The Metamorphosis DeFi Lending Needs.

Transcript

[Music] foreign [Music] I'm allowed to go excuse me is it good okay go so thank you everyone for joining today um so ample from mofo and today I'll be talking about decentralized lendering and so together we'll be exploring what's the current state of decentralized lending and what are the different paths forwards in order to improve the state of the space so the agenda is going to be the following first we'll be discussing a little bit about Morpho and the currency of Morpho and then we're going to introduce uh

what we call the two pass for defy decentralized Brokers versus protocols and finally we'll talk about the midterm office's default lending needs so about Morpho very quickly so mofo is what we like to call the most efficient and liquid decentralized learning protocol on ethereum so um it's called morph Optimizer for the current version it provides the same liquidity as other compound the same risk parameters but with better rates in total we have something like 900 million dollars deposited in the protocol at the moment and we're the third largest on on ethereum so

the problem that we're going to tackle today is that morph Optimizer in its current form is very limited and after a year building on top of avian compound because as you might have understood we are a landing protocol but we're also built on top of existing Landing protocols we realized that D5 lending was really limited and for three different reasons the first reason is that you have limited efficiency so you know more for optimizer's purpose is to improve the interest rates of Aven compound because they have large interest rate spreads between the landing API and the borrowing API so you have a rate inefficiency but you also have leverage like inefficiency like for example you can't take as much leverage as you can take on traditional Finance for us there are too many trust

assumptions like we've seen countless hacks in the last like two three years a good chunk of them were Landing protocols due to small contractual risk economic risk will come back to that and finally for malfo Optimizer we just see too much capitals like lending is to D5 lending is still nothing compared to like the trillions of dollars that are traded every day in the over collateralized Landing markets of traditional finance and so we literally spend the last year thinking about what could be the next way for decentralized Lending and so one of the first things we realized is that there are literally two ways defy lending could go

you have the decentralized broker's path and you have the protocols path and this does not only apply to D5 lending it applies to D5 as a whole so decentralized Brokers so decent price Brokers are literally very similar to what you have in traditional Finance hedge funds Brokers basically people that manage your money except here it's decentralized so you have an open infrastructure and you have decentralized governance so the obvious examples of those are like the maker of the world the Ave of the compound of the world are an open infrastructure so they run on Smart contracts everybody can actually like audit what's happening and you know changes in their protocol are transparent and indecentralize

you can recognize those patterns through mainly two things the first thing is they're smart contracts so the smart contract of decentralized Brokers is very product oriented so what I mean by this is like literally the code run by you know D5 platforms such as Ave or maker is quite similar to the one of like a web 2 startup in some sense so you have a ton of arbitrary choices in it and you have a very vast you know code base and the purpose is really to have as much features as you can and as much up in your nation as you can in order to you know provide a very passive ux for the user and the second pattern from which you can recognize decent price Brokers is

reliance on statistics and economic model in general so usually not everything is done in the code so you should look at maker not everything is done in the code like you have the Dao that is handling the all the risk management part and risk management in D5 is often you know the part that is not handled in the code directly but rather through the dial and so you know there is this piece from Jesse well done from 2019 which you know divides protocols into two categories you have the incomplete platforms and the complete platforms complete platforms are more like the uniswap Canada of the world where the code handles everything and the incomplete platforms are the platforms where you're not successfully able to put everything in the code so

you let your Dow basically manage those parts and this model is about to the product is opposed to the protocols model so we Define Protocols are really primitive something that is independent of anything not dependent on a dow not dependent on whatever uh external part is and lays out a foundation for like future like abstraction layers to be built on top and I think the best possible way to think of a D5 protocol but a natural protocol is actually to you know make the parallel with internet protocols like HTTP SMTP are actual protocols there have like multiple uh properties that I'll be describing in a

