Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Ep 63: Productivity Enables Security: The Economics of Blockchain Settlement
63 Plays7 days ago
Ep 63: Productivity Enables Security: The Economics of Blockchain Settlement
Recommended
Transcript
Introduction to the Policy Layer Podcast
00:00:06
Speaker
Welcome to the Policy Layer, where we talk about the real decisions shaping crypto, tech, and the future of trust. I am Silvia Sanchez, your host from the Avalanche Policy Coalition. We aim to bridge the gap between the builders of Web3 and the people that are shaping the rules. On our podcast, you'll hear from policymakers, developers, academics, and others working at the edge of tech and regulation.
00:00:32
Speaker
We ask the questions that everyone's actually wondering and we keep it easy to follow. This is the Policy Layer.
Blockchain Settlement vs Traditional Banking
00:00:43
Speaker
So when you tap your card at a cafe, the feeling is instant, but settlement can take a few hours or even a couple of days. In blockchain and crypto, we're out here expecting settlement to be fast and final,
Economic Security in Proof-of-Stake
00:00:55
Speaker
like now. So today we're asking what actually makes a proof of stake blockchain secure at the settlement layer and not in a cool marketing statement that sounds abstract,
00:01:05
Speaker
but from the standpoint of economics. I'm your host, Silvia Sanchez, and this is the Avalanche Policy Coalition. And today I'm joined by two very special guests. We have Professor Fahad Saleh and Professor Coase John, authors of the paper, Productivity Enables Security, the Economics of Blockchain Settlement.
00:01:24
Speaker
We're going to break down why real usage, can make a chain harder to attack and what that actually means. So thank you both for joining us today. Thank you for having us. Thank you for having us.
00:01:35
Speaker
Awesome. So I think we can get started with the title because the title itself is a mic drop. You have this bold statement, productivity enables security. And when you say productivity here, you mean the economic value of blockchain execution. And what is that in plain language?
00:01:53
Speaker
Fees, useful apps, real demand. How would do you break it down? Well, essentially, it is anything that anybody ah finds to be useful that requires using a blockchain.
00:02:07
Speaker
So um in economics, we talk about ah preferences and we don't ah tell people what they like or what they don't like, what their preferences are or what their preferences aren't. We just take as given that there are certain things that people like.
00:02:22
Speaker
um And because they like these things, they're, for instance, willing to pay for them. And so part of what we're saying here when we say productivity enables security is that if you have a blockchain platform that has nothing on it that is useful to anybody, that has nothing on it that makes anybody in the world happy about anything, then you won't be able to secure it because it's not actually generating any economic value.
00:02:45
Speaker
However, if you have, for instance, decentralized applications that certain people get some value out of, and this could be, for instance, a trading application that people get value out of because it gives them a lower cost of execution relative to, say, traditional intermediation mechanisms, it could be also something like I don't know, a meme coin that people attach value to for some reason that we're not necessarily concerned with here.
00:03:11
Speaker
um But ultimately, as long as you do have something on the chain um that ah that people value, that people are willing to pay for to interact with, um that is going to serve as a foundation to being able to secure the chain.
00:03:27
Speaker
um And so the more productivity on the chain, the the the more stuff that's useful on there, ah the easier it will be to secure the chain. Amazing. Thank you. Anything you'd like to add, Kose?
00:03:39
Speaker
um but Yeah, just to sort of a simple way of looking at it is, you know, there is a lot of talk now that we can do things, traditional finance kind of things on online, on the blockchain.
Blockchain's Role in Traditional Finance
00:03:55
Speaker
You know, this is an exciting part of what the future could be. So this is sort of consistent with that. I mean, that you know that's one of the one of the aspects of productivity of the blockchain.
00:04:07
Speaker
So, you know, this is in general, you can think of it as in general, if you're going to use the blockchain for various things, ah that is productivity of the blockchain. And that actually, in our calculus, as you're going to see, that makes it easier to make the blockchain secure.
00:04:27
Speaker
Amazing. Okay, so we have we have already defined this concept and I think it's a great starting point.
