7. Technical Challenges

Key Takeaways

The following content is provided under a Creative Commons license. Your support will help MIT Open Courseware continue to offer high-quality educational resources for free. To make a donation or to view additional materials from hundreds of MIT courses, visit MIT Open Courseware at ocw.mit.edu. Thank you again for all being here. We're going to talk a little bit about the challenges of blockchain technology. Um I'm apologizing in advance. I'm supposed to be like across campus for a 4:00 meeting. So, I won't have much time right at the end to do the little wrap with uh students coming up. I will note also that if you want to come see me, I'm open to it. Next week's a great week, by the way, cuz I'm here all four or five days. Um but I don't have uh set office hours. Just email me. Uh copy Dylan, who's the new uh course administrator. Uh there was a swap out from Ryan. Or copy Talita or Sabrina or something. But uh just shoot me an email and then uh I'll set something up uh with you if you want to follow up either on your projects or it's a question about uh anything around blockchain. Um I also want to thank we don't usually have people here with jackets on, but um we have uh six or eight uh veterans who have served our country, and I thank you for your service. Uh they're here to observe us. I I don't know whether we'll scare them away or not, but uh thank you for for joining us. So, today's topics um are going to be around the Of course, we're going to go through the readings uh a little bit. Uh we don't have Larry Lessig, and it's a little bit more relaxed. So, I might be doing some cold calling. Uh uh if that's all right. Um I'm going to go back a little bit to the technical features in a quick wrap in two slides or three slides, but I just want to do that as the setup again. And of course, cuz you all love hash functions so much, uh it's just a way to uh uh uh bring it back to some of the technical features to set up really what are some of the issues. Um we have uh I think it's lecture 11 and 12 where it's just uh what I call act two is the economics, but I want to set up a little bit about the economics. Uh you saw that in the reading. Uh the the the 21st Geneva report that Simon Johnson and Neha Narula and Mike Casey and Jonah and I wrote. So, now you all I only assigned you seven pages out of it. So, I hope that you read read the seven pages. Um but some of the costs and tradeoffs, the challenges of blockchain technology that are very real. I'll give you my own perspective on where I think this will sort out over the next three to 10 years. Uh so, I'll do some predictions. Uh Vitalik Buterin has also talked about a trilemma, and I want to chat about that, and that was one of the readings if I recall. Um he's such a uh leader in this community that when he writes and says something like this, it was relevant. I think that everybody's under understand what uh Vitalik Buterin's kind of trilemma is, even though that some people think he's uh mistaken. Um Some possible solutions uh to this. Uh we have uh who's attending today, Madhu Rao's, who is actually uh one of the developers on some of the solutions um uh around zero-knowledge proofs, and he might get called on. He works over the Digital Currency Initiative. I'm hoping you're ready. Um and and uh why I think governance is the most challenging piece. So, with that, the readings. I have a list of everybody that hasn't spoken yet. So, the goal is to speak. That's what class participation is. Look, I'm going to be lighthearted about it. I It's not that long ago I was a student, really. I remember all this, you know. Um you want to get your name off this list. Just I just want to say that kind of encouraging. So, should I do it alphabetical from the list as to who wants to tell me? No. No, you look like you're ducking your head. What's your name? What's that? Wendy. Wendy, what did you take from uh the the seven pages of the Geneva report? Did you read Did you do the readings? So, what did you take from uh uh the great work of Simon Johnson and I helped him out, you know. Anything about the business challenges of blockchain from the the readings? So, so one one challenge is time, latency. It takes a long time to do it. Wendy uh raises raises. Yes. If you could say your first name. Catalina. It really helps, Sabrina. I'll get you off the list. So, you would It's self-motivation to say your name. So, Catalina. So, it's sort of related. They're they're they're not uh alone, but performance, scalability, the time it takes to do a transaction, uh other challenges. Yes. First name. So, how to do micro payments? You want to tease that out? Why why is there a problem with micro payments? All right. So, how to do micro payments and the small micro payments, partly because they're so small, maybe relative to the fees and the cost of the network. Alexis. All right. So, there's there's economic incentives that are involved here. We're now moving a little bit away from all that, you know, that stuff, the the broccoli that I said that we were all going to be eating about hash functions and so forth. Akira. Okay. So, Akira just raised a bunch of points about privacy and security about the individuals and the regulators. Uh uh Does anybody want to tease that out a little bit more? Um Okay. So, you have that natural public policy tension that doesn't only exist around blockchain. It It Jihee. I just I could not hear anything back here. Um so, if people can speak up a little bit. Okay. Do you want to say it again? Yeah. So, so inherently, the regulators want to look into the details, whereas those banks have a high incentive to keep that on the private side. So, on the one side, there's a commercial interest to keep things private. On the other side, the official sector might want to peer in. And then interestingly, on top of it, layered on it, the official sector also wants privacy for everybody other than the official sector. So, like in Europe, there's a new uh requirement that wasn't in the reading, so don't worry. But is anybody familiar with the uh the directive, the privacy directive called GDPR? Um I don't remember your Aaron, I'm sorry. Aaron. Aaron, do you want to tell the class a little bit about GDPR, or if you I'm not super familiar. I just know that it's like a big deal right now with going after social media. Uh Stephanie. Yeah. So, my understanding is that um especially when it comes to like advertising to consumers, they have consumers in like the EU have to like really like check certain boxes to agree to be advertised to as opposed to just automatically getting that. Right. So, uh And Joaquin. It's it's a privacy law. You have also the right to opt out of being um um I don't being like tracked everything you do. Right. Michael. Is it Michael? Yeah. I was going to say we worked on this company I worked on at the summer. We had to put uh purging mechanisms into our database. Right. So, it's a remarkable new law. Europe is in a sense uh if you wish to say either more privacy protection or ahead of the US, or you know, and each jurisdiction has their own cultural and political norms, but Europe as a whole has moved further in a sense. You have a right to be forgotten. You have a right to access the information as well. And so, how to be forgotten in the context of an immutable blockchain is an interesting just technical set of issues. Yes. That was the question I was going to ask. Kyle. It's my name. But um Say your first name. I'm Kyle. Kyle. All right. I worked for a company this summer that processes transactions, and it was our understanding, speaking with lawyers in Europe, that under GDPR, you're allowed to request your transactions, cuz the transaction counts as personal information, you're allowed to request your transactions to be erased from a ledger, which obviously opens the door to all kinds of fraudulent behavior. I'm just curious to know if you've heard of any sort of resolution to that. I I haven't. I was speaking at a conference earlier today here at MIT with a bunch of member companies to the computer science and AI lab. And one of the participants said they thought they had a technical set of solutions to it. So, we're going to talk more about the privacy issues and GDPR in the public policy session next week. So, I'll try to I'll try it. Kyle, remind me and I will try to get more up to speed on that. Kelly. I found the it's specifically within the fourth