EP 295: Compound — An Automated Money Market for Ethereum Tokens

Episode transcript and additional resources, papers, articles and podcasts on the Compound protocol and how it works.

Published in

Epicenter

38 min readJul 15, 2019

In this week’s episode of Epicenter, Robert Leshner, founder of the Compound protocol, provides us with some basic information on how the protocol works on top of Ethereum, how it allows users to lend and borrow ERC-20 tokens with a duration-free interest model, as well as insights into the protocol’s business model and future plans for governance.

Subscribe to the podcast on Apple Podcasts, Spotify, YouTube, Stitcher, or Google Podcasts to never miss an episode.

If you haven’t already, make sure to subscribe to the Friday Newsletter to receive additional links and resources in your inbox.

Additional resources about the company and Compound protocol

Videos & Podcasts

Website / General Info

Blogs / Articles

Developer Resources

Episode Transcript

Meher Roy: So today we have Robert Leshner who is the Founder of Compound. Compound is this fascinating money-market protocol on Ethereum that allows people to make interest on the Ether DAI 0x and other tokens. Robert welcome to the show.

Robert Leshner: Thank you excited to be here.

Meher: So I’m curious to know your story Robert of how you entered the cryptocurrency space and what led you to build a money market protocol.

Robert: So I’ve had a slightly long journey. So I originally noticed Bitcoin in about 2011 while I was working at a wealth management business as an interest rate analyst and economist and I was dismissive at first I said, oh you can’t create money. This will never work and my first attitude was one of hostility, but as it began to prove itself in the wild for a while, I eventually became interested in actually mining and participating in the growth of Bitcoin. I learned a lot I was lucky to have probably broken even on my mining hardware, but it got me interested in what cryptocurrency was possible to create then when Ethereum came out I was also initially dismissive. I thought that it was an order of magnitude more complicated than Bitcoin and couldn’t work either and so I, you know, stop paying attention to it for a while until the DAO hack immediately caught my attention, in some ways I thought it was a positive that you could create an organization on top of Ethereum even while flawed and even while you know, having issues it proved that there was something possible that we could create decentralized autonomous organizations, financial markets, assets and entirely new economic functions and I was immediately hooked.

Meher: And then like a money market protocol, there are not too many entrepreneurs working on money market protocols. So why why money markets?

Robert: Well in the real world, you know money markets were the most boring financial instruments, but I think when it comes to crypto the one of the most interesting financial markets simply because it unlocks so many more applications to use cases. Just having a boring liquid short-term risk free rate, well, it’s not sexy. It’s fundamentally important as a composable structure for other use cases.

Friederike Ernst: So when did you start creating Compound?

Robert: So I started creating Compound in the late summer of 2017. I began raising capital and hiring the team really with the idea of exploring whether it was even technically possible and feasible at the time, you know, not many applications were launched on Ethereum of significant complexity and I wasn’t even confident that we’d be able to technically create a protocol like this

Friederike: Interesting. So can you give us a brief overview of what Compound is and how it works?

Robert: So Compound is what we think of as a liquid pooled money market where there’s an interest rate set by market forces and anyone around the world can participate in this part by supplying assets to it and earning an interest rates or borrowing assets from it and paying an interest rate. The market is designed to replicate the equivalent of an overnight rate in that it’s designed to have no credit risk, and it’s designed to be liquid and it’s designed to fluctuate relatively frequently.

Meher: Just for reference. Could you explain to us how money markets work in the traditional financial ecosystem?

Robert: So in the traditional financial ecosystem money markets are really a set of instruments which have very limited credit risk and that have interest rates that closely mimic the short-term interest rates and it’s really just a series of assets that collectively generate an interest rate that’s overnight in nature for the end users of it.

Meher: What kind of entities need a money market?

Robert: So every entity in some sense our needs a money market. So whether you realize it or not large amounts of wealth are deployed into money markets when they’re not being more productively used elsewhere. You have these long-term capital investments in economies or capital is invested for long-term goals, and then you have the excess, the things that aren’t actively invested the things that are between Investments the things that are mostly cash sitting in accounts. These are the assets that find themselves in a money market earning the best possible rate, even though it’s not going to be a high rate but it’s a way that wealth is passively deployed very frequently.

Friederike: So how does it work in practice? So say I have five Ether sitting in my account? What would I do with it?

Robert: So what you would do is send it through as you would send it to the Compound protocol, you don’t specify a duration that they’re supplying it for and you don’t specify interest rate that you’re willing to accept you’re simply supplying it and earning the prevailing interest rate, whatever that might be and it might be a very low interest rate if there’s not a lot of demand for ether, it might be a high interest rate if there’s a lot of demand for ether, but you’re basically passively accepting the prevailing terms of the market.

Meher: And then essentially there are a bunch of users that are supplying ether to the protocol. So this is some kind of pool of ether that is held by the protocol and then the protocol lends it out to other money market borrowers.

Robert: That’s exactly right and the keyword that you mentioned is pooled, you’re not actually lending assets to another user. It’s not peer-to-peer and that way it actually ensures that there’s additional liquidity because you don’t have to wait for another user to directly repay what they borrowed.

