#454 Scaling the Blockchain: Srini Parthasarathy on Fixing Gas Fees and Decentralization

[00:00:00] 
 

Mehmet: Hello and welcome back to a new opposite of the CO show with MeMed today. I'm very pleased joining me from the US Srini. The, the way I love to do it is I keep it to my guests to introduce themselves. So tell us more about you, your background and what you're currently [00:01:00] up to. Just like a, uh, I would say teaser for the audience.

Mehmet: We're gonna talk about blockchain and of course, maybe. Intersection with ai. So the floor is yours. 

Srinivasan: Absolutely. Thank you, Mehmet. Uh, it's absolutely my pleasure to be here. Um, so my name is Srini. Um, I, uh, I'm the CTO at Shardeum. And Shardeum is a, um, layer one EVM based blockchain. Uh, we'll talk about that a little bit.

Srinivasan: Uh, before I came to Sham, um, I was the CTO for a company called Silks. And at Silks we were building a blockchain based game. Um, specifically around race horses, so you could buy. Uh, assets, uh, in the metaverse, you could buy stables, you could buy land, you could buy your race Horses that actually had a representation, they were real race horses, uh, born in the US and that raced in real races.

Srinivasan: And if you owned an NFT every time the horse won something, you would, you know, you would win something. And it was important for it to [00:02:00] be on the blockchain because, you know, it dealt with ownership, it dealt with providence. So all those things were important to us. So, um. It's, and before that I was with DraftKings.

Srinivasan: I helped build out their, uh, blockchain infrastructure for another most popular games called Rainmakers. I hired out the whole team, we built the infrastructure. Everything was on blockchain. Uh, it is pretty cool. I also did a small stint at AWS, um, I was trying to interest, um, AWS in sort of building infrastructure for public blockchains, but there are.

Srinivasan: Sort of more focused on private blockchain. So, uh, you know, uh, so that's sort of, uh, but prior to that, um, my first brush with blockchain was when I was trying to, uh, sell my own startup and, uh. We were encouraged to find a way, incentivize dentists and patients to use our product, but not have to [00:03:00] pay for it, but, you know, accept something else.

Srinivasan: And that's how I discovered, uh, uh, this whole idea of using crypto and, and blockchain. This is fascinating journey for me, but I'm very happy to be here and I'm, uh, sham. We are, we are doing some amazing things. We are solving really hard problems, not just in, you know. Distributed technology, but also in blockchain doing something very unique and not too many companies doing that.

Srinivasan: And, you know, um, that's what excites me. The technology is so fantastic. 

Mehmet: Great. And, uh, thank you again Trini for being here with me today. Now for people who might not be very much familiar, uh, with Sheridan, like, let, I like to ask it this way. So what is the motive behind Sheridan? You know, what's the main problem that you are aiming to solve there?

Mehmet: So if you can take us like maybe in a journey, how things are done today and how you are doing it in a different, uh, way also. 

Srinivasan: Yeah, uh, [00:04:00] actually that's a great question. So, um, so the, one of the things that interested me in Shady was, uh, when I was at Silks we were selling resources NFTs for resources for about $1,200.

Srinivasan: But, um, users are paying nearly $180 to $200 in gas fees. And clearly that was, uh, you know, a terrible user experience. And because it was, uh, it was an Ethereum. Every time somebody floated a meme coin or something like that, the gas prices were so. Unpredictable that it had a huge impact on how, uh, you know, how we sold our horses or our, our assets.

Srinivasan: So at that time we were like actively looking at layer two blockchains and a bunch of other, uh, options for us to see how we can reduce gas prices. And so that's how Sham float into my, into my horizon and, um. Actually, let me take a step back and so [00:05:00] let's, this is a useful visualization and you know, please feel free to jump in.

Srinivasan: It's obviously, you know, I'm abstracting a lot of it, so it's, I'm living looking at a 50,000 foot view of, so if you think of a blockchain as, as a spreadsheet, right? It's a public spreadsheet that anybody can read. Um, every row in the spreadsheet is like a block and every cell in that row is like a transaction.

Srinivasan: So it has some constraints so anybody can see it, anybody can view it, but only obviously one person can write into that spreadsheet at any point in time. And this is where all the consensus mechanisms comes in. So every different, every blockchain, uh, has its own mechanism to figure out who gets to write that, that, that row in the spreadsheet.

Srinivasan: And of course, uh, blockchains also have this constraint that once you write a row, you can't update it. You can write a neuro to say that the override, uh, you [00:06:00] know, to say that the, the cells have new values, but you cannot edit the existing role. It's a simple, uh, visualization, um, that makes so, so in, this is how traditional blockchains are.

