How Bubbles Power Breakthroughs

0:00:02 (upbeat music)
0:00:07 Pushkin.
0:00:12 There are these moments when people make
0:00:17 huge technical advances and it happens all of a sudden.
0:00:20 Or at least it feels like it happens all of a sudden.
0:00:22 You know, this is happening right now,
0:00:26 most obviously with AI, with artificial intelligence.
0:00:28 It happened not too long ago with rockets
0:00:32 when SpaceX dramatically lowered the cost of getting to space.
0:00:36 Maybe it’s happening now with crypto.
0:00:39 I’d say it’s probably too soon to say on that one.
0:00:42 In any case, you can look at technological breakthroughs
0:00:44 in different fields and at different times.
0:00:48 And you can ask, what can we learn from these?
0:00:51 You can ask, can we abstract certain, you know,
0:00:54 certain qualities, certain tendencies
0:00:56 that seem to drive toward these bursts
0:00:58 of technological progress.
0:01:03 There’s a recent book called Boom that asks this question.
0:01:05 And it comes up with an interesting answer.
0:01:08 According to the book, one thing that’s really helpful,
0:01:13 if you wanna make a wild technological leap, a bubble.
0:01:16 (upbeat music)
0:01:21 I’m Jacob Goldstein and this is What’s Your Problem.
0:01:24 My guest today is Bern Hobart.
0:01:26 He’s the author of a finance newsletter
0:01:28 called The Diff, and he’s also the co-author
0:01:32 of a book called Boom, Bubbles and the End of Stagnation.
0:01:35 When Bern talks about bubbles,
0:01:37 he isn’t just talking about financial bubbles
0:01:40 where investors drive prices through the roof.
0:01:44 When he says bubble, largely he means social bubbles,
0:01:46 filter bubbles, little groups of people
0:01:49 who share some wild belief.
0:01:51 He really gets at what he means in this one sentence
0:01:54 where he and his co-author write, quote,
0:01:58 “Transformative progress arises from small groups
0:02:01 “with a unified vision, vast funding,
0:02:05 “and surprisingly poor accountability,” end quote,
0:02:07 basically living the dream.
0:02:10 Later in the conversation, Bern and I discussed
0:02:14 the modern space industry and cryptocurrency and AI,
0:02:16 but to start, we talked about two case studies
0:02:19 from the US in the 20th century.
0:02:21 Bern writes about him in the book and he argues
0:02:24 that these two moments hold broader lessons
0:02:26 for how technological progress works.
0:02:28 The two case studies we talk about
0:02:30 are the Manhattan Project and the Apollo Missions.
0:02:34 So let’s start with the Manhattan Project.
0:02:37 And maybe one place to start it is with this famous
0:02:42 1939 letter from Albert Einstein and other scientists
0:02:46 to FDR, the president, about the possibility
0:02:48 of the Nazis building an atomic bomb.
0:02:53 – Right, so that letter, it feels like good material
0:02:55 for maybe not a complete musical comedy,
0:02:56 but at least an act of a musical comedy
0:02:57 because the whole–
0:03:01 – It’s kind of a springtime for Hitler in Germany-ish.
0:03:02 – There is this whole thing
0:03:04 where you have this brilliant physicist,
0:03:08 but he is just kind of the stereotypical professor,
0:03:09 crazy hair, very absent-minded,
0:03:12 always talking about these things where no one,
0:03:14 no normal person can really understand
0:03:15 what he’s talking about.
0:03:19 And suddenly, instead of talking about space time
0:03:22 and that energy and matter in their relationship,
0:03:24 suddenly he’s saying someone could build
0:03:25 a really, really big bomb
0:03:28 and that person will probably be a German.
0:03:31 And that has some very bad implications
0:03:31 for the rest of the world.
0:03:34 – So now here we are, the president decides,
0:03:35 okay, we need to build a bomb,
0:03:38 we need to spend a wild amount of money on it.
0:03:40 And this is a thing that you describe as a bubble,
0:03:41 which is interesting, right?
0:03:44 ‘Cause it’s not a bubble in the sense of market prices,
0:03:45 it’s the federal government and the military,
0:03:49 but it has these other bubble-like characteristics
0:03:51 in your telling, right?
0:03:53 Maybe other meanings of bubble,
0:03:56 the way we talk about a social bubble or a filter bubble.
0:03:56 Tell me about that.
0:04:00 Like why is the Manhattan Project a kind of bubble?
0:04:01 – Why is it a bubble?
0:04:02 Because there’s that feedback loop,
0:04:06 because people take the idea of actually building the bomb
0:04:09 more seriously as other people take it more seriously.
0:04:12 And the more that you have someone like,
0:04:14 the more that you have people like Oppenheimer
0:04:16 actually dedicating time to the project,
0:04:17 the more other people think the project
0:04:19 will actually happen, this is actually worth doing.
0:04:20 So you have this group of people
0:04:24 who start taking the idea of building a bomb more seriously.
0:04:27 They treat it as a thing that will actually happen
0:04:29 rather than a thing that is hypothetically possible
0:04:31 if this particular equation is right,
0:04:32 if these measurements are right, et cetera.
0:04:34 And then they start actually designing them.
0:04:37 – The Manhattan Project seems like this sort of point
0:04:40 that gets a bunch of really smart people to coalesce
0:04:42 in one place on one project at one time, right?
0:04:45 It sort of solves the coordination problem.
0:04:49 The way, whatever you might say, AI today is doing that.
0:04:52 Like just brilliant people suddenly are all in one place
0:04:53 working on the same thing in a way
0:04:56 that they absolutely would not otherwise be.
0:04:57 – That is true.
0:05:00 And this was both within the US academic community
0:05:01 and then within the global academic community.
0:05:04 ‘Cause you had a lot of people who were in Central Europe
0:05:06 or Eastern Europe who realized that that is just
0:05:08 not a great place for them to be
0:05:13 and tried to get to the UK, US, other allied countries
0:05:15 as quickly as possible.
0:05:18 And so there was just this massive intellectual dividend
0:05:21 of a lot of the most brilliant people in Germany
0:05:25 and in Eastern Europe and in Hungary, et cetera.
0:05:29 They were all fleeing and all ended up in the same country.
0:05:32 So yeah, you have just this serendipity machine
0:05:35 where it was just, if you were a physicist,
0:05:37 it was an incredible place for just overhearing
0:05:40 really novel ideas and putting those ideas,
0:05:42 putting your own ideas to the test
0:05:44 because you had all the smartest people in the world
0:05:48 pretty much in this one little town in New Mexico.
0:05:52 – Right, so the Los Alamos piece is the famous part.
0:05:54 It’s the part one has heard of
0:05:57 with respect to the Manhattan Project.
0:05:59 There’s a less famous part that’s really interesting
0:06:01 and that also seems to hold some broader lessons as well.
0:06:06 And that is the kind of basically the manufacturing part.
0:06:09 How do we enrich enough uranium to build a bomb
0:06:11 if the physicists figure out how to design?
0:06:14 Talk about that piece and the lessons there.
0:06:16 – Yes, so that one, you’re right.
0:06:19 It is often under-emphasized in the history.
0:06:21 It was more of an engineering project
0:06:22 than a research project.
0:06:23 Though there was a lot of research involved.
0:06:26 The purpose was get enough of the,
0:06:27 get enough enriched uranium.
0:06:30 So the isotope that is actually prone
0:06:33 to these chain reactions.
0:06:38 Get it isolated and then be able to incorporate that
0:06:39 into a bomb.
0:06:41 They were also working on other physical materials
0:06:45 because there were multiple plausible bomb designs.
0:06:47 Some use different triggering mechanisms.
0:06:48 Some use different materials.
0:06:51 And there were also multiple plausible ways
0:06:54 to enrich enough of the fissile material
0:06:55 to actually build a bomb.
0:06:59 And so one version of the story is you just go down the list
0:07:01 and you pick the one that you think
0:07:03 is the most cost effective, most likely.
0:07:07 And so we choose one way to get just U-235
0:07:09 and we have one way to build a bomb.
0:07:11 – U-235 is the enriched uranium.
0:07:12 – Yes.