Friederike: So in essence there’s a supply pool and then me as a borrower, I can borrow from that supply pool and the interest rate that I am being charged is determined by what exactly?

Robert: It’s determined by market forces, which are enshrined in an interest rate model. So philosophically there’s really two ways that interest rates would be set they can be set by user specifying the terms that they’re willing to participate in where they can be set algorithmically, you know, given the complexity of developing an on chain system we actually opted for the second approach which is setting interest rates algorithmically and the Compound protocol uses an interest rate model to determine the interest rates at a given time and fundamentally, it’s based on supply and demand when demand to borrow an asset is low interest rates are low and when demand to borrow an acid is high interest rates are high and at any given point in between, given the preferences of the individual users, you get an equilibrium interest rate. If it’s too high, people stop borrowing and it goes down if it’s too low people are attracted to the market and it goes up.

Friederike: So would it be fair in your opinion to say that Compound to money markets is very much like Uniswap to exchanging tokens and that there’s an algorithmically governed automated market maker for the interest rate or the exchange rate and there’s not really someone who’s actually setting the rate other than that automated market maker.

Robert: That’s exactly right. Yes. It’s extremely similar to Uniswap in a lot of ways and that the system functions without an order book at all, you know, Uni swap functions without an order book for trading and Compound functions without an order book for interest rates.

Friederike: But does it also mean that I am never guaranteed a set interest rate neither as a lender nor as a borrower, I can only know what the current interest rate it is, but it could change either way after even a few blocks.

Robert: That’s exactly right. Yes. So that’s the one thing that you’re giving up control of when you use Compound versus a system that has duration, in some sense you’re basically because there’s no duration to your participation not saying I’m going to lend for 30 days or going to borrow for 30 days. You’re basically taking the best prevailing terms every 15 seconds. And so this can fluctuate, as the markets get larger and have our supply and borrowing in them the interest rates becomes significantly more stable, but in the short term, it is a little bit unpredictable. I think over time this will continue to improve and we’ve seen tremendous stability increases as the protocol has existed for longer amounts of time, but you’re right there is no guarantee.

Friederike: Do you think this limits the number of use cases in some sense? Because for instance I wouldn’t use it for financing my house for instance. Right?

Robert: Correct. It’s less useful for certain use cases and more useful for other use cases. So it’s less useful if you’re making a long-term purchase or investment. It’s just a terrible way to finance something. In 2008, you know people who are using variable rate financing for their mortgages, were dismayed when interest rates changed. Variable rate financing is not great for long-term uses. It is great in the short term and it is great for machines and other smart contracts and applications. So we actually think that having this short-term interest rate that’s interactable with smart contracts is actually great because smart contracts are very bad at managing time. They’re very bad at saying do something for 30 days, but they’re very good at saying do something then stop doing. So we actually think that the ideal users aren’t necessarily humans, but it’s going to be other applications.

Meher: The idea here would be that whenever there is a smart contract application that has some access to assets for some period of time they should by default. Just let it out to Compound and make some return on those assets.

Robert: Exactly, a smart contract is able to say I have extra assets that I’m sitting on proverbially I want to earn the prevailing short-term risk free rate. And as soon as I need the assets back, you know, I create the function call to retrieve them, you know with an indeterminate amount of time. It’s really designed from an application first use case.

Meher: Yeah. So in the future there might be some kind of library that I as a smart contract developer just import and all of this functionality all of this interest is automatically made for me from for the smart contract from Compound.

Robert: Exactly and a lot of our focus going forward is going to be creating those easy application hooks so that you know different smart contracts and different off chain platforms can interact with Compound.

Meher: So compound is not risk-free though. Right? Would you would you say that Compound is risk-free or do I incur some risk when depositing an asset with Compound?

Robert: Well, it’s designed to be an approximation of the risk debate. So as with any smart contract platform, there’s always the risk of code vulnerabilities. The protocol’s been audited the code went through formal verification. But as with anything involving smart contracts, there’s non zero risk. Our goal is to minimize that to 0 as much as we can but it is, you know, nonzero it took, you know, probably close to 10 years for people to say that Bitcoin is safe fundamentally at a protocol level and I think it’s going to take the same thing with something like Compound where you know, one month after launch, you know, there’s questions about its code integrity, one year after launch is going to be significantly fewer questions, 10 years after launch it’s going to be almost assumed that it works and it’s been battle tested.

Friederike: Back to ways in which loaning money or tokens can in principle go wrong, right? So basically you need to collateralize those borrowers. So this is something that we haven’t talked about yet. So you need to put up in excess of a hundred percent of the value of the thing you’re borrowing in some other token. So what actually happens if this falls below a certain threshold, is my first part of the question, the second part being if I actually lend out my money on Compound, so if I’m a supplier, am I always guaranteed that I can get back my tokens at any point in time.