Srinivasan: So, you know, if you have, so what happens is. Uh, you know, if you have a spreadsheet and you have like a say a billion rows in that spreadsheet, it quickly becomes unmanageable. So an interesting way of solving this problem is to say, you know, I'll take this, uh, this single sheet and break it up into a. Uh, into many smaller sheets.

Srinivasan: So if you look like a, a telephone directory, I'm dating myself. But if you ha, if you look at one of those, you know, the names are organized by, you know, in all the, a's are grouped into one, all the B'S are grouped into another, and so on and so forth. So you can imagine a similar system where you have the spreadsheet, but it has say, a thousand sheets and a thousand sheets.

Srinivasan: Have data distributed in some particular way. Partitioned in [00:07:00] a way so. You don't have to look at the entire, uh, all the records at any point in time if you need to validate something in that. So you just need to look at the one sheet. So this is the concept of, you know, this is like arding. This is what Ardio does.

Srinivasan: So it takes, it takes a huge space in, it takes the so, uh, spreadsheet and sort of breaks it up into smaller sheets, right? So, and so now you have, uh, a different kind of scalability. You are now each. Node, if you will. If it's handing, it's just looking at that one sheet. It has to deal with a smaller subset of data than looking at everything, so that makes it, uh, easier, faster processing it scales better.

Srinivasan: The beauty of Shardeum is that you don't have a fixed number of sheets. That number of sheets can also vary. It's dynamically changing. So if you have a small dataset, you might only have a hundred sheets. If you, as your data increases, that a hundred may become [00:08:00] 200,000, 10,000, and it can scale linearly that way.

Srinivasan: So this is this dynamic state shorting that, that Shardeum has. Um, I, I, so here's another part, right? So sha dium doesn't do consensus at block levels. So what does that mean? So it means that today in traditional blockchains, you can only write a, a row into the spreadsheet at a time. Allows you to say, update a single cell within that row so you don't have to wait for that whole block.

Srinivasan: You can just. Change individual cells on that, that makes it really powerful because it totally, it changes how fast your, your, your transaction settles. The, uh, the consensus is at a transaction level, not at the level of the block. It also eliminates things like, you know. So one of the biggest problems in blockchains is that, uh, you know, because block producers, they can see what blocks are coming afterwards, [00:09:00] so they can reorder transactions, they can reorder the cells in their row because they can see what's coming next, for instance.

Srinivasan: And so it allow, it allows them to do some kind. It gives them a little bit of control that. They probably shouldn't have. But at sha dium it's not possible because there's, there's no concept of a, uh, writing it as a block. It, each individual transaction is consent separately. So all of this makes for a very scalable, very secure, uh.

Srinivasan: Um, and it also makes it the, actually, the more it scales, more decentralized, it gets, so it really solves the, in the blockchain space, and that's what is the most exciting thing I see about Shadi. This is why I was completely drawn to Shadi. The, the team is fantastic. You know, we, this is a really hard problem to solve, even in any distributed system and doing it in the context of decentralization.

Srinivasan: It's a, it's a, it's a very hard problem to solve. Uh, that's a space I allow. 

Mehmet: Great. Yeah, absolutely. [00:10:00] Absolutely. So, and this is, you know, maybe one of the problems that the existing, uh, you know, solution like Ethereum and Salon cannot do, right? So they have, they face, uh, scalability trade offs. Now, I. Let, let me ask you this, you know, linear scalability that you're talking about.

Mehmet: So the more you scale, the more this has become. So what does it mean for, you know, the, the users and what does it mean for the developers on, on top of the, on the top of the blockchain? 

Srinivasan: So what it means for users is that, uh, you know, the, because. So if you go back to our analogy, right, so instead of looking at a single sheet with a billion rows, you are now looking at small sheets and the sheet will get smaller and smaller as we have more of these.

Srinivasan: What that means is that any validate anybody who wants to participate in the node. The what they need to validate, the resource required for that becomes smaller and smaller. So you can, you can think of [00:11:00] how, that's why we call it, you know, decentralization for everyone because the, the cost of. Adding yourself, uh, to participate in a Shardeum node is very small.

Srinivasan: That means that we can include a lot of people to participate in our network, sort of making it more secure and it makes it linearly scalable. So as, so what it means is that, so going back to our analogy, so, so we may start out with say a thousand sheets, let's say, right? And as the transaction volume increases.

Srinivasan: We may dynamically change that to 1,200 sheets, or 1500 sheets, or 2000 sheets. So that's how it scales linearly. And as the number of, and it's because it's dynamic, the, the, the volume of transactions that a node has to handle doesn't increase, but it can actually, it can be the same or it can decrease. So the, the, the transactions per second will still remain the same.

