JOSH: Hello, and welcome to Clearer Thinking with Spencer Greenberg, the podcast about ideas that matter. I'm Josh Castle, the producer of the podcast, and I'm so glad you've joined us today. In this episode, Spencer speaks with Sam Bankman-Fried about the strengths and weaknesses of cryptocurrencies, the differences between Bitcoin and Ethereum protocols, and the prospect of using Blockchain for novel applications.

SPENCER: Sam, welcome! It's good to have you on.

SAM: Thanks for having me.

SPENCER: One topic I don't think I've ever talked about in my podcast before, amazingly, is crypto. I know that you just know a ton about this space. So I'd love to dig into it a bit with you and get some of your thoughts on it.

SAM: It started to feel a little bit like 2017 again where you're just walking down the street and see two people walking and you're like, "You know what? They're probably talking about Bitcoin." It's not quite at that level. But, boy, has that question been coming up a lot more recently, ever since Bitcoin got above $20,000.

SPENCER: Absolutely. I remember in the previous crypto bubble a few years ago, it got to the point where people I knew would be in a cab, and the cab driver would say, "Oh, do you work in finance?" They'd say, "No," and the cab driver would say, "Well, do you know what cryptocurrency I should invest in anyway?" [Bankman-Fried laughs] It really reminded me of these old school stories from Benjamin Graham talking about, "Once your barber is asking you what stock to buy, it's probably time to sell." I'll just briefly mention my history with cryptocurrencies. I didn't really invest in Bitcoin. But back in the day, when Ethereum was launching, I just saw this project, I was like, "Wow, this is some crazy project." I was just really impressed by the insane scope of it. I was like, "This is made by some mad scientist. Okay, I'll buy a few $100 in this crazy thing," which I fully expected would most likely just go to zero. Anyway, it turned out great. It went up thousands of percent. I invested a very small amount so it wasn't that big a deal. So then I found myself sitting on a non-negligible amount of Ether later on, and was thinking to myself, "Should I hold on to this or should I sell it?" What I did is, I sorted all the crypto projects of the time — this was a few years ago — by their market cap, so basically looked at what is the biggest cryptocurrency, the second biggest, the third biggest, and so on. Then I started reading through their white papers — not Bitcoin but other newer projects — and then just one after another, as I read through their white papers, I was like, "This doesn't really make sense," then I would go on to the next one, and be like, "This doesn't really make sense either." My overwhelming feeling was that a lot of them didn't really hang together if you really thought about what they were saying. So then I just sold most of my crypto. I've never really been involved much since then. Anyway, that's my history. Any reaction to that so far?

SAM: I don't disagree. There's a lot of projects in crypto. The vast majority of them, it's really hard to convincingly argue that they're in it for the tech, shall we say.

SPENCER: I think it's one of these situations where this whole space has gone up in value so much, that you get this weird combination of really smart people who are really into technology who are really excited about it, together with people who are just like, "There's a way to make a bunch of money really quickly. Let me jump on the bandwagon." And as it's gotten more and more popular, maybe that ratio has changed. In really early days in Bitcoin, I am willing to believe that it was maybe mostly people that are just interested in this cool tech or these libertarian ideals of a system that completely runs itself with no government intervention. But it really seems like now, it's just hard to avoid scams. It's hard to avoid people that are just basically way overclaiming and their technology doesn't make sense at the end of the day.

SAM: That's true of most of them. You also just run into this thing where almost all projects end up looking either it makes no sense or it's so close to just a copy paste of an existing project. It's completely unclear that the changes make any sense. So to the extent it's good tech, it's only because it just copy pasted an existing one.

SPENCER: I really want to talk about both the good stuff and the bad stuff. We're kind of hammering in a little bit. Tell me about some of the things that you're excited about in this space, and where you see stuff that you think has more potential.

SAM: I'll maybe start with why crypto in the first place. Why are we even here? What's steelmanning even the existence of crypto in the first place? There's a few ways to get it. One of which is just, if you want to launch a new currency and you're not going to back it with a nuclear arsenal, then who decides who has how much of the currency? And that's actually not an easy problem to solve. It's not one that's really ever been solved before crypto, or at least not in the last many, many years. What crypto does is it gives you an opportunity to do that if you decided it was important to launch a new currency that wasn't backed by a state. Why would you want to do that? Well, one answer is... Put yourself in the position of someone living in a country where they just don't trust their country's currency at all. And that's true of really a significant fraction of the world, that their fiat currency is basically just a piggy bank for their politicians. It's massive, massive corruption which leads to inflation or hyperinflation. If you're in one of those countries, you might very well want some other option. It's actually very hard to use another fiat currency because you don't control that. You're reliant on banks. But if your country isn't enthused about having banks storing another country's currency, you're out of luck, or at least it's not easy to figure out what to do. Crypto is one answer to that. That's just one value prop.

SPENCER: It's funny you mentioned that one in particular, because I've advised a project a little bit called 'Reserve' which you probably know about. They're specifically trying to apply crypto to helping in situations of currency hyperinflation and saying, "Can we give people something that they can hold on to that's not going to lose its value instantaneously?"

SAM: Absolutely. That's one of the clear-use cases. We actually see that. There's real, immediate demand for this. Some of the countries that have shown big crypto demand are countries where this is a big deal and where people are basically thinking, "What's the safest thing that I can put my wealth in?" Crypto has a 20% chance to crash it to zero; alright, still better than putting it in a bank, which is an interesting foreign concept, I think, to a lot of Americans.

SPENCER: That's such a good point because, if you think about it from the point of view of someone living in the US, where it's basically pure speculation... If you live in a country where the currency is kind of screwed and is losing value constantly and maybe the whole government will collapse — that's your benchmark — well now, crypto looks really, really good.

SAM: Exactly. In some of these countries, on top of all that, if you put too much money in a bank account, the government might just take it. So banks aren't necessarily safe. Banks are sometimes among the less safe places to park wealth.

SPENCER: Well, India had that huge move to eliminate the high-value bills. I don't know exactly what the effects of that are, but my understanding is, a bunch of people who had their whole life savings in these large bills had trouble converting them. You can imagine just how ridiculously upsetting that is to lose a lot of your money because of some currency change.

SAM: Absolutely. So that's one case. I'll give a couple other cases of why crypto could be cool. One of them is, let's say that you go to a store to buy some bread or you want to transfer some money out of your bank account to somewhere else. How long does it take for that money to reach its destination?

SPENCER: It's amazing how slow and antiquated the whole banking system is and how much it costs to send money around.

SAM: You're paying 5% on these things and part of that's because it takes a day for the other side to even be aware you're considering sending money. It takes two months for those dollars to clear. Two months! It can be recalled for two months. It's absolutely insane. As it turns out, when you start diving into ACH transfers and credit card payments, this is where a lot of the fraud comes from, people buying something and then seven weeks later saying, "Eh, never mind, not gonna pay for that one," and they just tell their bank, "Please send me the money back."

