My guest today is Balaji Srinivasan, an angel investor and entrepreneur. When it comes to the future, he’s the single most creative person I know because he’s so technical, innovative, and polymathic. Talking to him is an experience unlike talking to anybody else, which I tried to replicate in this conversation.
A little bit about Balaji. He’s worked as the Chief Technology Officer at Coinbase and a General Partner at Andreessen Horowitz. In the world of academia, he holds a BS/MS/PhD in Electrical Engineering and an MS in Chemical Engineering, all from Stanford University. He’s also taught at Stanford, where his online course has reached 250,000 students worldwide.
This episode is a whirlwind through Balaji’s interests. We started by talking about his production function. We talked about what holding all those degrees from Stanford taught him about learning, how he identifies talent, and what building and selling two companies for more than $100 million taught him about management. We also talked about his interests in genomics, how to reverse aging, and why living forever is the ultimate goal of technology. At the end, we built off the ideas I talk about in my online writing school called Write of Passage to talk about his plan to fund online writers with a project called MediaFund.
Find Balaji Online:
2:42 – What Balaji learned about how to learn from his extended time in academia and why he doesn’t read the instructions until he has to.
4:21 – Why knowing philosophy and history is so integral to starting a successful company.
6:54 – Why Balaji thinks we are severely underutilizing the collaborative potential of the internet.
12:45 – How remembering references to knowledge instead of the knowledge itself gives Balaji a better way to argue his points.
13:47 – Why searching for people who are “hungry and can teach us something” serves everybody who is involved very well.
19:39 – The “tour of duty” and how to create a great strategy for developing and managing yourself and your team.
24:25 – The movement from a centralized century to a decentralized century and why Balaji feels the future is moving more towards his lifestyle.
31:19 – How technology hyper-deflates the market of everything it touches.
38:23 – How the past is wrapping back around to the future and how the evolution of education is leading the way.
44:49 – Why abstraction means progress as a culture up to a certain point and can become harmful beyond that.
48:57 – How to optimize your information diet to make you smarter, more effective, and more honest about where you spend your energy.
54:07 – The future of online education and why it doesn’t end with Wikipedia.
59:32 – New ways to look at incentive structures for writing and how it inspires technological and social growth.
1:04:27 – How to bridge the gap between Hollywood, big data, and education.
1:12:43 – The future of the internet and why the pseudonymous economy seems likely to Balaji.
1:15:04 – How we can use a “crypto oracle” to create an unfalsifiable history of our digital information.
1:21:31 – Why a worldwide ledger of record is the future we need in an information-driven world.
1:26:59 – Why Balaji believes that the pinnacle and goal of technology is to help humans live forever.
1:32:50 – How to build a digital country through writing.
1:39:34 – Why genomics needs more attention from the general population and technology.
1:44:42 – Why writers will be the future of millionaires and billionaires.
David: Welcome to the podcast. The thing that I find very interesting about you is that you’ve straddled the world of tech and academia like few other people. What do you think that academia with your background, with a BS, an MS, a PhD, what did all that teach you about studying and learning?
Balaji: Well, I was a career academic for the first years of my career. And I thought I was actually going to be a professor at Stanford or MIT or something like that. And what do they teach you in learning? Well, first is I did learn how to learn quickly. And the way that I learned technical content quickly is I just start doing problems. I don’t even read the text until I have an issue. And the reason is that, especially in technical topics, often if you know, for example, differential equations or you know statistics, or you know Maxwell’s equations or something like that, you can often jump into an area and just start calculating and quickly run into a roadblock. And then you realize what you actually don’t know.
There’s a great series of books on this called Schaum’s Outlines. They’re these yellow books from the early 2000s where you can do Schaum’s outline of probability and statistics or Schaums outline of accounting. And it’s amazing how many people have book learning of something, but you just give them Schaums and ask them to do the first 10 problems in accounting or something like that.
In accounting there’s an aspect where it’s not fully technical. It’s related to the time and place, accounting treatments are, but electricity and magnetism, you can give them something like this and they may not be able to do it. Whereas for sure, somebody who can do those questions can explain to you how the equations work. So that’s how I jumped in on technical topics is just go start doing problems.
David: How does your learning work differently when you’re learning technical learning versus more sort of liberal arts style learning? Whenever we talk. You seem pretty well versed in those things too. What do you think the similarities and differences are?
Balaji: I think that with liberal arts style learning, I actually only really got into it after I left school. Well, I liked fiction. I liked reading fiction when I was in college, but I really didn’t care about politics or anything like that or history growing up in the 90s, it was into history. Who cared? Right. But I got into liberal arts as a way to win arguments.
So what I mean by that is, history is most interesting backwards. You go to Berlin, for example, why is the city the way it is? Well, let me tell you about something called, the Cold wWar and the Berlin Wall. How did that happen? Okay, well then here’s World War II. How did that happen? And I think history is most interesting backwards where you can understand how people got esconced in the positions they are today, and then it’s relevant.
You have a contemporary hook. You have some aspect of life experience in which to stick that, right? There’s other people, for example, they’re fascinated by things like hieroglyphics or the pyramids, and they want to understand the context of that. And that’s also obviously legitimate and that’s interesting, but at least for me, I hook all of that stuff to a purpose, which is for example, how best to run an organization.
Well, the philosophy of the ancients actually does have something to say about that because you’re managing people. There’s aspects of human nature that haven’t changed in thousands of years, and it’s worth knowing what the great classic liberal arts’ curriculum has to say about that. And so from kind of a very pragmatic perspective, you get to apply philosophy. This is also similar to the founders of the U.S. That’s why they had to learn all this philosophy and history and so on, because they were actually building a political system. That’s applied history. On a small scale that’s what building a company is. You’re building a set of rules that scale the executive, that scale the CEO, and those rules are like laws, but internal to the company and there’s penalties for breaking them or whatever and rewards for following them. And this is in its own sense, similar to running a small city or a large city depending on how big the company gets.
David: It seems like there’s a lot of people in technology. Paul Graham comes to mind, Peter Thiel comes to mind, you come to mind, who would be academics in another life like what you said. Why did you leave academia for tech, besides the obvious that the power law of wealth in tech is much higher than in academia? Is there anything else going on there?
Balaji: So it’s interesting. Personally, I’m not money motivated. I know there are people who are, but I basically wear the same T-shirts that I’ve worn in grad school. I don’t own any cars or anything like that. I look at money as a tool to build that which I can’t buy. For example, Elon is doing space SpaceX because he can’t buy a trip to Mars. You can have all the money in the world but you can’t buy a trip to Mars. Just like 200 years ago, you could be the richest man in the world, but you couldn’t buy an iPhone. And so that’s what I care about. And I also care about mathematics in the sense of, in the year 3000 as a metric for whether civilization is advanced, we might all just be uploaded, artificial intelligences or whatever. At that point, I think a useful civilizational ruler is what mathematical problems can that civilization solve?
David: Why? That’s not intuitive.
Balaji: Yeah, it’s not intuitive. So the reason is that if you’re really good at math in the year 3000, for example, you would have computers, right? So you would have the infrastructure required to support computers. You’d have enough of the civilization that people can focus on higher level things. You’ll hopefully be enough at peace that people will be able to focus on that type of thing. And mathematics is cumulative so that if you can do differential geometry, you can do differential equations. Because of that, it’s this kind of … Oh, it’s also something that space-time in variant, right. So we think math should be space variant, physics may vary in space, at least, let’s say, physical law might be different, I don’t know, a billion miles away. So far it doesn’t look like that, but it might be.
So because of those properties, it’s not dependent on the specific civilization. The Incas, the Aztecs, the Zulus, the Parsis, every group on the planet has the same math, right. That’s the past and the present and the future. So one it’s universal, right. Two is, it’s hierarchical, it’s cumulative. So you’ve got a ruler, a ladder. Whereas for example, fashion, we couldn’t say if we’ve progressed in fashion. Fashion is just cyclical. Fashion is what people think is cool at any given moment. And that’s a function of just preference at that time, whether it’s hoop skirts or Victorian clothing or whatever. So A, math is universal, B, it’s cumulative and hierarchal, C, implicitly, it says, you’re at peace. You’ve got a supply chain. You’re focused on higher things. You’ve got computers. You’ve got whatever comes after computers, maybe it’s brain machine interface or something like that, that allows us to compute all this stuff. Just to give you an example, do you know the massive math program was?
Balaji: This is something I’ve thought about a lot. I think cryptocurrencies will actually help us with this, but I think the internet is really under exploited for collaborative work. Okay. So there’s a thing 10 years ago now, that is called a polymath project where a research mathematician named, I think it was either William Gowers or Terry Tao, both of them are Field’s Medalists. Maybe it was both of them, but the polymath project essentially posed an unsolved math problem, and just ask people to sort of collaboratively solve it on a block. And it worked.
It’s not something that you would imagine would work because you might think isn’t math sort of like basketball where there’s LeBron who’s really good. And no one else is going to score like LeBron. So how would one of the best mathematicians in the world get help from the crowd? You might argue that. Right. And certainly the terms that they were using were accessible only to a small group of specialists. Still collectively, that group was actually able to make serious progress and able to solve this.
And that’s something where you use the Internet to put a lot of brains together, asynchronously, right? You couldn’t have had a conference going on for months and months and months. And having all those people in one spot, all those folks who could work on it, they could sleep, drink coffee. And they contribute this little snippet, or they had this bad idea. And the bad idea would not be worthy of publishing in a manuscript.
It’s just an offhand comment. Oh, do you think about frobnicating this or whatever. That’s a made up word, but you know what I’m saying, right? And so there’s offhand comments that weren’t in publication grade that weren’t making an airline trip for, that you didn’t know the person, you didn’t know Terry Tao well enough to call him up and say, those actually got surfaced and integrated into something.
And broadly speaking, it’s actually kind of interesting how little collaborative work is done on the internet. I mean, we have Wikipedia. That’s one of the great ones. There’s obviously GitHub with open source. There’s Google docs. You might say, well, that is a lot of collaborative work. And I agree, but I think there’s so much more that can be done. Why aren’t there more things like Google docs where you can collaborate in real time and see the thing coming together?
You can see people fighting in quasi real time on Twitter, but you can’t necessarily see them building things together. And that’s partially because the division of labor is challenging. Rewards are challenging, approval of digital work is challenging, all that type of stuff. But I think that with cryptocurrencies and other kinds of things, you might be able to get there, where you incentivize tasks for sub units. Anyway, so that’s one last aspect, recusing back up the stack to do something like the polymath project you need civilization. And that’s also something historical.
