Flash Talk by Brenda Loya - Co-founder & CEO, Tellor

Hi everybody, my name is Brenda. I'm the CEO and one of the co-founders of Teller. And today I'm here to talk to you about why Defi needs better security and oracles than TRDFI ever has and how we can actually achieve that without becoming TRFI. TRDFI is optimistic and I mean everybody here should be I think familiar with what an optimistic system looks like. you can operate and transact very very quickly and then disputes are open.

Anybody can come and dispute. Right? So in triify what ends up happening is because they operate under a closed system no anonymous party can just come and run away with your money right and they can do high frequency trading because they have very reliable dispute mechanisms very robust. so robust that it's it's they're backed by governments and if you can't pay the money back, if you spent it already, you can always go to jail. Um, this is an example of a dispute that happened between Leman and Intel and Leman and City uh for some derivative products back in 2008 and they didn't actually get resolved or settled until what is it seven and nine years later.

and JP Morgan. They had a case of market manipulation from 2008 to 2016 and they weren't actually fined until fined until 2020. This is slow, right? Um I know that they can do very high frequency trading and we're trying to compete with that but we don't actually have this dispute mechanism that they actually have. So as stratify a as defi it is much harder everything that we do because every decision that we make could have irrevocable consequences right if you choose a very fast oracle you might be more subject to manipulation on the asset if you use a slower one it could be stale if you use a market oracle something you know can happen to the actual asset and be more easily manipulated fundamental or fixed also it could um if if you fix something or hardcode an asset to one for example for a stable coin and all of a sudden the collateral completely depects you're in trouble again right so there's no onesizefits all kind of fix for us it's much harder what we're doing uh because we are operating on an open system and anonymous parties can come and run away with money and these are just a couple of examples over the last couple of months of things that have happened related related to the oracle.

Obviously, when you use a single source oracle with low liquidity, it's always a bad idea. And that's exactly what happened with Binance with the whole Athena debacle. Athena ecosystem operated, you could say, perfectly in its ecosystem with redemptions and liquidations and mint, minting and redemptions versus what was happening in Binance, right? Binance decided to use a very low liquidity their own order book as their oracle and that is always a bad idea. Um and it was clearly shown on what happened just in their market.

Um, Elixir, this is a really interesting one because in May of this year, uh, somebody manipulated the price of DUSD on a curve pull on Ethereum. And that corpo had very low liquidity, but it was an a big enough of a trade that had an a direct effect on the VWAP that was quickly picked up by the oracle reported on Avalanche Oiler and 530,000 liquidations happened off of that. And then Elixir um some of its collateral was came from Stream Finance. So XUSD uh Oiler, Morpho, and Elixir had hardcoded XUSD to be equal to one. And what ended up happening when Stream Finance reported a 93 million loss from an external um fund manager, their XUSD completely depected or completely fell apart, right?

So their collateral for Elixir was basically worth nothing or worth a lot less than one or than um than it could have been. But Odelli they assumed it would always be by hard coding it to one. And so then Elixir had a bunch of um redemptions because it was hardcoded to one, but nobody could liquidate and and have this rebalance properly because it was hardcoded to one. So, there is no onesizefits-all moon. Well, um I'm still I haven't seen the full postmortem on this one, but the Oracle or the calculation it was way way higher than it should have been.

And they also ended up with about a million dollar loss earlier this month. So how do we address that as a and on this these are a lot of other examples that happened over the years of different things that have occurred where now we can we can we can see people are trying different things to to mitigate this risk but one size doesn't fit all and what in the conclusion of all of that to me is that we need to start seeing this as A we cannot say that we if we implement one thing it's going to work and everything's going to be perfect right so we have to look at security in terms of layers um the over the past years we've seen over and over again the that the actual oracle protocols don't get hacked the way that people manipulate or or hack the oracle is basically by manipulating the actual asset and taking advantage of the changes in liquidations or markets for these uh assets and whenever they're not optimal, whenever they're able to change the dynamics for price discovery, that's when they're able to attack the oracle. But the actual protocols, no matter who you use, um you're going to have to provide them a definition or a question of what it is that you want. And obviously if you don't have enough sources for that for that asset or that asset becomes delisted over time um you you have to continue to monitor what is happening within your definition that can have a direct effect. So you first start off with a proper definition that is robust hopefully and then those assumptions keeping you continue to monitor the assumptions over time because they will be tested and then make sure that you have your Oracle protocol obviously choose whatever it is that you need and within within your parameters you might choose anything from a centralized oracle to a more decentralized oracle but it's important that you implement them properly and then obviously you should always assume that something wrong is going to happen with the Oracle because even if it's not at the Oracle level we are operating under deterministic in a deterministic environment once you deploy your contracts but a lot of what happens within that deterministic environment it it's it's correlated to something that's completely deter uh subjective and at most probabilistic of information that's coming in from uh price discovery outside of your ecosystem.

So always assume that something's going to go wrong. Um that's pretty much uh you know a rule of thumb. Make sure that you have fallbacks to address when things do go wrong and things can happen um or things can go wrong when they come from they can come from the bottom up from the data definition or your liquidity or your sources. um or as we saw with the Levana protocol when the actual osmosis chain was attacked um it was attacked to create a diff a similar uh situation to what we had on Black Thursday in 2020. So these attackers basically congested the network so that they could control when the oracle would update and take advantage of that.

So it can come from the bottom up or it can actually come as an attack to the chain. But ultimately you have to have those fallbacks to be able to protect your users because ultimately all the bad PR is going to come back to you as a founder. Um and obviously the the as last resort you can always use your uh your community and actually fork your protocol. This is obviously the last thing you want to do. It's a very difficult thing to do.

We always say that, yeah, you can always fork, but if anybody has ever tried forking anything, it's really hard, especially with something that's already established. And as I talk about the social layer, uh I'm sure everybody remembers earlier in the in the year when uh Hyperlquid um got had an attack via the jelly token and they decided to overwrite the Oracle input. So in the past they hadn't really been very um they hadn't used that type of control, right? But they decided to use this as uh this specific hack started eating into their own treasury and that just showed everybody that they had all of this control. That's not I don't think that's either good or bad.

It's a choice that they made and I'm you know it's their community. Uh it was lots and lots and millions of dollars where they were at risk. So I you know I wouldn't want to judge them. But that does put into play now if the government comes and asks them tomorrow you you know do this or do that um they have no record they cannot say that they have credible neutrality anymore in my opinion and I think that that is a balance that is very difficult to strike especially if you want to maintain that um or maintain that you are still an open and decentralized protocol. But all this to say that Trotfi does not have to deal with the type of problems that we do because they have a much stronger dispute mechanism and Trophy has only updated their systems and been able to or been willing to provide better services to customers now because DeFi has become so big that it was impossible for them to ignore us.

And as I see more and more institutional investment coming into the into the industry, you see a lot of people really catering to Trophy and shifting uh how they work, right? Um and I just like to say uh that I think ultimately we're going to have something that operates very close to TRFI and actual DeFi. And we need the DeFi because Stratfi will not continue to improve if it doesn't have any competition. We saw that for for many many many decades. They didn't improve.

They didn't do anything very novel until we became a strong competitor. And if we cannot if we just cater to them and not maintain that independent system, they'll ultimately go back to being what they were before and we'll end up with a bigger debacle and a bigger de recession that we saw in the 2008 uh breakdown of the economy. And with that, that's uh all I have for you guys. Thank you so much for being on my talk. And uh