands of years in the past and crash landed. I think that's actually a more viable explanation than many.
WALKER: So speaking of the United States as the dominant superpower, if we fast forward ten years and American hegemony over much of the world has ended, what do you think the most likely explanation for that will be?
LUCKEY: Ten years? Well, that rules a bunch of things out. Probably collapse of our financial system. We've already talked about how risky some of this is, but ten years is not enough time for a lot of things to play out. Like, I don't think in ten years, for example, we're going to have our military wholly, completely outclassed. I think it's not quite enough time. And the US is doing a lot of things right. It's probably not a culture loss, right? We don't lose our soft power because the United States, everyone loves our movies, they love our music, they love our food, and we're also really good at getting along with all of our super friends around the world. So it's probably not a cultural collapse. We would probably be an economic collapse would be the biggest risk. That was the plot of a game called Home Front that Ubisoft released, I guess, twelve years from now, in 2011. It was that the United States suffered a massive, overwhelming economic collapse, and then China and North Korea teamed up to leverage that financial collapse of the United States to perform an invasion of the US Homeland. And then they basically invade the West Coast and the East Coast at the same time and managed to show up at a point when we just don't have our s*** together and everything falls to pieces. So I'd say not that's likely to happen, but definitely if there is going to be a ten year collapse, it would probably be on the economic side.
WALKER: Yeah. Okay. That's interesting, that notion of China launching a knockout blow kind of first strike. So in his book The Future of War…
LUCKEY: I think they also launch an EMP. So we're in economic collapse, then they launch an EMP. It wipes out our grid, it wipes out a lot of our advanced weapons, and then they invade.
WALKER: Okay. So in his book The Future of War, Lawrence Freedman argues that English and American futurology has tended to obsess over the concept of a knockout blow. This kind of goes back to The Battle of Dorking in 1871, partly because Western thinking about war had long been dominated by the classical model of winning a decisive battle. Are you aware of China's equivalent mental models of war?
LUCKEY: I mean, I'm aware of what people say their models of war are, but I think if you look at their political structure, it is fundamentally very long term oriented. And I don't think that's just hype. I think that their political structure is oriented in a way where one of the nice things I'm not arguing for nepotism here, but one of the nice things about nepotism and about having the political elite be kind of this paternal lineage where it remains the same families in positions of power for long periods of time, as in China, is that it does motivate people to think in terms of the long run. They're not thinking, what can I do now in my moment in power before I hand it off to in abstract, the rest of the country? It takes advantage of that instinct of what can I do to make things better for my direct successors and my children to then continue on this path that we have strategically? And I think China has been taking advantage of that. Their political system is very…
LUCKEY: Exactly. And I think that actually has been an advantage for them. It allows them to think not about a knockout blow within the term of one politician or one general, but instead, how do we really get into a better position over the course of 100 years or 1000 years? And I think that's guiding a lot of their military thinking in ways that are just kind of unthinkable in the United States.
WALKER: Which risk is bigger: AI misalignment or China getting AGI first?
LUCKEY: I would say definitely China getting AGI or really any country. I've often said that I'm more terrified of bad people with bad AI than really good AI. I think you're more likely to see rogue nations or hostile nations using limited, very limited even like at the level it exists today, artificial intelligence to create autonomous weapons and bioweapons and things like that. Really bad people using primitive AI I think is actually a much bigger threat to the United States than this more hypothetical of… it's a new AI and it's super intelligent and it decides it's going to wipe us out. Obviously that is a risk, but it's just, in my list of risks, it's far down the list.
WALKER: But to what extent do we actually need to solve the alignment problem to beat China in the AI race in the first place? Because the American public won't accept the deployment of unaligned AI.