David: It’s funny. I feel like your answer there with the polymath project speaks to the next question I want to ask. And whenever we talk, I feel like I have more links per minute with you than anyone else I know. And it has to do with the way that you think and the best I can describe it is you sort of think in hyperlinks. You’re like a high context thinker, probably more than anyone I know. And honestly, it kind of makes it annoying to talk to because I can’t actually get up the stack of what is needed sometimes. What’s going on there in how you think your brain works?
Balaji: I think it said, I try to store the minimum amount of information than uncompacted. So when I’m talking to somebody on something like this, I may do it verbally like we’re talking now, but the reason … what is a link? What is a citation? What is a reference? A reference is reinforcements. What it means is, it’s not just you who’s thinking this.
Here’s this person and all of these people who have reviewed this piece of paper, and that’s actually 50 people you’re bringing with you. And they’re all research mathematicians, right? Oh, and that actually represents 50,000 people. So arguing from references is summoning reinforcements to any discussion. So again, it’s about, okay, why do you learn history? To win arguments, right. Why do you argue from references? Again to win arguments. Because it’s not you, it’s you and what army. It’s this army of scholars.
David: Let’s move into talent. You’ve sold two companies for more than $100 million. What have you learned about gathering, recruiting and identifying up and coming talent?
Balaji: Up and coming is a very important component of it. Rabois has a good saying, which I sort of, Keith Rabois, which I sort of kind of independently come up with in so many different phrasing. His phrasing is, hire geniuses no one knows yet. Right. And mine was sort of, “hungry and can teach us something,” and everyone has different versions of this.
But “hungry and can teach us something” is basically like, okay, you’ve got somebody who is at the beginning of their career. Right. And one thing I found that at least I’m wired for, and a lot of other people are not wired for is I’m almost, I shouldn’t say, anti credential because I understand the value of it. But I’m actually most excited when I see a smart person who does not have that credential because that’s a good trade.
I can hire them. I can pay them reasonably well, I can help them level up. I can give them essentially the biggest opportunity they’ve ever had in their lives. And I say that in the sense of like, I’m not trying to thumb my chest or anything like that, but I have recruited people out of the middle of nowhere, folks living with their parents in Canada, folks in Malaysia, folks in Poland, people who are truly in the middle of nowhere.
And because I could see that they could code well or write well or both, I would just pluck them off the Internet. And I didn’t care if they had an accent or anything like that. They were smart people. And often they had some personal circumstance where this person could out code and outwork lots of the folks with the perfume degrees from Harvard or Stanford or whatever.
And I’m not saying by the way, the people at Harvard Stanford are dumb. I’m not saying anything like that, but from a price performance standpoint what you want is, as Rabois says, the geniuses no one knows yet. And it’s also something where, by giving somebody the biggest opportunity they’ve ever had, they’re hungry. Right. And it’s another bit like why can they teach us something?
Well, if I learn something from their writing, if I learn something from their code, their writing is actually pretty important getting to something that you talked about because, especially today, when you’re moving to remote everything, the way that you engage with somebody is, do they write clearly? For example, did they write stuff like a mystery novel where you have to kind of listen for a while before figuring out what they’re saying, or did they put the headline first?
Of course, there’s difference between casual conversation versus writing something for instructions. Right? You’d write it. And then you pull it to the beginning and you communicate in the headline. Then you communicate it again in the subtitle, you can communicate it again in a slightly different way in the opening sentence, in the opening paragraph. And it’s like this Russian doll thing… at least that’s a newsy style of writing.
And that’s how you should write internal memos. That’s how you should write slacks is most important first. Right? So basically I looked for folks that are like that, that are under priced relative to their potential and I try to level them up. Now, the issue with doing that, by the way, here’s actually an important tip on that is, the most challenging thing to do as an entrepreneur, but it’s also absolutely necessary, is to hire people who are better than you.
First, why is it necessary? Well, let’s say you’re getting into robotics. You may not know robotics, but you need to find somebody who will do the job better than you could if you were doing it. And the reason is, if it’s much worse than you could have done, then you’ll basically look at it. And you’ll constantly like, ah, if only I had the time I could get in there and fix this thing. And if there’s high stakes for the customers, and if you have the power to do it, you feel dragged in that direction. Where if you hire someone who can’t do the job better than you can, you feel that you should do the job. Okay.
Why is it hard to find such a person? Well, obviously, if somebody is better than you in this skill, why would they want to work for you or work for your company? So the answer I’ve come up with on that is, thinking of skill as a vector, right? You may be worse at robotics, but you’re better at mathematics or programming or something like that. And so then you go to that person as essentially, at least the way I think of it as like an academic peer.
And you say, look, you’re a disciplinary expert in robotics. I’m not a 10 or a 9 like you are, but I have done the equivalent of Schaums Outlines. I have self studied and I’ve gone my way to a six. Right. I can tell you what degrees of freedom are. I can tell you how manipulators work and the basic sensors and actuators. I know something of the problems in your field. And so I’ve done my homework.
And what happens is somebody who’s a nine in that area is actually pretty impressed if somebody has put in the work to at least learn the basic vocabulary of their space. It’s like, “hmm, not bad.” For example, if you’re an engineer and there’s someone who’s self-taught, and maybe their code isn’t well structured or whatever, but they’ve managed to get a data analysis together and make some graphs and charts and scrape some stuff, you’re always like, okay, that person is resourceful. I respect that. Okay, let me talk to this person.
And then if you have strengths in another area, well, then there’s an ongoing trade of knowledge. And again, this is how I like to recruit. Money is important. It’s an important tool. It has to be there. It’s a powerful thing for alignment and so on and so forth. I’m not against money, but fundamentally what I look for in the folks who we hire, if I hire people are folks where I can learn from them and hopefully I can teach them something. So you really learn something you level up in your skills.
And so that vector way of thinking about skills is the way that you can hire somebody better than you, because they’re better than you in one or a few critical key areas. Hopefully, you’re better in some other areas. And then there’s a mutual, positive-some exchange over the course of the relationship. By the way, speaking of the course of relationship, I think Reid Hoffman’s concept of a tour of duty is useful, because a company is not a family, right?
A family is about unconditional love or whatever. Not to say that’s bad, just that’s not how a company is. A good company is conditional love since you have to deliver. On the other hand, a company, it shouldn’t be mercenary because if it’s mercenary, look, everybody has a bad day. Everybody has bad week at times, sometimes even a bad month or something like Corona. Of course these are our fellow coworkers and you have to have some mercy and so on as well, right. Even if the customer doesn’t, because the customer, you don’t care whether someone had a bad day at Coca-Cola’s plant. You’re just, Oh, wow, why is this bottle all crushed? Okay. I’m just going to buy the Pepsi.
It’s not out of malice or anything like that. It’s just like the interface you’re seeing is not a human. You’re seeing a crushed can or something. There’s a failure somewhere in the supply chain. So you shrug and just pick the other thing. Right? So the customer is genuinely merciless. So you have to somehow buffer that. And I think the intermediate is this concert of the tour of duty, where you say, look, here’s what I’m going to try to give you. Also, I think it’s also useful by the way, for folks to write out what their goals are. Right. Where do they want to be in a year or three years?
David: Wait. But that obvious.
David: Are you saying that because people don’t do it or are you saying it because it’s undervalued or are you saying it because you have some different insight into this? This feels like something that my dad would tell me right before I go to college or something that like, “thanks, Uncle Tom!” So what are you trying to say here?
Balaji: Yeah. Yeah. So what your dad is telling you is actually non-obvious and good. You may have internalized that already, but basically writing out what your goals are, it’s amazing how many people don’t do that. And the thing about it is it’s not necessarily that you think about it every moment, but by writing out what your goals are, so many people just random walk through life.
And there’s something else that’s important. This gets into like management and whatnot. But basically before somebody joins, if they write out what they would consider a success, for example, bull, bear, and base. So the bull case, the bear case and the base case, right? So bull, bear, base is a useful thing to do where you kind of have folks right out, and this is a for themselves. What would be good, what would be amazing and what would be a fail in four years. And basically, it’s actually always fun to look back on this stuff four years later. And you’re like, wow, I was so awful on these things. And I was dead on on this, right.
Point being that by doing that, they self-constrain, and they actually take some responsibility for their own life. This is the same thing where whenever you’re dividing up labor, at the junction point between a manager and an employee, right, you should do a two by two table, which is, what did the manager expect from themselves? What do they expect from the employee? What does the employee expect for themselves? What does the employee expect from the manager. Okay. That two by two table is so simple, but you should do that at every one-on-one and you should do a one-on-one a week.
And what it does is it rectifies miscommunication. And by putting into words, what you expect of yourself and the other person you’re holding yourself accountable in front of them, just like they are holding their self accountable in front of you. And both of you know what you expected from the other. If a week later or two weeks later or three, etc., I mean, look, everybody doesn’t always hit everything. New things come up. It’s not meant to be a signed in blood thing.
But that two by two table, if you track that in a simple Google doc, right over time, it’s super useful because it also, as a manager allows you to point back and say, Hey, we agreed on this at this time. Or I did deliver what you asked of me and the employee can do the same. It’s like, Hey, you can’t say that I didn’t deliver because I did do this or that. I’m not saying it as an adversarial thing, but it is something which is useful to keep aligned. Right. And so that’s what I mean, like this concept of write out what your goals are, your uncle Tom or your dad is a smart guy.
David: So last question about personal productivity, but it’s another very biology thing. I feel like you very purposefully live in the future. And I feel like you actually make certain decisions. Like we were together, we had dinner in San Francisco and it was like 11 o’clock at night. You’re like, Oh, I’m just going to hotel tonight. And then I’m going to Uber to the hotel and then sort of figure it out. And I was sort of just watching that. I was just like, man, it’s not like hotel tonight. Uber is so futuristic, but menu of your decisions and sort of the options that you give yourself are very mobile first, things that I think people will be living in 10 to 15 years.
Balaji: Right. Right, right. So why is that? I’ve thought about that. And it just so happens in some sense that my preferences and way of life, the world has become more like me over the last 20 or 30 years. This is not something that I could … I feel really bad for somebody who is a morning person, is extremely nine to five, wants a routine, doesn’t like computers, wants to come into the office every day and high five somebody, essentially the kind of person that the 1950s was built around, I’ve nothing against that person. Okay. That’s a perfectly fine person. He makes the world go round in many ways, super reliable, daily kind of person, right.
That morning person type mentality is just the opposite of the way the world is going. It’s going from synchronous to asynchronous. It’s going from in-person to remote. It’s going from like pull the same lever a thousand times to think deeply, and then write a piece of code that just does it. It’s going from essentially the centralized century to a decentralized century.