0:07:14 – And that, by the way, is the way normal businesses
0:07:16 do things in normal times.
0:07:18 You’re like, well, we got to do this really expensive thing.
0:07:19 We got to build a factory
0:07:20 and we don’t even know if it’s going to work.
0:07:23 Let’s choose the version that’s most likely to work.
0:07:25 Like that is the kind of standard move, right?
0:07:26 Yeah.
0:07:26 – Right.
0:07:29 And then the problem though is that if you try that
0:07:32 and you just, you got unlucky.
0:07:33 You picked the wrong bomb design
0:07:34 and the right fissile material
0:07:36 or right material is wrong bomb design.
0:07:39 You’ve done a lot of work which has zero payoff.
0:07:40 – And you’ve lost time, right?
0:07:43 Like crucially there is a huge sense of urgency
0:07:45 present at this moment
0:07:48 that is driving the whole thing really.
0:07:49 – Right.
0:07:51 We could also do more than one of them in parallel
0:07:52 and that is what we did.
0:07:53 And on the manufacturing side
0:07:55 that was actually just murderously expensive.
0:07:57 If you are building a factory
0:07:59 and you build the wrong kind of factory
0:08:03 then you’ve wasted a lot of money and effort and time.
0:08:05 So they did more than one.
0:08:07 They did several different processes
0:08:10 for enriching uranium and for plutonium.
0:08:11 – All at the same time, right?
0:08:13 And they knew they weren’t going to use all of them.
0:08:15 They just didn’t know which one was going to work.
0:08:18 So it was like, well, let’s try all of them at the same time
0:08:19 and hopefully one of them will work.
0:08:20 – Yes.
0:08:22 – Like that is super bubbly, right?
0:08:23 That is wild and expensive.
0:08:28 That is just throwing wild amounts of money at something
0:08:29 in a great amount of haste.
0:08:31 – Yes, yeah.
0:08:33 And if you, so if you believe
0:08:35 that there is this pretty linear payoff
0:08:37 then every additional investment you make
0:08:39 has, it doesn’t qualitatively change things.
0:08:41 It just means you’re doing a little bit more of it.
0:08:43 But if you believe there’s some kind of nonlinear payoff
0:08:47 where either this facility basically doesn’t work at all
0:08:49 or it works really, really well,
0:08:51 then when you diversify a little bit,
0:08:54 you do actually get just this better risk adjusted return
0:08:57 even though you’re objectively taking more risk.
0:08:58 – Interesting, right.
0:09:01 So in this instance, it’s if the Nazis have the bomb
0:09:04 before we do, it’s the end of the world as we know it.
0:09:05 – Yes.
0:09:06 – And so we better take a lot of risk
0:09:08 and that’s actually rational.
0:09:12 – It reminds me a little bit of aspects
0:09:13 of Operation Warp Speed.
0:09:17 I remember talking to Susan Athea, Stanford economist
0:09:19 early in the pandemic who was making the case
0:09:22 to do exactly this with vaccine manufacturing
0:09:24 in like early 2020.
0:09:26 We didn’t know if any vaccine was gonna work
0:09:28 and it takes a long time to build a factory
0:09:30 to make a vaccine basically or to tailor a factory.
0:09:32 And she was like, just make a bunch of factories
0:09:34 to make vaccines because if one of them works
0:09:37 we wanna be able to start working on it that day.
0:09:39 Like that seems quite similar to this.
0:09:40 And it worked.
0:09:42 – Yeah, yeah, I think that’s absolutely true.
0:09:45 That you, you know, the higher the stakes are,
0:09:47 the more you wanna be running everything
0:09:49 that can plausibly help in parallel.
0:09:52 And depending on the exact nature of what you’re doing,
0:09:54 there can be some spillover effects.
0:09:57 It’s possible that you build a factory
0:10:00 for manufacturing vaccine A and vaccine A doesn’t work out
0:10:02 but you can retrofit that factory
0:10:03 and start doing vaccine B.
0:10:05 And you know, there are little ways
0:10:06 to shuffle things around a bit.
0:10:10 But you often wanna go into this basically telling yourself,
0:10:13 if we didn’t waste money and we still got a good outcome,
0:10:14 it’s because we got very, very lucky
0:10:16 and that we only know we’re being serious
0:10:17 if we did in fact waste a lot of money.
0:10:21 And I think that kind of inverting your view of risk
0:10:22 is often a really good way to think
0:10:25 about these big transformative changes.
0:10:26 And this is actually another case
0:10:29 where the financial metaphors do give useful information
0:10:31 about just real world behaviors
0:10:34 because at hedge funds, this is actually something
0:10:36 that risk teams will sometimes tell portfolio managers
0:10:38 is you are making money on too high a percentage
0:10:39 of your trades.
0:10:41 This means that you are not making all the trades
0:10:43 that you could and if you made,
0:10:47 if you took your hit rate from 55% down to 53%,
0:10:49 we’d be able to allocate more capital to you
0:10:50 even though you’d be annoyed
0:10:52 that you were losing money on more trades.
0:10:54 – Interesting, because overall,
0:10:57 you would likely have a more profitable outcome
0:10:59 by taking bigger risks and incurring a few more losses
0:11:02 but your wins would be bigger and make up for the losses.
0:11:03 – Yes, and this kind of thinking, you know,
0:11:05 it’s very easy if you’re the one sitting behind the desk
0:11:07 just talking about these relative trade offs.
0:11:09 It’s a lot harder if you are the first person
0:11:11 working with Uranium in the factory
0:11:13 and we don’t quite know what the risks of that are
0:11:15 but it is just a, it’s a generally true thing
0:11:18 about trade offs that if you, about trade offs and risk
0:11:20 that there is an optimal amount of risk to take
0:11:23 that optimal amount is sometimes dependent
0:11:26 on what the downside risk of inaction is.
0:11:29 And so sometimes if you’re too successful,
0:11:32 you realize that you are actually messing something up.
0:11:33 – Yeah, you’re not taking enough risk.
0:11:37 So we all know how the Manhattan project ends, it worked.
0:11:42 I mean, it is a little bit of a weird one to start with,
0:11:45 you know, the basic ideas like technological progress
0:11:46 is good, risks are good.
0:11:47 And we’re talking about building the atomic bomb
0:11:49 and dropping it on two cities.
0:11:53 And it’s, you know, it’s morally a much easier question
0:11:55 if you think it’s the Nazis, sorry,
0:11:56 but the Nazis are absolutely the worst
0:11:59 and I definitely don’t want them to have a bomb first.
0:12:01 You know, there is the argument
0:12:02 that more people would have died
0:12:05 in a conventional invasion without the bomb.
0:12:07 I don’t know.
0:12:11 I mean, how do you, what do you make of it?
0:12:13 Like, obviously the book is very pro technological progress.
0:12:16 This show is basically pro technological progress.
0:12:19 But like the bomb isn’t a happy one to start on.
0:12:21 Like, what do you make of it ultimately?
0:12:23 – Yeah, it’s one of those things where you do,
0:12:27 it does make me wish that we could run the same simulation,
0:12:28 you know, a couple of million times
0:12:30 and see what the net, you know,
0:12:32 loss and save lives are different scenarios.
0:12:36 But one thing, the bomb, if you,
0:12:41 I guess from like a purely utilitarian standpoint,
0:12:44 I suspect that the there’ve been net lives saved
0:12:47 because of less use of coal for electricity generation,
0:12:49 more use of nuclear power.
0:12:51 And that is directly downstream of the bomb
0:12:53 that you can build these, you know,
0:12:57 by design uncontrollable releases of atomic energy.
0:12:59 You can also build more controllable ones.
0:13:02 And then getting the funding for that would be a lot harder.
0:13:05 – And presumably we got nuclear power much sooner
0:13:06 than we otherwise would have
0:13:09 because the incredibly rapid progress
0:13:12 of the Manhattan Project, that’s the case there.
0:13:12 Fair.
0:13:15 – Which I don’t, I don’t think, you know,
0:13:19 if you let me push the button on what I drop an atomic bomb
0:13:22 on a civilian population in exchange
0:13:25 for fewer people dying of respiratory diseases
0:13:27 over the next couple of decades, you know,
0:13:29 I would have to give it a lot of thought.
