Banner image for 2019 Financial Markets Conference: Mapping the Financial Frontier

Policy Session 1: Will Blockchains Lead Us to a Trustless World?

Some have promoted blockchain technology as a new type of database that will eliminate the need for banks and other trusted third parties. What is blockchain technology? What are its advantages and limitations? How is it being used in practice?

Transcript

Raphael Bostic: Good morning, everyone—it's good to see you all. Well, it's good to be here again. Thank you for joining us for the 24th annual Financial Markets Conference. I'm Raphael Bostic, president and CEO of the Federal Reserve Bank of Atlanta, and I'm really excited to be here. We had a great kickoff yesterday with Simon [Johnson]. I hope you enjoyed his presentation, and I know at our table we had a lot of discussion around the issues that were raised. It was really thought-provoking about the future of the U.S. economy and how we can use basic research and development investments to spread prosperity in a pretty significant way.

And the other thing that was interesting was this notion about geographic policy as well, and trying to think about to what extent are we missing out on opportunities dispersed through the country that could allow us to be more productive. His presentation, that discussion, was actually a very fitting backdrop for this conference, so we are going to talk about what's going to happen in finance in the next decade. And the intersection between finance and real production is clear to all of you, and the stability of the macroeconomy—and the robustness of the macroeconomy—is very closely linked to the stability and robustness of the financial system.

So for us, this is really important. We have roles in both the financial sector, but also from a macroeconomic perspective, and so this is really an intersection that helps us bring together some of the best thinkers in the space—the most creative folks—to help us really get our mind around these issues and think about how we should be positioning ourselves as an institution to be a good steward of the economy and also the financial sector.

Now the theme is "Mapping the Financial Frontier," and the overarching question here is: What does the next decade hold? Now, this is a big question, and if I could answer this in a day and a half I'd probably have a different job, but we're going to start the conversation, at least, here and hopefully it will continue over the next couple of days and then the next several months and years.

Now, for today we have a really interesting program. This morning's sessions, we're going to explore the implications to the financial system of emerging technologies such as blockchain—and I know at our Bank, we've been having lots of conversation about blockchain. What is it, what its potential can be—and it has been touted as sort of the next "thing" that will become the future, and so we wanted to explore that. We have a group of folks who are going to help us demystify what the blockchain is and really understand its potential and also limitations.

Then in the afternoon, we will turn to the economics and regulation of data. As you know, today we can harvest and analyze more data faster than ever before, and this can definitely help businesses and firms be more productive and more creative, but it also brings some risks around privacy and other things. And so the conversation this afternoon will really identify, highlight and grapple with those tradeoffs.

Then later tonight, as I'm sure you're all aware, we have the privilege of having Chair Jay Powell here. He is going to discuss his views on the tensions between financial innovation and financial system stability, and I think that will be quite interesting. It's always good to hear from the chairman, and I think you will enjoy that as well. So hopefully this will be a very interesting day for you—very thought provoking—and at this point, I will just turn this over. So Warren, I think the stage is yours.

Warren Weber: Good morning, everyone, and welcome to a session on blockchains and trust. Before I introduce our distinguished panelists and start the session, I just want to follow up on something that was brought up last night in terms of jobs and whatever—and maybe get you a little more upset about blockchains. I've been reading a book by [Michael] Casey and [Paul] Vigna called The Truth Machine. It's about blockchains, and there's a quote in there that says, "If the future foreseen by this book comes to pass, we'll witness the biggest employment shake up the world has ever seen, and this time the most vulnerable jobs are..." and then they list the ones that were vulnerable in the past. But today the most vulnerable jobs will be the accountants, the bankers, the portfolio managers, the insurers, the title officers, the escrow agents, and the trustees. So I hope that's whet your appetite for us to talk about blockchains. So without further ado, let me introduce David Yermack, who is a professor of finance and business transformation at NYU Stern School of Business, and he's going to lead us off for about 20 minutes.

David Yermack: Thank you. I'm going to speak from the podium. Thank you all for coming out today, and thanks to the Atlanta Fed for inviting me. I'm here because in 2014, I offered the first university course in the world on cryptocurrency and blockchains, and this was in conjunction with a law professor at NYU. People thought we had more or less lost our minds. We overloaded our teaching in the expectation that the thing might be canceled and just kind of collapse under its own weight, and that first year we had 30 students, and then the second year we had 100, and the third year we had 230—at which point we split it into...we now offer it four times a year, in smaller sections. But the basic reason for this is very simple, which is jobs. The finance industry has in its possession now probably the biggest change in infrastructure and technology in about 700 years, and I've been asked to give a 20-minute introduction to show basically how this works—but also what some of the possibilities are, especially with attention to the wholesale banking market. I want to begin with a very simple observation that it's ironic we're sitting here in a central bank research conference talking about blockchains and cryptocurrency, and the reason is because the creator of this is a person named Satoshi Nakamoto. Nakamoto disappeared in 2011 and has never really been identified, who this person, or group of people, may have been. But one thing we know about Nakamoto is that he or she hated central banks. And in fact, the whole reason that this technology exists was an attempt to usurp the control that the central banks have over the economy.

So very famously, the first block of the bitcoin blockchain, which was created on January 3, 2009, it has the headline from that morning's Times of London, that the chancellor's on the brink of a second bailout for the British banking system. And this is there really to announce to the world that there is a better way and to throw down a challenge to the existing financial system that we are going to put you out of business if you don't adapt and recognize that there's a better way to do payments and transfers.

Nakamoto also had an active email and blogged, for a couple years at least, and some of the quotes have been harvested, and I've picked one out about central banks where Nakamoto wrote, "The root problem with conventional currency is all the trust that's required to make it work. The central bank must be trusted not to debase the currency, but the history of fiat currencies is full of breaches of that trust." So at various turns, Nakamoto really condemned central banks. And the timing of this, at the very bottom of the financial crisis in 2009, I think was no accident. But what has happened in quick order is that the legacy banking system, beginning with the central bank but also all the commercial banks, has recognized the value of the underlying technology and is rapidly moving to build it out in their own infrastructure.

So I picked this slide simply because it's a very recent story, that is just meant to be representative. It's about the central bank in Thailand using a blockchain to consider moving the fiat currency—the Thai baht—onto a national blockchain that would also integrate the banking system of domestic banks. I think almost every major central bank in the world is now undergoing this exercise, and it implicitly raises the question, not only: do we need these commercial banks going forward, but do we even need the central bank? And if these things still exist in 10 years, they may in fact look very different than they look today.

So I want to begin with just one slide to explain how a blockchain works, and what is the breakthrough and why are people so excited about this. Fundamentally, a blockchain is simply an accounting ledger where people enter transactions. And we've had accounting ledgers since the time of the ancient Egyptians. For the last 700 years, give or take, they've been organized with the principle of double-entry bookkeeping that everybody in the room probably learned in school. Double-entry bookkeeping simply means that for every debit there's a credit, there's a liability for every asset, and the security comes from the balancing of the balance sheet, that you have to enter two transactions that offset each other, and that requires a degree of internal consistency that makes fraud maybe a little bit harder.

But there is in fact plenty of fraud, that if you want to steal off of a ledger in double-entry bookkeeping, you just have to adjust both sides, and for Ernst and Young or PWC, or whoever the auditor is, it's very much a needle-in-the-haystack exercise to see if something has been tampered with. Now a blockchain is also a ledger, but it works very differently. You have along the top here three—I'm sorry, I keep pressing the wrong button. You have three transactions here, S(0), S(1), and S(2). They could be payments: Alice sends 10 [dollars] to Bob. They could also be things like birth certificates at the government registry, or they could be real estate titles—really any assets that you wanted to enter into a ledger.

And in each block here, there is simply one transaction that gets merged with the prior block, and typically these are coded in hash codes. So transaction one is the combination of the new data plus the inbound data from the last block, and this is called a blockchain, because each block of new data is linked and merged together with the old data from everything that had been entered up to that point. And what this does is really two things: one is that it makes it all but impossible to decrypt the blockchain and change the ledger, because two entries that are identical, even if S(1) equals S(2), the outbound blocks will not be identical because they are being merged with different inputs. And this is actually a very simple adjustment to the way that we keep track of data, but once you realize that it makes every transaction different in the chain, it's a profoundly more difficult problem for anyone who wants to hack or steal or somehow corrupt this ledger.