Robert: That’s a great question. So there’s really two pieces here, the first one is is there the risk of a borrower defaulting? So the way the Compound protocol works is you have to maintain excess collateral in order to borrow from the protocol, anyone who’s going to borrow from the protocol puts up a multiple of what they’re borrowing similar to how Maker DAI backs and how a lot of other systems work and that creates a significant risk of buffer to prevent users from at any point having borrowed more assets than they have collateral on the system. Compound V1 ran from September of 2018 to present and didn’t have any events where a user defaulted on their borrowing, the excess collateralization at least over the past year was enough of a cushion such that even if the value of what someone has borrowed has gone up or if the value of the collateral has gone down the system itself is still completely safe. There’s always the risk that you know, the market doesn’t liquidate them fast enough. We’ve created an incentive mechanism where the community is incentivized to rapidly liquidate users once they have less collateralization and is required and that system works relatively effectively. We’ve seen a variety of thoughts and software packages and different users participating liquidation process, but it’s incentive driven. It’s not it’s not a guarantee and it’s worked flawlessly for the past year, but it’s not a guarantee that it always works in the future. And the second piece of your question was is there the ability to always get my assets back. So that goes to liquidity. So the Compound protocol incentivizes but also does not guarantee liquidity. So it’s possible theoretically for every single token in a market to be borrowed and when you want to withdraw your supply every single token has already been barred. That’s theoretically possible. What the protocol does to prevent this is it uses this interest rate model and the interest rate model says the less liquidity there is the higher the interest rate is and the more liquidity there is the lower the interest rate is and when liquidity becomes scarce interest rates go up attracting new supply and incentivizing the repayment of borrowed assets and we’ve seen this play out again and again and again in the markets from V1 to present. Whenever there begins to be a scarcity of assets interest rates go up and it attracts new assets and it is a really elegant and beautiful process to watch when there’s a short-term spike do to borrow demand over time the interest rate returns right back down to the equilibrium where it is intended.

Friederike: Is there flow and a cap for the interest rate?

Robert: There is. So the interest rate model does have a minimum and a maximum parameter to it. It’s basically a straight line between the two and there is a cap. So a smart contract can be on chain exactly how the interest rate model works and it is defined, you know in smart contracts and at any point in the middle, it’s defined by the liquidity and utilization of the market.

Friederike: What’s the flow and what’s the cap ?

Robert: So it varies by asset. So right now for DAI the interest rate to borrow ranges from 5 percent to 17 percent and for ether it ranges from I believe it’s 2 percent to 30 percent and all the other tokens are 2 percent to 30 percent with the exception of USDCwhich is much closer to the DAI interest rate model.

Friederike: So how do you set these minimum and maximum interest rates? And what’s the rationale behind that?

Robert: So long term the process is going to be handed off to the community. Our goal is for each community to govern its own interest rate model. So, you know in Compound V2 to we’ve created this concept called C-tokens. We haven’t gone into C-tokens to see how they work yet. But the very simple story is when you supply an asset to Compound you actually get back a token that denotes your balance and these are also going to be governance tokens. So over time all of the different suppliers of any given asset are going to be able to govern the interest rate model for their markets. All of the suppliers of ether are going to be able to set the ether interest rate model and all of the suppliers of DAI are going to be able to set the DAI interest rate model. We’re planning to start to deploy the system closer to the end of the year but until then they are essentially centrally set by our team. They’re designed to be relatively immutable we’re not changing the model. We basically set it at the launched Compound V2 and now it’s really runs on autopilot until decentralisation. But you know, the model was designed by the developers.

Meher: And then looked at Compound as a supply and this was a while back. It was very attractive to supply DAI, but not so attractive to supply Augur. Why do a particular assets behave that way? Why does DAI usually attract a greater interest rate, but other assets don’t?

Robert: That’s a great question and it’s one that we’ve been constantly asking ourselves and we’ve started to do a little bit of data analysis on this. It turns out that stable coins are fundamentally the most desirable assets. They look the closest to money and if you’re going to borrow something you want to be able to borrow something where you know how much you’re going to owe me a year. If you borrow ether if you borrow Augur if you borrow a token of volatile value, it’s uncertain how much you’re going to own a year. So stable coins are just much more attractive to borrow than other assets are and it also goes for use cases people are borrowing DAI and USDC in order to make purchases, make purchases of other crypto, make purchases of real-world assets, just make purchases. Whereas borrowing the token the primary use case is right now to short sell the token. So, you know, you haven’t seen too many people borrowing Augur to create prediction markets that they wouldn’t otherwise be able to afford to create, you’re seeing it mostly for very limited short sale use cases.

Friederike: So in a way when you by DAI and in a way, it’s a short sell DAI, you’re going long ether right? So I mean basically is that a sign of the of the bull market or did you see that in the bear market as well?

Robert: So we actually saw that there’s some interesting correlations between borrowing demand on Compound and the price of assets. When people think that it’s a good time to buy borrowing spikes and when people think that you know, by buy I mean it’s a good time to spend the borrowed asset, it’s a good time to spend DAI or spend USD coin. You see borrow and demand go up. It’s the same thing for when an asset seems overvalued. If for example 0x seems overvalued you’ll see people borrowing 0x and sending it off to Coinbase or their favorite exchange to sell it. It’s when people think something is overvalued they borrow it knowing that the liability goes down in the future relative to other things. And so it’s the same thing for stable coin, when everything else looks cheap the stable coin looks expensive and that’s when you see borrowing demand starts to spike.

