Epicenter - Learn about Crypto, Blockchain, Ethereum, Bitcoin and Distributed Technologies - Kolten Bergeron & Robert Lauko: Liquity – The Decentralized Borrowing Protocol

Episode Date: March 26, 2021

Liquity is a decentralized borrowing protocol that allows you to draw interest-free loans against Ether used as collateral. In addition to the collateral, the loans are secured by a Stability Pool containing LUSD and by fellow borrowers collectively acting as guarantors of last resort. Liquity as a protocol is non-custodial, immutable, and governance-free.We were joined by Robert Lauko, CEO & Co-founder, and Kolten Bergeron, Head of Growth, of Liquity. We chatted about how the protocol is built and the mechanisms used, how to borrow, and the stability pool and liquidations.Topics covered in this episode:Robert and Kolten's backgrounds and how they got into cryptoWhat led Robert to create LiquityWhat Liquity is and the liquidation mechanism usedThe function of the stability poolThe process of existing troves taking on the debt of undercollateralized trovesLiquity vs Compound & MakerDAOLUSD redemptionsHow the Recovery mode worksThe purpose of the LQTY tokenHow the algorithmic monetary policy worksEpisode links: LiquityLiquity docsLiquity on MediumLiquity on GitHubRobert on TwitterKolten on TwitterThis episode is hosted by Sunny Aggarwal & Zubin Koticha. Show notes and listening options: epicenter.tv/384

This is Epicenter, episode 384, with guest Robert and Colton from Liquity Protocol. So we're on today with Robert, who is the CEO and founder of Liquity, and Colton, who is the head of growth at Liquity, here to talk today about their very interesting new lending protocol that they're building on Ethereum. And we're going to be diving deep into a lot of the mechanism design and sort of the reasons for why they, you know, there's a lot of lending protocol. protocols already today and why they felt the need that like, hey, this is going to bring something really new to the table. But before we dive into that, this is actually a very special episode as well because we have a brand new co-host joining us this time. Zubin Kotitsha, he is, he's been on the show twice already, you know, talking about Open v1 and the more recent episode just last week with

Open v2. And, you know, we just had him on. And I've known Zubin for a long time. And we've had him on and he's just been, you know, always really interesting and great at a lot of this stuff. And so we just, you know, we're wanting to bring him on as a co-host. So I just want to give Zubin a chance a quick second to just intro himself as well to all the listeners. Hey, hey, guys, I'm Zubin. Thanks so much for that, for that intro, Sunny. So Sonny and I go way back, actually. We went to college together. And then right after college, we were living together in SF. And it was just like this crypto house, which was just like the best time. ever. I always knew Sunny as the podcast guy because he would always wake up really early in the

morning and be like, guys, be silent. I have a podcast. But some of the really cool kind of moments with some of the top minds in the space have come out with that out of that process. And it's been so cool to see. And so I was, you know, really excited to join. As for me, I'm co-founder and CEO of Open, which is a decentralized options protocol in Ethereum. And we are currently allowing people to trade call and put options on Ethereum. And the goal is to allow for Ethereum in all sorts of ERC20s in the long run and build the first highly liquid options market in DFI. But, you know, I have two epicenter episodes talking about that more in depth. For now, happy to turn it back to Sunny and be chatting with Robert and Colton.

Awesome. Thanks. And so, yeah, guys, can you tell us a little bit about what is your guys' background? How did you guys get involved with the space? How did you guys start working together? And what led you to Liquity? Yeah, hi, everyone. And thanks for having us on the show. It's a great pleasure to be here.

So, yeah, let me tell you a bit about my background. ground first. So I'm Robert Lauco. I'm a Swiss resident. I used to grow up in, or used to grow up in this crypto valley, which has become famous over time. But then later on, I like move to Zurich. That's where I'm based now. And I'm a lawyer by profession, actually, but I always had a keen interest on technology and computer science in particular. And while working as a lawyer, I just realized that when blockchain became a thing and it was even a thing in my hometown, that it's really like something I wanted to dive in and I just started doing my own research. And after a while, I was lucky to get in touch with people from DFINITY.

It was a small company back then, but it led me to my first job, basically, in the blockchain space. And I became the first employee in Switzerland. I had a chance to work on their protocol as a researcher, but also helped with operations. And yeah, and while working there, I also realized that I got more and more interested in the D5 space, which was starting to unfold in the ecosystem around Ethereum. It was mainly a compound and maker back then in 2018. And then early 19, I started to think about how I would do things differently and maybe more efficiently than Maker when it comes to borrowing crypto assets. And that's how the idea came about. And yeah, in November 2019, I left Infinity to fully focus on my own project, Liquity, which is all about borrowing.

And, yeah, I mean, we set up the company later. Like maybe Colton can tell a bit more about how he joined as a head of growth. Yeah, so my sort of crypto journey is like really long and windy. So I'll spare a lot of the details. But like long story short, I was kind of the weird kid in like 2014 and coffee. talking about Bitcoin, this new magic internet thing that people didn't really understand at the time. And then I ended up graduating in 2018. So that wasn't like, you know, the best time to work in the space.

But after graduating in 2018, I was like, okay, crypto is slowing down quite a bit just because of the chaos that happened in 2017. I'll go work in my other industry that I'm also passionate about, which is e-sports. I spent about a year there, mostly as a graphic design. motion graphics designer, a video editor, content producer. And then after that, I wanted to make my way back into crypto and ended up coming across the position at the Stellar Development Foundation, where I spent about two years working on community and ecosystem development. And then I wanted to make my way back into Ethereum.

And through a mutual connection, I was introduced to Robert and Liquidy. We ended up hitting it off. We had a lot of good conversations. And after a couple months, I ended up joining the team, which was like three months ago now. So it's been full seam ahead ever since then. Awesome. And so, yeah.

So Robert, like what led you to want to build liquidity? So, you know, you mentioned you were, you got interested in defy because you saw a lot of these like existing, you know, you mentioned compound. So, you know, there are a number of existing lending protocols and such on Ethereum. You have compound. You have Avey. You have ones that output a stable coin like Maker. What was the sort of weaknesses that you saw with some of the existing systems that led you to want to design a new one? So I was looking mainly at the way those systems liquidate positions that become under collateralized because those loans, they all need to be sufficiently collateralized.