SPENCER: It's really interesting. This was an issue that the system was built so long ago, and then this patchwork of upgrades was made to try to improve it, but you're still building on some crappy really old thing.

SAM: That's exactly what's going on. And it works about as well as you'd think. You have these 5% charges everywhere, credit cards taking multiple percents on half of all payments in the States. Part of that sum is making money, but a lot of that is the cost everyone pays to the system. They're taking on lots of risks because of the lack of settlement of this and that gets externalized.

SPENCER: I agree with you. This does seem like one of the most promising reasons for cryptocurrency to exist. In theory, you could transfer money extremely quickly with extremely low fees. But I'm wondering to what extent has that panned out so far? And to what extent do you think it will pan out?

SAM: Well, it's a good question. I can tell you, I do it a lot. But that's because I spend all day talking to people in crypto.

SPENCER: If you need to send money to someone, you're just like, "I'll just shoot this over to you." What is the fee that you end up paying, effectively? It's really tiny, right?

SAM: It's a really good question and a topical one. It depends on what blockchain you're using. Anywhere between a 100th of a penny and $100 per transfer, depending on the blockchain.

SPENCER: Does it scale a lot with the amount you're sending or not really?

SAM: Not at all.

SPENCER. Okay, so it's just a fixed price. But the fixed price is anywhere from less than a penny to $100. That's a huge range.

SAM: It's a huge range. Some blockchains, it's always 100th of a penny; some, it's between $1 and $100. This is something the crypto industry is currently reckoning with in extremely slow motion.

SPENCER: Why is it so expensive? Is it because of the energy use of the proof of work or proof of stake systems or what's going on there?

SAM: What is going on there? There's a bunch of different blockchains and they're all independent systems, their own analogous transfer mechanisms. There's a ton of parameters but basically, only one matters, which is how many transactions per second the blockchain can support. You could maybe say, "No, it's actually data per second. Let's see how big a transaction can be." But whatever. You can think of bits per second as button clicks per second; they get similar results in general. You can imagine that if that number is 50,000 versus if that number is ten, what's going to happen?

SPENCER: Is it basically, if there aren't enough transactions they can support, they basically have to charge a high fee to reduce the demand to use a system essentially?

SAM: Exactly. Whether they're literally auctioning it off, whether there's some other bespoke process which is in real time changing this in order to get it to line up with what it needs to be, one way or another, if you only have ten transactions a second, those are going to be expensive transactions because people are gonna be bidding them up. That's a little bit of an insane thing. To give a sense of this — because how much is ten? Is that a big number or a small number? — large global applications like Facebook or the New York Stock Exchange or something like that, tend to take somewhere between ten and 500,000 transactions a second. That's like how many tweets there are per second, for instance, in the world. So if that's your prior, that answers the question, how many transactions you need. If you want this thing to get huge and the answer is ten, ten is just obviously not enough.

SPENCER: Got it. What are the limiting factors on why some of these systems just can't handle that many transactions?

SAM: This is maybe not a very respectful way of putting it. When the blockchains are created, they just type the number in. Some of them typed ten, and some of them typed 50,000. That's not the whole story but that's embarrassingly a piece of the story. But why did they type in ten? In some cases — with Bitcoin, for instance — the answer really is basically that, when it was created, Satoshi was like, "Ten transactions per second, that's huge. I'll be lucky if this thing ever grows to ten transactions a day." Turns out that he got very lucky.

SPENCER: You can't blame him too much because it literally didn't exist.

SAM: Exactly. But then you look to the newer ones and Ethereum is the one that has this big problem right now. Ethereum is about ten per second. Why is that? I honestly think if it was created today, that they would've chosen a much bigger number, but part of it is a philosophical discussion. What the constraint on this is, basically, somehow people have to be verifying these transactions. The blockchain has to be adding blocks to the chain, whatever that means. What that means is that whatever your decentralized nodes are — your various people running computer programs that are processing the blockchain and computing what's happening — have to be doing this in real time. That then gives rise to the question of how fast can computers generate these sets of transactions and communicate them all across the world. You've used computers before so I'm guessing you're aware that ten is not a very big number as far as computers are concerned. That sort of hasn't quite answered it. Then you get to the positive answer, which is that one of the big motivating philosophies of Ethereum was that anyone should be able to run a node. You shouldn't need a powerful computer for it. You should be able to do it basically on your smartphone. Your graphing calculator should be able to. Everyone in the world should be able to run a full node processing every transaction that happens on the blockchain. Now you start to put really tight constraints on the compute of this system. That's where some of the small transaction limits come from.

SPENCER: Got it. What is each of the nodes actually doing? Could you walk us through that a little bit?

SAM: It's a little bit different in different blockchains. There's two types. In one of them, there's proof of work versus proof of stake. The general answer here is that users of the blockchain are submitting requested transactions to these nodes. They're basically saying, "I would like to please send this transaction. Please send seven tokens from my address to this other address." Then the blockchain nodes have to process that and basically take all the transactions that everyone has attempted to submit, tack that on to the whole history of the blockchain and compute the new state, and then broadcast that to everyone else. People might be submitting transactions to different nodes; you have to coordinate on this. You'll each be broadcasting to each other, which transactions they've gotten recently. That's sort of the core process here.

SPENCER: Then once a certain number of the nodes — a certain percentage of them — all agree, then everyone's like, "That transaction has occurred."

SAM: That's right.

SPENCER: Got it. Just for those who are less familiar with cryptocurrency — because everyone hears about it all the time now but people may not fully understand how they work — I just wanted to break down some of the history of this. Bitcoin, the way that I think about it is you essentially get a really big number that you can prove you own. If people are willing to pay you for that really big number, you can prove that you transfer it to them so that now they own it. Is that a fair summary?

SAM: That's basically right. You have the password to your wallet and if you want to send coins out of your wallet, you need to send a hash of your password so that you can prove that it's actually you without actually giving away your secret key. That's the core process that's happening when you're submitting a transaction to the blockchain.

SPENCER: Then whenever you go transfer one of these big numbers to someone else, you can prove that you now have transferred to them and they own it because essentially, all of these nodes are recording this transaction saying, "This person transferred to this person," so anyone who wants to know who owns it now can go look at that ledger and say, "Ah, I see who owns it now."

SAM: That's exactly right. And the cool thing is that you don't theoretically need to rely on these other nodes. Theoretically, what you can do is to say, "I have downloaded the whole history of what's happened. If those nodes go dark, I can just start broadcasting everything. I can broadcast all the proofs of all the transactions that have been sent."

SPENCER: Can you just elaborate on where mining comes into this? Why is that necessary?