Friederike: Super interesting. So what was the rationale between not actually putting the base rate at zero. So I mean, why is that why is there a couple of percentage floor implemented? Because it sounds like you guys are big believers in the markets and this seems like something that in principle the market should be able to fix no?

Robert: Yeah, you would think that the models theoretically start at zero percent but we know that the model today is designed to find an equilibrium and the model doesn’t have to start at zero the model can start at two percent for most assets knowing that there should be some nonzero cost to borrow the asset otherwise incentive to actually start to break down a little bit the other way. If it’s too easy to borrow an asset eventually it creates perverse incentives. And so we wanted there to be some floor in general for borrowing costs just because we think at least we’re healthier model, but when the community takes over the interest rate models will see people potentially set the floor at 0 that will be exciting to watch.

Friederike: Yeah, that would be super interesting. So I mean if you look at if you look at the legacy banking world currently at least here in Germany, if you take out a loan against a house, you’re actually looking at the interest rate on the order of 1% fixed for 10 to 15 years or so. So, why are the interest rates you see on Compound larger than what you see in the legacy world.

Robert: That’s a great question. I think it comes down to the fact that Compound fundamentally allows you to borrow assets using other crypto assets as collateral and this opens users up to non-bank lending and it really is a great service for people who aren’t currently a part of the normal banking world who don’t have access to one percent mortgages to borrow. Houses are very known and understood high-quality collateral, crypto is not considered to be that high quality. And so the interest rates are going to be higher. And it’s also because it’s open to users who otherwise wouldn’t even be able to borrow. If not for the crypto that they have.

Friederike: That’s a great pitch but do you think that actually read that is informed by the people who are currently using Compound are those people who wouldn’t be able to get a mortgage on a house?

Robert: Well, it’s people who wouldn’t be able to take out a mortgage on crypto. I’m if you’re buying a house, then yes, you know, it’s very easy to finance a house because the house has you know, a very predictable value but crypto’s not there yet. And so it’s people who are using crypto to be able to purchase, to borrow assets to purchase other assets with and so it’s a much higher velocity borrowing. You’re not locked into a 10-year mortgage. It’s extremely short term and it’s based on the value of your crypto. And so I think it’s definitely a different user base of different use case than mortgages.

Meher: Let’s get into these C-token. So what are these Compound tokens? And why did you implement such a feature?

Robert: So C-tokens are really a tokenized fungible representation of the balance that you’ve provided to Compound. So when you supply an asset, let’s just call it DAI, to the protocol you receive a C-token that represents your balance and the interest that you earn is actually represented by the C-token increasing in price relative to the underlying asset. So when you supply DAI to Compound we give you D-DAI and over time that C-DAI that you hold in your wallet becomes worth a slightly larger amount of DAI and you know over time whether it’s a block an hour a day a week a month seven point four hours that C-DAI is convertible into more underlying DAI. This also unlocks tons of other use cases. So it allows us to have a governance token for each market right advocate so that C-DAI is also going to be the governance token to control the interest rate model for DAI and it’s also a transferable asset, you know the example we like to use is you know, it allows you to take your balance and do more with it. In the original version of Compound once you supplied enough to the protocol, that’s it. You can’t do anything else. You just sat there you had a hot wallet connected to the internet and you just had to you know, keep it that way, with C tokens it unlocks more use cases, one is you can actually send them to cold storage. So this is my personal favorite use case. You can basically interact with the smart contract, you know supply an asset to a market where if you earn interest and then take that representation of your balance and send it to a cold storage offline address that’s never interacted with the internet before and you can fundamentally be earning interest will cold storage and that’s super powerful. It unlocks more programmability from other smart contracts and other applications other smart contracts can control compound balances, which they couldn’t do in version one and it just opens up like this entirely new basket of opportunities that we haven’t even conceived of yet. We think this is the biggest upgrade from V1 to V2.

Friederike: So basically in the traditional word when you loan out something the thing that you’re loaning out is gone, right so someone else has it and now on compound when you get back, you know Compound whatever you lend you lent out you actually have a fungible token that you can still use. So it’s like you’re only lend out part of what you had because basically the C-DAI C-ether or whatever you have now is still transferable and it’s worth something and that to me sounds like a fundamental game changer. Do you feel the same way? And if so, what do you think would be the effects of this?

Robert: Well, we do think it’s a game changer. I think it’s so early. We only launched Compound V2 less than a month ago. I think it’s actually almost a month to the day actually, and we’re just starting to see developers experimenting with C-tokens. I think in a couple months we’re going to start to see fabulous new applications built that we couldn’t even conceive of today and I’m just eagerly awaiting watching where the community goes with this.

