On The Brink with Castle Island - Mike Cahill (Douro Labs) on the Evolving Oracle Landscape (EP.476)

Episode Date: November 20, 2023

We sit down with Mike Cahill, CEO of Douro labs, to talk Pyth's retroactive airdrop and launch. In this episode: A brief history of Pyth What was the problem with legacy oracles before Pyth How is Pyth is technically differentiated from other oracles The importance of introducing confidence intervals into prices How does Pyth achieves cross chain data delivery Disintermediation in the oracle market Major proof points that convinced Mike that Pyth could work Pyth's traction on various chains Which blockchains Pyth does particularly well on Mike addresses the 'collusion FUD'? Why Pyth oracles are not like LIBOR How the Pyth retrospective airdrop works What's happening with Pyth governance on a go forward basis? Further reading:  Pyth, How the Retrospective Airdrop works On The Brink ep. 364, Solving The Oracle Problem

Matt Walsh and Nick Carter are partners at Castle Island Ventures. All of these expressed by them or the guests on this podcast are solely their opinions and do not reflect the opinions of Castle Island Ventures. Guests and host may maintain positions in the assets discussed in this podcast. You should not treat any opinion expressed by anyone on this podcast as a specific inducement to make a particular investment or follow a particular strategy, but only as an expression of their personal opinion. This podcast is for informational purposes only. Hello and welcome to On the Brink.

I'm Nick Carter today. We're sitting down with Mike Cahill, the CEO of Dora Labs to talk about Pith. We've had Mike on the show before about 100 episodes back. Of course, Pith has made a lot of strides and has become one of the dominant Oracle providers serving a number of different blockchains, launching Pith 2.0, and now, of course, they're launching a retrospective air drop. I will share some links on that on the show notes. It's always great to have Mike on the show. And as a disclaimer, Castle Island is an investor in Pith. So just as a heads up there.

Let's dive right into it. Okay, I'm sitting here with Mike Cahill, the CEO of Duro Labs. This is a very auspicious day, very exciting day. The retroactive Pith AirDrop has gone live today, and governance is now live, and the token is live. Very exciting. We've been awaiting this day for some years now. Mike, thank you for joining us.

Thanks, Nick. Yeah. super excited to share this with you today. Awesome. So Pith, give us a little brief history of Pith. Give us a refresher. You have been on the show before, but for our newer listeners, give us the history of Pith. Yeah.

So Pith started in 2021 by a group of trading firms and exchanges that included Jump, Virtue, LMAX, GTS, Miox, and a few others. I was working at jump at the time, and we were really the leads on the technical development from the early days. Quickly, the group had grown or has grown to include many other trading firms and other contributors to the network. Cut to today, there are 100 publishers, institutions within the network, and much of the technical development is done by Dural Labs, the company of which I'm the CEO of. Dural Labs and Jump are entirely independent of one another. They have no direct relationship, shared ownership, but we do have some shared Jump DNA in the form of some

alumni. So there's about 20 people at Dural Labs and about half of us have have worked at a jump at one time. So Pith was able to amass in a relatively short time just an incredible roster of publishers. So a lot of trading firms, exchanges, venues, basically, that are contributing data to the network. How did that happen in the first place? How did you manage to achieve that incredible acceleration? The idea resonated so well with the audience of contributors to the network. So there was basically a void within the market data space where we had trading firms accumulating troves of market data and then never monetizing it.

And at the same time, having the value of market data from exchange. exchanges being worth many billions of dollars. Now, there are some complexities to why you couldn't really open source this in the off-chain world. But when it comes to solving this for an on-chain world or solving the Oracle problem, this ends up being a much more attractive solution. So just being able to describe the idea, I can tell you from the very first days of the concept around Pith, we would present this as an idea to trading firms. And within five minutes, they say, yeah, I get it. I'm in. And it was just mind boggling to see how fast it resonated with them. They said, yep, we get it. You know, blockchains give us the ability to make an entirely

different type of market structure where some of the components are modular and composable. And one of those components can be the market data feed and we can get into a position to contribute to that market data feed. We would be monetizing our data for the very first time. That's cool. And it would create something that the blockchains need. So it was one of those like win, win win things and it just resonated like wildfire. So we had, we've had a pace of three new publishers joining every single month. And these are all household names for, well, most households that are probably listening to this podcast. So it's all the large trading firms. I don't know if you can say household names generally, but maybe within the, you know, very crypto markets focused households.

