The Oracle Era with Marcin Kazmierczak and Jakub Wojciechowski

Transcript

Maybe I'm wrong, but I haven't seen any like calculation there. They said just okay, this is the price you should use for settling the trade. Done. That case closed. Innocent. So, it's about the scalability and it's about the the versatility. So, that's the that's the difference between us and and chaining. Could you kind of break down the difference between push oracle and pull oracles? And so Morpho, Pith, or this vault company RE7 basically didn't claim any responsibility and the user lost its funds. If you're an oracle, you deliver the price. Period. Welcome back to the rollup, your home for technical talks and open conversations that shape the

modular future. Nothing on today's show is to be construed as financial advice. You can see our full disclosures list at the rollup.co/ventures. Sit back, relax, and enjoy today's episode after a quick word from some of our awesome partners. Ladies, gentlemen, anons, and dgens, it is stable coin season brought to you by your favorite, the best in stable coins, Fracks USD dollar stable coin, backed by the best in the business, the Bidd Fund by Black Rockck and Superstate. This is stable coin season. It is the yieldbearing stable coin by FRA coming soon to a chain near you. What's going on guys?

Welcome back to the rollup. Today we have an episode to discuss some of the latest happenings in the oracle landscape in crypto amongst other interesting dynamics in the fast and cheap blockchain scaling space. Uh not notably a lot of roll-ups and fast applications on them. So uh lots of fun topics to discuss today. Marson, Jacob, welcome to the rollup. Thanks for having us. Awesome to be here. Hi guys. Great to be here. Fantastic. Um, yeah, I mean I think uh what what better way to start a conversation about oracles than to discuss the recent happenings in crypto regarding oracles.

Uh, two very interesting things have happened in the last 48 hours actually. One was Poly Market was hacked for $7 million because of an UMA Oracle uh dispute uh problem which we can talk about uh maybe Rob if you want to pull some information about that up that that'd be cool because I'm a little bit out of the loop there. But the more pressing thing in my view that got crypto traits attention was Hyperlid. So, uh, kind of TLDDR and then I'd love to hear y'all's opinions about what kind of what happened and, you know, why this is problematic. Um, so basically there was this coin Jelly Jelly on Hyperlid.

Um, and essentially how how Hyperlid works is there's this uh vault called HLP and it essentially takes the opposite position of all the traders on the exchange. And of course, it's quite profitable because people trading on leverage in crypto are dgens and aren't very profitable. And so, this has been a very successful like low-risk yield product in crypto for a while. And essentially what happened is somebody took out a a massive um I I believe it was a massive long position on this Jelly Jelly token. Um and so this HLP product had to take a large short against it. Um, and then somehow they pumped this token uh from like really low market cap, maybe a couple million

all the way to like 10 or 20 million. So very low liquidity, low market cap token. But this this person trying to exploit HLP essentially took out like a multiple million dollar uh long position on it, forcing HLP to take a massive short. Um what Hyperlquid did was they essentially reverted the price of their oracle and they overrode their oracle through their their validator set, reset the price to give the attacker a very small loss and the and the profit of HLP went like very down a little sideways and then right back up to where it was.

And so this all transpired while Binance and OKX decided to list futures of this Jelly Jelly token immediately as they found out that Hyperlid was under fire, suggesting that they are a bit worried about Hyperlquid's dominance uh perhaps. And so this is an example of an Oracle uh provider uh being basically reverted back to the validators of hyperlquid which is very very small set. So I'll shut up now but that is my TLDDR. Jacob Marson what's your take from the Oracle lens on this situation? Uh what are the the good and bad and ugly here and what

can we learn as an industry from this event? uh feel free to be very very blunt. I I think that the audience would appreciate some cold hard takes. Yes. I would say like Oracle is usually a very boring topic, very technical one until something like this happens and it's great because everyone recognize that actually it's pretty pretty important, crucial and core to the protocol. basically like setting the price have has got like enormous power of controlling like what happens to the protocol and also like hyper liquid is like pretty pretty interesting animal because they started with a super

simplistic model for the oracles basically and we are seeing like in many occasion like we we talked to a protocol we we describe like what Oracle does and they say nothing fancy basically we can we can take the the price from Binance We do not need to, let's say, pay you guys anything. It's like super simple. It's basically like two hours of work, get a response from from Binance, HTTP, put that on chain and jobs done. What what's a waste of money like building an Oracle service and it works until it doesn't as it occurred. Like it's like a lot of tricky questions, a lot of balances, a

lot of trade-offs in building oracles. You just need to put some check and balances like how to actually sanitize the price, how to curate the sources in first place, whether you should rely on the price of jelly being taken from such a low liquidity market, etc. So, yeah, I I'm not going to call anyone blame here, etc. For me it's like an amazing showcase like how oracles are important and it should be let's say handled by the external professional and unrelated third party for for the good of the protocol.