SAM: A question that you get to is, who decides which transactions are added to the blockchain, each block? You could say it's all the ones that are submitted, but then you're like, "How do you decide which transactions were submitted?" You can prove a transaction was submitted by giving the hash of it but it's hard to prove that one wasn't. To get to these awkward questions like, "Who is collecting all these? Who has the ability to censor it? Who is the person who can just decline to add some block to the blockchain?"

SPENCER: Right. Because if you run a node, and you're just like, "I'm gonna ignore that transaction. I'm going to pretend a different transaction occurred." Then now you have everyone saying different things. How do you know who to trust? Which nodes do you believe, basically?

SAM: Exactly. So what Ethereum and Bitcoin did — which is called proof of work — is that there's a random number-generating contest. Every ten minutes or so, in Bitcoin's case. Every six seconds or so, in Ethereum's case. Whoever wins that big number-generating contest, whoever gets lucky, and finds the right big number, gets the right to produce the next block. If you want your transaction to be included there, you can effectively bribe the next block producer to include it and say, "I will pay you $1 if you include my transaction in the block." Presumably they're gonna be like, "Well, I don't care. No skin off my back. Sure, I'll take your dollar."

SPENCER: Could you elaborate on that payment part? I don't quite get that.

SAM: As part of your transaction, as part of this signature that you're uploading, a thing that you can say is, "In addition to everything else, I would like to please transfer $1 from my address to whoever mines this block."

SPENCER: And that's to encourage them to do it faster, is that the idea?

SAM: That's right. Otherwise, why would they bother including your transaction? What's the incentive for them?

SPENCER: I thought they got minted Bitcoin automatically.

SAM: They do do that, but they do that whether or not they include any actual transactions in the block. They could just publish a block that's empty. So then how do you incentivize them to add your particular transfer? The answer is: you can pay them for it.

SPENCER: Okay, got it. That's the incentive to actually include transactions on the blockchain. That makes sense. So far, what we've built up is: we have this system of proving who owns different numbers, of transferring numbers to other people in a way that you can trust; even though there's no one single trusted party, all of the parties working together create this crowdsource trust, if you will. Why is it a currency at all? Well, if someone is willing to pay you for something, then it's worth something. By definition, if someone believes that that number representing a unit of Bitcoin is worth something, then that's what it's worth because you can sell it to people for that. That's the currency aspect. Do you want to add anything to that? Or should we go into Ethereum?

SAM: Nope, that sounds about right to me.

SPENCER: Okay, cool. So then, Ethereum comes into play. Do you want to give us some background? What was the theory in doing that was different?

SAM: Ethereum's super cool and completely revolutionized the blockchain industry. What it did was: you have Bitcoin which is the first blockchain and what Bitcoin does is, it's a blockchain that lets you send Bitcoins to each other. It's exactly what you think of. What Vitalik and the Ethereum creators realized was, this protocol that we're writing for transferring Bitcoins is very close to Turing complete. What if we just make it Turing complete?

SPENCER: Just to unpack that, the Turing complete is the idea that... A system is Turing complete if it can compute anything that can be computed. Or in other words, if you can do anything a computer can do, right?

SAM: Exactly. Any normal programming language is Turing complete. All languages are in the end, equivalent, modulo speed, and syntax, etc. So I realized, whoa, what if we just actually make this fully Turing complete? Then, in theory, what transactions could you be producing? Well, you wouldn't be restricted just to producing something like, "Please send one token from my address to this other particular one." You would be able to send arbitrary computer instructions.

SPENCER: And these instructions would actually be running on the blockchain. They'd be running on these nodes, these other people's computers that are being paid to do this.

SAM: That's right. So when you include a transaction in your block as a miner, you just do whatever it says. In Bitcoin's case, that means you just do whatever transfer it says. But in Ethereum's case, that means you run whatever program it's telling you to run. And wherever that goes, that goes.

SPENCER: It's this huge swarm of computers willing to run whatever code you send them essentially, as long as you pay the transaction fee.

SAM: Exactly. Now what you have is not just a currency, but a global decentralized computer, where anyone can write and upload programs to it, and then they just autonomously run automatically forever.

SPENCER: People might think, hearing that, that the cool thing about it is, you get these servers that you can use to do computation, like Amazon Web Services. I think that's misleading. What's actually cool about it, has more to do with the fact that this service that's permanent and anyone can expect, will always do the thing that you claim it will do because they can just look at it and verify. Is that right or am I mischaracterizing that?

SAM: That's basically right. If you go to google.com and try to search for something, you're pretty sure that if you type in 'printer,' you're gonna get things related to printers. But there's no proof of that. Google could just change their domain one day, that if you go to google.com, and type in 'printer,' it just flashes green. That's what it does. It'd be a little weird, but they could do it. There's no guarantee that it's going to do what you think it will. If you upload a program onto a blockchain, one of the really cool things that happens is that anyone can just look at the transaction you submitted to the blockchain uploading that code. They know exactly what that code will do and will always do. It's not up to you anymore (the person who uploaded it), what happens when that code is run. Because the code's just there and the nodes are just progressing it forward. So you can just prove to people that your program will do exactly what you say it does.

SPENCER: So it's not the computation in the cloud that's interesting; there are lots of services that will do that. It's the verifiability aspect.

SAM: Yes.

[promo]

SPENCER: Let's just go through a few examples of what you could do with this. There's the smart contract where you and I could maybe make an agreement which says, "I'm gonna wire Bitcoin into this system. You're gonna wire Bitcoin into this system. Then if the temperature in Texas is greater than a certain temperature on a certain day, one of us will get paid out. If it wasn't, then the other would get paid," or something like that. Is that an example?

SAM: Mostly, although there's one catch there, which is that the blockchain doesn't know about the temperature.

SPENCER: So you need an oracle that it can tap into? Do you wanna explain that?

SAM: Let's say that you want to do that. You wanted to have something where we both sign a transaction, saying that I'm going to owe you a number of Bitcoins equal to the temperature in Texas in a week. Then the nodes go to propagate this transaction in a week. They're like, "We're gonna run this code." The problem is that the temperature in Texas is not a built-in variable to the blockchain that your code can reference. The blockchain doesn't know what the temperature is. So when these nodes go to run that code, they're like, "This is written in English instead of a programming language. I don't know what to do with that."

SPENCER: Couldn't we agree, though, that we're going to trust some website where it's going to parse it from there?

SAM: Yes. You can't quite trust a website because in fact, it doesn't even know about websites. The blockchain knows about nothing but the blockchain itself. What you could do is a two-step process where, first of all, we have someone who just uploads, chooses some address on the blockchain, whatever some website says. Then you can have the smart contract say, "Look at this address on the blockchain; that's how many Bitcoins I'll load up."

SPENCER: I see. As long as we trust whatever system is doing the uploading to the blockchain, and we believe that it really will be the temperature in Texas (according to a reliable source), now we can actually have our smart contract automatically pay one of us based on the temperature.

