Another key difference is the infrastructure, we looked at things like Ethereum, and Tenorment, and others, and very promising, but we didn't feel it was designed for scale. And so with that, we had to redesign the way we think about how does scale look like in this space. And so going and drawing back on my knowledge of system design, I also, before this, I worked on high-performance computing and server computing, places like Argon, Sandia, and Los Alamos as well. And those systems are designed to have massive palisim to support the largest applications of the world around protein folding, and they're very hard scientific challenges. And you think about the way design a system, it should be supporting high levels of parallelism, it should be highly concurrent, it should make sure that programming models integrate very carefully to the system so it can support those key innovations. And at the end of the day, it looks like a decentralized database to most people. It is just very fast, highly performant, very low latency, and so you can build on top of it the kind of internet applications you see today, except for their based on Web3. They are owned by the users. They are a permissionless database that everyone can interact and agree on settlement on. And so those two things, I think, are the key kind of pieces that we started off in 2018 with. I think another key differentiator is upgradability. And the ability for the system to be able to change very rapidly over time through technology advances is super important because this space is moving very, very quickly. And so being able to design that system where we said, look, we have a mechanism for being able to change your consensus protocol as a part of a governance update. That's voted for on chain. The ability for us to evolve the way council looked at over time and support account abstraction natively. Or the way that we are able to kind of scale out in many different ways, from even from leveraging more resources on a single machine, to also going across multiple machines over time as a way to kind of scale up and scale down the system when you think about cloud infrastructure. Also very, very key. And also new features like on-chain randomness, which is very interesting and maybe one of the best important use cases of blockchain today. So it's a combination of starting off from how do we support billions of users from a systems perspective and from a language, a smart contracts development perspective to how do we also iterate really quickly with technology over time. And the last piece I would say is providing that user experience that looks exactly like the internet. So we can support those five billion internet users today, while there's only maybe tens of millions of kind of web3 users. And growing that audience into the larger internet audience is very key for us. So those are the kind of key differentiators, I think, from the optos perspective, from the philosophy standpoint. From a practicality standpoint, what we've also shown is that optos has the highest performance author in terms of scalability. We run, we do things differently in terms of transparency. We, there's a lot of noise in the space, okay, we're the fastest, we're the most scalable, so on and so forth. These experiments are often not repeatable. They're not end to end. And so what we've done is taken a very different approach, like from our history, we look at things like there are lots of database benchmarks out there. And there's a way to have measure these databases against each other through TPC, for example, which is a benchmark in suite. We don't have that in blockchain today. So we started to build out those processes, we started to put out those infrastructures and frameworks for us to say like, here's our framework for evaluation. Here's our experimental setup. Here's the machine configurations. Anyone can run those experiments, have verifiable and repeatable results. And what we were to demonstrate was we could get more than two billion transactions per second, sorry, per day, in a, in a, in a mean it like environment that 70 plus known operators witnessed and verified, which is more than 14 times what Visa can do. And that's instant settlement, unlike when Visa were kind of settles and the kind of settles again over time. So I think, those are kind of the things that we think really separate us from, from other infrastructures out there.

Super interesting. Okay. So a big piece, like you mentioned, is the programming language move, which was developed in meta first, then upgradability, this very scalable infrastructure, this web to like UX. And, you know, being able to, to test this transparently with an, a system that others can also try. That's, that's a really interesting piece as well. Is that available? Yeah, absolutely.

Yeah. So if they, you know, we published a media post about it, we put it on Twitter, people are welcome to go ahead and take a look at the open source code that we provided. We run the same experiments, use the same hardware on GCP. And they should get the same results, if not better today, because the system's actually improved over

time. So digging a bit into the infrastructure piece, like this scale level infrastructure, how, how do you actually achieve these, was it two billion transactions a day? Yeah. So

that's more than 25,000 transactions per second. Yeah. Okay.

Because, you know, there's this very well known, um, blockchain, trilema, where, you know, you have to make trade offs in security, decentralization, and scalability. So where's the trade off in, in app tools? Like, is all this scalability coming at the expense of decentralization or of security?