And I just realized that speed, it's all about speed. the faster you are able to like sell off or somehow make sure that your position gets bailed out in case of an under collateralization, the lower you can set your collateralization ratio or the the less extra margin you need for the same like system security. And then like I started like thinking how to make this just more efficient and quicker like than an auction for example or On the other hand, I mean, you can also just sell off collateral at the discount. I mean, that's what like the early version, like the first version, the single collateral die was doing. But even with that, I mean, you lose some portion of your collateral due to the fact that you need to sell it at a discount.

And that's like how, yeah, I came to the conclusion that why not like make it so that instead of start, instead of like starting looking for a buyer or a bidder once your position becomes under water, you can like start earlier and have something around like an insurance or stability pool, as we call it, where people can already become guarantors of borrowers that may need to be liquidated in the future. So you don't need to find somebody because you already have them available in your system. And that's, yeah, how the idea came about. Cool. So what do you see as kind of the fatal flaw of like the liquidating systems that exist?

Do you see it as like a contributing factor in Black Thursday? And you saw that as important to fix. Could you sum up liquidity as like really doing liquidation? better or is there kind of a broader design choice that you seem to keep making that that makes it different? It's definitely, I think, like the initial idea that like led me to start designing it, but it's by far not the only like thing that separates us from those other systems. Like there are other aspects regarding how we achieve price stability, but also the fact how we deal with governance or the fact that we don't have like voting or governance mechanism in our system. But yeah, I mean, the initial like improvement was all about making liquidations more efficient.

It's interesting because why is this like last, this fast liquidation is very important. And it's interesting to me because I think it's been a while. It's maybe been like six months or so, but we had UMA protocol on about six months ago. And they were actually taking the opposite take where they were saying, hey, these liquidations don't need to be that fast. And like, Maker is actually liquidating too quickly. We want to actually have a slower system for, like, providing Oracle data that triggers liquidation. So what do you see is the benefit of these fast liquidations versus what they're saying, which is like, hey, if the price is crashing very quickly, something's going bad. We want to, like, take it slow and steady rather than, like, force.

liquidating as soon as possible and people are buying a depreciating asset. Yeah, I think the main underlying issue is that you need to make sure, at least if you want to keep your currency stable, is that you want to make sure that your debt can always be covered, like fully covered. Now, I mean, there is some extra margin there because, I mean, those systems they are over collateralized. But if, yeah, it takes some time to liquidate. your position, like from the moment on when you trigger this process, like an auction that may take up to six hours or longer, then during those six hours, the price can, of course, drop even more. And to account for those further price drops during, like, the time it takes to liquidate the position,

you need to have a higher margin in the first place. So if you expect, like, a maximum price drop during the six hours time span of, like, let's say 20%, you need to add those 20% to the, like, margin that you already wanted to have in the first place. And of course, by adding more like margin or by requiring higher collateralization, you make the loans less capital efficient. Yeah, I think there's a lot of questions we'd like to dive into there. But before, let's actually, just zoom out for one second and make sure we give all the listeners a good idea of what the liquidity protocol is in the first place. And can you Can we describe the protocol at a high level first?

I'm sure. Yeah, I'm happy to. I mean, there are like three main differentiating factors. Like we talked about like Maker Dow as our main competitor, but liquid is a bit different. I mean, our main focus is about borrowing. We want to make borrowing as attractive as possible by doing basically two things, by first of all, requiring less collateralization, like the minimum collateralization ratio is only 110 percent, which is, I think, the lowest on the market you can find. Then the other differentiating factor or benefit is that we are not charging a recurring fee or interest rate.

We do charge like one-off fees, but that's like a bit different because it's more predictable for the user than a variable interest rate. And then last but not least, we have this kind of very, like, extreme or radical decentralization and non-governance approach where we don't really have a governance mechanism. everything is immutable. And we also have like an incentivization mechanism, which allows us to not be like required as a company to run a front end or a web interface because we have like incentivized third parties that will run those front ends, which allow people to access the protocol. So those are the three main points I would like stress here.

Yeah, you mentioned. something really interesting, right? Kind of a motivating force is making borrowing as attractive as possible. Would you describe that as one of the main goals of the protocol to make borrowing as attractive as possible? And then what steps do you think you guys do differently, if that is one of the main goals, that will make liquidity much more attractive than some the existing alternatives. So yeah, I mean, the two main points I already mentioned them briefly, like, or like a lower minimum collaterization ratio, which not only like allows you to have a higher

like security as a borrower for the same like say, collect characterization as you would have a different or another system, but it also allows you to be become more risk seeking if you want to and do leverage. So you can achieve a much higher leverage, up to 11% or 11, sorry, 11x by going up, by maxing out your possibilities in our system. And that's like more than what you could do on a similar protocol. And then the other thing is interest rate, which is like maybe less obvious. But the problem with those interest rates is that once you have a position open, you may have opened it at the time and the interest rate was low,

but now you have a debt and the interest rate increases. and that's what just currently happens on our competitors system like Maker. And then you feel obliged or you have this kind of pressure or uneasy feeling as a borrower to repay your debt earlier than you wanted. And by replacing like a recurring fee by something which is one-off, which is an upfront fee when you take out your loan, you have this kind of nice feeling that you wouldn't be obliged to pay anything later. So you can keep your debt or position open as long as you want.

Yeah, and I want to add on top of that, it's like floating interest rates have become this sort of norm in defy, and they're not a great deal for borrowers, especially if those interest rates are floating upwards. So a lot of the times you take out debt and you're expecting to somehow maximize your profit by using that debt or using that leverage. And it's a lot harder to do that if you don't know your cost to borrow or if your cost to borrow is constantly fluctuating. So let's say you're intending to borrow for a year and the interest rate changes constantly over the course of that year. It's a lot harder for you to quantify the outcomes that you'll be able to get from that leverage. And so knowing your cost to borrow up front gives you a lot of advantages. Could you explain a little bit about, you know, how do you get this collateral ratio to be much lower?