Cool. Anything to add there Marson before we kind of transcend into the poly market story as well? Well, I believe the question is uh you at the end of the day, Hyperliquid didn't uh you know got much damage even so they did profit from the whole situation. It is pretty um surprising how how they approached that but at the end of the day the market was a very low cap and there was like a large long position looking on the Oracle for pers for lending markets for real world assets for proof of reserves all of those are within the Oracle umbrella but all of those animals are behaving very

differently uh some are updating just once every 24 hour some are even less frequent and for purposes It's very important to have super accurate pricing because of the long and short positions. H over the years we've seen I believe at least 15 or maybe even more like oracle manipulation mispric mispricing events and so on. hyperlquid. The reason it's so peculiar is because they actually own the whole set and they could basically print even higher or lower number at the end of the day, right? Like literally if you understand that they took it from from almost the finger at least to my knowledge like they didn't give any like formula how they came up with this specific number to settle uh the market.

Maybe I'm wrong but I haven't seen any like calculation there. They said just okay this is the price you should use for settling the trade. done that case closed, innocent, whatever. So, it's very very interesting. So, yeah, those are my two there. There's a couple knock-on effects I think for hyperl specifically where I think uh now you know that you have like this like this back stop. Like for example, they call uh Trump being in office like the Trump put. Like, oh, like you know, stocks and everything's going to do well because you got the Trump put. You know, it's kind of like now you've got like this hyperlid put where you know if anything ever happens to HLP, you know, you know that they are going to step in and make

sure that their depositors are going to be uh taken care of regardless of how. And so that's going to, you know, increase more deposits likely or at least make people think it's it's less risky, probably lower yields. And it also puts it puts the exchanges in a really weird spot where they just showed their hands. Like they basically said like we don't even care. We're willing to just try to destroy other exchanges in the space. And that has knock-on effects where smaller exchanges are now going to rather than you know taking more risks that they might be more uh serious about which pairs they uh they list and how they manage their margin requirements. this could lead to less

capital efficiency as users have to post more collateral for certain trading. I also saw that this guy a guy um posited that this may lead to more like M&A. So there are some interesting perspectives here on this situation, but we can kind of move past it and move into this this poly market debacle which kind of went under the radar a bit more than this hyperlquid one. Well, yeah, both of these kind of happen around the same time frame and I think these are two s like very contrasting occurrences because you have hyperlid come in and provide that back stop for better or worse decentralized or not and then on the other hand you have the uma oracle and this poly market bet that also

got manipulated to a certain extent but you didn't have a backs stop you didn't have someone come in and provide some sort of you know correction to the manipulation. So, I'll I'll kind of, you know, phrase or frame what happened. Um, and then, yeah, would would love the takes uh from our two guests here. So, there was a a market on Poly Market. Um, it was an it was a Ukraine uh themed market and it was around the mineral deal between America and Ukraine. Um, and what happened was that this was somewhat of a contentious deal. There was some reports that said it was around and and it was signed. Others said that it hadn't been signed yet. Uh and according to the market rules, this

really should have resolved as too early. And that's something that uh the UMA Oracle has in place in order to protect things if someone proposes a result uh before there is a consensus of opinion out there. Um and so what happened was that uh you know this market was proposed and then the way these markets work for resolution everyone votes on whether or not that proposal is correct or incorrect. Now, in this market in particular, you had a user who's actually somewhat public. They have an ENS tied to their address. Uh, their name is born too late. Kind of a funny ENS. And they have accumulated. It's like one of those like born too late to just like live in the metaverse and like have welfare. Born too early to like experience Mars.

Exactly. Exactly. Um, and and so I mean that's pretty indicative of the culture poly market. Like there there's some funny individuals over there. and they th this this particular user has accumulated a stake of UMA tokens worth uh upwards of probably like$2 to3 million. It's 1.3 million UMA tokens. And this puts them in the top five of UMA stakers. Um and UMA uh the voting process works so that it's not one vote uh one address, one vote. It's if you have more UMA, you get more voting power. So this person was able to accumulate a large chunk of UMA to theoretically sway the outcome of this of this market. Um they voted yes when