Meher: Me personally, it seems that we are almost going to this like internet of composable risks, so when I have ether I can put the ether up into Maker and get a derivative asset which is the DAI. Now, of course the DAI has some additional risk on top of Ether because now this DAI is now dependent on the Maker system, the well functioning of the Maker system. Now I have this DAI. I can put it into Compound and I can get Compound DAI out of it. But now Compound DAI is exposed to the risks inherent in Maker as well as Compound right and now maybe imagine like this future version of Truebit or some system like that where I want to let’s say run a verifier some kind of node and I put Compound DAI as my stake in that node, and then I run that node and that node makes me some interest in a system like in an off chain system. And so when I do that I could issue the third asset which is like staked Compound DAI. So if you think of stake Compound DAI, it’s like this composite asset with three risks or four risks almost, there is a risk of Ether, risk of Maker, the risk of Compound and the risk of whatever staking solution you are using to stake Compound DAI. So you have composed four risks together you have like and you have like multiplied those four risks into an asset and you are earning in like returns and all of those four assets at the same time, but you’re also exposed to the risk of all of those four assets. So it almost feels like we are moving into this world where risks effortlessly composed together and create like synthetics that represent all of those risks as a bundle and all of those returns as a bundle.

Robert: Absolutely. I mean, I think that each of these layers when composed together can either add or remove risk, right? So you’re adding the risk of each platform individually. When you create a system that you described you have the risk of Maker and there’s risk associated with Maker you have the risk of Compound and there’s risks associated with Compound. There’s the risks of Truebit and there’s you know associated ones, but they can also offset each other interesting ways where it’s possible that two layers in conjunction are actually a better asset with less risk than either them individually. A great example would be you know, if there’s a layer of this just buying insurance. It’s possible that that insurance is a waste of money and you know the event never occurs, but by combining it with Compound, there’s income that offsets the cost of insurance and it’s possible to the net product of those two layers is less risk and more upside. And I think all of this is in just such an early chapter that we’re going to see, you know, a thousand new experiments proposing different layers together and experimenting with the use case to come up with systems that are fundamentally creating value. And that’s going to be really exciting to watch.

Friederike: Do you think this will change the concept of what money is because basically to us money something that’s fungible, something that has no inherent value but something that you can exchange for a lot of other things, but now that you can actually lend against anything for anything else more or less or at least I guess that’s the future vision, do you think this somehow makes money as a fungible thing obsolete?

Robert: I don’t think it makes it obsolete. I definitely think that you know, we still think of money as a sort of like base unit right where we still think in terms of euros and dollars fundamentally and even if we’re using crypto assets to exchange value, I think in the back of our minds, you know, we still go back to euros and dollars and so it definitely creates more money like properties for every crypto asset, you know, eventually, you know Augur tokens are a little bit closer to money than they were, you know a few months ago, but I don’t think it truly puts them on the same footing as money, it definitely just makes everything a little bit more fluid and liquid and you know easier to use.

Friederike: Many things indeed are much more fluid than previously. So basically if you if you borrow against something the you you run the risk that the interest rate on that actually changed substantially. So am I warned in anyway if I borrow an asset and the interest that I’m being charged just increases by a large amount.

Robert: So the base protocol is just a series of smart contracts. And what’s great about having an open protocol is that anyone can build new experiences on top. What you just described as an awesome product idea and I hope that someone in the community out there builds an application to interact with the Compound protocol that does exactly that, that builds in offline communication systems with you maybe through text message or email alerts that does something like this. The base protocol doesn’t, it doesn’t have the sophistication of product. It’s really just fertile ground for developers in the community to build experiences like that. It starts off as very, you know, simple series of smart contracts and eventually will be, you know, a beautiful series of applications like you just described.

Meher: So what are some of these statistics around Compound, how much has collateral been put into compound, how much has been loaned out? How much has been borrowed etc?

Robert: Yeah. So in terms of the overall numbers, you know, we host a dashboard at Compound.finance/markets with all of the real-time information in the protocol and one of the best parts of a transparent series of smart contracts is that you can view this information on chain, you can audit the information, you can inspect it, manipulate the data in any way you want because everything is available for you. Since launching a month ago the markets have grown in size. There’s currently as of this Monday morning, 42.3`million dollars of assets in the protocol that have been supplied and there’s about 10.1 million dollars of assets in the protocol that are actively being borrowed. The largest market is ether by far, ether is most frequently used as collateral and the most popular borrowing assets are DAI and USD coin.

Friederike: Last night when I checked this you could borrow DAI for 13.5% right? It’s probably still similar. If you actually take out a CDP on Maker you’re charge 16.5% stability fee. Why would anyone actually go to Maker, why don’t people just borrow this on Compound?

Robert: Well, the advantage that Maker has is that Maker has a community of stakeholders to MKR token holders that are incentivized to use Maker in to grow the ecosystem and to shepherd it along. If you’re an MKR token holder, you would probably rather use Maker then Compound. I know a lot of Maker token holders love using Compound but you know, I think there’s an economic incentive to use Maker and pay a higher interest rate than something like compound at the end of the day. I think there’s always going to be users who prefer to use the Maker ecosystem. They’re the ones fundamentally creating DAI and for a lot of users, you know, a small difference in interest doesn’t matter. They don’t have to be a complete interest rate parity between the two systems, you know Maker could still be a little bit more expensive and I think the system still works. What we have seen is that the interest rates on Compound and the Maker stability do move relatively in lockstep, it’s not a perfect correlation, but they’re definitely positively correlated.