Secure Settlement in Proof-of-Stake
00:04:33
Speaker
And if a blockchain is basically a giant group chat agreeing on what happened, just in super simple terms, what would secure settlement mean from the way you're presenting it in your paper? And what counts as breaking it in a proof-of-stake world? So if I use your group chat analogy, then settlement basically means that everybody in the group chat, or at least the honest participants of the group chat, let's say, agree that, ah let's say, the trade happened.
00:05:04
Speaker
So for instance, let's make it concrete. Let's say I am sending Coase one AVAX, right? um And so what we need is everybody in this group chat agrees that I have sent it to him and now he holds it Now, what would be sort of a failure of security?
00:05:22
Speaker
Well, a failure of security is, for instance, we get to a place where everybody says, yeah, yeah, yeah, Coase has the AVAX. And then five minutes later, or it could be five hours later, it could be five days later, suddenly a bunch of them say, no, no, no no actually, Coase doesn't have it, but HOD still has it.
Control and Economic Assumptions in Proof-of-Stake
00:05:40
Speaker
Right. So we need so. So one thing I'm emphasizing there is it's there is a temporal aspect here in the sense that you need kind of forevermore that everybody agrees that ah that the transfer has been made.
00:05:53
Speaker
um And it's not just about, oh, at this moment in time, everybody agrees, but forevermore. And that's actually kind of what makes it a little bit non-trivial. Amazing. um If you'd like, we can go to the next question, just going a bit more into the model. So in your model, the attack the attacker tries to disrupt settlement by buying enough of the native assets to control a big chunk of stake. So what would be a really easy way to explain why buying tokens can translate into power over settlement? Sure. So if we go back to your group chat analogy, ah it's we ought we we perhaps have to refine it a bit. And this get this gets into sort of proof of stake specifically, which is to say that...
00:06:39
Speaker
um the the way permissionless blockchains are set up, and a proof-of-stake blockchain is an example of a permissionless blockchain, probably the most prominent example. um i Who gets to be in the group chat and how much they count um matters. And really what proof-of-stake does is it dictates who gets to be in the chat and how much their voice matters. So in a proof-of-stake context,
00:07:03
Speaker
um ah You essentially have to go buy some of the new native asset on the blockchain to be a participant in this group chat. And your weight essentially depends on how much of that native asset that you have bought and and what we say staked, um which is basically giving yourself a voice inside of of this group chat.
00:07:25
Speaker
ah So if you are just somebody, you know, living your life, not touching this blockchain in any way, not holding any AVAX, you're not in the group chat. But you can go in the group chat. All you have to do is get some AVAX and, quote, stake it, which basically you could think of as, like, entering the chat.
00:07:43
Speaker
And then you have a voice that is proportionate to the amount AVAX that you have staked. Okay, really cool explanation. Kose, what are your thoughts on this?
00:07:55
Speaker
it It is specific. it It depends on the protocol that you use. So we are not getting into it. For example, how things are going to work, how you get to control a block, that depends on the protocol. So Fahad explained very well the protocol for proof-of-stake protocols. So proof-of-stake protocols, they actually conduct a lottery And the probability that you will be winning the lottery depends on how much you've staked as a fraction of the total amount staked.
00:08:33
Speaker
So that is why if you want to have a high probability of controlling the next block, you better have ah ah fraction the the right fraction, ah ah you know, of a stake.
00:08:49
Speaker
So that's why that's why you have to buy the native asset, you stake it, and then the fraction that you have staked as opposed to the total staked native assets, that would that would determine the probability that you'll succeed.
00:09:07
Speaker
So in order to, Fahad explained it earlier. So for example, let's say somebody is interested in reversing one of the transactions which has taken place earlier.
00:09:20
Speaker
In order to reverse the transaction, he needs a large enough stake. Is here what I'm saying? So that's why if you don't have the large enough stake, you have to go buy transaction the large enough stake from the pool of native assets, which is not already staked. So in other words, if it's already staked, let's assume that you cannot disturb that, but you have to look at the already unstaked pool of native assets.
00:09:53
Speaker
You have to go buy up enough of that coins to assure yourself a a high probability of being able to control the next block, which you can use if you want to, to reverse the transaction, which is, of course, ah unsettling the the transaction already settled.