chapter, there was something mentioned about what makes blockchain uniquely qualified to solve a lot of these solutions and I found that it kind of coincided with what Michel said last class that there are significant trade-offs that often come down to the cost of trust. But, it still begs the question with so many technical challenges, why is it still such a something that's so sought after? So, I'm hoping that we can clear that up a little. Well, we'll give it a shot. Other thoughts or questions from the readings? Uh I was talking earlier with Leo about the layer two. About the layer two? Yeah. And my point about that is that it's okay about having a second layer to provide the the efficiency and the high performance, but it's running offline. Yeah, so we are starting to trust in third second layer that is runs offline and then goes and put inside the blockchain. How how feasible how how can we trust? So, the question is about possible solutions to address performance and scalability and I have a few slides on that, but in essence, if the if the principal protocol, the Bitcoin protocol or the Ethereum protocol or maybe tomorrow be EOS or some other protocol has some performance issues, could some activity be moved to another channel? Um that channel could be called a layer two in the lightning network, which was there was a reading about that. That could be with a little bit different technology called side chains or sharding, which I think was an optional. Yeah, I did that optionally. I didn't force you all. So, there's a number of alternative channels. And though the technologies are to technologists importantly different, sharding and side chains and layer two, for this purpose for a moment, let me blur over the differences. Um the question that uh Lior Dros asked is, "Well, is that meaning that we have to trust?" I would contend we already have to trust the protocol that the Bitcoin core developers have written. It's open GitHub, it's open code, but very few people are actually going to investigate it enough to be assured there's not a bug or an error. But, I agree with you that the the core the Bitcoin core or the Ethereum core has been living in, if I can call it, the technological and commercial swamp. It's been attacked by so many viruses and bugs, you have some reason to trust it, but you should never be 100% sure. The side the side chains are less tested, but I do agree with you you have some trust in unless I misunderstand the question. I thought it was a trust in the in the underlying code. Uh my my main point is working with second layer that's offline. You are not really transacting in the the the in the blockchain. Yeah, it's off chain. Yeah. May I say like this? So, the question is is if you're off chain, should you be more worried? I was addressing just a narrow part about the code. You're saying, "Should you be worried because it doesn't have the same validation models?" So, can I hold that question until we get to the slides cuz I think that I think you do actually have some pretty good validation, but I think you're right that it's a it's a valid question. Is the validation in these off chains still work? There was a question back here. I was I was just sort of going to respond to the issue of like sort of trusting these offline mechanisms. It's not in the same vein, but like 90% of according to this of like daily trading volume occurs in these crypto exchanges. So, that's also like sort of happening off the chain. So, I think that like in this community of people who are like currently participating, there's no reason that like we're not going to trust sort of like third party vendors, so to speak. So, you're raising the point and it's a sort of an irony of the whole ecosystem that the majority and in some cryptocurrencies over 90% of the actual daily transactions are happening in a very centralized ways on exchanges, particularly the centralized crypto exchanges, where um I can't remember, but was about half of you have owned Bitcoin at some point in time. Um but, can I ask how many of you have actually operated a full node? So, there's the two technologists at the middle table and Hugo, who's also, if I remember right, engineering PhD student. Okay. So, we have our three PhD students who have operated full nodes. Honors to you. Um but, for most of you, and it was half the class of owned some Bitcoin, you've trusted some other authority to hold your private keys. I'm not saying you're wrong or right, but it's an interesting and an important point about this ecosystem. So, let me let me unless there's other points, just go through some of how I thought about these things and laid it out. And I I of course the study questions we've been talking about, so I'll come come back, but we will talk a little bit about hard forks and and we we I didn't hear anybody talk about interoperability, so we'll come back to that as well. We've talked more about performance and privacy issues. Um just back to the technical features, it's just repetition. Sorry, but I do think it's worthwhile. There's of course the cryptography and timestamp logs. So, the bedrock of this technology that we did three or four lectures ago. You will find that in permission systems, you'll find it in permissionless systems. That is a bedrock. You will find it in Ethereum, Bitcoin and 1600 others. There there will be some shifts around. The hash functions might be a little different, but that's kind of a bedrock of this technology. The network consensus is not necessarily always the same as we talked about. Sometimes it's proof of work, sometimes it's proof of stake or in the permission systems, the consensus is really are you amongst the club? And then there's kind of some form of a club deal. If you're the Australian stock exchange, the only member of the club is the Australian stock exchange. But, in other permission systems, it's 20 banks or 15 banks that are sharing that some delegated randomized authority to say what's the next amendment to the And then the transactions code and ledgers shifts largely dependent upon whether it's a transaction ledger or an account ledger. So, transaction ledgers have to have some way to record transactions. Account ledgers have to have some way to record the change in accounts, which Ether calls state transitions. But, either you have to record a transaction or you have to record a change in the state or a change in the account. An account would say that would be like the income statement versus the balance sheet, but you you kind of need to record that. And basically all of these technologies, as I understand it, have some way to keep those ledgers. Although there's multiple ways to do it and as we talked about last week, some have one Merkle tree and some have four or five Merkle trees and so forth, but it's it's embedded uh in this and they could have different scripting. Questions? Just That's like the thumbnail just to remind you about the technology. It's the T in MIT. So, just a quick fact checking question on the ledgers. So, for a transaction versus account in an account based ledger, does it say like if you send $5, does it say who you're sending that $5 to? Or does it just say your account went down $5 and somebody else's went up $5, but it doesn't matter where it came from. So, I'm going to take, as I understand Ether and Midor's you'll bail me out, but um you you put a state transition uh uh instead of a transaction input, it's a state transition input and that state transition input does have one account going down and another account going up. So, if you investigate that, you can see where it came from. You can see both sides of it as I understand, and there's a receipt ledger. There's actually a receipt Merkle tree that then keeps this state transition happened. Did I Did I roughly get that right, Madars? I believe so, yeah. Oh, wow. All right. Madars is actually does this for a living. I mean, he's he's the one of the founding people of Zcash, so um and and a bunch of other wonderful things. I'm not going to go through each of the details, but there was about 15 or 20 details in these slides about the differences between Bitcoin and Ether, and I just use it to remind cuz it's saying, "Okay, why Why Why did the professor Why did the Why did Gary put it up there? This Professor Lessig, that was nice for Larry, but for me, you don't need." But, in essence, the big difference is account-based versus transaction-based. The kind of big