Casey Handmer - Terraform Industries

from a scientific perspective, why this lake, why so deep? Like, usually lakes are not in mountains. What explains the depthness of the lake is my question.

That's a good question. Yeah, so most lakes are actually fairly ephemeral. They tend to silt up, right? They tend to fill up with sediment. And so there are a handful of extremely deep lakes worldwide. Lake Tahoe is quite deep, and Lake Baikal in Russia is quite deep. And those lakes are deep because they're on rift valleys, and the crust is actually spreading apart in those areas faster than the sediment is accumulating. And thus they're quite deep. The Sierras has on the order of 10,000 lakes, and they're all pretty cool. It's a beautiful place. And most of these lakes are glacial lakes, so they're areas where the ground has been gouged out by glaciers. Normally rivers and stuff run downhill, but glaciers can run uphill to an extent. They can get kind of pushed up. So as they go, they can carve out these scallops in the ground. And then, you know, the glacier retreats and melts and those scallops fill up with water. And I took my inflatable raft and my sonar buoy up to the 20 Lakes Basin in August with my son as part of a kind of shakedown temporary test and found a lake up there called Steelhead Lake. It's actually the site of the Hess tungsten mine that was 30 meters, 33 meters deep at the deepest point, which itself is staggeringly deep. Like, it's deep enough that you can fit, you know, an absolute, like, Amsterdam Max container ship in that lake.

Right.

which is kind of mind-boggling. And so I thought maybe this lake up in the Sierras could be 30 or 40 meters deep even. It's a similar size, but it was far, far deeper than that. And it was really mind-boggling. Like, I just kind of jumped in the kayak and started paddling around. And within like literally spitting distance of the shore, I was already in 20 meters of water. And I was like, okay, this is weird. I grew up in Sydney and Sydney is regarded as one of the finest natural harbors on earth. It's It's about the same size as the San Francisco Bay, but it's far deeper. You can fit very, very large ships into Sydney Harbour. And in the parts where I used to sail, which are relatively close to the ocean as opposed to up the river, it was not unusual for the harbour to be 20, 25, 30 meters deep, which is of course incredibly deep by even modern standards. And when Captain Cook first kind of sailed by and the first fleet showed up, they were like, oh, fit the entire British Navy fleet in here with ease several times over. And yet this lake, which is at the top of this mountain, it's like literally 13,000 feet up in the mountains, is far, far deeper even than that. It's really mind-boggling to think about. I mean, and part of the reason it's so deep is there's not much sedimentation happening up there. The lake is on top of a mountain, so there's no catchment that really drives dirt into the lake over time. The lake is frozen for, you know, 10 months of the year, so there's even less, you know, water flow in that area. There's a small lake just upstream of it that probably captures most of that sediment, to the extent there's any sediment at all. But it's— but yeah, like, how the hell it got gouged out that deep, I don't know. There's like pondweed and stuff. Um, there were fish— fish were stocked in there in the '30s, but, but I don't know if any are still in there. I didn't see any fish in that lake. It's, uh, you know, it's brutally cold most of the year, so you have to wonder how the fish would survive without sunlight for, for 10 months.

And how do you decide of those expeditions? You have like a formula? It's just a hunch? And can you explain that hunch if there's one?

I don't know. I've always liked doing weird stuff. So, you know, it has a point, right?

Because you're full, like, you're all like this Mars exploration type of guy and you're like hyper curious about the planet itself.

Yeah.

Right. So obviously—

Well, it looked like a good shape problem that like I could maybe go up there and solve. And I like the physical aspect and I like the camping and I like going in the Sierras. And so all those different things kind of came together. And then after a few years of kind of tossing the idea back and forth, I was able to figure out how to make it happen. and blocked out the time in my calendar and went up there. And yeah, it was extremely physically strenuous to do it. I don't know if I'd be rushing back to do it anytime soon. So how long did you do it? So me and my partner, we started at— we started Tuesday evening because we went up Shepherd Pass, which is— it's one of the harder passes in the Sierras. And we spent 2 nights on the pass, one night about halfway up and one night near the top. And then on the third, The 3rd day, we popped over the top and down the other side and made it almost all the way up to the Talayna Basin on that day. So on the 3rd night, we camped once again just below the lake. Generally speaking, when you're up in the mountains and you're acclimating, it's good to kind of get as close as you can without getting too high. And then you sleep for a night, let your body adjust a bit, and then go up again the next morning. Then after you've done your hard climb, you have breakfast and head And your friend, is he as weird as you are, or is this— Oh, he's much weirder than I. This is Edward Jacobs. He's a ride-or-die buddy now. He came on one of my camping trips and lived to tell the tale. I'm actually somewhat notorious in my family for like dragooning, you know, sisters and brothers and siblings and cousins and so on into like wildly improbable adventures that involve walking far too far. And And then them kind of like limping, but telling the tale after the fact. I remember at a family event, probably 20 years ago now, maybe not quite that long, maybe 15 years ago, with my cousin, we were staying at a place called Drouot, which is in New South Wales in Australia. And it's on a beach that's called 11 Mile Beach. I think it's called 7 Mile Beach, but it's actually 11 miles long now because the Shoalhaven outflow was moved 4 miles to the south. But my dad dropped us off at the old outflow and we decided to walk 4 miles south Until we reached the modern outflow and then turn around and walk back again. It's kind of weird, like beginning a hike on a beach at 10 PM and, and you're— the city you're walking to is not even visible over the horizon. But after 3 hours of walking, it is visible over the horizon. And then after 3 more hours, you kind of, you get closer. And I think we arrived just around dawn. We were pretty sore the next day. So that was, that was a good, good, good time. Yeah, it's just kind of things I can— kind of things I like doing, I guess.