Friederike: Interesting. So currently you allow borrowing against six tokens, right? So Ether, DAI, USDC, basic attention token Augur and 0x, and basically the soundness of the operation depends on these tokens. So basically if these tokens are of such a nature that they crash in price and predictably, or for instance, if you list some sort of Shitcoin that is artificially inflated and borrowed against and then the market collapses and the borrower defaults, I mean that that would be a bad situation. Right? So basically we saw this recently on Poloniex, there was a token listed called clam and clam was clearly pumped and borrowed against and then the market collapsed and the Bitcoin holders pool took a severe loss on that one, so what’s your governance solution for deciding which markets are valid and which ones to support in essence on Compound?

Robert: Yeah, so there’s really two pieces to this. The first one is that in Compound V2 there’s no longer a global standard calabro factor, which represents how useful collateral is. In the very first version of the protocol for simplicity and to be able to ship the protocol one of things we did is we said all assets were the same, you can borrow against any asset in a similar fashion. Everything is equally useful as collateral and this was a simplifying assumption which we knew was a starting point. And so we only listed five of the largest and most liquid assets that were least likely to be, you know pumped and then dumped. In Compound V2 we actually broke that simplifying assumption and what we did is we said that each asset has its own collateral that represents how useful it is as collateral based on the liquidity and the volatility of the asset. Large liquid assets like ether are great collateral and clams which are extremely thinly traded low market cap assets are terrible collateral, but in Compound V2 the protocol can actually support assets that aren’t just the absolute largest and most liquid they’ll just have a lower collateral factor. So, you know with ether you can borrow 75 percent of its value. One day the protocol might be able to support something like clams and you might be only able to borrow five percent of its value because it’s thinly traded. The protocol can support every asset because of this variable usefulness of collateral factors and the second piece of this is you know, we’re actually allowing the community to select which assets the protocol supports. So we like to call this semi centralized governance, at the end of the day, you know, the developers still control the admin of the protocol, but the important decisions of what occurs what assets are supported when and how we like to give to the community to vote on. So, you know, we initially held a vote for which stable coin to list, you know DAI came in first place, USDC came in second place by just a hair and everything else lost dramatically, and we basically like allow the community to express its views on what to support. The same process is going to hold going forward, you know, if the community thinks that we should list clams for some reason I’ll think they’re crazy. But you know, that’s a very strong signal table. It doesn’t mean that it has to be useful as collateral and you know, and as I can even be supported without it being collateral at all, we can create a market for clams. You just can’t use it as collateral. You can borrow it took a keen interest rate. You can theoretically actually have the Colony X style Market, but without the risk and that’s something we’re really excited about.

Friederike: So that’s super interesting. So what’s the current governance model? So how does the community actually voice their opinion on which token should be listed or not listed.

Robert: So we actually have a page on Compound. We’re going to be bringing it back shortly. Basically, what we’re going to do is we’re going to freeze at a certain point in time a list of addresses that were users of the protocol. So when we start to vote will say these, you know, 3,000 are able to vote and then each one gets one vote and we basically allow them to vote on a list of assets, you know similar to how some exchanges have let people vote for which asset to list and it’s a very similar process. People sign a transaction with their key basically saying, you know, I vote for this asset, but the whole thing is productized and very simple web interface where you just click a button and it assigns the message and it says I vote for Maker to be the next asset.

Meher: Many of the governance questions around Compound the emphasis is to have users vote like the holders of C-tokens can vote and do something or the users of Compound can vote to list assets. Why push these decisions on the users rather than let’s say create a Compound token and have the holders of that token be responsible for these decisions, are users the actually the best class of people to be making these decisions?

Robert: I mean that’s a fantastic question. The answer is we’re not sure yet. I mean it’s still so early in governance. I think Maker is the first example of a widely held governance token that gets used by the community and we’re just watching this really unfold. I’m extremely excited by Maker and that governance token, but we don’t want to rush into determining the long-term shape and form of the protocol until we have a lot of conviction on what it should look like, you know in some ways., yeah, maybe the system would run better if there was a token, but we haven’t made that determination yet.

Friederike: So basically make us a formidable protocol but it’s well known that there are a small number of Makers who have a lot of sway in where the Maker ecosystem are going to navigate right? Is your way of implementing this user voting a way of safeguarding against that and do you hope that this will translate into network effects for Compound?

Robert: Right. So I definitely think that it helps prevent against a concentration of power but it’s also, you know more advantageous because it allows us to actually change how we allow the community to express itself before deciding on any specific path. You know, once you have a governance token, you know, it’s distribution its ownership, the concentration of power is relatively permanent and you can’t go back from that. You know, we’d like to start off with a much more set series of lightweight approaches that we know can evolve that we know can change so that we don’t wind up with something that’s permanent. You know. The last thing you want to do is, you know, like distribute something of value or wealth and power and have to change the model later. We’d rather, you know, conduct a series of experiments transparently, you know hand-in-hand with the community figure out what works and then eventually slowly begin to enshrine that in product and economics.

Meher: So currently one of the big governance acts that need to be done is the is the setting of the interest rate parameters right, and this is done by your firm like the firm presumably controls some multi signature account and that is able to set these interest rate parameters that are used by the protocol.

Robert: Yes, that’s right. So we’ve designed the interest rate models that the protocol uses at launch these are going to be relatively rigid models until we hand off control to the community. I would like by the end of the year to be very close to having each C-token control its own interest rate model.