LUCKEY: I mean, it's a little bit too late. I would say that the success of Large Language Models was like putting up… I talked earlier about Unidentified Aerial Phenomenon and Oculus where proving that it was possible was enough to make everyone else divert resources to trying to copy it. I think that the success of Large Language Models like ChatGPT in a highly public, well recognized way has made it impossible for politicians even in China to ignore. And it's now clear that you can do at least that well, and therefore probably much better. And so I'd say there's almost nothing that we can do to slow them down at this point. The thing that might have slowed them before was just they believed AI would be important, but it was in a more abstract way: “Oh, it'll do things of some kind, I'm sure it's going to be important.” But now that you have politicians using ChatGPT typing stuff in, they can intuitively understand that it's a huge deal. And I'd say the flare has already gone up.
WALKER: If the US gets demonstrably closer to creating AGI, does that make it more likely that China invades Taiwan, to deprive the US of GPUs?
LUCKEY: I'm not so sure about that. Maybe. I could imagine that happening. I just think China's ambitions with Taiwan are so domestically focused that's going to dominate their thinking. And of course they're thinking about the influence on the US and the rest of the world. But let me put it another way. If Taiwan was making nothing but dirt, China would still be very interested in taking over Taiwan. It's a fluke of history that Taiwan happened to pick probably the one single most important thing in the entire world they could have picked, high end semiconductors to invest in, subsidise and then build. Like, what an incredible bet that played out there, right? For a centrally planned government focus like that to play out correctly is pretty rare in history. But they've made themselves basically so valuable that the entire West cares about protecting this little tiny strip of land in probably the most politically contentious, most dangerous area of the world that we would otherwise never even consider armed action in. And so I would say maybe they'll go to Taiwan to deprive us of GPUs, but I think that primarily they are internally focused by the fact that they see Taiwan as a rogue government, a rogue state that needs to be brought to heal so that they can achieve a unified China and fulfil the heavenly mandate.
WALKER: Yeah. To what extent was that bet on chips the result of just luck or contingency versus being planned?
LUCKEY: I mean, it was everything like any big revolution, there was a lot of luck involved that it worked out because other countries made similar bets, but they just didn't have quite the same level of success. Like, I would say South Korea definitely had a huge bet on this, and South Korea is pretty good at making semiconductors. They do make some high end stuff. They just didn't quite get to be as good as Taiwan did at making these high end, very advanced process semiconductors. Japan also tried. So it was a matter of I think there was some luck involved, there was some grit involved. There was the government being willing to kind of go in, all in on doubling down on this advantage that it had emerged. But, man, what a smart one. I mean, is there anything that Taiwan could be making that we would care about more than high end semiconductors? Agriculturally I don't think there's anything that they could do. Culturally, I'm not sure there's much they could do. Maybe if they had been like, the Hollywood of Asia and making half of the media that we watch, maybe that would have been equally powerful, but that's quite a long shot. Everyone's trying to become the next Hollywood, right? Canada is trying to do it. Australia is trying to do it. Everyone has tax credits for trying to promote filmmaking. But, man, what an incredible achievement for Taiwan.
WALKER: Yeah, kudos. If you could wave a magic wand and have the US Military use advanced software in the way you want, how much more effective could it be? Like, what's the force multiplier?
LUCKEY: Well, it depends on how you measure it, but by certain measures, ten times more effective, because the power of AI is not necessarily the ability to do things that you couldn't do before. That's sometimes the case. Like if you're building autonomous submarines that previously you couldn't fit a person into them, and you didn't have communication so you kind of need AI on board. But a lot of times it's doing the same things you would have done before, but being able to make better decisions faster. So a decision that would have taken you days of collecting intelligence and having people look and correlate different dots and deciding what it means now could potentially be decided to an equal level of fidelity and sureness in literally one minute. And so in that sense, to make something happen a thousand times faster, that can be a really big deal. Maybe you don't need your airplanes to be twice as fast, maybe you don't need to go twice as far if you're able to make all of your decisions at an equal level of quality or higher in half of the time. And I'd say that there are huge gains applying AI even to legacy existing weapons systems. And I would take a look at what we did during the War on Terror, where we were working with a coalition of different countries around the world. And a lot of our decisions were through many layers of meetings, phone calls, PowerPoints, text messages. And you would make maybe 1000 important decisions about what was going to happen in these wars over the course of a year. What we need to do is use AI so that when there's a fight in the Pacific, we can make 1000 equally important decisions in the course of an afternoon, because that's the pace at which the fight is going to happen.