00:10:15
Speaker
Okay. Yeah. Right. Absolutely. OK. And I think also something worth mentioning is that you argue that prior literature kind of treated the attacker like they could buy a massive position at today's price, kind of like Costco bulk pricing for an attack in a way. So what assumption were people making and why is it unrealistic? So the the assumption in the economics literature is frequently referred to as the price taking assumption.
00:10:47
Speaker
um which is to say that economic actors observe a price and they take it as given um and then they can basically purchase as much as they like at that particular price um but of course as you were saying that's not really true now a lot of times the idea is Well, but if you're a relatively small economic actor, um you're not probably going to to affect the price. You're not going to have what we call price impact.
00:11:19
Speaker
And so kind of to a first approximation, it's reasonable to say that you will just that you think you will just trade at the price that you observe. um But the problem here is that this is a specific context where ah we're not thinking about people who are, for instance, small retail investors doing small trades.
00:11:43
Speaker
we are When we're talking about attacking a blockchain in the way that we're discussing, by construction, you are talking about a big buyer, and that's putting it lightly. You're talking about a massive buyer, right? So, for instance, if...
00:11:57
Speaker
30% of the asset is staked, which is not a particularly high ratio for proof of blockchains, um then in order to kind of take control of that group chat we were talking about earlier,
00:12:09
Speaker
you're going have to go buy another 30% of this asset and stake it. But 30% is not a small purchase for any asset, right? We could be talking about buying 30% of all of, you know, the T-bills in circulation or 30% of all of a particular commodity in circulation. Um,
00:12:32
Speaker
That is not a small amount, no matter what market you're talking about. And so what really is is happening in this case is that people were using an assumption that is kind of innocuous in most settings, in a particular setting where it makes no sense whatsoever.
00:12:47
Speaker
And so we're we're kind of pointing that out and then then correcting for it. And that's a really good good baseline. I think that that's basically where um I think that we have now clarified that and we can actually start discussing a bit more your price impact function, um having already gotten this out of the way. So this one is for Coase. To just go a bit deeper in that you derive the price impact function where buying a big fraction of circulating supply can make the cost blow up.
00:13:18
Speaker
So um just dissecting this a bit more, what should people understand about circulating supply versus total supply in this context? So um the important thing to understand is that the assets that are staked are not assets that you can buy, for instance, at Coinbase.
00:13:38
Speaker
are on some decentralized exchange on the blockchain of relevance because these assets are, ah ah in our language, out of circulation.
00:13:49
Speaker
ah Frequently, they're actually sort of stored and custodied on chain. It depends on the exact architecture of the blockchain we're talking about. But the point is ah the the control of them, the custody of them,
00:14:01
Speaker
is not such that they can be liquidly traded in the normal way when you trade and purchase crypto assets from a centralized exchange or a decentralized exchange or so on.
00:14:12
Speaker
And so part of the point that we're making is that when you have ah more assets staked, um while on the one hand it requires purchasing more for an attacker to to successfully attack, in some sense there's an even more powerful effect, which is those assets that are staked are not freely available to trade, which means the assets the attacker has to buy has to be bought from a smaller share of available assets.
00:14:41
Speaker
um So, for instance, um If you just think about different staking ratios, you can see kind of how powerful this is. So, for example, if 30% of the assets are staked and the attacker needs to to match that with, again, 30%, then because are staked, that means that 100 minus 30%, that is 70%, is actually the only part that is freely floating and available to purchase.
00:15:09
Speaker
And so you'd have to buy 30% of that 70%. You can't get the full 100% in circulation. And that that's going to be quite expensive. But think about the following.
00:15:21
Speaker
Imagine that the staking ratio wasn't 30%, but 50%. Now, if the attacker wants to match that, they've got to buy 50%. But there's only 50% in the circulation, which means they actually have to go and buy the entire inventory of every centralized exchange and every decentralized exchange on the face of the earth. And if it, God forbid, so happens that somebody is sitting on these assets, just keeping it in their wallet, not allowing it to be traded at a decentralized exchange or a centralized exchange, guess what?