In many, many ways, all the stuff that I have been into, it’s just a predictor of just by my nature I seem to gravitate to this stuff. For example, machine learning and genomics at Stanford in the early 2000s was something I thought was, I shouldn’t say old hat, but machine learning was something I had been doing for years by 2006. Right. And to my surprise, 10 years later, everybody’s talking about that concept. I mean, of course there’s new developments in it. Of course there’s deep neural networks. There’s a reason that people are talking about it and whatnot, but it was interesting to me to see people like, Oh, the gradient. I’m like, yeah gradient. Okay.
And so that was something I had to calibrate on. Just to give you a few examples of this, machine learning genomics, robotics, that was … we built a clinical genomics lab that had robots in it in the early 2010s profitable. Things obviously promote work, Soylent, bitcoin. Teleport built a search engine for digital nomads. That was a little too early. It was a great company actually in many ways, but had that been around in 2020, it would have had probably millions of users just given how many people are still searching for that. Right. So we set that up in 2014, sold that in 2016 or 2017.
So digital nomadism, remote work, startup cities, exit, like all the stuff that we’ve been talking about in terms of physical exit, from a location. And I gave this talk at Y Combinator in 2013, talking about Silicon Valley’s ultimate exit, it’s happening.jpg. We are literally physically exiting. All the moving trucks are here. We have de-centralized into the cloud. We’re finally achieving our destiny of becoming this digital people. 2020 is a year the internet starts.
And I mean, there’s a 50 other things like that, but even something like Soylent- why? Because I want it to be maximally productive. I wanted to basically be able to sit and have something nutritious and then just be able to work all day. And so the thing about it is that the 9:00 to 5:00 person wants to work at a defined period of time consistently and then go home, and then go and do sports or whatever. Right. And that’s fine, nothing against that. It’s good.
Whereas, what I want to do is maximize the number of hours I’m working every day, including weekdays and weekends. But I may not work at the same time because I might want to work for 16 hours one day, and then sleep and then think about it and then wake up and work again. Right. And so just for whatever reason, the world has a climatized itself to this style of work. So whatever my personal preferences are, it’s going to become the future. This is something that it’s not something that’s like a good or bad thing. It just so happens.
I’ll give you one more example, by the way, in the mid 2000s, there’s a period between when Google maps and the iPhone came up. Okay. So 2004 to 2007. And during that period, what I did was I would, I’m sure other people did this, but just to give you an example, I would go and Google maps would give you a screen by screen directions. So what I do is I would charge my laptop. I would screenshot every screen of the directions. And then I’d keep my laptop on the passenger side seat facing. So it’s like a proto-iPhone doing Google maps, right, where I could just pause and … you understand. Right. And I also printed it as a backup, just in case the laptop ran out of power.
And so I knew that mobile was going to be a thing, that Google maps was going to be an app. And just stuff like that where it’s just like weird idiosyncratic things that I do. For example back in 2015 or something like that, when there was some wedding or something people were doing, I wrote a semi-serious proposal. It was kind of meant to be a joke, but also semi-serious on how people were going to Skype into their weddings in the future. And that’s happened this year. Okay. Five years later because of Corona.
And the thing about that is, ultimately what was holding that back was a cultural thing. But guess what, you’ll get more attendees to your wedding if they can all come virtually, right? You might actually see them and be able to catch up with them. The remote socialization tools are not yet there, obviously. And what people will say is nothing beats in person and so on. And maybe that’s true, but I think in VR and whatnot, you can probably get something which is a pretty decent party experience even if you consider a wedding special. And so I do think, and it will be way cheaper.
And so if 80 or 90 percent is good, but one one hundredth the cost, a lot of people are going to do that. So just for whatever reason, it just seems like my preference is starting to become the future in 1, 10 or in 20 years. I’ll give you one more example. Okay. November 2019 I tweeted something like, has anybody done full like Zoom conferences. Here, I’ll find that exact tweet. It’s pretty funny. Everybody ends up doing that. Just like obviously, a few months later due to corona. November 26, 2019, “Has anyone held a fully remote conference via Zoom or VR or something” Of course they have, right? So just kind of funny.
David: So I want to transition into teaching where education is in a weird spot because when technology works, it’s deflationary where we get the essence of technology is you get more for less. Costs go down, value goes up. And teaching has been in a weird place where that hasn’t been happening. Costs have been going up so much. And you, when you were teaching at Stanford, you taught class about cryptocurrencies, data mining, statistics, computational biology. From teaching in-person, teaching remotely, what did you learn about teaching and the differences between in-person and virtual?
Balaji: Yeah. Well, so the deflation aspect is important. It’s funny. People talk about Baumol’s cost disease, that jobs that have no rises in labor productivity, have salaries going up and whatnot. And folks act as if this is like super mysterious, but has a great graph that shows that the inflation is here. It’s just not evenly distributed, where prices are up in education as you point out. They’re also up in healthcare, or at least until recently, they were up in real estate, and they’re down in computers and telecommunications and so on. The rough lesson of that is anything that the technology industry touches, prices hyper deflate, and everybody becomes equal because everybody’s got the same iPhone or Android phone experience. They’ve got the same Google, they’ve got the same Wikipedia, they’ve got the same Stack overflow, the same GitHub, anything that is digital.
For example, Steph Curry also uses Uber. So you actually get a consumer experience where to have… The consumer economy is actually another form of equality. At massive, massive scale, you produce the same product for everybody. There’s not that much really price differentiation between a top end iPhone and a low end one. It may be five-x or 10-x, I guess you can get the gold thing or whatever, but that’s purely a conspicuous consumption thing as opposed to a functionality thing. So every area of technology is touched. The price hyper deflates. Every area of the state has touched due to regulations or subsidies, that’s what impacts and impairs increase in labor productivity. Just as one example within medicine, there’s a requirement to have the doctor in the loop for many kinds of diagnostic decisions, even if artificial intelligence can do it better than 99% of doctors, even before it’s deployed.
So you get into a production system, you calculate and you collect all the data, it will become better than every doctor for many things. Maybe you have some MD supervising it. It’s not something where it’s just a program that runs by itself without any supervision, but that kind of thing, which is automating the doctor way, you can’t do because of regulations subsidies. By the way, a subsidy is much like a regulation in the sense that it sort of has a particular way of doing things cast into stone. The state has got taken over and that’s why, for example-
David: A subsidy changes gravity, right?
Balaji: Exactly. That’s right. That’s why we’re in 2020 and we still have yellow school buses, we still have number two pencils or whatever. I think we do. In many ways, you still go to a school because all that stuff is in budget appropriations, and corona has finally changed that. But until corona, essentially this lagging aspect of the system, law, medicine, education, finance, real estate, all of these things were sort of lagging areas that were regulated or subsidized and technology had partially reformed them, but not fully. Now, they’re just dragged into the future where you have digital education. It’s all being put online. You have digital courts where folks actually can attend court or whatever virtually. You have much more telemedicine because they finally repealed these laws that were stopping doctors in Michigan from treating doctors in Illinois, like interstate licensure. So the reason that all that stuff, the prices have increased is 100%, in my view, a function of the state either subsidizing or regulating and thereby, distorting market forces and stopping new technology from entering it.
I shouldn’t say a hundred percent. Let’s say it is both the state and it’s the state’s indirect impact in terms of the cultural stuff the state sets up like, “Oh, you’re supposed to go to college before you get a job.” That type of stuff. Now, that’s all falling apart. Who wants to go to Zoom classes? People are realizing that a lot of K through 12 is actually babysitting as opposed to education. So maybe you can unbundle those and you have a learning pod and you have a babysitter at a learning pod and education is coming remote from a world-class teacher or learning apps, which these are really quite good. Those learning apps will let your child learn at their own pace. Maybe they’re really fast on math and so on reading, or vice versa. And now, it’s not the industrial age, factory style production of every child the same, which is simply not true.
There’s no such thing as a sixth grader. There’s no such thing as a 30-year-old. You have 30-year-olds who are billionaires, you have 30-year-olds who are in prison and you have 30-year-olds who are everything in between. And you have basically this huge spectrum of experience and difference. Actually, that goes very young. There’s 18-year-olds who are extremely good basketball players, 18-year-olds who are extremely good mathematicians. There’s 18-year-olds who have quite a lot of differentiation by that age already. So I think that having an educational system that speaks to that is important.
We’ve mostly gone after higher ed so far in tech, because it’s been a little more amenable to choice. There’s the concept people are adults and they’re able to choose between things. So you’ve had Coursera, you’ve had you Udacity, you’ve had Udemy, you’ve had all that stuff, which is good. You’ve got Lambda School and so on now I’m an investor in Lambda School.
David: Write of Passage.
Balaji: Yeah. But now, it’s finally going into K through 12 and of course, lots of kicking and screaming. People are freaking out about it, but yeah. So finally, tech is getting into education. I think it will bring those prices down.
David: Yeah. I want to sort of get back to the very concretely, what kind of advice you would give to somebody who wants to learn how to teach because your MOOC was, to the best of my knowledge, one of the biggest MOOCs in the world. But as you think about that, I have a comment. So one of my Write of Passage students, she’s so talented and she would routinely ask these incredible questions. Her name was Taylor and I would talk to different people and they’d all be like, “Wow. She’s really smart.” So eventually, we looked her up in terms of the survey data and I was like, “Wait, she’s 17 years old. How could this be? That she’s 17. She’s asking all these great questions.” You know when you meet someone who is a really good talent, but like an idiosyncratic talent? You’re like, “I just need to talk to you. So I call her up and I’m like, “Hey, what’s your story?” She’s from an hour North of Victoria, Canada. She lives on a small island, and she took online courses her entire way through high school because she didn’t want to go to the local public school. So I was like-
Balaji: Her sophistication is just way off the charts.
David: Sophistication is way off the charts. But then there was something else too. I think that humans are like this and the internet is beginning to expose this. She was more polymathic than almost anyone I know. So I was like, “What are you interested in?” She’s like, “Oh, I’m really interested in sort of theoretical math and physics, but I also do drawings.” And I was like, “Okay. So are you thinking going to college?” She’s like, “Yeah, I’m going to go to college, major in the liberal arts.” And I’m like, “Wait, what? I’ve never met a 17-year-old or just about anybody who has studied all these things in depth,” and I think that then this spoke to what you were just saying about age is just a number here. There’s different kinds of aptitude that are becoming legible now and now, you don’t need to be on the conveyor belt of how old you are. You could be on different kinds of conveyor belts based on wealth or success or talent in a certain area. And I think that’s really interesting.
