So, we are ready for our next report. For this conversation, I’d like to invite Mike from Dual Apps to talk to you about all the things we can do to make DeFi a lot better. So please give Mike a big round of applause. Let's talk about how to make DeFi better. Let's start with the first slide—this is how I find friends at crypto conferences.
So, DeFi sucks. Of course, we are all involved in DeFi development here. So, we need a little context on this. DeFi sucks when compared to the boss. The boss is traditional finance ( traditional finance ( TradFi).
We are obviously trying to build new financial rails. Of course, there are problems with TradFi or CeFi. Think about trying to change the country. The onboarding there is pretty terrible. Sending money is always accompanied by fees.
Getting a decent interest rate is very difficult. But when it comes to things that don't happen every few months and don't annoy people, things are pretty good there. Robinhood, Charles Schwab, NASDAQ—these are prime examples of what I'm talking about. You actually have extremely low fees, zero fees on Robinhood and on Schwab. The execution time of operations is incredibly fast.
500 milliseconds or less for retail, around 300 milliseconds or less for small institutional trades. And if you trade as a large institution in a data center on one of the top exchanges, then we are talking about less than one millisecond. And trust, of course, is based on trust in brands. So if they are wrong, you can sue them . So there's no programmatic trust here, but they don't "dump" you all the time.
They, of course, use trust in certain ways, for example, by introducing certain commissions. Now, your average DeFi app is pretty weak compared to these three categories. For the most part, the commissions will be far from the consistent zero commissions you get on some of these trading apps. Your transaction time is typically around two seconds or more on most blockchains most blockchains , and trust in applications is mostly relatively centralized and unbranded. So there's really no one to sue.
We've all heard of " We've all heard of " rug pulls." Pyth makes things a little better. Today we will delve into some of these topics regarding the three categories and how it helps improve them. But we are starting to make progress towards TradFi. So, what is stopping DeFi from competing with TradFi?
The sad reality of new technology is that there is one thing that new technology is famous for: technology is famous for: security and a new form of trust. And then there is everything else, that is, existing market players. The reality is that the people who really care about this thing get on it very early, and they're all very frustrated with how long it takes for others to join. The problem is that everyone only joins when other elements become on par with the existing players. Now we will look at an example to show how this actually works.
If you go back to the If you go back to the 90s, Amazon was the first type of online shopping. It competed with the Sears catalog at the time, right? The Sears catalog was like the Amazon of its time and existed for over a hundred years. You flip through a page You flip through a page and you can load 10 images in about a second, right? To flip through a page of 10 pictures on Amazon via a 56k modem would take about two and a half minutes to load.
Of course, it wasn't nearly as good as its competitor's. Here we are . Oh, I'll walk you through it . Not as good as the competitor's. And here are all the things you do when you start talking to someone who doesn't quite understand the benefits of connecting to a store: real-time inventory updates, no need to call or do a mail order.
They don't understand any of this because all they see is that the Sears catalog is much faster. Now fast forward 10 years and you get a hundredfold speed improvement. You can download 10 images in about one second. And that's when things get interesting. People usually don't know what's going on on the backend, but that's because the infrastructure is constantly improving.
So, we moved away from 56k modems. After 10 years, we had broadband access. These lines were just being laid, so they weren't available to everyone at the same time. Then fiber came along, and we started moving into mobile technology and 5G. Now you can go into your phone, open Netaporter and watch real streaming video without any loading time, and it all feels very seamless, and people who are n't tech-savvy just say, "It feels better," right?
It's that innate ability to notice when there's no friction or when everything doesn't feel like some kind of awkward compromise. Now, with regard to blockchains, as I mentioned, the main innovation that I'm sure is exciting for all of us is trust, right? So, before the advent of the web, trust was personal, completely non- scalable; you could only interact with those you knew, or you could communicate via the web. Now you can interact or communicate with communicate with anyone in the world, but it was difficult to achieve trust. Everyone has heard the story about buying an Xbox on eBay and getting a box of rocks, so there are a lot of issues with trust, credibility on the internet.