0:13:32 – I don’t, I’m not gonna, I’m not gonna push that button,
0:13:34 but I’m never gonna have a job where I have to decide
0:13:35 ’cause I can’t deal.
0:13:39 Okay, a thing you mentioned in the book,
0:13:41 kind of in passing that was really interesting
0:13:46 and surprising to me was that nuclear power today
0:13:51 accounts for 18% of electric power generation in the US.
0:13:54 18%, like that is so much higher than I would have thought
0:13:57 given sort of how little you hear
0:13:59 about existing nuclear power plants, right?
0:14:01 Like that is a lot.
0:14:05 – Yeah, yeah, it is a surprisingly,
0:14:06 it’s a surprisingly high number,
0:14:09 but also nuclear power, it is one of the most annoying
0:14:12 technologies to talk about in the sense that
0:14:14 it doesn’t do anything really, really exciting
0:14:17 other than provide essentially unlimited power
0:14:19 with minimal risk.
0:14:24 – And some amount of, some amount of scary tail risk,
0:14:25 right? – Yes.
0:14:27 – Like, I mean, that is what is actually interesting
0:14:30 to talk about, sort of unfortunately for the world
0:14:32 given that it has a lot of benefits.
0:14:34 There is this tail risk and once in a while
0:14:36 something goes horribly wrong.
0:14:39 Even though on the whole, it seems to be clearly less risky
0:14:41 than say a coal-fired power plant.
0:14:46 – Right, and the industry has, they’re aware of those risks
0:14:49 and nobody wants to be responsible for that kind of thing
0:14:51 and nobody wants to be testifying before Congress
0:14:54 about ever having cut any corner whatsoever
0:14:56 in the event that a disaster happens.
0:14:59 So they do actually take that incredibly seriously
0:15:03 and so nuclear power does end up being in practice
0:15:05 much safer than other power sources.
0:15:07 And then you add in the externality
0:15:09 of doesn’t really produce emissions
0:15:13 and uranium exists in some quantities just about everywhere.
0:15:16 – Yeah, no climate change, no local air pollution
0:15:19 has a lot going for it, always on.
0:15:22 Okay, let’s go to the moon.
0:15:26 So you write also about the Apollo missions,
0:15:28 US going to the moon.
0:15:32 It’s the early ’60s, was it ’61.
0:15:35 Kennedy says we’re gonna go to the moon
0:15:36 by the end of the decade.
0:15:39 There’s the Cold War context.
0:15:41 Kennedy announces this goal.
0:15:45 What’s the response in the US when Kennedy says this?
0:15:46 – Yeah, so a lot of the response,
0:15:50 at first people are somewhat hypothetically excited
0:15:52 as they start realizing how much it will cost,
0:15:54 they go from not especially excited
0:15:57 to actually pretty deeply opposed.
0:15:59 And this shows up in,
0:16:01 there was someone coined the term moon doggal.
0:16:03 – Yeah, moon doggal, I loved moon doggal,
0:16:05 I learned that from the book.
0:16:08 It was Norbert Wiener, like a famous technologist,
0:16:11 not a crank, right?
0:16:12 Somebody who knew what he was talking about
0:16:16 was like, this is a crazy idea, it’s a moon doggal.
0:16:17 – Right, and this really worked its way
0:16:18 to popular culture.
0:16:23 If you go on Spotify and listen to the Tom Lehrer song,
0:16:26 Werner von Braun, the recording that Spotify has,
0:16:28 it opens with a monologue that is talking about
0:16:30 how stupid the idea of the Apollo program is.
0:16:35 It’s, and this is again someone who is in academia,
0:16:37 who’s a very, very sharp guy,
0:16:40 and who just feels like he completely sees
0:16:43 through this political giveaway program
0:16:45 to big defense contractors
0:16:47 and knows that there’s no point in doing this.
0:16:51 You write that NASA’s own analysis
0:16:55 found a 90% chance that a failure,
0:16:57 a failing to reach the moon by the end of the decade.
0:17:00 Like it wasn’t just outside people being critical,
0:17:03 it was NASA itself didn’t think it was good work.
0:17:06 There’s a phrase you use in the book
0:17:09 to talk about these sort of bubble-like environments
0:17:10 that are of interest to you.
0:17:12 And I found it really interesting,
0:17:14 and I think we can talk about it
0:17:15 in the context of Apollo.
0:17:18 That phrase is definite optimism.
0:17:19 Tell me about that phrase.
0:17:21 – Yes, so definite optimism is the view
0:17:24 that the future can and will be better
0:17:27 in some very specific way.
0:17:30 That there will be, there is something we cannot do now,
0:17:31 we will be able to do it in the future,
0:17:33 and it will be good that we can do it.
0:17:34 – And why is it important?
0:17:38 Like it’s a big deal in your telling in an interesting way.
0:17:41 Why is it so important?
0:17:43 – It’s important because that is what allows you
0:17:45 to actually marshal those resources,
0:17:49 whether those are the people or the capital
0:17:51 or the political pull to put them
0:17:53 all in some specific direction
0:17:54 and say, we’re going to build this thing,
0:17:56 so we need to actually go step by step
0:17:59 and figure out what specific things have to be done,
0:18:01 what discoveries have to be made,
0:18:04 what laws have to be passed in order for this to happen.
0:18:06 And that is, so it’s definite optimism
0:18:07 in the sense that you’re saying
0:18:09 there is a specific thing we’re going to build.
0:18:11 It’s the kind of thing that can keep you going
0:18:13 when you encounter temporary setbacks.
0:18:15 And that’s where the optimism part comes in,
0:18:20 because if you have a less definitely optimistic view
0:18:22 about that project, you might say the goal
0:18:24 of the Apollo program is to figure out
0:18:27 if we can put a person on the moon.
0:18:29 But I think what that leaves you open to
0:18:31 is the temptation to give up at any point.
0:18:35 ‘Cause at any point you can have a botched launch
0:18:39 or an accident or you’re designing some component
0:18:41 and the math just doesn’t pencil out
0:18:43 you need, you know, it’s going to weigh too much
0:18:46 to actually make it on track.
0:18:47 And you could say, okay, well that’s how we figured out
0:18:48 that we’re not actually doing this.
0:18:52 But if you do just have this kind of delusional view
0:18:54 that know if there’s a mistake,
0:18:55 it’s a mistake in my analysis,
0:18:58 not in the ultimate plan here,
0:18:59 and that it is physically possible,
0:19:01 we just have to figure out all the details,
0:19:03 then I think that does set up a different kind of motivation
0:19:06 because at that point, you can view every mistake
0:19:09 as just exhausting the set of possibilities
0:19:10 and letting you narrow things down
0:19:12 to what is the correct approach.
0:19:14 What you sort of needed was this sort of
0:19:16 very localized definite optimism
0:19:20 where you could imagine a researcher thinking to themselves
0:19:23 or an engineer or someone throughout the project
0:19:25 thinking to themselves that, okay,
0:19:28 this will probably not work overall.
0:19:30 But the specific thing I’m working on,
0:19:31 whether it is designing a space suit
0:19:33 or designing this rocket
0:19:35 or programming the guidance computer
0:19:39 that one, I could tell that my part is actually going to work
0:19:40 or at least I believe that I can make it work.
0:19:42 And two, this is my only chance
0:19:43 to work with these really cool toys.
0:19:46 So if the money is going to be wasted at some point,
0:19:48 let that money be wasted on me.
0:19:51 And I think that that kind of attitude of just,
0:19:52 you know that you have one shot
0:19:54 to actually do something really interesting,
0:19:56 you will not get a second chance.
0:19:58 If everyone believes that it does become
0:19:59 a coordinating mechanism
0:20:00 where now they’re all working extremely hard,
0:20:04 they all recognize that the success of what they’re doing
0:20:05 is very much up to them.
0:20:07 And then that ends up contributing
0:20:08 to this group’s success.
0:20:13 So it’s like this, if I’m going to do this,
0:20:14 I got to do it now.
0:20:15 Everybody’s doing it now.
0:20:16 We got the money now.
0:20:18 This is our one shot.
0:20:19 We better get it right.
0:20:21 We better do everything we can to make it work.