00:15:55
Speaker
The attacker just cannot get the assets that they need at that point, right? And and and I think that's that's a really good a good way to put it. That's like the whole spicy part, like how staying how staking changes the whole battlefield. And I think this really helps to drive forward the concept that staking is so much more than, oh yield and just locking your tokens aside, but it's essentially adding more security. It's basically the backbone of, of I think, these proof-of-stake mechanisms. And based on what you've said, so it would be like staking removes tokens from circulation, or at least temporarily, which essentially raises the attacker's price
Introduction to Core: Avalanche Crypto Wallet
00:16:35
Speaker
impact. So it's kind of like a... liquidity shrink rate that makes the attacks or the attempts of attacking more expensive and it would be kind of like trying to buy concert tickets when 70 percent are locked by fans who can't resell you you try to do it but essentially you you can't you just can't you don't have enough a percentage of it
00:16:59
Speaker
Ask yourself, what's a wallet for? Bank cards, train tickets, spare change, maybe even a picture of your cat Fluffy. But what about a wallet that does more?
00:17:10
Speaker
Meet Core, the native crypto wallet for Avalanche. Mobile, desktop, however you work, it works around you. There's no messing around with exchanges. Upload funds directly from your bank account or Venmo just as easily as sending a PayPal transaction.
00:17:26
Speaker
You've probably got enough passwords to remember already, right? Log in with your Google or Apple account. No seed phrase memorization required. It's that easy. With multi-chain compatibility, Core lets you view and manage assets across Avalanche and other supported networks. All in one place.
00:17:45
Speaker
Pretty awesome. Then there's a whole world of crypto apps and tools to explore direct from your wallet. Get the wallet that does more. Download Core today.
Economic Productivity and Security in Blockchain
00:17:56
Speaker
So now and walk us through your Core Chain reaction. Higher execution value increases user welfare and some of that gets transferred to investors via staking rewards, which increases staking investment and then security improves. So what's the cleanest way to narrate that loop?
00:18:16
Speaker
Right. So this kind of comes back to, I think, an earlier part of the conversation um where if there is value being generated on the chain, there then are essentially things we can do ah from that due to that value creation to secure the chain.
00:18:34
Speaker
And sort of the the first thing to to understand, to to understand this point, is that by and large, most of the incentives for people to stake, most of the so-called rewards that stakers receive are through a mechanism in economics that we refer to as seniorage.
00:18:53
Speaker
ah which essentially ah means that it's new units of the base asset that are just being printed and happen to be given to a certain subset of the population. So in this particular case, we would be, for instance, printing new units of AVAX and giving it to the people in that group chat, the the so-called stakers.
00:19:13
Speaker
um What this then means though is that we're not giving it to the people who aren't in that that group chat. And so therefore it kind of creates an incentive for people to enter ah the group chat.
00:19:26
Speaker
um And in in the context of our analogy, more people being in the group chat is the thing that secures the system because it creates a higher barrier for somebody to kind of invade this group chat and take it over.
00:19:37
Speaker
um And so the the point though is that because ah the incentive mechanism is sort of this supply side thing where we're just actually increasing the amount of asset, we're not, we don't have a production technology.
00:19:51
Speaker
um what What we're really doing from an economic perspective is we're essentially imposing an inflation tax on the people who aren't in the group chat. And the thing is the value from an inflation tax is only worth something if if there is actually initial an initial base of real value there. right And so if the blockchain is actually producing real value, that means there's real demand for the underlying asset.
00:20:18
Speaker
And the idea is that this is this this, of course, would have to be very valuable for the for the people in the first place to be you know demanding things on the chain. But if we just put a little tax on that,
00:20:29
Speaker
right Then we can take you can funnel just a little bit of that value to encourage people to get into this group chat, which turns out to secure the entire system. And, you know, you could even make an analogy to sort of what really happens in the real world to think about this in some sense, which is that, for instance, actual ah countries have to secure ourselves in sort of a more physical way of for instance, like in the United States, there's a very large defense budget. And, you know, how are you going to fund that massive defense budget? Well, a big piece of that is taxation.
00:21:01
Speaker
um And of course, if there's no actual productivity in the economy, then taxes wouldn't do anything, right? if If nobody's actually working and we say the tax rate is 50%, well, 50% of zero is zero, right?
