Scaling the EVM with BlockDAG on Waterfall Network

Conor Svensson: Cool. Well, Sergii, if you start by introducing yourself, I'd love for everyone to learn a bit more about who you are and what it is you do, and then we can move on to the Waterfall Network.

Sergii Grybniak: Sounds great. Yeah. Thank you. so, yeah, thank you, Conor, for having me here, and thank you for inviting me. So, my name is Sergii Grybniak. I'm a PhD in computer science. I have 18 years of experience in different software development and basically, I was a technical co-founder, team member, of the project, which was later acquired by one of the top ten banks in the United States. It was web two, but not web three, but fintech. I hold multiple positions at IEEE. I am a vice chair of the standard group for working on recurring transactions for EVM. In particular, my blockchain journey started in 2017. All the time I was a builder in Ethereum inspired ecosystems. Basically me and my team were developers of one of the first layer tools for Ethereum based on plasma. And at that time, again, it wasn't that complicated of a development, but it was a big deal back in the day. So our GitHub working prototype of plasma cache was actually first on GitHub at the time. Later on we decided that we probably wanted to do something more sophisticated. And actually I started my PhD research on the topic of direct heterocyclic graphs. And Yeah, that's how Waterfall was created. So Waterfall is actually the result of my PhD. And yeah, I also was involved in advising one of the governments on their, basically web three strategy. So yeah, that's in a nutshell about me.

Conor Svensson: Thank you. And one of the questions I'd really like to ask as well is just about DAGs themselves. What was it that brought you into them and made them so appealing that you wanted to start really studying them and researching them?

Sergii Grybniak: Well, thank you for this question. Basically I  was exploring various topics. So what we all are talking about is mass adoption and there are multiple levels of this, right. There should be usability level, there should be compliance level, some sort of KYC, etcetera. And like digital identity in particular. This was, by the way, one of the topics I was considering. There is an infrastructure level, so it should be focusing on this. I was considering the fact that I again, as I said, all my background was around infrastructure. I started to look into infrastructure direction and directed the cyclic graphs because they were like a very promising direction. They enable parallelism on multiple levels depending on the particular network architecture. I remember these things when there were projects, at least one project which was saying that yeah, the more decentralized it is becoming, the more scalable it is becoming. And again, I was excited about it because many projects are saying, yeah, we kind of solved this, we are ready for mass adoption. If you look at them, basically on architecture for example, let's say start counting the number of validators. you understand that basically they're doing the same trade off in this famous or infamous trilemma, which actually boils off scalability, security and decentralization and which eventually boils down to the dilemma of trade off between scalability and decentralization. So there are technologies which are scalable and some of them to a certain degree decentralized. And basically DAGs are more decentralized. So basically it kind of multiplies itself. But so far it hasn't been developed and at that time for sure it wasn't developed by anyone. Maybe some disappointment about ducks because yeah there was excitement about them. And then something happened. So that was an interesting and exciting topic to research. I actually went through a good portion of at that time existing technologies and also through the concepts which at that time hadn't been realized, materialized to the products. But yeah, that's basically the path. And I can say that I am not disappointed by spending like multiple years researching this topic.

Conor Svensson: And so DAGs, effectively they address that security, what is called I guess the blockchain trilemma there about trading off security and decentralization. Then at what point did you feel like that Waterfall Network needed to be created and here's a really good opportunity for a new type of I would say blockchain but I guess DAG network.

Sergii Grybniak: Well we at some point. So basically my PhD research, we can say it wasn't strictly connected to only the theoretical part and then going into the practise. So basically it was kind of an interconnected process and we were also doing some, I mean a sizable number of independent tests and we essentially were thinking actually to preserve the same architecture, which most of the currently popular technologies are preserving. Again, there are different names for this but also there are, let's say many of them can be called delegated proof of stake, as people are delegating something to some basically limited number of validators and these validators are doing the job for them. So basically they achieve scalability in this environment. Let's say from a software development perspective this is an easier task to solve because the less decentralization you have the faster the entire thing works. Then we actually decided to try to push it forward to try to combine it with a potentially high level of decentralization attempt. Let's do an attempt to potentially make it more scalable, the more decentralized it is becoming. The more scalable it is becoming. And yeah that was the motive. We went through eight testnets over this time. So our first testnet was developed in 2021 and yeah was presented in October and since then we actually went through eight testnets and finally recently released the mainnet. And so far it is getting interesting results and we are very excited. We are still excited. We don't believe that we achieved this statement, the more scalable it is becoming the more decentralized it is becoming, the more scalable it is becoming. So it is not yet accurate to say this about us but we have intermediate results which makes us excited and we still don't know about anything which can potentially prevent us from getting to this point where the statement will be accurate. So yeah, that's what, so this driving.