How would you compare that to what you're doing at Terraform? Obviously requires immense grit. You have a hunch that this can scale, right? But how does it compare?

Yeah, it's actually kind of similar in that you need a kind of fairly specialized set of skills to make progress at all for it even to be possible. And then you need to show up and show up and show up and show up and just day after day after day, keep on grinding away until it's done.

How do you select your teammates? I saw some members of your team, like, How do you select for grit? Let's start with that.

Well, I think to an extent they select in, right? So they can kind of see from the outside that it's an environment that rewards people who are willing to put in a lot of work. And people who are willing to do that obviously want to go and work in that kind of environment because then they can have a competitive advantage. Their natural characteristics can be rewarded. And then the work itself, you know, it's its own test, right? Every day you show up and you kind of get shaped by your environment and tested by it. And it's the same thing as if you're a professional mountain porter or something carrying loads up and down mountains, like you get stronger and stronger over time.

Right, Sherpas, right?

Exactly.

Okay, how far are you like with the last station that we talked about that I shared my screen last time we spoke? You're literally producing gas with solar energy, right? Which is quite original. I like the idea. Makes all kinds of sense. You've shown proof of concept, right? How far are you? What are you chugging here?

Liquid death.

Yeah, death water.

H2O.

You feel badass like chugging this?

It's like, do I feel badass chugging it? Yeah, that's a good question. How many how many moles of that's like twenty moles. It's just I just had another podcast just before you, and I needed to get. To get some, uh, some liquid in me before I—

Metal, hard, hard metal energy in you.

Exactly. Um, sorry, I forgot your original question.

Yeah, how far are you with the team? Maybe if I ask that like, let's say I'm a teammate of yours, right? I'm in charge of like one specific system, but am I talking to the rest of the team? How does— how do I make sure that this specific component integrates with the rest of the work?

How do you separate Yeah, I mean, we have 3 core subsystems and the interfaces between them are pretty lightweight, quite deliberately decoupled, right? So, you know, but they sit right next to each other and they say, well, you know, we ran the math and, or we ran a test and it looks like we can hit, you know, 99% instead of 98.5% purity. Does that help you? And they say, actually, it does help us. Thank you very much. We'll take 99%. Or they'll say, it doesn't make any difference, in which case, you know, maybe you put in a bit too much effort. Um, and, uh, yeah, and then it's just a matter of making sure everyone stays on the same page.

And how much is enough hardness for a team member? Like, because there's such a thing as burnout. Arguably, I experienced it in my last ultra race. The 94-kilometer out of 125 was my first DNF, right? What's— I'm sorry, someone— and also, like, what's, uh, what's not enough, right? Like, have you ever outsourced something to a teammate and was like too easy and just they went to SpaceX because it was like hard?

Yeah, a few of our teammates have left for SpaceX over the years. I feel like it's not a critique. It's maybe a vote of confidence in our ability to create and develop talent that people are able to move on to other jobs where they can contribute in the wider environment. SpaceX is a huge organization and it does things at enormous scale. And it's an incredible environment, I think, to try and push and succeed. if that's your thing and you're excited about rockets, and I totally get that. If you want to work in a small organization where you have a big impact on the outcome and so on, then obviously SpaceX has grown beyond that point now. I think you just kind of take it as it comes. I mean, like lots of people get different projects and over time those projects change and evolve and someone solves one problem and you give them another problem. You say, was that too easy, too hard? Give you another one. You know, you're ready for another level of responsibility. And actually one of the The greatest pleasures as a father and as a manager is taking people in as relatively green and inexperienced early career people and saying, well, this is how it's done. And today you're responsible for this, but tomorrow you're responsible for delivery. And the day after you're responsible for the part working correctly. And the day after that, you're responsible for the person who receives the part being happy about it. And each of those things are a kind of a big stair step, level up in terms of internalizing responsibility for extrinsic factors over time. And you really want to climb up the earlier steps in that thing as fast as you can. This whole like, oh, not my job, not my problem kind of attitude doesn't get you very far. If you want to have an interesting career, you've got to be able to take responsibility for successful operation of your system in the wider world. And by success, I mean People are desperate to have it.

And how do you speed date with someone to know that they have it? I was green, I was full of potential. You were green as well. I don't think they obviously took you to your full potential, right? Like at Caltech.

I don't think they're even trying to.

