Balaji Srinivasan is an angel investor, tech founder, philosopher, and author of The Network State: How to Start a New Country. He was formerly the CTO of Coinbase and General Partner at Andreessen Horowitz. Please support this podcast by checking out our sponsors:
- Policygenius: https://www.policygenius.com/
- Blinkist: https://blinkist.com/lex to get 25% off premium
- Notion: https://notion.com
- Onnit: https://lexfridman.com/onnit to get up to 10% off

EPISODE LINKS:
Balaji's Twitter: https://twitter.com/balajis
Balaji's Website: https://balajis.com
Books:
1. The Network State: https://thenetworkstate.com
2. Reputation and Power: https://amzn.to/3eyQiF6
3. Exit, Voice, and Loyalty: https://amzn.to/3TpuLNP
4. Seeing Like a State: https://amzn.to/3MJfvcD
Articles:
1. Bitcoin Is Civilization: https://www.commonsense.news/p/is-bitcoin-anarchy-or-civilization
2. Great Protocol Politics: https://foreignpolicy.com/2021/12/11/bitcoin-ethereum-cryptocurrency-web3-great-protocol-politics
3. Regulation, Disruption, and the Future: https://bit.ly/3CKY9Ys
4. The Measured Man: https://theatlantic.com/magazine/archive/2012/07/the-measured-man/309018
5. Secret Document That Transformed China: https://npr.org/sections/money/2012/01/20/145360447/the-secret-document-that-transformed-china
6. Hollywood's Missing Movies: https://reason.com/2000/06/01/hollywoods-missing-movies
7. New York Times Truth Campaign: https://adweek.com/agencyspy/droga5s-first-big-campaign-for-the-new-york-times-is-all-about-the-truth/126274

PODCAST INFO:
Podcast website: https://lexfridman.com/podcast
Apple Podcasts: https://apple.co/2lwqZIr
Spotify: https://spoti.fi/2nEwCF8
RSS: https://lexfridman.com/feed/podcast/
Full episodes playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4
Clips playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOeciFP3CBCIEElOJeitOr41

OUTLINE:
0:00 - Introduction
2:21 - Prime number maze
28:33 - Government
44:50 - The Network State
54:54 - Pseudonymous economy
1:18:40 - Exit
1:33:21 - Building a network state
2:21:08 - Wikipedia
2:58:38 - Fixing science
3:25:06 - Fixing the FDA
4:56:14 - Longevity
5:15:12 - Donald Trump's ban from social media
5:44:31 - War
5:51:39 - Censorship
6:05:39 - Social media
6:21:54 - Wokeism and communism
6:40:42 - Cryptocurrency
6:58:15 - AI, AR, and VR
7:10:14 - Advice for young people
7:43:08 - Regulating logic

SOCIAL:
- Twitter: https://twitter.com/lexfridman
- LinkedIn: https://www.linkedin.com/in/lexfridman
- Facebook: https://www.facebook.com/lexfridman
- Instagram: https://www.instagram.com/lexfridman
- Medium: https://medium.com/@lexfridman
- Reddit: https://reddit.com/r/lexfridman
- Support on Patreon: https://www.patreon.com/lexfridman

Donald Trump was probably the biggest person ever to be removed from social media.

Can you still mend the case for it and against it?

Everybody who's watching this around the world basically saw, let's say, US establishment or Democrat aligned folks just decapitate, you know, the head of state digitally, right?

And they're like, well, if they can do that in public to the US president, who's ostensibly the most powerful man in the world, What does the Mexican president stand against that?

Regardless of whether it was justified on this guy, that means they will do it to anybody.

Just like the bailouts, as exceptional as they were in the first area, everybody was shocked by them.

Every single bill is printing another whatever, billion dollars or something like that.

The following is a conversation with Balaji Srinivasan, an angel investor, tech founder, philosopher, and author of The Network State, How to Start a New Country.

He was formerly the CTO of Coinbase and general partner at Andreessen Horowitz.

There are chapter timestamps, there are clips, so you can jump around or, like I prefer to do with podcasts and audio books I enjoy, you can sit down, relax with a loved human, animal or consumable substance or all 3 if you like and enjoy the ride from start to finish.

