Guest User

June 10, 2022

Understanding Crypto

Understanding Crypto 2: Prof. Igor Makarov: Economics of the Crypto Ecosystem

Guest User

June 10, 2022

Understanding Crypto

Igor Makarov is an associate professor of finance at London School of Economics. His research covers theoretical and empirical topics in capital markets, with a focus on information economics, limits of arbitrage, and cryptocurrencies. He has published in leading academic journals, including the Journal of Finance, the Journal of Financial Economics, and Review of Financial Studies.

Dr. Makarov has won several professional awards, including the 2012 NASDAQ OMX Award for his work on CDS auctions and the 2007 Crowell Memorial Prize (second place) for his work on sources of systematic risk. He received his M.Sc. degree in mathematics from the Moscow State University, an M.A. degree in economics from the New Economic School, and a Ph.D. in finance from the MIT Sloan School of Management.

Understanding the complexity surrounding cryptocurrencies is essential in making the correct decisions regarding investing in DeFi technology. To help us understand the basics, we talked to Dr. Igor Makarov who is an expert on cryptocurrency and Bitcoin, particularly Bitcoin and the associated mining processes. He is based at the London School of Economics, where he serves as an Associate Professor of Finance. Dr. Makarov is also the author of several papers focusing on DeFi and crypto markets in general and has provided new insights surrounding governance and mining processes. In today’s show, we learn about the basics of cryptocurrencies, mining and the future of DeFi. In particular, we talk about the role of intermediaries, what drives the prices of Bitcoin, how concentrated mining processes are, the role DeFi in increasing governance, the upsides and downsides of cryptocurrencies, and much more. Tune in to make sure you don’t miss out on advice from a respected figure in the industry, Dr. Igor Makarov!

Key Points From This Episode:

A brief outline of Dr. Makarov’s professional background and experience. [0:00:39]
What role do intermediaries play in the traditional financial system. [0:02:36]
Find out if economic rents that intermediaries collect are unwarranted. [0:03:36]
Dr. Makarov explains the complexities of cryptocurrencies and the elimination of fees. [0:06:06]
How rents are different on cryptocurrency exchanges. [0:09:44]
Systemic risks associated with the traditional banking system. [0:11:24]
Whether Bitcoin and DeFi can improve banking by reducing systemic risk. [0:13:22]
Learn if blockchain or DeFi ecosystems can exist without human intervention. [0:15:06]
Why it is unlikely decentralized autonomous organizations will improve governance. [0:17:06]
Breakdown of the potential problems that concentration of ownership could have on governance. [0:19:53]
Opportunities where cryptocurrencies and DeFi can improve the traditional financial system. [0:21:44]
Some of the potential benefits of a permissioned distributed ledger system. [0:24:39]
Why is it important to understand the Bitcoin ecosystem. [0:26:22]
What are the limitations of understanding the Bitcoin ecosystem. [0:27:12]
How Bitcoin addresses are associated with with real-world entities. [0:29:36]
Ways to differentiate between addresses belonging to individual investors and those belonging to intermediaries. [0:31:42]
What happens when you send Bitcoins to an exchange. [0:32:49]
Details on how Dr. Makarov calculated the concentration of Bitcoins. [0:33:25]
How did Dr. Makarov gain an understanding of what Bitcoin is used for. [0:37:14]
The role exchanges play in influencing the total volume of Bitcoin transactions. [0:39:00]
Why exchanges are essential to the overall network. [0:41:06]
The challenges in enforcing KYC and AML rules for the crypto ecosystem. [0:41:56]
How fungibility could effect the use of cryptocurrencies in the marketplace. [0:45:19]
Whether most important cryptocurrency exchanges are generally decentralized. [0:46:36]
What portion of Bitcoin transactions does Dr. Makarov consider to be economically meaningful. [0:47:06]
Why most Bitcoin transactions comprise of irrelevant transactions. [0:48:08]
What the meaningful Bitcoin transactions are being used for. [0:50:06]
Why the estimates of illegal activities by Dr. Makarov differs from other studies. [0:52:56]
A more in-depth explanation of what is driving the price of Bitcoin. [0:55:54]
How Dr. Makarov was able to identify individual Bitcoin miners for his paper. [01:01:57]
Dr. Makarov explains why the original vision of DeFi has not been realized. [01:05:52]
Reasons for the concentration of miners in the Bitcoin space. [01:06:42]
What are the risks are for the Bitcoin ecosystem when mining concentration is high. [01:08:24]
How Dr. Makarov determined the geographic locations of mining operations. [01:10:52]
What the function of a mining pool is. [01:13:36]
How the concentration in mining power affects the security of smaller proof of work blockchains [01:16:20]
How concentrated the ownership of Bitcoin in the hands of individual investors is. [01:22:40]
We find out if cryptocurrencies are democratizing financial services. [01:25:03]

Read the Transcript:

Igor, what role do intermediaries play in the traditional financial system?

Intermediaries perform many important functions. To name just a few, they help allocated sources to their most productive use. They also provide efficient means of transferring wealth from people with surpluses to those with deficits. And they also help transfer wealth across time and states. Often intermediaries exist to alleviate important frictions in financial markets, such as arithmetic information, address selection, and moral hazard. They're not always successful, I'd say, because intermediaries themselves are subject to the same frictions. The important question is where the intermediaries help create more efficient financial system and how we can structure and regulate them to achieve this goal.

Are the economic rents that intermediaries collect unwarranted?

It's an important but difficult question. Certainly there are areas where I feel intermediaries' fees are excessive. And just to give you one example, even today, we see that many banks charge very high fees when we try to, say, swap British pounds for the US dollars. Banks often charge one, 2%, irrespective of the size of the transaction, and these transaction entail very little risk for banks because they can hedge these transactions with a foreign exchange market. The fees are excessive, but to say that these are excessive rents, that might be difficult because it might be that these fees are used to cover some other costs, for example, bricks and mortar costs. I think it's fair to say that, in general, we recognize that intermediaries are in a position to charge excessive fees and excessive rents, and for this reason they are subject to a right set of regulations.

But to say in any particular case whether fees are excessive or not, it's not so easy because it's all about the level of rent. And we know that we should leave some rents, because rents are important for firms to innovate and invest in new technologies. Also, intermediaries are charged with a number of important societal goals. If we talk for example about banks, it's not just a nonprofit, but we also want banks to safeguard access to the financial system. We want them to comply with anti-money laundering laws. We want them to check KYC, know your customer, norms. And of course this is a costly activity and usually it is a bank that carry or bear these costs. For banks to have incentive to comply with regulation, we also need to leave them some rents. It's always a question about the level of their rents, right? And then we need to decide whether it's excessive or not. That's not so easy.

One of the things that I've seen DeFi and maybe crypto more generally take aim at is the fees, the level of fees and costs in the traditional financial system. Does the type of decentralized financial system that we see crypto enthusiasts kind of promoting, does that eliminate rents in the system?