Crypto markets, financial markets, too. Yeah. So there are some that you wouldn't expect. Susquehanna, Jane Street, DRW, Hudson River Trading, flow traders, Optiver. Like, those are not names that I think would be confined to just a crypto markets. They've existed long before. the crypto markets and do a lot of other things. So I think it's more expansive than if I just told you it was trading firms and it was crypto, you would think, okay, fine, this includes Oros, Wintermute, you know, matrix port, like a bunch of very crypto-native names, but it really creates a landscape that is highly inclusive cross-assets. And so a lot of these firms leaving the sort of exchanges and trading venues aside,

these are normally consumers of data. But so in this case, they're also. basically producing data and contributing it to the system. Exactly. Yeah. So there's like some elements of the Airbnb model here where previously there was just hotels that were providing the supply for places to stay. And you can think of trading firms as having like a whole bunch of houses or apartments and now being able to rent them out for the first time.

They were just sitting on this as almost like a found commodity or found resource. and it's a big business. So it allows them to enter into the competitive business of financial market data, really high quality financial market data. And there's been a bit of a transition, I guess, in the Oracle space in crypto, where you had kind of third parties taking Oracle data and inserting it on chain. now the way Pith envisions it is that the publishers are kind of the originators themselves. They're kind of closer to the source.

Is that fair to say? Yeah, exactly. So Oracle Model 1.0, call it, was a good one for bootstrapping the very first DFI primitives. And this is the model that every other Oracle really follows. And it's predicated on the assumption that all the data that you need for blockchains is available on the public internet. And the reason why I say that is because blockchains are public. So as soon as you publish the data to a blockchain, it's de facto public there. So the assumption was we can get all the information we need from the internet,

but we need to be able to publish it in a trustworthy way. And so the thing that was jerry rigged up was let's go get a bunch of nodes who all will publish the same data. And then it'll be really easy to filter out exceptions. They're all publishing from the same source. So it could be that they're coming from Yahoo Finance or Coin Gecko or Coin Market Cap, but you get 10 people to report the price from Coin Market Cap. If one of them is wrong, it's very easy. You know, the other nine are correct.

And so you just filter it out. And, you know, it worked. It's very jerry-rigged, but it worked. And it worked okay for stuff like compound that don't require low-latency oracles. They don't require low-latency updates. In fact, the update speeds for these first oracle networks were on the orders of once an hour or every 50 basis points and still are actually. Now, I have worked in trading for over 15 years at pretty much every spot within the market microstructure. So I started off at an

investment bank. I went to a high frequency trading firm. I went to an exchange. And then I was at Jump Crypto when we were working on the idea. And so it was very native to me to think about the types of data that needs to be brought on chain needs to be the lowest latency. If you're using something that comes through these jerry rig tops, it's going to be super slow. Like, that's just like a high frequency trading firm's eyes would turn into dollar signs if somebody told you that that's how they were constructing a price. Like, you know, that's, that's silly. But because we were so focused on building defy and saw this as a critical bottleneck, it wasn't like, well, we can make this as an opportunity. We're just like, this will never work if it's not solved. And so that was why we wanted to do

it. The opportunity set within all of blockchains working is so much greater than like whatever tiny little trading opportunity there would be to arbitrage some slow Oracle. And so that was the idea behind Pith. And it required an entirely different architecture. So in the old model, it's nodes scraping from internet websites and then pushing those updates on chain at predefined in intervals. So as I mentioned, once an hour, every 50 basis points. The architecture that Pith was designed with was get the publishers to publish directly, cut out all the middlemen. Don't go to the internet first. Don't let somebody then watch the internet to then publish. Do it directly on chain. You solve two things here. Number one, you cut out all that latency,

right? That's the number one most important thing. And then number two is you actually ensure that there is the appropriate licensing to be able to publish that to a public blockchain. If you're taking data from Yahoo Finance, you're probably violating the terms and conditions of that website. You can go click on them and say, hey, what can I do with this? Typically, you can just use it for personal consumption. You cannot do what they call uncontrolled broadcasting. Now, if you're a trading firm, you have your own data, for instance, the trades that just occurred, well, there's no dispute. Right. You have that data. You know that you are allowed to publish it. If you're in exchange and you're selling data, but you also wanted to publish it, fine. It's your data. You can,

you can publish it. And so that was how we decided to create this architecture that was future proof from being bottlenecked at some point. Because at some point when the old model, which was like scraping stuff from Yahoo Finance, when they start getting into U.S. equities and trying to publish the price of Apple on chain, NASDAQ is going to come and sue all those 10 nodes who are publishing the price and say, hey, look, no, you cannot publish this. This is uncontrolled distribution. You violated terms of conditions, taking it to court. This is a, you know, $3 billion revenue business for us.