00:21:14
Speaker
um The idea of there being value in that sense is like analogous to the idea that, for instance, in an economy, there's actually value being produced. And so only if there's value being produced can we put, for instance, a small inflation tax on it in order to then funnel the value to encourage people to enter this group chat, which in our case turns out to be something that secures the system.
00:21:34
Speaker
Got it. Amazing explanation. Anything you'd like to add to that, Koz? No, i think I think this is, a in my mind, this is a difficult ah question, and I think Fahad did a very good job. So, you know, i it's a chain that you ask, right?
00:21:53
Speaker
And I think Fahad explained the chain, that is, the productivity of the system enables... um you know Basically, to increase the staking rate, and then you know you can continue with the chain.
00:22:08
Speaker
Once the staking rate is higher, ah that improves the fraction that is staked, and of course, shrinks the pool that is available for the attacker to buy. you know So the whole chain now goes from this first part of the chain that Fahad expected. Right. Okay. And you also mentioned that there are realistic conditions where proof of stake can be secure, even if the attacker's benefit is arbitrarily large. So what are the conditions doing the heavy lifting?
00:22:43
Speaker
Right. So what what is doing the heavy lifting um is ah essentially ah the... ah the ability to engineer sufficiently high staking ratios.
00:22:58
Speaker
So it kind of comes back to that whole point about productivity. One way to think about it, for instance, is that um if you go back to my sort of real-world taxation example, um you you don't want the tax rate to have to be 100% because that wouldn't work because it would definitely disincentivize anybody from doing any actual work, right? And then it wouldn't work at all.
00:23:20
Speaker
um So you have to be able to sort of i tax at a level that you can induce the staking ratio to be sufficiently high, such that the circulating supply will be sufficiently low, such that the price impact will be sufficiently high, such that the cost of attack will be sufficiently high.
00:23:40
Speaker
And this, but that's a long chain, um but but but this, the the thing that that that sort of underpins it goes back to this point about productivity, right? So if the underlying chain is sufficiently productive, then you won't need, for instance, a very high tax rate, so to speak, in order to get that staking ratio sufficiently high, to get that circulating supply sufficiently low, to get that price impact sufficiently high, to get that cost of attack sufficiently high.
00:24:09
Speaker
um And that's why the first word in the paper is productivity. If you have an unproductive blockchain, if you have an unproductive economy, you can tax at 0%, at 100%, at 50%. None of those numbers will really generate any serious revenues because you don't have a fundamentally underlying productive economy.
00:24:25
Speaker
And so you're either taxing at a high rate with a very small base or you're taxing at a low rate and maybe still even a small base. But if you have productivity in the underlying economy, then you can even, for instance, at a low tax rate, induce the sort of security that you would like. Amazing. And also something um to just mention is in your results, longer average investors hoah investor horizons increase staking, shrink circulating supply and make the price impact div diverse. So intuitively, why would patients make the chain harder to attack? Sure. So ah this gets um into maybe a slightly different intuition that's that's worth thinking about, ah which is that um you want that's those staked assets to actually ah be illiquid for this channel to work.
00:25:19
Speaker
And what that investor horizon is really doing is it's saying this stake capital really is going to stay safe. and is not going to be able to come out of staking in order for ah for, for instance, an attacker to access it. Because, for instance, if you think about, you know, all the examples we've discussed, let's say, for instance, the 50% staking ratio case, if the attacker could somehow convince all of the people who are currently staked to unstake and just dump their assets onto, say, Coinbase or something like that, that's going to essentially start to reverse the price impact channel that we're describing.
00:25:56
Speaker
um And so what that horizon is doing is essentially counteracting that. And practically the way to think about it is that most proof of stake protocols, rightfully so, implement withdrawal lags and limits on the number of on the amount of capital that can flow out per unit time.
00:26:15
Speaker
um And so it's not, you know, we we frame it as an average investor horizon, but the truth of the matter is it's actually a more general concept, which can be controlled at the protocol level, which basically affects how quickly stake capital can come unstaked.
00:26:29
Speaker
And as a general practical guidance to people, you know, designing proof of stake protocols, we would say that you should have ah things in there that make it hard to instantaneously unstake.
00:26:42
Speaker
um it's ah It's a good thing that these protocols tend to already have those things, but that is an important piece because if you make it such that I can immediately unstake, then the attacker will just basically go after that state capital also, and it will actually bring back down the cost of attack.