The second thing you get out of this is that if something does change—if a thief goes in and tries to change one of the earlier entries—it will throw off not only that entry, but every future entry in the ledger from that point forward, so anybody looking at the ledger would be able to tell right away that something was wrong, and exactly the point where the problems begin. In other words, it's really a self-auditing ledger, and the immediate implication is we really don't need auditors anymore once people start to use this. And that's not the only profession, but the list from Vigna and Casey I think was pretty much spot on, that many jobs in banking and finance are essentially people keeping track of lists of things, and transferring entries from one entry in the ledger to another.

This does this with great ease, and in a way that is completely transparent if somebody goes back and tampers with a ledger. So very famously, bitcoin has never been hacked in 10 years. It's by far the most secure financial ledger that's ever been created, and people have come to realize that there are many other records where the integrity of the data—not only in finance, but in health care, and vital statistics, immigration, border security, almost any aspect of government or business where data integrity is important—could probably benefit from this technology.

Now, this is a timeline that shows when the major players in the financial system made public announcements that they were doing some type of experimentation or beta testing or trial project involving blockchains, and you see that the huge cluster is in the fall of 2015—that there had been, prior to that, one or two groups of people in the New York Stock Exchange...in fact, one of the early recognitions that this could be used for stock markets and not just banks. But you see really a gold rush of companies, all piling in. This would have been basically seven years on. Nakamoto posted the first white paper about bitcoin in October of 2008, and it's really not until October of 2015, give or take, that industry reaches a consensus that this is a technology that can't be ignored.

So up until that point they had gone through the seven stages of ignorance, denial, ridicule, trivialization. But by 2015, they realized that there is an existential threat, and that if banks don't figure out a way to turn this technology in their favor, there's a huge risk that they won't really be needed in the future. And so you begin to see some pretty aggressive research and development taking place.

Now what we've come to realize—and I think this has taken a few years—is that the model that's used in bitcoin and ethereum and all of the famous decentralized currencies: this will not work very well in the financial system that we already have. So this is sometimes called the "public blockchain" or the "open blockchain," and it's open in the sense that anybody with a computer can opt into it, so any of you who wishes to be a node on the bitcoin network could actually go back to your hotel room and do it right now. There's no prequalification or barriers to entry or rules of admission, and the blocks are created in an open competition every 10 minutes, where people try to decrypt certain types of puzzles, and if you're the first person to do it you get a reward of 12 and a half new bitcoin—which, at current prices, is worth about $95,000. So many people try to do this.

But the system has issues, and probably the main problem is that it is too transparent and that everybody who is a node gets a copy of the ledger and can see what everybody else is doing. And this runs directly into really two distinct but critical issues: one is really a regulatory issue, that almost every country has laws and rules about data privacy and protecting the identity and the behavior of your customers and so forth, and you would quickly violate those rules if you moved to an open blockchain, because everything is more or less laid open for people to look at.

And the other problem is really a marketing problem, that there are huge clienteles of potential customers who simply don't want their data...you know, the entire Swiss banking system is founded on this principle that privacy is valuable to people. And I think for young people, it's maybe different. But I think ultimately the exposure of the data is probably the number one reason that people don't wish to lead their whole lives on open, public blockchains.

Now, other issues are also important, and we're going to hear more about these in the next session. But it turns out that the competition to update this thing every 10 minutes on bitcoin, and the other blockchains are much faster: it's every 12 seconds on ethereum and so forth. But these are expensive—they take a lot of resources, they use energy that is equivalent to a small nation, and so forth. And so until there is more of a breakthrough in the consensus protocol, as it's called, I think the cost of mining is simply much too high to sustain this on a large scale.

We also have a problem of disagreements that occur on blockchains. So there is a competition every 10 minutes to create the next block. Sometimes there is a tie, or a near tie, where people disagree about who actually won, and this creates something called a "fork." And if you bought coffee at Starbucks, you may be on one branch of the fork but not the other, which is very unnerving to both the merchant and the customer, and so it takes time to resolve these things, and it's never the case that a transaction is approved with 100 percent finality. Instead, we can make statements that this is 99.99—you know, all but epsilon, it's probably in the books for good. But this is a very different model than we're used to. We have clearinghouses that certify with finality, the transactions are there forever, and I'm not sure we're ready to move to a market structure where things are expressed in probabilities instead of black-and-white "yes" and "no."

You also have forks where there are policy disagreements. If you really don't like the conduct of a market, you can propose a new form of the code and recruit volunteers from the network to run your version instead of the version that everybody else is running. This is called a hard fork, and it doesn't occur by accident but rather by deliberate usurpation. It's a little bit like Swiss democracy, where ordinary citizens can try to amend the federal constitution, and sometimes they do. But this can get very messy, to have schisms and multiple versions of the truth and so forth, so I think all of these have emerged over time as interesting problems for academics, but in practical terms they're real issues that limit how you might apply this in the real world.

So what we have moved to is something called the "permission blockchain," or the "permission ledger." This is a graph from the BIS [Bank for International Settlements] annual economic report, and at the top it shows essentially the open blockchain...I'm sorry, this is the legacy financial system where there is perhaps a clearinghouse in the middle and then everybody going against the clearinghouse, who is the one keeper of the master ledger. In an open blockchain, everybody can deal with each other and exchange data—that's the one down here. A permission blockchain puts a boundary around who's allowed in and who can see what aspects of the database, and in a permission blockchain you can quickly get past things like disputes about the rules. You can't really have a hard fork based on a disagreement, because there is one powerful gatekeeper who basically decides what software is going to run and if and when it's going to change.

I think it's fair to say that the permission blockchain is a hybrid solution between the traditional system and the open, free-for-all blockchain that was introduced by Nakamoto. It clearly has limitations, and above all the gatekeeper remains extremely powerful. The main thing Nakamoto was trying to do was get rid of the gatekeeper and have a completely decentralized system with no leadership and nobody in charge. But in a permission blockchain, whoever has the power to grant the permission is essentially the trusted third party, and the banks are all trying to appoint themselves as trusted third parties, or the head of such a system.

So I think the leading American bank at the moment in experimenting with blockchains would have to be JP Morgan. They've gotten a lot of publicity in the last three months, really, from launching the JPM Coin, which is meant for interbank settlements. They also have an international remittances platform that has grown from 75 to more than 200 banks that resembles the Ripple international network of cryptocurrency for remittances.

But the irony is that the head of the bank, Jamie Diamond, is deeply suspicious. Very famously in the fall of 2017, Diamond gave this—it can only be described as a rant, where in a conference kind of like this one he just unloaded on bitcoin. He said, "It's a fraud, people are going to get killed, it's a scam, it won't stand the test of time." You can read Diamond's quote carefully, and every sentence is actually untrue. The man simply didn't know what he was talking about. And he finished his remarks, saying that if he caught anyone within the bank dealing with this stuff they would be fired the next day. And then the newspaper was full of a long list of the blockchain projects that he apparently had no knowledge of, and so forth.

So this is kind of like the White House these days, where the old man has certain views and then nobody does...in fact, they're working on very different things. [laughter] And Bitcoin went up 30 percent in the next month after Diamond unloaded on it, at which point he walked back all his comments and said it's actually a very interesting technology. But JP Morgan is not unique in that most of the banks in fact have big internal R & D that is trying to scale this up in a permission format for use in the main financial system.

So I've written a paper that summarizes some of these in a fairly concise manner. I just want to talk about two of them, and then we're going to get on to the panel. I think by far the most significant for the banking industry are interbank settlements, for which a closed or permission blockchain with a specially created coin that can be exchanged among the participants on the blockchain, looks like a much better way to reconcile and do all the offsets that typically occur every day among major banks in the financial system.