Meher: And so how does that work? So you have a group of ever-changing token holders and they need to set essentially two parameters on the on the interest rate model, which let’s say is like the floor and the ceiling of the interest rate for approximation. How does a group of ever-changing token holders decide on these two important parameters, it seems like an unsolved governance problem.

Robert: It is, that’s why we want to be careful with it. The thing that gives me confidence is that the incentives are in place for the C-token holders to set the correct interest rate model, assuming that they’re selfish, you know, if you’re a supplier of capital you want the highest return possible and the highest return possible comes from the most amount of usage times the highest interest rate. The interest rate is too high people won’t use the product if it’s too low you don’t make enough money and there’s actually, you know, an equilibrium to even for the C-token holders that they’re incentivized to do this correctly. I think the model can be simplified a little bit where they just have to express a single number. What’s the you know slope of the curve so to speak, we can create it where there’s just a single parameter that they have to vote on and from there it could be as simple as a medium. It could be something more complicated, but I think we can boil it down to a relatively usable experience and it can start off with just one market being governed by the community. You might say Augur has an interest rate model set by the suppliers of the token and you know continue to have the other markets using a semi essentially set interest rate model.

Friederike: Can you envisage a situation where basically, they’re different pools with different business logics as to which collateral to allow for lending out a certain a certain asset. And I mean basically the interest rate will inform itself from that basically from knowing what the quality of the asset is that is permissible. Do you think there’ll be different pools with substantially different interest rates that allow different collaterals?

Robert: Well, we actually want to take it one step further so really long term and in the governance, what we’d like to do is allow the C-token holders themselves to say what’s acceptable collateral to borrow from their market. So instead of having a globally recognized collateral factors each market is going to be able to express what’s acceptable collateral to borrow that asset and they’re very incentivized to handle that correctly. And so a great example of this is, you know other stable coins could be even better collateral for borrowing DAI then ether is just because there’s more price volatility between ether and DAI. It might be that USDC is you know hyper good collateral to borrow DAI and other assets less so and so when you actually allow the community to express this granularity, I actually think you get a more efficient system because each market knows how it works and they can set their own risk parameters and they’re incentivized not to accept clams as collateral. And so I think long-term this goes hand-in-hand with just completely decentralizing the governance and allowing each market to really control its own destiny.

Friederike: I think there’s a larger question here of whether people by and large car enough to make these decisions or whether it’s fair to have this democratic process because for financial decisions, it always seems to be the case that there are few people who care a lot and then a lot of people who don’t really care.

Robert: Yep. Oh, I agree and it’s possible that we also, you know take the lessons of other governance approaches and have a delegated proof of stake type model where people just vote for who they want the chief economist in their market today where the DAI holders elect a Chief Economist on the sort of majority rules, you know way and then that address gets to set the model. Maybe one day we’ll see, you know public elections held for the Chief Economist of each market. There’s going to be someone who’s an expert at Augur. There’s gonna be someone who’s an expert at DAI and you’re going to see, you know, experts start to control the models, you know through a delegation of the C-token holders. Maybe it’ll be exciting like that. We’ll see.

Friederike: Well, that sounds great. So there’s one last thing we haven’t talked about yet in terms of the protocol itself. So there has to be a price feed right that actually informs the decision of whether liquidation can be forced or not. How do you actually obtain this price feed? And is there a way to manipulate it? So basically, I mean this is seen even in legacy markets that you know, you have pump and dump schemes to force liquidation through margin costs. And I mean even even even in legacy markets that are tightly regulated we see this constantly. So what’s your take on this, is it a bug or a feature and has Compound a way of handling this?

Friederike: In the short term the price feed is relatively simple. It takes an average of prices across Coinbase, Poloniex, Bittrex and Binance and post it on chain. And there’s some safety mechanisms hard-coded into the smart contracts to prevent, you know, extreme deviations and price in the short term unless there’s human manual override and approval. It’s a relatively simple system. It still requires there to be an organization, which is ours that maintains a system but long-term, you know, our goal is to move to a truly robust decentralized perpetual on chain system that’s not susceptible to really any real world flaws and you know in the next couple of months were going to be announcing the next generation of how we handle price feeds on chain. We think that this is a great area for improvement because what we want at the end of the day is for us the developers to be able to disappear entirely and the system works, it’s perpetual it’s robust, and it doesn’t need any involvement whatsoever from any of our developers.

Meher: Any comments on how this new price system will work?

Robert: We’re not going to reveal it yet. But I will just pre-announce it here that something new and big is coming soon.

Friederike: So how is Compound currently financed?

Robert: So far Compound is a venture-backed business. We look like a very traditional Silicon Valley tech company in that there’s Venture Capital firms that are providing as capital to develop this platform. We have not gone the route of creating a token who are conducting an ICO and we’re following a very boring path so far.

Meher: And I’m actually curious about some of the network effects around the Compound protocol. So are there any network effects? And what is the nature of those Network effects?