You know, you need to shut down. And so that was that was the insight we saw. When we looked at the solution, the first oracles are like, these are not markets people. They do not understand the nuances of financial markets. We need to get in here and solve this in an appropriate way. So, you know, one interesting thing that I noted, obviously following the development of the market data market in crypto is, you know, traditionally in Tradfi space, the exchanges monetize the data. In fact, that's a huge revenue driver for them, really meaningful. In crypto, they've been reticent or unwilling to do that historically. So most exchanges, barely commercialize their data. What is your explanation for that status quo? It's just too early in the market. It's purely the nascency and fragmented competitive nature.

You know, back in the 20 years ago, exchanges didn't make any money on market data. There was obviously consolidation and the largest exchanges started to make more money. And it's gone up every year since. It's 20% of the revenues and $6.5 billion in 2022 revenue. Now, there are exceptions to the crypto market data kind of paradigm. Coinbase started charging for market data a few years ago, not really enforcing it to the extent with which, say, you know, NASDAQ enforces their market data licensing, but sort of doing it with a light touch. And that has, from what I understand, been becoming a little bit more institutionalized, let's say. Now, you know, Binance doesn't currently pay wall or charge for its market data.

And my view of that is we're still very early in the growth and success of crypto as an asset class from institutions trading it all the way out to, you know, just kind of common acceptance of it being something that people want to be exposed to. As the market matures, what I would expect to happen is for the same. composition of revenue to transpire within these markets. So market data becomes a bigger component of it. Perhaps things like preferential access become part of the product offering. I think that there'll be, you know, probably more creativity around the make or takeer fees, although there already is lots of creativity there. And so I think it's just a,

you know, it's just a matter of time. Then there's the crypto markets where, or, the like the defy markets where it's it's different as well because you know as we talked about the off-chain world is kind of fully integrated um the market data for nasdaq has to come from nasdaq because that's where you're trading um but when you go and build a dex on chain if you're trying to build a competitor to the binance perpetuals market you can use some of the composability and say actually you're going to want to get the best possible market data feed in the world. So I'm going to use PIF.

That's all they do. They specialize. They focus on this. I'm going to use the best possible bridge for it. And then you go choose whichever bridge. And there's going to be some segment of expertise that you as a builder were going to focus on. Maybe it's around the front end you acts, but you want to outsource some of the other components like the matching. Maybe it's around the matching itself and you wanted to bring a new novel, a novel architecture to that component, but then outsource some of the front end development.

And so all those options are available. And I think why, I think that's the reason why we will see market data continue to evolve and become a more meaningful revenue component of the markets, both in crypto off-chain as well as on-chain. So in terms of, you know, you're talking about high frequency, low latency, as compared to the prior oracles that were dominating for that, Solana was the obvious. Domesol for your outputs. You've also had an evolution there. So you had Pith 1.0. Now we're on Pith 2.0. Tell us a little bit about the technical evolution there.

Yeah. So that's all correct. So for everyone that isn't fully aware, the idea for Pith was always to be cross-chain. When you look at the landscape in 2021, and you want to be cross-chain, you have to choose the blockchain with the fastest block time. if that's the thesis and the core thesis around Pith is that defy needs trustworthy data at low latencies.

And it worked out really well. There are some characteristics and attributes that were table stakes for Pith to be successful. So one of the criticisms of the 1.0 version of Oracle's is that they're mostly black boxes that then publish on blockchains. So from the chain link model, you have not. no idea where their data sources are. They don't tell you. They just tell you that they're from the internet and their third party. But here are the nodes.

So those, you know, those 10 people that scrape the data, they'll tell you who those are. They don't tell you what the aggregate is looking like. They just tell you when it gets published onto a variety of blockchain. So on ETH, it gets published, you know, every hour. And there's, you know, maybe 150 symbols. But then on Avalanche, they have only 50 symbols available. So there's a lot of black box type situation. tables stakes for Pith to be, you know, the preeminent solution for solving the Oracle problem was it needs to be fully auditable.