So the JPMorgan coin addresses this market, but there's actually an older one sponsored by UBS, the Swiss bank, which is called the utility settlement coin. This one's been trialed for about four years, but I think the design of these is very similar and they're each trying to recruit banks to their platform. I think if I'm a bank I would probably join both of them. I'm not sure if these are going to be competitors or be mutually exclusive, but it's meant to essentially eliminate a lot of the duplicative recordkeeping, to eliminate the need for arbitration of disputes, and simply to improve the speed and accuracy of transfers between banks.

Now, how much money could you save? One consulting firm looked at this when UBS rolled out their coin. They said that the net expense in this system is $65–80 billion a year globally, and they think they can save about $20 billion a year by transitioning to this settlement coin. So you're talking about a savings of 25 to 30 percent. And this is very interesting. This would get the attention of any banker, and is obviously a first-order opportunity. And the way you would really think of this is getting rid of 25-30 percent of the jobs in the back office, the people who keep ledgers—which was exactly the quote that we heard at the very top of the session.

Another of what I think is probably the highest value opportunity is in shipping and logistics, especially import-export, cross-border shipments, and remittances. So this is another BIS diagram where they're showing the old way, where two people are trading with each other and each is relying on a bank either to lend money or to collect money—letters of credit and collateral are being posted, and so forth. But everybody is keeping their own ledger and is communicating only with the person who they have privity with.

If you can move to a system where there is one shared ledger, or distributed ledger—only one version of the truth—it eliminates the possibility for disputes and greatly speeds up the movement of goods, the release of collateral, and so forth. The opportunity here is vast because the amount of capital tied up in worldwide shipping runs into the trillions, and there's a pilot project undertaken by IBM and Maersk, the big container company in the port of Rotterdam, that has estimated that you could maybe speed this up by 50 percent—that for the time it takes to move inventory around the world, to get iron ore from Australia to China, the paperwork delays alone are slowing it down by a factor of two. And so if there is say 10 trillion [dollars] in capital tied up in this, you could recoup 5 trillion [dollars] from transitioning to such a system. I think the money is so big that this will happen without a doubt, it's simply a question of who's model and how quickly.

I've highlighted a project by a consortium called R3. R3 is based in New York, and they have recruited approximately 70 or 80 banks, insurance companies, a few stock exchanges, I think even a couple of government's central banks belong to R3. And they have a platform called Corda—which is essentially a template on which you can build a permission blockchain in which many people are now trialing these different platforms.

R3 is not the only game in town. There are various others, called Hyperledger, the Ethereum Enterprise Alliance, and many others—some of them focused on industries such as insurance and accounting. And it's a very exciting time. There's a lot up for grabs, and I think in 20 or 30 years all the financial data in the world will be kept on blockchains, but whether the same banks and audit firms will be keeping those ledgers is I think very much an open question.

And I think the real issue for the industry is: Of the technology companies, what role will Amazon, Google, Facebook, Alibaba, and so forth play in this market? Because I think they have the comparative advantage in this field, and if you look at what Alibaba has done to the banking system in China, I think very easily this could happen in this country with Google, Facebook, Apple—any of the big tech companies—playing a similar role of trying to become essentially the guardian of the data, the processor of transactions, the role that historically has been performed by the banks. So let me stop at this point, and we're now going to have some rebuttal and then audience Q and A.

Weber: Thank you very much, David—a very great start. So far we've only had one question, which I find rather surprising from this group, given what we saw last night. So I'm supposed to remind you that you can post your questions by going to pigeonhole.at—and if you need the password, it is caps: F-E-D-F-M-C. So I expect to see more questions popping up. They're all going to pop up here for me. You're going to have to put up with my choosing which ones get answered, but that's why I'm getting paid the big bucks to moderate this session. Our first panelist is—Keith is going first. Okay—Keith Pritchard, and he's with JDX Consulting.

Keith Pritchard: So I won't walk all the way over there, I'll just stand here. So just a 30-second background on myself: I've got 30 years in financial services technology—24 of those working for Mr. Diamond at JP Morgan, not directly for him. The last two years I was on secondment to the DTCC [Depository Trust & Clearing Corporation] to—if you remember the diagram with the sudden peak of interest around 2016, and there was DTCC in there. I was on secondment to DTCC to help them look at how they might use blockchain distributed ledgers, and you may or may not know that they're in the phase of testing a replacement for their credit derivatives process system, which supports the credit derivatives market, onto a distributed ledger platform. So that's my background.

I wouldn't disagree with anything that David said about the capabilities and the functional ability of distributed ledgers and blockchain. I would just want to highlight here one thing: distributed ledger and blockchain are two different terms. Distributed ledger is a way of making sure that everybody involved sees the same data. Blockchain is one way of achieving that, and that's something I'll come back to. But before we start or initiate any project that's looking to use a blockchain to solve a particular problem, I think there are always two questions we should be asking ourselves. First of all, is the capability that the blockchain brings: is that solving a problem that we have? So if it's solving a problem, great. But if we don't have the problem, why do we bother? And the second is: okay, great, distributed ledgers, blockchains bring that capability. Is there an existing technology that also brings that capability, and if there is: maybe distributed ledgers and blockchains should just be part of our toolkit. So we have it there, as technologists. We have it there in our toolkit, and when a problem like that comes along, we sort of do a beauty contest and say, "Okay, which of these tools best solves that problem?"

So those are the two questions we should always ask about blockchain and distributed ledgers before we launch any project. And as David said, the two key...if you had to pick out the two key things—okay, what is blockchain/distributed ledgers? What are the two key bits of functionality they bring? I would sort of classify them this way: The first is the distributed ledger part—so, making sure that all the nodes of the ledger, all the participants, see the same data. So, for example, if we created a distributed ledger that supported the processing of interest rate derivatives, and we were all participants in that market, and we were all members of that network: the distributed ledger element of the capability would mean that we all saw the same data at the same time.

So going back to those two questions I said we should always ask. The first of those: Is this solving a problem we have? Great—we all see the same data, but is that a problem we have today? Absolutely, yes—and similar to some of the stuff that David presented: when I was at JP Morgan on secondment to DTCC, I worked with some of the other big investment banks, and we looked at just in the credit derivatives market, how much of a bank's expenditure was to do with the technical and operational processing of reconciliations and resolving the reconciliations.

And it was—I can't remember the exact number, but it was something like 2.8 billion [dollars], 3 billion [dollars], so 50-odd percent of their processing costs were to do with reconciliations. It's absolutely making sure that if we were on this distributed ledger network for derivatives that we all saw the same data—that's a problem we have. So that gets a big tick.

Second question to ask of that particular functionality: are there other ways, other technology, existing technology, that can solve it? Again: absolutely, yes. One of the first things I did 25 years ago when derivatives were just becoming regular traded instruments was to build a platform for JP Morgan Global Bank. And we had a central database, and then we simply spat out all the transactions to each local office as a way of keeping a distributed ledger. Now, there are problems with that, but there are other technical ways of achieving a distributed ledger, you don't have to tie the two together. To get a distributed ledger, you don't have to have the blockchain to do it.

So that's the first part, the distributed data. I'm actually a big fan of that, and think is a real benefit to certainly the investment banking industry. The second capability actually comes to the trust question, so that's around distributing functionality. So the way of thinking about that, you can go back to this distributed ledger that we're all members of. Think about an essential counterparty, or a clearinghouse today: for cleared interest rate derivatives, that clearinghouse calculates the cash flows, calculates the margin calls, and it tells us all, it says, "Here's the answer." And in fact, it moves the money across our accounts, so that's a model way of a central, trusted service provider.

What the capability of distributing functionality to the nodes, the ledger does, is that it allows us to avoid having that central, trusted counterparty. So go back to our two questions: Is trust in central service providers a problem we have today? And again, I'm going to limit my comment on this to the investment banking world: not really. Is there anybody in investment banking? So not everybody trusts the DTCC, or don't really trust the London Clearinghouse, or CLS and their settlement—don't really trust them.

Now, we will moan at them about being too expensive, too slow, but—trust. So that's one area where certainly in investment banking, I don't think trust is an issue. So, yes you can do it this way—and then I've got to think about some of the downsides. Again, this interest rate derivatives, distributed ledger that we're all members of: we can have David running his trusted service provider that does all that calculation for us, or we can all do it.