in reality the market should have resolved as too early. Um and people are speculating as to why this person voted this way. Now the way this happened um and the way this works is basically like you know all these people vote and it's the wisdom of the crowd and so people are basically saying Uma is saying that you know the the oracle worked as intended because everyone voted the outcome resolved um but obviously polyarket voters are very upset because they they feel it should have resolved too early. Now, I I think there's a couple things to take away here, and then I do want to hear from Marson and Jacob. You know, the real indication here is that 1.3 maybe 2 to3 million of economic stake is

just simply not enough. The the the amount of economic stake that is securing these markets. Um there was uh you know, these markets get upwards of seven maybe 10 or 15 million dollars in volume. And so for a top five voter in in voting power to be uh you know only two to three million dollars of economic security that's just simply not enough security around these markets to protect them from corruption. Um the other thing that I think is interesting here is that they didn't come in and and it would have been Poly Market that would have came in and provided a backs stop and refunded market traders out of their own pocket. They chose not to come in and do that because according to their uh to their

um that this to their uh to their uh spokesperson on discord was that uh this was not a market failure in their words and so they didn't provide that back stop where in the past they actually have provided that backs stop. So I think those are two particular takeaways. Um, and then I've got one more thing that I'll kind of bring up after, but those are those are the two important things that I I think are happening in the poly market sense. Jacob Marson would love to hear your your thoughts on this this outcome. Yeah, happy to start as you may see like Oracle is becoming interesting again.

Every day we get a lot of interesting challenges about what's the what's the correct model whether you should be like on one end of extreme trust the trust the code trust the protocol even when it evidently brings incorrect price because it was manipulated that's a very good philosophical question like what's the what's the true whether it's like a social construct whether it's like some arbitrary fact or let's say ultimate true probably not a topic we we are able to to answer in doing this this podcast, but also it leads us to a way that Oracle should be like very pragmatic and very open about their their mechanism.

And indeed the problem was that no one was actually looking into the the the the details of this security model. It's it was more like a social experiment done by UMA. It's interesting experiment. We we are talking to many financial uh protocols and people were afraid of playing this game with Uma although it's like very interesting concept but still at the end of the day you do not want to leave let's say the the answer for the price of if that you have like few billions to to let's say this this social coordination experiment. So in in in my view that was like too too

progressive and there was like not enough of check and balances but fortunately I think that the landscape evolves for example we are we are very much engaged in the in the shared security movement openly speaking it's still very early but at least there is a capital to power that concept so so with you get like over 10 billions of dollars so that's a significant money it's not also so easy, let's say, to to tactically switch the the voices because you get like professional operators.

There are some check and balance and basically it's like a well-designed system and it's very hard to to create like properly decentralized security and safety mechanism. So, so kudos to we are slowly migrating our security there. But also, I would say like with Oracle, you you need to be decentralized to avoid a situation. And then even the temptation of printing a price because in my view is something unacceptable. But also you cannot be let's say naive and jump to let's say the this oversimplistic model of giving all of the the power to token holders because you could be the game. It's the white is crypto.

Yeah. to Jacob point uh maybe coming back for a moment to hyperlquid like in the end of the day the users of hyperlquid are happy how the uh path like ended right so they their core goal is to protect their core users the ones that are staying with the protocol even though it's against probably like the regular logic how you would settle the market in Uma case uh and let's say more even poly market so u they didn't decide to take action I believe that was also uh probably a good decision because if they had taken the action there would be the question okay so now you're questioning the oracle that is underneath so actually we should pause the whole protocol we should like shut down all the stuff over here and then rethink like how we are logically like

approaching that topic right I I'm not right now aware what's the volume on some markets that can be potentially manipulated and also I'm not an expert in this uh not price rather uh oracle that dictates what is the truth of resolution of of the market. But I believe right now people will be also looking at UMA token how it behaves and trying to also create some other uh engineering attacks like more like capital attacks on on the markets that are resolving and also like if poly market team is listening to that like we are happy to jam about it. uh maybe we don't have let's say a direct solution right away but if there is like openness

to discuss uh we have some ideas around ZK implementation and also like probably um narrowing down the possible markets that can be created right because you can also just trim a bit the universe of uh markets that you can create or at least trim the amount of volume that can be uh put into a market that can potentially be skewed in that Yeah, there are um qu there is quite a vast universe of markets on poly market like I I am curious as to like any oracle that that could resolve um some of the markets on there like the earth is the earth flat is on is on there like

is you know are they going to find that missing airliner that that went missing you know Malaysia 19 or or what you know Like there is no like one reputable news source out there that we can just trust. And I think you know if there was it would be easy to just plug in like okay whatever this news this source of truth is we just plug that in we use that to resolve the market and there would be no dispute. So yeah, I mean it's a fas prediction markets in general I find fascinating and I'm curious as to like how uh how uh any oracle would be able to resolve uh some of these like rather vague or ambiguous markets.