moment but the first and most important one is you know being trustless so you have literally no dependency on anything just like okay there is just the code that you have to trust and the e-sharm or the blockchain underlying that you have to trust and otherwise you don't have to trust anything so usually it comes with immutability code and you know different different patterns the second part is being neutral or unopinunated and basically it's about providing a oh well is there a Neko or something or is it just me okay so basically it's about providing a layer that does not take any choice for the user such that anyone is able to use

the protocol even though it's super primitive to you so the good example of that is the IP protocol itself before IP protocol was actually a thing so the networking layer you had several different other protocols that were much more opinionated you had net bias you're not you had Apple talk Etc and what ended up being you know the dominant primitive was actually the most primitive the most in opinionated one just because everyone could be connected to IP and and so we believe there is a certain category of D5 platforms and the greatest example of all in in that category is probably uniswap where you have the least number of dependencies and the least you know

amount of opinion and you could see that through Unison V1 UNICEF V2 unit sub B3 units of before like unisoft before is really about removing as much up in your nation as possible removing features charges like the smart Contracting itself does not make any choice so okay so if we apply this model to Lending well let's talk first about decentralized Brokers so decentralized workers again in the case of lending would be the Aveda compound the Euler of the world where the risk management is handled by the Dow and so what's really awesome about this model is that the Dao in the case of Harvey you know there is a protocols with five billion

dollars into it and the Dow is going to monitor the 500 risk risk parameters of the protocol so at the Ave protocol has more than 500 risk parameters and every user of the IV protocol has to subscribe to this model but they don't have to care really about it and so what's nice about it in some way is that all users are treated equal so that gives access to huge amounts of liquidity so with a decentralized broker model you're able to fungibilize all the liquidity of all the users such that borrowers can come in and borrow straight away you know tens of Millions with their capital and the second really good thing about decent trash Brokers is that they

abstract all the complexities away it's just that when you're using Avid you don't have to care about you know the liquidation threshold or the collateral factors and or you know all those kind of things are obstructed away for you because you trust the Avid Dao you don't trust the code you trust the Avid Dao as well to be able to make the changes on all those 500 risk parameters so the problems in that model are obviously trust assumptions so as I mentioned you have to trust the dial and the way the dial performs changes in those risk parameters is questionable every day in 2023 and average you have a

proposal to update those parameters and the thing is that the avidao is not just proposing it's validating proposition from third parties and third part is that I have calculated those parameters so let's say I want to you know change the risk parameter of diet from 75 to 80 this is a proposal done by our third party which does not reveal how those calculations are made so as a token order I'm not a risk expert like I don't have three phds in what you know risk management and everything but I have to trust a third party in the proposition it does without you know knowing anything on how he came to that conclusion so that's on the risk management part obviously you have Smart contracts

unfortunately we went through a bunch of hacks recently in learning like the most famous of those was obviously the hardware hack which introduced like a bug was introduced in their last upgrade uh and even though it was audited uh and and the rest of the code was edited multiple times well you know a bug was still there and mainly due to the fact that the code base was very large and very hard to add it at the same time and the second point I want to mention is like how limited efficiency and scalability is in the decentralized broker model so when you look at the Ave protocol or the compound protocol or the mofo protocol

as it is now you have a bunch of Assets in the same pool so you get the idea that those parameters need to be updated so this like operational costs that is factor in because there is a fee in the other protocol that you have to pay to actually pay those contributors that are going to basically update those parameters Etc this not only does this it's just because you're mixing up all assets together you have to consider the worst asset when you do your risk modeling and so that turns out directly into you know Pro rates per capital efficiency and permission listing you can't list as much assets as you want just because all the assets of the baskets are tied together and so

that raises the question what can we do and after again after more than a year of thinking about it and have like Morpho is the largest Landing protocol built on top of a landing protocol obviously so we're the largest integrative Landing protocols in practice we have more than one billion dollars like on Landing protocols and at the same time we built a landing protocol I would serve doing so and and really what we believe is truly needed is to First establish core Primitives so basically remove the need for the Dow to basically be trusted because the Dao is just about validating third parties and those third parties are completely trusted and not only this is a trust assumption but