Balaji: Yeah. And I think also, Blake Ross, for example, famously coded Firefox when he was a high school student. There’s more and more examples of this young talent. I think actually, we’re sort of going back to the future where in the 1800s, kids worked very early on. The reason that Jebidiah and Abigail would have six kids is because they were actually cashflow positive in the sense of those kids were quickly picking cotton or they were going in and mending fences, milking cows, doing all these farm type things. So the thing about that was those kids, they were not infantilized. They were very quickly given a level of responsibility that was commensurate with their ability to follow instructions. They were doing chores as soon as they could understand what chores were.
So because of that, folks became adults relatively quickly. They married early, all the type of stuff. Then what happened was by the late 1800s, early 1900s, sending your kids to work early on meant sending them now increasingly than farm, sending them to a factory where there’d be wage labor. The difference, and this sort of a flip way of putting it, but I think in an interesting way of putting it is those factory owners were not equity aligned. They didn’t love the kids. The parents, when they sent their kids out to do chores, you wouldn’t work the kid to the bone, of course. You love your kid and you weren’t going to harm them. These chores would be good for them. They’d learn some discipline or whatever.
The factory owner would not necessarily have such a constraint. So child labor obviously got a bad name because some unscrupulous factory owners would push these kids too much. So what happened was basically a backlash against children doing any kinds of work. So that led to the 20th century and basically, the post-war rise of teen culture and something we take for granted, really wasn’t a thing because teens didn’t have consuming capabilities.
David: I thought Hot Topic was a 18th century staple of malls. Is that not true?
Balaji: Right, right. So, so essentially… Yeah, no, exactly. This entire kind of post-war consumer culture with teens and young adults being so important and whatnot, it was something where those kids were used to be producers rather than consumers. I’m not saying a hundred percent, but that was a big part of the trend. It was only like a prince who could afford to not work until they were 20 something and just learn the fine arts. So because of that, this reaches its apotheosis with the idea of being on your parents’ health insurance until you’re 26, right? So this kind of period of infantilizations, period of, “Oh, you’re not independent. You’re not making your own way. Failure to start. Failure to launch,” all the type of stuff that people talk about is in part, coming from this well-intentioned thing of, “Okay. Let’s stop children from being exploited by unscrupulous factory workers,” et cetera.
Now, what’s, I think, happening is we’re coming back to the future, which is thanks to remote work, thanks to pseudonymity, thanks to cryptocurrency, thanks to software development and online writing and so on, you can have a kid like your example of Victoria, like Blake Ross who did Firefox, who can set up a pseudonym online and on the internet, nobody knows you’re 14 years old. They can do professional software development or writing or anything. So long as they’re sending the right symbols and getting the right symbols, getting their work done on time, they can work and crucially, this is important, a parent can supervise them. You can know, because they’re at home, you can just watch them, they’re just typing things into a screen. They’re not working at the Triangle Shirtwaist Factory. They’re not at risk of having their finger singed by some bad piece of equipment or something like that. The maximum they could do is spill coffee on their laptop if you have them drinking coffee and their teens.
So their injury profile radically drops. The degree of exploitation… It’s still possible of course, but many of the concerns of physical damage associated with child labor go away. So now, what you have, I think are lots of kids who are going to be remote working at a much earlier age. Also, you combine that with the fact of folks needing more money. The US economy is contracting and folks are now… Basically, family units are getting crushed back together because everybody’s stay at home all the time. So you’re sort of getting this interesting steam punk, Neo modern, techno archaic future, and this is something I think is a deeper point, but our future is more like the 1800s and the 1700s than in the 1900s. We’re sort of going in reverse. 1950 was kind of a mirror image moment, and that’s the peak of the centralized century. You go forward and backward in time, you get more decentralized.
David: Top of the mountain. Yeah. It’s funny. It’s interesting to see the adulting meme on one side, #adulting, “Oh, I’m 27 years old. I have to go do my laundry, and this is such a big moment in my life that I’m going to get out of bed before 2:00 PM on this Saturday to go do by my noble duty of cleaning my clothes for the world.” And that contrast with what we were talking about with metallic, but also I’m on the record saying that there’s going to be a teenage billionaire by the end of the decade, because if you take everything that you just said and then you add the leverage that you get with software, to me, that seems like almost a logical conclusion. I guess my question here would be, let’s move into writing and getting people to write about these technical subjects at a high level. That isn’t really happening that much.
So there’s the incentives for high level companies to be founded that are decent in terms of it’s pretty clear how you can make a lot of money doing that, but this is something that you and I are both really interested in. How can we use the internet to incentivize and allow people to write really well about super technical subjects, that it doesn’t quite feel like Substack is there, even though I’m a big fan of the company and know that they’re onto something, they only feel like a part of this project of increasing the quality of information on the internet.
Balaji: Yeah, I think that’s right. So by the way, there’s certainly our teenage billionaires, but they were born rich rather than built rich. And by that, I think that’s a very useful kind of one-liner. Rich it obscures more than it… I think born rich is different from built rich. Did somebody hand it to them or did they build their fortune? And there’s a huge, huge difference between those obviously. But I agree with you that we certainly have 20-something billionaires. We’ll probably have a teenage billionaire if we haven’t already had one. I do think that all these people who kind of tweet, “adulting,” and I’ve actually feel that’s… So it’s interesting. There’s an aspect of that, which is related to progress in the following sense, which is the more things that can get done without you thinking about them, the more civilizational progress is.
For example, I hit a key to send an email and there’s so many things happening. I depress the key, there’s the capacitance that changes when I hit that key, the wireless keyboard, okay, 80211B, or Bluetooth will send it to the laptop, laptop will trap that event and it’ll turn into packets and it’ll… So 500 things that are happening there, and you’re not thinking about any of those things. You’re certainly not milking the cow to see it happening. Exactly. So progress is abstraction. And that person who’s saying, “Oh, I’m adulting,” is like, well, all those things that they were doing were behind levels of abstractions. They didn’t care about them. The issue though, is that you do too much abstraction and a lot of people, just humans being humans, they get alienated from that and they think it’s easy when it was actually ridiculously hard to put all those things behind an easy interface. Really, really, really hard to do.
David: It’s like Amazon. “Wait, the boxes don’t just fall from the sky and land on my front doorstep?” No, there’s entire supply chains. There’s trucking companies. There’s how that cardboard gets made. It’s like all these things are extracted away.
Balaji: Exactly. It is really hard to make something easy. So the problem is… Yeah, that’s a good one-liner, right? Yeah. So it’s really hard to make something easy. So the issue is that if you’re a producer, if you’re a founder, you have a respect for everything else out there because you’re like, “My God, that was a lift to do that from scratch. That was very difficult.” And then you’re also actually much less critical of products on the internet because you’re like, “I have some empathy for the person on their side. Yeah. This product sucks, but I’m sure they’ve worked hard on it,” and you’re like, “Hey, do better next time,” or whatever. “This is what I wanted to see.” You’re just more constructive in your criticism of it.
David: So when it comes to getting people to write good ideas, technical ideas online, there seems to be a big hole in the market. For example, yesterday I was walking the beach in Los Angeles and there’s this guy. He’s taken a selfie, and so I go up to him and I’m like, “Hey, can I just take the photo for you?” Because it was kind of awkward when he was doing. So he’s like, “Yeah, sure.” So I snapped the photo for him. We ended up talking and he had been a geologist for 37 years. He’s based in Houston. So I’m like, “Oh, so you work in oil and gas?” He’s like, “Yeah. How’d you know?” I was like, “It’s kind of obvious.”
So then I looked down at my foot and I have a big thing of tar on my foot and I’m like, “Why is this?” And he’s like, “Oh, that’s actually pretty normal when you’re on the beach.” And then he starts talking to me about all the developments that have been made in geology. I was like, “Wow, I had no idea that any of this existed,” but crucially, that information, at least some level of it, isn’t really well written online because there isn’t an incentive for him to take that and then publish these very technical ideas and it seems like this is as close as we can get to a free launch of raising the quality of average online information, saying Wikipedia isn’t the end of history here. It’s actually just that one.
Balaji: Absolutely. That’s right. That’s right. So, okay. So this gets into kind of send the stuff that I want to fund and I’ve been thinking about. So I have a landing page by the way, if folks want to go and put in their info at mediafund.com. This is just the landing page right now. It’s a twinkle in my eye, but let me talk about what we’re thinking about here. So typically, and I’ll get to your point on incentivizing the useful content online. Typically, people will put, let’s say, war reporting and Kim Kardashian reporting at opposite ends of the spectrum, like, “Hey, this is super serious. Pulitzer,” and this is fun infotainment or whatever. I actually argue they’re both infotainment at one end of the spectrum, and the other end is news you can use or tutorials. The differences, is this piece of information directly relevant to your life? Are you going to in particular, spend the effort to check every line of this information?
For example, with the tutorial, it has built in fact checking because you have the tutorial on the left and you have your terminal on the right, and you’re typing in line by line to check that it works. Or if you’ve got a YouTube video that is showing you how to sew something, or how to build a table. You’re checking every line of that, every frame of that as you go and build the thing. So they can’t, “lie,” to you, not that they would want to. They can’t misrepresent it. They can’t distort it. They can’t understate the complexity or overstate certain things. It’s instructional, and because it’s instructional, it’s leveling you up and you’re also fact checking it, which is non-obvious. People don’t think about that part. Similarly, news you can use is like the weather, or a stock price. This is something where it is upstream of an action. If rainy, then umbrella. If high stock, maybe I sell. If low stock, maybe I buy. So that’s information that guides decisions. It’s right next to the tutorial stuff.
On the other end of the spectrum is I read about some distant issue in so-and-so country and unless I have relatives there, or if have businesses there or operations there, it’s really not relevant to me on a daily basis. The thing is that people might say, “Oh, you should be concerned about this and that,” the problem is that there’s 7 billion people on the planet and to even know each of their names, this is just kind of a toy experiment, but there’s 86400 seconds in a day. So in 12 days, if you learn one name a second, 12 days, you could learn the names of a million people. Okay? And it would take you 120 days for 10 million people, and 1200 days for a hundred million people, and 12000 days for a billion people and basically, your whole life to even learn the names of the 7 billion people on the planet.
So therefore, you cannot possibly care about them equally. You have to somehow triage or somehow prioritize with the assumption being that to even care about someone, at least you need to know their name to personalize it. So you have to rank order what you’re paying attention to. The infotainment stuff should be filtered out of the feed. When we we’re thinking about this is the analogy between your diet and your information diet. So your diet… Look, having a cookie from time to time might be fine, but if you’re only eating cookies and you’re not eating healthy food, you’re going to have negative metabolic consequences. Your health is going to be messed up. Your life is going to be worse and so on. In the same way, I’m not saying never have fun, never watch something interesting, but I do feel that Twitter and a lot of social media are like a restaurant that’s learned to put sugar in their food.