Yeah, it's it's it's a tough question. Um and subjectivity is kind of the that's what I layer is also trying to bring into the picture with their kind of inter subjective model. But subjectivity when it comes to uh blockchain is difficult. It's it's a problem yet to be solved in terms of determining the right answer. One more example for you guys and I think this ties us back into the more kind of redstone centered conversation. Earlier this month, uh, a user that goes by the name of Jamus lost about 14 ETH due to a faulty oracle upgrade, um, involving CBE, Coinbase ETH, and Pith. Um, so

essentially what happened was, uh, the CBE USD price feed from Pith remained frozen for seven minutes. So basically just wasn't updating and there was a price ratio distortion that triggered unnecessary liquidations. basically that user lost their capital when it was not actually supposed to be lost. Um, and then Pith comes out and says basically that we were accurate and that someone else either Morpho or these this this other kind of RE7 labs or the user should have had an extra scheduler to update the feed. Seems a little bit odd to me. um that the creator of of this vault, you know, they came out and said,

"Okay, like yeah, there was a timing mismatch, but they also said that pushbased oracles, uh this is how they work and you know, next time they'll be better." And so Morpho Pith or this vault company RE7 basically didn't claim any responsibility and the user lost its funds and you know it was just kind of like another day in DeFi. Um, so I know Pith has like started to in introduce redemption rate feeds and stuff like that, but what went wrong here, guys?

And how is how is Redstone uh not susceptible to a similar uh incident? Um, and what do you guys think about this one in particular? Yeah, so I think all the parts are right. like there was a possibility to to push the price on chain but obviously you you you shouldn't delegate that to the end user to some you know normally that was trying to earn a bit on the CBE I think it's cruel just to say it's it's your fault and and basically like pushing data on chain it's it's not a trivial task honestly it's like very

few companies that can do it right and also like we admit that It's it's a it's a hard task. We we spend more than two years building a lot of let's say fallback mechanism. So so we are running a set of nodes not like a single script that you can push price and then just another shadow collection of nodes that are watching the existing nodes and intervening if there's anything wrong. Also we are using the gelato automation for some of the part as a as a third layer of defense etc like a different algorithm for for the gas calculation you need to have like tons of redundances you need to have like monitoring

checkers all of this crazy engineering hard and and something that is very familiar to the standard web two architecture all of these control systems and a lot of redundancies and it's not easy to implement. So when I'm also like hearing that there is a solution when basically like anyone is able to to push data on chain I think it's like too much of an ask because it it's it's not easy and if an Oracle is is also claiming to provide a service it should take the responsibility of actually delivering the service and and do it to the best of

their abilities. Yeah, I can just double click what Jacob said like if you are an oracle, you deliver the price period. Like this is your job at the end of the day. Naturally like there are those self-s served models and some people can opt for them. But then if someone is opting for them and everything is transparent then the person that is you know accepting that it's self-s served is probably responsible for that. We can also play that game. But at the end of the day, if you're playing with, you know, millions of dollars and sometimes even billions of dollars, probably it's not worth taking the risk. And this should be like very reliable and robust infrastructure consistently delivering

price by price by price by price. And there is a reason also like for so long chaining was the only player like throughout 2020 21 and 22, right? because they were the first one probably to at scale provide that resiliency and um ease of acceptance of them as a solution let's put it that way because people are just uh not afraid that it can mispric at one point right um and p itself like we respect the team a lot and I I see them on many of the conferences and I believe they bringing a lot of uh also fresh wave to some of the topics but specifically in terms of push price feeds It's not the first time we've also seen

people not being uh fully let's say u satisfied with at least the redundancy of delivering of of that right and that's also the reason they've been mainly promoting themselves as the pull oracle not the push one like rather the pull oracle and then people try using them in the push model which okay works until it doesn't as we already mentioned over here and with oracles it's about 30 365 days a year 247 and making Sure. Especially when the market is stress tested, the volatility is high, gas spikes, RPCs are not responding as you would expect, right? Especially on other networks than Ethereum. Like Ethereum is fine, but with new networks, many times RPCs are also playing funny games when

you're trying to deliver the feed on chain. And we specialize just in that like we have never had a mispricing or downtime event since the main launch of Redstone. And this is one of the core things we want to make sure we keep on delivering for our clients. Marson, could you kind of break down the difference between push oracle and pull oracles? I I think it's relatively straightforward like at a high level, but maybe you could highlight some of the some of the key differences. Of course, in pull model by principle, you have the feeds available in a sort of a layer. In PIF case, it's called PIFFET. It's a fork of Salana. So they keep the updated price in this uh for of Salana and then they use wormhole for

crosschain data delivery whenever it's needed. So for example there's a user transaction on let's call it Ethereum or arbitrum or other chain that requires that price. So then Warhol is responsible for delivering that data package. In redstone case, we also have pool model. Uh we keep those prices in data distribution layer that is a kind of like a a decentralized cache where the feeds are updated all the time and example gearbox or curvenance are using that in production or even uler and whenever there is a transaction via uler uh user that utilizes redstone as a price feed then it's delivered on chain.