0:20:23 – Yes, fear of missing out.
0:20:25 – Yeah, FOMO, right.
0:20:25 So FOMO, it’s funny.
0:20:29 People talk about that as like a dumb investment thesis,
0:20:29 basically, right?
0:20:31 It’s like a meme stock idea,
0:20:36 but you talk about it in these more interesting contexts,
0:20:36 basically, right?
0:20:38 More meaningful, I would say.
0:20:43 – Yeah, so in the purely straightforward way,
0:20:45 the idea is there are sometimes
0:20:47 these very time-limited opportunities to do something.
0:20:49 And if you’re capable of doing that thing,
0:20:50 this may be your only chance.
0:20:51 And so missing out is actually something
0:20:53 you should be afraid of.
0:20:56 So if you actually have a really clever idea
0:20:58 for an AI company,
0:20:59 this is actually a time where you can at least
0:21:00 make the attempt.
0:21:03 So yeah, we do argue that missing out
0:21:04 is something you should absolutely fear.
0:21:06 – So what happens with the Apollo project?
0:21:10 Just in brief, like talking about just how big it is
0:21:13 and how risky it is, like it’s striking, right?
0:21:14 – Right, yeah.
0:21:16 So it was running, the expenses were running
0:21:20 on like a low single-digit percentage of GDP for a while.
0:21:23 So a couple percent of the value of everything
0:21:25 everybody in the country does
0:21:27 is going into the Apollo mission.
0:21:31 Just this one plainly unnecessary thing
0:21:33 that the government has decided to do.
0:21:35 – Right, and this is one of the cases
0:21:37 where there was a very powerful spillover effect
0:21:41 because the Apollo guidance computer
0:21:44 needed the most lightweight and least power consuming
0:21:47 and most reliable components possible.
0:21:49 And if you were building a computer conventionally
0:21:51 at that time and you had a budget,
0:21:53 you would probably build it out of vacuum tubes.
0:21:55 And you knew that the vacuum tubes,
0:21:56 they’re bulky, they consume a lot of power,
0:21:57 they throw off a lot of heat,
0:21:59 they burn out all the time,
0:22:01 but they are fairly cheap.
0:22:06 But in this case, there was an alternative technology.
0:22:08 It was extremely expensive,
0:22:12 but it was lightweight, didn’t use a lot of power
0:22:13 and did not have moving parts.
0:22:15 And that’s the integrated circuit.
0:22:17 So transistor-based computing.
0:22:19 – The chip, well, we know today as the chip.
0:22:21 – Yes, the chip.
0:22:23 – You read that in 1963,
0:22:28 NASA bought 60% of the chips made in the United States.
0:22:32 Just NASA, not the whole government, just NASA, 60%.
0:22:33 – They actually bought more chips than they needed
0:22:37 because they recognized that the chip companies
0:22:41 were run by very, very nice electrical engineering nerds
0:22:43 who just love designing tiny, tiny things
0:22:46 and that these people just don’t know how to run a business.
0:22:49 And so they were worried that Fairchild Semiconductor
0:22:51 would just run out of cash at some point.
0:22:54 And then NASA would have half of a computer
0:22:56 and no way to build the rest of it.
0:22:57 So they actually over-ordered.
0:23:00 They used integrated circuits for a few applications
0:23:02 that actually were not so dependent
0:23:04 on power consumption and weight and things.
0:23:06 So that critique of the Apollo program
0:23:07 was directionally correct.
0:23:10 It was money being splashed out to defense contractors
0:23:11 who were favored by the government.
0:23:12 But in this case, it was being done
0:23:15 in a more strategic and thoughtful way
0:23:17 and kind of kept the industry going.
0:23:21 – So you talk a fair bit in the book
0:23:26 about the sort of religious and quasi-religious aspects
0:23:28 of these little groups of people
0:23:30 that come together in these bubble-like moments
0:23:31 to do these big things.
0:23:35 And that’s really present in the Apollo section.
0:23:39 Like talk about the sort of religious ideas
0:23:41 associated with the Apollo mission
0:23:43 that the people working on the mission had.
0:23:46 – Yeah, I mean, you name it after a Greek God
0:23:49 and you’re already starting a little bit religious.
0:23:54 So there were people who worked on these missions
0:23:56 who felt like this is part of mankind’s destiny,
0:23:58 is to explore the stars
0:23:59 and that there’s this whole universe
0:24:01 that is a universe created by God.
0:24:02 And it would be kind of weird.
0:24:04 We can’t second-guess the divine,
0:24:06 but it’s a little weird for God to create
0:24:07 all of these astronomical bodies
0:24:09 that just kind of look good from the ground
0:24:11 and that you’re not actually meant to go visit.
0:24:13 – You talk about somewhat similar things
0:24:18 in other kind of less obviously spiritual dimensions
0:24:20 of people coming together
0:24:24 and having it kind of more than rational.
0:24:27 You use this word thymos from the Greek meaning spirit.
0:24:29 Like what’s going on there more broadly?
0:24:31 Why is that important more generally
0:24:33 for technological progress?
0:24:38 – Because the, so thymos is part of this tripartite model
0:24:42 of the soul where you have your appetites and your reason
0:24:45 and then your thymos show like your longing for glory
0:24:50 and honor and this kind of transcendent achievement.
0:24:54 And logos reasoning, it only gets you so far.
0:24:55 You can reason your way
0:24:57 into some pretty interesting things,
0:24:58 but at some point you do decide
0:25:01 that the reasonable thing is probably
0:25:03 to take it a little bit easy
0:25:06 and not take certain risks.
0:25:10 And it still is just this pursuit of something greater
0:25:13 and something beyond the ordinary,
0:25:14 something really beyond the logos, right?
0:25:16 Like beyond what you could get
0:25:19 to just by reasoning one step at a time.
0:25:24 And I think that that is just a deeply attractive proposition
0:25:26 to many people.
0:25:29 And it’s also a scary one because at that point,
0:25:31 if you’re doing things that are beyond
0:25:34 what is the rational thing to do,
0:25:35 then of course you have no rational explanation
0:25:38 for what you did wrong if you mess up.
0:25:40 And you are sort of betting
0:25:41 on some historical contingencies.
0:25:43 – That’s the definite optimism part, right?
0:25:45 – Metting on historical contingencies
0:25:48 is another way of saying definite optimism, right?
0:25:52 – So, back to the moon.
0:25:55 So we get to the moon, in fact, against all odds,
0:25:57 we make it.
0:26:01 And there’s this moment where it’s like,
0:26:05 today the moon, tomorrow, the solar system.
0:26:07 But in fact, it was today the moon,
0:26:09 tomorrow, not even the moon.
0:26:10 – Right.
0:26:11 – Like what happened?
0:26:15 – Well, you had asked about what these mega projects
0:26:16 have in common with financial bubbles.
0:26:17 And one of the things they have in common is,
0:26:19 sometimes there’s a bust.
0:26:21 And sometimes that bust is actually an overreaction
0:26:23 in the opposite direction.
0:26:28 And people take everything they believed
0:26:33 in, say, 1969 about humanity’s future and the stars,
0:26:35 and they say, okay, this is exactly the opposite
0:26:37 of where things will actually go,
0:26:38 and the exact opposite of what we should care about,
0:26:40 that we have plenty of problems here on Earth,
0:26:43 and why would we, do we really wanna turn Mars
0:26:46 into just another planet that also has problems
0:26:49 of racism and poverty and nuclear war and all that stuff?
0:26:53 So maybe we should stay home and fix our own stuff.
0:26:54 In public policy, you’d actually need for there
0:26:58 to be some kind of resurgence in belief and space.
0:26:59 You need some kind of charismatic story.
0:27:02 And perhaps to an extent, we have that right now.
0:27:03 – Yes.
0:27:05 – Maybe Elon’s not the perfect front man for all of this,
0:27:07 but he is certainly someone who demonstrates
0:27:10 that space travel, it can be done, it can be improved,
0:27:13 and that it’s just objectively cool.
0:27:16 That it is just hard to watch a SpaceX launch video
0:27:18 and not feel something.
0:27:21 – Yes, so good.
0:27:26 I wanna talk more about space in a minute.