Yeah. So what happened there? I guess like what's— well, first, how do you spot these people, right? Is this like they've done like crazy ultramarathons and worked on ultra hard projects? And then how do you fully exploit their potential? Exploit might not be the right word, but just like fuse with them and just 1+ 1= 10.

You develop potential. I mean, actually, like elite sports or endurance sports are a pretty good filter actually, because it's impossible to succeed at those without having a high degree of intrinsic motivation. And fundamentally, maybe it's just something you're born with, but it's extremely difficult to succeed. without having the ability to drive yourself really hard. And so it's just, it's an unfakeable signal, right? It's an honest signal if someone's able to do that. But yeah, that said, there's plenty of people who are successful here who've never really distinguished themselves in terms of athletics. That's fine. I mean, generally speaking, in an interview, you're just looking for evidence of a track record in that direction, right? And, but for early career people, often there isn't much evidence yet, so you've got to develop it. In terms of exploiting potential, I think it's more like developing potential. It's more like helping people reach their limits. And, you know, if you've worked in extreme sports, you'd understand that like having a coach can often be extremely helpful. They can help you kind of get past internal barriers and see things in a new way and push you to limits of your ability in a way that maybe you can't necessarily do as well by yourself. And I see my own job as basically in that. It's helping people get the most out of their career and out of their lives.

How do you manage your stress, you know, like while getting up there? What goes on up there, like sleeping a couple of nights and just like walking for a fucking long distance with 60 pounds on your back and in business when it goes bad, you know, like how do you manage that up there?

Well, it's an interesting thing. I did a fairly heinous trip in the mountains in Germany and Austria a couple of years ago. And in the process realized that something about running a startup had changed the way I approached fear and risk and so on. Yeah, basically, like, starting a company is a hell of a drug. It really changes the way your brain works. Yeah. So last week I was up in the mountains having a great time. Like, almost every minute I was just like, you know, at the peak of my game and having— I was just like in rapture. It was incredible. The worse the situation got and the less comfortable my climbing companion got, the better time I was having. Yeah. We're like, Scrambling across a couple of miles of boulders to cut a corner on a trail. And it starts raining halfway through. And I'm like, oh, I'll put on the rain jacket and keep on going. And I'm just thrilled because for the first time in my life, I actually have a rain jacket that will keep the rain off because I actually went out and bought a nice rain jacket at the age of 39. And meanwhile, he's like, oh, these boulders might get slippery or something. I'm like, I'll probably be fine there. They're granite. Granite doesn't really get slippery. In the same way that say limestone does. So I was just having a great time. At one point, my sleeping mat stopped working and I got a bit cold. But I was like, I don't know, it's like you feel cold, but you weren't cold, if that makes sense. I've been in situations on mountains and stuff that are much worse. In terms of startup, every now and then you have moments of terror, right? It's a bit like Wagner, right? You have just long periods of just grinding away and then moments of sheer terror. But again, after you've done it, after you've been through that process 3 or 4 times, it doesn't frighten you in the same way anymore because you've kind of learned the coping skills and the ways of compartmentalizing and approaching these problems. I don't think it's for everyone. I'll put it that way. I think for a lot of people it would be a pretty permanently negative effect on their mentality, almost traumatizing in some ways. But for me at least, maybe thus far, at least, it's been fine. But for me, it's kind of neither here nor there. And yeah, I've been having a good time.

Is it another physics problem, like cortisol in your blood and need to downregulate with something else?

I guess there are ways of mentalizing it like that. People say like, oh, stress is a terrible thing. But actually stress is the only thing that creates growth. You know, if you work out, like you have to stress out your muscles to make them grow. And if you want to get smart, you've got to try hard problems to make your brain get better. And yeah, it's basically stress. Stress is a nootropic. Anxiety and stress help you think better. They can also make you crazy. So, you know, there's kind of a balance to be struck there.

Yeah. Are you a biohacker? Do you, what's your stack? Do you have something or just like raw dog it?

I'm pretty natty in that regard. I mean, I'm happy to share with your audience the things that I find tend to improve my mentality when I'm getting stuck in a grind is like, you're on the checklist. Have you slept recently? Have you eaten enough food, eaten a balanced diet? Have you had good interactions with your friends and family? Have you exercised? And have you interacted with nature? And so one of the nice things about going up in the mountains actually is that you get many of these things for free, right? So it's cold in the mountains. So when the sun goes down, you have to lie down in your sleeping bag and stay warm. So it's like basically mandatory 8 to 10 hours of sleep a night, which is probably twice as much as I normally get. You're off the internet because there's no service up there. So that alone like saves you hours a day. You're doing a shitload of exercise, which is, you know, provided you don't get horrendously injured, is really good. Generally speaking, Maybe not eating the healthiest food, but you're eating decently healthy food and the exercise you're doing more than makes up for that to the extent that it's not healthy. You're often up there with friends, although sometimes I go by myself, and you're in nature and having a great time. So it's certainly the most helpful thing for my wellbeing is to go up in the mountains and have a horrible life experience.

What's your relationship with the internet and Where's it going?