Regulatory Focus on Blockchain Security
00:27:02
Speaker
Yeah, exactly I mean, this is important. So in other words, remember in our argument, we were using that the stake capital is somewhat illiquid, right?
00:27:14
Speaker
That is, the the attacker is not ah getting, you know, he has to go from the unstaked pool to the Exactly how much that holds depends on um how illiquid the staked capital is. And as Pahad was saying, you can make it part of the protocol.
00:27:36
Speaker
That is, it takes time. If you need to withdraw staked capital, it takes time, which is in in reality the case. So in other words, it's part of the financial engineering that you can do ah using the rules of the protocol.
00:27:55
Speaker
Okay, okay, thank you. And just as we we get close to wrapping up, we get close on on our time, um if you had some moments with financial regulators, policymakers looking into this type of technology, what would you want them to understand from this paper about proof-of-stake settlement finality, especially the idea that real economic usage, that productivity can strengthen security?
00:28:21
Speaker
So the the thing that I would, the first thing i I would like them to understand is that rather than thinking of it as a binary, that, oh, blockchains are secure or they're not secure.
00:28:35
Speaker
um In some sense, what we're highlighting is that the the nature of the design of the blockchain protocol very much affects how secure it is. And so if we're gonna have a conversation about having, for instance, ah ah traditional assets on chain, tokenized and so on, the conversation should shift from, ah oh God, what if a bad actor acquires these assets after we put them on chain to how do we ah constructively contribute to the conversation about engineering the chain so that these things don't happen?
00:29:11
Speaker
um Because, again, part of the point we're making, and this touches on sort of the last part that that we were discussing, is that the these the proof of stake chains can be secured against essentially arbitrarily large incentives.
00:29:23
Speaker
so it's not ah it's not So you shouldn't start from a place about, you know, will a bad thing happen or won't a bad thing happen? It's more about um how do we constructively get place where we can securely do the things that we want to do.
00:29:37
Speaker
Because the the value, for instance, of doing a lot of these things comes from, for instance, like the liquidity that we might get or the increased access that that that that might arise from doing these sorts of things. And so security is really just kind of, I think, what regard a lot of people think of as a necessary condition.
Closing and Resources
00:29:53
Speaker
but then they make the mistake of kind of not realizing that, for instance, the security we observe today is a function of the design that we see today. And the design can, and in fact does over time due to upgrades change. So let's have the constructive conversation of how we design these things so that it is sufficiently secure to do the things that we see economic value ah from.
00:30:20
Speaker
Amazing. Thanks, Fahad. Koz, your concluding thoughts? Yes. So um i like I like what Fahad said. In addition, you know, I think in terms of policy, there is a lot of conversation on, ah you know, using the blockchain you know, to to, for example, using ah financial services online on the blockchain, you know. So in other words, this is sort of a ah related point that is precisely those blockchains where we can have lot of productive activity,
00:31:02
Speaker
you know, those also can be engineered ah to to also make them secure. So, you know, that's sort of a that's sort of an interesting point that these things come together, which which which is, I think, going forward, you know, how we think of doing finance on chain, you know, it's ah related to these points that we're making in the paper.
00:31:27
Speaker
Absolutely, and and thank you both for delineating that in a really helpful way. And i think, and we've seen that in proof of stake, settlement security isn't just about the protocol rules, but also about market structure. And if an attacker has to buy their way into control, the cost is not linear and price impact can be a real barrier, especially when staking reduces what's available to buy.
00:31:53
Speaker
So we could continue talking about this, but we're out of time. But for anybody listening who's curious and wants to discover more about this, you can check out their paper, which we have linked in the description of the episode, and just take a closer look at their model, at their conclusions. So once again, Professors Fahad and Coase, thanks for your time. Thank you for your work, for your research, and for joining us. um And yeah, we'll catch you all next time in the Avalanche Policy Coalition.
00:32:21
Speaker
Thank
00:32:24
Speaker
We hope you enjoyed this episode of the Policy Layer. If you want to learn more, check out avalanchepolicy.com for free educational resources. And also follow us on social media at avalanchepolicy to stay updated.