Robert: Well, I think it’s very early in the story of Compound. I think you know, it might take a number of years for us to truly be able to answer that question. It looks like there’s network effects, the bigger the markets the more stable the interest rates and the more stable the interest rates the easier it is to use, you know, we’ll see I think the biggest network effect is going to come into play when you know, we have more markets and there’s more utility that has created by having different pairs of assets that you can use. I think if there’s 20 assets in the system works, it’ll have more of a network effect on after 6 assets. And so I think it’s early, you know, I’d love to come back to this question in a year and we’ll compare and contrast today to then?

Friederike: So what’s currently your business model?

Robert: So right now our business model is to develop extremely widely used infrastructure for decentralized finance and figure out the business model later. You know, one of the things we are really focused on is creating widely used dependable safe financial infrastructure with the Compound protocol, and you know, our vision is that if you know this can power a trillion dollars of economic activity, there will be a very big business model and it doesn’t make sense to worry about until we get there.

Meher: But my understanding is that the Compound protocol does charge a spread between the lenders and borrowers and that is what already constitutes a business model today.

Robert: Yes, that was the case in V1. So in V2, there’s still a spread. It’s a percent of interest that set aside into our reserve. But this reserve is to further decrease the liquidity risks of liquidation. It’s basically a reserved that’s just held by the protocol. So there’s still a spread there, but that’s not necessarily the business model.

Friederike: There’s other landing protocols such as dYdX and Dharma, how would you say does Compound compare to them?

Robert: Well it’s interesting because all of them are cousins. All of us are exploring decentralized finance together. All three products are fundamentally different, you know one is a peer-to-peer lending protocol that all of the order matching is handled by an organization. One is a margin trading protocol that uses pooled borrowing and lending facilities very similar to Compound but it’s the foundation for a margin trading exchange and then there’s Compound which is probably the simplest of all three and simplicity can be an advantage or a disadvantage, we’ll see, but all it is is a pooled approach without exchange case capabilities. And so I think all three are very different products taking very different approaches, I think Compound is you know, really exciting because it’s so simple, I think it makes it easier to integrate into other smart contracts, but we’ll see you know what this looks like in year two.

Meher: Have you ever had the pressure for any of interest rates to be competitive against the other protocols for example, dYdX might be offering a slightly higher interest rate than Compound and then you needing to change your parameters to accommodate that?

Robert: So our models are relatively, you know stable, we’re not like, you know, pulling the lever every time, you know, something happens when the market, you know, it’s an interesting model that sets the interest rates not a bunch of people and so, you know, we actually don’t look at other projects on a day-to-day basis. That’s just not how we think about it. You know, they might look to Compound is sort of like the base rate because it’s by far the largest market and sort of plan around Compound but Compound doesn’t plan around other projects.

Friederike: You said earlier that you would like for Compound to be something that other companies build on top of, are there companies already building on top of Compound right now?

Robert: There’s a few and we’re starting to see more and more developers building on top of the protocol every week. We have seen a series of projects launched sort of in tandem with Compound over the past couple of weeks. We’ve seen Compound protocol integrated into a project called Open which is a decentralized exchange. We’ve seen CDP Saver integrate Compound as well as instadam being able to sort of interact with compound alongside Maker in both cases. We’ve seen a project called Zerion build a really beautiful web interface around the protocol. We’re starting to see more and more integrations of Compound into other applications and systems. I think, you know, give it a year and you can see, you know, a pretty vibrant ecosystem spring up around Compound, at least that’s our goal. And our goal is to help any developer who wants to build with Compound, you know, we try to be, you know, very supportive of other developers with you know, engineering resources, mentorship, design and just you know, anything we can do to help developers we will.

Friederike: So what for you would be the the key metrics for success for say the next year or five years or all of Compound’s future?

Robert: That’s a great question. We actually measure success in terms of the number of applications that enable, you know, I think at the end of the day, you know Compound long-term isn’t going to be used by users going directly to the Compound protocol. I think over time the Compound protocol is going to be a base layer for other applications to be created and users aren’t going to have to fiddle around with the compound protocol. They’ll just go to other applications. So for us the success metric is the number of experiences that we enable to get built, you know, pretty soon. we’re going to start to, you know, really start to reframe how we show off Compound, to not be Compound but the projects built on Compound and that’s that’s how we think about success.

Meher: Very interesting. So Compound will sort of recede into the background over time.

Robert: That’s the goal. I mean, I think success really looks like Compound being a part of many other applications, but on its own not being something that you have to use directly, being able to use Compound through other services.

Meher: And so what are some of the cool features apart from the price oracle that did your team is working on?

Robert: So we’re working on a longer term, we like to think of it as you know more decentralized oracle system, after that I think we’re going to be really focused on building out is you might call an SDK’s and API’s for other developers. It’s less about changing what happens on chain and more about enabling new use cases and we’ll be doing that in parallel with transitioning governance towards the C-token holders. So, you know at the end of the day we become less and less important and the community becomes more important.

Friederike: That’s maybe a lovely note to end on. Thank you so much Robert for coming on the show. It’s been super interesting and I look forward to seeing what Compound will be up to.

Robert: Thank you it’s been such a pleasure to join you on Epicenter. I can’t wait for you and the rest of the community to follow Compound’s progress and it’s been great to be on the show.

Meher: Thanks a ton for taking the time.

Robert: Likewise. Thank you.