0:27:28 So it’s interesting, these two stories
0:27:30 that are kind of in the middle of your book,
0:27:31 they’re kind of the core of the book, right?
0:27:34 These two interesting moments that are non-financial bubbles
0:27:37 when you have these incredible technological innovation
0:27:40 in a short amount of time that seems unrealistic,
0:27:44 unrealistically fast, impressive outcome.
0:27:48 And they’re both pure government projects.
0:27:51 They’re both command and control economy.
0:27:55 It is not the private sector, it is not capitalism.
0:27:57 What do you make of that?
0:28:00 – I would say there’s a very strong indirect link
0:28:02 for a couple of reasons.
0:28:07 One is just the practical kind of,
0:28:08 the practical kind of reason that personnel is policy
0:28:13 and that the US government in the 1930s,
0:28:14 the US government was hiring
0:28:16 and the private sector mostly wasn’t.
0:28:18 And so all the ambitious people,
0:28:20 basically all the ambitious people in the country
0:28:21 tried to get government jobs.
0:28:23 And that is usually not the case.
0:28:25 And there are certainly circumstances
0:28:27 where that’s a really bad sign.
0:28:28 But in this case, it was great.
0:28:30 It meant that there were a lot of New Deal projects
0:28:31 that were staffed by the people
0:28:34 who would have been rising up the ranks at RCA
0:28:36 or General Electric or something a decade earlier.
0:28:38 Now they’re running New Deal projects instead.
0:28:40 And they’re again, rising up the ranks really fast,
0:28:42 having a very large real-world impact
0:28:43 very early in their careers.
0:28:47 And those people had been working together for a while
0:28:48 and they knew each other.
0:28:49 There was a lot of just institutional knowledge
0:28:52 about how to get big things done
0:28:53 within the US government.
0:28:55 And a lot of that institutional knowledge
0:28:56 could then be redirected.
0:28:59 So you have the New Deal and then the war effort.
0:29:00 And then you have this post-war economy
0:29:02 where there’s still, it takes a while
0:29:04 for the government to fully relax its control.
0:29:07 And then very soon we’re into the Korean War.
0:29:11 So, yeah, there was just a large increase in state capacity
0:29:13 and just in the quality of people making decisions
0:29:15 within the US government in that period.
0:29:20 – We’ll be back in a minute
0:29:23 to talk about bubble-esque things happening right now.
0:29:27 Namely, rockets, cryptocurrency, and AI.
0:29:30 (upbeat music)
0:29:37 – Okay, now to space today.
0:29:41 Bern and I talked about SpaceX in particular
0:29:42 because, you know, it really is the company
0:29:45 that launched the modern space industry.
0:29:48 And there’s this one key trait that SpaceX shares
0:29:51 with the other projects Bern wrote about in the book.
0:29:55 It brought together people who share a wild dream.
0:29:56 If you go to work at SpaceX,
0:29:58 it’s probably because you believe
0:30:01 in getting humanity to Mars.
0:30:03 – Yeah, yeah, it’s not just that you believe in the dream,
0:30:06 but when you get the job, you’re suddenly in an environment
0:30:07 where everyone believes in the dream.
0:30:09 And if you’re working in one of those organizations,
0:30:11 you’re probably not working nine to five,
0:30:15 which means you have very few hours in your day or week
0:30:17 where you’re not completely surrounded by people
0:30:19 who believe that humanity will,
0:30:21 people will be living on Mars
0:30:24 and that this is the organization that will make it happen.
0:30:26 And that just has to really mess with your mind.
0:30:30 Like, what is normal to an engineer
0:30:32 working at SpaceX in 2006 is completely abnormal
0:30:35 to 99.9% of the human population.
0:30:37 And, you know, most of the exceptions
0:30:38 are like six-year-old boys
0:30:40 who just watched Star Wars for the first time,
0:30:41 go Mars, it’s crazy.
0:30:43 – Yeah, I mean, really, as I went through the book,
0:30:45 I was like, oh, really the bubble you’re talking about
0:30:48 is a social bubble, like the meaningful bubble.
0:30:50 Like maybe there’s a financial bubble attached to it.
0:30:52 Maybe there isn’t, but what really matters
0:30:54 is you’re in this weird little social bubble
0:30:58 that believes some wild thing together
0:30:59 that believes it is not wild,
0:31:00 that believes it is gonna happen.
0:31:02 Like, that’s the thing.
0:31:03 – Yeah.
0:31:05 – And has money, and has the money
0:31:07 to act on their wild belief.
0:31:08 – Yes.
0:31:10 And so, you know, getting the money
0:31:13 does mean interacting with the normie sphere,
0:31:15 interacting with people who don’t quite buy into all of it,
0:31:20 but they, when you have these really ambitious plans
0:31:21 and you’re taking them seriously,
0:31:23 you’re doing them step by step,
0:31:25 some of those steps do have other practical applications.
0:31:27 And so that is the basic story.
0:31:29 It was not just straight shot,
0:31:32 we are going to invest all the money Elon got from PayPal
0:31:32 into going to Mars,
0:31:34 and hopefully we get to Mars before we run out.
0:31:35 – Yeah.
0:31:38 – It was, you know, we’re going to build these prototypes,
0:31:39 we’re going to build reusable rockets,
0:31:43 we’re going to use those for existing use cases,
0:31:45 and we will probably find new use cases.
0:31:47 And then once we get really, really good at launching things
0:31:50 cheaply, well, there are a lot of satellites out there,
0:31:52 and perhaps we should have some of our own.
0:31:53 And if we can do it at sufficient scale,
0:31:56 then maybe we can just throw a global communications network
0:31:59 up there in the sky and see what happens next.
0:32:02 So, yeah, that’s, you know, the intermediate steps,
0:32:05 each one, it’s basically taking the thermos like the spirited,
0:32:08 you know, here’s our grand vision of the future,
0:32:09 and you know, here’s my destiny,
0:32:11 and I was put on Earth to do this and say,
0:32:13 okay, well, the next step is have enough money
0:32:14 to pay rent next month.
0:32:17 – Right, what do I gotta do tomorrow to get to Mars?
0:32:22 So, is there a space bubble right now?
0:32:25 – I think so, I think there is,
0:32:30 I think there are people who look at SpaceX and say,
0:32:32 this is achievable, and that more is achievable.
0:32:34 They also look at SpaceX and say,
0:32:36 this is a kind of infrastructure
0:32:40 that there are things like doing manufacturing in orbit,
0:32:41 or doing manufacturing on the moon,
0:32:44 where in some cases that is actually the best place
0:32:45 to build something.
0:32:48 – Basically, because SpaceX has driven down the cost
0:32:50 so much of getting stuff into orbit,
0:32:54 new ideas that would have been economically absurd
0:32:58 20 years ago, like manufacturing in space are now plausible.
0:33:00 And so this is sort of bubble building on itself.
0:33:03 And like, why is it not just an industry now?
0:33:05 Why is it a bubble in your telling?
0:33:10 – It is the feedback loop where what SpaceX does
0:33:11 makes more sense if they believe
0:33:12 that there will be a lot of demand
0:33:16 to move physical things off of Earth and into orbit,
0:33:18 and perhaps further out.
0:33:21 If they believe that there’s more demand for that,
0:33:23 they should be investing more in R&D,
0:33:25 they should be building bigger and better rockets,
0:33:29 and they should be doing the big fixed cost investment
0:33:31 that incrementally reduces the cost of launches,
0:33:33 and only pays for itself if you do a lot of them.
0:33:35 And then if they’re doing that,
0:33:37 and you have your dream of,
0:33:39 we’re going to manufacture drugs in space,
0:33:42 and they will be, like the marginal cost is low,
0:33:44 once you get stuff up there.
0:33:45 Well, that dream is a little bit more plausible
0:33:47 if you can actually plot that curve
0:33:49 of how much does it cost to get a kilogram into space,
0:33:54 and say, there is a specific year at which point
0:33:56 we would actually have the cost advantage
0:33:57 versus terrestrial manufacturing.
0:34:00 – So it’s this sort of coordinating mechanism
0:34:03 that like you also write about with Microsoft and Intel
0:34:04 in the like 80s, 90s, where it’s like,
0:34:06 oh, they’re building better chips,
0:34:08 so we’ll build better software.