as I mentioned earlier this is an efficiency bottleneck and the scalability bottleneck and so the problem with this approach is that well if you become very primitive it becomes very very hard to use like you know we've seen extraordinary approaches like the Azure protocol that you know is bringing you know oracle-less lending but you know it requires much more management from the user the let the instead of you know putting back the the risk management on the gauntlets or the chaos of the world or the avidau it's pushed back to the user and so the user has to take on that complexity and this is why we need something else which is the abstraction layer we need to rebuild the two-lang we need to rebuild the abstractions such that as we were mentioning in the

previous talk and I think it was the purpose of one of the questions is like yes some users are risk experts yes some users are expert Traders or expert LPS and those guys will be able to sort of delegate their users will be able to get their funds to those experts for them to sort of perform their operations instead of them and this is how you get the most efficient and the most trustless and almost open Financial Service and recently we announced that as Morpho we decided to go that path so even though the protocol was the biggest gross over the last year in the world D5 space we decided that that was not the correct approach and we have been working over the last six months on our

own protocol that is independent obviously from other Landing protocols but that is also taking this exact right approach Like A Primitive and abstractions on top that will rebuild in order to provide the same passive ux but with the efficiency at its core cool um do you have questions so everybody can hear me um these new positions that the users of your new

protocol may open and will it be I guess like a analogy would be similar to how liquidity providers on uni V3 would be able to specify unique positions is that what I'm kind of reading into what you're saying yeah exactly so yes indeed there is a strong parallel it's obviously different because The Primitives are different on the side we'll have like on one hand you have dexes on the other it's Landing but you can make a case for a parallel between units of V2 and units of B3 where basically on units of V2 you come up with like a passive experience that is fully managed for you and what's managing here is another dial is the formula XY okay uh while on units will be free you have to decide yourself what's your risk profile and you sort of

like what's your sort of piano strategy Etc um so yes obviously we feel like complexity has to be pushed back to the user the you know the first layer of the user in order from for this to to to really be efficient at the core and then we feel like we should rebuild abstractions such that you know the risk manager like The Gauntlet the chaos the black analytic average doubt of the worlds basically rebuilds what we today call like the avivault where it's fully managed for the users but we feel like this is more Layer Two than a core primitive and and yeah so to answer your question that's going in that direction definitely and will the positions be represented in nfts or will it just be a direct kind of

postcode interaction yeah so the product was not public yet we're aiming at releasing a a white paper in there you know soon not not public date at the moment to share uh and probably those details will be answered at this moment awesome very exciting any other questions so when the new protocol be like extension on the current phone or just dependent new version yeah so it's completely independent in the sense that we're still doing Landing like our code like the Morpho team is laser focused on doing one thing and do it well which is

lending and we've been doing that for the last two years and even though moff Optimizer was a great success so I don't know if your family was the first protocol but it's built on top of Aven compound it's optimizing their interest rates uh we feel like this is the decentralized broker approach this is great but has a capped growth and we feel like the cap is around 10 billion approximately and above that you you can't really grow further while traffic is doing like literally trillions every day and so we're taking a different approach this is a new protocol obviously you know you'll have like potential links between the two but this those are meant to be like separate from a design perspective connection