SAM: That's right.

SPENCER: What are some other cool things you can do on the blockchain that are hard to do another way?

SAM: The first thing that I would say is: anything you can do on the blockchain, you can do in other places if you have trust. You can do it on AWS if you trust that it's gonna do exactly what you think it will. But what are things you can do on the blockchain trustlessly? Well, you can run an exchange. Typically, when you run a stock exchange or cryptocurrency exchange (whatever it is), what you're doing is you're just running some centralized process that collects users' funds and transfers them between users when they click buttons. There's the worry: what if you just run away with all your users' money? Or what if you print random trades that make no sense that are not what people asked for? What stops you from doing that? The answer is your reputation, the law and stuff. But if you code the exchange into the blockchain, then the answer is, the only thing that ever could have happened was exactly what should happen. You can't lie to the blockchain, so to speak.

SPENCER: As long as we read the code and can say, "The code really does that," then it will do that. Nobody can change it at that point.

SAM: Yes. That's one example there. Another one is, think about Twitter. Everything that Twitter or Facebook are, are things that you could run on a blockchain in a trustless way.

SPENCER: What would the advantage of that be? Let's say someone builds a Twitter on the blockchain.

SAM: Why bother? It's a good question. First of all — say what you will about this — but there's been a lot of controversy about whether Twitter and Facebook have the right to censor their users, to what extent they do. What if it's a politician? Do they have a mandate to do that? Are they even allowed to? Whatever you think about that, you're at their mercy. A lot of people are not currently very happy about that fact. If you add it on a chain, it would be theoretically uncensorable. What you can do then — and this is the classic trick to deal with censorship on decentralized applications — is you split apart the protocol from the GUI. What you can do is: this protocol, you can upload whatever you want to it, any transaction. There's no censorship in it. But anyone who's making a website that displays your Twitter thing, your decentralized Twitter, can choose whether their GUI that they're hosting, their website displays various transactions on it.

SPENCER: You can have one back end, and lots and lots of Twitter alternative clients that are all reading from this back end and processing different ways. They could make different decisions about how to sort people's comments or who to put at the top or what to censor or whatever.

SAM: Exactly. It gives power to the users, it gives power to diversity, and lets you have more than one answer to this question, depending on whatever people want for it. I think that's super cool. I think it serves the best answer that I've heard or seen for how to deal with censorship on social media.

SPENCER: That's really interesting. One critique that is sometimes given towards cryptocurrency stuff in general, is that it's a very bad database. Imagine a database that's really slow, and you can only append to it, you can't go back and delete parts. How sympathetic are you to this critique of it?

SAM: I'm moderately sympathetic. Maybe just to spell that out a little bit more, what are the ways in which it's shitty? First of all, latency. A blockchain can basically never be faster than 100 milliseconds. Why is that? Well, it's geographically decentralized nodes. That's how long it takes light to travel around the world.

SPENCER: And that's the best case scenario. I'm sure that's very optimistic.

SAM: Exactly. You're never gonna have programs running with less than 100 milliseconds latency. Does that matter? Well, it depends on what you're doing. Think about Twitter. How much does 100 milliseconds of latency matter for Twitter?

SPENCER: Probably not much.

SAM: Not much. No one cares. The website has more lag than that. That's totally fine. Now think about HFT firms trying to trade with each other. How much does 100 milliseconds matter there?

SPENCER: You have these firms that will build a computer right next to the stock exchange to cut a few milliseconds off.

SAM: Exactly, to save a microsecond. It's fatal for some applications and basically irrelevant for others. That's one constraint and it just means that, conveniently, 100 milliseconds is about human reaction time. If something is only constrained based on needing to be fast enough for humans to use it in a way that flows, it's basically okay, at least theoretically, if you can hit this theoretical limit.

SPENCER: Just to react to that for a moment, I really don't know why this is, but a number of times when I try to use GUIs (graphical user interfaces) that are hooked into some blockchain thing, it seems like they're really laggy, or they work really slowly or don't work well.

SAM: Yeah, it really does. Let's go back to this pesky little 'theoretically' thing. I said theoretically, it can be 100 milliseconds. What is it in practice? Well, what blockchain are you using? This is also just a property of a hard coded number in the blockchain, basically, and as it turns out, for Bitcoin, that number is minutes. So it will lag submitted transactions by minutes.

SPENCER: Is it because you have to wait for another node to confirm the transaction?

SAM: It's not even that. Yes, you do have to do that. But you could be optimistic. You could see one node, confirm it and be like, "Alright, 99.9% chance this is going to be confirmed. I'm just gonna go ahead and display the answer so they know it's coming." The problem is that blocks only appear every ten minutes on Bitcoins. So no node is going to know about it for ten minutes, or five minutes on average. That's not great. That's like putting a little bit of a dent in this theoretical throughput. Now, how about Ethereum? Ethereum is, I want to say about ten seconds or so — I forget the exact number — it's faster, but it's never going to be faster than waiting ten seconds to see the results of what you do.

SPENCER: Waiting ten seconds feels kind of shitty.

SAM: It does feel shitty. And that's the best case scenario. But remember, ten transactions per block — not very many — the demand is for 1000 times that right now. There's auctions for each block. If you submit a transaction, if you're not paying through the roof, it's probably not going to make it into the next block. You're probably going to have to wait till a low demand block to get it in if you're not paying $200. So in practice, you might have to wait five minutes for your transaction to make it into a block.

SPENCER: Interesting. Now if you're sending a bunch of money across the world, it's a lot better than the alternatives. But if you're trying to do something more real time, that's a huge barrier.

SAM: Exactly. It's one of these things where, for 10% of use cases, that's okay; for 90%, it's prohibitive. We've made some progress versus the alternative, which is way worse than that, but it's still not great. That's a thing you run into but there are faster blockchains. There are blockchains which are consistently getting every submitted transaction in every second, or more quickly right now. And that number is getting faster. So this really is a property of the blockchain you're using, and obviously how fast the GUI you're using is, as well.

SPENCER: What are some other limitations of blockchains that serve as a system?

SAM: Other limitations of this are that, if there's only one node, it's not really doing a whole lot of good in terms of decentralization. If there's just one guy in Louisiana who's doing all the blocks, and it's just him and no one's verifying it, you don't really have much trust there. This only really works if you have (let's say) hundreds of nodes processing the entire blockchain. Note at that point — this gets to your shitty database point — it's not the most efficient system in the world; you're probably doing 100 times as much work as you need to be.

SPENCER: You'd have to have 100 databases when you only really need one, in some sense.

SAM: That's right. How bad is that? Well, I don't know. How big of a cost was each database? Was that the limiting factor, database cost? If so, that's not going to be great for your project in crypto. On the other hand, if you had three orders of magnitude to spare on that, then maybe it's totally fine.