Well, that's an interesting one. I'm an elder millennial. I'm 39. So I remember the first time I had, I think we called it ADSL, internet connection, and I discovered Wikipedia at about the age of 17. And back then Wikipedia was small enough that you could actually read the whole thing if you were curious enough, which I was. So over the course of about a year, I read pretty much all of Wikipedia. This is long before it became a kind of lightning rod in the culture war. And it's just fascinating, absolutely fascinating, because I'd always been a very curious person. And if the things I was interested in didn't have an accessible book in a library that I could reach on foot, then it just wasn't part of my world. I didn't get access to advanced organic chemistry textbooks until I was in college, but I was interested in it since I was 6 years old. So it's kind of interesting, right? It's like I spent 12 years or 12 or 13 years just kind of theorizing about about your biochemistry and organic chemistry before I had the ability to actually get access to people who knew what they're talking about in this area. Of course, that's completely insane. Like nowadays, you can ask your AI, you can go on the internet, you can download a textbook, any of these things are possible. So it was super interesting. And then obviously, internet was also a way to provide communities for people with strange interests and predilections. And initially for me, that was Facebook, and it was how I kind of stayed in touch with people in my college.

Wasn't it MSN?

I had MSN Messenger and stuff, but just for me personally. And then, but in recent years I found my tribe on X and my tribe are by and large—

Didn't Facebook fuck up the internet, right? I guess the last renaissance was LLMs taking info on the internet and making it very valuable. But didn't social media like made internet less fun for most humans?

I would say like the revealed preference of most humans is that they like social media. including me. But you kind of have to know what you're getting, right? Like, you go on X to get sanity damage and enjoy it. It's like drinking, right? It's like, I know it's killing brain cells, but it's fun. And the other thing is that X is a place that attracts lunatics like me who enjoy building companies. And so, you know, my contemporaries, kind of peer group today, are mostly entrepreneurs of one flavor or another.

Wikipedia, I'm a fan, still am. Like, I still read lots of shit on Wikipedia and go into very deep rabbit holes. They're always looking for funding, right? Like, they're sort of struggling and OpenAI, like, scraped their shits, right?

Well, no, not really. I mean, we get the cost of, the cost of operating Wikipedia is basically nothing, right? Like, it's a, it's a, it's a, you could serve the entire thing on, on a cloud, uh, a Cloudflare bucket for—

Not really heavy also, just like text, right? Like—

It's text. Yeah, you can, I downloaded the whole thing 10 years ago onto a USB drive just, just for shits and giggles. You can do it. But the Wikimedia Foundation spends a shitload of money on social causes and activism, and that's what they're raising money for.

Interesting. Were the big LLM boys in the right to just like fucking get everything on the internet? I think from one standpoint, they added gigantic value to humanity because they repacked that value, right? But On the other end, there's a lot of people shit talking and pretty angry about them doing that.

I mean, it doesn't matter what they would've done. It doesn't matter who they would've done it. Someone would've found a reason. You're probably old enough to remember 10 years ago in the early days of woke, there was this fun game that the kids used to play on the internet, which is like, tell me anything and I'll tell you what's wrong with it. I'll tell you how it's problematic or how it's concerning. And it's just It's just a dumb game, right? Because it doesn't actually make the world better. And so, yeah, you can say, oh, well, here's this amazing, miraculous new technology where every single person on Earth can have a personalized, hyper-intelligent robotic assistant cognitive prosthesis that can help them solve any problem in their entire life and democratize them essentially for free and without permission, without interdiction from hostile foreign governments, you name it. Basically, anyone can get a personalized expert consult on almost any subject. Well, let me tell you what's wrong with it. They went on Wikipedia and they downloaded data that a bunch of millions of contributors had given to Wikipedia for free and they took it for free. How dare they? Well, give me a fucking break because I read Wikipedia when I was 17. And if I was able to, I would read it today. It's just too big for me to read in one lifetime. But it's free for a human intelligence to read. It'd be free for a robot intelligence to read. And the thing is more to the point, I'm going to kind of get on my soapbox about this. I'm old enough to remember when Google, whose original mission was to organize the world's information and make it accessible, had an extremely ambitious project to scan all the world's books and then make those books accessible to people. So, for example, the vast majority of books ever published, ever printed, are no longer in print and cannot be obtained for any amount of money from any dealer whatsoever. There might only be a handful of copies left in a library in the entire world. So if you don't happen to have access to that library, which necessarily means living in that city and having the appropriate credentials, the information in that book is not accessible to you. The original rights holders of that book are often long dead. The publisher might not even exist anymore. But because of the way legal principles are set up, obviously someone still owns in many cases the copyright to that information, and that's fine. And Google, being Google, having almost infinite resources, went about this the right way. They went to all these different publishers and they sought to obtain permission from them to scan and then distribute the information from these ancient intellectual property, kind of properties that were not even no longer economically productive. They were forgotten. They were, I mean, like, which publishers make money? The publishers that make money are publishing like fucking Harry Potter and stuff like that, right? Like the Bloomberg terminal or something like that. Like, no publisher on earth is making money selling a book that was first published 100 years ago. And so it's not even a kind of economic thing. But what happened was enough of the world's publishers got together in a big class action lawsuit and basically forced Google's hand, that Google was then forced on pain of basically extreme violence and coercion to delete their priceless records, priceless scanned copies of many of these books that they had sought to obtain and then distribute legally via paying the rights holders and so on, paying out the rights holders. So all of humanity was deprived of the benefit of that project. And of course, Google didn't try again.