WALKER: Is there a way in which this causes adversaries to underestimate US capabilities and therefore increases the likelihood of a miscalculation and war? Because obviously software has leverage over the kinetic world, but a lot of it is not observable. It's not really something you can show off in a military parade.
LUCKEY: It's an interesting question. I think the obvious examples are going to be the ones like robotic submarines, where it's just they're going to see that and say: “Oh, I understand how AI makes that possible.” The rest is probably going to be the same way. We show off a lot of our capabilities through military exercises. If we can show combined joint all domain command and control across five nations, all working where every sensor is a sensor for every effector spread across five different nations, and we're making decisions about what to strike in matters of seconds that previously would have taken hours, I think that is something we can demonstrate. Obviously China keeps an eye on those and that's one of the reasons that we have these exercises. So I think the real point is we have to make sure they are aware of these capabilities to some extent. It's good to have a few aces up your sleeve, but you don't want to hide all of your advanced technology development and then leave China more confident that they're going to be able to win and then end up in a fight that didn't need to happen in the first place.
WALKER: If the US Military moves from like a cost plus model to a product model, what do you think the stable equilibrium is in terms of the number of players? Could we go back to something like at the end of the Cold War when there were 107 major defence companies?
LUCKEY: I think you probably don't end up there only because there are certain institutional advantages that you have. I think large companies have advantages even just in purely financial ways. Like, there's a reason that the common private equity model is: “Okay, I've got 20 construction companies, I'm going to go find a profitable construction company and I'm going to fire all of their accountants and salespeople and lawyers and et cetera. I'm going to integrate them with my large organisation and boom, they've gone from being a profitable company with lots of overhead to an even more profitable company with much less overhead because of shared resources.” I think the same idea applies to defence. For example, if a company like Anduril can reuse a lot of the autonomy and communications and even like actuator modules we've built on certain aircrafts or programs, if we can reuse those in future products, we're going to be much more competitive than a company that doesn't already have all of those things to learn from. We're going to be more competitive than people who don't necessarily have a deep bench of talent to pull from and tell you what not to do to make sure you don't screw up a program. So I don't think it's going to be a huge number of small companies, but I bet it's a few dozen. I could imagine it being 12 or 24 companies that have the… What is it today? It's something like five companies have 80% of the spending. I bet it's probably more like one or two dozen will have 80% of the spending, which is very healthy in comparison to what we have today. That does, of course, put a cap on where Anduril can get to. I've long talked about this, but I think it's more or less impossible for Anduril to ever, even in theory, get to above maybe $100 billion in revenue a year. And that's an upper limit. The largest defence company in the world, Lockheed Martin, they have $67 billion a year in revenue. I feel like maybe Anduril does really well internationally and really well in the US and we end up capturing a huge fraction of the defence budget. But at some point, the governments are going to say, we want to have competition. We want to have multiple vendors. We don't want to have one mono defence company.
And so you get to a certain point and I think the government will say, we're literally not going to give you contracts because we've given you a large enough cut. And I look forward to the day when we lose a contract where we should have won on the merits, but instead, we are just too big. One of the things that Facebook taught me is to set goals for yourself that are so big that they'll get you hauled in front of US Congress for antitrust concerns, because if you can get there, then you're doing really well. Howard Hughes managed to pull that off, too. So I'd say I hope we get there someday.
WALKER: So US defense spending is currently about 3.5% of GDP. If the military switched from a cost plus model to a product model, how much lower could you get that?
LUCKEY: I'd say it wouldn't just be the switching from a cost plus model that would help. It would be the incentive alignment that would then occur, which would result, I think, in much lower cost for taxpayers and government. I don't think that it would be as much of an impact to GDP as people think, only because, yeah, that's 3.5% of our GDP but much of that today is in the form of manpower and facilities and a whole bunch of things that are not necessarily run on a cost plus basis. So you're only impacting kind of procurement of new technologies and systems. But I think that you're talking about a gain certainly in the tens of billions to low hundreds of billions of dollars. Like, I could imagine a world where you're saving $200 billion a year in the US. That's pretty good because…
WALKER: They spend about 700 billion a year.