difference is Ether does seem to go faster, but it doesn't have a lot of throughput. They still both use proof of work even though Ether says they're going to move to proof of state. They're not there yet. When they get there, we'll all know together. Um the economics are a bit different, of course, as well. Um and um But, all of these details are part of the reason why uh there's problems. And so, you read in the Geneva report a little bit about a professor uh an economist from the 1930s. Does anybody want to take a crack at it, or should I just do my slides? Here we go. I know, it's just like Coase's theory. Yes. Yeah, the fact that like uh everything should be analyzed on the like cost-benefit analysis, so that like if one wants to use the blockchain decentralized network, you should take into account like all the benefits in terms like reducing cost of trust, enhancing security, but also like the cost of like switching to a decentralized system. Co- Coase is a uh is an economist from the 1930s who wrote extensively about the cost and empirically about the corporation. Kelly? Yeah, basically like trying to understand why transactions would aggregate into a firm. Why Why uh move all your activity into one? Right. So, in a much earlier time, way pre-Bitcoin, but a different Why does economic activity cluster into a firm rather than, if it was truly market-based, I might just be uh selling my services? In essence, he was asking the question, "Why don't we have a fully gig economy in the 1930s, where everybody's free labor uh and and and and even capital and labor meets individually, and we collect up together?" Um That was kind of the body of his question that he was So, centralization versus decentralization 80 years ago, studied by a great economist. So, just thinking about it here a little bit, I think that when you go from decentralization to centralization, you tend on the centralized side, you get capture. You get economic rents, and you do have a single point of failure. In some sense, the resiliency of the system, whether it's in finance, where you worry about systemic risk, the one clearinghouse, one central bank, one government, if it's knocked out, uh it's so relevant to the economy at large, it brings it down. Or, if it's one database, you have a single point in essence of failure. Economic rents is an ability to collect excess profits. I assure you, everybody in this class wants to collect economic rents. We started out as venture capitalists and entrepreneurs, but we have a motivation and incentive to be monopolists. But, we don't want to do it illegally, of course. I mean, we we just want to get there, you know, by dominating the market. I'm sorry, was there a question here? You're just You're shaking your head, and I didn't know. Um but, on the other side, there's there's the benefits. The The Y scale is not written on here. The Y scale is how How are we going to think of it? But, I think of it as kind of cost. So, the Y scale up the Y scale is like greater cost. Decentralization, the big cost that come there is coordination. You You have a lot of collective action issues. If the hundred or so people in this room were trying to do something together, you have to figure out how to do it collectively. Uh and coordination. And that's true of every blockchain uh that you could think about. Governance relates to coordination and collective action issues. And then, security and scalability. These two lines are not in any of the readings, but they try to capture what And depending upon the slope of the two lines, you might say that a market might tend towards more centralization. In theory, if I had changed the slope, it would be further to the decentralized side. Right? If the cost of decentralization is a lower slope, and the cost of centralization is higher, we will tend more towards decentralization. So, it's just a a way that uh to visualize Here are the costs of centralization, which are basically capture rents and single point of failure. Not that there aren't other costs of centralization. Here are the costs of decentralization. Um I think in each one of the applications, when you're thinking of use cases, it's worthy This is kind of a core thing. Uh will this application lend itself uh to a a low slope decentralization curve and a high slope centralization curve? Are there a lot of economic rents? Are there real problems with single points of failure or capture? And if it's a low slope decentralization curve, meaning there's not much cost to the governance and coordination and scalability issues, then you're going to be more towards decentralization. That This isn't any reading or book. It's just, you know, a shot at trying to visualize it. Sean, any question? Um no, I don't. But, one one question I have is slightly irrelevant from from this area is that why are all the all the pri- privacy co- the privacy coins are off the hot water? Uh what what of the a lot of them are off the hot water with the Bitcoin as opposed to So, Sean's question is, "Why are the privacy coins forked?" Not all of them, but but but many of them are forked off of Bitcoin or one of the major coins. Madars, you want to say Did you You have one privacy coin. So, Bitcoin has a very robust and well-established code base, so there's a lot of high-quality code Could you speak up? Oh, okay. Bitcoin has a lot of high-quality code, so you can build up on it. So, it's natural to like add privacy on top of Bitcoin in your fork, rather than writing from scratch. So, what what Madars is answering is There's something that's freely available, the Bitcoin core code. It's actually under a copyright license here at MIT, which makes it free. That was Satoshi Nakamoto's decision. It wasn't that Well, maybe Nakamoto does work here. It's a clue. Um but, it's it's it's been developed, and it's knocked around, as I call it, in the proverbial swamp. I mean, with all these attack viruses and so forth. And so, Madars is saying, "Build off of that, and basically get that code for free, and then fork." Is that Um So, the challenge is We talked about it. Uh performance, scalability, efficiency, privacy, security. What there was less talk about was interoperability, governance, and collective action. And I'm going to dig into those two more because I it feels like uh that's worthwhile. I also believe that the first bucket, performance and privacy buckets, are more susceptible to fixes. Though that might take three or five or even eight or ten years to happen, I think they're more susceptible to the to the bright men and women that are in these fields addressing themselves to the computer science and cryptography of the space, whereas governance and collective action might be solvable, but I think it's it's sort of inherent in the human element uh and the commercial arrangements that governance and collective action are the harder of these four buckets. Um that's just one person's read of it. Uh but, we'll go through some of the reasons uh as well why I kind of I I get to to that uh view. There's also commercial use case challenges. We're not going to dig into that much here. That's mostly the second half of the semester. But, I just wanted to mention that's a real thing that a lot of folks are saying, "Well, I have to make sure that this is the best commercial application and so forth, and can I make money?" I mean, ultimately, um uh on it. And then, uh we are next week going to talk about the public policy issues and challenges, and they all kind of intersect in a way as well. So, Vitalik Buterin, I think there was a medium post I had you all read. Uh Bo, do you want me to tell you tell us a little bit about what you think uh is is uh Vitalik uh not only a brilliant computer scientist, but does he get the economics of this right, or you think he's off? You can be on it There's no right answer. I I know people that feel both ends of the spectrum about his trilemma. So, I'm setting you up that The Geneva reading, which I did first, it seems like it makes sense, but his He's basically saying choose two. Um build cuz you can't have all three. Right. It's an old saying about what How many of you have ever hired a contractor to fix something in your kitchen or renovate renovate something? I mean, I have. I'm a little older, right? You know the old saw that it's good, quick, cheap, but you can't you can't get all three. You can only get two out of three. That's sort of the contractor dilemma. But but what do you