Biology is a fascinating mind who thinks deeply about this world and how we might be able to engineer it in order to maximize the possibility that humanity flourishes on this fun little planet of ours.

Also, you may notice that in this conversation, my eye is red.

And also, if I may say so, from a life well lived.

To support it, please check out our sponsors in the description.

And now, dear friends, here's biology, Srinivasan.

At the core of your belief system is something you call the prime number maze.

we can start anywhere, it's with mathematics.

A rat can be trained to turn at every even number or every third number in a maze to get some cheese.

But evidently it can't be trained to turn at prime numbers, 2357, and then 11 and so on and so forth.

And frankly, if most humans were dropped into a prime number maze, they probably wouldn't be able to figure it out either.

Actually, it'd be pretty difficult to figure out what the turning rule was.

And so, the thing I think about a lot is just how many patterns in life are we just like these rats and we're trapped in a prime number maze.

And if we had just a little bit more cogitation, if we had a little bit more cognitive ability, a little bit more, whether it's brain-machine interface or just better physics, we could just figure out the next step in that prime number maze.

We could just see it, we could just see the grid.

Like that's a big thing that drives me, is figuring out how do we actually conceive, understand that prime number maze that we're living in.

So understand which patterns are just complex enough that they're beyond the limit of human cognition.

Are the limits of human cognition a feature or a bug?

I admire Ramanujan, I admire Feynman, I admire these great mathematicians and physicists who were just able to see things that others couldn't.

And just by writing it down, that's a leap forward.

People talk about it's not the idea, it's execution, but that's for trivial ideas, for great ideas, for Maxwell's equations or Newton's laws or quantum electrodynamics or some of Ramanujan's identities, that really does bring us forward, especially when you can check them and you don't know how they work, right?

You have the phenomenological, but you don't have the theory underneath it.

And then that stimulates the advancement of theory to figure out why is this thing actually working?

StatMech arose in part from the kind of phenomenological studies that were basically being done where people are just getting steam engines and so on to work.

And then they kind of abstracted out thermodynamics and so on from that, right?

So the practice led the theory rather than vice versa, to some extent that's happening in neural networks now as you're aware, right?

And I think that's, so just something that's true and that works, you know, if we don't know yet, that's amazing and it pulls us forward.

So I do think that the limits are more of a bug than a feature.

Is there something that humans will never be able to figure out about our universe, about the theory, about the practice of our universe.

Yeah, people will typically quote Cordell's incompleteness for such a question.

And yeah, there are things that are provably unknowable or provably unprovable.

But I think you can often get an approximate solution.

You know, Hilbert's problem is like, we will know, we must know.

At least we should know that we can't know.

Push to get at least an approximate solution, push to know that we can't know.

At least we push back that darkness enough so that we have lit up that corner of the intellectual universe.

Okay, let's actually take a bit of a tangent and explore a bit in a way that I did not expect we would.

But let's talk about the nature of reality briefly.

I don't know if you're familiar with the work of Don Hoffman?

I know Roger Penrose has his Road to Reality series for basic physics getting up to everything we know, but go ahead, tell me.

In modern physics, we start to question of what is fundamental and what is emergent in this beautiful universe of ours, and there's a bunch of folks who think that space-time, as we know it, the four-dimensional space, is emergent.

It's not fundamental to the physics of the universe.

And the same, many argue, I think Sean Carroll is 1 of them is that time itself, the way we experience it, is also emergent.

It's not fundamental to the way our universe works.

Anyway, those are the technical term, I apologize for swearing, those are the mind fucks of modern physics.

But if we stroll along that road further, we get somebody like Donald Hoffman, who makes the evolutionary case that the reality we perceive with our eyes is not only an abstraction of objective reality, but it's actually completely detached.

We're in a video game, essentially, that's consistent for all humans, but it's not at all connected to physical reality.

In a very distant way, but the simulation says that there's a sort of computational nature to reality, and then there's a kind of a programmer that creates this reality and so on.

Now he says that we Humans have a brain that is able to perceive the environment and evolution has produced from primitive life to complex life on Earth, produced the kind of brain that doesn't at all need to sense the reality directly.