Megan um will there be liquidations on the platform or the lenders just be left with bad debt yeah so uh let me think so obviously being able to provide this liquidation service that is guaranteeing lenders the it's another level of passivity to do risk management that to Auto liquidate yourself in some sense what I mean by this is that the game Liquidators are playing is an even stronger game in sort of like reactivity than risk management in

itself and we feel like asking lenders straight away to take over that job and removing the need for liquidations is a bit maybe too early for the current D5 Persona so yes we've been working on liquidations awesome last one I promise thank you okay thank you run on top of the new version of Marvel so what do you mean run on top it's just like can you rebuild it no optimizing the new version of Morpho oh yeah yeah so um

again let me think I think it's gonna be especially complicated to do so and again after building more Optimizer for two years I think it's going to be a great challenge but you know like the concept of Optimizer is like really tied to the fact that Landing Protocols are using an interest rate model rather than an actual you know audiobook with exit liquidity on the decks or something so so yeah all this to say that there is always room for a mofo Optimizer in the in the world where you have interest ramp models it's just more or less easy to build and to automate at the Smart contract level

hi uh thanks for the presentation um I had a question about ajna because it sounded like you want to make the job of suppliers in this case a more complex in exchange for potential Capital efficiency but you just mentioned that um you're working on liquidation mechanism so I would assume that you would use some oracles because that's the main difference for without Schneider writing they're not using the oracles and letting the the lenders trigger liquidations yeah so good question um my personal take on oracles is that they're extremely powerful and

to me it's really hard to think of a protocol that does not have an oracle unless it is a DEX itself trading Protocols are probably the only protocol that has the notion the very notion of price embedded at its at their core and when you look at ajna well in their design they're rebuilding your Market you know it's not just a lending broker it's also a trading protocol and it's written on the website I think and when you're looking at just doing lending and doing extremely well and as efficient as possible I don't think that should be the job of the landing protocol to do and when you look at like the new oracle's implementation of Unison before I think we can be very confident about decentralized oracles again personal Tech like you know

oracle-less is also a thing but uh I'm just think like the pain point is not sufficiently here to come up with a full implementation where the pain to solve is removing oracles foreign I think we're out of time anyway but oh okay last one hello yeah so how do you think about um

upgradability of Internet protocols and like security like because I think this is one of the main advantage of having like someone who can have access to the smart contracts stop it or put it in safe mode yeah how do you think about that in Morpho yeah so so when I was describing the two approaches decentralized Brokers versus protocols one side decentralized broker approach is really about traditional Finance world where you would have like time logs Post systems and you're basically managing the thing and making sure you know everything goes well and you have like convenient options and features to be able to handle emergency events etc etc

on the protocol side of things it should not even be a question of like the protocol should be designed in such a way that there is no notion of management like you don't manage HTTP you can improve HTTP by doing HTTP 1.1 1.2 Etc but you know like you don't upgrade the HTTP like people have to upgrade their clients basically and so the way we're thinking about it is that immutability like as soon as you have an upgradability pattern is that you're thinking about your code base as not a protocol but more of a decentralized broker or decentralized fund where you're going to add features you're going to have a large code base and you're basically running a product not a protocol in the sense of like how I

Define it before does that answer the question or yeah can you hear me yeah yes let's say you have a hack and like for example you can imagine how you can have a cool down period like to let's say we can only withdraw 100 million in let's say 10 minutes and how do you make like when you have an Internet Protocol and you don't try to put cooldown periods or anything like that because you don't know when you have a hack you cannot have any

access to the to oh yeah the smart contract right but the thing is that the product will never gets hacked UNICEF never gets hacked and when you're designing the protocol you're you're thinking of it extremely differently it's not about software engineering it's about protocol engineering it's very very different where where basically it's made in such whether it's impossible to act like there is strong invariance uh not necessarily but you could have strong environments you can have formal verification like of the entire code base in such a way that no hack is possible and if the protocol gets hacked there's nothing to do about it you just drop the Primitive and it was just the wrong one and so we like it would not I don't know if that answers the question but you know

you would not put cooldowns on unisub for example to withdraw liquidity just because you know that the thing is perfect and it's like a protocol in the again in the definition that we have like we can have different definitions but yeah and so so yeah I think it's just a way of thinking when you design a protocol it has to be in such a way that it's literally not hackable uh and there is no no notion of like reserve or no notion of like you know Common funds of everybody this kind of things okay make sense thanks [Applause] [Music]

ha [Music]