Anyone should be able to choose to look on-chain directly to verify that the publishers are publishing accurately where the data is coming from, have it be all in the open so that no one has preferential access in advance. It's all aggregated on-chain. And for that, we needed a public blockchain. And so Solana was the obvious choice. So Pith then grew. It became the largest Oracle used on Solana within seven months of its deployment, about 50 applications, and over 90% of the market share there in terms of the total value secured.

But Solana also became very popular. And so there were these NFT minutes that would go on. And there was volatility. And there was a lot of defy big protocols like Drift and Mango and Zeta were becoming incredibly popular and there was all this volatility. Now, if you're competing in a general shared runtime and one application in an Oracle and another application is doing an NFTment, you have these priority fee markets, not everything is going to land on chain.

Well, oracles are pretty important to have on. And so we realized that we were chewing up about 50% of the block space just to be able to publish the almost 100 symbols at the time. And so if we needed 50% of the block space, it was going to be a problem. We couldn't grow to more symbols. So the solution to this was to go create our own environment built on Solana technology. So there's an SVM, Solana Virtual Machine, application chain called PithNet. And that's where all the PIF activity now occurs. This is the exact same transparency characteristics as Solana Mainnet. So you can still use a Solana Block Explorer to validate the providence of every

single aggregation. All of those publisher keys are there. And it only runs the Pith application. Right. So there's no competing for block space with other applications. The publishers or the validators in this network. So it has some of the decentralization requirements that we wanted to secure. But from here, Pith is able to increase dramatically the scale of symbols. There are over 350 symbols, increased dramatically the number of publishers before there were 50 on Salon, and now there's nearly 100, and be able to distribute to all the different blockchains at the same latency. So there's this one updating high-speed market data feed on PithNet, and any of those

updates can be delivered to any of the blockchains that Pith is connected to. It's connected to 40 different blockchains today, and any time a Pith price aggregate is created, that Pith price aggregate can be delivered to any of those 40 different blockchains. So Pith doesn't just dominate on Solana. It's actually dozens almost of blockchains where you have pretty overwhelming market share. Is that kind of a deliberate strategy was go to some of these smaller, newer blockchains and be the kind of de facto leader there? Yeah. We realized because Pith launched on Solana, When there were already oracles, right,

there were other Oracle models available on Solana. And then there was Chainlink was going to come. And I remember, you know, Twitter was telling us that Chainlink was going to eat our lunch when it deployed on Solana. They never got above 5% market share. And now they don't even list it as a ecosystem that they're available on. So that was very telling for us. If you're early and you have a better product, then you can, create the kind of canonical oracle paradigm for that ecosystem.

When I say to people that Pith is live on 40 blockchains, they just don't believe me at first. And then they start to like say, what do you mean by blockchains? And then they unpack like, oh shit, you really, really is. Pith is really on so many blockchains. And then I tell them the next stat is that on half of those blockchains, Pith has over 50% market share. And then on a third of those blockchain, Pith has over 90% market share.

That's just bewildering. So Pith is the dominant and canonical oracle for all Cosmos chains, for instance. So anyone that uses a Cosmos application that requires an Oracle on any of the variety of Cosmos chains from, say, injective or any of the others, it's most likely going to be Pith. The market share is commanding there. If you use an application on another high-throughput chain like Optos or Suey, there's a greater the 90% chance that it's the Pith Oracle. And so we saw a lot of value in establishing beachheads of deployment within chains

where the features that Pith optimizes for, like low latency and high throughput, are part of the offering there. That then means that Pith has a bit of catching up to do when it comes to Ethereum Mainnet because it hasn't been deployed there all that long. all of this deployment, these 40 different blockchains has only occurred over the last 11 months. So in 11 months, Pith has deployed on on so many different blockchains. But over time, you know, the intention is that as protocols upgrade and look for a V2V3 and perhaps are doing something on a layer two within the EVM landscape, that's when there's an opportunity for Pith

to kind of step in. And we've seen that happen on many protocols who had longstanding relationships with previous oracles. One example of that would be synthetics, who in their mission to upgrade to a V2 that was an Oracle-based perpetuals market, they looked at the landscape, re-evaluated, and governance decided that the best possible Oracle at the time was Pith. They deployed Pith on optimism, and today Pith is the primary Oracle for synthetics, and will be when they deploy on base as well. So you also were telling me a bit about your confidence interval approach and how that can really help consumers, I guess get more transparency into the data and also manage volatility. So I think that's kind of a unique thing to Pith.