Web 3.0 and crypto give you, you know, the principle of "it's impossible to be evil" rather than just "don't be evil" just "don't be evil" . And we have this programmatic trust. So if I use the same parallel to compare DeFi to its competitor, say Charles Schwab, it's 300 milliseconds of turnaround time and zero order flow fees. Let's look at how DeFi has progressed from a user and adoption perspective through the lens of infrastructure.
So, 2020 is the “DeFi summer,” where blocks lasted 12 seconds, roughly 12 transactions per second. Push oracles at that time were updated every 24 hours or 50 basis points. So you had to consider the worst case scenario: either the oracle's price update time or the block time. You can't do a lot of calculations with limited space. And so we got DeFi 1.
0 apps that were pleasantly retro and cool, had great tradeoffs to make them work. But they weren’t the kind of things that could outperform Robinhood, outperform Robinhood, like first- generation AMMs or lending protocols; these were truly experiments for those who wanted to delve into decentralization. A year later, Solana came out and really changed the operating system: the block time was reduced from 12 seconds to 300 milliseconds. This gave us much more expressiveness when writing smart contract applications. We first saw derivatives trading platforms online.
We also saw some warrant books online. Behind the scenes, you were getting Pyth prices you were getting Pyth prices that were updated on every Solana block to use synchronized block updates with oracle updates in 400 milliseconds, which is pretty fast. And then, in 2023, I think the rest of the world realized that instead of convincing people that productivity is unimportant, decentralization and productivity should be improved. So, we've seen EVM achieve the same performance, around two seconds, say, in superchains or some L2 solutions, although some of them actually run a bit faster. So we saw some sort of Solana-style speed everywhere.
The Pith network is designed to strengthen the DeFi infrastructure so that dApps can compete with regular applications regular applications , and it does so in three different directions, which we compared in the first metrics. So, it's trust, speed, and cost. Currently, the Pith network has three product lines, which we will now look at. The first of these is probably the most famous. These are price channels.
The problem of oracles for blockchains is well known. You have a blockchain that has programmatic trust built into it. There is proof-of-work or proof-of-stake, as well as validators who support this stake with a high level of trust. But once you add something external to the state, you have to get it from a trusted source, otherwise you completely destroy that trust, right? And this is, in essence, the oracle problem.
The first generation of solutions to this problem focused on building trust but didn't consider speed, which is what I was showing you at the beginning. To solve this oracle problem quickly, you need to understand the mechanics of how market data works. The market data industry for financial products is very developed. This is a business with revenue of about seven billion dollars a year. There are about six major exchanges that make most of this money.
This data is most valuable at time zero because it can be retrieved for free in about 15 minutes. So they depreciate very quickly. The closer to time zero you transfer this data to the blockchain, the more valuable it is. The business valuable it is. The business model of large exchanges selling their data is incompatible with blockchains.
They sell them through subscriptions. Each user has their own subscription. It is limited by the number of users they have . Blockchain breaks this model because it is a shared state, meaning there can only be one subscription. Moreover, these large institutions, large exchanges, have no incentives to develop blockchains.
They are not interested in this . Therefore, the first problem of getting this data into the blockchain is where to find the sources. The Pith Network solved this innovatively by reaching out to big data owners who had never participated in the data market before, namely large trading firms. So if you think about how financial markets work in the traditional sense, there's the average user or trader, and behind the scenes there are very large trading firms: Jane Street, Susquehanna, DRW, Jump Trading. These firms typically take on a huge share of the volume that is executed across various asset classes.
They are only on the side of the buyers of market data. They actually have to buy this market data back, and this is something that has been bothering them for a long time . So the number one innovation for the network was to get data from them to get data from them . They never monetize them. This will be the first time they will be able to monetize their data, and they have a huge amount of it.
Put this together and you can create a very comprehensive view of all markets with a baseline cost of effectively zero because they don't compare it to the alternative cost of selling the market data elsewhere. And that's how Pyth creates this price. There are over a hundred institutions, large institutions including all the major crypto exchanges like Binance, and all the way down to smaller ones. There are also traditional exchanges, American stock exchanges such as MIAX, IEX, MEMX and CBOE Global Markets, which own the VIX index. And they provide real-time data to a place called Pythnet, which is a public blockchain.