I think the answer would be not necessarily. And here we need to step back and ask ourself, "Well, in what circumstances we can expect excessive rents or essentially not excessive rents," right? And if we think about excessive rent, it's usually a sign of limited competition, and competition can be limited for a number of factors. And the usual factors are some barriers to entry, asymmetric information, switching costs, some network externalities. Now, if we think about the centralized finance, their premise that, "Well, we have an open access system, permission-less protocols, anyone can participate, so the rent should be dissipated." But free entry is not synonymous to more competition because there could be some endogenous barriers to competition that comes from these factors. And what we show in our recent paper that we wrote for Brookings Papers on Economic Activity Conference, which is available as an NBR paper and SSRN Network, that there are several natural points in decentralized finance where rent can be accumulated.

I'm not going to go through the whole list of this point, but just to give you some example. First rents can be accumulated at the level of protocols like Ethereum, and the reason the rents can be accumulated there are very similar to what we observe in the case of Visa or Mastercard's network. Here, we see some strong network externalities. Developers are more likely to develop the applications for a network where already there are a lot of customers. Also, customers are more likely to join a network where there are a lot of applications. And so already established networks, therefore, can charge potentially high fees, and smaller networks will find it difficult to compete with these large networks. An additional complication here is that the security of the networks also very often is linked to the network size, be it like in proof-of-work protocol like in Bitcoin, or proof of stake protocol like Ethereum is now moving towards. In both protocols, larger networks are usually more secure and therefore smaller networks or potentially rival networks might find difficult time to compete with these networks.

Just to give you another example, we can think about decentralized exchanges. Here,, similar to centralized exchanges size matters. If we have large exchange, then large exchanges usually have significantly high liquidity, and as a result, large exchanges can charge higher fees compared to small exchanges, and small exchanges would find difficult time competing with large exchanges. It's a similar mechanism as in the traditional financial system, work in the centralized finance or in this crypto space, and therefore we in general should expect that rent will be accumulated at different points or in different levels of the system unless we put some regulation.

Are the rents higher? I've messed around a little bit on the Ethereum blockchain, and the fees to do anything are crazy.

Yes, right now the fees... I've been also surprised when I look at the level of the fees. Last year, we have numbers, and Ethereum collected about $10 billion in fees, and that's compared to Visa network of these were like $24 billion, but that's in revenues, not just in fees. If we look at the fees per transaction, Ethereum fees were 100 times that of transactions in the Visa network. Having said this, it is true that people recognize that proof-of-work protocols are expensive because of all the expenditures that we need to spend on electricity. And that's the reason behind the push to move to proof-of-stake protocols, which supposed to be cheaper, and we already see that some networks that move to proof-of-stake protocol or started as a proof-of-stake protocol, there the fees are smaller. But again, it doesn't mean that in the future, the fees will stay small or they would be at the lowest possible level. It could be the case that once the network is large and important, then it might choose to charge, say, fees which are above the marginal costs.

Yeah, that's super interesting. In Satoshi Nakamoto... the pseudonymous Satoshi Nakamoto's... early writings on Bitcoin, they wrote that... this is a quote... "Banks must be trusted to hold our money and transfer it electronically, but they lend it out in waves of credit bubbles with barely a fraction in reserve." In the traditional banking system, how much of a problem is the type of systemic risk that Nakamoto was worried about?

I think here it would be fair to say that we give banks our money, not just to hold them. We want actually to earn some return, right? Holding money would be a quite a straightforward problem to solve. What is difficult to solve is that, if we want to make our money work well for us, say, or in general in the financial system, that kind of goes back to one of the functions of the financial system, is that we want to allocate resources to the best possible use, right? And we, as savers, we have some excess of liquidity and we want to give the liquidity to, say, firms of other people who have good ideas and can invest these resources for us. In that sense, it's almost to the extent that future is uncertain, to the extent that we do not perfectly know which projects are good or bad, because it depends on so many factors, right?

The information which we need to take into account while making these decisions is spread across many people in the economy. We need to aggregate all this information, and then again to go through possibly different views on what is a good technology, what is a bad technology, right? That's what makes this problem quite difficult. And we should expect that, well, we make sometimes mistakes. The loans that banks as intermediaries make, they are not perfectly safe. As a result, there could be some states of the world where banks make some losses. Of course, what we want, we want the most efficient system possible, right? We want to minimize those mistakes, but we need to recognize that the problem is really difficult. It's impossible to solve completely this problem.

Does solving the systemic risk problem in the way that Bitcoin and DeFi propose improve banking?

Well, I think here we are not yet there. It's not that they propose to solve the systemic risk problem. I wouldn't say so, because right now they operate under what we call narrow banking mandate, where if people would like to borrow, say, some coins, they have to post collateral which is even in excess of the amount of what they want to borrow. Clearly we can see is that it might serve some investors, for example, if investors want to take some leverage, et cetera, but a lot of people in traditional financial systems and not in a position to borrow some funds and put some liquid collateral. Those people usually just don't have actually liquid assets to put as a collateral to begin with.

That's one of the frictions, right? People have some labor income, but that happens to be in the future. And then the question is, well, how can we borrow against this future labor income which is not liquid, which is not tradable. That's what makes the problem difficult in a way. So far, what we see in the DeFi space, that it's largely abstracted from this really difficult problem, but at the same time, the problems that is important for very large part of population, especially those who are maybe in most need for resources, for funds and would like to borrow against the future, while anyone there would see some problems, but I think we will touch upon those problems a bit later.

Some of the things that we hear a lot about with blockchain is trustless, that it's trustless, or that code is law. Can a blockchain or DeFi ecosystem exist without human intervention?

That's a really good question. And that really depends what we mean by "without human intervention." If we mean where the smart contracts can be executed without human intervention, in an automatic fashion, sure, we can write those smart contracts. But if we ask a somewhat, maybe more relevant question and interesting question, whether the system can, for example, function without any recourse to legal system or whether we can automate the investment process, and I think very quickly we can get to the conclusion that, while that's probably not possible, because no system is perfect. And even if it's perfect at a particular point in time, it doesn't mean that, say, one year, five years from now, the system would be as efficient as or maximum efficient from the point of view of some new technology, et cetera, so we would need to make some changes.

Once we need to make some changes, we would need to rely on humans to make those decisions, right? What kind of protocols we want to adopt. Also, if again, we see a lot of problems with even smart contracts, right? Smart contracts, because for smart contracts to execute without any human interaction, this must be a complete contract, as we say in economics, where we envision all possible states of the world, and we say, what should happen in every possible state of the world? Which is costly, right? For this reason, very often in the real world, we leave some contracts incomplete, and then we also allow ourself to resort to the legal system to effectively complete those incomplete contracts, and that could be cheaper than writing a very, very complete contract where we envision all possible states of the world.

Would decentralized autonomous organizations improve governance compared to the existing financial system?

That's another important question, and the short answer would be unlikely. And the reason being is that the governance problems in DeFi space are very similar to what we see in the normal corporate space. Here we also have different types of stakeholders. These are developers, validators, miners, users, investors. The interest of these different types of investors are not necessarily aligned, right? For example, if we think about users. Users generally prefer lowest possible fees, but validators, for example, like higher fees. We need some set of rules, and we usually call it governance rules, that ensure that different types of investors don't fight with each other, that investments are made in the efficient way, so that minority stakeholders are not expropriated by some powerful insiders. But in terms of economics, the same frictions that apply in the corporate world and in crypto space, they are very similar.