So tell us about the notion of adding confidence interval data to the data itself. Yeah. So, you know, very first principles in. creating the world's best Oracle network, you have to, and if you're going to design it in an entirely bottoms up first principles architecture, you say, well, what would be the most optimal thing to be able to publish as well? If you're highly constrained and you say, well, we're just going to have to deal with the data that's on Yahoo finance or coin gecko because, you know, that's where we're getting our data from, you know, you can only output the price that they output.

And the price that they output is a single price and it updates every 30 seconds. So, be Because we were going to redesign the entire infrastructure, build it up from the ground up, which is going to take a tremendous amount of work, but was the right way to do it, we had a whole lot of new toys. And so one of those was being able to articulate in real time what the confidence is of any particular price. And so Pith will publish a price. And that price will be the combination of all the inputs from the publishers. But it also publish a confidence interval. And so the confidence interval comes into effect typically during fast moving markets or

markets where there are idiosyncrasies within the market structure of a particular exchange. I'll give you an example. At any given point, there's not just one price of Bitcoin. There's a price of Bitcoin on Binance, on Ku-coin and Upbit, so on. Usually those prices agree. And they tend to agree because high frequency trading firms are incentivized to align them back into place based on arbitrage opportunities. But every now and then, they diverge.

And that divergence needs to be articulated in the market data feed. One reason why they might diverge is upbit. It's the largest exchange in Korea. Sometimes we'll have discounts or premiums based on the idiosyncrasies of monetary policy in Korea. So it's a highly restrictive, non-delivable currency that Korean won. And so it's very difficult to freely transact in it. So if you're an arbitraiser, you don't have defungeability on the fiat side to be able to smooth out those markets. And so it's not uncommon to see the markets be out of line during volatility for, you know, pockets of geographies like that. Or it could be that there are just very fast moving markets and they're unaligned for a while. So the confidence interval represents this phenomenon.

in real time. And every price comes with a plus minus number as well. So the price of Bitcoin could be 36,300 plus or minus 200. Users, application users can take that information and they can determine what they'd like to do based on it. So if that confidence interval becomes very wide or the price becomes uncertain, maybe it's a good idea to limit some of the transactions that are happening. Now that you know that there's a bit of uncertainty in the market, maybe you want to limit it. Or maybe you want to be more conservative, perhaps with contributions of collateral towards a loan. It's typically during these ephemeral market spikes where things go wrong. For instance, if you looked at like what the attacker did in the Mango instance, they figured out that

the price for Mango was coming through another Oracle called Switchboard. They manipulated the direct feeds where the Mango token was being traded by making the market go up on very little liquidity. And then they took advantage of it by taking out a big loan on the artificially inflated price of Mango at the time. If they had looked at the confidence intervals, because that exchange was out of sync with the rest of the exchange that were traded in Mango token, they could say, you know what, we don't want to let, this high contribution margin go towards this big loan to value position that this person is taking on, because then when it reverted back to kind of equilibrium, it was all this bad debt in the system. So it's just to maybe summarize what's going on here. In V1 Oracle world, in some senses, some of these DFI protocols were insulated from significant volatility

is because the oracles didn't update frequently. And so you would get this kind of like unreality where oracles would be slow to report and that would occasionally protect D5 protocols from volatility, albeit because of a flaw in the actual technical delivery of the data. And then as we move on to higher frequency oracles, you now want to take highly compressed sort of medianized data, and almost decompress it in a way and add in additional metadata such that these protocols are

being fed reliable up-to-date data, but with the additional metadata such that they can sort of manage their reaction to extreme volatility. Is that kind of a fair way to summarize it? That's a perfect summary, Nick. Thank you. So it always kind of cracked me up that, you know, occasionally defy protocols were protected from volatility, not through any sort of deliberate means, but just because the oracles were relatively slow to update. It is totally wild. And it's something that, you know, we'll tell, I think, future generations when, and we all know the instances of it, right? It was like the three hours loans. and everyone's watching the Oracle.