Srinivasan: But the system will be able to handle, uh, a larger [00:12:00] transaction volume, if that makes sense. 

Mehmet: Yes. But for developers from from developer's perspective, so people who want to build application on top. Ah, yes. Any, any benefits? Yeah. 

Srinivasan: Yeah. So, so interestingly, uh, we are completely EVM compatible. So what that means is that any, uh, con any, any smart contracts that you built on, on any other EVM uh, project should just.

Srinivasan: Transparently work on short dm. That's one. The second benefit is all of this complexity that I'm talking of is abstracted behind the CVM, uh, you know, API. So in theory, a developer should not see any change except that their transactions be faster, their settlement will be sooner, the finalization will be sooner.

Srinivasan: Uh, they should only see benefits of this, but all of the magic happens behind the scenes. So they should have, you know, they would not have to, they'll not see any of that. 

Mehmet: Right? So, [00:13:00] uh, now you talked about security, uh, Sunni, but you know, like if you want to like open it more like, uh, from, because the way.

Mehmet: Blockchain. It was like also always introduced that it's like, of course we talk more about privacy, we talk about, uh, you know, kind of anonymity also as well. But when it comes to security, and we've seen like, you know, some incidents, so what does the add-on value that the she them can bring when it comes to, uh, secu securing the, the, the blockchain?

Srinivasan: Yeah. So, um. So in traditional blockchains right, the, the consensus mechanism, figures picks who can write that role into that spreadsheet? Right. Um, so in the case of Sham, because the data is charted, uh, any [00:14:00] specific node that's processing, say one sheet, it only has a, uh, uh, a limited view of the entire system.

Srinivasan: So we use a, uh, our own, uh, consensus mechanism called Proof of Quorum. And it's a fairly sophisticated, uh, mechanism where we a, a group of, uh, um, to make it very simple. So, uh, a, uh, we, we select a preselected set of nodes. Uh, not preselected. A random set of nodes are selected to participate in, in a consensus for handling a certain set of transactions.

Srinivasan: Uh, we call that, you know, that happens in, uh, in a unit of time called a cycle. This is com selected at random, and even within that group. How the, uh, the consensus you need to have within that. For each set of transactions, we [00:15:00] select a smaller group. It's randomly selected a smaller group of, uh, validator nodes that have some stake in that transaction.

Srinivasan: Maybe they are, uh. You know, maybe it's like, let's say I'm transferring, um, some tokens. A is transferring some tokens to B and the shard that owns the, uh, you know, uh, the address of a and the shard that owns address of B would participate in that group, but there could be other groups so that there's some kind of overlap.

Srinivasan: So all of this hap this selection happens at random and it changes in every cycle. So. No validator knows, uh, no node will actually know all of the transactions in the system. And even, uh, even within that, uh, in that group, each validator node knows only a very small piece of the entire state. So it'll be extremely hard, uh, for somebody to, uh, to break in, to, to compromise that system, which is.

Srinivasan: [00:16:00] The selection is randomized and it changes all the time, and it's there are, you need like to breach that. You need like a super majority to, uh, uh, to, to make a transaction actually, uh, be verified. So like two thirds of, of the selected in this selected set of nodes have to agree that this transaction is valid.

Srinivasan: For it to be, uh, to, so that really adds a, um, adds multiple layers of security to the, to the protocol. 

Mehmet: Got you now, like who are you seeing, you know, uh, the most, I would say excited about this? Like, you know, uh, any, any like specific maybe use cases that you, you see people, uh, whether like it's users, developers who adopted the she dom, uh, like maybe any partnerships also that, uh, that you have done that you can share that with us.[00:17:00] 

Srinivasan: Um, so. I think Shaio, like I said, you know, for people, uh, like before I came here, uh, you know, so, you know, people who are developing applications say using NFTs where gas price was an issue, it's, this is a very attractive solution for them. Anybody building defi applications? It's a super, uh, you know, this is, this is very exciting.

Srinivasan: You know, I can scale linearly infinitely without impacting gas prices, keeping them, uh, you know, gas prices are really low. All of that would be very interesting to any Defi developer. Um, so it's like, uh, you know, while our primary target market would be companies and developers building, uh, you know, DAPs and, and, but.

Srinivasan: Even as a pure token transfer protocol, I can see companies that would just wanna do like, you know, um, uh, uh, [00:18:00] money transfer companies, you know, even they should be interested in this because the gas prices are so low and it scales, uh, you know, um, so all of these are. Uh, very val, uh, valid, uh, target users for short.