And we have like huge amount of academic research that studied corporate governance, what rules make governance essentially kind of good corporate governance, what rules don't work. And while governance rules somewhat different across different countries, it's almost uniformly accepted truth that for good corporate governance we need to resort to the legal system. Usually market mechanisms alone will not solve the problem, because again, incentives may not be enough. Sometimes we need to go after people who make bad decisions and punish those agents. But if you think about resorting to the legal system, it goes against the maxim of decentralized finance, which says that, "Well, we should not trust to anyone, right? We should rely to essentially trust this trust system." And that's why I think that this experimentation with decentralized governance, while interesting, but in the end, it will kind of converge to the conclusions that we make in the traditional financial system, that we probably would still need to resort to the legal system. And in that sense, it would be very similar solutions that we have in traditional firms.

One of the things that we're going to talk about later is the concentration of ownership and mining in Bitcoin specifically. Does that type of concentration cause problems for governance in DeFi?

Yeah. It could. For example, to give you maybe just simple example, think about, we have miners in Bitcoin and this is a proof-of-work protocol, and it's based on the ideas that miners have to solve some difficult mathematical problem, and for this they invest in expensive hardware, right? And hardware is very specialized. It's kind of designed to solve one particular problem, so we might think that, well, actually this protocol, for example, is not very efficient and we would like to move to more efficient protocols, right? Because Bitcoin, we know that it cannot process a lot of transactions. It's relatively slow, so we might think that, well, maybe it's good to have something else. But from the point of view of miners, who already invested a lot of resources in the hardware, so they incurred a lot of fixed costs, this move would destroy some of their investments and therefore it would not be necessarily in their interest.

And so naturally they can resist the move. And of course, if every miner is small and doesn't have any power, maybe again it would be easier to persuade the community to move to the new protocol. But once we have just very few powerful insiders, then this move could be more difficult to accomplish because they can say collude with each other and block those investments. And that's not just the fictitious story. We have some accounts that, indeed some worse... okay, maybe not worse... but at least this kind of fights did happen among Bitcoin miners and people who are in charge of Bitcoin protocol.

Yeah. It's super interesting to think about. Where do you see opportunities for crypto and DeFi to improve the existing financial system?

Right. I think it's also an important, but not an easy question. That's almost like I'm not even sure if it's a billion-dollar question. It could be much-billion-dollar question. That's what the industry tries to find out. I think the underlying distributed ledger technology holds lots of promises. It can reduce transaction costs, and that includes, for example, shortening settlement time. It also can make the system less fragile and prevent, for example, kind of to increase the resilience to some particular failures. Usually people talk about cryptocurrency, so decentralized finance, when they say, "Well, improve over the traditional system." That's where we need to be little bit more precise, right? What do we mean by decentralized finance? And usually, when people talk about decentralized finance, they mean a system which is based on permission-less protocols and where the access is anonymous and unrestricted.

If we are to adopt that definition of DeFi, then I would say here the question becomes less clear, because if we compare this system with a system where, say, validators... there are still a lot of them, but we supervise those validators, for example and if we have, say, different countries allow a certain number of validators, right? It's not even one country that controls the network, so it's quite distributed. But nevertheless, still we know who validators are and if necessary, we can go after them if they misbehave and do wrong things. If we compare these two systems, then it becomes less clear that this permission-less, anonymous system actually would be significantly more efficient compared to the system where we have some control over developers. We know who they are, and we can impose some constraints on them, on validators, so that they don't misbehave.

In that sense, it was a long conversation, but ultimately I think comparing these two system, it becomes less clear where the current designs will be able to significantly improve upon the traditional system. It certainly pushed the traditional intermediaries to be more efficient, and we can see that now they adopt new technology, which makes transaction fees lower. And again, more leaner, but so far we haven't seen that they solved problems which the traditional financial system couldn't solve.

You were talking about like a permissioned distributed ledger system. Is that right? Just now?

We can call it this way. I think in a way it depends really also what is a permission, right? If we call permissioned any system where we know who validators are, yes. Then that's what they call the financial system. But in some sense, we can envision a system where, for example, the number of validators is not a constant, so we allow almost anyone who wants to be validator enter and be a validator. It's just that they would be subject to some regulation. And if we think about some other societal functions, for example, enforcing KYC norms and, say, oh, anti-money laundering laws, then we can see how that system might be more efficient that the existing system, right? Because with the permissioned network, we can mandate validators who is processing transactions that only come from verified or certified users.

They don't necessarily need to know who stands behind particular cryptocurrency addresses. They just need to know that those addresses are certified, and they would have some agencies that would provide the link between the certified addresses and particular individuals, like digital ID, et cetera, right? While in the current design, this function in particular is very difficult issue.

Okay. That is interesting. That's like a way of enforcing KYC protocols, for example, by requiring people to adhere to certain things, to participate in the network.

Yes.

Is that right?

Yes.

Huh. Yeah, okay. That's interesting. Why is it important to understand that Bitcoin ecosystem?

Oh, Bitcoin is oldest and still by far the largest cryptocurrency, and it's often seen as a barometer for the whole crypto space. We can see that many products are first offered essentially in the Bitcoin space. Think about futures, think about ETFs or pension system. And Bitcoin also was the first cryptocurrency to be adopted as a legal tender in El Salvador. It's quite natural to look what's going on at Bitcoin and understand the problem, because again, it's almost half of the market capitalization of the whole crypto space.

What are the limitations of understanding the Bitcoin ecosystem?

The main problem comes from this anonymous and unrestricted access. As you know, the way Bitcoins are stored, they're stored in Bitcoin addresses, which are just alphanumeric streams protected by cryptographic algorithms. If we step back and think about how traditional financial system works, we have centralized intermediaries, which are charged with safeguarding the access to the financial system, and they achieve this goal by collecting and protecting personal information. Cryptocurrencies, in contrast, they protect privacy by not collecting this information. As a result, we really don't know who stands behind those crypto, say for example, Bitcoin addresses, right? And decentralized finance, while Bitcoin protocol is not particularly well-suited for decentralized finance, but say protocols like Ethereum, where we have smart contracts, they actually make or shift the balance in favor of personal privacy even more, because there in the Bitcoin space, we have centralized exchanges and a lot of centralized intermediaries that can do KYC checks, and a lot of them actually are charged with these functions now.

But in decentralized finance world, we have now decentralized exchanges where you can completely bypass KYC checks and say, for example, trade with other people using smart contracts, right? In this world, it becomes really difficult to, for example, to understand whether we trade with some sanctioned entity or not. And also enforcing taxes becomes quite difficult, right? In the centralized, when we talk about exchanges, exchanges see how we trade, they can keep track of our trades and provide reports to IRS. And also to tell us where you realized such and such profit. Again, with decentralized exchange, when we don't really know even who trades and for example, it's just tracing all transactions along the long chain, it becomes quite a difficult problem, so there is problems.

How did you associate Bitcoin addresses with real-world entities?