David: They’re giving us intellectual diabetes.
Balaji: Exactly. That’s right. They’re basically stimulating us. I actually think that, you could make this totally quantitative where there are devices, for example, they’ll monitor your blood glucose. So you can see after you eat a cookie or a banana, you can see a spike with some time delay. So you can actually see what it’s doing to your body. You can see it on screen, which is actually pretty awesome. Okay. The equivalent is you are reading Twitter or whatever, or social media, and we should be able to see a spike in cortisol or something like that. “Oh, that stresses me out.” “Oh, that concerns me.” Or it’s cloying, emotionality. You know what I mean? Like, “Oh, it’s so cute, this puppy,” or whatever. So it’s meant to provoke an extreme emotional reaction and it’s coming in through our eyes and ears in much the same way that Hollywood has learned to put sex and violence into movies because it’s stimulating. It’s like you see this and your reptile brain is like, “Oh, violence. I must pick up an ax and join in.” It’s real.
It’s exciting. It’s cool. You pay attention to it. So we need is, I think, a different form of media that is more like the news you can use in tutorials and the infotainment. It’s all about relevance and skill building. It’s like an optimized information diet. So this gets back to write down your goals. Your dad was smart. So what skills do you actually want to build? What is your intentionality? To some extent, following people on Twitter is so sort of there, but I think a better… One framework for this is the mnemonic framework by Michael Nielson and Andy Matuschak. So the idea is rather than reading recreationally as we do now, if you’ve read in the lean forward way if you, if you read to remember, they’ve got a site, quantum.country where they have spent a lot of time on essentially, an application where you read and it immediately prompts you for whether you recall the stuff. It’s got a question there and it asks you whether you remembered it.
It’s optimized for mobile and so on. So you can still do it in a lean back kind of way, but then sometimes they’ll ask you to break out a pen and paper and actually learn something. Crucially, because they send these kinds of email reminders to people, they find that people’s retention is actually much higher than they would expect. It’s a finite amount of effort that gives an exponential amount of retention. So it shows you can spend your time in a much better way by learning things very fast. This is kind of related to the shams thing I talked about earlier, where you solve problems and if you solve the problems, you understand what’s going on and then you can fast forward and skip through like a placement test kind of thing. So I think that a big part of the next generation of content for allowing people to level up can’t be like the current social media where you interact with it by liking and RTing. It should be about learning and earning.
David: It’s purely passive. What you’re saying is a much more active approach. So I open up Twitter and there’s… It’s interesting. It is very personalized in a way, but in what you’re saying, I’ve never thought about this before. It’s actually very unpersonalized. It has no idea. Here we go, dad. It has no idea what your goals are.
Balaji: Exactly. That’s right. Exactly. So one way of thinking about it is social media and Netflix and so on are optimized to just consume as many minutes of your day as possible, literally addictive. There’s a totally opposite design paradigm, which is give the maximum value to somebody in a minimum amount of time. How can I give somebody a hundred dollars in one second? Now, what we do with earn.com before we sold it. Actually, earn.com is still this. Coinbase earn is like that. We can give you $5 of value in a few seconds. You go there, you answer some questions on cryptocurrency, you earn some cryptocurrency. That concept of tasking over the internet, you might say, “Well, how could you run a hundred dollars in a few seconds?”
Well, let’s say you’re a doctor and you are really good at diagnosis. I could probably get you a stream of images, which you could help label. We might automate a machine learning algorithm. [crosstalk 00:54:28] Yeah, yeah, yeah. Exactly. People have actually proposed this. And it’s actually, “Is this cancerous or is it not? Is this cancerous or it’s not?” And it sounds flip, but it’s actually something where you’re downloading the information out of a skilled practitioners brain and into a machine-readable format. Those zero and labels from a thousand radiologists could train an algorithm, or it could supplement an algorithm where the algorithm can’t decide and it kicks it over to a human for the hard cases.
David: Well, it’s interesting because this is one of the things that Mandelbrot talks about in, “(Mis)behaviour of Markets,” is that a human’s ability to use their perception to gauge what a graph looks like, this is where humans are really good. Nassim Taleb was very influenced by that book. If you look at his citations throughout the Incerto, he’s always citing Mandelbrot. A couple of weeks ago, he went on Twitter and it was sort of like a riddle of different numbers, and which one is more likely just with statistics. Taleb said, “This is something that you want to eyeball.” So there’s a certain class of, we could call them eyeball problems, which computers are way worse at solving than humans, and then we could get Tinder for cancer for those.
Balaji: Yeah. It’s interesting though because once you can teach a machine to do something, it’ll usually do it better than a human.
David: Once it’s made legible.
Balaji: Yeah, exactly. So for example, chess, or go, it is non-trivial to teach them those eyeball problems, very non-trivial. You often have to think very deeply about what a human is actually doing, extract the training information. There’s various techniques, right. But once you do, it surpassed it. It’s kind of like with flight there’s aspects of… We don’t yet understand how agile the bumblebee is, or how it can fly, how it does. At least that’s apocryphal. We don’t fully understand it. And there’s aspects of how birds are really agile that we haven’t yet been able to figure out. But we have been able to make machines that fly faster and higher than any animal that we know of. So, basically, there is something to the idea that once we can put it into the machine format, we can exceed, and it’s just non-trivial to do that. I mean, brain-machine interface, another great example of this. You just turn it into something where you can read signals from the brain, and you’re like, “Okay, well, that’s what an A is encoded as.” Boom. There’s devices now where you can think and type. It’s telepathy. It’s pretty awesome actually.
David: So, let’s talk about media fund, which you mentioned earlier. So, let’s take all of this back into, very tacitly now, how do we create incentives? You were talking about this twinkle in the eye vision. What is that thing?
Balaji: Yeah. So, one thing I’ve been thinking a lot about is that if you, as a writer, add value to your reader, you should get a cut of that. Okay. So, the venture journalism or venture media model of which there’s a few, but basically there’s kind of two of them. Right? One is, let’s say you’re interested in fusion energy to solve climate change or genetic testing to help with COVID or self-driving cars to reduce traffic fatalities or encryption to protect privacy. Right. So you’ve got some social problems. So you are concerned about a problem, or you’re just excited about technology or either one, and you’re writing about it, and that inspires an entrepreneur to go and start a company, and it could be writing by the way, but it could also be video. It could be images because… Let’s call it full-stack media, right?
So, you inspire someone to go and start something, and that could be, for example, the movie Minority Report inspired them to the Kinect at Microsoft, or you had Nicholas Kristof in New York Times wrote an article that inspired Bill Gates to go and do these decentralized toilets in Africa, which didn’t require connection to a central grid. So those two are cases with big companies. But there’s many start-ups that were inspired by a piece of writing. And the thinking is, okay, you inspire them. What if you get a cut? Where they seek investment through a form on your site, and you get an affiliate referral, right? So they click through, they set up, let’s say an AngelList profile. We have the record of the fact that they listed email@example.com as a referrer, right. Or they came from author.com’s domain, just like we track referrals all the time now with internet advertising, right.
And if we invest X dollars, 10% of that goes to the author and they could take half of that inequity and half of that in cash for example. And so just as a toy example, let’s say that there’s an entrepreneur, their company’s valued at 10 million, investors put in 500K, 50K of that goes to the author and 25K of that is cash today and 25K is speculative equity. And then in 10 years, that’s worth two and a half million dollars because that fusion energy company exits for a hundred X. Right? These are not totally insane numbers. This is possible. It goes from a 10 million valuation to a so-called billion-dollar valuation. So, here’s what that does by the way, that changes the incentive structure of the writer to go from quantity to quality. Now, it doesn’t matter how many page views you get, or rather that’s not the only metric. What you’re trying to do is find the next Zuckerberg or Vitalic of fusion energy and attract them to your blog or your outlet.
What that may mean, by the way, is you only have 10,000 readers in the world. These only people who care enough about this fusion energy as news they can use as opposed to curiosity, but your discussion is pegged at a very high level. Right? You have someone who is able to add value to the best people in the world, and they look back, and they’re like, “Okay, this is awesome.” Like what Ernest Cline did with Ready Player One. He helped inspire Oculus. He could’ve gotten a cut of that.
David: Yeah, it’s interesting because I’ve seen this with Write of Passage where, when I first started the school, I wanted to have the world’s biggest writing school, and then I wanted to have the world’s smartest community of online writers and it has totally changed how I write. I haven’t looked at my page view metrics in months. I used to look at it every single day. And so what ends up happening is you begin to think in terms of attractors. And right now, there aren’t really metrics that incentivize people to actually attract higher-level people and who could build a company like this. But once you do begin to have those incentives, you then… And I’ve realized this in the last couple of months, oh my goodness. The kinds of articles that I write now have a level of meat and nutrition that I was actually incentivized to write intellectual junk food when Google Analytics was my proprietary measure of how well I was doing.
Balaji: Exactly, that’s right. You will get what you optimize. Right? And so, if your dashboard is how much… So, by the way, there’s another aspect of it, which is how much your users have learned. So the mnemonic dashboard, how much they pertain for your writing. You’ve actually increased your human capital. And then how much they’ve earned. Did they raise money or and not everybody’s going to be an entrepreneur or a founder. It’s a relatively small percentage of people. But there’s a more repeatable process, which is recruiting. So. You didn’t just level them up and teach them fusion energy, or what have you. You may have gotten 0.1% of them are 0.01%. You made them a founder, but 1% you got them a job at a fusion energy company, and then you get a cut of that.
David: Why is this so important? Why writing? Why are you focused on this? Why is this the thing that you’re now attracted to? That’s not really clear to me.
Balaji: Huh. Okay. That’s a great question. It’s because media and social media are upstream of everything. Code is how you have machines know what to do and media is how humans know what to do. And technology has focused on software, which is how we can gain lots of value out of machines, but we have focused less than we should have. I mean, it’s funny to say that because social media is so huge and whatnot. We focused a lot less than we should have on first-party media in terms of increasing people’s human capital. And again, there’s been a lot of efforts here. There’s online education. There’s a lot of good stuff out there, but what we really haven’t done is seek to build the full stack narrative. Oh, that’s something else by the way. Everything I’ve described up to this point may be sound like, eat your vegetables, which is education versus infotainment. But there is Thesis-Antithesis-Synthesis where you can make it edutainment. Really entertaining, but also highly educational, textbooks that read like novels. And I think that’s possible to do it just hasn’t really been tried.
