DeFi's Security Ceiling: Why Lending Can't Be Like Uniswap

Transcript

We just had kelp dowo here. We've had like attack after attack after attack. And to me it raises the question and I'm curious you know for the rest of the group do we have to change our default like thought model of transactions are legitimate? >> It's a good question but you you risk giving up concepts like immutability and the concept of nonreourse and finality. >> Do okay. So I will grant you that on finality. Yes. Because if we slow down transactions processing, agreed. But immutability is just that we can't change the history of a chain. If there's like a 280 million exploit and it goes into a queue and everybody looks at it during the delay period and says,

"Oh, that's not correct." And so a transaction spits out that then reverses it, you still have an immutable chain where the history is transparent, known, and unalterable. It's just a question of under what terms do we let them sort of transact down the chain, right? >> Yeah. It's interesting. I mean, you're asking all the right questions. I was nodding my head there and you start to see how things naturally get centralized over time. So, like over time, people flow risk to trusted parties. They're trusted because they're regulated and capitalized and have all sorts of oversight and auditing. We have to find a way to keep this going with code and technology, not bring it back into the system. You

know, one would think that there's a way to transfer the risk of each transaction. Maybe it's not, you know, at the pool level, maybe it's at a transaction level. Um, there's more work to be done. >> It's a major setback. Like this puts DeFi back a couple of years. >> I think >> Go ahead, F. Oh, I was going to say, yeah, you you can um you can try to build these mechanis, you know, I thought thought deeply about this kind of stuff as a sort of mechanism builder myself. I we used to try to think about how we could introduce these kind of mechanisms into uh into our protocol and it becomes quite difficult because there's there's all sorts of nuances here and and sort of edge cases. When you think you fix

one thing, it's like pushing a bubble under the carpet. You just sort of move it somewhere else, but the bubble's still there. Um you know if you want to do you know rate limiting for instance on protocols you need to know what are you actually limiting you know what's the actual unit of of of limit there is it the underlying asset well that the amount of asset in circulation can change so that needs to sort of adapt over time it's value could certainly change you know a lot so then you need to start to bring in oracles and third parties and and that that introduces new dependencies and things. So adding adding these mechanisms in principle sounds like a great idea and people often ask what often ask builders like me why why aren't there more of these and it does come down to just being very

very difficult to do it without adding new trust assumptions into the mix and uh and that car carries a cost ultimately. So I strongly agree that it carries a cost but let's look at the counterfactual which is that in 48 hours defi TVL has gone from what 99.5 to 86.3 billion I think the market may be speaking that the cost is what is currently there is just rapidly becoming unusable >> too high >> I agree yeah it's totally agree >> so Michael I would ask you like as a builder yourself and I know some others that have been posing this question to people I kind of see two paths paths here. Path number one is increasing like call it slowness, checkpoints, multiple

redundancy in the system one way or another because Chris has said the latency problem is real and if we're going to be optimistic about transactions with zero latency, I don't see a pathway to interdicting these things. caveat if we continue to build relatively complicated things because as a counterpoint let's look at a model that's largely been successful and stood the test of time which is unis swap right if all of our LP pools are segregated and unalterable and to upgrade we've got to like withdraw from V2 and go to V3 and withdraw from V3 and go to V4 and all of these could sort of blow up in their own self-contained box yes unis swap is a very simple thing.

They are not building complex like layers upon layers upon layers. But is the future I guess I would ask you as a builder more likely to be actually we need to simplify a lot of these things and remove the attack surface area like go back to basics in DeFi or is it if we want to retain this complexity we need to think much harder about security which may as you noted lead to certain forms of centralization. the the yeah, this this came a up a lot through my time built building uh building a lending protocol. Ultimately uh a an AMM or an exchange is uh is very self-contained. It really doesn't care

about the outside world. It just it just looks at what's in inside and it you know it's uh it has to try to to balance a single invariant uh in the in the pool. If you look at a credit market or lending market, there all sorts of dependencies and the credit markets really care about the outside environment that they that they sit in. So they care about the volatility of the assets. They need to know the prices of those. The prices aren't intrinsic to the assets. So then you have to have some kind of third party report on those prices. Liquidity is a massive massive issue. You know, a DEX doesn't care what the liquid environment is outside of the DEX, right? It's uh you can it can operate with $10,000 or 10 billion. It's it doesn't matter. But lending protocols as we've seen, you know, with this uh,

you know, with the the liquidity crisis we're in now, very very heavily dependent on what's what else is going on. And I think that's the that's what makes it so difficult to build a, you know, fully truly immutable unis swap like lending protocol. We we've tried we've tried to get as close to that as possible. You know, Moro also building with similar sort of mindset. um trying to trying to keep as much of that as you can while still offering a viable capital efficient lending market. It's um it's just I I think but at this point I'm I'm resigned to say that it's just not possible for credit really. We've had seen many teams come up with all sorts of uh interesting ideas on how to do this and I just don't think it's it's possible. So then we have to accept uh this is the state of the world now.