There are essentially three ways, but only two ways are practical, accessible for researchers. First, some people post their Bitcoin addresses on the web, so by scraping cryptocurrency blocks, we can link people and those addresses. The second way is that we can interact with real-world entities. Think about exchanges. When we send Bitcoins to an exchange, the exchange provides us with a particular Bitcoin address where we need to send our Bitcoins to. By collecting this address and then using clustering algorithms, we can find addresses which belong to a particular exchange. And we can repeat this or we can apply these methods to other entities. For example, if it's a marketplace, the marketplace would again provide us with a particular exchange where we need to send, say, Bitcoins if we want to buy particular goods.

Finally, the third way is that when people interact with the entities that enforces KYC norms, for example, exchanges, then they exchange because it enforces know-your-customer norm, they can link addresses that the exchange knows an address where I sent my Bitcoins from, and it can make this correspondence, but that's usually private information of the exchange, which is not available to researchers. But this information might be available to regulators if this exchange is regulated and in the jurisdiction which is in the domain of regulators, which is in the domain of legal agents.

This is all in your paper, Blockchain Analysis of the Bitcoin Market, so listeners know what we're talking about here. In that paper, how did you differentiate between addresses belonging to individual investors and those belonging to intermediaries, like the exchanges you were just talking about?

Yeah, that's a good question. First, why we would like to differentiate between intermediate and individual addresses? And here what we are after is to understand the concentration of Bitcoin holdings. The problem is that when investor sends a Bitcoin to an exchange, they forfeit ownership one of the Bitcoins to the exchange, and then the exchange usually collects all Bitcoins and mix them together and stores in a so-called cold wallet. These are the Bitcoin addresses stored on a particular device, disconnected from the internet for security purposes. If we see Bitcoins in this cold wallet, this is an aggregate claim on Bitcoins of many people.

Can I just ask a quick question? When people are sending their Bitcoins to an exchange, that individual no longer owns their own Bitcoin on the blockchain. They're trusting a centralized exchange to keep an account of the Bitcoin that they've deposited. Is that right?

That's correct.

Okay.

And that's one of the selling points of decentralized exchanges, because with decentralized exchanges, you don't need to give up the ownership. You are always in control and then transaction, if it's executed, you just swap the ownership of one coin for another coin. But that's not what's going on with centralized exchanges. The exchanges keep their own record of who owns what, and also this record is adjusted, based on your trading behavior. Pretty much the same way as it happens in other centralized exchange.

Right. I was going to say, that sounds a lot like a regular exchange.

Yeah. They are regular exchanges. It's just that we deal with one Bitcoins and if we send them to a centralized exchange, it's exactly the same. It's indistinguishable from any other trading instruments. Yeah. Now coming back, if we want to understand the concentration of Bitcoin ownership, well, we cannot do much about holdings within exchanges because we don't see who the ownership of Bitcoins within the exchange. The exchange has this information, but we do not. For what we do in our paper, we try to only look at individual holdings of Bitcoins outside exchanges, so as a large intermediary. And for this reason we would like to separate these two groups. How do we do this? And the general idea is that accounts of exchanges or as intermediaries, they should be more active compared to individual accounts, with a caveat that cold wallets might look also similar to individual accounts.

We need to be very careful not to confuse cold wallets of exchanges or other intermediaries with individual accounts. How do we do this? For this, we essentially follow two steps. In the first method we scrutinize addresses with very large Bitcoin balance, the so-called Bitcoin Rich List. Any address or cluster of addresses with more than 1000 Bitcoins. And then we first check whether it belongs to a known intermediary and we have a kind of a large set of known intermediaries. If it doesn't belong, then what we do, we build a network of clusters that interact with this particular address. Because again, the worry is that it might be cold wallet, and we would like to see whether it's linked again to any cluster that possibly look like intermediaries.

And again, we do quite careful analysis. It takes quite a lot of time. We tried to automate it, but not everything can be automated. In difficult cases, we have to examine by hand manually and kind of presume whether it really looks like an individual account or intermediary account, and we try to be here quite conservative because again, what we want, we want essentially measure of... The preview of the result will show that concentration is really high, and therefore we rather would like to make a mistake, misclassifying some individual addresses as intermediary addresses, but not vice versa. For example, if we have a whale, an investor with very large holdings on the exchange and this investor takes out Bitcoins from the exchange, then in some cases we can actually misclassify this whale investor as kind of address at intermediary address, because it would have a link to the exchange. But again, we do so that to be conservative in our estimate of concentration of individual holdings. This is one method.

And the other method, because we cannot scrutinize all addresses in the network because there are more than 600 million of these addresses, what we do for smaller addresses with smaller balances, we have a rule of thumb that if particular cluster of addresses was not active for a year prior to where we look at the concentration, then we deem these clusters or addresses as belonging to individual investors. Because again, intermediaries are more likely to interact with these clusters. That's what we do.

So fascinating. How did you gain an understanding of what Bitcoin's actually used for?

Well, here I think it's good to step back and ask what makes a cryptocurrency valuable, right? Where does the value come from? And here, well, first cryptocurrency, if it's used as a transaction medium, can also obtain some value, similar way to money, right? If that's the case, we should see that Bitcoin and we would see Bitcoin flows to some marketplaces, be they legal marketplace or legitimate marketplaces. We also can see that a flow of Bitcoins to what is called payment processors. These are again, special entities that process payments say, for example, for some websites, et cetera, or ATMs. Yeah. That's number one.

Number two. Well, if Bitcoin is used as a store of value, if that's the primary essential use of Bitcoin, in principle, we should not see a lot of trading for them, because people would just occasionally buy Bitcoin and hold it. Yeah. Finally, the third reason could be just a speculative mode. If people not just hold Bitcoin, but they try actively to predict what will happen to Bitcoin price in the next moment and, based on their expectations, they would either increase their position or decrease their position. And in the third case, we should expect to see a lot of volume to exchanges, from exchanges and a lot of volume on exchanges if that's the case. And that's what we find in the data.

How large of a role do the exchanges play in the total volume of Bitcoin transactions?

We estimate that exchanges and over-the-counter trading desks, they account for roughly about 75% of what we call off-chain Bitcoin volume, right? That's the volume which is recorded on the blockchain. This volume... we can see this volume when people move their Bitcoins from the private wallets to an exchange, essentially from an exchange, and also when people send from one exchange to another exchange. There is also off-chain volume. Did I say, by the way, off-chain or on-chain? I might have confused. In the first case we were talking about on-chain volume, the one which is recorded on the blockchain, right? Off-chain volume is the one that happens on centralized exchange. It's the one which we do not see on the blockchain itself. It's just we have normal limit other books and then exchanges cross volume on these other books.

What we show in the paper, we mostly analyze on-chain volume, the volume on the blockchain, but we show that exchange volume on the blockchain, while it's smaller than off-chain volume on exchanges, but it's comparable to that volume and also correlates quite highly with the volume. In a way, our results suggest that actually when people move their Bitcoins to an exchange, it's an indication that they are more likely to trade quite soon on the exchange, and in principle that, again, this is a motive for holding Bitcoin, so this speculative kind of motive is important.