So like this table, according to Donald Hoffman, is not there.

Like, not just as an abstraction, like we don't sense the molecules that make up the table, but all of this is fake.

So, you know, I tend to be more of a hard science person, right?

And so, just on that, people talk about qualia, is your perception of green different from my perception of green?

And my counter argument on that is, well, we know something about spectrum of light and we can build artificial eyes.

And if we can build artificial eyes, which we can, they're not amazing, but you can actually, you can do that.

Obviously, we can build recording devices and for cameras and things like that.

Well, operationally, the whole concept of your perception of green, you see green as purple, I see green as green, or what I call green, doesn't seem to add up because it does seem like we can do engineering around.

So The Hoffman thing, I get why people more broadly will talk about a simulation hypothesis, because it's like Feynman, many others have talked about how math is surprisingly useful to describe the world.

Like very simple equations give rise to these complex phenomena.

Wolfram is also on this from a different angle with the cellular automata stuff.

It's almost suspicious how well it works.

Yeah, but on their hand, it's like, yet we're still also in a prime number maze.

Also, within the constraints of the non-prime numbers, we find math to be extremely effective, surprisingly effective.

So maybe the math we have gets us through the equivalent of the even turns and the odd turns, but there's math we don't yet have that is more complex or more complex rules for

This is probably all RH, we're all just rats in a cage.

I know that gets like very abstract, but there are unsolved problems in physics, like the condensed matter space, there's a lot of interesting stuff happening.

My recollection, I may be out of date on this, things like sonoluminescence, we don't know exactly how they work.

And sometimes those things that are like at the edges of physics, in the late 1800s, I think Rutherford, somebody, I think it was Rutherford said, basically all physics is being discovered, et cetera.

And that was obviously before quantum mechanics.

That sort of edge case, people are looking at the bomber and the passion series and seeing this weird thing with the hydrogen spectrum and it was quantized.

And that led to the sort of phenomenological set of observations that led to quantum mechanics and everything.

And sometimes I think the UAP stuff might be like that, right?

People immediately go to aliens for UAP, like the unidentified aerial phenomena, right?

And people have been, there's surprising amount of stuff out there on this.

The UK has declassified a bunch of material.

You know, Harry Reid was a senator, has talked about this.

It's not an obviously, it's not an obviously political thing, which is good.

It's something that, is there something happening there, right?

And people had thought for a long time that the UAP thing was like American kind of counter propaganda to cover up their new spy planes that were spying over the Soviet Union to make anybody who talked about them seem crazy and hysterical or whatever.

But if the UAP thing is real, it could be atmospheric phenomena like the Aurora Borealis or the Northern Lights, but some things we don't understand.

It could be something like the Bomber and Passion series, which were the observations of emission spectrum before quantum mechanics.

So that's like another option as opposed to doesn't exist or Lilligreen men.

It could be physics we don't understand yet.

Do you think there's alien civilizations out there?

So there's a lot of folks who have kind of written and talked about this, you know, the Drake equation, which is like, you know, multiplying all the probabilities together.

There's perhaps more sophisticated takes like the dark forest, you know, which says that if the universe is like a dark forest where the dumb ones that aren't hiding our presence.

There's 1 calculation I saw and I haven't reproduced it myself, but basically says that the assumption that other civilizations have seen ours is wrong because when you have a spherical radius for the electromagnetic radiation that's leaving our planet, as that sphere gets larger and larger, it gets like smaller and smaller amounts of energy.

So you get farther out, you're not getting enough photons or what have you to actually detect it.

I don't know, I actually haven't looked into the math behind it, but I remember seeing that argument.

So actually, it is possible that it's so diffuse when you go past a certain number of light years out that people, that an alien civilization wouldn't be able to detect it.

That's more basically about signals from them, from us, to be able to, signals colliding enough to find the signal from the noise.

Yeah, Hanson has an article called Grabby Aliens.

And so there's a- He's been on this podcast.

I like him, he pushes boundaries in interesting ways.

I like him overall, he's an asset to humanity.

Grabby aliens, so he has this interesting idea that the civilizations quickly learn how to travel close to the speed of light,

we're not gonna see them until they're here.