For example, there’s this famous book on statistical mechanics, opening to the 1975 book, States of Matter by David Goodstein. “Ludwig Boltzmann who spent much of his life studying statistical mechanics, died in 1906, by his own hand. Paul Ehrenfest, carrying on the work, died similarly in 1933. Now it’s our turn to study statistical mechanics. Perhaps it will be wise to approach the subject cautiously.” All right. That’s a great hook. At the beginning, it’s like forbidden knowledge. Oh my God, what did they see, kind of thing. But of course, it’s a very technical topic. So, I think that can be done where you apply many of the techniques of Hollywood and of screenwriting and so on to this kind of content. And it just hasn’t really been tried yet. I mean, again, it’s somewhat, these educational apps are out there. I’m not saying there’s zero, but I feel it hasn’t been direct in the gun sites.
David: I think you’re missing feedback loops as a critical component of this. As someone who teaches and one of the things that we do is, we try to make as much of the student experience legible as possible. And so for example, during class, at least once I will ask a question and I will get hundreds of paragraph responses from my students, I can also follow the chat, and after every single lesson, I have five to 800 chats that I can look at to gauge the temperature of the class. We have our entire circle forum of places where students are engaging and interacting and I look at all this. And so I can actually understand how my students are feeling so much better than when I would go to college classes, I would sit down, I’d go to the class. I wouldn’t give much response at all, except for maybe head nods.
I think that what you’re saying here is that what we can actually do is begin to understand what parts of the educational experience have this drop-off rate and then in the same way that Netflix has a lot of data on how their TV shows are actually performing. If we can then apply that to the educational experience, it can be so much richer. I’m not sure that necessarily it hasn’t been tried before. I think that there’s actually a 10 X, 100 X of legibility that then allows us to gain visibility into what students are feeling and where they’re struggling.
Balaji: So, that’s really good. So basically, first of all, that’s good feedback on feedback. Let me sort of integrate that with this because I agree with you since having taught, that’s also super important. The aspects where I was thinking about that were, A, the mnemonic dashboard. So you can see where your students are learning. So that is feedback of a kind. B, is I think every new media company needs to be a media community as well, where you have a personal relationship with your 10,000 readers in, for example, locals.com is like a social network in a box. You put up your content, but you also have a relationship with your readers. A newsletter is just a broadcast, but you really want like a mini social network.
I mean, there’s comments on it, of course, but maybe you want something persistent, like a chat room where it’s not just… It’s like a guided chat room, perhaps where you’re posting first-party stuff. And there’s a continuum between a blog where somebody comments, and there’s one moderator and then a group blog, and then where there’s a few moderators and then social media where there’s no moderators beyond Twitter, and it’s just a free for all.
Who’s posting? Who’s giving feedback to each other? I do think that the blog or group blog model is actually under-appreciated where it’s a moderated conversation. A group blog with a social network, I think, is a less explored format on the internet that I think we could do a lot more of a little bit like the faculty of a university and all of their grad students being in the same chat.
David: Right. What I’m hearing, you say, is you get to kind of create your own university here. That the university starts with the network, you become a magnet for talent. Then you then have your own incubator coming out of a university that then creates start-ups like a YC model. And then you take a cut of the magnetism and the network that is built.
Balaji: That’s right. That’s right. And the thing is I’ve had this business plan, and so on since 2013, I’ve just been doing other things since then. And with now where just tech is in 2020, maybe it’s the right time for something like this. The big thing about it is it’s extremely scalable because once you can actually… so the part that’s not scalable is that personal feedback. But the part that is scalable is everything associated with mnemonics, everything associated with content that trains people and so on and so forth. So you might do something where, for example, if you want to become an expert on platinum and platinum mining and the industrially uses of platinum. There is some stuff out there it’s not bad, there’s Wikipedia, there’s so on. But there isn’t to my knowledge at least, a platinum university. Right?
And so that kind of thing where it’s contemporaneous, because a lot of textbooks don’t get updated fast enough, and so on. This would be the place where you’d come to become a world expert in this keyword. And I think that building a community of those folks interesting question is always the question of, do it fully decentralized or do it centralized, view it centralized, you have a consistent format and style and aesthetic. If you do it decentralized, then you can have a hundred people who are motivated as CEOs and that may get you more in the long run, right? So, there’s arguments on both. Y Combinator is decentralized because Paul Graham goes and funds a thousand people, they do different things, but something that’s achieved a vision is typically centralized. Elon is centralized. Both are reasonable. I do, however, think that there’s a big gap here, and it’s still… what’s funny about it is despite all the online education, I think, as you said, Wikipedia is not yet the end of history. We’re not yet there on this.
David: Yeah. What I hear you saying is that there’s a last mile problem with knowledge. Where, when I was talking to this oil guy on the beach in Los Angeles, he already had all the knowledge. That’s actually the hard part. But what’s happening is he’s not sharing the knowledge. A lot of people think, and this is why I’m so loud about trying to get people to write. A lot of people think that writing is starting at square one where, “Oh my goodness, I need to start writing.” But people who have years of experience or have read X number of books traveled to Y number of countries, started Z number of companies. They’re actually at mile 24 at a marathon, where they actually gaining the experience is the hard part, making it legible to the world is what we are struggling with as a society, in terms of how many experts there are versus how few of them actually share their knowledge. My question to you is I kind of think of you sometimes as the pseudonym guy. How does that factor into this? And in terms of the future that we’re going to?
Balaji: Yeah. So, I gave a talk last year called the synonymous economy, and that’s maybe useful for folks to just Google and watch it on YouTube. And one thing I think is going to happen this decade is that people are going to stop using their real names online and start using pseudonyms more and more. And I also think the specific kind of pseudonym that will get popularized will be a crypto domain name. So, David Perell.ETH, well, that’s actually, it’s got your real name before it, but you might actually use fubar 97.ETH. And the thing about this, that’s powerful about crypto domain name in particular, as a pseudonym. People here may not know what a crypto domain name is, but you can Google ENS. That’s the Ethereum name system, or handshake, and read about it.
It’s a domain name, like David pearl.com or fubar.com, but it also has encryption and payments and all the other stuff built into it. You could send an encrypted message to that crypto domain name. You could send a payment certainly to it. You could send other things like digital art or digital certificates, a certificate of graduation of Rite of passage. You could confirm that that person graduated Rite of passage at a particular time.
You could see their reputation because they can log in to online services with that crypto domain name. You can see whether they’ve staked a balance, for example, they staked a balance of a hundred bucks so that they’re a trustworthy actor when they log into a platform. And they’re also likely to be real and not a troll. You can do things like they accumulate karma under one name, and they can transfer it privately to another pseudonym. So this way, you can build up karma in a community and then transfer to a pseudonym and then speak more freely in that community than you normally would. Right? All of these are types of things which I think allow for people to speak their minds freely on the internet and earn pseudonymously without being attacked. And I think that’s going to be a huge part of the 2020s is this synonymous economy with crypto domain names as being the kind of pseudonym.
David: The other thing that you’ve been talking about as we’ve spoken recently, and you’re talking about this with truth and verification, is something that you call the ledger of record. So why don’t you give us just a brief background of Oracle’s The Oracle problem, and then let’s work our way up the ledger of record.
Balaji: Yeah, totally. So to introduce first, the concept of crypto Oracles. Right now, if you want to find out the weather in Chicago, Illinois, back in 2014, January 15th. You’ll go and look at the weather channels records on that. If you want to look up stock prices, you’re going to go and probably look up, let’s say, Bloomberg’s records on that. If you want to know real estate prices in a particular area, you look up Zillow’s records on that. So we have a system right now where there’s tons of tons of data that’s online. It’s in these databases, but it’s siloed. It’s behind paywalls. It’s not all that queryable. It’s in these companies’ data sets, but it’s not out there on the internet. And, an important point is that many stories today are effectively wrappers around this kind of data, for example, sports articles are wrappers around the feed of data from the NBA, and financial articles are wrappers around the feed of data from Bloomberg and many political articles are wrappers around the feed of data from Twitter.
And so if you think about it, they literally have embedded tweets. This is progress in a sense where you have the letters sitting on top of the numbers. Each of those posts is fundamentally a numerical data structure. Even a tweet, even though it’s words there, there’s numbers that say when it was the timestamp, where was it posted from, what device, what account, all the type of stuff. It’s literally a data structure, you can print the so-called JSON of it. So, the concept is all right, if all of these articles are based on feeds of data, what if we took all of the feeds of data and we put them not just online, but on-chain. Okay. So, on-chain is the next step after online. And what that means is it’s not just accessible on a server that might go down at some point, it is posted on a blockchain, which will never go down.
So you saw the permalink problem. You saw the link rot problem. Blockchains don’t go down since they’re distributed. So, and there’s other kinds of data structures you can use. You can use IPFS or a system like that for decentralized data storage. Okay. Blockchains are not the only thing, but you post it on-chain or the equivalent. And what you can get is, especially with post-it on-chain, you’ll get timestamps, you’ll get a digital signature as to who posted it. You get a hash of what was posted. And so you can prove mathematically who posted what, when. Right. Or at least what entity would this digital signature, what data at what time, and why is this important? Because it’s digital history. It’s unfalsifiable history. So to give you three examples of how this record of all of these facts could resolve things.
Six years ago, Tesla had a issue where NYT wrote an article saying that their car had stalled out on the side of the highway. And they pull the instrumental record, and they showed that that article was not true, that it had not actually stalled out in that way. Right? And so that was something where the instrumental record, the ledger of record was able to falsify the words on top. The numbers took precedence over the letters. Second example, last year, the Brazilian fires, Emmanuel Macron retweeted a photo of supposedly of the Brazilian fires, and the New York Times printed this tweet, put it in an article. And it turns out that photo was taken by a photographer who died in 2003. So the photo was at least 15 years old. And so again, a timestamp basically show that the story about the Brazilian fires was false.
And that was actually a big deal because people were talking about international intervention. Oh my God, Brazil is letting the Amazon burned down and we need to stop it, et cetera. But the information was fake. It was not or at least some of those photos were false. Third example in China recently, two years ago, a court in Hong Sao went and proved that a patent dispute, the litigant was incorrect because the defendant had actually put their patent or the contents of patent on-chain before that infringement had actually occurred. So a third example of where a timestamp resolve the dispute. And to make this even more obvious, the crypto space is worth hundreds of billions of dollars. And the whole reason is because you timestamp as to who got what money, when. Who owned one scarce property when. Right.