Just to be clear, was that 75% of on-chain volume is through exchanges?

Exactly. Yes.

Wow. Okay. That's a lot.

Yeah. That's a lot.

Other than transaction volume, how important are exchanges to the network?

Well, exchanges are important, because especially in the Bitcoin network, exchanges are almost always centralized, and these are the points of entry and exit of funds to networks. Whenever people would like to buy extra Bitcoins, they would need to send, in most cases, their funds to one of the exchanges, so again, OTC trading desk, and that's how they buy or how they cash out their Bitcoin. They play really a central role, I think, in this ecosystem.

We touched on earlier the idea of permission systems that could help enforce KYC in a global financial system, which is a pretty cool idea. Practically speaking for Bitcoin, what are the challenges in enforcing KYC and AML rules for the crypto ecosystem?

Yeah. Here I would actually probably make a distinction between the Bitcoin system and say, Ethereum, which allows also decentralized finance solutions. And the distinction is not completely black-and-white because Bitcoin also develops some solution that allows to run some decentralized applications. But right now, if we think about that, Bitcoin mostly is centered on centralized... say, exchanges on centralized intermediaries. There in principle, if we regulate all exchanges and all exchanges adhere to strict KYC norms, we would have something that looks very similar to the traditional financial system, where we have centralized intermediaries, and then the Bitcoin protocol works like as a payment system that connects different intermediaries. We can replicate that system. The problems come that not all exchanges are usually like that. They're not all regulated or at least some of them are in offshore areas and only lightly regulated, so they may not adhere to the same strict norms, KYC norms.

If that's the case, then we have some potential entry nodes where some illegal money can enter the system. But if we now think about more about decentralized finance, right? That's where the real problems come, because there we have decentralized exchanges, for example, so people can easily trade on those exchanges and then it's very difficult to know, right? Whereas we trade against a sanctioned entity or even to trace all the transaction to calculate taxes. And what really matters then, it's also the adoption of cryptocurrencies, because if we have a very wide adoption, we might even envision a system... and that's, I guess what people, when they talk about adoption, they envision that for example, cryptocurrencies can be used not just for trading, but also to buy some goods.

Then if we can trust, again, these two systems, right? In the traditional system, we have the centralized intermediaries that control who enters and leaves, how money enters the system, right? And the benefit of the system is that we need to do checks only at a limited number of these nodes, right? When we go to shop, if we use a credit card issued by a certified bank, we know that these are already clean money. We don't need to worry. While if we don't have those agents who would kind of safeguard the entry to the system, that's what currently houses decentralized finance world. Then we run into problem, because I have some cryptocurrency, and I would like to buy some good, but then what? Is it the shop that needs to check whether this is essentially the source of my funds... legal, not legal. It becomes really complicated problem.

Interesting. That sounds like an enforcement problem, but it also sounds like it could be a fungibility problem. That accurate?

What do you mean by fungibility?

Fungibility, like if a cryptocurrency has a history attached to it, where it came from a sanctioned area or something like that, and that could affect whether it's accepted.

Yeah. I mean, you see, it's quite popular to say that, "Oh, because we have digital footprint, because we can trace all the transactions on the network, we effectively can maybe eliminate illegal transactions or be able to go after the offenders." The problem is that it becomes quite difficult to implement because we would need to trace along potentially very long route. But also this tracing only bites if we can identify who is behind this particular chain of addresses, right? But if, for example, we can use our cryptocurrencies to pay for goods, and then a particular store doesn't check because it's probably impractical, imagining that a store would have to go through all the potential chains for each customer, right? Then it would be really difficult problem. Somebody would have to do this in the system. And again, currently the current design of most cryptocurrency system is that essentially nobody checks as entry, and that introduces many problems.

We touched on centralized and decentralized exchanges. Are the most important cryptocurrency exchanges generally decentralized?

No. Right now we're still in a world where the largest exchanges are centralized. These are Binance or the Coinbase. But again, these different exchanges are subject to different level of regulation. Exchanges which are in the US, like Coinbase, they're more regulated compared to exchanges which are in some offshore, like Binance.

What portion of Bitcoin transactions would you consider to be economically meaningful transactions?

Yeah. Here, I think it depends what we mean by economically meaningful transactions. In the paper, we treat any transactions where the ownership of Bitcoin changes as meaningful. And this includes for example, trading. If I own some Bitcoin and I sell it to you, that would be meaningful transaction. We estimate that about 10% of all nominal volume on the Bitcoin blockchain as a meaningful transaction, and the rest means what we call some superfluous volume or fictitious volume. These are the transactions where the same entities essentially just moves funds between accounts that it controls. It's the same as if I take some Bitcoins from my left pocket and put it into the right pocket. Those transactions, we don't treat as meaningful because there is no change of ownership.

Why are all of those changing from one pocket to the other transactions happening?

That's because two main reasons. One is structure of the Bitcoin protocol. The way it works is that to understand what is the current balance in a particular account, we need to trace all of them, and the transactions that involve this particular account. And then, suppose somebody at some point sent 1000 of Bitcoin to a particular address. If we would like to make a payment of just one Bitcoin, then actually Bitcoin protocol is structured so that we have to send the whole balance that we received in the past and then collect the change. What we would see is that, again, one address would receive one Bitcoin and then the other address... either the same address or newly created address, but again, that would belong to the same party... would hold 999 Bitcoins. That's one reason.

The other reason is the so-called no address reuse kind of Bitcoin protocol because of anonymity concerns because of this potential traceability. The Bitcoin protocol encourages users not to reuse the same address more than once. Many wallets, therefore, are programed in a way that when we send our money or our Bitcoins, then we would kind of generate new addresses in every transaction. And as a result, what we see is a long chain, what is known as a peeling chain, where Bitcoin travels from one address to another address, and so on, and so on, and so on, and so on. And again, this is just because we would like to preserve anonymity of the transactions, and that's why we see this type of behavior.

If 10% of Bitcoin transactions are economically meaningful, do we have information about what those are? Like, is it retail transactions? Or what is it?

Yeah, well, as I said, so out of this 10%... we already touched a little bit, I guess in one of the previous questions... out of this 10%, 75% transactions for which are related to exchanges, and some small percentage goes to some illegal activities such as document marketplaces or transactions related to spam or to some ransom. We also see some transactions that go to gambling. We see transactions associated with payment processes where people use Bitcoin for, again, maybe to buy some airline tickets on some exchanges, but those are very, very small fraction of this remaining 10% of Bitcoin volume. Overall again, majority of volume is linked to exchange volume.

Just to put a finer point on this, how much illegal activity... because this is a common question... is being facilitated by Bitcoin?