Now, so I've spent my career building and maintaining derivatives processing systems. They're complicated beasts, with a ton of technology, a ton of code, a ton of servers you require. We would all have to have them, we would all have to have them if we had a model where we distributed functionality, we all do our processing, then send it back into this process you call "consensus." I think people have called it "consensus" because the word "reconciliation" gets a bad rap. Consensus is reconciliation. So we send our results saying, "Here's what I got, do we all agree?" Yes, we agree. Right now we'll commit that transaction to the database.

So not only do we have the issue in terms of distributing functionality there, it's solving a problem I don't think we really have. It actually makes things worse. It makes it a lot more expensive, and a lot slower. So for me, distributing functionality in the form of smart contracts and—if you want to grab me later, I can have a diatribe about smart contracts—I think is completely pointless.

So anyway, to summarize: we need to think of blockchain and distributed ledgers as a group of capabilities, not just one thing. It's not "either you get it or you don't." We should look at those capabilities and say, "Is this solving a problem?" I think the data distribution, the capability, is very, very useful, very powerful, and the examples that David talked to actually are all about data distribution: the Maersk example, the trade finance example, are all about people seeing the same data. The functionality that allows you to distribute processing power to the nodes I think is completely a technical capability looking for a solution, and really—and again, certainly I'll talk about investment banking space—I have not come across yet a sensible use case where that's applicable.

I think I'm out of time, so I'll stop there. I just want to give you a flavor of where I'm coming from, which is the realist point: it's neither 100 percent good, 100 percent bad, it's somewhere in the middle. I'm sort of maybe two-thirds nearer the "it's more smoke and mirrors than reality," but I do think there is some benefits in there.

Weber: Thank you, Keith, thank you. And our second panelist, or third one, is Martin Walker, and he is the director of banking and finance for the Center for Evidence-Based Management. By the way, questions are starting to come in, so...

Martin Walker: I'm not going to go through my background, because it's actually very similar to Keith's. So there are an awful lot of opinions about blockchain. Anyone who has had anything to do with it has got an opinion. But I think the key thing that almost everyone who's come across blockchain will agree about, that the claims about blockchain are absolutely extraordinary, and to understand how valid these claims are requires a much deeper dive into the how: how, specifically, does blockchain and related technologies actually solve or improve the challenges we face in building almost any kind of IT system. So we're going to talk about it, and actually just go back to some of those basic fundamental challenges. I'm sure as I go through them, many of them will seem very familiar to you.

So one of the primal challenges in any IT system is the GIGO problem: garbage in, garbage out. If you put garbage into a system, if you're lucky you'll get garbage out. If you're unlucky, the whole system falls over. Another challenge: implementing the right programmatic logic in the system. How do you make the system actually do what you really want it to do? And everyone has experience of bugs in systems, or gaps in systems. And in times when technology isn't quite so fashionable, people have lots of negative opinions about their IT systems and their vendors.

Another fundamental challenge is: how do you actually model and structure data? Systems run off data, and you need to actually structure that data in such a way that the system understands what to do with it, and people actually understand what that data is actually representing. Another fundamental challenge, which is common in many systems, particularly in areas like trade finance, supply chain management, is: how do I make sure that the digital representation of reality actually matches onto the physical world? And almost every system has that challenge, whether you're talking about people, or whether you're talking about pork bellies.

Now, a big challenge—particularly in financial services—is actually making sure that the systems that the various parties to a transaction actually have a consistent view of the truth. So we're talking about breaks, exceptions, et cetera. That's actually a big challenge, as Keith has mentioned, and Professor Yermack mentioned that actually as a big source of costs. But it's not just between the organizations. If you look at the infrastructure of an investment bank, there are dozens of systems who actually receive the same data about trades, for instance, and maintaining consistency is a big challenge.

And the final challenge is security—huge topic, but the fundamentals are: How do we make sure only the right people get to see the data? How do we make sure only the right people get to operate systems? And how do we make sure that no one actually interferes with the programming logic or exploits the bugs? Any of those familiar? Familiar challenges? So I think many of you will actually be thinking, "These problems aren't unique to IT systems. These are problems in creating any kind of business process, any kind of organizational structure." And that's very true. A lot of the problems in building good IT systems are nothing whatsoever to do with the technology. They're all about people, socioeconomic factors within organizations and between organizations. So how do the various forms of blockchain...sorry, I'm just going to get some water.

How do the various forms of blockchain actually address these challenges, and how does that compare to what we do at the moment? So for the GIGO problem: how do we make sure the right data goes into systems? Well, we train people, we put validation into systems where the input is coming in from another system, we try to make sure that feeds in correctly. Now, in the blockchain world—and I've seen this in many pilots—there's this assumption that, "I've got a blockchain, it's immutable, and the data is just right—it's there." I've seen this over and over again. But to actually have a blockchain-based system, you still have this fundamental challenge: How do I make sure the data going to the system is correct? It doesn't matter if it's immutable if it's wrong.

Programming logic: how do we get that programming logic right? You may have heard people talk about smart contracts—would I do it with a smart contract? It would be fine. Smart contracts are just programs, written by the same kind of people, in the same kind of languages, with the same challenges as writing any other kind of program. And what you actually find with some of the blockchain-based technologies is it can actually be harder to program business logic onto a blockchain-based system than conventional systems.

So, data models: we approach this in similar kind of ways. We have people who are trained, who studied computer science, et cetera. We test the models. We also spend a lot of time, when it comes to actual data models, in talking about and agreeing on standards about how to represent things. If we...the entire financial system, global financial system, actually relies on the fact that people have agreed to standards and data models in things like SWIFT messaging.

So you look at blockchain. I take the supply chain management system, I've digitized it, I've put it on the blockchain. The fundamental thing you have done in that digitization is you've come up with data models. Now, doing that in isolation, doing that in a pilot—if you don't actually get other people in your industry to agree and to participate, how are you going to model stuff? It's not going anywhere—you still have exactly the same challenges.

So, consistency between systems: now, in capital markets, where I spent most of my career, particularly in processing trades, derivatives, effects, repo, et cetera, I did some very, very detailed analysis, I looked at exception rates, I looked at breaks, and the most effective way to get consistency and to reduce those breaks was actually more centralization. If you look at capital markets, the trades which flow front to back through systems with the least problems are actually the trades people have executed on a centralized system, an exchange trading platform.

Now, blockchain-type solutions, as my fellow panelists have touched on, have a somewhat different approach. It's, you have people operating the same programming logic on the same pool of data, and ideally getting the same consistent results. And that is an approach that can work, but it's not a panacea because there are a few issues there—having the same programming logic replicated many times, operating on the same pool of data, is actually incredibly inefficient. Having a distributed ledger and blockchain actually goes in opposite direction to what we call "distributed systems," where different systems do different aspects of processing. That's basically the concept the internet is based on.

So another aspect of that is, even if you actually want to adopt that model—everyone runs the same system, looking at the same data pool—you don't actually have to have blockchain-specific technologies to do that. So, security...a common approach to security is, well, you lock up your system and you try to stop bad people getting to it. You also try to limit the access of legitimate users to the areas they should have. If you worked in any kind of large organization, you review permissions to the system on a regular basis. You pay people to try to actually test the security of your system. One of the most fundamental ways of actually making sure your system is secure, which is one failure to do [and] which causes a great deal of actual hacking, is keeping your software up to date.

Now, blockchain-based technologies—you talked about cryptography, encryption—it comes with this aura of, "This must be secure, this must be really good." But when you actually look at where those technologies have come from in terms of general approach, it's actually come from a really, really bad direction in terms of actually creating secure systems. It's come from the view of you distribute data to everyone. It's come from the perspective that once something's done—you're going back to bitcoin, and Satoshi's paper. He specifically designed a system which was irreversible. This is one of the reasons why cryptocurrency exchanges are persistently being hacked. You need one piece of information, you can drain the account, and there's nothing you can do to get the actual money back.