SPENCER: Well, word is, you have to write huge files, hundreds of megabytes. How is that going to work?

SAM: Another good point. Yup, at some point files just get too big. As it turns out, storing a JPEG is not too bad; storing a movie is impossible on a blockchain, on the fastest blockchains, that is. So what can you do there? Well, there are things you can do. They're kinda shitty. Basically, you need to have some side product, which is blockchain-like, but optimized for big data and storing it and whatever you can do with it, it's not a great answer.

SPENCER: I mentioned this briefly, but there's this issue that these are append-only databases. Part of the fact that it's trusted means that you can't allow people to delete stuff because you want to be able to see the entire history of what happened. To what extent does that actually create limitations in practice?

SAM: Not a lot. You can build whatever you want. You can't censor the past, but you don't have to display the past. And you don't have to reference it.

SPENCER: So you could just say, "Okay, now, everyone, this thing is deleted. Just remember that." You're just appending. But doesn't that create this increasingly long history that you have to process in order to understand what's happened? Or is that not really an issue?

SAM: Yep, it absolutely does. That is a constraint; it's not the biggest constraint because that turns out to just not be super expensive, all things considered. But it can be awkward to process; it's not free. Given what blockchains are doing, that tends to not be the limiting factor, because the computers have to remember it, but hard disks are cheap. That's not really a limiting factor here. So you can just have all the nodes storing it in their hard drive somewhere, and obviously agreeing on what it is; they don't have to rewrite down the entire history on each block. They just have to write down the newest block, and then everyone remembers separately what the previous blocks are, and can confirm that that is consistent with it.

SPENCER: I guess one way of looking at this — this is how I think about it, curious to get your reaction — is that in many ways, these are very bad databases. But they have this really amazing and interesting property of: you don't have to trust anyone in the system. The whole system, through this crowdsource mechanism, you can prove that it will do the thing that it claims even if you don't trust any of the individual actors. To me, a big problem that I see is that we take this really cool technology, and we forget the fact that there's only one really good thing about it and, in a lot of other ways, it's not very good. Is that a fair characterization or am I being unfair?

SAM: I think it's sort of fair. I think there's maybe one other really cool thing about it.

SPENCER: Great. Let's go into that.

SAM: It's composability. What does that mean? Let's say that you want to build a consumer stock trading app. You're like, "Great, there's lots of examples out there. Mine's gonna be different. It's going to be magenta; that's what makes mine cool." What's step one, in general? Step one is, "I have to go build an exchange and a broker dealer and liquidity and the whole financial ecosystem. Or license it." I don't know if you've tried licensing complex financial software from 16 different vendors. That's not an easy thing to do. The fact that Robinhood has solved this problem doesn't mean that you have solved this problem.

SPENCER: The way that normal software works is everyone has their own version. It's really hard. And there's huge transaction costs to using someone else's, unless that's their whole business. There are some systems where the whole business is, 'we're gonna offer an API,' but for everyone else, it's all proprietary.

SAM: Exactly. You can look at Plaid as an example of that. It's a $40-billion company whose sole goal is to build an API to make it less shitty for applications to interface with some subset of US bank accounts, in a way that is not native but is less bad. Eighty percent of the cost of starting a business is you have to reinvent the wheel in 17 different ways. Now, let's say you want to build DeFi Robinhood. You want to build your magenta trading app, and it's going to be a crypto-based trading app — everything's gonna look magenta — you've got to figure out how to make things magenta; that's on you. But DEXs are built into the blockchain. They're open. You can just build a GUI on that DEX.

SPENCER: These are things that other creators have made, added to the blockchain. Then because by nature, all this code is just out there — anyone can inspect it — you can actually just integrate directly into your system.

SAM: Exactly. And you don't have all these 'but it sucks in 38,000 ways' when you interface with a counterparty because they're constantly going to be shutting you off for reasons you don't understand. All the 'gotchas' that kill this, it's literally just exactly what it says; there are no 'gotchas.'

SPENCER: That's great. Okay, so we have these two really, genuinely cool properties of these systems. One is this trust thing. The second is composability which, I think, is a really great point. Then we have a bunch of downsides. Like you have to have 100 of these databases running simultaneously. There's this high latency, and maybe you can't use it with really large amounts of data sometimes. I guess the way I think about this when I'm looking at a crypto project is basically saying, does this make sense on crypto? In other words, is it actually leveraging the strengths of crypto and does it suffer from the weaknesses of crypto, as opposed to, why aren't you just doing this as a normal startup and just building normal software?

SAM: 100% agree, and a lot of crypto projects fail that test. A lot of them, you're like, "I don't understand why you're putting it down on the blockchain. No one has ever asked for that to be on the blockchain. It makes no sense. It's gonna be a terrible product because of the downsides of blockchain. And it doesn't really get to utilize any of the cool properties."

SPENCER: The cynical way of looking at this is they're just using it as a funding model because they think they can sell this token and get a lot more funding than if they actually did it as a normal startup.

SAM: How dare you imply that? [laughs] I mean, that's exactly right. One thing the world has learned is that selling tokens is a great thing to do if you're a scammer. Frankly speaking, it's the easiest way to raise five million dollars right now, without having a real business.

SPENCER: It's sad because it can start to overshadow some of the coolest projects that are actually trying to do something real, just the fact that there's so many of these (essentially) scams out there. To be fair, I think some of them are genuinely scams where the creator knows it's a scam, but I think probably far more than that, it's in this weird gray area where it doesn't really make sense, but maybe the creators of it can rationalize it. What do you think about that?

SAM: Completely 100% agree. You talk to these people, they're fucking sincere. They believe in their project. And if you start poking holes in it, they will either fight back tooth and nail, or they will sound extremely sad, as they're sort of like, "Are they right?" You actually pretty rarely get them saying, "Okay, fine, whatever. We all know it's stupid. Can we just continue the conversation about the token?" You actually very rarely get that.

SPENCER: Of course, to anyone listening who has a crypto project, we're not talking about your project. We're talking about most of the other projects. But yours, I'm sure, is totally exempt from this.

SAM: That's right. Everything but ours are pure. But we always think of what we build as, "Let's build something great." Then if a token makes sense for it, let's introduce a token into it. But step one was, "Let's build something."

SPENCER: When I'm thinking about this, I think it is really important to use an analogy here, which is when the internet bubble occurred, back in the late 90s, a bunch of the companies really made no sense. It was just people really excited about the internet. A bunch of it was total hogwash. These companies were getting these huge valuations that weren't justifiable, and then the whole thing came crashing down. But the internet was obviously an incredible innovation. It was a staggeringly huge innovation.

SAM: That's right. In fact, if you just blindly bought all internet stocks, you'd be fantastically wealthy. Yes, 95% of them would have been obviously ridiculous and gone to zero. But also, you would have invested in Amazon and Google and that would have, I think, made up for all the other ones combined.