think? Do you think he's right, or do you think you could maybe over time get all three? Good, cheap, quick, scalable, decentralized, and secure. My very unsophisticated knowledge of the technicalities behind this, um I think he's right. You think he's right? Who wants to take the other side just for have some fun and some debate? Sure. Uh go go at it, Leonardo. So, Yeah, there you got it. We got it. We got the name. I think one of the facts we're talking about the time that systems have had to develop. So, the image, for example, Visa has had 60 years to develop a system that works. Some of those currencies have three, four, five, 10 years. So, they will I think even put a number. His personal opinion that I actually thought that less than 5% they will not overcome the hurdle. Uh I don't know if that is right or wrong, but I I think the fact that it's so recent, I think the jury's still out. Right. So, Leonardo's point is it's recent. This is a new technology. Yes, maybe Vitalik is right that only 5%, maybe only 1% will succeed. But to say that no one will deal with these three points in in a simultaneous satisfactory way. And ultimately, it has to be satisfactory in a commercial way. You know, taking the risk and trade-offs. So, Leonardo takes the other side. Anybody want to say why Leonardo Hugo, were you Which side are you taking? Leonardo's side or Vitalik's side? Somewhere in the middle. Okay. Um So, I I think that there is a trade-off, but I agree that it might take time to improve all three simultaneously. I mean, one thing that happened just this last week in Bitcoin was there was a vulnerability discovered, and like as somebody who doesn't know how to check the code base really, like I'm not a computer science person, so I never like check for bugs or anything like that. I don't know how to do that. I kind of just take the software as it stands, download it, and install it. Um and when they say I need to update my software because there's a bug, I'll update the software there's a bug. And that kind of feels like uh more of a centralized system, but you're getting that security. But then the decentralization comes something from the fact that the network is still spread out over so many nodes. So, like there are trade-offs, but I think you can kind of build on one and then climb another cliff and then like kind of build on each, but not at the same time. So, Hugo's somewhere in the middle. Um I'm I'm probably an optimist enough on the human condition and that that technologists will solve more than Vitalik. So, uh I'm not saying I'm all the way where Leonardo where where Yeah, there uh would be, but I'm probably closer to Leonardo than to Vitalik, but that's just my points of view. I also find it interesting if it's all right if I say Hugo's an engineering uh doctoral student here. And yet, he's not checking the code, not any, right? Because right. So, there's a trust issue that in the marketplace, the trust of the code. We all also trust Facebook and Dropbox and and and broadly speaking, the internet as well. Priya? I mean, I was going to say that there are several examples throughout the evolution of human interaction where these three things have been sorted. So, it might not be that any In some systems at any one time, all these three nodes are working, right? Like for our current payment system, there are moments of vulnerability. Yes, but then you catch it sooner or later. So, I feel like it's maybe not it's about having it all perfect right now versus will you get to a point where all three are mostly in place and working. And I think I like how Priya said working enough. Right? It doesn't have to be scalable to the place where it's millions of transactions a second. But maybe it needs to be faster than seven or 10 transactions a second. Or Ethereum, 20 transactions a second. Um we had uh I I should also remind everybody Tuesday nights we have dinner. Simon Johnson uh treats every Tuesday night. Uh it's not required to come, but it's 5:30 to 7:00 that we have an outside speaker around the blockchain space, and you're welcome to come. Michelle is here who uh Michelle Fierro runs it. mailing list, so please um I will put my name on the board later. So, Michelle will put her name on the board. Anybody that wants to come, but this past week we had somebody speaking about uh the scalability issues, and when his company did a $25 million initial coin offering, the day of the offering, which only had 40 to 50,000 purchasers. So, 40 or 50,000 purchasers on that day, that means the smart contract had to be triggered numerous times that day. Took over a third of the entire Ethereum network on that day. And it's sluggish, and just to close and settle on his initial coin offering, he was saying, "Geez, that that's not the scalability, you know, we want." So, we know that's where we are today. Um Visa runs around 20 to 30,000 transactions a second. DTCC, which settles all the stock and equity trades here in the US, has to be available to co- transact at least 100,000 a second. Most seconds it's five or 10, or some seconds it's 30,000, but the Securities and Exchange Commission says, "No, you have to be rated for four times your average, roughly." Um so, this gives you a sense of the scalability issues just in the current environment. Um if one layer is the internet of things on top of it, there's somewhere and I've heard different estimates of eight to 10 billion devices currently connected to the internet. And that's likely to grow as more refrigerators and street lights and traffic lights are tied into the internet in the next five or 10 years to 50 to 100 billion devices tied to the internet. If they start communicating to each other, will it be a blockchain for internet of things? Can't do it at these types of scaling numbers. Or can you do it uh in some alternative um method? Um proof of work is also has a bunch of energy consumption. Um we didn't have that a lot in the writing. We chose not to put a bunch of that in the Geneva report. Um one estimate is that it's 200 million kilowatt hours per day. That's equivalent about 7 million US homes on average. Just give you a rough uh uh digiconomist estimate. Um that's 1/3 of 1% of all the world's electricity. Just to scale scale it if you Now you can have a nice dinner you know, dinner party conversation point. Um it's the it's the electricity of the country of Austria. Is anybody Austrian? No, I just was, you know, um So, it it's that's one set of trade-offs of proof of work as well. But it also costs a lot of money to run the banking system. So, I think that when somebody says, "Well, it's terrible, it's challenging, all this electric cost." Yes, we always want to lower costs, but the US financial system is 7.5% of our economy and costs 1.5 trillion dollars. So, the payment system around the globe costs 0.5% to 1% of the global economy, which is more than 1/3 of 1% of the world electricity cost. So, I'm just putting it in It's back to this question of which cost of trust, which costs. I'm um um neither a maximalist or a minimalist, as you recall. Um So, what are some of the alternatives? And we're not going to dig into each of these. Side chain, sharding, layer two payment channels. Anybody want to take a crack? They're not all that identical. It It's uh uh Madhu and I had a conversation earlier today that I could even get my head around because I get confused. But does anybody want to give the basic We were talking about it earlier. The basic tenant cuz you had a reading about the lightning network. As to what's what's the economic thing and technical thing that's being attempted in all four of these types of things? Was James was that a hand up? You're going to give it a shot? Give it a shot. I'll give it a go. Um Most of these are on the chain uh sorry, most of these are off the chain, but some is on the chain. The idea is you transact off the chain and balances are traded and the net of the two of the results goes onto the chain. So, you try and speed up by processing off the chain where you can have thousands and millions of transactions and there's only the net amount that goes in the chain that is fewer So, James has summarized it as it's like saying there's this chain this this this this this this channel if you wish that the water is running in or the digital money is running in that only has a certain speed. It can only take a certain amount of performance. Why not take a lot of activity and put it in a side channel which is called a payment channel actually, but a side chain or a payment channel or a layer two all with slightly different technical features. Um and maybe do millions of things off here and only put some here. It is not new to blockchain and Bitcoin. We already have that in the world of finance for decades in some way or another where some activity can't go to a central settlement system. And recalling ledgers, the central bank whether it's the US central bank or any central bank could have been set up that all of our deposit accounts were directly with the central bank. And in a sense, the side chains in finance right now are 9,000 commercial banks. 