So blockchain-type technologies have evolved away from that, but typically they have evolved away by dropping blockchain-type features. Most of the enterprise-type blockchain systems do not actually distribute data to everyone participating anymore—particularly a few things like Fabric, Corda—and generally, "security" has actually meant moving further away from the blockchain-type model.

So just in conclusion: Are people building plausible and interesting systems, particularly in financial markets, using blockchain-type technology? Yes, they are—in part using blockchain-type technology. Is the blockchain the magical ingredient there? No. Typically the magic ingredient is smart people having smart ideas. Are people starting to build generic systems based on blockchain, which look like they're going in interesting directions? So we mentioned Corda, Fabric, Ethereum, etc. Yes, they are heading in interesting directions, but typically they're heading in interesting directions by dropping more of those blockchain-type features. So Corda, that was mentioned earlier—it doesn't even have blocks in it anymore.

So, do I think, on a personal level, the level of hype around blockchain distributed ledger is a problem? Yes, I genuinely do, because it feeds into this belief—and you get it about every decade—that IT is a form of magic, and you just have to repeat the magic incantations—whether it's blockchain, AI, machine learning—and everything is going to get solved, as opposed to, you actually have to think through problems and solve them. But worst of all, I think it distracts from the genuinely extraordinary: computers and software are extraordinary things that will let you do amazing things. And they have been extraordinary not just the last two or three years, not just during the dot-com era, but since their invention. And they have been progressively changing the world since their invention. And all that really holds us back from actually doing ever more wonderful things with computers and software is our capacity to exercise our intelligence, including our critical reasoning to question whether this is the right solution, and our imagination. And that's what I have to say.

Weber: Thank you. I just have a couple of comments, and then I'll open up Pigeonhole to your questions. One is, I think what this discussion has made clear is that blockchains really make the issues of trust and identity and fraud different from the ones that would be in, let's say the conventional type of accounting systems or financial systems that have been in place, and I'm wondering if maybe regulators ought to start thinking about those issues differently because of blockchains being out there.

The second thing is, David I really liked your discussion of trade finance, and the difficulties, and all three panelists brought up the "consistency of data" issue. And I'll throw out sort of another topic that's blockchain-related because it relies on blockchain, but I think something that you're going to see coming in the future is what's called "triple-entry accounting" or "triple-entry bookkeeping," which—at a very, very basic level, that's about all I know about it—is: you have an asset on your books but it's a liability on somebody else's books, and how do you make sure that what you think the asset actually is, is what the other person actually thinks the liability is. And so this triple-entry accounting, as I understand it, is: you, the person with the asset sends the information to a blockchain, a centralized person on a blockchain, and so does the liability person, and only when they both agree is it really kind of made or put as a permanent thing on a blockchain.

So I think there is some interest in that. I don't know how big it is, but just to throw out an idea of something that may also be coming down the road. Maybe we'll be seeing a future conference on triple-entry accounting. So, also I think I'm a little bit less skeptical of what David in his paper called "public"—I would prefer to call them "permissionless"—blockchains in the financial services industry. I'm not sure I share all of his skepticism about them. I'm a little bit concerned that if you go to the permission blockchains, what you're doing is sort of locking in the—I don't want to call them "monopoly rents," but they're close to what financial institutions have now, and maybe you'd be blocking some potential innovations that could come about because you let more people get in on the action. But those are just some wild thoughts I have. In terms of questions, here's one I want to throw out—but first of all: do you have other comments that you want to make on each other's presentations?

Yermack: Quite a few, but it might be better... [laughter]

Weber: Okay, we'll go to the questions with that, all right. So, here's one question, which is, since there—well, everybody can read it, so any panelists want to comment on that?

Yermack: This is a very interesting issue that...there are billions of unbanked people in the world, and you know it's really in the underdeveloped countries of the Southern Hemisphere, and you have the opportunity to use this technology to draw them into a peer-to-peer financial system without the need to qualify them for accounts and to pass the anti–money laundering rules and so forth—

Weber: I'm going to stop you for one second, because there's a second one that is very close, and so maybe we can combine the two of them. It all has to do with kind of what this may do in terms of restricting access to the banking system.

Yermack: There has really been huge interest, with groups like the United Nations becoming involved and using this technology as a platform to draw in the unbanked and get more of the world on platforms of financial services, the great success of this has been M-Pesa, in Kenya, which is used by I think 87 percent of the population. It's a telecom-based payments platform, and it carries at this point more than half of the GDP [gross domestic product] of the country.

I was asked last summer to speak in Mauritius at an African economics research group about the success of this technology in exactly this setting, and after a fair amount of research on a cross-country basis, it turns out that what's critical is the electrical grid. And this shouldn't come as a surprise, that fintech, in the end, needs access to electricity and even more to data networks and so forth, and to the extent that countries are unable to develop them, this technology simply can't be used until you can plug it in and boot up a computer and so forth.

So there's a huge need for infrastructure spending, and this may ironically become even more of a barrier, given the spirit of the second question, that there's a pretty big capital requirement to scale up in this industry. And maybe outside NGOs [nongovernmental organizations], maybe donors and so forth can help create this in developing economies, but I think it's been a much harder slog, and not nearly as successful as people had hoped. There are a lot of utopian beliefs about financial inclusion tied to blockchains, but I think infrastructure has to precede that, and that's going to be expensive.

Weber: Given the earlier discussion, we should point out that M-Pesa, which has been very, very impressive, does not have a blockchain but is very, very centralized—and also has the back-up of the Kenyan Central Bank behind it, so it's a government-insured system, in a sense.

Pritchard: I'm going to make a sort of related comment to this, I guess, which is—I'll be interested, David, in your thoughts on this. If you look at the bitcoin network—and as David explained, you rely on what they call the "miners" to try and put the blocks together and obviously that takes a lot of electricity, a lot of effort, and sometimes you're trying to create a block, and somebody else does it before you so you don't get any reward. And the idea is that the miners get paid a little bit of bitcoin.

And at the start, when the bitcoin network was created, a whole bunch of bitcoins were set aside as a pool to pay the miners, and every now and then the reward you get for a certain amount of work goes down—because obviously, that pool of money for the miners goes down and it can't ever run out, so every now and then they half the reward. So at some point, the cost of mining is going to be more than the reward you get in bitcoins. So, why would people mine? And I was just thinking this the other day, thinking, "Well, why would anybody mine?" Well, you need these miners, you need the people to create the blocks to keep the infrastructure going.

Yermack: There is an answer to that, which is: the miner gets two rewards. There's the organic reward that you described, but individual transactions also have tips that a miner gets if they put the transaction into the block. And the reason the reward is halved every four years is to try to transition to an entire auction fee-driven system, where you would actually pay to cut the line, for faster processing. And Nakamoto left a 130-year interval for this to occur.

Walker: If I could just say something about financial inclusion: so even though I've spent most of my life in technology, I do try to look at things from the human economic perspective. If you look at what some of the key things that drive financial exclusion, one of the fundamentals is actually identifying people, so if you look at one of the biggest things, which is driving increased financial inclusion in India—so they have a centralized system, a centralized ID system, which allows you to identify people, which is one of the fundamental costs in KYC. If you look at M-Pesa, owning a cell phone in Kenya, and many other African countries: you can't just go out and buy one, you have to identify yourself. So the KYC is done.

So, identifying people: nothing to do with blockchain. The other factor is: What are the actual incentives for financial institutions to actually service the financially excluded? Often the financially excluded are excluded because they don't actually have much money, so how do you tackle it from the actual incentive for institutions to actually service people who are excluded? And it's not just in the developing world. Even in England, you have poorer segments of societies who don't have bank accounts, who don't have credit history, the banks don't necessarily want them. And that's a real problem. More cost effective, incentives, but also cheap and effective ways of identifying who you are, are some of the big tours, regardless of the technology, to drive more financial inclusion.

Weber: Excellent point—how one establishes one's identity is very critical, especially in undeveloped countries. What kind of records can you put in place that can't be changed by a government or some person? And we all know that identity is critical in terms of getting credit, and to the extent that a blockchain-type technology could help establish identity, a trusted form of identity, may unlock all kinds of credit opportunities in development in those countries.