So you believe—

so that shows what happens if you try to go in the front door. So why the fuck would you go in the front door? Because if you try and do the right thing by people and by customers and so on, the incumbents who are rent seekers, who just happen to have been the people who are privileged by legislation written back when information was distributed by printing press, will still try to fuck you. So why bother?

Do you believe in patents and trademarks and all that shit?

There's obviously a— there has to be a way to extract value from intellectual property and monetize it in order to incentivize the creation of new intellectual property. On the other hand, it's pretty clear that extending IP for 75 years to the pests of the Disney Corporation or something is not in anyone's interests. And it's obviously, everyone involved has derived the necessary amount of value from those properties at this point. And that, like with many things about these different principles and values need to be balanced in a society, and a society that's able to take, say, the property of Mickey else and then recycle it after 75 years is a society that's much more dynamic and creative than one that is like, oh no, some guy thought of this 100 years ago and now you can't touch it forever. And yeah, so generally speaking, when it comes to things that I create and I generate, I go out of my way to give them away for free.

If you make it to Mars, would you like to run your own civilization?

That seems like a very difficult problem, very hard thing to do.

The civilization or getting to Mars?

No, the running it part.

They're running?

You said, would I like to run my own civilization? I said, I'd like to potentially go to Mars and potentially contribute to industrial kind of development there. But actually, having line responsibility for the successful execution of an entire civilization is probably beyond my pay grade. It's a very difficult task. Someone has to do it, but it's a very difficult task.

So humans, but humans is a physics problem, no? Managing humans, it's a science, right? Don't you think you could install systems the way you would with cracking the current first principles problem you're cracking right now?

I mean, I'll be really blunt. The management of organization of large numbers of humans at scale is a problem that has beset and befuddled humanity since we first started throwing up pyramids.

Well, we have corporations. They work. Why not select people?

Well, they work, but they work better than the alternative. But thousands and thousands of books have been written on on managing people and it's still an unsolved problem, right? Still a problem that people don't really, um, people don't really know how to—

Yeah, the fact that corporations fire their employees all the time, you couldn't really fire your civilians. I mean, governments fire their civilians all the time, right? They put them in prisons and hoose them.

Well, a small fraction of people. Yeah. Um, yeah, I mean, So I certainly think that of all the different economic and political systems that have been tried throughout history, the ones that we are using right now seem to be surprisingly effective. And certainly it is the case that a kind of giant syncretic political— sorry, economic system composed of millions of independently coordinating, voluntarily associating individuals and corporations, all cooperating in a nonviolent way beneath the umbrella of some democratically elected lawmaking and law enforcing apparatus. It seems to be relatively effective because it means that you can kind of fuzz the problem. You can generate thousands of different styles of corporations and management structures to solve different kinds of problems at different times to different kinds of people and just kind of experimentally determine what works best. Yeah, fine. Great. Wonderful. Do I want to be the guy at the top driving it? I mean, like, look what happened to Obama, uh, you know, over, over his 8 years of his presidency. He went from being a young man to being an old man in 8 years.

Yeah. A DAO. Yeah. It's like redistributed power and possibly, yeah, AI with, I don't know, 50%, 51% of voting rights because I believe more in AIs generally than humans for judgment. Or at this stage in like 30, 40 years, I'm pretty sure we'll get AGI. What are your thesis on that?

Oh, probably much sooner than that. Yeah. And it's an interesting idea to say, well, in what ways can the LLMs or their descendants assist with human governance problems? And I think that it's actually not a completely sui generis thing. I think that the bureaucratic state, which came about in the 1800s at large scale in Europe and since then in other places too, is an early instantiation of an artificial intelligence. You've got this kind of large state bureaucracy, corporate bureaucracy, where the individuals within it don't actually have very much autonomy and they execute from a fairly finite set of instructions, a whole set of operations that more or less autonomously and constantly discharge the duties of the state. But there's, as I said, there's very little leeway for individual expression within that organization. Essentially like COBOL, the original business-operated language, right? Business-oriented language was designed to substitute for the parts of both government and private enterprise that would've been executed by people who were called computers or people who were called calculators back in the day. And then you could automate that part of the business within these mainframes that would do these batch mode operations and still do actually for much of our banking system, for example. And so you just basically take the additional parts of that bureaucratic apparatus and you hand them over to the LLMs for the execution of that kind of stuff. And that's already happening. I mean, you see that happening in private enterprise, you see that happening in businesses at scale. And it's already the case that The preferred front door, if you like, for a lot of the services that we use at Terraform is an LLM that belongs to, or is trained upon, or post-trained for, I don't know, banking apparatus or something that our internal LLM then goes and integrates with or interfaces with, which is far more efficient than the alternative. Yeah. So in some sense, like the corporation is a prototypical AI. Because it has lifelike properties and information processing properties and learning properties and so on, but it's not a human. It's composed of humans and machines, but it's not a human. So we'll see. But I don't know if that necessarily solves the part of governance that is hard.