Conor Svensson: Force of decentralization as you said, it's kind of really underpinned how the network's been designed. And this is demonstrable through the number of nodes on the network and where you intend to get to around the, the actual node counts. Would you, would you like to talk a little bit more about that side of things?

Sergii Grybniak: Yeah, what we have right now, we have like a proof of stake mechanism, very similar to the one Ethereum has. So we have a number of validators and this one or multiple of such validators can be set up on the server. So technically one validator can be one node. And the requirements at this stage of the network is actually two core cpu for gigabytes of ram which is very low. And this node right now is able to handle up to 32 validators at the node. We have powerful servers where these validators are set up you know, significantly grouped. Basically the more devices are able to join. And this is why we are kind of working to push the hardware requirements low and to make it accessible for people with inexpensive devices. Because for example let's say, I mean if somebody bought their laptop like ten or maybe even 15 years ago and they have it stored in their storage room, right, they can take this laptop and they can potentially run a node. Chances are that this laptop will allow them to run a node. High chances are to run a node for the Waterfall Network using their old laptop and using their unused Internet traffic. So we do believe that basically this level of decentralization is important for certain use cases such as IoT devices, the pinnae. Again there are some views on potentially launching the client for mobile devices but there are at least a few pretty complicated tasks to be addressed before that. And in general this kind of narrative that you preserve low hardware requirements, high scalability and high decentralization means that you actually have to solve pretty complicated engineering tasks and you need to address these tasks algorithmically. So it actually involves a lot of research work. I'm the head of research for Waterfall Network. I'm actually Head of the research department. So we do the work one step before the developers. We are developing a priori evidence of the possibility of this to work. Then it does some modeling on this topic. Then we pass it to our development department and they actually provide us with empirical evidence whether it works or it doesn't and then we actually push it to production. So yeah that's how we do it. And yeah currently we are actually doing kind of our next milestone towards separating the network to kind of stronger connecting groups. A concept also of virtually unlimited sharding which essentially can allow us potentially because currently we operate like on the one shard. And again when we are going to the concept of virtually unlimited sharding which by the way was already introduced. I cannot say that it is only our idea. This was presented in the past. But to our best knowledge this virtually unlimited sharding concept hasn't been implemented yet by anyone. we are approaching it differently and we are thinking that considering the fact that our current one chart is able to be both scalable and both decentralized and again these are not strictly defined parameters.

Conor Svensson: Right.

Sergii Grybniak: It's not like for example, you know you have, I don't know, like 1000 validators or 100 validators and you are not decentralized enough. And then you have 10,000 validators, you are decentralized enough. So it's not a strictly defined term. So we can say that we're pushing it. We are pushing the boundaries of this, of what is kind of the average in the industry, right. And then yeah, basically with the ability to replicate it virtually unlimited amount of times it can bring us to the point where first we are the more decentralized we are becoming, the more scalable we are becoming.

Sergii Grybniak: And also it can bring us to the point that we are both scalable and decentralized in every subsystem of this entire system. Because there are examples where sharding was implemented, and there are examples where layer twos are implemented and there are many other examples. So sometimes people, and this is more like a quality of blockchains, right, that still have to do this trade off between scalability and decentralization. The way projects are addressing that is they are basically doing this kind of certain number of replications and they are not scalable and decentralized on one level of them. But considering the fact that they are replicating it some number of times they can say okay, the entire system is decentralized and it is scalable but if you go back to the subsystems they also do this trade off and it possess certain, again I don't want to criticize anyone, actually there are good technology projects and there are different promising ways, but we just approach it differently and we are aiming to achieve this, to solve this, both tasks. And Yeah, that's a complicated task. It may take some time for us to implement all this. We are sure to meet certain challenges on our path and what we are doing, again, something which we are planning to do in future, our vision can change, if we meet certain difficulties on this path. But after these years of working on this we still are not aware of something which can prevent us achieving this. And yeah, basically, again I need to probably highlight that what is virtually unlimited, virtually unlimited means that it can be unlimited in theory, but practically it will still be limited by something, because there is always the limitation of the amount of devices, amount of traffic, amount of users, etcetera. So it's not going to be unlimited. But at least in theory it can potentially become unlimited and then at some point we can say that it will be a very big number. 