I mean, 1 of the things is, so here's, for example, a mystery that we haven't yet done, right?

Which, or we haven't really figured out yet, which is abiogenesis in the lab, right?

We've done lots of things where you can show macromolecules binding to each other, you can show evidence for the so-called RNA world, abiogenesis to go from non-life to life in the lab.

You can show microevolution, obviously, with bacteria.

Lots of other aspects of fundamental biochemistry origins of life stuff have been established.

There's a lot of plausibility arguments about the primitive environment and nitrogens and carbons snapping together to get the RNA world is the initial hypothesis.

But to my knowledge, at least, we haven't actually seen abiogenesis demonstrated.

Now, 1 argument is you need just like this massive world with so many different reps before that actually happens.

And 1 possibility is if we could do atomic level simulations of molecules bouncing against each other, it's possible that in some simulation we could find a path, a reproducible path to abiogenesis and then just replicate that in the lab.

But that seems to me to be like a mystery that we still don't fully understand, like an example of the prime number maze.

And again, there may be some biochemist who's like, oh, Bhaji, you didn't know about X, Y, and Z that happened in the abiogenesis field.

I freely confess, I'm not like, you know, au courant on it.

We'll copy your voice and it will pronounce it perfectly correctly in post.

1 thing that I do think was interesting is Craig Venter a while back tried to make a minimum viable cell where he just tried to delete all of the genes that were not considered essential.

And this was like almost 20 years ago and so on.

And that thing was viable in the lab, right?

And so it's possible that you could, you kind of reverse engineer.

So you're coming at the problem from different directions, like RNA molecules can do quite a lot.

You've got some reasonable assumptions as to how that could come together.

You've got like sort of stripped down minimum viable life forms.

You can see microevolution, you can see at the sequence level, you know, if you do molecular phylogenetics, you can actually track back the basis.

There's actually, so It's not like there's no evidence here.

But this, in my view, is a fascinating area.

And actually, also relevant to AI, because another form of abiogenesis would be if we were able to give rise to a different branch of life form, the silicon-based as opposed to carbon-based, to stretch a point.

You give rise to something that actually does meet the definition of life for some definition of life.

What do you think that definition is for an artificial life form?

When will it give us pause that we created something that feels by some definition or by some spiritual, poetic, romantic, philosophical, mathematical definition that it is alive and we wouldn't want to kill it.

1 is Francis Crick of Watson and Crick, before he died, I think his last paper was published on something called the claustrum, okay?

And the thing is that, you know, sometimes in biology or in any, you know, domain, people are sort of discouraged from going after the big questions, right?

But he proposed the claustrum is actually the organ that is the seat of consciousness.

It's like this sheath that covers the brain.

And again, I may be recollecting this wrong, but you can look.

But my recollection is in mice, if you disrupt this, the mouse is very disoriented.

It's the kind of thing which Watson and Crick were all about structure implies function.

They found the structure of DNA, this amazing thing, and they remarked in this very understated way at the end of the paper that, well, obviously this gives a basis for how the genetic material might be replicated and error corrected because helix unwinds and you copy paste it, right?

So he was a big structure function person and that applies not just at the protein level, not just at the level of DNA, but potentially also at the level of organs.

Like the claustrum is kind of this system integrated level, right?

It's like the last layer in the neural network or something, you know?

And so that's the kind of thing that I think is worth studying.

So consciousness is another kind of big, abiogenesis is a big question, the prime number of maize, consciousness is a big question.

There's folks, gosh, there's, I think, so this 1 is something I'd have to Google around, but there was a guy, I think at Santa Fe Institute or something, who had some definition of life and like some thermodynamic definition.

But you're right that it's gonna be a multi-feature definition.

We might have a Turing test-like definition, frankly, which is just if enough humans agree it's alive, it's alive, right?

And that might, frankly, be the operational definition.

Because viruses are like this boundary case, are they alive or not?

Most people don't think they're alive, but they're more alive than a rock, in a sense.

Well, I think in a world that we'll talk about today quite a bit, which is the digital world, I think the most fascinating, philosophically and technically, definition of life is life in the digital world.