For sure.

The part of governance that obviously ages people with a high degree of executive authority.

I think we'll be Neuralinked by this time, right? So I think that should solve it much better. AGI, even Neuralink and getting to Mars, is that dependent on cracking limitless energy, which is what you're doing now?

No, no, we could do it without limitless energy, although more energy helps. And I certainly think that right now our civilization is is largely preoccupied with kind of these insane questions of structural decline, which are brought about by mostly energy scarcity amongst other forms of rent-seeking. And that lifting that— so if the energy was limitless, it would be free. It doesn't have to be free, it just has to be cheaper. So we do need to increase supply, but we don't have to increase supply without bound.

And that can— and what is your ultimate dream? Like, what's the ultimate version of Terraform in like, I don't know, 20, 30 years from now?

20, 30 years from now, I think there's a good chance Terraform will have pushed drilling decisively out of the market. So you'll have humanity consuming something like 5 times as much oil and gas as it currently does. Most of that oil and gas will be produced locally to where it's consumed. And so relatively little will be shipped around the world in tankers and so on. It'll be cheap by today's standards. So in 2026 dollars, we're talking like $30 a barrel or something like that, kind of 1968 level prices. And it will no longer be the kind of primary source of the kind of ongoing industrial impoverishment of the developed world. Yeah, I think that's a nice future to look forward to. I think you expand the privilege of aviation in particular. So you say like, For a lot of people, having enough oil to light a kerosene lamp at night is a big deal, or to run a car or motorcycle is a big deal. But actually, if you say, well, for the developed world at least, what does it look like to increase your per capita energy consumption by a factor of 10 if you've already got an air-conditioned house and an air-conditioned car? And it looks like using, consuming just a shitload more aviation mostly. And then obviously— On a per-human-capita basis, much more AI.

Isn't global population falling down and isn't the world merging towards more like solar and doesn't give a shit about gas?

Yeah. So the human population's not yet falling, at least generally worldwide, but it is certainly the case that it is on track to fall. And it is also the case that the economy as a whole, as it grows, will continue to electrify where it can. And some parts of the economy are easier to electrify than others. But the parts that are actually particularly energy intensive are actually really difficult to electrify. So if you need extremely high temperatures for certain industrial processes, or you need your energy to be mobile, say in the form of rockets or or long-distance shipping or high-performance aviation, then it needs to be a liquid, typically a liquid or a gaseous chemical fuel, which is where we come in.

And I asked you your dream, your vision in 20, 30 years. Weren't you like a guy that wanted to go to Mars? It seems like this vision is mostly like capitalism. How's— yeah, what's your number one dream? Is it just helping the world also? What are your goals here, your incentives?

Well, I think some people say, well, we'll go to Mars and all the problems that I face in my world, in my life here will fall away. And it's always been a place that I think has attracted dreamers for that reason. But if you think that the Mars civilization is going to somehow allow humanity to step beyond capitalism, I think that you're vastly mistaken given that Mars will demand of humanity far higher levels of productivity and innovation and performance than anywhere here on Earth. It will necessitate even more extreme forms of capitalism, hypercapitalism, if you will. And what Terraform is doing here is— sorry, what Terraform is doing here is definitely like a piece of the space puzzle, because when humans go to space, they will need hydrocarbons as well for making plastics and things. So that's kind of the piece of the puzzle we're working out here. But as of today, 8 billion potential customers here on Earth and zero on Mars. So it's pretty clear to us where our business is to be built.

Why hypercapitalism on Mars? Right. Also, what will happen to human productivity once AGI is here statistically?

Well, those are 2 separate questions. I think As far as Mars goes, one of the most compelling and beautiful accounts, at least contemporary accounts of civilization on Mars, is the Mars trilogy by Kim Stanley Robinson. And Stan, as he is known to his friends, is a compelling author within a literary science fiction tradition that kind of anchors itself in the American left. And I've written a technical commentary on Red Mars. Which was quite fun to write. But the great weakness of the non-capitalist economic means of organization is that while they sometimes sound good in a story, they don't actually work in reality. And so it's important to realize that the economic dynamism that you would need in space is not the sort of thing that's delivered by a kind of top-down command economy or something like that. That was the first part of the question.

It's just sheer survival out there and it's super hard to create new things on a very—

Oh, and then productivity and so on. Yeah. Yeah.