Yermack: I think we can't let this go, because you said you don't need a blockchain but this has actually been held up by many people as a textbook case where a blockchain would be of critical value, because it's a ledger that can't be rewritten after the fact. There's a massive problem of identity theft—certainly in my country—and there are many cases where refugees have their documents taken at the border by oppressive governments and so forth. And the whole point of a blockchain ledger is that once the identity is put onto the blockchain, it can't be changed, and the United Nations is very interested in this for stateless refugees, of whom you know there are millions around the world.

But more generally, the Indian Aadhaar project is a biometric ID, generally based on your fingerprint, and there has been basically undermining of the database and hacking since it was created, which has been a major deterrent to its scaling up in use. And a blockchain, as an unhackable ledger, addresses this problem.

Walker: I think this goes back to some of my initial points about garbage in, garbage out, and how do you maintain that link between data—a representation of something—and people. Any system, if it's controlled by a disreputable party, is not trustworthy, so when people talk...for instance, people used to talk four years ago about Honduras's land registry—I think it's Honduras. And if—and this is not meant as a comment about Honduras, because that never went anywhere—but if you're in a country which does not have the rule of law, and people have access to the authority, it doesn't matter who is running the ledger or whatever is on a blockchain, because there is recording information, and actually power to action that information about who you are and what you have. And these are two different things, and they both need to work together to actually support things like financial inclusion.

Weber: If the information is being misrecorded, then it doesn't matter that it's immutable.

Walker: Or if the authority just doesn't care what's on your blockchain.

Weber: Okay, I'm going to go to this one because I think it's an issue that ought to be discussed. Did it come up? No? There we go. I think this is an issue that a lot of people do not completely understand.

Yermack: The thefts of bitcoin, which in fact occur with some regularity, typically come from third-party companies that are called "wallet providers," where they will hang on to your bitcoin and secure them typically for a fee of 1 percent per year. And it is these wallet companies that have basically been drained of bitcoin because they are too casual with the passwords, or the—as they're called more formally—the "private keys." When bitcoins are stolen, we can see exactly where they go and it's not the network itself that gets hacked it's the people who are basically very sloppy about safeguarding their coins, and this is a completely different issue.

Weber: And a lot of the, if you talk about "it's the wallets"...some of them, it's the hard wallets of the exchanges that are getting hacked, and there was the big hack of Ethereum out of I forget which exchange, but then that caused one of the hard forks for Ethereum.

Walker: That wasn't a hack.

Yermack: I'm not sure using the word "hack" is right.

Walker: That was exploiting a bug, in a smart contract.

Yermack: Someone with a little smarter contract, that competed with the first.

Weber: Right. And Ethereum basically then said, "We'll hard fork and undo that, because of the bug."

Walker: Well, Dave is quite right. The losses are actually where the bitcoin network interacts with the real world. The thing is, as I touched on—it's one fundamental design flaw in bitcoin, which was deliberately put in there, which makes it an incredibly insecure system, which is irreversibility. Satoshi Nakamoto kept going on about irreversibility. Just imagine: you walk into the vault, there's all the bitcoins. I just need one piece of information, and a smart phone. It's like, I throw the gold out the window, I can see it's out the window, but there's nothing I can do to get it back. You just need that private key, and it's gone, and because it's a decentralized network—and "decentralized" is another word for unaccountable—there's nothing you can do.

Yermack: I think, again, greatly misunderstood. The point of irreversibility, and some of the other aspects of blockchain, are really to make people behave better because they know that once something is entered into the ledger it's there forever. I don't keep my credit card receipts. I'm very sloppy about swiping my Visa and then getting the bill at the end of the month—because I know that there's a dispute resolution transaction, and I'm paying 2 percent for that, with every transaction.

Walker: If you go back to Satoshi's paper, he said he wants to build an irreversible system because he thought reversibility, and the ability to cancel things, was a big source of costs.

Yermack: It adds costs, and it adds irresponsible behavior, too.

Walker: It adds that, also, but if you look for instance at the Bank of Bangladesh fraud, which was around a SWIFT network: most of that money was actually blocked, because we have systems where you can stop things and cancel it and reverse it.

Yermack: It was blocked because there was a typo that one person in New York entered after they had sent 81 million [dollars] to a casino in the Philippines.

Walker: No, it was the money that after concealed plans most of that sort of disappeared, but the fundamental thing was: you can block transactions, and you can reverse them as long as the money is still in the destination bank.

Yermack: And the bank found out a month later that it been robbed when it read about it in the newspaper.

Walker: But this key point is: reversibility, the ability to cancel things—most of that money actually got stopped from actually ending up with the crooks. There are still flaws in the system, but with systems which are close to the original bitcoin, you can't. It's designed not to be blocked, not to be canceled, not to be reversed. That's a design feature.

Weber: Okay, let's try this one. It's...

Yermack: Yes, this is a really good one. I've had the pleasure of getting to know the creators of the first blockchain in 1991—these are cryptographers from Bell Labs, Stuart Haber and Scott Stornetta—they have started something called the Blockchain Corporate Governance Foundation and recruited me as a collaborator. IBM is putting up some of the money. The first thing we're doing is going to the asset management industry and trying to get commitments from money managers not to backdate...there are all kinds of shenanigans in asset management, and it's private capital, the hedge funds, the private equity funds, that will merge failing funds together, that will backdate investments, that will do window dressing at the turn of the quarter.

It's much harder to do this in open mutual funds, because there is the accountability of the redemption, that every day people can withdraw money. But in private, opaque capital there are all kinds of reporting problems. And if you had a blockchain system that entered every transaction irreversibly, you would take away a huge moral hazard problem in the asset management industry, where people cook the books and there's no way to tell if they're doing it. So I think this is a great question. It's in fact one of the very highest-value opportunities to apply the technology.

Pritchard: I think I agree with what you said, Dave, but one of the practical things to think about with blockchain is, as David was describing in his presentation, each transaction is a little block that's added on, and you get this long chain. And it brings all that transparency and addresses this problem as you describe. But then we come down to some of the practical things about, how would you use it? So again, I'm thinking back to my experience around derivatives—again, quoting numbers from JP Morgan, when I was there last we had 2.4 million derivatives and 50,000 bits of activity every day, so that chain suddenly becomes really, really long, with big, chunky data—lots and lots of transactions. How do you run a report—a report which said, "All right, show me all the trades that JP Morgan made, or Goldman Sachs" would take forever to run, so actually what a lot of these platforms have—certainly a lot of the Ethereum-based ones have—is something called an "enterprise database" that sits alongside the chain.

So they recognize that actually I can't do all the things on the chain, because it's a linear structure that I need to do for reporting, so they have a relational database, in the case of a couple of Ethereum-based solutions I've looked at, sat alongside. You've then immediately...and I don't like that as an answer, for exactly the point you raised: you've then immediately got away from that the fact you can't change it, because you're now actually doing all the reporting off this side database.

Walker: It's the same draft report of the database hanging off the side, and all they record is the hashes.

Yermack: Even on bitcoin we now have something called the lightning network, which is a bunch of off-chain transactions that are periodically reconciled, but you are reintroducing the trusted third party, and it's perfectly fine for the banks to say, "Oh, we're very trusted" but you know the whole point of this was to remove that as an issue in the whole thing.

Pritchard: Which is my concern with this. I actually get that a bitcoin-type network can work and bring all the things you talk about. It's actually relatively straightforward, it's not doing too much complex functionality. If you look at something like derivatives, or at any other complex transactions we have, you start getting into all these practical problems. I don't think this type of technology actually, in its fully-blown distributing functionality—no centralization—actually works for anything much more complex than moving an asset from A to B. But I then take your point that that actually can be very valuable in itself.

Yermack: Yes, and it's still early days. We are at a primitive stage and simply the increasing velocity of computer processing will solve some of these problems, but some really need design breakthroughs that have not yet occurred. And I think, like a lot of new technology, some of the value here is agenda setting: that when you look at it through the lens of distributed ledger that's irreversible and so forth, you start to realize that there are design flaws in the general financial system that you hadn't really viewed as points of vulnerability. But we're years away from...