So you have this 110% collateralization ratio. I mean, obviously the lower the collateralization ratio, the higher risk, the, somewhat the system faces both individual CDP holder but also just the system as a whole faces and so what enables you to get this lower rate compared to things like maker at 150 percent or like you know I think some of the other protocols have like even higher like synthetics goes all the way up to like 750 percent or things like that so yes it's really the fact that we can trigger liquidations instantaneously without a need to find anyone to bail out the borrower. So by having a stability pool, which I can maybe briefly explain,

so you can think of it as an insurance pool where people can put their LUSD tokens. LUSD is the name of our native stable coin or token. So any holder of LUSD can place those tokens in that pool any time. Also withdraw them basically at any time. But then when the system needs to liquidate a borrower, which drop below the 110% threshold, then all the system needs to do is basically take out an amount of tokens from the stability pool, which corresponds to the debt that is about to be liquidated, and then basically just burn those LUSD tokens and pay off the debt.

and in return what it also does is that it takes the ether like the collateral held by the borrower's position and it gives it to the pool. Now the pool will obviously be filled or will be filled up by not just one but like many people. So everything happens like prorata like everybody who deposited the share would get like we would lose some LUSD in case of liquidation in proportion to his deposit the share. but also gain ether collateral from the liquidated borrower in proportion to the deposited share. That's the idea. So if I'm understanding this correct, what's happening is there's a stability pool

which allows people to deposit LUSD, and it's basically saying that they get first rights access to being the liquidator of choice of the protocol. So instead of having an auction process, where people are competing for to be the one to liquidate a CDP. It's saying, hey, you know, the protocol has this preferred liquidator, which is this stability pool itself. So what's the benefit of this?

Like, is it to reduce the competition or? So on one hand, you say it's about speed. Isn't this actually scary for the people who deposit in the stability pool as well? Because when you have this like liquidators who participate in an auction, If ETH is flash crashing, you know, they want to actually, or if they think it's going to drop further, they might want to withhold putting in a bid in the auction. But the stability pool is kind of forced to buy ETH and even if it's expected that ETH is going to continue to crash, right? So that's right.

There are risks involved if you become like a stability depositor. But we chose this like 110% by looking at. at the past data also of like not just the ether price, but the ether price that we can get from Chainlink, which is our Oracle. And we just saw that this 10% margin is about the maximum that could happen or based on the historic data that could be like the maximum price change between two like price feed updates. So theoretically, if you as a stability depositor, you're running like, say, an automation tool that would fire sale or sell. like the ether that you get from the liquidated position, like immediately,

when you get it, then you should be able to recoup your, like, loss in the first case. But, I mean, that's just the worst case. In normal situations, liquidations would happen just slightly below 110%, maybe at 10% or 108. So there's still this 8 or 9% net gain, which should compensate for, like, the risk of a loss in, let's say, in a black swan event. You achieve this much faster liquidation mechanism through the stability pool, and it allows for much more capital efficiency, a rate of 110%. So doesn't there still need to be an on-chain transaction to start a liquidation?

How do you kind of incentivize that on-chain transaction, and how do you ensure that on-chain transaction gets into the mempool and gets into a block? sufficiently quickly as well. Yeah, so what's pretty cool about liquidations is that anybody can execute them externally. So basically, we have what's called a liquidation reserve, which basically is, it's almost like a fee, but it's a fee that you get back if you pay back your debt. So, for example, you open a trove, you put aside 200 LUSD or 200 LUSD's added to your debt, but that's kind of like stuck in your trove as this reserve. for a liquidator whenever they actually execute a liquidation. So the other part of that is a 0.5% of the collateral also goes to the liquidator. So if I'm somebody who's running a bot who likes to execute liquidations, anytime I liquidate a trove, I receive 0.5% of that troves collateral,

along with 200 LUSD as a sort of flat rate. But what's also cool about liquidity is that you can liquidate troves in bulk. And because of the way it's implemented, you'll actually save money on gas by doing that. So let's say there's 30 troves eligible for liquidation. It'll actually probably be more profitable for you to liquidate all 30 in one transaction versus trying to liquidate just five of them or something like that. So that's kind of the way it's designed for liquidators who might be running bots or, you know, monitoring the troves. So this is super cool because it's for a stable coin platform pretty revolutionary to have instant liquidations. But for a lending platform like compound, you already have instant liquidations. How would you compare liquidity to something

like compound for instantaneous liquidations? I'll add that I'm not totally familiar with the compound liquidation mechanism. I'm mostly familiar with makers liquidation mechanism because I've been kind of just studying it in a little bit of detail as well as their 2.0 liquidation mechanism that they're working on. But I don't know, Robert, if you have anything to add regarding compound. Yes, I think compounds liquidation works by letting people like repay just a fraction or like the part that needs to be repaid in order to push the collateralization ratio back to a healthy range. So it's like a partial liquidation in a sense, not like a complete liquidation. If I'm not not mistaken. But the difference is, I mean, they still like have higher liquidation ratios than

we do. And so as far as I know. And the other difference is of course like the interest rate that like with compound is again like a recurring fee. And in our case, it's a one off. Issuance or borrowing fee, which like leads to a different like experience as a borrower. Got it, got it. So it's that combination of both instantaneous liquidations and this new way of doing flat fees for a borrow that's unique and kind of uncharted territory to a certain extent. Yeah, exactly. Another thing, though, that I saw that was a bit different about Liquity is that in Maker, dieholders don't have any clear. claim on collateral, or at least until there's a liquidation, and they can buy that. But like, it seems in Liquity, an LUSD holder, which is the stable coin, which maybe we should

have mentioned that earlier, but LUSD is the stable coin in Liquity, they can, at any time, you can redeem your, even if you're not a CDP holder, you can redeem for collateral. And how, so how do you do that? Who gets, whose collateral are they redeeming here? Is all the collateral getting pooled together or do people still have individual CDPs or troves? And so then who gets liquidated? Yeah. So people do have individual troves.

But what's really cool about redemptions is that it's this mechanism almost for like bringing back or bringing debt back into the system and then canceling that debt out. And so the way it works is whenever the system storage troves in this way that it's from the most, risky to the least risky or from the lowest collateralization ratio to the highest collateralization ratio. So when somebody comes along and it's profitable to redeem that LUSD, say there's this arbitrage opportunity LUSD is floating below the peg, they can redeem that LUSD at face value against the riskiest trove or set of troves in the system. So that redeemed LUSD is actually used to pay off the debt of the riskiest troves within the system in return for their collateral. So as a as a borrower, you don't incur a net loss.