And then it's super wealthy people also that we would ship there first, no? Or—

Well, I think by definition, if you're educated enough and skilled enough and connected enough to kind of even be in the running to be hired as an astronaut, you'll You'll be statistically one of the wealthier humans in the history of humanity. It does not necessarily mean you'd be some hyper-billionaire. Okay. In terms of human productivity on Mars, well, the whole point of going to Mars is to build a kind of a second branch of humanity, which is industrially self-sustaining. Mars's environment is actually fairly hostile to human life. For comparison, it is somewhat more hostile than the summit of Mount Everest, for example. which is famously fairly hostile to human life. And so it's impossible to survive on Mars without advanced technology, which means that your industrial civilization there cannot just be like Hawaii 2.0, where in Hawaii you can basically survive in a pair of underwear indefinitely because the environment is so benign. It's much more like a nuclear submarine or something. So you need to have the ability to manufacture and maintain all this machinery to keep you alive there. And all that machinery today on Earth is produced by maybe 5 or 6 countries on Earth. So Germany, Japan, China, South Korea, United States is basically it. And so you say, well, the smallest of those is probably South Korea, and South Korea is on the order of 50 million people. So in order to have the ability to potentially have a complex enough, diversified enough economy, to be able to produce microchips and robots and rockets and pressurized systems and spacesuits and all the rest, all the other stuff you need to survive there. You would have to have an economy that's at least as developed as South Korea's with a population in the tens of millions of people. That seems like a very tough thing to do. So you kind of look at that number and you say, well, maybe, maybe, maybe if we get really, really good at automation and factories and humanoid robots and so on, you can get by with only a million people. Right. But even then you're saying, okay, well, those million people have a productivity equivalent to South Korea, then their per capita productivity is going to be in the millions of dollars per person per year, which is comparable to an employee at Netflix or something like that. Yeah. So basically another way of looking at it is to say, well, the amount of output on a per capita basis has to be so much higher than it is even in today's world that the per capita allocation of robotics or automated production or something has to be extremely high. And thus the human has to sit kind of at the top of this growing pyramid of abstraction from actual material labor or contact with the parts themselves, which is easier to envision in the case of a compiler for a computer or something than in the case of building up a robotic capacity. Another way of thinking about it, you say, well, the first time humans land on Mars, a team of 50 people land and they set up the factory that does steel. They make steel. And then after 2 years, another, the population doubles. And after 2 years, the population doubles again. And after another 2 years, population doubles again. So every 2 years, this team of 50 people has to basically double the output of their factory. And maybe you can double it once or twice without too much difficulty, but to double it 20 times in a row requires obviously extreme reliance on automation. And the people in the factory are no longer like, pouring ingots one by one by hand anymore. They would necessarily have to be essentially supervising the automatic emplacement of extremely large conveyor belts and things like that. I wrote a book about this about 10 years ago that I think kind of covers the details quite nicely.

Yeah, I think I read it or the Google Docs at least version of it, right?

Yeah, there's a Google Docs version.

Yeah.

I should probably revise it at some point. I've had a few ideas since then.

Yeah, for sure. That'd be great. AGI and human productivity in the next 10 years.

Okay. That was the other question you asked. So opinions vary on this front. And it is certainly the case that while the US economy is growing right now, essentially all that growth is in the AI space, physical construction of data centers and so on. And so some people think that that kind of scarcity of human intelligence, amongst other things, is one of the fundamental bottlenecks on economic growth today. And that, or maybe some other aspect of AI development, such as the liberalization of the construction of new power plants in the United States, will lift that constraint from humanity and enable a higher rate of economic growth. And kind of finger to the wind, I think that with optimal industrial policy with current technology, United States could grow perhaps 4% per year, maybe 5%. And you say, well, maybe with the advent of AI doing a lot of the work, you could double that to 10%. And 10% growth per year would be pretty extraordinary. That'd be roughly equivalent to the 5 fastest years of Chinese growth in the last 35 years. And also, like, 10% growth per year, is extremely exciting from a social perspective because it means that the opportunities are once again available for young people in a way they haven't been for generations. Other people think that Baumol's cost disease, blah, blah, blah, means that growth will, if anything, slow down and increasing dependency ratios and aging the boomers and so on will mean that essentially all the value generated by AI and then some will be captured by stringent overregulation and a few other things. That's also possible. And then some people think that we'll get recursive self-improvement and the AIs themselves will build their own economy that then essentially grows at the speed limit of the construction of new AI factories, which may be significantly faster than what humans can do, maybe 100% per year, something like that. So it could be anywhere in between. I guess we'll find out.

Chop most jobs, like 99% of jobs?

I think that's a very reductionist point of view. I think that it's natural when new technology comes along to assume that this is going to result in massive economic upheavals and everyone's going to lose their jobs and starve. But actually, if you look at the scoreboard, the reality is that post-industrialization, the introduction of new technology has basically resulted in improved quality of life and life outcomes across the board for pretty much everyone. So it's certainly the case that in today's world, there are very few people who are the lifelong driver of horses down at the local manor, and they drive the rich people around on their horses, like in Pride and Prejudice or something, right? That job has disappeared. Do you or me or anyone else lament that our children won't have the opportunity to be the keeper of the horses for the Darcys of this world? I don't think so. And so my instinct is that higher levels of wealth creation, economic growth will actually drive increased demand for higher quality and higher volumes of consumer goods, and that those in turn will drive increased demand for labor and increase the value of human labor and basically continue the story of massive increase in per capita GDP.