Walker: To make this work, however you do it, you fundamentally need to get the participants to agree to put the data somewhere—which is the hard bit. If you're talking about an enterprise-type blockchain, more often public. Someone's got to run that, so you're basically getting back to a degree of centralization. So they sound like the big challenges, rather than the blockchain bit. It's: how do we get people to do this, and who do we get to run it? As opposed to, "We're going to put it on a blockchain..."

Yermack: Many of the incumbents naturally look to...the DTCC is a great example that we were talking about before. That's what they do now, and they see this as a way to consolidate and entrench themselves further, and maybe that even makes the most sense. But you're absolutely right, in that the data concentration, and the person who controls the data being able to charge monopolist fees and to discriminate against who has access to the platform. All of these are already big problems, and they may become even bigger problems.

Walker: Well, yes—I totally agree with that. And it's interesting that whether you do this on a blockchain or centralized system, you're still fundamentally trying to meet the same challenges of actually how to create utility. And you can see a cycle, particularly across financial markets, of how do utilities get created? And there are many areas where there isn't standardization, there isn't utility, because the participants, it wasn't in their interest to do that.

And if you look back to the 1960s, the crisis of Wall Street—actually, dematerialization that birthed the DTCC—that required a huge push from regulators, because the brokers weren't signing up for that. And then you also see this—again, it's not technology, but there's an economic cycle around utilities as well that...let's say whoever gets together to create a utility is successful and makes money. Then it gets into people's minds that, "Hey, we can sell this utility and we can make even more money." And then the utility starts operating in their own interests. So there are some very interesting cycles around trying to achieve this, and building utilities, but the big drivers are actually economic and legislative rather than technical.

Weber: Let's follow that one up with this question—did it come up?

Yermack: You know, you need a credible regulator. [laughter] It moves the problem. I'll just leave it at that.

Walker: I think if you are building any kind of permissioned ledger, for whatever function shared between parties, how that is actually built—the technology—should kind of be material to the regulator. It should actually come down to what it actual outcomes: Is it secure? Is there someone responsible for running it—whether it's DLT [distributed ledger technology] or not, particularly if it's permissioned. It's kind of immaterial, or it should be from the regulator perspective, as long as they are looking at those kind of things which are actually set out in the BIS, for instance, in the principles for market infrastructure. There are some very good principles there, but it's DLT. I don't think if it's permissioned, I don't think it needs to be regulated in any specific way compared to other infrastructure.

Yermack: We spend an enormous amount of time on this issue in my course, because I teach with a law professor and there's an emerging practice area called "blockchain law," which is growing very quickly. There is a very short-handed phrase that "code equals law," that things that we mediate with arbitration or litigation can instead be delegated to algorithms and protocols. But when these things run off the rails, lawsuits are filed. And I think what's happening in this country right now is that the regulators are taking the laws that we already have, which are mostly from 1933 or 1934, or even 1914, and trying to fit a lot of this technology into case law that predates the invention of the computer. The discussion of the Howey Test from the 1940s is how we should think about initial coin offerings—it's almost comical.

And I think you will have regulation as long as there are sovereign governments—that this is really not "if" but "how." And you have, I'm not sure I would call it indifference or apathy, but I think the regulators in this country are both playing for time—they want to understand this better before the rules really get written—but there's also a degree of competition, which is a little bit unattractive. In various times the commodities regulator, the securities regulator, the banking regulators and so forth, have all asserted control over Bitcoin. And the real truth is that you have cryptographic assets that are a new type of property and really demand new regulation, and how these laws are going to be written over the next couple of decades will be very interesting, probably the basis of competition between different nations. Singapore and Switzerland...

Weber: Certainly different regulatory agencies.

Yermack: Yes, it's a fast-changing area, but it can't be avoided and it's going to be quite interesting.

Weber: You see the SEC [Securities and Exchange Commission] right now struggling over whether security is given in new technologies.

Yermack: Yes, and this is fun to talk about in the university [laughter]. We love to argue about this. But in the real world it's a little bit different.

Pritchard: It's interesting because you could sort of throw this on its head and say maybe we won't get distributed ledger platforms—and again, I'm thinking about the world of complex processing for derivatives—that we won't get those unless the regulators force them. Now what I mean by that is if you think—go back to the DTCC's processing for posttrade credit derivatives: the only reason—which is a centralized utility that the industry uses—the only reason that came about was the regulator saying, "You guys are out of control with the way you process credit derivatives, you've got to go and do this, you've got to standardize it." And having been on numerous industry forums and panels and working groups where we've tried to get agreement amongst the investment banking industry, that is phenomenally difficult for so many vested interests that I have half a suspicion that the only way you would ever get a distributed ledger for any major investment banking functionality or market is if one of the regulators just said, "Look, guys, you've got to do it."

Weber: Okay, there we go. Anybody want to take that one?

Yermack: Yes, I don't think it's up to the people. [laughter] You know, when I first got interested in this was in the fall of 2013, the U.S. Senate held two days of hearings about bitcoin, and there was testimony from [former Fed chair Ben] Bernanke who said, "I think this is an interesting technology that has a role to play in the financial system." And for about a day, I was baffled. Why would the chairman of the Federal Reserve give this an endorsement? And then it's exactly this, I realized that you could collect every dollar of tax, you could defeat money laundering, and the central bank could really finally target monetary policy to regions of the country or specific account holders. I think this is really a fascinating area, the possibility for a central bank digital currency, but there's no question that it makes the central bank incredibly powerful.

Now, is there any way for the people to stop the Central Bank from creating this? I know that in the UK, Mark Carney had to go into Parliament and disavow a scheme that was laid out in a series of academic papers by the Bank of England because they saw him as a usurper. But this is a very interesting question for the future about whether we have central banks on blockchains, which makes them much more powerful.

Weber: But for CBDC—central bank digital currencies—they don't have to be on a blockchain.

Yermack: Yes, right, but that's the premise of the question.

Weber: That's the premise of this question, but I just think we ought to point out that when we start talking about the wholesale end for the central bank digital currencies, that's where I think the talk or the interest is in the blockchain. But I think when you're going to the retail level that their thinking is, it's going to be basically everybody has a checking account only it's with the Fed, and that is a very different issue. Some of the issues you raise are of course that, but it has nothing to do with a blockchain at all.

Walker: I think that it's a common thing to actually confound the central bank digital cash with blockchains. Blockchain is just one way to implement it—you can have a centralized solution. And I think that the point the question is getting to is valid, whatever method of implementation you use, because in many countries, physical cash is starting to disappear. I think I use actual cash in the UK once a week, maybe less, and you either rely entirely on commercial bank money—so the bank knows everything you've done—or you put in place some form of central bank digital cash, where again—whether it's in a blockchain or centralized system, it's going to leave a record—a record that isn't there if you're using physical notes.

Weber: So the only stuff right now that doesn't leave a physical record is that paper stuff.

Yermack: There's a lot of news just this week about the Indian demonetization, because the government appears to have been reelected despite this very controversial attempt to nullify the cash. But what Modi was worried about is pretty much the same thing that's the basis of this question: using cash to conceal black market activity, especially tax avoidance.

Weber: Right. Okay, we have time for at least...several more, okay. Let me...I'm really quite interested in a stablecoin, so...What do I see is the potential for stablecoin? First of all, what is stablecoin? It's a token, a digital currency, that is supposed to maintain a one-to-one exchange rate with the U.S. dollar, although there are now over 50 of these where they go to various different currencies, and some are actually with commodities. Why are they coming forth? My view—I would be interested in the reaction—is that right now most national fiat currencies are not available to purchase digital assets of any form, and are not...so if you want to buy an asset, somehow you have to transfer your dollars to an exchange, you may have to run it through bitcoin or some other currency, and then you can buy the digital asset. If you want to sell a digital asset, what you're going to get paid in is some form of digital currency, which if you then want to have dollars you have to go through the expensive and time consuming process of having it wire transferred back to your bank. What the stablecoin does, in my view, is allow you to hold the purchasing power of whatever you sold your asset for, to hold that in something that you think is going to maintain relatively constant purchasing power until you want to go ahead and buy something digital.