So you can think of it in a simplistic way as on demand only when it's needed then it's delivered to the destination chain and let it be arbitrum avalanche monat or whatever else right like it's pretty like abstracted away whereas the push feeds is delivering the price to the destination chain a specific one then one one pusher let's call it pusher for simplicity um is for Ethereum the other one is for avalanche the other one is for um let's call BNB chain and then every price feed has its own deviation threshold. So the difference between the last updated price on chain and the one that is available in the offchain environment. So what's the difference

between the two and the moment it crosses the threshold for each asset is a different one. Let's call it it can be 1% it can be 2% it can be 10 bips. The moment it crosses then the next update is sent. And the second usually condition is the time interval. So it can be 10 seconds, it can be one minute, it can be 24 days. So it's very much dependent on uh the asset itself. And usually the more performant chain you go to, the lower is the either deviation threshold or sometimes even just the uh update condition based on the time interval. And Redstone is an oracle that delivers both of those models like across many blockchains. I think in push

model we have right now almost 40 35 for sure. So we are uh going like very broad like story protocol para chain uni chain and whatsoever. Um and in the wellestablished protocols such as a compound forks of those morpho uler fluid all of those are relying on the push model. So the one that was popularized by chain link and is also utilized across many chains. The pool model is usually either for derivatives and perpetuals or some new models. So people who had already the flexibility and plasticity of the protocol to implement the pool model because the integration is slightly different. We

talked about we talked about pith we talked about UMA um we talked about the hyperlquid oracle like generally you know it it feels like it's different strokes for different folks. There's you know different products different oracles for different use cases. Some are more flexible, some are more directed. When you guys are looking at the Oracle landscape in general, how do you guys see yourself comparing to some of the existing solutions, Chainlink, Pith, UMA, etc. Where do you guys see Redstone fitting in? And how do you guys stand out? What do you guys do in the oracle space that's different architecturally and uh technologically than than the rest of the oracles?

Innovating finance, global, institutional, and onchain. Deep liquidity to drive capital efficiency. Now maximize yields on institutional grade liquid staking with M. Unlock trillion dollars of capital via Fitbitcoin. All of this secured by Mantle's treasury of billions in stables and ETH capital pools. This entire ecosystem is coordinated by a vast network of seasoned VCs that are aiming to create the next generation of banking and financial technology. This is Mantle. Okay. So, let me hang up the the

technology and let's refer to our big friend Chain Link. So to to give a comparison I I think it chaining make a very good decision when they started in 2017 to focus on on the on the security and doing things like extremely well but it was targeting the only on only single network and basically there was only mainet then and they done extremely amazing work in terms of stability security performance but right now it's 2025 five and I don't know if you're counting but we are probably approaching like a thousand of networks different source you know some

are like a very ephemeral some are like super interesting fast growing etc so we're double tapping on on the marine uh answer for the push model actually like on 56 networks with a full push model while chain link which yeah as you may expect that slightly per pockets than our and much larger headcount is probably on on a dozen maybe like 20 networks. So there is something wrong in their their design that we fix with Redstone which is about the scalability. So basically we we decoupled different services in Oracles like there is like separate service that takes care of the quality of data prices

and the price discovery and we can reuse the same information across different venues. So, so the same pricing data could be consumed on Ethereum mainet on on arbitum on bar chain and know any any other roll up that we just created like 15 minutes ago. It's still the same quality of data and we can also quite easily expand with the set of re layers. We got a network of re layers. We just need to configure them and they can start pushing price whenever you want. So it's more about this this small but very important architectural decision like how to make Oracle truly scalable.

So we can adapt to different use cases. So, so we can change a a single model module in in our system for example to adapt to the high security of the proof of reserve systems when we need to do like a tons of checks for example for the proof of reserve of of the Bitcoin networks when you get dedicated indexers and we we need to push the price infrequently like once a day but it needs to be of extreme quality or we can focus on the on the speed when we work with networks such as Mega when when basically we are like loading the network with I think thousands of requests per second. So it's about the scalability and it's about the the versatility. So that's the that's the

difference between us and and ching. The Mega ETH integration is particularly interesting because you guys and Megga ETH like th this is from my understanding an application of this architectural difference to the fullest extent basically tapping into the most potential like the speed at which mega ETH is trying to operate necessitates an extremely agile oracle. How do you guys anticipate pushing price feeds as fast as Mega ETH wants to move like at the millisecond level?