Or maybe that would be 2 economies, one that is AGI produced, one that is human produced. AI is invention as well. It's like recreating—

That's still the case. Right now, I'm speaking to you with the help of a mass-produced webcam, mass-produced computer, mass-produced cables, mass-produced microphone, mass-produced headphones, wearing a mass-produced shirt. And that means that even I'm relatively prosperous, can afford all this stuff, stuff that my parents frankly couldn't even dream of when they were my age. But if I want to, and do have the financial resources, I could go and contract with a friendly engineer and have them build me a custom pair of headphones. Unlike any other set of headphones ever on earth, they'd be made by hand. And if they did a really good job, they'd be maybe nearly as good as these mass-produced ones at 100 times the price. So if you want to, you can already go into the handmade Etsy economy where everything is kind of a little bit crap and a little bit weird, but it's quite a bit expensive. But if you want that human touch, you can pay for it right now.

Yeah, but almost no one buys that. We buy value, we buy utility, right? So my thesis is like one—

Well, enough people do. More people are making money through Etsy doing handcraft stuff than ever before.

You'll have to convince me on that. I haven't spoken to anyone rich off Etsy and talking to 5,000 founders and CEOs.

I know you're not going to get rich like handcrafting tea koozies on Etsy. That's not the point, right? The point is if you really want to handcraft tea koozies, it can pay for itself.

Sure. But okay, coming back to the question of AGI, right? Like, do you believe in UBI? I think right now is as far as my brain can think. It's a system that would pay these people that wouldn't have a job. Because as a CEO, I wouldn't hire someone because, oh, it's a human, nor would you, by the way. Like, I don't think you give much shit about that. You want to take care of your family first. You want to take care of your community after hiring a random someone rather than outsourcing to an AGI. time to have it done. I don't think guys like you and I will like be out of jobs 'cause we're so competitive and we're wired differently. We're wired for competition and like just not resting on our laurels.

Yeah.

That's 1%. But I also think that AGI would create so much value. And then there's probably the Neuralink aspect in that we could have Neuralink inside us. I think with human magic, it could create even something higher. I think there will be laws as well preventing AGI from doing X. And I think humans will be at the bottleneck of various decisions still, right? What do you think about that?

Well, I mean, hiring is really difficult. And so these AI tools give me the ability to let my existing employees level up way faster than they otherwise would, which is good for them, right? more responsibility and more experience and they learn faster and earn more money than they otherwise would. And they get to be part of a small and tight-knit and extremely effective team that's able to increase its output far faster than it would be able to if it just had to do everything by hand, including calculations and spreadsheets. So that's just kind of, that's the nature of the beast.

You're recruiting like the—

Yeah, but I mean, at the cutting edge of your economy, you're going to necessarily be employing 1 percenters. I mean, like, yeah, like it's not the case that you can go out and develop new technology just by like throwing darts at a crowd, right? Like, let's not bullshit ourselves here. Okay. As far as UBI goes, I think it's an idea that a lot of people have reached for over time. And I think in the past I was much more amenable to the idea, but currently I'm deeply concerned that it creates an incentive structure that that actually doesn't help bring out the best in humanity. And it's just too easy to pay annoying people to go away. And so I don't think that UBI is the answer to the problem. I think, well, I mean, aside from like in the current world, UBI is just fiscally not possible. It's just not possible. But even if it was possible, I think the better future to look forward to is one where But basically humans can find self-actualization doing a wide variety of things that are interesting to them, while the drudgery, boring bullshit, time-wasting, life-wasting jobs are abstracted away by AIs that do them to the extent they need to be done at all. That the economy is productive enough that the imbalance between income and the cost of goods and services is such that with essentially a wide range of incomes, you can you can purchase the things that you and I take for granted at relatively low prices. And so the effect is that pretty much everyone is universal high income, but not universal high income in the form of Donald Trump writing us all checks to feel good about ourselves. Universal high income in the sense that you can draw income or salary doing a wide variety of interesting things that's adequate to more than meet your needs. And that's actually much more productive Vision of the future, right?

Or some kind of value coin, right? Like in which you would be rewarded. We could calculate the value that you add to the greater group, like utilitarian style, and then you could be paid off that even if it's not an economical activity. We could calculate all of that biologically with sensors, right? I think that would make sense. That would—

Um, I mean, that's— the Soviets tried that in the '60s. The Soviets attempted to use early computers and linear programming to generalize labor theory value to basically calculate the true production cost of, and therefore price of everything in their economy. And it didn't work for a couple of basic reasons. But it turns out that markets are actually the mechanism you need to balance supply and demand in a distributed dynamic sense within an economy. And so, yeah, I think we already have a mechanism for economically deciding what the value of particular activity is. And it's just, it's called a market.

Do markets need supervision every now and then? What's your—

Markets have failures, of course. Markets have failures and markets need regulations in order to maintain a fair and open level playing field.

Like a random stallway, 80% free market, 20% intervention.

If we did 80/20 for the rest of humanity, I think that'd be just fine. Unfortunately, in the Anglophone West in particular, it's trending more towards 50/50, and that's obviously unsustainable.

Casey, thanks for being on again for a round 2. I'll let you go to your other activities. Where can people find out more about you?

You can find me posting on X.