9,000 commercial banks are dealing with our money flows and then sort of net settling to the central bank's ledger in what's called digital reserves. And in fact, even the banking system, the 9,000 banks in the US have their side chains. Visa, MasterCard, First Data, the all the money processing. So, there's already a layering. I look at layer two and side chains as kind of taking a similar economic approach and technical approach that's already been around, but in a new way. I grabbed a chart from 2015. It The details don't matter, but this was I did it cuz it was 3 years old. This was one, you know, person's truth coin. A one person's view is what side chains. Basically, what James says, lots of activity over here and only a few things go over to the main chain. The visual is what I wanted to get across. It was just that it's kind of Think of it as loads of activity over here and then we only settle at sometimes to the main chain. Another visualization of a different is the lightning network. Again, a lot of activity then settle to the main chain. Um questions. Zack, there seems to be a a trade-off. A lot of people are proponents of making the block size bigger. A lot of people say side chains. I I have trouble understanding like the trade-offs between those two. So, like what what it like why not just a bigger block? What's the problem there? So, there's a there's a series of trade-offs of economics and technology. The more you put inside the let's call it the main chain. That that that blue the blue boxes at the bottom so so to speak. The the more you're putting in there you uh weight it down. There's there's there's more processing of course and more storage and so forth. Um but also there's some there's too much latency. In Bitcoin, it's every 10 minutes and you're not really sure until three, four, five, some would say six blocks to an hour go by. So, economically, if you want high uh frequency low latency short time periods you might say, I can't get that on the main chain because the main chain wants to have low latency. Every 10 minutes is low latency now. Uh low latency to be more secure to keep the mining cost and the proof of work up. So, there's some economic and technological both cross currents uh uh with that. Unrelated to what I just said, but overlapping, there's also a bunch of minor and mining pool operator economics as to whether they want big blocks or small blocks and part of the split last year was uh uh but uh it was sort of more motivated around uh local politics rather than global politics as as as uh the former speaker of the house Tip O'Neill said all politics is local. I think some of the debates last year was about local economics and the economics of miners. Medores was in probably in the middle of some of those debates, but would you have a different view as There was a big debate last year as to whether the Bitcoin block should go bigger or stay the size. It was not the only reason, but it was part of the reason we have now Bitcoin Cash and Bitcoin because uh Bitcoin Cash has a bigger block size and a shorter 2 and 1/2 minute um processing time. There's something that can be said about just um mining centralization. The bigger blocks you have only the miners that can handle the enormous blocks will be able to stay in business. And less decentralization means less security. So, there's incentive both from decentralization and security to keep the blocks smaller, not bigger. So, what Medores is saying is is is there's also a bit of economics around centralization. The bigger the blocks, the fewer miners can handle it. The fewer miners, the more centralization and thus less secure and maybe even economic rents cuz every centralized system can collect economic rents. Uh Aline? Yes, so another problem is that if you have bigger blocks, they take longer to propagate to the network. Uh and in sort of non unintuitive ways, uh if that happens, you get more accidental forks in the blockchain. And people hate accidental forks, especially miners hate accidental forks cuz they lose coins when their when when their blocks aren't So, Aline's saying a technical feature as bigger blocks uh are more likely uh to take time to propagate through the network and thus you might end up uh inadvertently having um uh uh more um uh chains that are are are discredited in a sense cuz they they're there was work being done until the first one gets propagated. Uh my question is about the keeping track of the of the transactions. That's right. So, Leonart Leandro Did I know? Leandro. Okay. Uh was he asking, how do we validate the lightning network and how do we assure that that's is uh though though Yeah, because if we're working with nets uh to put in the in the chain uh how do we really keep record of everything that's Okay. So, the side chains are not recorded gross on the main chain. They're in essence recorded net. And in lightning network I said I wasn't going to get into the differences, but here I go. The lightning network is more a bilateral network. It can be it can take on the the the feeling of multilateral because I could have a transaction with James, James could have a transaction with Kelly and it feels like it's three of us, but it's bilateral James and Gary, bilateral James and Kelly as I understand it. And so, those individual transactions, while they're recorded if keep me going here, recorded uh in the lightning network, they're not on the main chain. We ultimately then uh net settle to the main chain. And we actually in a sense pre-fund or pre it's a it's a form loosely of escrowing at the beginning. Um So, James and I might be messing with each other, but we're bilateral and so we have a another approach to the trust in addition to the computer code. It's James and I might have other reasons to trust. Joaquin. Sorry, and and what keeps me out of double spending once on the lightning network and another one on the blockchain main network at the same time? Um because there's a there's a I want to be careful cuz I'm using the terms loosely. There's a form of pre-funding. It's not that you actually fund onto the main chain, but there's a little bit of partitioning. Is that Medores All right, I keep looking at Medores cuz he's actually coded this. Um so, that's what protects you in essence that James and I, if I'm saying, well, I'll send you a one Bitcoin and tomorrow if the sun does come up tomorrow, you'll send me a half we're partitioning that one Bitcoin or his half Bitcoin uh uh until we then close out that it's written into the scripting code and it's written almost like a smart contract, but it's not called smart contracts um uh to sort of partition or you might loosely think of it as escrow, even though technically it might be different. But stop by and we can And if not, uh some of our colleagues at the Digital Currency Initiative like Taj uh Draga who programmed the Lightning Network. I mean, that's an MIT uh collaboration with others. And we don't we don't promote it just cuz it's MIT. It's like one of the leading ways to do performance. It it happens to be MIT. So, let me talk about other ways to do performance and kind of move on. We already talked about alternative consensus protocols. You've seen this slide. I'm just bringing it back because it is a way to deal with scalability. It's a really critical important ways. Proof of work is one of the issues about scalability and generally I'm I'm summarizing I'm simplifying the sense, but generally all the alternatives have some way to randomize or delegate the node that will do the next block. It kind of all comes back. How do you add another block? And Stuart Haber in the 1990s when he started with all this blockchain stuff and put it in the New York Times had a central authority and he set up that company Surety and he put it