SPENCER: That's really interesting. I guess my feeling about crypto is a similar thing: we're in this really crazy bubble where a lot of it makes no sense. Eventually a bunch of stuff will die off, and we will be left with some truly valuable things. Personally, I don't think it's as big a revolution as the first internet was. But it's still very sizable. I imagine there will be a bunch of really, really valuable and important projects coming out. Is that what you're feeling, too?

SAM: Totally agree. It's a little sacrilegious to say... I think in crypto, you're supposed to say it's as big as the internet. It's not. That's not to malign crypto. I think it's just hard to put into words how important the internet is.

SPENCER: I mean, it's ridiculously important. It's like one of the top inventions of human history, right?

SAM: You have to go back to tools as the next thing to compare, and the amount that it will revolutionize society, maybe the industrial revolution. But it has completely changed the world and changed almost everything about the world. Think about us right now: we're talking over the internet thousands of miles away, about a new industry which is built on the internet.

SPENCER: Absolutely. It's the base layer of so much of what we do today. Increasingly so. It's also just insane, the extent to which the internet is still changing things after so many years; you still have companies changing their business models. It's really amazing.

SAM: Completely agree. Part of what crypto is — one slice of this — is, let's combine the internet with finance. Instead of taking a non-internet based financial system, and trying to put internet GUIs on top of it — which is always going to have the problem that the base thing is not internet-native so your GUIs can't be faster than the speed of the underlying systems — let's just start from scratch and build finance natively online. I think that's one piece of what crypto is doing.

[promo]

SPENCER: Let's talk about some of the actual applications that are happening today. It seems clear that the number one application is speculation. Do you agree with that?

SAM: [laughs] Yeah, hard to argue.

SPENCER: Basically, it's people buying crypto because they hope it goes up in value, or I should say up in price. That seems clearly the number one reason most people get involved in this stuff. Then we have some other things, there's some applications like gambling, which is pretty easy to do when you're building on the blockchain. I think there are literally gambling protocols where you can put money in and it's randomized, like roulette. You have things like collectibles; non-fungible tokens are all the rage these days. Let's talk about that briefly. You want to explain what a non-fungible token is?

SAM: Basically, it's, 'what if you put collectibles on the blockchain?' What that means is, "Here's a token. This token represents Lebron James." You're like, "What do you mean this represents LeBron James?" "This is the one and only LeBron James token." That's what an NFT is.

SPENCER: It gets into this weird philosophical thing. What does that even mean? You've seen this, for example, for Pepe the frog. There are Pepe the frogs on the blockchain. Then you're like, "This is the unique copy of this Pepe the frog." You're like, "Well, what are you talking about? I can download the image of it. There's a million copies of the image all over the internet. What do you mean this is the unique one?"

SAM: With collectibles, there's always a little bit of that, like, "No, no, this is the physical jersey that LeBron wore, literally this cotton." I don't know how meaningful that is but it's something.

SPENCER: Right, a copy of the jersey that was made in a factory doesn't have the same feel to a human mind as the actual jersey he wore, even if they've both been washed and you couldn't actually detect that he'd ever worn it.

SAM: Right. You then look at some of the NFTs which are literally just a JPEG tied to an Ethereum address. You're like, "Well, what if you copy paste the JPEG and upload it to a new Ethereum address? What's the real one?" I don't really have anything to say that's going to make that sound better.

SPENCER: [laughs] Let's go disturbing. It's not the copyright. Because you might say... At first, when people were telling me about this, I was like, "Oh, you mean it's a way of transferring copyrights?" That makes sense. But that's not what it is, right?

SAM: Well, it's not necessarily. You could do it that way — but you don't have to — to make it an NFT. What really frustrates me about NFTs right now is, you can have NFTs that are meaningful. One example is something tied to a physical object, an actual physical object redeemable for it. Another is (let's say) skin in a video game where, whoever owns this token literally gets to use the skin in the video game. If you try to copy paste the token, the video game's not going to respect your skin token.

SPENCER: I like that because the video game could actually verify that you have the right to it because it can look it up on the blockchain. That's really cool. That actually makes sense. But the physical object, how does that work though? Because you have this number representing the fact that you own the physical object. I guess if you go to a court of law, you could prove it to the judge. Is that the idea?

SAM: Yeah. Maybe even the warehouse housing it will mail it to you if you want, if you sign something from your address.

SPENCER: I see. Because they can look it up, "Oh, yeah, this person really does own this. They can prove it to me using their private key," or whatever. I think that some NFT was sold for 69 million? Does that sound right?

SAM: Yep, that's correct.

SPENCER: [laughs] Explain this. What's going on here? This was a bunch of art that was made and...?

SAM: Oh, boy. You might be hoping that I'm gonna give you something really satisfying. If so, you're gonna be disappointed there. I can send you the NFT; it's a lot of pictures combined into one picture.

SPENCER: Anyone can just look at it, right? It doesn't prevent... [interrupted]

SAM: I'm looking at it right now. That's right. I can 'save image as' if I want.

SPENCER: What can the owner of this do that you can't do?

SAM: I don't know. That's a good question.

SPENCER: I see. [Friedman-Bank laughs] Do they own the copyright on it? Could they say you can't make a production or you can't make a poster unless I give you the permission?

SAM: I will check on whether they can do that. Honestly, I would guess they do not. That's not generally the case with NFTs. They could have been done that way.

SPENCER: It could be done that way, but it's actually not even standard. Because that would make a lot more sense if it's like, "I own the copyright to this. I'm the only one that can make prints of it," fine. But this is something more. It's almost a philosophical claim. It's like crypto magic dust that says, "I'm the one that owns this and not you, because I have this number and you don't, even though you can do everything I can do with the image."

SAM: I completely agree. [laughs]

SPENCER: Often, with human stuff, there's this layer of real things, like does this person have food to eat? Do they have shelter? Are they able to cope with their environment? Then there's the magic layer. For example, in law, if you sign a contract, it's like magic just occurred, and now you have to do the thing. Then we try to make that magic real by having courts. We try to bridge the magic layer with the physical layer. Another good example is a promise. A lot of people feel, if you made a promise, you shouldn't break it. There's something magical about having said the words, "I promise," that didn't exist before you said it. We try to make promises tighter with things like courts, the certain type of promise you can actually get enforcement. It feels to me like NFTs are almost pure magic; they're just magic completely devoid of any reality.

SAM: Maybe what I would say is, NFTs theoretically could be much more than that. But that doesn't mean they are. It could have been the case that most high-profile NFTs sold clearly had lots of things tethering them to the world. But that's not the case in many of the actual cases so far.

SPENCER: I really do like the idea that you could have a liquid copyright and things like that. It makes sense. You can imagine the person developing the technology. They had some cool ideas for it and then I don't know what exactly happened or what went wrong with it.