So that's what I think the future of these is, as we get more and more stuff out in the digital world, I see there will be more and more demand for these currencies. What I find amazing right now is that one of them—Tether, which is by far the biggest—seems to be maintaining its value even though it purportedly said, "We will maintain a USD [U.S. dollar] in some form, somewhere, against every Tether that is out there," and has recently announced that well, it's only really about 76 percent in USD, the rest of it is in some short-term stuff that was borrowed by the other part of the enterprise that is overseas, both of us bit-financed. But I can see I think a future for them, because I see more and more stuff going digitally, more and more assets being tokenized, more and more need to have the ability to buy the stuff. And yet, we all tend to still think in terms of our national currency—at least, I think most of us do—and you'd like to have that value be stable, so you think you know what you're going to be able to buy in the future.

Yermack: There are two ways to create a stablecoin: one is the reserve—and it's surprising, I completely agree with you that the Tether holding its value, I wouldn't have expected. And the other is a trading strategy, where you try to run a currency board, and this'll work about as well as Argentina in the 1990s.

Weber: There are some of these guys in my view are trying to be central banks with their own currency, and what they're doing is they have their token coin, and then their equivalent of government bonds that the Fed uses is their own second token, which is only going to have value if the first token has value.

Walker: There's no government on the other side of the transaction that's willing to accept these as payment of their taxes.

Weber: That's right, absolutely right.

Yermack: Let me just throw out that: what if the stablecoin is the Facebook dollar? It's not a bunch of kids, but it's a network with two and a half billion people on it, and all kinds of capital standing behind it. Is that something that the Federal Reserve would see as a competitor?

Weber: I'm not in the Federal Reserve any more, so don't ask me. [laughter]

Yermack: But we've been asking that question since the first lecture that we gave in our course back in 2014, and I think there are some very deep-pocketed technology companies that arguably can defend the parity of a dollar-based stablecoin quite well—and get the seigniorage revenue from issuing them. This may be a very good line of business for Google in a couple of years.

Weber: Let's take it back to: your deposit actually is a stablecoin—not that you can buy digital assets with it, but that's what demand deposits are, is really a stablecoin that's supposed to maintain a one-to-one. And there are many cases in history prior to deposit insurance when it didn't hold one-to-one, and what is doing it is quite deep pockets called the federal government. And if the pockets are deep enough, then I think they'll be able to do that, but there's always a question of: Will the people behind Facebook, will Zuckerberg basically say, "Oh, yes, I'm willing to throw several billion in there to keep the..." I don't know.

Walker: I'd like to make a couple points. So you talked about digital assets: virtually every financial asset is digital at the moment. What we're really talking about when you're talking about using stablecoins to purchase digital assets is purchasing cryptocurrencies, and tokens issued by initial coin offering. So stablecoins are basically just facilitating this whole cryptocurrency bubble, fundamentally, and if you look at stablecoins—the better ones—it's just another form of e-money. There's a segregated account in a reputable bank somewhere. The worst ones, you've got no idea what's really there. And I think that's not really a sustainable long-term thing, that these things exist to help sustain the cryptocurrency bubble, make it easier to trade cryptocurrencies, and there's a form of basically regulatory arbitrage as some regulators haven't got their head around the fact there's this new-ish form of e-money with a technical veneer but without the proper controls around it.

Weber: Anything else? Okay, that's...

Yermack: This is what we just were speaking about. I would venture a guess that the dollar is going to be the last central bank to go crypto, the export of the suitcase of $100 bills, very valuable franchise for the United States...

Weber: Oh, is it ever!

Yermack: And it's not going to be forfeited too easily. And so I think as long as there's demand for money laundering, and it's kind of one of the timeless aspects of a rich society, that there will be plenty of Russian gangsters and petrodollars and you know, whoever the demand is coming from—that's not going to stop, and I think the dollar is probably safe as long as we keep printing them on paper and exporting them. But if they go crypto, if it's all crypto money, anything goes.

Walker: Well, coming from a country that used to be the global reserve currency, which came from running large trade surpluses and funding a lot of battleships, I think there's a lot more fundamental things which would—fundamental economic trends—that would threaten the dollar's position as a global reserve currency than blockchain or crypto or any of that kind of stuff.

Yermack: And they are?

Walker: And they are.

Yermack: No, what are they? Because we've had trade deficits since the 1970s. Nobody really cares.

Walker: I know, it's amazing. You have an amazing economic model here. [laughter] But I think Britain also managed to go through that cycle. We had a lot of trade deficits, we fought a couple of very large wars we couldn't afford. Eventually it chips...

Weber: You had some outside help.

Walker: With a lot of outside help and a lot of outside funding.

Yermack: We just have to be the least worst country.

Walker: I totally agree. I think you've hit it on the nail. As long as you're the least worst, you'll be the global reserve currency. But those sort of long-term trends that put you in the position where if something better comes along, they've been going on for several decades now.

Weber: Okay, I threw up another one because I didn't know you were...I thought you were close, but...anyway.

Yermack: The answer to this question is "no," because you can make the cryptography more complicated, even faster than the progress in computing hardware. So you know, you bring your quantum computer. I can just make the hash codes that much more difficult for it to invert. The cryptographer will always win this battle head-to-head.

Pritchard: And I think there's another angle, which when you look at actually what quantum computing is it's not necessarily a speed-up for sort of linear solution providing. It deals better in multi-variable optimizations. Actually, it's not a particularly good technology or technique for solving cryptography problems anyway.

Weber: But I thought that one of the problems...one of the things that I had read that quantum computing could do would be to basically take your private key, and...

Yermack: You just have to complicate the encoding of that key.

Pritchard: But it's the type of problems that quantum computing is good at accelerating, and it's good at accelerating multivariable optimization-type problems, not cracking a code. It's actually just a linear processing power problem.

Walker: Just an interesting thought experiment: Let's just say a group of miners got 51 percent of the mining power, which they do occasionally, or you had a new technology and you took control of, say, the bitcoin ledger. There are all these thousands of distributed versions of that ledger with a version of the truth, so people can go back and check. Now the interesting question is, given what happens in the crazy crypto world, is if it would be obvious if someone is actually corrupting it and making things up. Now, what I think is interesting is, would anyone speculating in the crazy crypto world actually care that it's corrupted? Just like so many of the exchanges, and Tether, seem to be quite corrupt. Would they just keep pouring their money in? It doesn't matter, it's...

Yermack: I think they probably would, but... [laughter] You know, this question I worry about in the hands of a state actor who kept the computer secret. If [Russian president Vladimir] Putin showed up one day with such a computer that we didn't know about, you wouldn't have time to prepare to defend against it. And I think the motives wouldn't be financial, but it would be an act of geopolitical aggression. And I think that's really where the concern needs to lie, is about the motivations of people doing it for sabotage reason.

Weber: Okay, I think we have time for one more—let's... [laughter] I'm not allowed to edit these questions, all right? So I might have left out the last line.

Pritchard: I think you just put technologists to head up every company, every bank.

Yermack: I was going to say: executive edit, NYU Stern. We have programs that people could enroll in, so that's what we do is we educate people.

Walker: I think...touching on the points I raised in my talk, there is no point just educating people about blockchain unless you educate people about the fundamentals of how systems work. Otherwise you're still just encouraging magical thinking, and you do not have to have gone and earned a computer science degree or have been a programmer for years to actually understand: what are those fundamentals. And I think those fundamentals are something that many board members and senior managers would actually really benefit from learning today, regardless of whether after the 101 course it goes on to the advanced course, which is "Blockchain or Not?" But I think that's a really necessary thing in our society as an economy, is where everything actually really depends on information technology. It's very hard to do anything in any kind of firm without making changes to computer systems. But the decision makers don't really understand what's going on below the surface.

Weber: I think with that, we're a couple minutes early, but we started a couple minutes early, so I'm going to thank the panelists for I think a very stimulating and enlightening discussion. So thank you.