in the New York Times. And what Nakamoto consensus is is he said, "Well, no, we're not going to have Haber and and and a central authority. It's going to be decentralized." So, these other consensus protocols generally have some randomized approach to delegate the selection uh of the next block. It's not always that way, but they may also have a mechanism to do a second thing, a second touch. Uh Silvio Micali's Algorand. Uh he's a professor over in the computer science and AI lab and Turing Award winner. He's got a company that has an interesting thing. It's like a jury selection. It's like picking somebody for the jury uh that's that's picking the short group of 12 nodes that might do something. Uh and every block has that selection process, but then there's another broader group that then can check the work of the jury. So, often there's kind of a second uh automated way uh because trust isn't there ensuring that there's a quick second check. Did they decide guilty or or innocent correctly? Uh so to speak. Uh again, I apologize if I'm a little oversimplifying Silvio's brilliant work, but it's um So, it could be proof of stake, proof of activity, proof of burn as we talked about proof of capacity and as I mentioned last week, um there's not large-scale uses, but Dash and Neo both have some form of this going on right now. Um and Ethereum has a big project. Uh I'm I'm confusing their two projects. There's Plasma and there's uh Casper. Casper is their project to get to proof of stake, but they're not there. Privacy and security. So, uh I'm trying to remember who raised the contradictory tensions. Um The contradictory tensions is law enforcement and regulators want more transparency even though the FBI did you know, sort of figure out some Russians were using Bitcoin to mess in our elections. Um they want some more transparency and financial institutions users and even some regulators want less transparency. So, it's it's not kind of goes both ways. But these I think are also truly solvable. Well, for for consumers, there's Dash and Monero and Zcash. And there's even mechanisms called mixing and tumbling, which I truthfully can't tell you the difference, but I can tell you regulators talk about mixers and tumblers and privacy coins. When when I go to some regulatory conferences cuz they sometimes invite me as a former um that middle slide, the the the privacy coins and the mixers and tumblers uh the the finance ministries and the law enforcement stuff, that's where they kind of get worried. Um Madars, you want to come up here and tell us anything about Zcash or you want to do it from there as to uh what inspired you to do uh a privacy coin that a bunch of law enforcement folks don't like? Oh, I don't mean I mean, but you know Honestly um And it's legal. I mean, it's a coin. It's real. Um So just like cash can be used for illicit purposes, also systems that provide strong privacy like Tor can be used for illicit purposes. So, that said, privacy is a human right and we shouldn't be giving up our financial independence just because I want to buy coffee. I don't want to reveal all my other transactions. I think that there are mechanisms how law enforcement can get their regulatory objectives, but privacy I think is fundamental right, so we should fight for it. And so, when did you start working on the project? Um I think it was 2014 when we started writing the paper and then So, you started with a paper? Uh so, we got like prototype code base. We we put it uh open source and uh then there was a company that got formed uh to launch the project. So. And I think Zcash now is somewhere around a billion dollar market cap if it Fluctuates wildly. Fluctuates, so So, that's why you're here. It went down. No, no, you don't need to answer that. That's all right. Privacy. But but in essence, what Madars is saying that he came to this. What were you doing in 2014? Were you I was a grad student here at MIT. I was mostly working on zero knowledge proofs. On zero knowledge proofs? Yeah, and it seemed to be like a natural application like Bitcoin plus ZK proof techniques. Uh maybe there's something there and So, here a talented graduate student at MIT with the with the collaboration of others said, "Here's this cryptographic mechanism called zero knowledge proofs, which we'll chat about in 30 seconds. And here's something called Bitcoin. Why don't we bring them together and we can promote in his own words uh some human rights. Just as you buy a cup of coffee and you don't have to say who you are, you could use this new Bitcoin enhanced zero knowledge proof Zcash uh shot." Uh I have a question regarding how do you define illicit activity in a way that What if living in a country that has capital control and if I use let's say Monero or Zcash as a way to get the money out, does that count towards illicit activity or how do you define it? So, I don't I don't Fortunately, I don't have to define illicit activity. Uh but generally societies come together through their uh uh reasonable mechanisms, whether they're democratic societies or not, but they come together through their legislative branches and their executive branches and their courts and define some things that are not allowed. Uh but generally speaking when I'm using the term in this class, I'm thinking about four or five buckets. Most societies do not want to shrink their tax base, so they want economic activity be inside the tax envelope rather than outside the tax envelope and that's usually the words that finance ministers call that is a tax base. How much is outside versus inside? Secondly, uh most law enforcement uh and most societies do not want to have the the money rails, the banking and other ways you can move value to facilitate otherwise illegal activity. So, it's using money to facilitate otherwise illegal activity. So, the otherwise illegal activity might be drug running. The otherwise illegal activity might be terrorism. It might be child uh um slavery, literally. So, it's whatever the otherwise illegal activity you know, is to use money and that's generally called money laundering uh or other things. So, you're absolutely right. Another thing is is for some countries uh less than a majority, but some countries have capital controls. They're trying to maintain the value of their fiat currency relative to other fiat currencies. And in an effort to maintain some either fixed or relationship they have capital controls and and thus in those countries, they might say illicit activity also is running around the the capital controls. But it's it's each country, each society. And Shawn, you raise a good point as to what does it mean? I I I mean it uh not to show any value. I'm saying there's a there's a series of these things that each society comes together and says, usually around the tax base, usually around trying to not use money to facilitate otherwise bad stuff. And in some countries, the uh capital controls. I saw a hand here. Uh Daniel, no? Was there And we're going to do more about illicit activity next week. Uh or about guarding against illicit activity. Hope the correction got filmed, too. Uh um So, there's another set of security issues around private keys. And to most of us that have passwords, you know, if you lose your password, there's usually, in essence, a backdoor that somehow the the platform, whether it's Facebook or even at Bank of America, if you lose your password, there's a backdoor and they can say there's a way to validate with enough probability waiting that I'm Gary Gensler and they'll give me a new password. I mean, in some circumstances, it's a high bar and there's some biometrics involved, but in most cases, it's a pretty low bar and they'll give you sort of another password if you can like me remember the answer to who my high school girlfriend was or something, you know? These questions like I remember, it's Irene. But then I've just given up I've just given it up. That's terrible. Uh I have to change it. Um So, uh but custodial of private keys is a very real thing and you've read about the hacking and we'll read more about it when we get to crypto exchanges. It's it's a very dominant issue not just for individuals, but for institutional actors. How does a hedge fund or more likely, how does BlackRock or Fidelity as an asset manager uh secure custody in a way that works? And it's an asymmetric risk. It's a tricky risk. For most of finance, they don't have custody any longer of the securities. When