SAM: I think I do know what happened. I don't know if it's a great look, but it's money. Halfway through that process, you build your NFT exchange and like, "Shit, wait! I haven't gotten the cool NFTs yet." And you're like, "Wouldn't it be great if we get a partnership with League of Legends for their skins?" or all these things. And they're like, "Oh, that's hard. That's not going to happen tomorrow. I can email them; they're just not gonna respond. But I have this NFT exchange, and all my users are asking if they can buy NFTs on it." Then there's this pause and you wonder. Then they break out Microsoft Paint and they make some JPEGs. They upload them and they sell them.

SPENCER: It's really a self-fulfilling prophecy with this stuff, where the more people believe that this NFT is worth $10,000, the more that that becomes true because it's really an intersubjective truth. The value of something is the value that people assign to it. So the steelman argument is, "It really is worth that because people are willing to pay for it." And the skeptic argument is, "Are you fucking kidding me? This makes no goddamn sense. You're paying for nothing. This can't possibly last because there's absolutely no actual value creation in here."

SAM: I think one thing is, you create one NFT and it has a high valuation; you create 10,000 and they all do. At some point, you're just like, "Wait. That can't be. They can't all be." The thing that everyone is agreeing... It can't be the case that anyone can make a new NFT and then everyone that creates that has a lot of value, too. Something's wrong there.

SPENCER: Because it's not generating value. Imagine that the thing you're doing is helping actually facilitate faster transactions or something like this.

SAM: It's difficult to do and so it can't happen much.

SPENCER: Right, there's some kind of limit on it. But I guess the point I'm making is, if you build a system that's actually producing value in the world — people are better off because of using your system than they would have been — then you're creating this positive-sum interaction. There's more value being created. Whereas in NFTs, the value is all just people's belief, and therefore there's a limit to it. And the more NFTs you create, essentially, that balloon is just gonna pop and you're gonna get a massive game of musical chairs of whoever's left holding that NFT is screwed, basically.

SAM: You have this other nasty property of them. One fundamental property of non-fungible tokens — that's what NFTs stand for — is that they're non-fungible. What that means is, it's not like there's ten million of an NFT forming a liquid market. There's one or maybe 20. Basically, by definition, there's no liquidity. By definition, you're not going to have lots of buyers and sellers at the same time of the same NFT and so you're not just relying on market prices being higher rather than lower. You're relying on someone being like, "That NFT, I'm going to pay a lot of money for that specific NFT right now," in order to be able to sell it for a lot of money.

SPENCER: So this person who bought this NFT for $69 million, it's not even clear that, in the next week, they could sell it for half of that. Maybe they could, but who knows?

SAM: Right. The concept of market price is only somewhat meaningful if there's no liquid market. They're specially designed to crash in some ways.

SPENCER: It's really interesting. Another question about this that I don't really get: when people have these collectibles, you mentioned this idea of... Couldn't someone else upload another copy of it and then just claim that it's the real one? Is there anything that prevents that?

SAM: Eh, no. [laughs]

SPENCER: You could imagine a system that at least tries to detect if someone had uploaded the same JPEG or done some clever thing to see if there's a really similar JPEG on the system but you're saying it's not even... Someone could just upload another one and just sell that one.

SAM: That's right.

SPENCER: At least with Bitcoin, you can say, "I'm the only one who owns that Bitcoin. I transferred it to you." If you could just upload another copy, it feels like it defeats the system in some way.

SAM: Of course, you can mint a new cryptocurrency and be like, "How about Bitcoin 3? Anyone care about my Bitcoin 3s?" There's a little bit of the ICO craze; it sort of looked like that. Instead of minting a new NFT, your NFT was your new blockchain or your new fork of an existing blockchain.

SPENCER: Let's talk about some other applications of crypto. Another one seems to be doing stuff on the black market, essentially doing stuff that governments are trying to prevent you from doing because you can do it in a way that they can't see you. Famously, some of these systems — like Silk Road on the Dark Web — I think they would only let you transact in cryptocurrencies.

SAM: I'm not too sure. But I think so.

SPENCER: But in at least some of these systems, you couldn't even make transactions. You can go on to these systems and try to buy drugs or try to buy weapons. Famously, I think one of the founders of one of the systems tried to actually hire a hitman on his own system and got caught by the FBI. But I'm wondering, this clearly is an application of cryptocurrencies because the government doesn't want you to do some things; with crypto, you can do it. It's not fully 100% anonymous but, if you're clever about it, you can be quite anonymous. There's no name inherently linked to what you're doing. This certainly seems to be something that's happening. Do you have a sense of how big a market it is?

SAM: It's less big than you might think, with some exceptions. It's not most of what's going on in crypto; it's a really tiny fraction of it. One of the things you run into, even if you can wiggle your Bitcoin around, somehow you can't get a start with dollars and end with dollars. In practice, those are the gateways that have been enforcing anti-money laundering. Any reputable crypto exchange that has a fiat gateway is going to be running real stringent KYC on people trying to send tens of millions of dollars in and out.

SPENCER: That's 'Know Your Customer,' right?

SAM: That's right.

SPENCER: They do due diligence to make sure you're really the person you claim and you're not on any wanted list.

SAM: Exactly. It's that thing where you upload a selfie holding up your ID card and today's date.

SPENCER: So if someone wants to just go on the Dark Web and try to buy some drugs, nobody's gonna be able to stop them (probably) from doing that using crypto. If they're careful about it, maybe they'll get away with it. Maybe there are some ways to trace it if agencies are sufficiently clever. But if someone wants to move (say) $100 million, it's presumably not going to start in crypto. So they're going to somehow have to turn $100 million into crypto. Then on the other hand, if they want to do much of anything with it, other than buy some things on the internet, they're gonna have to turn it back into some kind of dollars or some currency on the other side. So that's really where the huge barrier to using it in these bad ways is. Is that right?

SAM: That's exactly right. The regulatory enforcement has focused on these nexuses on places that will let you do massive fiat transactions for crypto with no KYC. Those get shut down by governments fairly quickly. There's one or two in Eastern Europe, which were fairly prominent for a little while, and no longer exist. That's the biggest check on this system. In general, it makes it quite difficult for criminals to launder large amounts of money through this in ways that are that much easier than other means they have. If you can convince a bank and the global banking system and the global criminal databases that your idea is totally chill to be sending tens of millions of dollars in and out, then you're going to be able to do it in and out of crypto as well. But you first have to get a bank account if you're comfortable with it. You have to pass standards checks. So at some point, in order to be able to do this for large sites with crypto, you have to be able to do it for a large size without crypto.

SPENCER: Interesting. I imagine for person-to-person transactions, you can probably evade the law easily.