So chatbots, essentially creatures, whether they're replicas of humans or totally independent creations, perhaps in an automatic way, I think there's going to be chatbots that we would ethically be troubled by if we wanted to kill them.

They would have the capacity to suffer, they would be very unhappy with you trying to turn them off, and then there'll be large groups of activists that will protest, and they'll go to the Supreme Court, of whatever the Supreme Court looks like, in 10, 20, 30, 40 years, and they will demand that these chatbots would have the same rights as us humans.

I saw that Google engineer who's basically saying this had already happened, and I was surprised by it because when I looked at the chat logs of it, it didn't seem particularly interesting.

On their hand, I can definitely see, I mean, GPT-3 for people who, you know, haven't paid attention shows that serious step-ups are possible.

And obviously, you know, you've talked about AI in your podcast a ton.

Is it possible that GPT-9 or something is kind of like that, or GPT-15, or GPT-4, maybe?

But- Yeah, for people just listening, there's a deep skepticism in your face.

Yeah, the reason being because, you know what's possible?

It's possible that you have a partition of society on literally this basis.

That's 1 model where there's some people, just like there's vegetarians and non-vegetarians, right?

There may be machines have life and machines are machines, you know?

You could definitely imagine some kind of partition like that in the future where your fundamental political social system, that's a foundational assumption.

And, you know, is AI, does it deserve the same rights as like a human, or for example, a corporation is an intermediate.

Do you see the thing which is how human are different corporations?

So for example, Disney is considered about as human as like a dog.

But like Exxon, I may be remembering this wrong, but they had like a level with like human at 1 end and like rock at the other.

Does it have to do with corporate structure?

it's about people's empathy for that corporation, their brand identity.

But it's interesting to see that first of all, people sort of do think of corporations as being more, like the branding is really what they're responding to.

that's what, I mean, they're also responding, you know, I have a brand of human that I'm trying to sell, and it seems to be effective for the most part.

Although it has become like a running joke that I might be a robot.

But I mean, in that sense, I just, I think, I don't see a reason why chatbots can't manufacture the brand of being human, of being sentient.

I mean, that is the Turing test, but it's like the multiplayer Turing test.

Now that actually a fair number of chatbots have passed the Turing test, I'd say there's at least 2 steps up, right?

1 is a multiplayer Turing test where you have chatbots talking to each other.

And then you ask, can you determine the difference between end chatbots talking to each other and clicking buttons and stuff in apps and humans doing that.

And I think we're very far off, or I shouldn't say very far off, at least I don't know how far off we are in terms of time, but we're still far off in terms of a group of end chatbots looking like their digital output is like the group of N humans.

Go from the Turing test to the multiplayer Turing test.

Another definition is to be able to swap in and you're not just convincing 1 human that this is a human for a small session, you're convincing all humans that this is a human for end sessions.

Remote work actually makes this possible, right?

That's another definition of a multiplayer Turing test where basically you have a chatbot that's fully automated that is earning money for you as an intelligent agent on a computer that's able to go and get remote work jobs and so on.

If you could have something that was like that, that was competent enough to, I mean, because everything on a computer can be automated, right?

Literally, you could be totally hands-free, just like autonomous driving.

As a challenge problem, if you were Microsoft or Apple and you had legitimate access to the operating system, just like Apple says, can you send me details of this event?

A decentralized thing could, in theory, log the actions of 10, 000 or 100, 000 or a million people and with cryptocurrency, you can even monitor a wallet that was on that computer and you could see what long run series of actions were increasing or decreasing this digital balance.

So you start to get, at least conceptually, it'd be invasive and, you know, there'd be a privacy issue and so on.

Conceptually, you could imagine an agent that could learn what actions humans were doing that resulted in the increase of their local cryptocurrency balance, okay?

There may be better ways to formulate it, but that I consider a challenge problem is to go from the Turing test to a genuine intelligent agent that can actually go and make money for you.

People obviously have trading bots and stuff, but that would be the next level.

It's typing out emails, it's creating documents.

See all Lex Fridman transcripts on Youtube

Balaji Srinivasan: How to Fix Government, Twitter, Science, and the FDA | Lex Fridman Podcast #331