Yes. In our estimate, and if we look at this 10% of meaningful transactions, we estimate that only 3%... not of the 10%, right? 10% becomes 100%. But 3% of economically meaningful volume in 2015 was related to illegal activities. By the end of 2020, that number became 0.4%, 40 basis point. But it doesn't mean that illegal activities don't matter because the volume also and the price of Bitcoin have been growing over time, right? We estimate that in 2020, for example, illegal activities were around maybe $1.6 billion, and that's only in Bitcoin. I think firms like Chainalysis or Bitfury, they estimate the illegal activities across all crypto, essentially within the whole crypto space and the number, if I remember correctly, it's about $15 billion, so it's not a trivial amount. It's something that we certainly should be concerned. It shows that as a percentage of the total activity, it's not a large number. If we think about value of Bitcoin, it's unlikely as that value of Bitcoin comes from these illegal activities as a means... From those transaction, it's more probably because of investors' desire to speculate in this market.

There's another paper that I read, a 2019 paper by Sean Foley, if I remember correctly, and they found like 46%, I think, of the transaction volume they found to be for illegal uses. Do you know why your number is so much different from theirs?

Yes, of course. That paper received quite a lot of attention and a lot of people cite this number. There are two main differences. One is that they intentionally exclude transactions associated with exchanges, so their set of known entities is very different from ours. We have much more precise data, so we can identify significantly more exchanges. They use so-called walletexplorer.com there. As far as I remember, it's less than 200 exchanges and most large exchanges like Coinbase or a Binance, that kind of exchange, etc. they're missing. That's one again. The exact comparison, therefore, is difficult, but again, one thing is that they don't count any transaction associated with exchanges in their calculations, so their base is potentially small.

The second important factor is that how they decide which transactions are used or which addresses are used for illegal volume. And what they do, they start with some darknet marketplaces, and then they look at addresses or clusters that interact with these darknet marketplaces. And they say that, "Oh, if a particular address receives more than 50% of its flows from this darknet marketplace, then it's also an illegal cluster. And they repeat this classification of clusters in a affinity recursive method. But what we show in this paper is effectively, this involves double counting of volume because they do not remove in their analysis and of this billing chain transaction... as I said, people try to clean up their Bitcoins that come from illegal marketplace, so they come through marketplaces.

And therefore what we see that very often, a particular Bitcoin would be sent to some addresses. It's not necessarily that the ownership of this Bitcoin changes. It's just that because we want to make it clean, we would mix this particular Bitcoin with some clean Bitcoins. And again, we split and now, which is kind of becomes gray Bitcoin into further parts, where like one half goes to another and we mix it with another clean Bitcoin and so on and so on and so on. What they actually count, they count this flaw as volume, even though, again, it's not necessarily that Bitcoin ownership changes. We just count volume that comes to a particular darknet marketplace and we think that's the right measure of illegal transactions., so again, the Bitcoin that comes to some other illegal entities.

Okay. You already touched on this next question, but I do want to ask it explicitly. Based on the documented uses that we were just talking about, what do you think is driving the price of Bitcoin?

Yeah, that's a good question. An interesting question. I think here it's good to start with some facts about Bitcoin pricing. And actually it is related to what we did in our earlier paper, which is published in Journal of Financial Economics. There we showed actually that Bitcoin price is related to assigned net volume, so net buying volume, we can see whether it's a buy or sell order. We can calculate what is the net order flow, and what we show in the paper that actually net order flow explains 80% of variation in the Bitcoin price. This is really a large number. We see that this number is significantly smaller in other markets like foreign exchange markets or equities markets. And it actually it means that a news component is small. So the price is mostly driven by investor sentiment. When investors get excited and they start buying Bitcoin and the price goes up and vice versa.

And actually in that paper, we also estimate the price impact, which show that it's quite stable over time, and 10,000 of Bitcoins induce a permanent price impact of about three to 4%. And that means that whales, or investors that control a large amount of wealth, play an important role in this market. Just to give you some example, right? We have this MicroStrategy, famous stocks that they now hold about 130,000 of Bitcoins that they accumulated over these three years. Our estimates suggest that actually MicroStrategy alone is responsible for about 40 to 50% appreciation of Bitcoin. If MicroStrategy is going to sell their holdings, then actually we would expect a very large Bitcoin price decline about 40 to 50%, so a lot of systemic risk associated with these guys.

Now I think people are always interested, oh, what will happen in the future? Right? Past is the past. Here, I think the future price will really depend on new inflows of capital because Bitcoin is unproductive asset, right? It's all about the willingness of people to hold, essentially, their capital in Bitcoin. And the inflow of new capital will depend on the lobbying efforts to bring serious institutional flows to the system. I think at this point, retail investors who wanted to invest in Bitcoin have already invested. The question, therefore, essentially institutional money is the only untapped resource, at least on a large scale, and we can see that there are some signs that institutional funds can indeed arrive into the space. Last month Fidelity announced that they will allow Bitcoin in their 401k plans. Personally, I wouldn't put my retirement savings in Bitcoin.

It is true that we can try to speculate on the short ride movement because of this price impact, if we know that a lot of institutions are going to buy, then we know the price will go up. But again, in the long run, I think the price appreciation of Bitcoin is limited because again, it's tied to the gross rate of aggregate wealth and again, that's usually not a large number. It's definitely not what we kind of... would not support this huge price appreciation of Bitcoin that we have seen in the last 10 years. But again, it will be difficult.

Right now, I cannot predict what will happen to this institutional money, whether they will succeed. And if they succeed, whether essentially there will be institution starts putting, say, pension money into Bitcoin. Here again, we might have some concerns, right? For example, because of the price impact, if a manager of a particular fund already has some prior holdings of Bitcoin, we can see how his interest could be conflicted, because he or she would know that, well, maybe it's not a good idea to put pension money into Bitcoin, but at the same time, once you know that your own holdings might go up, you might benefit. There should be concern. But let me stop at this.

Before we started really trying to understand... Well, I guess that's not true. We had a few people on our podcast a while ago when we were trying to understand this stuff, and we decided to go a lot deeper, which is why we're talking to you. But we talked to Marco Di Maggio from Harvard, and he told us when he was on that Fidelity has been mining Bitcoin since the early days, so it would make sense that they have, like you're saying, an interest in increasing the price.

Yeah. I was not aware of this fact, but it is consistent, let's put it this way.

Yeah, no. That's super interesting. It sounded like, when you were talking about flows and prices, that sounds like Ralph Cogen's stuff on price elasticity in stocks. Is that the same kind of thing? Like Bitcoin has inelastic price?

Well, it's certainly related, and the price impact in the Bitcoin market, I think, is the order of magnitude, or if not two order of magnitudes, larger compared to what we see in equities. It's mostly investor sentiment and not so much news. From time to time, we have some news that a particular exchange is hack, or I don't know, some regulatory improvement or approval of some particular exchanges, right? There are news, but quite infrequent and not like in equities.

In your paper, how did you identify individual Bitcoin miners?

A short answer would be that we follow the distribution from pools to individual miners. Most mining is done in pools. That's because the protocol provides incentives for miners to insure each other, because if I'm a small miner, the probability that I ever win the race and we will be able to verify the block and collect the reward is very small. What miners do is they organize themselves in pools and then, again, they share the reward. And the way pools operate is that they collect this reward using particular addresses. Very often they would include... Bitcoin protocol allows actually to include some text into Bitcoin addresses. So you would see that, oh, this particular block was mined by such and such pool, for example, by Antpool, so we would know that this particular address is associated with a particular pool, right? We can see that, oh, okay, on this picture, these green addresses, these are the reward collection addresses. That's when pools collect the rewards.