I started on Wall Street, there was still the cage, c a g e. There was a physical cage where the the the remnants of paper stock certificates were still in the cage. I didn't start so long ago that it was before DTCC. It was already things were getting electric, you know, digital. But there was still a physical cage for some physical paper certificates. If you lost the paper certificate, you could still go to the government or the company that issued it and backdoor and get a new paper certificate. It took time, it was hard it was to authenticate it, but in this circumstance, if you lose the private key there's not the backdoor issue or to get the next one. So, it's a very interesting issue, not just a technological issue, but and a cybersecurity issue, but it's a whole set of financial custody issues and asymmetric risk if you're Goldman Sachs or Fidelity and you lost a key or it got hacked and it was billions of dollars. So, it's just a interesting uh I I don't think it's unique to blockchain, but it's rather specific to blockchain and finance and how it overlaps. So, some of the solutions, and I do think there are solutions here, are some of the things that that Madan and Neha Narula, who runs the Digital Currency Initiative, are working on. And they're working on using two cryptographic primitives we're not going to deeply go into. We did hash functions and we did digital signatures. Those are algorithms or they're called cryptographic primitives. Well, there's dozens cryptographic primitives, math algorithms uh Well, the other two that are used a lot in this field are zero-knowledge proofs and less often probably is Pedersen commitments. And I put up there my words. I got Madan to help me write this one. Um but my words is zero-knowledge proofs let someone prove a statement is true without revealing the details of exactly why that statement is true. You might say, "Wait a minute. You can prove something's true." It's sort of like if you walk into a bar and they need to know you're 21 to get a drink. Let's make this tangible. What do you need to prove that you're 21? You need to prove that you were born before 1997, September 27th. But you don't need a lot more details. And so there's some computer scientists here at MIT that actually did the foundational work on zero-knowledge proofs 20 to 30 years ago, Silvio Micali and others. Uh for which I think was part of why they won the Turing Award. Um and amongst other work. Um so, zero-knowledge proofs are a very interesting cryptographic mathematical puzzle-solving that Madan used for Zcash. Neha Madan is using for something called ZK Ledger, which was an optional reading. Um my gut tells me there are ways that we can go forward that regulators and the official sector can get their transparency they want and the financial sector at the same time can get the privacy they want. That the two can actually coexist through the modern methods of technology. Uh Alexy? Is that a hand up or just a waving? Or no. All right. Um you want to add anything, Madan, since you're the co-author of this the ZK Ledger paper that was optional? A little louder, please. Yeah, well, um I just wanted to correct the record about the about Zcash stuff. There was an influential middle coin called Zero Coin protocol developed at Johns Hopkins. So, Johns Hopkins developed a middle coin called Uh middle protocol called Zero Coin. Zero Coin. And that was using Pedersen commitments and Zcash uh didn't use Pedersen commitments. There's like lots of very interesting history behind it. And Pedersen commitments are yet another cryptographic primitive or algorithm uh which interestingly they're similar to hash functions where you take a bunch of data and you squish it together in a sense, you compress it and get a commitment, but you can actually add and subtract them. It's it's a interesting thing where you can commit to data like a hash, but you can also add and and and sort of subtract commitments. Uh so, it it has some interesting features. Um if you're deeply interested, I will probably line up Sabrina to help you or Madan might help because I've at the edge. But what I'm saying from a business side is my hunch, and this is the we're at the we're at the cutting edge here at MIT of some of the folks trying to figure out how to do privacy and security uh at the same time. Um questions on that? Cuz we're going to get to the tougher things. Interoperability. Linking blockchain applications to legacy databases or linking them to each other. So, you might want to link a uh blockchain application. If you're if you're thinking about a payments protocol how does that payments protocol in blockchain world link to the fiat? Cuz ultimately, if you're doing, let's say, remittances and you want to move money from here to Mexico, somebody wants Mexican peso, they might be starting in US dollars. How do you how do you operate basically with three different systems in that case? You know, your in- ingenious innovative startup but the US dollar fiat system and the Mexican peso system. So, that's a form of interoperability. Uh and the and the challenge around it or blockchain to blockchain if we've got 1,600 of them. Or even interoperability of the main chain and some of these layer two and side chains. That's an easier interoperability because it's kind of coded right in. But it's always uh it and this is not a new thing. Banking has had interoperability all the time. Take my example of the US to Mexico. To move US dollars and convert it to Mexican peso is in two entirely different banking systems and two entirely different ledger systems. So, we have to have this question of interoperability even pre-blockchain. But it's just bringing it to this new technology. Um it raises cost to trust in coordinating the transfer assets and information across chains, in essence. Or as we talk about across ledgers. So, it's a it's an issue that's it's been around. We just got to sort of see how we solve it here. Um One solution, it doesn't mean it's the right solution, it's the only solution, but one solution is to do through some decentralized mechanisms including side chains or one of the favorites of the director of the Media Lab, Joi Ito, thinks maybe if we have layer two we should also have layer zero underneath all of these these coins, underneath Ethereum and Bitcoin, maybe there's a layer that we can technologically create. Nobody's done this yet. But Joi is a visionary. Uh Joi had the first internet service provider in Japan at the age of 23 when he got $1,500 of computer equipment and put it in his bathroom. And that's how he started. Yes. Yes, his bathroom. It was the only real estate he had. Um uh So, you know, which way does this go? You hadn't heard that about Joey? No. So, it's a way to start a company. Um and and I think far more work needs to be done. Um So, it may be solvable. I'm not I'm I'm uh I'm not sure. And then, consensus required for software updates. It's a tough one. Open source software updates which are not backward compatible. Like, can I update the software, but then you can't use it for the 500,000 blocks that are already out there in Bitcoin or in some or the millions of blocks in Ether. So, the problem often happens that the older versions won't validate all the new blocks. And if they won't validate all the new blocks, I I'm simplifying again. Think of like Excel, and you get that update on Excel or Word for Window, and you can't open your old files. Um I mean, it's a you know, it's a rough lay definition of what what this issue is. And so, it leads to something called hard forks. And uh this little visual on the right-hand side is basically uh what happens is you can't validate all the old blocks because the new software is kind of going beyond it. A hard fork would happen as if you took two megabyte blocks, if you made the block size bigger. Um uh the the old the old software would not take that. If I've got this right, that would be a hard fork. Um And so, that's a that's an issue, and it's happened. Uh the Ethereum network has Ethereum Classic and as Ethereum because of a hard fork uh that was uh encouraged by Vitalik Buterin. The Bitcoin network has one from last year where it was this debate about block sizes. Um so, it's most software for decades has dealt with how do we update software, but they can push it to us, and we get it. Uh and we hit a button, and we get it, and then after a while we get annoyed, and we don't update uh if you're