Conor Svensson: This is really really helpful to hear that and conceptualize it. But I guess the other question is why are more projects not following this route? Because from what you're saying you're getting the best of both worlds. You're getting the ability to massively scale the network but also you're getting the decentralization benefits as well. And it seems that that's certainly The result that many different networks want to achieve.

Sergii Grybniak: Well this could be actually, there could be different cases and there is no generic answer which will be relevant to everyone. But, for example, there are numerous examples where there's kind of like partial decentralization. And when you are working in the conditions, when you are doing something which has already been done by someone but you are introducing some new level of innovation, it puts you in a more predictable state as a project, right? So you can make more precise road maps, you can basically manage the risks more about delivery of the projects. And Generally there could be different stakeholders involved in this decision making and sometimes the teams can find themselves okay, we have to launch it, let's say in one year and this is it, right? And In this reality when teams are in these conditions they of course are choosing the safer route and then basically they pick this direction and then they deliver this on time or with some reasonable delays. But in general that could be one of the reasons. And I actually  think that this may be the reason for many technologies, many popular technologies, to be that way to address this task in this way. There is also this narrative in the industry. So you know, that decentralization is not that important. Let's actually focus on scalability because it is important. Let's focus on, I don't know, usability levels because this is important. And for people to use this and let's focus on digital identity because it is also important. Who cares about you know whether it is like 100 servers or 1000 servers or like I don't know, whatever or potential number of servers, and how decentralized the consensus is? Again we are currently talking about the number of validators but still there are many of our validators that are set up on the same server. So we are still talking about this kind of theoretical direction. But our consensus essentially designed to be decentralized and this is very decentralized, more decentralized than most of the players on the market. And we have passed the test with achieving consensus with almost 1.6 millions of validators set up on five continents and three islands of Great Britain, Taiwan and Japan. And Basically yes, so far this is the largest test on the decentralization public test ever conducted by anyone and more validators than most of the popular, then all of the popular proof of stake technologies have combined. Sometimes we actually hear the question okay, why do you need this? And the thing is this, people who are saying that yeah, decentralization may not be as important, they actually have a point somewhere because there is a big number of cases where this wide decentralization is nothing that important. And technologies which are saying that okay, let's not focus as much on this, let's make it decently decentralized or use case specific by use case specific metrics and let's focus on other things. But our technology also can work in this environment because if your consensus is designed to work on various high levels of decentralization, it is way easier to work it in the centralized environment. And yeah these use cases are open for us as well. But if these technologies will go to the area of use cases where this high decentralization is important, it is very difficult to make something which is designed to work with. For example let's say 100 to work on the scale of say 100,000 validators because it affects how the consensus should be reached. So basically it is a task where they actually will need to start from the beginning, to actually design it, to be able to work in this environment, how to redesign it, how to close these gaps. So in our case we actually will not need to go that far to the beginning because we initially are building it with the goal to be able to fit both types of use cases where decentralization is not that important and where decentralization is important. And we believe that this is where this is a task to be addressed in the future. And even if somebody is focusing on it right now it may be more important going forward.

Conor Svensson: It's definitely fascinating to hear the R and D angle there in terms of willingness to take on the be that cutting edge technology in order to achieve these greater feats. So we've got just a minute left, but just very quickly, what's coming from the Waterfall Network? And if people want to learn more about the network, what's the best way for them to?

Sergii Grybniak: Well, I think, I think the best way to do this is actually to join our community channels. We frequently post updates there. We do at least, I mean, usually we do. We try not to break this rule to actually do at least bi weekly AMA session where people can ask us questions. And yeah, there is a website, waterfall.network. There is a section docs.waterfall.network. People are able to review the documentation and basically review what is already working. Right. And they can do some tests, they can check the tests we have done and all this. We are expecting many, a lot of important things, happening soon and with some big announcements, but I actually would like to preserve it. It is easier to report on what has been done and what is yet to come. And recently we have conducted the scalability tests on our network. And so far, this is at least one of the most scalable, and I don't want to make a bolder statement, but probably some people will be able to make this statement. It is one of the most scalable technologies, which are currently present on the market. And all of this with, EVM can be one of the most, or maybe even the most scalable technology in the EVM ecosystem. So, yeah, that's where we are and that's where we're going. Thank you.

Conor Svensson: Awesome. Well, Sergii, just wanted to say thank you so much for joining us for the session today. And yeah, it's been really insightful to learn more about what's happening with Waterfall and DAGs. Look forward to our next conversation.

Sergii Grybniak: Well, thank you for having me, and yeah, I wish you to have a great rest of your day and thank you for listening. Thanks.

Conor Svensson: Thanks.