Once they collect the block reward, usually what happens is that they accumulate rewards, say within a day, and then each day, or sometimes twice a day... that depends on how large a pool is... they would distribute the rewards to individual miners. And different pools use different algorithms, and we spend a lot of effort understanding those algorithms and kind of go through them. Here you can see what happens in the case of Antpool. Once they accumulate rewards, they send these rewards into what we call reward distribution addresses. And I show you in the next picture how it works. Here, for example, we can see that we collected rewards on this address and then we sent to this address.

Once funds arrive to this address, what happens is that they start distribute these rewards. We can see that this reward distribution address receives 100 Bitcoins and then they start sending it to individual miners. I can see that this particular miner represented gives its address and then it receives that amount of Bitcoin and so on and so on and so on, because there are so many miners who work for Antpool. Antpool was one of the largest pools and still remains one of the largest pools in the space, so it's not possible to distribute the reward in one transaction.

Then we can see that the change is collected in a particular, in a new address. And then this address also kind of continues sending rewards to individual miners. And this process actually goes for many times. Sometimes you have more than 100 distribution. Usually at every step we distribute more and more rewards until we are done. What we do, we follow this distribution pattern and kind of associate each miner with this Bitcoin address that the miner uses to collect the awards. And we can see that the same addresses are used repeatedly and from time to time, and we can see how... therefore, we have the reward history of any particular miner. Overall, once we remove the smallest miners and leave only those that collect at least $1000 in total rewards, we have about 240,000 miners going through this process. Let me now stop sharing.

That is fascinating. Crazy. You mentioned Antpool being a big mining pool. Are the Bitcoin miners decentralized? I think it's important to point out that that was part of Satoshi Nakamoto's vision, was this one CPU, one vote, I think. Is that holding true?

Well, I think we know that it's not entirely true, and today we have specialized mining farms that do mining. What we show is that, as I already mentioned previously, that in total we have 240,000 miners at any particular point in time, maybe about five to 10,000 miners active, but the mining power is distributed very equally. We show that often as few as 50 miners control 50% of aggregate mining capacity. Again, it's not one miner, not two miners, but quite a few miners.

Based on the built-in economic incentives, is miner concentration all that surprising?

I think the very fact that mining is done in pool is not surprising because that's essentially the economics of the Bitcoin protocol. When we started this project and we tried to look at concentration of individual miners, we actually didn't have a good prior what will happen. And I think now with the results, I would say it is surprising. I did not expect it to be so concentrated, but at the same time, some people who may be closer to this industry, they would think that, well, maybe the industry has some increasingly chance to scale. Therefore, maybe we should expect miners to be concentrated.

How does the mining concentration vary over time?

What we show in our papers, that concentration depends on the price level of Bitcoin. Whenever price increases and it becomes more profitable to mine Bitcoin, we see the inflow of new miners, so concentration falls. And vice versa, when Bitcoin price goes down... or for example, we have the so-called Bitcoin halving events, where the reward for mining your block half, then around those, or right after those events, we see that concentration increases because a lot of miners leave the space. And our results suggest, therefore, that large miners, these are the stable miners, and it's mostly smaller miners who either leave or enter the space.

What are the risks to the Bitcoin ecosystem when mining concentration is high?

I think there two main concerns and they're somewhat related. It's when concentration is high, it's easier for miners to collude and therefore they can do several things. First, they can launch maybe an attack on the Bitcoin protocol. That may not be maybe the most important concern because at the end miners hence have some stake in the system, so they may not necessarily want to attack the protocol that feeds them. But miners, if there are few miners and they're concentrated, it becomes also easier to collude, for example, to decide on future changes to the protocol, right?

As we discussed earlier in the talk, sometimes we might want to make changes to protocol to make it more efficient. But again, because miners are responsible for security of this protocol, we need to take into account miners' views on the protocol. And then of course, if they're already made investment and as a result of the new changes, these investments will be lost, then they might be reluctant to implement these changes. That's one concern. The other concern, it depends also on the geographical concentration of miners, which will only be touched in the next question. If miners are geographically concentrated, then it also becomes easier for some third party to take over for these miners or for the state to exercise control, and that also introduces some systemic risk to the system.

That's a crazy one to think about. That's like if all the miners hypothetically are in one country and that country's military comes in and just physically takes over the operation, that's kind of the... Is that the risk that you're talking about?

Yeah. That's the risk. Yes. In a way now, of course, with the China before, kind of the preview, right? We know that a lot of mining activity had been concentrated in China and then, in principle we might envision some scenarios that, for example, if there are some disagreements between two countries and China would want to maybe take control, in principle it could happen. Well, now I think we are past that point maybe, but nevertheless.

Okay. How did you estimate the geographic location of mining operations?

Yeah, so what we did is, because we have very good data about miners, we can see not just the flow of Bitcoins to the addresses, but we also see where they cash out their rewards. Then the idea is that if they cash out in some exchanges, which prevalent in a particular geographical area, then it's a sign that a miner belongs to this area. One limitation of our approach is that some exchanges don't belong to one jurisdiction, for example, like Binance cross jurisdictions. In principle, miners who are in different parts of the world, they can send to these exchanges. But with this caveat in mind, we did this analysis and we arrived to a number which was consistent with prior estimates based on an analysis of IP addresses of miners, which was supplied by mining posts. We estimated that about, pre-China ban of mining activity, about 70% of miners, individual miners, were located in China.

Are there other areas of the world where miners are concentrated?

Yes. Well, I mean, I think what I just said, right? This money concentration that was pre-China ban. What we see after the ban is that a lot of miners moved their operations to some other countries like Kazakhstan, Indonesia. Also now we see some activity in the United States. I think while mining farms are now not as concentrated as before, at least geographically not as concentrated, there are good reasons to believe that the ownership hasn't changed significantly. It's probably same people who previously owned those mining farms, they are still the owners of those farms. They just lays them in different parts of the world. And again, we haven't done analysis, our analysis. We didn't extend our analysis yet to post-China ban, but also a priori it's unclear why concentration of miners should change, because you would just move your operation elsewhere. But if you were large miner, you probably still is a large miner.

You talked about the mining. So there's large miners, and the large miners operate through mining pools. Is that correct?

Yeah, that's what we see.

Are the pools required to do what the individual miners want or could the mining pool...

That's I think, it's an interesting question. in principle, there was some argument. We know that pools are concentrated and a lot of them are tied to China, but some arguments were that pool concentration doesn't matter because if a particular pool misbehaves, then miners can move away from this misbehaving pool and join some other pools. And if we think about, well, how easy to switch pools depends actually on the power of individual miners versus pools, which again, depends on the concentration, because it's also how easy to monitor a pool, right? If we have tiny miners, it's very costly to monitor whether your pool behaves or misbehaves. Larger miners, maybe it's easier to monitor, but at the same time you also have more power over the pool. And also you can probably maybe dictate or a little bit to negotiate the fees with the pool or how the pool sets the fees.