0:34:10 And then because they’re building better software,
0:34:11 we’ll build better chips.
0:34:14 So this is like a more exciting version of that, right?
0:34:17 Because it’s going to get even cheaper
0:34:18 to send stuff to space.
0:34:22 We can build this crazy factory to exist in space.
0:34:23 And then that tells SpaceX,
0:34:27 oh, we can in fact keep building, keep innovating,
0:34:28 keep spending money.
0:34:32 – Yes, and so someone has to do just half of that,
0:34:34 like the half of that, that makes no sense whatsoever.
0:34:36 – That was SpaceX at the beginning, right?
0:34:39 That was like just a guy with a lot of money
0:34:40 and a crazy dream.
0:34:42 – Yeah, it just really helps to have someone
0:34:44 who’s eccentric and has a lot of money
0:34:46 and is willing to throw it at a lot of different things.
0:34:49 Like Musk, he spent some substantial fraction
0:34:52 of his net worth right after this PayPal sale
0:34:54 on a really nice sports car.
0:34:57 And then immediately took it for a drive and wrecked it,
0:34:59 had no insurance and was not wearing a seatbelt.
0:35:02 So the Elon Musk story could have just been this proverb
0:35:05 about dot com excess and what happened
0:35:07 when you finally gave these people money
0:35:09 as they immediately bought sports cars and wrecked them.
0:35:12 Instead, it’s a story about a different kind of excess,
0:35:15 but it’s still, I guess what that illustrates
0:35:17 is the risk tolerance. – Risk-seeking, yes, yes.
0:35:22 – Yeah, there’s a risk level where you are going for a joy
0:35:25 ride in your $2 million car and you haven’t bothered
0:35:28 to fill out all the paperwork or buy the insurance,
0:35:29 and that is the risk tolerance
0:35:32 of someone who starts a company like SpaceX.
0:35:35 – Okay, enough about space.
0:35:38 Let’s talk about crypto, formerly known as cryptocurrency.
0:35:42 Let’s talk about Bitcoin, and let’s talk about Bitcoin,
0:35:44 especially at the beginning, right,
0:35:47 before it was number go up,
0:35:50 when it was, it really was true believers, right?
0:35:52 It was people who had a crazy worldview,
0:35:55 like you’re talking about in these other contexts.
0:35:59 – Yes, so we still don’t know for sure
0:36:01 who Satoshi Nakamoto was,
0:36:02 and I think everyone in crypto
0:36:05 has at least one guess, sometimes many guesses,
0:36:07 but whoever Satoshi was, whoever they were,
0:36:09 this is the creator of Bitcoin
0:36:11 for the one person who doesn’t know yet.
0:36:16 – They had this view that one of the fundamental problems
0:36:20 in the world today is that if you are going to transfer value
0:36:21 from one party to another,
0:36:23 you need some trusted intermediary.
0:36:25 – You need a trusted intermediary
0:36:28 like a government and a bank.
0:36:31 Typically, in money, you need both governments and banks
0:36:33 the way it works in the world today, right?
0:36:33 – Yes.
0:36:35 – And Satoshi happened to publish the Bitcoin White Paper
0:36:39 in October, 2008, which was a great moment to find people
0:36:42 who really didn’t want to have to deal with governments
0:36:44 and banks when they were dealing with money,
0:36:45 at the financial crisis, right?
0:36:48 Right in the teeth of the financial crisis.
0:36:50 – Yes, so it is in one sense
0:36:52 just this technically clever thing.
0:36:54 And then in another sense,
0:36:55 it’s this very ideological project
0:36:57 where he doesn’t like central banks,
0:36:59 he doesn’t like regular banks.
0:37:02 He feels like all of these institutions are corrupt
0:37:05 and your money is just an entry in somebody’s database
0:37:07 and they can update that database tomorrow
0:37:09 and either change how much you have
0:37:10 or change what it’s worth.
0:37:13 And we need to just build something new from a clean slate.
0:37:14 And there’s also,
0:37:17 I think there’s this tendency among a lot of tech people to,
0:37:21 when you look at any kind of communications technology
0:37:23 and money broadly defined as a communication technology,
0:37:24 you’re always looking at something
0:37:25 that has evolved from something simple
0:37:28 and it has just been patched and altered and edited
0:37:32 and tweaked and so on until it works the way that it works.
0:37:35 But that always means that you can easily come up
0:37:36 with some first principles view
0:37:39 that’s a whole lot cleaner, easier to reason about,
0:37:40 omits some mistakes.
0:37:42 And then you often find that, okay,
0:37:43 you omitted all the mistakes
0:37:44 that are really, really salient about fiat,
0:37:46 but then you added some brand new mistakes
0:37:49 or added mistakes that we haven’t made in hundreds of years.
0:37:51 And so they’re, it’s full of trade-offs.
0:37:53 – It gets complicated, but at the beginning, right?
0:37:57 So the white paper comes out and I covered,
0:37:59 I did a story about Bitcoin in 2011,
0:38:01 which was still quite early.
0:38:05 We had shocked that it had gone from $10 a Bitcoin
0:38:07 to $20 a Bitcoin, thought we were reading it wrong.
0:38:10 And at that time, like I talked to Gavin Andreessen,
0:38:13 who was very early in the Bitcoin universe,
0:38:15 like he was not in it to get rich, right?
0:38:17 Like he really believed, he really believed in it.
0:38:20 And that was the vibe then.
0:38:23 And like he thought it was gonna be money, right?
0:38:27 The dream was people will use this to buy stuff.
0:38:30 And one thing that is interesting to me
0:38:34 is yes, some people sort of use it to buy stuff,
0:38:36 but basically not, right?
0:38:39 Like that, it would go from $20 a Bitcoin
0:38:43 to $100,000 a Bitcoin without some crazy killer app,
0:38:44 without becoming the web,
0:38:47 without becoming something that everybody uses,
0:38:48 whether they care about it or not.
0:38:50 That I would not have guessed.
0:38:52 And it seems weird.
0:38:54 And plainly now crypto is full of some people
0:38:55 who are true believers and a lot of people
0:38:56 who just want to get rich.
0:38:59 And some of whom are pretty scammy.
0:39:01 – Yeah, yeah, there’s like the grifter coefficient
0:39:03 always goes up with the price.
0:39:05 And then the true believers are still there
0:39:07 during the next 80% drawdown.
0:39:09 And I’m sure there will be a drawdown,
0:39:10 something like that, at some point in the future.
0:39:12 It’s just that that’s kind of the nature
0:39:14 of these kinds of assets.
0:39:18 Bitcoin, it was originally conceived as more of a currency.
0:39:21 And Satoshi talked about some hypothetical products
0:39:23 you could buy with it.
0:39:27 And then the first Bitcoin killer app, to be fair,
0:39:29 was e-commerce, it was specifically drugs.
0:39:31 – Yeah, crime. – Yes.
0:39:33 – It is a very, very libertarian project in that way.
0:39:36 So it doesn’t work very well as a dollar substitute
0:39:39 for many reasons, most of the obvious reasons.
0:39:42 But it is interesting as a gold substitute
0:39:44 where part of the point of gold
0:39:46 is that it is very divisible
0:39:48 and your gold is the same as my gold.
0:39:51 And we’ve all kind of collectively agreed
0:39:54 that gold is worth more than its value
0:39:56 as just an industrial product.
0:40:00 And then the neat thing about gold is
0:40:01 it’s really hard to dig up anymore.
0:40:03 Gold supply is extremely inelastic.
0:40:07 – And Bitcoin is designed to have a finite supply,
0:40:08 right? – Yes.
0:40:09 – It’s an important analogy, yeah.
0:40:09 – Yes.
0:40:12 – More generally, like, it’s what?
0:40:13 It’s a long time out now.
0:40:17 It’s 17 years or something since the white paper.
0:40:24 What do you make of the sort of costs and benefits
0:40:25 of cryptocurrency so far?
0:40:27 The costs are more obvious to me.
0:40:29 Like there’s a lot of grift.
0:40:32 It’s, you know, by design, very energy intensive.