And so, the simple kind of mechanism of the timestamp is… you think about Nixon. Right? With Watergate, what did they say? What did he know? And when did he know it? So digital history as to what happened, when, who did what, when, is way more important than people think, right? Go back to the timestamp, go back to the video camera, go back to the footage, find out what happened. If you had that in an undeletable way or a public way or all of the above. You’d have what I call the ledger of record, where you take… So what’s a crypto Oracle, by the way, coming back to that definition, imagine if the weather channel didn’t just post its weather online, where it could go down, but on-chain where it couldn’t. Whether if Bloomberg did the same if Zillow did the same. Right.
That would mean that you have a ledger of record where it’s like, you hit a button, and it’s for the record. So you’re posting this for the record. That’s undeletable, and such as undeletable, it’s also publicly accessible. You might make it something where you can do something in between, by the way, you could make it something where it’s accessible to those people who have a cryptographic key. Right? But the point is that this ledger of record, as you get feeds of not just sports scores and financial stuff. But every tweet, every real estate transaction, every coronavirus death, every weather channel post. You start to have a library of all the events that ever happened. And now, any story you tell about what happened in the past, better reference, the ledger of record. And the thing about it is, on the one hand, this is not that big a change in behavior. It’s like tweeting, except you tweet for the record by posting to ledger record. And it’s like linking, except rather than linking to a website, you link to a blockchain or a block Explorer, and that’ll show you this undeletable thing.
David: So why is this important besides the novelty of it? It’s not quite clear to me how this is something that’s worth being excited about. It sounds like, oh, cool this is a new technology, but the way that you’re talking about it, it’s like, “No, this isn’t a bunch of interconnected cables in 1985.” This is the Internet. And it’s going to have some fundamental changes.
Balaji: Absolutely. This is probably I think one of the most important… I mean, we have to be able to execute on it, but it’s already happening. You’re having these scripted Oracles be set up. This thing is going to be one of the most important technologies of this decade. If we can popularize it and make it a thing. And the reason is, let me give you a very contemporary example. Okay. How many coronavirus deaths have occurred? Right. This is a number out there I think the latest was about 170,000 in the U.S. And the issue is that folks will distress this, “Oh, these are folks who have heart attacks or diabetes. They didn’t actually die from coronavirus” and so on and so forth. Right. And so there’s layers, you just see that number.
Right. But if you had something where every hospital was posting a feed online. Okay. And crucially, not with private patient information. All right. And that’s where the encryption and permissioning and so on. These are important issues. I don’t want to gloss over them. There are things you’ll have to figure out technically. But they’re obviously giving summary statistics already, deaths and so on. So clearly, that’s not a patient privacy issue. So if there was a feed of every single coronavirus death or case, or release, or diagnosis was a single tweet worth of data. Okay. It’s a little more than a tweet, but just conceptually. And all of that was posted on-chain. Then anybody who is skeptical could hit enter and pull all of the raw data. You would not need to take it on faith. You’d have to see that a thousand hospitals, all of whom had reputations and names and had been trusted and so on.
A thousand hospitals were all going and posting this data over time. Once you’ve got that, that’s a level of granularity where if you have something where it St. Jude’s, and it’s Cedar Sinai, and it’s Harvard’s hospital. And by the way, for example, all the Chinese statistics, right. One of the issues with this with coronavirus in the early days was people were like, “Oh, I can’t trust anything that comes out of China.” Okay. You can trust something that comes out of China. You know what that is, is their blockchain transactions because they couldn’t falsify that. Right. There’s people in China who are sending and receiving Bitcoin, and those transactions are getting posted. And by the way, that’s also kind of the future I think, of international trade is, these folks in different countries, you may not be able to trust each other’s legal systems, but they can trust digital currency.
They can trust the on-chain substance. You can fully audit everything. You can see it all. Right. So the point of this idea is let’s say that folks in China had actually done this. They’d posted on chain right. Now, basically, it’s much more auditable, right? There’s still ways that you might be able to falsify it, but they start to get really expensive, depending on how it’s done. As an example, would that Brazilian fire’s thing, right? Very few people could have realized that the timestamp from 16 years ago of this journalists dying in 2003 would be relevant to stopping a war 15 years later. Right. And you’d have to in the blockchain mode, erase billions, trillions of records to go back and rewrite that thing from 15 years ago. Do you see what I’m saying? It becomes infeasible. You can’t do what Stalin did and Photoshop somebody out of a photo.
You can’t delete history anymore. And so that is where things start to get super interesting in terms of having a base of undeniable truth to realign society around. Because without that, you have something like COVID 19, which I think has really motivated a lot of my thinking on this, how do we get to consensus on those things that can be numerically demonstrated? And we always get, by the way, with this kind of discussion is the naive first order is, Oh, facts matter. And then the sophisticated, second-order is, facts don’t matter, man. Say it’s all narratives, and it’s all… you’re just competing tribes. And the facts don’t matter at all. Right. But the third order is actually facts really do matter because the narrative that is based on facts will in the medium to long-term have more contact with reality and therefore more technological and economic strength.
That last bit is critical. That’s why the U.S. eventually beat the Soviet Union is that communism required you to just lie more. They had to make up factory numbers because if you didn’t have prices, supply chains didn’t work, factories didn’t produce, and so on. So eventually wholly lies. Right. Like the kinds that animated the, Soviet Regime, unfortunately works surprisingly well in the short term, because you can bully people into conforming with them and so on and so forth. But in the medium to long-term they don’t work because your machines don’t work. You just become poorer as a society. You’re screaming these holy lies, but it doesn’t matter because other societies that didn’t believe them have exceeded you technologically and economically. That’s why this type of stuff is actually pretty important, the nature of how to get truth in a decentralized environment.
David: Awesome. What I want to do is I have a bunch of questions. I just want to sort of do rapid fire. We’re just going to skip around all over the place, but I want someone to sort of get a, I guess what you could call a Balaji explosion of polymath. Why is life extension the essence of technology?
Balaji: Ah, I’ve written an article on this, balajis.com, “The Purpose of Technology,” and it’s a non-obvious claim, but let me see if I can get there in a sentence. Which is, if the purpose of technology is to reduce scarcity, then the ultimate purpose of technology is to eliminate mortality. Let’s break that down. The first part, if X then Y, right? If the purpose of technology is to reduce scarcity… Why do we say that? Well, every technology that you’ve ever heard of, it’s usually described as, this makes something faster, or cheaper, or better, lighter, that type of stuff.
And why is it better if it’s faster? Why is it better if it’s cheaper? So if it costs you a dollar and it used to cost you $100, that means it takes a hundred X less money, but that also means it takes a hundred X less time because your life, at least some of your working life has turned into money, right? You spend an hour doing consulting, you got some money out of it. And rather than spending 100 minutes of your life, you spend one minute, now of time on using this because the technology improved it, it went 100X faster.
There’s actually a movie by I think, Andrew Niccol, called “In Time” that explored this idea of time as a currency. He puts in various dramatic hooks to make it seem like some people are hoarding time and immortal and others aren’t, but I don’t think that actually has to be scarce. The point being, that if you’re doing all these things in tech to produce scarcity, the ultimate scarcity reduction that gives everybody back lots of time is life extension or reversing aging and that should be technically feasible now. That’s the thing that people don’t get. It is possible to reverse aging.
It’s starting to happen in mice. The technology is actually moving, like Oliver Rando, David Sinclair, Laura Deming at Longevity Fund, there’s some really awesome stuff that’s happening in this area. So we should prioritize that because life extension makes everything cheaper. If you live to 150, as opposed to 70, if you knew that you had that many years, not just of being superannuated, by the way, you could also call it youth extension, right? Why can’t we drink a potion or take an injection or whatever and restore ourselves to how we were at 25?
And if you’re 25, by the way, you’re like, “Well, that’s fine. I don’t care.” But once things start to break, it’s like a car that starts to break down, you’re like, “Yeah. I wish I was 25 again. I could run faster. I could work harder.” Lots of folks want to be 25 again, right?
David: As a 25 year old I can tell you it’s pretty good right now.
Balaji: It’s pretty good now. That’s right. But basically what happens is just stuff that never bothered you, your elbow, your knee, or whatever, it just starts acting up. One thing by the way, that’s important though is, I made the analogy to a car breaking down, but it’s actually not a great analogy for the following reason. When you deal with mechanical things like cars, there’s a failure rate that you can model and some cars that are just lucky, never experienced any of those failures. And so, one out of 10,000 cars makes it 50 years, 100 years without being repaired, because the errors don’t cumulate. You might, by random chance, be able to make it.
Humans aren’t like that. There’s like a hard drop-off at a 120. If humans broke down like cars, there would be some humans who’d be living to a thousand, right? Instead, what happens is you have a very predictable and coordinated kind of thing where people kind of go gray, they get fat or whatever in kind of the same way as they age. So people believe, and now there’s a lot of evidence that shows, that this is actually a triggered and coordinated thing. Just like we know that you’re kind of genetically wired to go from a baby to an adult, it’s not like 25 would be your natural state forever, you’re genetically wired to die.
So maybe we can unwire that. And in fact, folks are working on that. This to me is also where, more broadly, we have to start talking in tech much more about our values in our valuations. Money is only a tool. And what really matters is building something you can’t buy. And that’s actually what improves life expectancy. Life expectancy by the way, used to be the metric of technological products. Most recently, it’s something where people are focused on stock prices or GDP, which are fine, but are ultimately human metrics as opposed to physical metrics.
Let’s say, they’re intra-human. I pay you X and you pay me Y, versus, I changed the physics, so this person went from 70 years old to 150. It’s a different scene, like a ratio scale variable versus one that’s not. Either, folks who are actually living forever or they’re living until each 400, or they’re not, whereas the stock price, you can kind of decide on that between two people.
So I think that broadly speaking, in the 2020s we’re going to start to see, not that we won’t see SAS apps and that type of stuff, that’s all fine, it’s good. But we’re going to also start to see more of a focus on the real stuff. Not just life extension, which is like the core of it, but brain machine interface, limb regeneration, curing deafness. I had an article that I linked there, like RNA injections that go and restore hearing, curing loss of sight with bionic eyes. A lot of the actual miracles, I think we can actually start making happen with technology and get the future that folks that have grown up thinking about
David: Yeah. Values to valuation. That’s a shot in the Balaji one-liner drinking game. I’m trying to think, is this a drinking game or is this bingo? I’m trying to figure out which one, but I could totally see 25 one-liner squares.
Balaji: Yeah, it’s good actually.
David: How would you use online writing to spin up a digital country?