It's a great question. I actually put out a very long post on Twitter about the blockchain trilema. And I don't think of it as so much of a trilema as more of like a, a stool in some sense. Like these are the legs in which every piece of infrastructure in the, in the web through space needs to build on top of. We need to be more secure. We need to have better performance. We also need to be, you know, you know, more decentralized over time. And so it isn't necessarily so much trilfs today as pushing these directions forward and kind of making those tools stronger. So, so let me give you an example of that. For decentralization, um, it's, it's, you know, there's many different ways of thinking about decentralization. The way we really focus on it is going to be around things like Nakamoto coefficient, which I'm not sure if you're familiar with, but that's the number of nodes in which kind of you lose security of the network. If you, you, you kind of, you lose, you, those nodes become malicious over time. We also have another, um, metric that's not so well known that we care about, which is called the liveness coefficient, which is number of, it's similar to Nakamoto coefficient, but the number of nodes in which it takes to kind of make sure transactions are being appended to the blockchain. And so a lot of people think about, oh, you know, you have 10,000 nodes or, you know, 100,000 nodes, whatever kind of your network that makes it, you know, more decentralized. But in fact, many of these protocols have a very low, they're not going to go coefficient. And those extra nodes don't really help with security. They just increase cost, they increase coordination, make it, make it very difficult. And they are, they're kind of dust in the system. Uh, maybe like, you know, this, this comes to kind of a vanity metric system degree. Um, you know, and, uh, it's, it's something that we, again, we try to focus on what's real. Uh, and, and I think that's, that's kind of where we think about from a node standpoint decentralization, like these two metrics are the key ones out there. But there's other aspects of decentralization that are also really important, such as the way, how do you do development? Uh, Aptos has been open source, uh, code, even since its earlier days as DM and back in 2018, under the Apache two license, anyone can contribute, anyone can take the code and fork and do what they want to do. And in fact, we saw that, uh, which was really cool. Like back in, uh, back in the day, um, some developers did take our code, fork it and kind of launching network, um, which is, which is really neat. And that's the kind of way we think about the ethos of web three being very helpful for others, uh, and supporting an industry, not just a specific project over time. And so it is that the sense of open source also community aspects. We have an AIP process or a process improvement proposal process where anyone can kind of contribute different, um, ideas, uh, and also implementations that could be a sort of time, uh, and then kind of also, like, how do you think about governance? Uh, so governance, uh, in Aptos is, is a bit unique. It uses on chain voting. Any token holder, uh, staking token holder can vote on different proposals. And the other thing that's really interesting about Aptos is that once you make those votes, those changes can be live immediately, uh, through the programming nature of the protocol. So it's a very, very decentralized process around governance. Um, and so it's, it's a government, you know, distribution is a very, very big topic. And it's not just about nodes. It is not just about, you know, your processes. It's really encompassing like your whole philosophy around how you're involving others, um, and how are you involving a community in the way you develop software and make sure changes go out with, um, with community, uh, kind of approval. Um, the other aspects of security and scalability, I'd say, the security is also very related to decentralization, right? If you have very few nodes in the network, uh, are very low, no-comical efficient, they can be compromised. And we've actually seen nodes that have been compromised because of that aspect of security. Um, and, you know, for this, you know, things like move, just move the bar forward. There's no trade off with respect to move, uh, except for maybe there's a smaller ecosystem that's not aware of it, but that will change over time as people kind of get familiar with like what, what's possible as opposed to having to live with what exists today. Um, and, and also, security aspect is around processes. Uh, so how do you think about this way you develop software? Do you have one code of your two code of yours? Uh, do you, uh, have a lot of testing that goes into dev nets, test nets, um, main nets also, you know, does it test on every code commit is a nightly, all these kinds of things around security. Uh, they aren't trade offs. I mean, they're trust in terms of costs, perhaps, but not trade offs in terms of the other aspects. And so that's where, you know, we see that, you know, things like that, us that have that enterprise grade security that you would expect, uh, uh, have a, have a differentiating, uh, play in the market as well. The last aspect around performance, this is just around mostly technology to us. How do you build amazingly scalable software that can leverage all the resources of your machine and beyond that to providing a really, really great, um, highly scalable system with a great user experience.

Okay. So, but specifically to the app tools chain, like how, how is it built? Like how is it a proof of stake chain? Um, how many, how many notes does it have? I know you say, I mean, that's not the only thing that's relevant, but, but still like, yeah, that's

just like, well, though, so I'm talking as a proof of stake based network. It uses a business, you've got taller and protocol. Um, and it is actually really cool. There's a paper called shoal, which describes our latest and greatest in consensus. And I definitely encourage you viewers to read it. It is, we've iterate our consensus protocol, I think four or five times already since 2018. And we kind of bring new improvements across the way where there's latency and or scalability through that. So that's, that's something that we were really excited about. In terms of the other aspects of node count, I think we're about 126 nodes today. I think the Macomikomayo is 18, which is actually quite good. I think that's the highest Macomikomayo efficient of any node network that's launched since our launch time. So we're really happy about that. And we'll see, I think we'll see more decentralization in that aspect coming over time. But that's very, a very pretty, pretty impressive network that's been live just about two years from now.

Nice. Can anyone become a node? Yes, anyone can

run a node. There are some minimum requirements in terms of staking, but any, any, there's, it's a permission of the system. Yeah. Okay.

There are other layer one chains are criticized because it takes, you know, a lot of resources to run a node. Is that something that the, your, you know, participants on the app just changed, would struggle with? That

is a good question. So I think this is one area where there is a bit of a tradeoff, right? So if you want to have a high performance and low latency network, especially low latency, I think that's where you have to have a great internet connection, for example, you have to have a scalable value infrastructure, like one that has decent amount compute power. And this is kind of with respect to the needs of the network. And so this is something that, you know, in order to get those great results, we did have to get, you know, pretty reasonably mid-tier powerful machines to get the 2 billion transactions in a single day and 25,000 transactions per second. So yeah, a bit of a tradeoff there. But we think a tradeoff worthwhile to make for a customer experience being really good. Cool. Yeah,

it makes sense.