0:40:35 Like I’m open to like better payment systems.
0:40:38 There’s lots of just like boring efficiency gains
0:40:42 you would think we could get that we haven’t gotten, right?
0:40:42 – Yeah.
0:40:43 – What do you think about the costs
0:40:45 versus the benefits so far?
0:40:49 – I think in terms of the present value of future gains,
0:40:50 probably better off.
0:40:52 I think in terms of, yeah, realized gains so far, worse off.
0:40:53 – Uh-huh, uh-huh.
0:40:55 So basically worse off so far,
0:40:57 but in the long run, we’ll be better off.
0:40:59 We just haven’t got the pay off yet.
0:41:01 This is actually something that general purpose technologies,
0:41:03 it is a feature of general purpose technologies
0:41:05 that there’s often a point early in their history
0:41:07 where the net benefit has been negative.
0:41:11 – What would make it clearly positive?
0:41:13 Like what’s the killer return you’re hoping to see
0:41:15 from cryptocurrency?
0:41:17 – Yeah, so I think the killer return would be
0:41:20 if there is a financial system that is open
0:41:23 in the sense that starting a financial institution,
0:41:25 starting a bank or an insurance company or something
0:41:27 is basically you write some code
0:41:31 and you click the deploy button and your code is running,
0:41:33 you have capitalized your little entity
0:41:35 and now you can provide whatever it is,
0:41:37 like mean tweet insurance.
0:41:38 You’re selling people for a dollar a day,
0:41:42 you’ll pay them $100 if there’s a tweet that makes them cry.
0:41:43 You know, that kind of thing–
0:41:45 – Weird incentives in your insurance business,
0:41:46 I’m gonna tell you right now.
0:41:48 – You get to speed run all kinds of financial history,
0:41:50 I’m sure you learn all about adverse selection,
0:41:52 but like a financial system where anything,
0:41:54 anything can be plugged into something else
0:41:57 and basically everything is an API call away.
0:41:59 It’s just a really interesting concept
0:42:02 and the fiat system is moving in that direction, but slowly.
0:42:05 – And just to be clear, like why is it,
0:42:08 why is that better on balance?
0:42:10 So for it to be net positive,
0:42:12 that has to be not only interesting,
0:42:15 but that has to like lead to more human flourishing
0:42:18 and less suffering than we would have in its absence, right?
0:42:22 – Yeah, markets provide large positive externalities.
0:42:25 There’s a lot of effort in those markets that feels wasted,
0:42:29 but it is like markets transmit information better
0:42:30 than basically anything else
0:42:32 because what they’re always transmitting
0:42:34 is the information you actually care about.
0:42:36 So like oil prices,
0:42:40 you don’t have to know that oil prices are up
0:42:41 because there was a terrorist attack
0:42:44 or because someone drilled a dry hole or whatever.
0:42:47 You, what you respond to is just gas is more expensive
0:42:49 and therefore I will drive less
0:42:51 or you know, energy is cheaper or more expensive.
0:42:53 And so I need to change my behavior.
0:42:56 So it’s always transmitting the actually useful information
0:42:57 to the people who would want to use it.
0:42:59 And the more complete markets are
0:43:02 and the more things there are
0:43:04 where that information can be instantaneously transmitted
0:43:06 to the people who want to respond to it,
0:43:08 the more everyone’s real world behavior
0:43:11 actually reflects whatever the underlying material constraints
0:43:12 are on doing what we want to do.
0:43:17 – The sort of crypto dream there is just more finance markets,
0:43:21 more feedback, more market feedback,
0:43:24 better financial services as a result.
0:43:27 That’s the basic view you’re arguing for.
0:43:29 – And it’s just a really interesting way
0:43:33 to build up new financial products from first principles
0:43:35 and sometimes you learn why those first principles are wrong
0:43:37 but that itself is valuable.
0:43:41 Like there is actual value in understanding something
0:43:42 that is a tradition or a norm
0:43:44 and understanding why it works
0:43:46 and therefore deciding that that norm
0:43:47 is actually a good norm.
0:43:48 – Good.
0:43:51 Last one, you know what it’s gonna be.
0:43:54 You tell me what the last one is.
0:43:55 – Is AI a bubble?
0:43:59 – Yeah, but you sound so sad about it.
0:44:01 Of course we’ve got to talk about AI, right?
0:44:02 Are you sad? – Yeah, of course.
0:44:02 – You talk about AI?
0:44:06 Like it seems exactly like what you’re right about.
0:44:08 Yeah.
0:44:13 When you hear Sam Altman talk about creating open AI,
0:44:18 starting open AI, he’s like, we basically said,
0:44:21 you know, we’re gonna make AGI,
0:44:24 artificial general intelligence, come work with us.
0:44:25 And when he talks about it, it’s like,
0:44:27 there was a universe of people who were like,
0:44:30 the smartest people who really believed
0:44:31 who that’s what they wanted to do.
0:44:33 So they came and worked with us,
0:44:36 which seems like exactly your story.
0:44:37 – Yes.
0:44:39 It turns out that a lot of people have had that dream
0:44:41 and for a lot of people,
0:44:43 maybe it wasn’t what they were studying in grad school,
0:44:46 but it was why they ended up being the kind of person
0:44:47 who would major in computer science
0:44:48 and then try to get a PhD in it
0:44:52 and would go into a more researchy end of the software world.
0:44:56 So yeah, there were people for whom this was,
0:44:57 it was incredibly refreshing to hear
0:44:59 that someone actually wants to build the thing.
0:45:02 – So you have that kind of shared belief.
0:45:04 I mean, at this point, you have these other elements
0:45:05 of what you’re talking about, right?
0:45:09 Like a sense of urgency,
0:45:12 an incredible amount of money,
0:45:19 elements of spiritual or quasi-spiritual belief.
0:45:23 – Yes, there are pseudonymous open AI employees on Twitter
0:45:26 who will tweet about things like building God.
0:45:29 So yeah, they’re taking it in a weird spiritual direction,
0:45:31 but I think there is something,
0:45:35 it is interesting that a feature of the natural world
0:45:38 is that you can actually,
0:45:40 if you put enough of a,
0:45:43 you arrange refined sand and a couple of metals
0:45:44 in exactly the right way
0:45:46 and type in the right incantations
0:45:48 and add a lot of power,
0:45:50 that you get something that appears to think
0:45:52 and that can trick someone into thinking
0:45:54 that it’s a real human being.
0:45:56 – The is it good or is it bad question
0:45:58 is quite interesting here.
0:46:00 Obviously too soon to tell,
0:46:04 but striking to me in the case of AI
0:46:06 that the people who seem most worried about it
0:46:09 are the people who know the most about it,
0:46:10 which is not often the case, right?
0:46:13 Usually the people doing the work, building the thing,
0:46:15 just love it and think it’s great.
0:46:17 In this case, it’s kind of the opposite.
0:46:21 – Yeah, I think the times when I am calmest about AI
0:46:24 and least worried about it taking my job
0:46:28 are times when I’m using AI products
0:46:31 to slightly improve how I do my job.
0:46:32 That is better natural language search
0:46:36 or actually most of it is processing natural language
0:46:39 when there are a lot of pages I need to read,
0:46:41 which contain, if it’s like a thousand pages
0:46:43 of which five sentences matter to me,
0:46:46 that is a job for the API and not a job for me.
0:46:49 But it is now a job that the API and I can actually get done
0:46:53 and my function is to figure out what those five sentences
0:46:55 are and figure out a clever way to find them.
0:46:57 And then the AI’s job is to do the grunt work
0:46:58 of actually reading through them.
0:47:00 – That’s AI as useful tool, right?
0:47:03 That’s the happy AI story, yeah.
0:47:06 – And I actually think that preserving your own agency
0:47:08 is a pretty big deal in this context.
0:47:11 So I think that if you’re making a decision,
0:47:14 it needs to be something where you have actually formalized it
0:47:15 to the extent that you can formalize it
0:47:18 and then you have made the call.
0:47:20 But for a lot of the grunt work,
0:47:23 AI is just, it’s a way to massively parallelize
0:47:25 having an intern.
0:47:26 – Plainly, it’s powerful.