Yes. So it's it's a very fun experience honestly like when you started playing on the devet like we we run some initial budget part like the first integration we run to 25,000 of transactions per minute in in in the early stage and we said should we keep increasing the the pace basically like does it make sense to update like quicker than one millisecond because technically yes we can we are getting data from from multiple exchanges and if the market moves quickly then basically we can that's I think more frequent updates than a millisecond but doesn't make sense for the users. So they're like interesting questions. So it's a very funny tech to work with if you got this kind of performance and yeah we are just waiting for the new use case. It's going

to be fascinating like to see like what people can get out of this let's say really extreme speed of getting new data whether that will be like per we are talking to some protocols that we'll be trying to gamify like how can you get most of the almost real time updates. Yeah. So one really interesting part about uh Mega E's design in general as well as uh single sequencer based L2s is the latency aspect and this is directly affected or you know directly important to what you guys are attempting to do there as well which is basically when you have a a single sequencer which for those in the audience is effectively like a piece of hardware in a data center. um you know being closer to this data center makes you have an

information propagation advantage. This is what latency is all about. Basically the speed of the of of lights and and the the uh the dynamic of your fiber optic cables or whatever that that you have um being closer to this data center means that you have an advantage an information advantage. Um, so May ETH is starting off with a single sequencer, but then going to start doing collocation uh rotations. So they're going to be rotating different sequencers across the world, I believe, um, over time. And so um, when you guys are thinking about building the Oracle integration for Mega ETH, how do you guys consider latency as a important

dynamic when thinking about that? And then is it any different than building a uh Oracle product for like a arbitrum one or or optimism in terms of the sequencer setup being somewhat similar? Uh h how are you guys optimizing for the best latency on on the market and um effectively how do you guys potentially if possible fight against this information advantage or disadvantage from other parties participating in in the network? Yeah, it's a pretty good question. It's great that you are let's say putting a focus on on some basic geography. So finally we can prove that the that the earth is not flat actually. Yeah, we get like a really tangible results because

yeah, it matters the the geography matters at at the level of milliseconds basically like it's up to half a second to travel the around the world and yeah it's like funny experience because I don't know if it's on purpose but when mega if not always is disclosing when they're like moving the sequencer so for example the test net was supposed to be probably in Tokyo and they ended up in Miami. I don't know why it's probably a nicer weather there, but we need to chase the the sequencer. So, so we need to collocate, try to be as quick as close in the physical proximity as possible in order to to have this latency and yeah, it's also a game of geography. You need to find a strategic

location. So, you need to take into account like where where the data is being sourced from. So also you reduce that but also the the the distance to the sequencer is crucial but it's a different game. We need to focus on the old-fashioned let's say models for for like trading application compared to what's blockchain was previously about like properly decentralized for example for a bitum doesn't matter where where we are pushing data from it's still like yeah basically this the same and especially for for Ethereum like the the data is being like properly propagated merged etc with mega is like a different came

from. Yeah. Like going back for a second about the question with pivot chain link because I want to double click. I'm on the forefront of chatting with the DAPs like dayto-day together with our go to market team and I can tell you that we feel more and more of those like free solutions being like compared like all the time. Like I I'm being asked about hey why you should choose between those three like what are the reasons like almost every other day or week and these kind of examples like the morpho style for push oracle is a tangible example like which ones are good for those kind of use cases and of course piv is probably like better performant right now on Solana for per markets and those

kind of uh applications but as Redstone my goal is to be the most comprehensive and useful Oracle for D5 applications. If you want to create something useful and innovative in DeFi and you need some custom implementation with an oracle, Redstone is there to help. We were the first oracle to support it. We're the first oracle to support liquid restaking tokens. We're the first oracle to create Lombard LBDC price feed. We are the primary oracle for security ties for their all tokenized funds including Blackrook Bidd Apollo Acred and so on.