And I think also we know that actually it's difficult to start a pool if you do not have miners who would support your operation, right? Because if you're a small pool, to grow to large pool, which composed of very small miners, nobody would want to join this pool because it doesn't provide any insurance, so usually pools are organized around large numbers. I think we shouldn't think about pools and miners as completely independent entities. Most likely, again, at least the largest individual miners, they are connected to pools and they might be maybe either directly related to these pools or maybe even also the one that start this pool.

Yeah. Okay, interesting. For your pinpointing the geographic location, you were looking at what the miners did with their mining rewards. In general, are miners holding onto their Bitcoin mining rewards?

The behavior we see on the blockchain is very heterogeneous. Some miners cash out their rewards almost immediately, but we see also some miners that accumulated very large sums. For example, we can see some miners accumulate like 20,000 of Bitcoins. But we studied this investment behavior and cashing-out behavior of miners in a separate project, but we are not yet ready to share the result, but it's an exciting area. The data allow us to do this analysis.

That does sound exciting. How does the concentration in mining power affect the security of smaller proof-of-work blockchains?

Yeah. Again, it's a good question. Here what matters is actually the structure of, in a way, of the industry. Let's put it this way, right? If we think about the security of protocol depends that the majority of mining power lies with good agents, that the agent companies behave. But at the same time, miners should be compensated for their work, right? They have to be at least break-even, and that should hold irrespective, whether they hold 50% or 60% of aggregate capacity. Because again, we do not penalize... The protocol doesn't penalize the size of miners. Now, usually what we know is that miners have to incur some significant, to make initial investment. To be able to mine Bitcoin, they have to set up these farms, which again, expensive. If we think about incentives to undermine the protocol, therefore, because we have these high fixed costs, then in principle, undermining is not very profitable, because if we undermine the protocol, it would be more difficult to recoup this fixed cost.

Now, if we think about Bitcoin or Ethereum, what happened in the case of this protocol that, as the price of Bitcoin increased, so did aggregate capacity, and those protocols actually operated at almost a maximum capacity. Well, partially because the price of Bitcoin overall was increasing, or at least it was profitable to mine most of the time. And as a result, the system was almost at the frontier of computing power. The difficulty of obtaining 51% was more like a technological problem rather than incentive problem by itself. Now what happens is that, well, we see the emergence of some marketplaces where computing power can be bought. For example, places like NiceHash and others. For Bitcoin, the capacity of those marketplaces is rather small compared to the aggregate capacity of Bitcoin, maybe 1%, maybe 2%. Even if a third party tries to buy all the capacity of this available in these places, they would not be able to change the Bitcoin system.

However, if we're now talking about smaller proof-of-work protocols, there 1% of the total Bitcoin capacity actually might be very often more than 100% of the total mining capacity available to these smaller pools. And therefore what happens, could happen and actually even happened in the past, is that some third parties would rent some computing power from those marketplaces and use this computing power to launch an attack on smaller proof-of-work protocol, such as Bitcoin Gold or Ethereum Plastic. Summarizing what I was saying, we can see that large proof-of-work networks, they impose negative externality upon the security of small proof-of-work protocols. And remember, we talked about some barriers to entry and limits to competition. That's another factor that might, again, give advantage to larger networks because Bitcoin, again, it's secure, but is leftover of computing power, which is left from these large networks. It can be used to undermine, for example, the security of smaller networks and would make those network less secure. And again, people might not necessarily want to join those networks.

Crazy. Not so easy to do a proof-of-work startup, I guess, securely?

Yeah. But the same argument also applies to proof-of-stake protocols, because usually the security arguments or at least the industry has been mostly concerned about so-called double-spending transactions, where we spent a balance in a particular account twice. Right? And then the idea that, well, if people have a large stake in a particular protocol, then it wouldn't... in place as a collateral of their stake, they would not want to undermine this particular protocol because they would lose a large portion of their value compared to how much they can benefit from this double-spending transaction.

But this argument, it applies to double-spending transactions, but it neglects the fact that, well, an attack may be not necessarily... or the benefit of an attack might not come necessarily from these double-spending transactions, but rather large networks might try to undermine their rivals, right? And in this case, it's similar to large multinational companies we know that sometimes buy their rivals to shut down those rivals. We can imagine a scenario where a large and established network would get some stake in this smaller rival and then jeopardize... For example, they would try to, I don't know, do the double-spending attack, but again, the benefit would come that people would just get concerned with the security of a smaller network and not use a smaller network.

Interesting. So they might be willing to take... I mean, it's like a lost leader product, maybe. They're willing to take a loss on an attack to make the other network go away?

Precisely. You essentially reduce future competition. It's hypothetical and it's not to say that people necessarily would behave in this way, but it's a concern. Let's put it this way.

Very interesting. How concentrated is the ownership of Bitcoin in the hands of individual investors?

Yes. We show in the paper that it's actually quite concentrated. To put it in a perspective, if we compare, for example, to the United States' wealth distribution, in the United States, we have estimates that top 1% of people control about, I think it's about 35% of wealth, and 0.1% control about 16% of wealth. What we show that in Bitcoin ecosystem, even if we... using our conservative approach where we do not count large investors which have direct ties say to large exchanges, like Coinbase, MicroStrategy... if we only use these individual holdings, we show that essentially 100% of investors control about 26% of all Bitcoin wealth. It's almost 10 times concentration, 10 times larger, compared to the concentration of wealth in the US, so the wealth is concentrated.

Interesting. So not exactly solving wealth inequality?

Yes. In a way it depends what will happen, right? If we push for high-end Bitcoin adoption, people who already control a quite significant amount of wealth, they might become even richer. Of course, at the same time, if Bitcoin price falls, it means that right now it's relatively small group of people will be affected. Again, there are a lot of Bitcoin users. And again, of course, some of them might, again, as for example, in the now infamous case of Luna, where holdings were also concentrated, a lot of people who are not necessarily rich also lost almost their savings. It's not to say it's not important, but nevertheless, right now again, holdings are concentrated.

All right. Last question, based on everything we talked about, which has been pretty far-reaching... it's been a great conversation... are, in your opinion, cryptocurrencies living up to their promise of democratizing financial services?

I think a fair answer would be not yet. Currently we do not see is that cryptocurrencies or decentralized finance solve the full range of problems that traditional finance solves. They are rather focused on some narrow set of problems, mostly focused around trading. We also have narrow banking that allows that being leveraged into transactions, which again, a lot of observers claim that it's mostly used for trading. It might be that, again, in the future, the underlying technology would be able to bring down the costs of transaction, and through this process, we kind of arrive to a more efficient financial system.

But again, as we talk today, it's less clear that to arrive to this point, we really need permission-less protocols and anonymous access, which creates some other problems like enforcement, et cetera, and adherence to KYC norms. Therefore, I think I'm not very different from what other people in the industry would say, is that yes, there is a promise, but we do not fully see a clear path how this promise will be fulfilled in the future, but hopefully.

Awesome. All right, Igor. Thanks a lot. This has been fantastic.

Thank you, Benjamin. Thank you, Cameron.

Thanks, Igor. That was great.