LUCKEY: They do, but again, a lot of that is on things that cost plus contracting one way or another doesn't impact. For example, we got to pay a lot of soldiers, we pay a lot of money for medical treatment of veterans, and maybe, for example, getting off cost plus contracts could reduce the cost of keeping veterans healthy a bit, but it's not going to be these radical reductions in cost you see on the new technology procurement side. But hey, there's not that many companies that could claim that they're going to save taxpayers $200 billion a year. So there's plenty of other things we could be doing with that money.
WALKER: Yeah, for sure. So in the late 1980’s, during the final years of the Cold War, defence spending was around 6% of GDP. Is there a way in which you could view the success of Silicon Valley in the last few decades as the peace dividend of paring back defence spending by freeing up resources and talent to be redirected to the private sector?
LUCKEY: I've heard this theory. I think to some degree it's true. On the other hand, I think that a lot of the technologies that have really underpinned all this stuff, they came directly out of the military. And usually a peace dividend is not necessarily a limited time thing. It's not a one time harvest. People see it as this ongoing great thing about the money that was going into defence, now going into the private sector. I'd say it's more like a harvest than a dividend. A dividend goes forever. In this case, you had created all these military technologies, and then there was this windfall where all those technologies were kind of seized and taken by private companies and turned into this economic explosion, creating things like the internet and GPS and a lot of the modern sensors that underpin our stuff microprocessor technology. But I don't think that actually goes on forever. I'm not an economist, but my theory is there's a big harvest, diminishing returns, and then probably you don't get the same return on that continuously until you probably have another cycle of massive government invention of new technology. Like, you can imagine a world where it's government investment that really makes quantum communication a powerful tool, where we can communicate without any signal spooky action at a distance, flipping q-bits back and forth across the planet. I could imagine a world where that is the result of a massive military spending push, and then we reap this dividend or harvest when the private companies take it and commercialise it down the road after it's no longer a controlled weapons technology. So that's kind of where I see things going.
WALKER: Final three questions.
LUCKEY: All right.
WALKER: How plausible is the sci-fi concept of uplift, and how do you envision its role, if any, in the context of future warfare?
LUCKEY: Well uplift for people who aren't familiar is this concept of taking a species that is not sentient, or perhaps on the edge of what we would consider human levels of sentience and then dragging it across that boundary. So making animals smarter is one way of looking at it, but it's specifically it's that particular bridge from not a sentient conscious, intelligent being to clearly an intelligent, sentient conscious being in the way that we define it as humans. I think it's a really fascinating concept. It used to be a very popular science fiction concept. I think it's become less and less mainstream as people have focused on other things, but it seems like it's actually pretty low hanging fruit. There's a lot of animals that are very close to human levels of intelligence, maybe not on the structure or learning side, but biologically, their brains are high glucose eating brains. They've got lots of neuronal folding. They've got plenty of surface area for human level sentience to work, in theory. So I'd say there's a few primate species that are pretty promising. Dolphins are pretty promising. African grey parrots would be really promising if we did a little bit more folding on the brain. They're a little too smooth brained right now to get the density that you would need. But your question was, how would that apply to defence or to… what was the question?
WALKER: Do you envision a role, if any, in the defence context?
LUCKEY: Well, absolutely. And I'd say I think that different people think in different ways. And there is value in having people looking at problems in a different way, even to the extreme limits where there are people who, I'll be honest, I would never really want to have over for a beer, but I would trust them with my life when it comes to analysing a particular technical problem because that's the way that they're wired. And vice versa. You don't want that guy being the guy who has to go around and convince using soft power and charisma, that people should get along with our nation or our country.