Balaji: Great question. Herzl did it, where he wrote Der Judenstaat, and that actually founded Israel. So proof of concept. Basically what happened was, he catalyzed a reverse diaspora where folks around the world were inspired by this and came to Israel and so on, and obviously not without controversy, but it showed that basically about 50 years after he wrote it, he got a state. So first that shows that it’s not crazy to think about doing it. It’s possible.
David: It’s here. It’s just not evenly distributed yet.
Balaji: Yeah, exactly. Also, the Federalist papers, with Publius and so on, the founding fathers went and wrote synonymously and that became the base of the new government. They argued with each other and so on about how it should be. So how’d you do it? I would say, A, you can write online, build a community, B, give everybody VR headsets and start actually building a city in VR. And you can do this with Autodesk and so on, you can actually import the models into VR. I think, by the way, VR architecture is going to be a big thing this decade, just like there’s folks who are now… So many people are publishers, so many people are video editors. A lot of people are going to become amateur architects because it’s becoming more accessible.
VR will start to make people think again about innovating in the physical world, because you can do everything on the computer and then hit enter. So A, you build a community online with a purpose that comes from the writing, B, you simulate the city or town in VR, C, you enforce governance and physical levels of civility, that’s an important term. It’s funny because the physical world is now becoming uncivil, but just remember back to 2010 when people weren’t beating each other up in the streets and you could mostly walk around, now that’s something where civility is novel. It’s funny, but it’s true, right?
What I said to somebody a while ago is, “Twitter is not the real world, but it will be”, and that’s happening. Twitter fights are coming into the real world. That’s another part of it. You want to have physical levels of civility. You probably have a digital currency. You have folks saved money, and then this community is able to then crypto crowdfund territory.
And that is actually super, super interesting because basically, everything you’re doing, you’re doing online. You’re doing the dress rehearsal for the whole community online. You’re organizing online. You build up a hierarchy online. And it’s founding as opposed to election, or a founding as opposed to inheriting. So, one thing I think a lot about is-
David: Start-ups becoming countries, basically.
Balaji: Exactly. Or cities to start, you know? And the thing is, there’s nothing illegal about doing the city. There’s nothing illegal about building stuff. All of that is totally possible. There’s unincorporated land, and so you’re working within the existing system, there’s nothing in feasible about it. But it’s a level up from going from starting your own company, start your own currency, start your own city, start your own country.
So I think we’re at the start your own city stage. And obviously you learn a ton by doing that. The city eventually becomes maybe a city state. But the thinking is, it’s a new model for legitimacy where it’s founding versus inheriting. The East coast of the U.S. I think is very much about inheritance. Obviously inherited wealth, you have these inherited fortunes, but also inherited names, where it’s a Kennedy or a Bush or what have you, it’s not like any one party or group, but it is its inherited wealth, its inherited names.
It’s also inherited institutions. You inherit a paper, whether that’s the Murdoch’s or the Grahams or the Saltzbergers or what have you, they inherit papers. And this is very different from the West Coast model which is founding. SpaceX wasn’t inherited. PayPal wasn’t inherited. Facebook was not inherited. These were built from scratch. Amazon was not inherited.
And the thing about it is when you inherit something, if you’re the 45th president or the 109th mayor, you don’t know how to build that system from scratch. It’s like an air to a factory. They don’t know where everything went, they didn’t build it up from scratch. So long as it’s normal times, the factory just keeps cranking through and your manager keeps cranking through the widgets, go through. It doesn’t seem like there’s anything wrong, but something ephemeral that’s hard to quantify has been lost, which is the know-how. “I know how to do it.”
So these folks who are currently running these East coast institutions could pretty much never have built them from scratch, most of them. For example, organizing the U.S. Military or the federal reserve from scratch.
David: This is the inverse of progress as abstraction. This is the downside of that.
Balaji: Yes. Yeah.
David: So this is abstraction. Then it comes down, but then we lose the process, knowledge of how the system actually functions.
Balaji: Exactly. That’s exactly right. And it’s alienation from how that was actually done. You start to just think, okay, this will endure, because it always has, and always will and it’s hundreds of years old and blah, blah, blah, right? But I don’t think that many institutions that predated the internet will easily survive the internet. That’s another one-liner if you want it.
So these inherited institutions are incompetent and increasingly less considered less legitimate. There’s like a loss of faith in banks. There’s a loss of faith in media. There’s loss of face in politics, loss of faith in education and higher education and so on. What I think is going to be legitimate is an alternative model, which is founding as opposed to inheriting. Why is this organization set up the way it is, the hierarchy it is? Well, this person built it from scratch with their own two hands by hitting keys on a keyboard. And it accumulated crucially, the backlinks came in over time incrementally.
It was proof point, proof point, proof point. You know, this person came in as an executive. This person came as a partner. This person came in as a customer. This person came in as a well-wisher or supporter. But they happened over time, it wasn’t like, “Oh, you win an election and boom, it’s all yours.” “Oh, you inherit something and it’s all yours.” It didn’t just get dumped in your lap, it was built up progressively over time, as opposed to a big bang and then another big bang.
And I think that’s a different model of legitimacy, which is a digital model. It’s an intranet model it’s already worked certainly for starting new companies and new currencies. I think it’ll also work for starting new cities and it’s think it starts with writing.
David: Awesome. Two more questions. Why is genomics important and why do you think that it’s under-covered relative to its importance? It feels like a big secret that has just been glossed over.
Balaji: Yeah. First, the reason genomics is important is that your genome is your OS. That is your operating system. And we now can read the source code for that. You can actually look at that on the computer and you can see how your genes help determine your body, your metabolism, your makeup. And you can click those genes and you can see whether you can metabolize caffeine, whether you’re predisposed to this or that cancer, all this stuff is there, but you’re kind of have to be sort of an expert to do it, but I think it’s going to become extremely accessible. And so that’s number one, is first, for the same reason you have a Fitbit or a scale, you’re going to measure yourself and the genome is going to be a huge part of that.
Number one, number two is once more folks have genomes, you can network them together just like you can with computers. When I say have genomes, I mean, installed genomes where you’ve got on your computer, you have some form some way of doing private aggregations. Because since it’s medical data effectively, it’s private, but since it’s statistical, it needs to be aggregated to provide value. And so ways of doing private aggregations across people’s computers will be a super important thing for genomics. That’s actually where genomics and crypto intersect. Since you want to basically calculate things like how many A’s, G’s, T’s and C’s are there on different people for this particular trait while protecting their privacy.
That’s number two. Second point is, lots of genomes have a network effect, just like lots of computers, have a network effect. But number three is we started to learn how to edit genomes. The first successful CRISPR therapies are now out there where folks are actually curing things like sickle cell by editing genes, restoring functionality of fetal hemoglobin in one case to compensate for the sickle cell deficiency and others.
And live gene editing is crazy. That’s like updating the software on your computer. The fact that that works is this enormous, enormous breakthrough should be page one headlines. I mean, maybe it was for a day or something like that, but it’s simply not something that most people are aware of. And finally, genomics is a great way for software people to get into biomedicine because genomics also encompasses ‘omics’ in general, which is metabolomics and it’s transcriptomics and so on.
So it’s looking at not just your DNA, but it’s looking at the small molecules that are floating around your body. Like when you’re metabolizing glucose, it’s looking at the RNAs that are floating around your body when you’re synthesizing RNA. And it’s basically getting a readout of your whole body, just like you’d get a readout of your computer state what’s going on with the RAM, what’s going with the hard drive and the CPU and so on. You get that kind of readout a Fitbit but for everything, you know?
David: You could comment, I was talking to my mom last night and she brought up such a good point that the fact that nutrition labels are static is kind of ridiculous. That you should be able to look at a nutrition label and the nutrition label should not just be this absolute thing, it should be relative to your DNA, your health and your biology.
Balaji: Absolutely. In fact, you could do that where if you had a genomics app and you barcode scanned a food or a drug, you could see whether or not it would affect you. One thing about that is, something that might get past the regulatory issues associated with something as this is, literally just see a table of all the people with different variants and what their reactions to this drug were. That’s literally just showing raw data. You’re not saying the drug works, you’re not saying it doesn’t, it’s literally just a table of the raw data.
This goes back to the ledger of record, where there’s so much that can be done with decentralized data collection, where you can start building up systems. Because the information is upstream of decisions. It’s the information that defines what drugs are legal or not. It defines whether we consider somebody good or bad, whether how many deaths for corona happened, whether you should buy or sell something, et cetera.
So the information on your genome is actually just one of these aspects of information that will help guide your decisions. The aggregation of all those things is another entry in the ledger record, which also helps guide your decisions.
David: Cool. Last question. I think that one of the big things that’s going to become true, and this gets back into some of the things we were talking about earlier with media fund, is that writers will go on to be millionaires, billionaires. And that writing is, like we’ve spoken about the way that you attract a network, the way that you start movements, the way that you basically get free marketing, advertising and networking for basically everything that you do.
And that writing online is the internet first credential that you would get at a university network that you would get at a university benefits of going to conferences. Like now you get to run the show. Why do you think that media writing online, and we can talk about media fund here to close., why do you think that this now has venture returns rather than being something like a journalist where if you’re an average paid journalist, you might make $80,000 or $70,000 a year. Even that might even be on the high end, whereas if you’re like a very well-paid one, maybe you’re making $250,000-$300,00?
Balaji: Because you’re no longer monetizing the content, you’re monetizing the action outside. I think like this whole model that I’m interested in is open source media, which is to the existing media, what open source software was to Microsoft. So all the content is free, you monetize the community and you monetize the actions that are happening outside. The content is there to train you and to help you do things in the real world.
And so for example, a textbook that teaches you how to do investing or something like that is worth way more than the 50 bucks that you paid for it. Yeah. I think that’s the answer. I think we’re basically radically under monetizing writing because we’re focusing on making people pay for the writing, as opposed to using the writing to level up people, increasing their human capital and taking a cut of that.
David: I couldn’t agree more. That’s why Write of Passage is about writing for free and then having the benefits of writing show up in other parts of your life. So given everything that we spoke about, is there any message that you want to leave the audience with?
Balaji: Sure. This is just a test right now, but if you go and sign up at mediafund.com, we do have a landing page there where we’re looking to fund writers who are writing about technical content at first, and we would basically fund you and then also take a cut of any investments that come through your site. And then you’ve recruits that come through your site for some of our companies. So that’s the concept of how we might make media millionaires.
David: Great. And how can people reach you? You can reach me at hello at perell.com and I will forward your email to biology. What’s the best way for you?
Balaji: Sure. DM’ing me on Twitter will work for now. And I’ve got open DM, so do that.
David: Cool. Thanks Balaji, this was fantastic.
Balaji: All right. Thank you David. Okay, bye.