You just lose a little bit of your exposure in return for your debt being paid off. That basically allows for people to feel more certain that LUSD is actually, you know, redeemable instantaneously for its value. And how do you kind of ensure that these redemptions are kind of in good faith and are not arbitrages. So for example, let's say that LUSD is actually trading at $1 on other markets, secondary markets, but the chain link Oracle, there's a little bit of lag. You know, it's not instantaneous. And so ETH might be trading at a price that implies like you can make money by doing an instantaneous redemption, right? So is there some kind of fee or how do you ensure that

that that is accounted for, that potential for Oracle lag? Yeah, so that's a great question. That's a point that we have looked into because it is like a potential, like, issue that we need to take care of. And the way we do that is that we have a redemption fee. And the redemption fee, even though it has a decaying mechanism, which means that it goes up with every redemption. And whenever there are no redemptions, it would slowly decay or exponentially decay over time.

but there is like a minimum fee at 0.5%, which makes sure that it can never decay below 0.5. And 0.5% also happens to be like the minimum threshold for the chaining Oracle in order to trigger a new price update. So of course, there can be like maybe like situations with like where the Oracle is a bit lagging behind, even though there is this 0.5% threshold, but it can happen that there's a little bit of is like a higher price hike. So it means that an arbitrage would have this kind of possibility.

If the price goes up faster in reality, then the Oracle keeps track of. And in that case, yes, there can be temporary situations where an arbitrage could make maybe some small profits, but it's not something that is there every time. And it would be, it already requires quite some sophistication. and also like the fact that the fee would already react and it has this kind of memory that it stays high and it goes down. It's not just that it goes up for a second and then it's back to zero or 0.5. Yeah. And I'll also add that default minimum fee also makes it slightly more expensive than uniswap whenever the LUSCE, E, Uniswap pair fee.

So like by default, you'd get a bad deal if you just made that trade because you'd pay more in fees by, you know, redeeming against the system if it wasn't. profitable to do so. This lowest collateral debt trove that's being redeemed against, is there any chance that this reduces the collateral ratio or will it always increase their collateral ratio? So it should always increase the collateral ratio because you are basically taking, like you're subtracting the same constant amount both from the debt and from the collateral. And given that the collateral, and given that the collateral, is higher than the debt, I mean, this would lead to an increase of the collateralization ratio. I mean, if it was a different, it would be different if the

debt was higher than the collateral, which means that the liquidation mechanism somehow wasn't able to liquidate it fast enough, but that shouldn't happen already because there are very strong incentives to liquidate earlier. And also, we are not allowing to redeem from troves that are below 110%. So those troves, would be eligible for liquidation, but not for redemption. So this should never happen. I think you can rule at that. And I think that's actually a good point because there was an example the other day that I was working on where there was a trove that I redeemed against that was right on the border of 110%. So it was really close to being liquidated. And that redemption saved that trove from liquidation,

which is a 10% haircut, right? Like losing your eth exposure is probably better than just losing 10% of your entire position. So in the event that you get redeemed against, in some cases, it's probably a good thing because that redeemer could have just saved you from losing 10% of your whole position. Another mechanism I came across was you guys have this thing called a recovery mode, which is basically triggered when the overall collateralization system of the entire system is below 150%. And so can you describe to us a little bit about how this recovery, how this whole recovery mode scheme works? So, yeah, I mean, it's relatively sophisticated because we have to think in different, like,

cases. But generally, like, it is triggered when the total collateralization ratio of the system drops below 150%, like all the collateral's values summed up divided by all the debt value summed up. That's the total collateralization ratio. So that would trigger recovery mode. And then what happens is that now every trove or position becomes eligible for liquidation if its own individual collateralization ratio is below the total collateralization ratio, which by definition is below 150%, because that's what triggers recovery mode. So that sounds scary, of course, because now suddenly people with let's say only one or a 10, sorry, one.

sorry, even with 140% collaterization ratio could be liquidated, but we do kept the liquidation loss. So you would never lose more than 10% of your collateral. Or like you never lose more than this 10% difference between your collateral and debt. So it's basically for you the same as in a case where you're liquidated, below 110% because you only lose this 10%

above your debt. So we kept this loss and we give a way to the user or to the borrower to get the surplus refunded. So let's say if you were liquidated at 140%, you would basically get back 30 of your collateral. Like you had a debt of 100, you had collateral worth 140,

and you lose 110 of your collateral, and you get back 30. So that's... the way it works, like the main mechanism. Is there an attack here, though, that can be done where, like, let's say I wanted to grief someone and they have, let's say they had 120% or something and I'm a giant whale. Could I, like, add in a bunch of liquidity at like 140% just in order, so much just to push the global ratio below 150% just to cause a liquidation of the person below me?

That's not possible because we prohibit every transaction that would trigger recovery mode. Like every borrower operation that would drop or decrease the total collateralization ratio below the 150% threshold would be banned. And also we have a set of bans and prohibitions during recovery mode. Like we make it so that some operations are not allowed during recovery mode as well as at the cost of it. to make sure that those incentives cannot be easily, like, abused. Yeah. So that whale would have to be very mean and dump the entire ETH price, not just to try to attack liquidity.

They'd have to aggressively dump ETH too. So another thing that was interesting here. So it seems like there are a few risks that will, in practice, push the vault to not be at, a vault or sorry, a trove opener to not be around 110%. So the first thing is like you could get redeemed upon. So you lose your, you know, eith position. And then you might have to, you know, repay your flat fee in order to open it again. The second is kind of this mechanism we just talked about, which is recovery mode,

where the lowest collateralized vaults will be in the liquidation state. And then, of course, there's just traditional liquidation. if you, for some reason, go below 110%, you're also in a little bit of a risk scenario. There is another mechanism, I believe, right? So if the entire stability pool, well, actually, we'd love to hear you guys describe it. What happens if the stability pool runs out of funds

and there needs to be liquidations beyond that amount? Whenever the stability pool is empty or emptied during a liquidation, then what happens is that we basically take the position that needs to be liquidated and chop it up in proportion to all the other borrowers in the system. Like you just look at their collateral, like you have a set of borrowers with their own collateral amounts. And what we do is we basically split up the. liquidated borrower's position both the collateral and his debt and just give those tiny proportional shares to all the other borrowers like every borrower would see a slight increase of his own collateral but also an increase of his own debt and given the fact that we do liquidations just slightly or in normal situations slightly below 110% it means that when you look at the

total net change, it's like a gain and not a loss because you get more collateral value than you get debt from a delquidated borrower. So that's how the system basically ensures

that debt can always be redistributed. We call it redistribution. And of course the collaterals redistributed according to the same principle. Got it. So if there are multiple different vaults or tropes, or troves, sorry, that are in a liquidation zone, right?

And the stability pool has kind of been exhausted. How do you choose or which part of these troves go to different troves that are not under collateralized? Does that make sense? I think I understand your question. But the general idea here is that every trove gets redistributed to every other trove in the system. Like it's an one to all mapping.