They're watching the centralized exchange. Like the centralized exchange is there. Fuck, it's not updating. This is great, though, because now it's not being liquidated. And it was just this weird thing. It was like, it's great because it's broke. I don't really know how to feel about this. That's not exactly the paradigm you want to be in

where it's protected by virtue of the actual technical flaws of the system. Exactly. You had a good talk at Breakpoint actually recently. you talking about some of the major proof points you wanted to see to be persuaded that Pith could be a dominant Oracle provider. Can you elaborate on that? Yeah, there were three things that we needed to figure out. So the thesis was blockchains need high throughput, low latency data. And, you know, proof point number one was, well, where are we going to get this high throughput data from?

if we go and ask the exchanges who are the, you know, kind of the largest data sellers today, they're going to say, you know, get out of town. We're not going to give this to you. We make too much money selling this data. Why would we ever experiment with something like the Pith Network that is going to build up blockchains? We don't even know how we feel about blockchains. This isn't great.

So number one was like, can we amass NASDAQ's data without asking them? directly. And we would have to do that by getting a big number of contributors. And so that was the proof point number one. And I described that process a little bit in detail earlier in the conversation. There's just all these trading firms that have all this data. And that was a eureka moment. And that is where we really, we really learned that we had captured like a little bit of a miracle in a bottle. And we were on to something. So I was proof. point number one, proof point number two is like, hey, now that we've got all this incredible data sourced, are people actually going to want this on chain when there's other stuff?

Because right now, they only seem to care about, you know, the trust element of it. This is adding a different complexity to it. And so proof point number two was, well, we'll be on Solana and, you know, chain legal deploy there, will Pith be able to stand out. And so I talked about that a little bit already within. seven months, Pith had over 90% market share and has maintained that today. So then the third one was, well, when Pith goes cross-chain, it's going to have this entirely different type of architecture to read an Oracle. You're going to have to pull prices from PithNet. No one has ever

done this before where they pull prices to another chain and they have to read it. Obviously, they understand why there's some benefits to it because, you know, having something something that updates every one hour or 50 basis points is certainly suboptimal. So are they willing to do the work? And we've deployed on 40 different blockchains. And there are now 250 applications that are using PIF in 2023. In this year, in the past 11 months, three new applications have integrated PIF every single week.

That's every week. There's three more applications that integrate PIF. And this was during the bare market. But these are just wild numbers of how much people really wanted to do this and really depended on the data. And so that was incredibly exciting to see. I think that was the one, I mean, that was the last proof point. And that brings us up to kind of where Pith is in its evolution in the network.

But it's, you know, it's certainly wanted. It's certainly needed. And the market has chosen that this low latency pole, or. Oracle architecture is the current best possible use case for an Oracle in all of crypto. One thing I want to touch on is this, I guess you could call it FUD. I mean, I don't like to call things FUD because it often seems like unfairly dismissive of maybe critiques that have some validity. But what is your answer to this critique we see around collusion potentially in the Pith network?

Yeah, it's something that I was, I would always just brush off for a really long time because it's, it's just so easy to disprove and it's incredibly naive. So I just assumed that it was like, it was kind of farcical. But I've just been seeing it more often. So I figured I should, you know, since it is so easy to prove, I should, I should talk through it. But basically the, the, the, the, the fund is something like, well, because there's all these trading firms and exchanges in the Pith network. Um, surely they are trying to manipulate the price so that they can take advantage of that bad price on chain in a protocol. And, um, and so I'll, I'll describe, you know, why that doesn't work. So first off, there is a hundred publishers in the network. And they're all announced. They are all, as we described, household names.

They're all, you know, they're all published. They're all public. But there are names like Siebel Global Markets, Jane Street's Tuscan and a DRW, also Binance OECS, you know, you name it. It's every big institution, many of which are regulated, you know, those big financial firms that operate in traditional markets are generally highly regulated. Okay, but still, like, let's say, all right, each one of those symbols that they contribute to has some subset of publishers for it.

And it usually ranges from say 15 to over 50. So the, you know, the collusion conspiracy theory goes, well, there's going to be some subset who figure out who the other ones are in a particular symbol. Now, there's a minimum number of quotas that have to be live. So let's say that we've gotten in a symbol you've got like 15, 15 publishers there. You have to figure out which. let's say eight are going to collude together. And then they're going to go and manipulate the price, such that the output is favorable to them.

And then they're going to take advantage of it on chain. Now, the problem with this is, first off, in the LIBOR scenario, you basically had an issue. Yeah, I was just about to mention this sounds reminiscent of the LIBOR. Well, the LIBOR, it is exactly. The reason why LIBOR worked as a scandal is because there's no way to check. it. It was a black box and whatever they said with the output. And you couldn't go back and be like, are you sure? There's nowhere to check it. It's like they created it. And so in this world, it's incredibly easy to check. You just say, well, Pith is supposed to be publishing the price of Apple right now

and it published the price of Apple incorrectly because the price of Apple, you can go look at Yahoo finance after 15 minutes when that data is free. Anybody can go look it up. It's in highly, highly easy. to go and check in retrospect. Just go look and say, all right, this was the price that was pushed on chain. Here's the liquidation. Oh, and here's the people who liquidated it. And then this is what they did with the pool of money. They distributed it.