like me. Um but the consensus uh remember, this was a graph that we had. The consensus always supports the longest chain. Um If the consensus is to adopt this new technology, and only 80% adopt it, it's a question of whether the other 10% will keep maintaining the shorter chain. And in Bitcoin Cash, they have. And so, in essence, now you have two currencies. If for some reason it atrophied, and they stopped maintaining it, then the value in essence in a commercial setting might go to zero. Was there a question? Yeah. Um So, broadly speaking, I think the toughest issue is about collective action and governance. How do you get a whole group of people to be moving in a similar direction? Um blockchain applications derive part of their value from participation of multiple parties on the network as well. There that that that multiple people are involved. It's remarkable in hindsight that Satoshi Nakamoto, whomever he or she was, got this many people There's nearly 100 people in this room studying this 10 years later. But somehow he solved a collective action issue because it was just software code back in 2009. But it's still the example. How does Silvio Micali, with a very clever blockchain adaptation in Algorand, how does he get people to start using it? And until he starts to get people to use it, where is the value? Or if you have an application that's a that's to be file sharing or for medical records, there's a medical records project here at MIT, but how do you start to get people to use it? And you know, these are solved every day in the internet space, but blockchain has a little bit greater rick wrinkle. So, there's a chicken and egg issue. Priya. I mean, um this is like heresy in this room, but is it because it isn't a real thing? Whereas like a medical record, it's really a real thing. Cuz I wonder about that a lot. Like, does it come to fruition because there's essentially not much of a real cost to it or So, Priya's question is is there some perception Can I use that word? There's some perception it's not a real thing, so it might not propagate, and there might not be as much con- consumer adoption. That I That may well be one of the commercial challenges. Ilan, did you have something more? Yeah, I think that Bitcoin the adoption of Bitcoin was because people were interested in the in in the innovation solution innovative solution, and then Ethereum, and then Algorand maybe in in few months it will happen. Uh and other blockchains are basically pouring money into the ecosystem, giving money to developers to develop solutions because they they they are betting on the success of that network. Every one of you as you're thinking about your final projects, this collective action issue has multiple features. One to me is the governance of the blockchain software updates, which we said a little bit about hard forks, and so forth, and how do you have consensus, and how centralized do the the governance stay? Which we'll come back to when we talk about the Securities and Exchange Commission and whether it's a token that's regulated as a security. So, there's that part of governance. But then there's the collective action issue that if you have a payments or medical records or or trade finance, how do you get folks to adopt? And in the banking sector, the banks are the big the big sort of elephants in the room, the big dominant incumbents. How do you get them to adopt? Or you're somehow competing away their profits and not having them adopt, which is more a commercial business issue about collective action. Um So, the financial sector, as we've talked about, favors permission blockchains that don't have as many They have some collective action issues, but they don't have as many collective action issues. They have far fewer scalability and performance issues because they say, "I'm not using proof of work." It might be 15, 20, or even 75 or 100 nodes, but they think that way they can secure their privacy and security. Now, Madhavan's and Neha's uh paper on ZK ledger might be a solution, and some of those banks might start using that. But I'm talking about 2018. I'm not talking about 2020 or 2025. Right now, they're favoring permissioned loop systems rather than permissionless open loop systems. So, next week we're going to be moving to um public policy. Oh my god. You're you're going to get to read my testimony. Uh all 28 pages of it. Yeah, look, I get it. But I was asked to testify in the House uh Agriculture Committee in July. It's a venue I'd been at a whole bunch of times. It was fun to be back in front of uh uh them. Um uh Chairman Conaway from Texas, and uh Colin Peterson uh from Minnesota. Um uh But yes, you'll get to read my that that it was I I knew that there was no legislation that was going to happen this year. I want to just give you the feedback, but some the Republicans run the committee. They get to invite as many witnesses they want, and then they let the minority invite one, sometimes two witnesses. So, I got the call to to testify cuz I'm like the old, you know, sea dog, and they're bringing me in or something. But it was fun, uh but I it sort of I was preparing for this class anyway, so I kind of wrote the testimony for you all. Congress thought it was for them. And it was. It was. Uh um um But that's Mark Carney, who runs the Bank of England, wrote this really beautifully written piece that he gave in the spring uh about a little bit the history currency and so forth. But Mark also runs the Financial Stability the Yeah, it's the the the uh Financial Stability Board, which has the finance ministers and central bank governors and securities regulators from 20 countries. So, it's the G20's finance heavies. I used to go to that, not as a finance heavy, but I used to go because they wanted me inside the tent rather than outside the tent cuz we were doing derivatives reform here in the US, and some of the foreign finance ministers uh had a different point of view. And when it got to the place when seven finance ministers, and it was an interesting group, the Russian finance minister and the UK and the South African and four others, wrote a joint letter to Secretary Geithner uh pointing out some shall we say observations on what we were doing. Um They had differences. Um I got invited, so I used to go. I got to know Mark very well, but it's a it's a good paper. Uh and and you'll get a sense really from I would say Mark is neither a Bitcoin maximalist or minimalist. But he does say, "Don't use the word cryptocurrency. Use the term crypto asset." So, it's kind of an interesting piece. And then, uh I don't know how many of you are Sloan fellows that are here. I recognize some of the Sloan fellows. I think about 20 or 25% of the class are Sloan fellows. You're going to get to see Joe Stiglitz in New York in a few weeks. And I think Yeah, this is the CNBC piece where Joe uh who's Nobel laureate at Columbia University has a a stark and distinct point of view about Bitcoin. I've had two or three lively conversations with Joe about this. Um later in the semester you'll read Paul Krugman and Nouriel Roubini. And and and there's a little video of Bill Gates talking about Bitcoin. I I want you all to be aware of the Bitcoin minimalist and understand what they're saying. And I would put Joe on a one to 10 scale at maybe one and a half. Uh or maybe two. Uh Paul you'll read Paul Krugman's piece a little later. He's kind of down there, too. I can't just They can't modulate between them. But I think it's really important to understand what some really great minds are thinking about this from that side as well. So, that's what the three things are for next week. Um and then uh the conclusion I think that um uh it does provide uh the networking, but it comes with costs. As we said, there are a bunch of trade-offs. I think the scalability, the efficiency, the privacy, the ones are solvable. I can't prove it, but I think in matter of years, and it might be three or 10 years, it won't be three to 10 months, though. I think a lot of that is susceptible to the bright minds of MIT and elsewhere's computer scientists. I think the challenges that are tougher uh really relates to governance. I think governance and collective action, and back to those two graphs, there really are places that are better to centralize than decentralize. And we're going to be exploring that for the rest of the semester together. So, thank you. Thank you for the veterans who who sat through all that.