They are aggregated with full transparency, and with Pythnet they become available for extraction on 80 different blockchains. And this is where the economy starts to work. As I mentioned, the economics of placing data on a single blockchain are inefficient because then it becomes available to everyone. Nothing is returned to the data owners. By hosting everything on a single application chain called Pythnet, and each request being paid for as it is used, this model is created .
So every time an application needs a price in Ethereum, Optimism, Arbitrum, or Solana, they get that price directly from Pythnet and pay a small fee. This economy can be returned to the data owners. And there have been three or two iterations of Pyth price bands up until now . The first was launched on Solana. And the second is migration to Python.
Solana was obviously only available for Solana, and now, as I said, it's 80 different blockchains. There are currently about 400 app users using Pyth price feeds. You can see what we get out of this through these three categories. So, first of all categories. So, first of all , trust, as I mentioned, is based on a public blockchain public blockchain .
All this can be checked. In fact, there are 110 different data sources that also confirm the chain. The cost is very low because, as I said, we get this data from new sources. We don't take them from those who are trying to get billions of dollars for their value. Plus , it works fast.
You know, 400 milliseconds—that's pretty close to the 300 milliseconds that we talked about in the Schwab case. The second line of products for the network products for the network is on-chain random number generation. It is called Entropy It is called Entropy . It has a very similar set of problems as price bands. This is an oracle problem.
So, if you run a gambling game, how can anyone know it wasn't rigged? How do I know that there is trust in this ? Entropy solves this using a “commit-reveal” mechanism . This is an old cryptographic technology that has been around for a long time. Two parties actually capture secrets, hash them, then those hashes are revealed, and the final result is hashed on the blockchain.
This allows programs to prove to their users that everything was done by an algorithm that cannot be faked. It's not very expensive here, about 5 seconds per update for Entropy, and it's pretty fast. The response time is about 400 milliseconds about 400 milliseconds . OK. And the third product for the network is called Express Relay.
Express Relay focuses on the MEV problem. Maximum Extractable Value or Miner Extractable Value is a term to refer to the total revenue that flows from the protocol to validators or miners. To take the best example— best example— liquidation in the lending protocol— validators typically either reorder blocks in their favor or create an auction where they pay for the right to be the first to insert, keeping the profit for themselves. Express Relay works differently. This is a new primitive that connects protocols directly to a verified group of searchers.
He allows them to run the auction themselves and keep most of the proceeds for themselves. This is the first thing. Thus, many of these costs do not go to validators, but remain in the application protocol itself and can be distributed among users. So, this leads us to lower fees, similar to what we've seen in traditional finance traditional finance . Um and...
um, yes, here we are . Another element to think about in terms of speed is time to market. If you can tap into this established group of searchers, you know, one of the big problems with the new protocol is eliminations. I need someone to keep an eye on them so that when the liquidation is triggered, someone can execute it. The new protocols actually combat this quite strongly.
An established network of searchers allows you to tap into this from day one and have liquidators immediately looking for opportunities. So time to market is how I understand speed in this context. And of course, regarding costs, they are passed on to users, not lost. So there are a large number of applications that already use Pyth products today. As I mentioned, there are over 410 of them on 80 different blockchains.
In fact, 60% of derivatives protocols on any blockchain use Pyth, according to DeFi Lama. That's about 100 apps. In fact, it enables more derivatives trading on blockchains than any blockchains than any other single protocol, blockchain, or oracle. And this number has increased about 10 times in the last year times in the last year . You can see the growth and market share that on-chain that on-chain derivatives are currently gaining on Binance.
And that's actually all. That's the main goal of the network, to main goal of the network, to make sure that there's an opportunity to continually improve the infrastructure so that these applications compete not just for decentralization but also for portability, for those things that make someone think, "Isn't there a better system? better system? " It makes everything better " It makes everything better , with all of these elements. And the Pyth network will not stop improving until we surpass traditional finance in every other way.
Steeply. Thank you very much.
Automatic transcript — names and jargon may be misspelled.