0:47:29 And you’re talking about what it can do right now.
0:47:32 I mean, the smartest people are like,
0:47:34 yes, but we’re gonna have AGI in two years,
0:47:36 which I don’t know if that’s right or not.
0:47:37 I don’t know how to evaluate that claim,
0:47:39 but it’s a wild claim.
0:47:43 It’s plainly not obviously wrong on its face, right?
0:47:44 It’s possible.
0:47:46 Can you even start to parse that?
0:47:49 You’re giving sort of little things today about,
0:47:50 oh, here’s a useful tool and here’s a thing
0:47:51 I don’t use it for.
0:47:53 But there’s a much bigger set of questions
0:47:54 that seem imminent.
0:47:56 – You know, there are certain kinds of radical uncertainty.
0:48:00 They’re, you know, I think it increases wealth inequality,
0:48:04 but also means that intelligence is just more abundant
0:48:08 and is available on-demand and is baked into more things.
0:48:11 I think that it’s, you know, you can definitely sketch out
0:48:12 really, really negative scenarios.
0:48:15 You can sketch out, you know, not end of the world,
0:48:19 but maybe might as well be for the average person scenarios
0:48:21 where every white collar job gets eliminated
0:48:22 and then a tiny handful of people
0:48:25 have just unimaginable wealth and, you know,
0:48:27 rearrange the system to make sure that doesn’t change.
0:48:30 But I think there are a lot of intermediate stories
0:48:33 that are closer to just the story of, say,
0:48:35 accountants after the rise of Excel,
0:48:37 where there were parts of their job
0:48:39 that got much, much easier
0:48:41 and then the scope of what they could do expanded.
0:48:43 – It was the bookkeepers who took it on the chin.
0:48:47 It turns out like Excel actually did drive bookkeepers
0:48:50 out of work and it made accountants more powerful.
0:48:53 – Yeah, so you, you know, within,
0:48:55 I think within a kind of company function,
0:48:58 you’ll have specific job functions that do mostly go away.
0:49:00 And then a lot of them will evolve.
0:49:03 And so the way that AI seems to be rolling out
0:49:06 in big companies in practice is they,
0:49:09 they generally don’t lay off a ton of people.
0:49:12 They will sometimes end outsourced contracts,
0:49:15 but in a lot of the cases, they don’t lay people off.
0:49:18 They change people’s responsibilities.
0:49:20 They ask them to do less of one thing
0:49:22 and a whole lot more of something else.
0:49:24 And then in some cases, that means
0:49:25 they don’t have to do much hiring right now,
0:49:27 but they think that a layoff would be pretty demoralizing.
0:49:30 So they sort of grow into the new cost structure
0:49:32 that they can support.
0:49:33 And then in other cases, there are companies
0:49:35 where they realize, wait,
0:49:37 we can ship features twice as fast now.
0:49:38 And so our revenue’s going up faster.
0:49:40 So we actually need more developers
0:49:42 because our developers are so much more productive.
0:49:48 – We’ll be back in a minute with the lightning round.
0:49:58 Okay, let’s finish with the lightning round.
0:50:01 It’s the most interesting thing you learned
0:50:04 from an earnings call transcript in the last year.
0:50:08 – Most interesting thing from a transcript in the last year.
0:50:12 I would say there was a point,
0:50:13 this might have been a little over a year ago.
0:50:16 There was a point at which Satya Nadella
0:50:18 was talking about Microsoft’s AI spending.
0:50:21 And he said, “We are still at the point.”
0:50:23 And I think he and Zuckerberg both said something
0:50:25 to the same effect and in the same quarter,
0:50:27 which was very exciting for NVIDIA people.
0:50:28 But it was like, we’re at the point
0:50:30 where we see a lot more risk to underspending
0:50:34 than to overspending on AI specifically.
0:50:36 – That really speaks to your book, right?
0:50:39 That really is like a bubbly as hell
0:50:40 in the context of your book,
0:50:43 like overspending like the Apollo missions,
0:50:44 like the Manhattan Project,
0:50:47 like the big risk is that we don’t spend enough.
0:50:49 – And also they know that their competitors
0:50:51 are listening to these calls too.
0:50:55 So they were also saying that this is kind of a winnable fight,
0:50:57 that they do think that there is a level
0:51:00 of capital spending at which Microsoft can win
0:51:01 simply because they took it more seriously
0:51:02 than everybody else.
0:51:06 – So he’s like, yes, we’re gonna spend billions
0:51:08 and billions of dollars on AI
0:51:13 because we think we can win Zuckerberg implicitly.
0:51:22 What’s one innovation in history that you wish didn’t happen?
0:51:31 – I wish there were some reason that it was infeasible
0:51:34 to have really, really tight feedback loops
0:51:37 for consumer facing apps, particularly games.
0:51:42 – Is that a way of saying you wish games were less addictive?
0:51:44 – Yeah, I wish games were less addictive
0:51:48 or that they weren’t as good at getting more addictive.
0:51:50 So I wrote a piece in the newsletter about this recently
0:51:51 ’cause there was that wonderful article
0:51:54 on the loneliness economy in the Atlantic
0:51:55 a couple of weeks back that was talking about who we,
0:51:58 we just spend that one of the pandemic trends
0:51:59 that has mean reverted the least
0:52:01 is how much time people spend alone.
0:52:02 And I think one of the reasons for that
0:52:05 is that all the things you do alone,
0:52:08 they are things that produce data for the company
0:52:10 that monetizes the time that you spend alone.
0:52:14 And so the fact that we all watched a whole lot of Netflix
0:52:16 in the spring of 2020 means that Netflix has a lot more data
0:52:18 on what our preferences are.
0:52:21 – So they got better at making us want to watch Netflix
0:52:25 and all the video games we’ve played on our phones
0:52:27 got better at making us addicted
0:52:29 to keep playing video games on our phones.
0:52:31 Yeah, that’s a bummer, it’s a bummer.
0:52:35 What was the best thing about dropping out of college
0:52:37 and moving to New York City at age 18?
0:52:43 – So I would say that it was,
0:52:45 it really meant that I could,
0:52:49 could and had to just take full responsibility for outcomes
0:52:53 and that I get to get to take a lot more credit
0:52:55 for what I’ve done since then,
0:52:57 but also get a lot more blame
0:53:01 where there isn’t really a brand name to fall back on.
0:53:02 And so if someone hires me,
0:53:05 they can’t say this person got a degree from institution X.
0:53:06 You know, I didn’t even,
0:53:07 I dropped out of a really bad school too.
0:53:10 So there’s not even like the,
0:53:13 not even the extra upside of, you know,
0:53:14 if I started up was so great,
0:53:16 I just had to leave Stanford
0:53:17 after only a couple of semesters.
0:53:20 No, it was Arizona State and I didn’t even party.
0:53:25 So yeah, but yeah, it’s that.
0:53:28 It’s just being a little more in control of the narrative
0:53:33 and also just knowing that it’s a lot more up to me.
0:53:35 – What was the worst thing about dropping out of college
0:53:36 and moving to New York at age 18?
0:53:39 – So one time I went through a really,
0:53:40 really long interview process
0:53:41 for a job that I really wanted.
0:53:45 And at the end of many, many rounds of interviews
0:53:49 and you know, work session and lots of stuff,
0:53:51 they, the hiring committee rejected
0:53:53 because I didn’t have a degree and that was on my resume.
0:53:55 So that was kind of inconvenient.
0:53:57 I guess another downside,
0:54:01 like it might have been nice to spend more time
0:54:06 with fewer obligations and access to a really good library.
0:54:09 (upbeat music)
0:54:15 – Byrne Hobart is the co-author of “Boom”,
0:54:18 “Bubbles and the End of Stagnation”.
0:54:20 Today’s show was produced by Gabriel Hunter-Chang.
0:54:23 It was edited by Lydia Jean-Cott
0:54:25 and engineered by Sarah Brugger.
0:54:29 You can email us at problem@pushkin.fm.
0:54:31 I’m Jacob Goldstein and we’ll be back next week
0:54:33 with another episode of “What’s Your Problem?”
0:54:36 (upbeat music)
0:54:39 (upbeat music)
0:54:49 [BLANK_AUDIO]