So to say, now maybe you're also wondering about the order of liquidating multiple troves which are below the threshold. So there we don't like, we initially wanted to like impose like a strict ordering, but then realize that this wouldn't work because an attacker could then clog the system with a huge amount of tiny troves. So that's why we had to relax like this liquidation ordering which we initially wanted to have, which would have been like that you can. only start liquidating the position or at the trove that has the lowest collateralization

of all troves. Now we relaxed it so everybody can be liquidated below the 110%. And you can also apply this batch liquidation that Colton just explained to make it more efficient. But then everybody would be on the recipient's side. If you are near the liquidation ratio, the, even though you are, let's say you happen be at like 111%, and you had more debt applied to you. Yes, you ended up on top because you have more collateral, but like you could have been added at a rate that actually pushes you below

the liquidation ratio. And isn't that actually, couldn't that cause like cascading liquidations? Yeah, that's a very good point. So we did some simulations and also just some static like calculations, so how that would play out. And we realized that it only really matter. or it only puts you in a position where you as a recipient can all like where you as a recipient would also become eligible for liquidation only if you're already very close like as you said maybe a 11 11 that would be like a range or like collateralization ratio where it could happen that you would become liquidatable just due to the fact that you receive those liquidated shares but for higher collateralized trade. that wouldn't really happen. It's very unlikely if you just look at the numbers. Yeah. And I think even if you're on that 111% threshold, you're so close that if a lot of people are getting liquidated and we're in a phase where we're doing redistribution, I mean, the odds that ETH drops 1% in that scenario is probably really high. So you'd almost like end up being in a

liquidatable position regardless of whether you got the redistribution or not because of ETH's volatility. And so I think what Zubin's question was is how much, like what, how do you decide the distribution of, you know, you're distributing both this debt and collateral to existing trove holders. How do you decide like how much of it goes to which person? I mean, I assume it can't be per trove because otherwise, then I just open a lot of troves just to earn a lot of money that way. Yeah, so the way it works is that it just looks at the collateral amounts held by all the recipients. like by every active trove owner, and it would just give you like a proportional share in proportion to your collateral

as a percentage of the total collateral in the system. Like somebody, let's say, who has a collateral worth $2,000 and another person who has like collateral worth $1,000, now the trove owner with the $2,000 would get like twice the share or $1,000. of the debt, but also of the collateral. Like, it's really proportional to your own collateral.

Does it make sense? There's almost like two tranches. So when it comes to, like, earning from liquidations, right? There's like two tranches, right? There's the people who provide in the stability pool first, and then if that gets depleted, then it goes to existing trove holders. Is there a world in which, like,

no one wants to deposit to the stability pool and everyone would rather just use their capital to provide CD to open more troves and like as long as they can like get other people to not put into the stability pool it ends up this ability pool never ends up being used and it only becomes like this like that second layer trance becomes the primary system so I think as long as the price of one LSD is below one at 110 like which is a hard price floor, sorry, which is a hard price sailing. I can explain it later. But as long as the price is below the one ten dollars, it would mean that basically every liquidation would lead to a net gain for the stability pool providers or the stability providers. So just looking at the incentives, it's not

really conceivable that nobody would want to become a stability depositor because you can, there is free money basically on the table. It's only, like conceivable if the price exceeds 110 because then like this kind of net gain for you as a stability depositor is turned into a net loss. But this is a very extreme scenario which we don't expect to happen exactly because we have this kind of hard price boundary at 110. So isn't there also additional rewards I think you might have mentioned where you get liquid tea tokens? LQTY liquidity tokens for being in the stability pool. Could you explain about that?

That seems very, very interesting. Yeah, so LQTY is the sort of secondary token of the system. It is not a governance token, but it is a token that allows its holders to sort of have a claim on fees in the system. So as you're earning LQTY for being a stability pool depositor, you can then turn around and stake those LQTY tokens in order to earn fees that are accrued from people who are borrowing from the system and then people who are redeeming from the system. So those fees end up going to the staking contract and are distributed amongst the stakers. So this is really interesting.

This is almost like the opposite of the paradigm that's existing right now. Right now we have lots of governance tokens which don't really get cash flows, at least at this point. but this is almost more like a token that does get cash flows but not governance rights. Yeah, and the reason there's no governance rights is because there's nothing to govern. So basically, everything within liquidity is handled algorithmically from monetary policies to everything else. So there's nothing to govern, which is like kind of a blessing for some, you know, like governance processes can just be long and boring.

and frankly, they could suck, and a lot of people don't enjoy them. And a lot of that also has to do with the culture you've built around your governance and stuff. But for the most part, governance can be a really hard problem to solve for. So if you can come up with a solution where you don't need governance and you don't need a community of people pretending they're the Federal Reserve or something like that, you end up in a pretty good spot. So that's kind of our opinion when it comes to that. how do you do protocol upgrades without a governance token? Have you thought about Lido? Liquidity V2. How do you work on

those kinds of protocol upgrades? Yeah, I think our current headspace is that it will be very similar to uniswaps approach where they deploy the system in an immutable way. And if we end up doing a V2 later or maybe a V3 after that,

whatever, we come up with, one, a migration plan and then two, we end up launching a completely separate system. Almost similar nature to what Maker did when they went from single collateral dye to multi-collateral dye or what Unoswap did from V1 to V2 and then eventually V3. Does the liquidity token act as, you know, in Maker, the other purpose of the MKR token is also active as just like global backstop kind of thing. Does the liquidity token take on any such role

in this system or not really? No, it does not. What I'm kind of seeing that what Liquity is doing is that in Maker, it's sort of every CDP is sort of this like standalone unit. And they, there's no sense that there's all these things are actually sharing a protocol. But what's happening in Liquity is it's like, hey, we're kind of taking the assumption that because some people are more conservative with their like collateralization ratio, some people are want to maintain 300% ratios or something like that.