And right now, this is by the way when we are recording it, it's not announced yet by by publication. I I hope it will. We are the primary oracle for mega if test net. So we are the official one. We deliver the price feeds over there. And let me double click on what Jacob said. We delivered on devet 25,000 updates per minute. It's crazy. I mean like this is the amount of updates that is already you cannot comprehend that number with your mind like how many updates per milliseconds even it is right um and we are going to keep pushing to the boundaries naturally like the performance is very much dependent on how mega itself is rolling out next updates to the test net and all the

other applications and then specifically about the prices we are going to deliver there it will not matter for RWA is that we are going to update once a day or even verbst fundamental rate that is also updated like roughly once every 24 hours it doesn't matter right but for bitcoin if or even the stable coins when there is like temporary arbitrage of USDT or USDC like mega gives you like just new map new spectrum of use cases because of its speed for the absolute professionals to tap into that super high frequency and that's where we are right now exploring many use cases that we haven't even seen being discussed on other ecosystems because they were just

outside of the possible map. So Marson uh Mega Eats advertising I think it's 10 millisecond or is it 100 millisecond block times or 10 millisecond block times? I think it might be 100. Um, regardless though, um, if you guys are pushing 25,000 Oracle updates per minute, then what is the like basically what I'm getting at here is if Mega East is claiming like they can do this many TPS and this many um this small of block times, but the Oracle provider can only update 25,000 or is currently updating 25,000 times per minute, then how does a chain have really really fast block

times, but is if the data that they're that that that they're getting um in each block isn't updating at the same rate as the block times, then how is that going to work? Like how is there not going to be a problem there if the chain is updating faster than the oracle can provide the data for certain price feeds? Let me just put a couple of numbers behind that. And I'm also very curious about this cuz uh there's 60,000 milliseconds in a minute and so 25,000 uh Oracle updates in a minute means it's about one every two milliseconds.

Yeah, absolutely. Right. So to to clarify all of that I don't know like what would be the final let's say uh block production rate with mega if like when we work on on devet it was like around one two millisecond per block time and that's basically when we are hitting that limit with 25,000 updates per single feed makes sense right now the test net is is way much slower I think they're making an upgrade but also like they allowed much many partners to join the work. That's probably what happens. So I think it's an order of magnitude slower right now, but we'll see like how it will evolve, whether they will be able to also let's say increase the performance with the number

of connections there are new connections are getting or whether it will stay at the current level of the test net. But yeah, you are right. It doesn't make like any sense to to push it more often than than a single block. But also we most likely will be pushing not only a single feed but probably like a few dozen of feeds. So then you got this multiplication. So the oracle has to update as many times. It has to be as fast as the block time essentially in in order for it to work. And so this must be new for you guys then like this must be a completely new dynamic because no other chain is this fast or maybe like avalanche and

stuff just you know I don't know is it new or is it has it been a big engineering lift? Yeah, it's it's pretty new. So, we design um a totally new node from scratch is you as being open like you need to sacrifice the the centralization. You you cannot be decentralized working on sub millisecond. So basically it's like a new design. It's like focus on the extreme performance but it's still like a very very centralized and yeah service. So you put first the reliability and the speed and then sacrifice

the the redundancy and the centralization. Maybe to give some more context on other networks in example on Ethereum like you have roughly like 12 second block time right and some of the feeds are updated once every 24 hours or even every an hour or even longer right so there are many blocks that the same oracle update is present over there and the uh blocker over here is not the performance of the oracles because I believe right now any reasonable oracle can update once every 12 seconds here it's more a question of the gas cost right because an ethereum mainet the gas can spike very high. So if you were to do those updates every 12 seconds, you can calculate how many updates there

have to be during a day and then especially when some project is doing their airdrop for example in Ethereum I remember when I layer was doing the airdrop like the gas like just went bananas and we've seen many other instances that the gas just went high. Oracle cannot say okay now for an hour I'm not updating because the gas is high someone has to pay that gas cost right and then there's a question is it an oracle then it's not the best business to run or is it the client that is requesting the price fee or is it a chain right and in Ethereum specifically case they do not cover the gas cost for oracles right so it's a question of economics as well understood all right um kind of zooming

out from the mega e um and into something that I know that you guys are working on which is more around the ZK oracle side uh but perhaps more of like a spicy pointed question for you is there's been a lot of talk around ZKTLS and actually quite a few use cases that have emerged one being uh powered by um a company called Opacity I believe and effectively that they're able to bring a lot of this this off-chain data onchain in a in a very trust minimized way in ter in in like a private Okay.