It's so far fetched because it's so easily trackable if they were ever doing it. Because the market is just representing a market that exists somewhere else today. And it just requires all these assumptions, you know, for this, um, kind of cartel of people who are trying to manipulate it and as soon as they do they've ruined the Oracle, right?

So this is like a one time thing where when they're busted like their own reputations are are wrought. The Oracle's raw like it's just it's just so incredibly far-fetched. Yeah. So you're, as I said,

today I think the retroactive air drop goes live. It's quite the undertaking. I mean, you're air-dropping. or rather Pith is being AirDrop to addresses on something like 30 blockchains, is that right? It's like a huge. Yes. Yeah, it's huge. Yeah, because it's a usage air drop.

So it's a retrospective usage air drop. So what's funny about it is Pith is a B2B service. Right. So your household defy user will have heard of. synthetics, but they may not have ever heard of Pith. But synthetics doesn't exist without Pith because it relies on it. So what's pretty cool about this is the Airdrop goes to the users of protocols that are using Pith. So synthetics users will all receive an air drop. Now, synthetics lives on Arbitrum. And so anyone who has been trading on it and using their EVM wallet will find out

whether or not they're eligible for the Pith AirDrop based on that activity. So there's 30 different blockchains, as far as we're aware, this is the largest cross-chain usage air drop ever, ever done. And there is over 75,000 wallets and users that are in scope or eligible. And that air drop goes live today. The other development is that now the PIF protocol is going to be controlled by the token holders. So, you know, obviously we see it. array or different governance strategies, what do you expect to see here from governance? What parameters will token holders have control over? Yeah, the idea was always to have this be a fully decentralized Oracle.

And the right mental model for it is to be run and operated by the stakeholders, the network. So publishers, the application users, and then the end users as well. So there will be some end users that say, hey, the application, I'm, you know, excited to be using is using Pith. I want a certain symbol to be listed there. Let me go to the Pith governance forum and propose this to be a new symbol. And they'll be able to have their say. So there's eventually going to be all the parameters within the network, but there's a few in the early days that will become important.

The first one would be the adding of new publishers. This is obviously important and ensuring the appropriate reward distribution or fee distribution to these publishers. The second is fees themselves. The way that Pith is delivered, each time a price is pulled onto another chain, there is a fee that comes with that price. It's currently set to almost zero, but the token holders can decide to change that. Major technology upgrades, it's my prediction that there will be contributors like Doro Labs,

who will have tech upgrades from time to time. This will then be run down with the community. They will have their opinions on it. There'll be other contributors that may propose different upgrades as well. And the community can kind of decide on that. So there's really like no limitations to what's sort of possible. Although in the early days, it's going to be pretty well defined around the attributes that people really care about

within the network and making sure that things are set up to grow. The first thing that will get set up is how the governance model will work. I'm personally a fan of representative governments. So councils rather than single token voting, which ends up becoming a plutocracy very quickly. So that would be sort of my opinion of it, but I'll make my opinion known. And it'll be up to sort of the token community to decide how to implement it. So I know Duro is now just one contributor to the Pith protocol, but what can you tell us about the future of Pith, future directions for the network? Yeah, it's difficult to predict.

But I think that what seems to be working really well is the continued deployment of additional asset classes, the onboarding of market makers. or publishers, data providers, and ensuring that the needs of Oracle users are met. So if I were to make a prediction, it would be that more of the same continues to happen with additional asset classes being added as there is demand for it on chain. And then just kind of speculatively, the Pith model is really around incentivizing publishers of proprietary data, right? So something that they own and they only own and incentivizing them to publish it to public blockchains.

And so that model works really well in coming up with a novel source for financial market data that is very, very timely. And so that model would also work for other things. I don't have a prediction today of what it could work for, but areas where there are sets of Waldgarten data that could be. incentivized to be made public, you know, those sorts of things could potentially fit into a Pith model. It's been great to follow your growth over the last couple of years and your success. Really important milestone today. Mike, thanks for joining us.

Thanks so much, Nick.