That they're, they're allowing, enabling some people to take on more risky ratios. And it has this sort of somewhat socialization of risk a little bit. But then shouldn't the people who are taking on those higher collateralization ratios be compensated for that? Because so, so currently you're saying that like the liquidation, liquidations are getting given pro rata. So once this ability pool is out, it's getting given

pro rata based off of collateralization amount. Shouldn't there be some sort of weight given as well to the collateralization ratio, where if you have a higher ratio, you should be getting rewarded slightly higher than people with a, even if someone has a higher amount, collateral amount, but lower ratio than you? So yeah, I think that's an interesting point which we also looked a bit into. So I mean, that would mean it would have like a super proportional or yeah, not sure what the right word is like share of the liquidation gains if the stability pool is empty. Now the problem with that is we are bound to some like implementation details around like keeping the troves in an ordered list. Because like it was quite tricky to do all this in a scalable manner. like implementation was not easy as I can tell you because

exactly due to the fact because the troves are so interrelated and interdependent rather than just being standalone beings and for that matter I mean it was just it's not possible to do any kind of redistribution that leads to reordering like that may breaks the ordered list because some troves would then suddenly like get like a higher or lower collateralization ratio than their neighbors in the list. And it turned out that the way we do the redistribution is more or less the only way.

It's not the only way, but it seems to be like there's only a very thin design space that allows you to redistribute that and collateral in an order preserving way. One thing I wanted to ask about shifting gears a little bit. You would mention that LUSD has a price ceiling of. a dollar 10 cents. It seemed like that was one of the weaknesses of Maker in a liquidity crunch is that die can be trading far above its peg. You mentioned that you have like a hard ceiling. How do you kind of create that that mechanism? Yes, so it's basically a simple arbitrage cycle that becomes possible thanks to the 110%

collaterization ratio because when the price of one LUSD is above 110, then you can basically take out the maximum amount of debt that you can get from the system against your ether collateral and sell the LUSD on the market for price that's higher than basically the value of your collateral. So you can theoretically even forget about your collateral because it doesn't matter for you whether your trove gets liquidated or not because you already made a profitable transaction just by opening a maximum collect or a lowest collateralized trove and selling your LUSD. Right.

Aren't there the flat issuance fees that we talked about earlier though that would make it effectively higher than $1.10? So that's right. So it's basically there is this 0.5. percent in like like because we cap it at 0.5 percent as a minimum but whenever there is a situation where this happens then we can be more or less sure that the fee would be around this 0.5 percent because like the way we set the fees or the way that the protocol like sets the fees is by measuring the amount of redemptions or the redemption volumes And redemptions only make sense if the price is below $1. So nobody would redeem is the price is above $110. So that means that the fee would eventually or relatively quickly go back to 0.5%,

which as you like correctly mentioned would mean that the price saling is not exact at 110, but at 110.5. As also with the price lower there, we also have the redemption fee, which makes it a little bit fluffy. So we have both sailings and floors are there like a bit not clear lines, I would say. I guess we should talk then a little bit about how this like these fees are decided. And so, you know, as you mentioned, this is meant to be governance minimized. This is done purely algorithmically. Can you talk a little bit about how about this algorithm?

How is this done? And like how do you have a purely algorithm, like input? into what the price of LUSD is. So the main way is that we can measure like the redemption volumes. And by volume, I mean the volume in proportion to the total supply. Like the total LUSD supply can be set in relation to the redeemed amount or the redeemed amount in relation to the total. LUSZ supply and then based on this fraction, we can determine like the amount by which the fee

needs to be increased. Like we have a base rate which serves as the base this for both. Sorry, do you mean like over like over a window like you're saying like oh in the last 24 hours this is what percentage has been redeemed? Or what do you mean by fraction? So actually yeah. So actually it's like every single redemption is like affects like some percentage of the total LUSD supply. Now what we do is we take the current base rate and we add basically the redeemed fraction of the total supply times 0.5, which is a constant parameter, we add this to the current base rate. So let's say the base rate is at 1%, and now somebody redeems 2% of the total LSD supply, then we calculate 0.5 times 2, which is 1. We add 1 to 1, which gives us like,

two as the new fee. And then it would decay over time. How long is this window for decaying? Yeah. So for the decaying, we just set like a half-life of eight hours. But we are still like tweaking the parameters. So that's something we are still finalizing. But I would say this one will stay probably at eight hours.

So it would just go down by 50% in eight hours. Why use this sort of indirect calculation of like, you know, trying to estimate whether what LUSD is training at instead of just using a Oracle, like, you know, using UNISO-TWOP, or if you're already using ChainLink for like your ETH price, why not just have it also tell you what LUSD is trading at? So this is also a bit like a chicken and egg problem. If you want to use an external Oracle for something that you are just issuing or creating, because, I mean, Chaining doesn't have come with like an LUSE Oracle in the first place. So you first need to have some kind of token that gets large enough to be like have a large enough market cap in order to be sufficiently traded so that you can get like a reliable price from like Uniswap or or an off chain Oracle.

So we think that by measuring our own like transactions or the redemptions we can basically just use what we already have as a signal. We don't need to rely on something where there is just less liquidity. Because, I mean, if only 5% of the LSD supplies, let's say, in a uniswop pool, then it's much easier to manipulate that price because you only need to manipulate a fraction of this 5%. Then in our system, which basically measures redemptions as a percentage of the total LUSD supply.

So we should have a much higher. resilience against manipulation if we use our own numbers. Zuban had an interesting attack that he was thinking about, which is can you hold like whoever has the lowest collateral ratio, can you hold them hostage by like threatening to redeem against them? And like if they want to regain their eth exposure, then they have to go keep adding back in eth and have to keep paying that issuance base fee. And so is this like an issue?

Maybe Zuban, maybe you want to explain the attack better than I did. So it's not necessarily an attack. It was like, can, what I was wondering is like there are some kind of, for being the lowest vault or trove, there are some kind of stresses that you put against the system that you should have to pay for. So I'm wondering like economically if there are ways for the system to extract that from you. So for example, anyone could redeem against like a very low collateralized vault, which is good for a system health, but is bad, very likely bad for that individual because that individual wants that eth exposure.

And so in order to get that eath exposure, they're going to have to pay an issuance fee again, right? So basically, they don't want to pay the issuance fee, so they would pay up to the issuance fee. so they would pay up to the issuance fee to a contract which like threatened to redeem against them if they didn't pay like this this floating rate that was like a that was like a weird thing that I was thinking about does that make sense I can like reiterate what exactly that means so yeah I'm not 100% sure, but maybe, I mean, like the idea here is that you should always be like exposed to redemption risk if you are the lowest trove and the prices below one dollar minus the redemption fee. So that's like a situation where the system where like it's the feature not a bug that

redemptions would happen because that's what we want. We want to like reduce the total currency supply by burning LUSD and by making MESD. and by making it maybe costly to reopen a new position because we don't want people to just being redeemed against them and then immediately reopen the same position because that would be a zero-sum game. So in those scenarios we want that. Now I'm not quite sure whether you're referring to a situation where the price is above $1 or maybe exactly at $1, and now somebody wants to do harm to some trove owner at the lower end of the scale. Yeah, so there's a few things there.