This is how it has to be. How do we what's the minimum amount of uh centralizing force that we can add back to these systems to make them uh not collapse so so spectacularly when they do um whilst maintaining the core responsibility and disintermediation composability and all the rest of it? I mean, if you >> So, we saw >> Go ahead, Chris. >> We saw we saw Drift partner up with Tether and I don't know, is it a centralization force that's going to restore security or is it just the resourcing that you need to have just to go through like inordinate amount of audits, bring in the the AI like clearly for this industry to move forward, we need to provide confidence in whatever protocol you design. So I guess is

centralization the only option or can you know like you're seeing this partnership between Tether and Drift don't know all the details yet but is it more of a resourcing issue? It's really hard for a startup to compete. >> I I really don't think it is. I think um you know uh we we we talked earlier about mythos right finding uh bugs and flaws in 26-y old uh operating systems you know the amount of the amount of manh hours spent on that you know is just more than anybody could afford to ever spend building a D5 protocol and yet there were still bugs that could be you know uh teased out there so I don't think it's just a I don't think it's purely a resource issue everybody's really really cognizant of the the importance of security and they really really are buying it's um it's very very

uh it's just a very very hard problem. I I I I sort of I think it's unsolvable at the at the code level. Uh so we need we need other we need other fail safes. We need other mechanisms that are introduced I I think at this point to to help safeguard things a little bit more than just relying purely on on on on code code alone. So working backwards, right? Like one of this is me putting on my uh business school professor hat, but looking at some of the lessons from history, I feel like there are two paths that we can go down here. One is exactly what you said, which is looking at tradi lending markets and starting to build in a lot of the fail safes in modern markets. Like great example, you know,

if you have call it $200 million of supply and somebody deposits $500 million, not making all of that available for borrow immediately, right? Like allowing that only to increase by call it 10% per day or something like that gives you a very long runway to interdict these things. said, "Yes, it obviously makes that protocol somewhat less efficient from like a I can't take a $300 million flash loan or, you know, immediately deposit something in size." But it might also be fair to say maybe that's a problem, not a feature. And there why why are you so intent on doing this like literally right now at 3:00 a.m. on a Sunday becomes a valid question along with many of the other

redundancies you build in. The other one though to go the other way on like a lending protocol is I know this makes it less liquid but there's no reason you can't structure it like an AMM of all of these are essentially bilateral offers for lended borrow and then you don't have this contagion risk right like I will lend >> $100,000 of USDT at this price and I will accept the following things as collateral and that's that you could have like these prepackaged boxes where each person can do their like lending and terms. The problem you run into there, which I think you know you've spoken to this in your earlier answer is that's less efficient than it might otherwise be. Like I I if I recall I

think there was an early version of like sushi swaps led protocol that was similar to that way back in the day. >> Exactly. >> Yeah. It just didn't it just didn't it just didn't attract enough um enough users. It was so clunky, sort of almost peer-to-peer like and uh and when you're competing against a pool model which is you know I can just deposit and withdraw whenever I like and there's so much liquidity that you know people think it's about the technology. It's really not. What a like really large lending protocols are selling is liquidity. It's that ability to go in and take out those huge loans and repay them whenever you like. That's the main product. When you've got liquidity that's that's king above all else. And so when you have these isolated you know almost

peer-to-peer like uh lending uh systems much much more secure as you said you don't need all these external dependencies and all the rest of it you just need to trust that the code a very simple piece of code works for that particular loan but then it's just how do people find each other that's the problem that I think uh compound originally solved and that the ave you know heavily modeled on the compound system you know followed on with and it's it worked so tremendously well that it's scaled up to billions and billions of dollars But with it has come this you know you know other challenge which is uh when things go wrong they they sort of tend to uh cause contagion in in a way that those isolated me models don't. So yeah we we we've we've experimented a lot in

DeFi at this point and I'm I'm I'm sort of resigned to the fact that maybe it's just not possible uh to to to get the best of all worlds and that we're going to have to sacrifice something somewhere along the line. I will say as a closing remark here as we are hitting time that I have resigned myself to the belief that I think you are probably correct about that that we are not going to be able to have our cake and eat it too and we're going to have to figure out which by the way there are multiple paths to tradeoffs but we're going to have to tolerate some of them because one thing I think all of us agree on is the current paradigm is not sustainable. If you like this segment, please like, subscribe, and tune in every Monday at 4:30 p.m. Eastern time. I'm Austin

Campbell, the host of Bips and Bips, along with my friends Ram Alawalia and Chris Perkins, and our slate of exceptional guests. Every week, we're going to discuss macro, crypto, and the collision of worlds, covering topics that move markets and shape the financial landscape. >> If you've been loving Bits and Bibs, don't forget that the show is transitioning to its own feeds on X, YouTube, and your favorite podcast player. If you're not already subscribed to Bits and Bips on its own channels, go there now and hit that subscribe button so you can keep up with our twice weekly live streams and macro meets crypto breakdowns. Bits and Biffs will only be on the Unchained feed for a few more weeks, so subscribe today to be ready for launch. You can get all the links at unchanged crypto.com/bitsandbbits.