And I think that there might actually be some really interesting ideas out there that cannot be conceived of on the human side. It could be that if we were able to, for example, turn dolphins into human level communicators and thinkers, it might be that there are certain problems that they're better suited at thinking. And by the way, this is not my idea exclusively. In science fiction, the trope is usually that dolphins are the really good space pilots. Either they're very good at thinking three dimensionally or navigating the wormholes or whatever it is, and I think there's probably truth in that. If we have so much diversity of thought in people, surely in other species that think in radically different ways, that structure their learning in radically different ways, surely there's things to be learned. And it's kind of interesting that we're basing all of our artificial intelligence experiments on human cognition. It could be that people are not the most efficient thing to model after. It could be that our way of thinking is not the most efficient thing to train after. It could be that other species are more efficient, that they get more done with less because they've had more evolutionary pressure to do so. I think there's even a lot to learn on AI from other species. As far as applying it to the military, the United States already actually has a significant number of members who are non human. We have a lot of canine units that are full of dogs that are some of the smartest, best trained dogs on the planet. The US Navy has a dolphin and a seal fleet, I don't know if you're familiar with this, but we've got dolphins that are members of the US Navy. And so it's actually not much of a stretch to imagine that we would take these animals that are already at the kind of pinnacle of what's possible and above what's possible in the natural world and then just drag them that little bit further.
WALKER: How important is it for people creating military technologies to have a general ability to forecast well? So do you want your product and engineering managers to be the sort of people who are early on Covid, early on Ukraine?
LUCKEY: Interesting point. I don't think it needs to be all of your leaders. I'd say the most important thing is you have someone in the company with the ability to do these things, and that you also have a certain level of mutual trust and respect where your product managers believe the person who is telling them these things. To have a specific example, I don't think that it's nearly as important to have a group of people whose real job is to manage people, manage programs, manage products, inherently be good predictors of the future of what China is going to be building in their military. It's totally fine for them to be just very good at their little portion of solving the problem very tightly, have a tight expertise in that area, and then to instead rely on someone else in the company who says: “Hey, my job is to look at China and to understand what they're doing and then communicate to you guys what the bet we're making as a company with regards to that is.” And it's a classic concept, specialisation of labour. I'm a big fan of it. So I would actually say I actually don't think that they all need to innately be able to do these things. They need to be really good at what they need to be good at. I need to be good at the things that I'm good at. And then I certainly have to delegate to a lot of people. When I talk about what I think China is going to do, you think I come up with all this stuff? No way. I talk to people who are way smarter than me, and they come up with most of it. And then I need to be smart enough that I can judge if they're full of s*** or not. But most of my opinions are borrowed, not my own.
WALKER: Final question. We're here in Sydney today. What is the biggest thing Australia is getting wrong about its current defence strategy?
LUCKEY: Oh, man. Well, that's dangerous when you're trying to do well in Australia. I'd say probably the things that it's getting wrong in its defence strategy. You could make the same criticisms that are true in the US. It's the classic ones, right? It's moving too slow, the timelines are too long. The sense of urgency is not universal. At the same time, I'd say that probably those weaknesses are identified by a lot of people. Like, you can have a whole organism itself that moves slowly, but then you have people who are inside of that organism that understand the problem, they're moving very quickly. For example, we're partnered with DSTG, Defense Science Technology group, and they really do understand the urgency of the problem. They have moved extremely quickly. I could fault maybe the broader defence apparatus for moving more slowly than a technologist like me wants to move, but DSTG has gone from… We went from meeting them to having a signed contract to partner on Building XLVUV in five months, and I can't criticise that. So probably the criticism is less the monolith and more parts of it are very fast, very functional, proving that you can do these things well. And then you have these larger organisms where institutional inertia is hard to overcome. And it's the same thing in the US. We have partners in SOCOM and in the Navy and in the Air Force who are very fast moving. There are other parts of the military that have been much slower, and I'm very frustrated with them. But it's not a good idea for me to talk about who they are and what their names are, because at the end of the day, I probably catch more flies with honey.
WALKER: Fair enough. I won't force you to do that. Well, Palmer, I’ve probably got 20 questions that I didn't get to ask you, but maybe I can have you back on the show. This has been so much fun.
LUCKEY: Thank you so much.
WALKER: Thank you.