I think like one thing that was interesting that you mentioned, it seems like a big motivating factor for liquidity is for, in order to allow much lower collateralization ratios. But at the same time, it seems like if you're at 111%, 115%, 120%, there are all these things in place to, for example, redemption, this mechanics we, just discussed to discourage you from being that, that like least collateralized or being near 110 percent. What do you think is like an effective rate, right, that people should aim for then, right? Because it seems like, you know, at 111 percent, 112 percent, we want people to get redeemed upon because it increases the health of the system. What is the rate then that you see as like truly healthy for the system to a certain extent. Yes. So I think here we looked a bit at the numbers from like our competitors and we saw that

like let's say the maker system normally hovers around 300% collaterization or at least the single collateral die had this kind of average. Now with multi-collateral it's really depends on the collateral type so it's hard to compare. But for ether I think 300% only, the maker system is not just an average, but it's also something which is relatively safe. Now in our case, we don't need that much because we have this faster liquidation, but what you can see there is that the 300% is basically twice the minimum, like it's two times the 150% minimum. So in our case, that would translate to 220%, which is two times 110.

So I think 220 would be like the lower end. of where we would consider our system to be safe, but we would generally recommend or hope that it would probably hover around 250%, so over time. But people are crazy these days, so who knows? It's hard to tell. Yeah, that's super interesting, yeah. Yeah, and I think it's an interesting point.

I just think it's an interesting point that, I mean, you really get more leverage for the same amount of safety because of the 110% minimum. So that's the important fact to consider. definitely don't want to be hanging out around 110% just because it exists. But if you're a maker user who is used to a 300% collateralization ratio, you can transfer over and hang around 250% and 220% and take advantage of better capital efficiency that way. Right. And it seems like there's a tradeoff there. You get a better liquidation ratio that

makes you feel comfortable as an individual to not get liquidated. But you also have these other kind of variables to think about redemptions, et cetera. So it'll be really interesting to see the dynamic between, like, Maker and Troves on liquidity and see what kind of market forms

between people shifting back and forth. Do you think this will become, like, a much harder or higher mental overhead for, like, people trying to open positions on Liquity? Because, you know, in Maker, I just have to think about the only thing I have to be thinking about really is the ETH price. But now here in liquidity, it seems that there's like all these different things I have to start like paying attention to. I have to

pay attention to, you know, how many people are redeeming LUSD. I have to look at, you know, what is the global collateral like ratio. I have to start paying attention to. What is this like stability pool rate? Because, you know, things in the system act differently. Like this is so much more I have to be paying attention to. And do you think this is a higher cognitive, this? This is a higher, this. going to this cognitive overhead is going to cause UX issues? It's possible, but I also think, you know, users are starting to become more advanced, so they want more advanced access to leverage. I think we've done a pretty good job in explaining how liquidity works, like, within our

documentation, explaining how it works within the UIs we've made available. And I think people are, I think people are going to get the hang of it, you know, like all new things. They come with headaches sometimes. sometimes they don't. And it just depends. Like, are people willing to sacrifice capital efficiency for maybe like one, one or two more things they have to consider, right?

And I even think I'm not totally convinced that redemption risk is like this massive risk that people will have to be thinking about all of the time because it only comes into play within certain situations. And, you know, I would rather be protected in those situations anyway. So I don't think it's something that will be just constantly. prodding in the back of their head as soon as they kind of get the hang of the system. And maybe one thing I would add to it is that we can also think of the markets in, like the boring market in segments. So for those who are willing to be on the safer side,

like maybe have a position that's less capital efficient, those people only need to maintain a collateralization ratio above 150%, which is what they are used to from like using, like EtherA on Maker. So they don't really have to worry about anything else because it's practically impossible that they would get redeemed against. Even if the system enters recovery mode, they would be on the safe side.

It only affects those risk-seeking people who are already, like maybe more D-Gen or more, yeah, just trying their luck. So I think those people, folks, they will need to be a bit more cautious or a bit more attentive to those changes. But, I mean, there are also trove automation tools that we encourage people to build for our system.

And at some point, we assume that this will all be automated. Like, at least the sophisticated borrowers would use those kind of toolings. Carlton, I heard you mention the UIs. So I noticed a lot of your docs, you guys talk a lot about, like, you know, incentivize front ends and things like this. So why, like, you know, I know like a lot of people care about like, you know, decentralized front ends. Like, you know, UtoSwap has a number of front ends and things like this. But like, you guys seem to put a lot of stress on it.

So why is this such an important focus for you guys? I mean, one of them is philosophical. The big, one of the cornerstones of liquidity is decentralization and immutability. And so one of the best ways to achieve that is not only at the protocol level, but at the front end level too, right? Like if you have one access point to your entire protocol and somebody shuts it down or you have like a DNS hijacking or something, you never know what happens these days. All of a sudden you put all of your users at risk. And the other side of it is we're starting to see the market evolve such that interfaces like Xerion and Zappa and Argent are starting to become more popular than the interfaces of the protocols that they're plugging into. So these like third party front ends are becoming a sort of staple in the industry.

And we think that, one, those are important, but two, it's also important to incentivize those. And not many people have really given that a try. We're kind of one of the first ones. I think Yerun is starting to experiment with it a little bit. I know zero X tried it way back in the day, but the incentive mechanism was maybe not completely fleshed out at the time. So I think that's one of the big reasons we want to focus on it is because, one, we can maximize decentralization not only at the protocol level, but the sort of product level where people actually access. the protocol. But two, we can find a way to incentivize this new, like, sort of budding segment of the crypto industry. Yeah, I thought that was really cool. So how does that mechanism work?

Is it like a Dow that's giving Liquity, LQTY tokens? From my understanding, it's not. It's actually built into the protocol. Yeah, it is actually built in. So we have this basically pool of LQTY rewards that is going out to stability pool providers. How that works for front ends is that whenever they facilitate a stability pool deposit through their front end, they tag it within an Ethereum address that's associated with their front end. And it tracks the rewards of the users that they sort of bring in to the system. And it also applies another thing called a kickback rate, which specifies how the rewards

are split between the users and the front end who facilitated that user's deposit. So let's say I have a kickback rate of 80%. I'm a front end. You guys all deposit through my front end. I'll receive 20% of the rewards that you guys earn over the course of your participation in the stability pool. And so that's distributed straight to my Ethereum address as my users earn those rewards. Got it. Would you expect to see kind of front ends competing for rates?