Um I I guess the question is why are oracles going to stand the test of time if ZKTLS is can be better and private? And how are you guys adjusting to this idea? And why am I wrong? Like why is ZKTLS not better currently? So I think right now the the LS is a bit like a like a yeti. I think some people claim that's like possible to do it like totally trustless way. I haven't seen like single implementation. I think there's a lot of progress. I understand like we started working with a few

companies on that. So but still you need to have this let's say trusted notary in the in the TLS system that will authorize that the the communication happened. So it's not like fully finished in terms of implementation. So that's one of the element. I think we are not there yet with the with the implementation of of cryptography and still you need to trust someone with the TLS. Second one is about performance. So proof generation still takes a lot of time. So it's it's counted in in minutes. So comparing to for example the mega if the

the other side of the spectrum that we talk about when we are talking about s milliseconds then we are talking into let's say like a few minutes maybe half an hour etc. It's a lot of time to to generate a proof and then there's like a cost of validation which is also like pretty expensive. So we cannot have like the the the oracle for not only ps but let's say like a standard pricing oracle that is being run on a on a z case. We are not there yet. But I I totally agree that it's like probably uh direction when the services will be moving and that's why we are also also implementing this approach especially

for for the solution that doesn't require speed like the proof of reserve but require like an extreme certainty that something So then you guys are going to go and implement ZK based oracles then and that is different from ZKTLS. Yes. So for the I think there will there's place for different set of solutions. So ZKTLS is great for for getting data data from the endpoints. There's also a very interesting stream of use cases when you are like generating ZK proofs for the old fashioned email protocol. I know

that a lot of let's say listeners may find it hard to believe but the trafi and the auditors are still running on sending an email that certify that you know this funds hold so many let's say assets under management. So that's another stream and yeah we'd love to probably pretty soon put like the first pilot using the the TLS notary. I think the the tech is getting there and it would be like fascinating to to offer something complimentary for the speed.

Maybe to tap into that a bit more. uh some of those ZK solutions claim themselves as Oracle killer or like chain link is going to be obsolete and all red stones of the world are not going to be needed which I believe is a bit of a a marketing gimmick and overstatement because there is a lot of customizability of oracles that is required that with ZK will not be possible anytime soon but as Jacob mentioned like for specific use cases around less frequency and uh usually more traditional finance institutions related. We do see already like requests and ideas how to implement that and personally I keep very good relationship

with some of the solutions around ZK and we just want to partner with them because we are a distribution channel. People already trust us and know that if we deliver something to the market, we are going to first make sure it's validated and it's secure and reliable, right? Uh so if we are to go with one of those providers, we would first do like extensive testing phase and then invite them to our client base like and then creating the specific solutions. So I wouldn't necessarily say it's a killer. It's more like even expansion of what is possible with us ourselves because we are not also an experts experts in ZK TLS and other zero knowledge technologies.

Yeah. and the and these guys are I guess from the Oracle perspective just to close this loop what value does ZK provide that by by bringing ZK and oracles together you you guys are able to do things that you weren't able to do before. So like what what is that that benefit and that advantage via adding ZK? For example, if we all agree that the that the source of truth is I don't know like a central end point or certain domain in internet for example the one that that published the results of the voting and everyone on the market let's say agree that that they're going to trust it then you can create using that TLS

notary a cryptographic proof that the data came from that source and it wasn't altered along the way so that's the value so you do needs to trust let's say like a cryptoeconomic mechanism. You basically trust the the cryptography that is based on the let's say the certificate on the domain and using the the also the knowledge about the notaries that were involved which is still pretty the trusted part. So that's how the zk can help the oracle space but also as Martin mentioned it's a part of the of the larger machine. So it's for us like ZK data will be another source of data as we're currently using

data from exchanges from the XS from certain aggregators so it will be like another source of true but basically as you may also like get from the conversation of us like in oracles you get different modules modules responsible for pushing data for validating data etc. So it will be great addition but still a small part of the whole system. Got it. Gentlemen, it's been a pleasure chatting today. I think uh I think we're about out of out of questions here and just um I think it's a good place to wrap. I I appreciate the the commentary around the state of the ecosystem as

well as the updates as to where you guys are focused specifically around high performance chains. I think that that becomes the norm in the coming coming months into 2026. So being prepared for that with other similar chains like Mega E I think is going to be a successful strategy. Um and I appreciate the technical breakdowns as always. Marson Jacob, it's been a pleasure. Thank you guys. Thanks for having us. Yeah, thanks. It was very fun. Uh maybe like just a finished touch. Uh we've got a lot of other stuff uh coming into the pipeline in the upcoming weeks around Oracle extractable value. Monet itself is you know expanding its test net and then

eventually coming to the mainet. Babylon is launching mainet and we are very much focused on the bitcoin narrative created bespoke indexing technology for gathering the data directly from the bitcoin chain to create the proof of reserves. So I would just say in the Oracle game we are here for the long term and we are here for the innovation on the tech side. So we'll keep on crossing paths and hope to come back to the rollup soon. Fantastic guys. Thank you so much. It's it's been a pleasure.

Yeah absolutely.