SAM: That's right. Although note that the other person is just gonna end up with crypto, and if they're a known criminal, they still have these dollars somehow at the end of this.

SPENCER: But as there are more and more things you can do with crypto directly, more and more purchases you can make, this becomes maybe a bigger issue.

SAM: Exactly, as you said. At that point, AML KYC is gonna have to focus on any purchases with crypto that are large. If someone tried to buy an apartment building with crypto, that's gonna have to go through these same procedures. Otherwise, they could use that as a way to get out.

SPENCER: That makes sense. We've been a bit hard on crypto today. [laughs] I wanna talk about some of the cool stuff. What's some of the cool, exciting stuff that you think is real and interesting?

SAM: I legitimately think that building exchanges and social networks on-chain could be really fucking cool.

SPENCER: Because it's anti-censorship at the core, but then you can have censorship layers that different people can experiment with. Is that the idea?

SAM: That's one thing. And I also think the composability is just really powerful. You can potentially get this quadratic or even exponential explosion of innovation, because everyone can just keep piling on top of each other's work rather than everyone having to reinvent the wheel and just go linearly from start to finish for their entire product line.

SPENCER: I do like the idea that there's this core social media layer. You can post on it. Then there's 100 different apps that you can consume it from, where some of them are focused on your closest friends and some are focused on following the celebrities and however you want to consume it. That is really nice and it's not locked in.

SAM: If you think about it, Facebook and Twitter are the same thing, just with different skins. They both have ways to send public messages and private messages to groups of people. That's what they are.

SPENCER: They put artificial restrictions on it. Twitter artificially limits the length because they choose to do that.

SAM: They display them in different ways but they're using basically the same protocol. You build the core protocol, then you can have lots of different flavors of apps all drawing from it, not having to build their own messaging system, and they're interoperable. If you send a message in one, you can access it across all of your social media apps.

SPENCER: That's cool. I like that application. What are some other ones that you think are really interesting?

SAM: Finance has gotten a lot of the attention here, but the financial ecosystem's super awkward. There's an opportunity to rebuild 30% of it on-chain. You're not gonna get all of it on-chain; some things are too resource-intensive but a lot of the pieces of this could be. You could just disintermediate a lot of giant costs in companies and delays that are there to deal with the other nasty parts of the system. You can have exchanges, borrow/lending, loans, interest rates, lots of different products with cross collateralization. You can have retail-facing UIs, you can have sophisticated investment products, all using the same on-chain protocols. It's actually pretty easy. Once you have the primitives built out, you can just create a button that anyone can click to make their own ETF. In two minutes, you have your ETF.

SPENCER: Interesting. So I'm like, "I wish there was an ETF that just tracked this particular industry, but there isn't one. So I'm going to construct one on-chain where if you buy it, it will move with the price of that industry. Now anyone who wants to can invest in that."

SAM: That's right.

SPENCER: How does that actually work in practice though? How do you make sure it follows the correct price?

SAM: That's a really good question. At the end, somehow, some way, that has to be tethered to reality. If they are natively crypto assets and you're buying a crypto asset, that is what it is. If you want to buy Tesla, you have to have someone somewhere who's like, "This token is a Tesla token. I'm going to literally buy Tesla shares and put it in a brokerage account." Then you can redeem your tokens for this share of Tesla and vice versa, things like that, or oracles that service this interface.

SPENCER: But what's the point then? If someone's gonna go buy Tesla and put it in a bank account, why not just have them create an ETF? What are we getting out of the fact that it's on blockchain?

SAM: Because you only need to do that occasionally. Most transactions don't need to go through that layer. Usually, you could write an ETF but then you have the same problem with this ETF and the brokerage layers around that. You can do almost all transactions within transfers, instantly on-chain, cheaply. No one needs to deal with brokerages. It's only when someone actually wants to cash out into the real world that you need to go through this clunky process. But you can abstract that away for everything else that happens.

SPENCER: I see. So most of the time, you're just buying shares in an ETF, and you can just do it with your existing crypto holdings. You can do it in a few seconds or a few minutes. There's not all the complications with it being tied into the actual financial system.

SAM: Exactly. As long as there's a way to redeem it, even if it's clunky, that's enough to keep the price in line. If it goes too far out of line, some liquidity provider will come in and do that arbitrage.

SPENCER: It's also interesting, the idea of tokenizing different things. You can have boring examples like tokenizing a stock. Then it's like, it's just another way to trade the stock. But you can have more interesting examples. Someone can be like, "I think I'm going to earn a lot of money in my career. Let me tokenize my future salary for the rest of my life and sell some of it now because I think that would actually be beneficial." It's a crazy idea but you could do that in theory, right?

SAM: Some people have done that. There are a few people who have.

SPENCER: Oh, interesting. What are some other interesting things that people have tried to tokenize or you think people might tokenize?

SAM: Trading strategies. You could have an on-chain algorithmic trading strategy that then gets tokenized by some pool. You can think of it as an on-chain hedge fund; you can think of it as an automated trading tool. The line between them sort of blurs.

SPENCER: But everyone can see the code, right?

SAM: Exactly, if you do exactly what it says it does. Or if you want there to be a human element, you could have limited scope. You could say, "Here's where the human inputs how much size to buy today." You can still give the human the right address inputting things, but not give them the buttons to steal all the lines. You can scope out what people can actually input.

SPENCER: Interesting. This person has limited control, and you can prove — looking at the code — that in fact, they can't take all the money out, but they can do XYZ. That's pretty interesting. Then I imagine you can tokenize things like real estate in a different area, or even a single house, or all these kinds of things, and then people could invest in them. It's also interesting from the point of view of startups. Have there been startups that have basically said, "We're gonna sell tokens to our startup. It's not a cryptocurrency. It's just a potential ownership in how well our company does"?

SAM: Absolutely. In fact, most crypto companies do this in the end. Those are the first movers in this space, the native crypto companies that realize that there's a way easier way of doing this.

SPENCER: Well, a lot of them just have a token, but it's not really equity, per se. But you're saying that, for a bunch of them, their token essentially represents equity in the business?

SAM: Or at least it's somehow tied to it, or has some way to draw on it such that it's not... [interrupted]

SPENCER: Linked to the success of the company, essentially?

SAM: Yeah.

SPENCER: Well, that was super interesting and informative, Sam. Thanks so much for coming on and for sharing all your insights.

SAM: Thanks for having me. It was a blast. It's refreshing to talk to someone who's engaged and thinks about new topics in a sharp way.

SPENCER: Oh, thanks! Appreciate that. Well, one thing I appreciate about you is you're willing to talk about both the good and the bad. I think with crypto especially, you can have the lovers and the haters and it's not always easy to get that nuanced perspective. So that was really cool.

SAM: Oh, yeah, it frustrates me, too. I think it does a disservice to the industry.

SPENCER: Awesome. All right. Thanks, Sam.

[outro]