Yeah, exactly. we expect to see the sort of kickback rate to be a competitive market, but also the features you offer your users to be a competitive market. So two things to touch on really quickly. We provide a front-end launch kit, which is basically this cookie cutter template that anybody can deploy just from the command line. It comes with built-in features and stuff. But we also provide a front-end SDK for more maybe advanced products who want to integrate liquidity into their existing system. And so those two types of ways to build a front end will also have a competitive advantage. So if you're launching the cookie cutter thing that we provide and you try to say, like, I want all the rewards, I don't think anybody is going to be excited to use your front end.

But if you provide a custom integration, maybe with some custom tools, then people are going to be more willing to accept some kickback rate that is a little more favorable to the front end. So we expect those two aspects of it to be an interesting competitive market. Another thing I kind of wanted to ask, especially for you, Colton, was what is the growth strategy, right? There's so many moving parts in making a stable coin platform to be highly liquid. And there's so many moving parts for also the borrowing aspect. And you have also a stability pool on top of that and LQTY as well. how do you bootstrap all of these different parts of the protocol and make this a really liquid experience for users?

Yeah, that's a good question. It's something we think about a lot, too, because I think, you know, it's really interesting. The state of the industry now is that you kind of launch this governance token and you expect like a community to just pop up around it. And it's kind of like it's like low hanging fruit a lot of the time these days. And so since we don't have that approach, we really have to focus on building these like paths to success for all of these different users within the ecosystem. And so what I've really been focused on is building this sort of foundation for a community to grow, whether you're a borrower, an arbitrator, a staker, a front end, like, you need all of these different resources to actually succeed. And so

that's where a lot of our focus has been. We think that the front end reward scheme will help a lot on the adoption front, but also on the user acquisition front. And as far as liquidity goes, We plan on providing some LQ2I rewards to an LUSD-E-E-Th uniswap pool to help bootstrap liquidity for the first six weeks. And then from there, we want to pursue ecosystem integrations outside of that that we think will be beneficial for the growth of the ecosystem. So, you know, like curve integrations, maybe shell protocol, et cetera. There's a bunch we've been considering and looking at and more pop up each day. So that's where we've been looking first, focusing internally and making. making sure we have everything set up and ready to go for launch, for a community to grow and thrive and focus on things besides governance.

So they can kind of succeed within the ecosystem. And then we want to start pursuing integrations as best we can in ways that benefit liquidity. One thing I also wanted to ask was kind of how you think about the regulatory question and especially having a token LQTY, which will get cash flows. How are you thinking about these kinds of questions and creating a decentralized protocol as well at the different regulatory regimes that exist? Yeah, that's also an important point for us. And I think you are referring here, or you may be referring to, like, US security laws, security laws, because like there it could matter whether you are issuing a token that has some like reward capture mechanism or not. But I think here we are taking a little bit of a different approach. So what we are really trying to do is making sure that the token is sufficiently decentralized and not just the token, but also the whole system end, including the front ends.

So that it doesn't really matter what your token is because the entire protocol is so decentralized that there is nobody who is like really an issuer. I think that's like the point. Now, I know there are like systems which have like this opinion that, oh, they are just giving out a governance token and that's not a security and that's it. And they are like somewhat more centralized because they have their own front end and they, maybe they even have like majority voting rights or at least like a large impact on the votes through their like own tokens. But I think that's, yeah, that's also a dangerous strategy because I mean a governance token can easily be turned into something that gives. you a cash flow. And I think that's like eventually the goal. I mean, we've seen it with Uniswap that they now have like, or even in version two, they have some kind of mechanism for diverting some income to the token holders, like the unit token holders. So I think our approach

is really about decentralization, which we think is the way to minimize legal risks. And then one other question kind of around this. How did you kind of bootstrap your team and how do you like fund development especially without a Dow that can give you know give rewards to to developers so i mean we are mostly VC funded um so that means we we had a few two um like rounds uh financing which gives us um a nice runway so that we can also like see what we want to build in the future, but currently we are really focused on just releasing and launching this liquidity protocol. And yeah, I mean, we are now not that much worried about like fundraising anymore, or at least for the time being. Yeah, and it helps that we're quite a

small team. There's about eight of us. So it's not like we're having to fund like, you know, 20 plus developers or anything like that. We're pretty scrappy. And so what's the roadmap here. So, you know, you guys aren't launched yet, but should be coming soon, I believe, right? So I think this week. Okay. So if it comes out after today, I guess, we could share that the official launch date is April 5th, which we haven't shared, which we plan to share, actually shortly after we finish recording this. That'll go out today. The roadmap so far has been wrapping everything up. We were able to fit in another audit in March. I think it was three weeks with CoinSpect. And then after that, the plan is to ship it and make it go live. And so where can people learn more and connect with the, and, you know, do you have like some community channels that you want to suggest? Or where can people just go to learn more about the information and launch and all that kind of stuff?

Yeah, definitely. The best place is Liquity.org. And then we also have docs. dot liquidity.org, which we think is quite extensive for mostly an FAQ. And then we also have technical documentation that you could find through that FAQ. So anything you want to learn about liquidity, you could do so through there. Hopefully, if not anybody, feel free to reach out and ask questions and we'll get them added. We also have a community discord, which is relatively active. We also have a Chinese-speaking telegram group that is quite active as well. So if you're

international in that way, feel free to participate there. And then Twitter, we're at Liquity Protocol, I believe. So those are the main sources, probably more to come. But other than that, those are the best ways to keep up. Well, thank you guys so much for coming on and sharing what you guys are building and really excited to see it come live in the next couple of weeks. Likewise. Appreciate it.

It's great to be on the show. Thank you. Thank you for joining us on this week's episode. We release new episodes every week. You can find and subscribe to the show on iTunes, Spotify, YouTube, SoundCloud, or wherever you listen to podcasts. And if you have a Google home or Alexa device, you can tell it to listen to the latest episode of the Epicenter podcast.

Go to epicenter.tv slash subscribe for a full list of places where you can watch and listen. And while you're there, be sure to sign up for the newsletter, so you get new episodes in your inbox as they're released. If you want to interact with us, guests or other podcast listeners, you can follow us on Twitter. And please leave us a review on iTunes.

It helps people find the show, and we're always happy to read them. So thanks so much, and we look forward to being back next week. Oh, my...