Mehmet: Got you. I've seen, you know, on the website, like you and I, I believe, like this is something many comments, so I ignore my ignores a little bit. So there's a lot of things like regarding open source and, you know, making things, uh, you know, very transparent how this is important usually. And you know, how, you know, you, you have adopted this transparency, I would say, um, at chart.

Srinivasan: Um. Open source is very important to us. We are an open source project and absolutely, uh, transparency is a key, uh, tenet of that. Uh, we are building everything. Uh, you know, it's like a goldfish in a ball. We are, we are building everything in the open. All our sources is on GitHub. It's all public. Uh.[00:19:00] 

Srinivasan: Anything that, you know, any, everything that we build is available on GitHub for, for people to look at. So that should inspire a lot of confidence in people that we are doing what we say we are doing. And they should, you know, they're, they can look at our code, they can see what we are doing, and we are.

Srinivasan: Uh, hugely supportive of the open source, uh, community. In fact, we have a whole specific department group just focused on, uh, enabling open source, uh, development and, and, and that, and, and building out that community. And we have a lot of people working on that actually. 

Mehmet: Great to hear. Great to hear. Uh, so I can't have nowadays any conversation without asking about the AI.

Mehmet: Um, so, so maybe in general you, can you share your, you know, as a, uh, veteran CTO you've been, you know, for a long time, uh, in the tech. So just your view on the combination [00:20:00] of AI with blockchain and maybe, you know, some of the use cases that you can share how you are adopting, you know, AI at shared. Um. 

Srinivasan: AI is, is an incredibly fast moving field, right?

Srinivasan: I mean, if you think of it, uh, whatever I, uh, looked at a month ago has already changed and it's become obsolete. Uh, so it's like, it's an interesting, um. Intersection between AI and, and blockchain. We use AI quite a bit. You know, all our developers use, uh, some form of AI in, in, in, in our product for writing code, for looking at building unit tests, uh, for, uh, you know, analyzing code.

Srinivasan: We use it quite extensively. Um, you know, the standard tools, GitHub copilot, um. You know, all of these things, I think there's a very interesting intersection between, so I, I can see how [00:21:00] blockchain developers would leverage all of ai, uh, to write better smart contracts, optimize for gas, uh, optimize in other ways, look at security, uh, issues.

Srinivasan: Uh, to see, you know, to look at, you know, what possible attack vectors they could mitigate. I can see how AI can really be useful, uh, for, for people creating DAPs on the blockchain. There's also a very interesting, in interesting intersection between, uh. Like, for instance, Shardeum because our gas prices are, are, are, are, are, are so low and all of AI gen AI development is tokenized.

Srinivasan: Shardeum could be a very, uh, useful, uh, tool to make that process of paying for tokens, uh, you know, cheap and, and, and easy, inexpensive, and, and, and easy to use. So that's one use case right out of the bat. Um, [00:22:00] there's also, uh, you know, um, we also see a lot of synergy between how blockchain does a great job of, of transparency.

Srinivasan: So anything that you, that you, uh, you know, uh.

Srinivasan: Let me, let me, uh, see if I can phrase this correctly. Sure. Um, the intersection between AI and blockchain is very, is very interesting because AI allows you to, to explore new things, uh, you know, look at data in different ways, uh, and, and, and create products that allow you to parse and, and, and, and. And, and, and look at different use of the same data.

Srinivasan: And where blockchain can help is make this process simpler, easier, and, uh, and less expensive. Yeah, you can, uh, I, I can see how blockchain can provide that layer of security transparency, uh, that AI actually needs. Because today when, you know, when gen [00:23:00] AI. Uh, you know, create something. We actually don't know whether it's, it's you today, today's technology.

Srinivasan: You need a human to look at what is produced by Gene AI and see if that's actually, uh, a accurate, and once you have a few iterations, it, you know, it's, it's hard to keep track of that. And I think this is where blockchain can actually play a role because, you know, it's, uh, the, the transparency that blockchain provides can be very useful for ai.

Mehmet: Yeah, absolutely. And I think there's still a lot of, uh, use cases to uncover in the future also as well too. Absolutely. Yeah. And yeah, to your point is, uh, you know, the AI is such a. Fast. You said one month. Believe me, sometime I feel like after one week, you know, some, some people, you know, by the time they see something, something else came out and I said, yeah, guys, you're so late to the game.

Mehmet: Right. This became obsolete. Yeah. Um, [00:24:00] so maybe one thing we, we can cover also, so. Uh, let's, let's be, you know, like realistic. Mm-hmm. This domain, the blockchain domain, uh, regardless what people look at it or something like this, but it, it proved, the technology proved itself right? And this is why now we started to see some faults who.

Mehmet: Maybe work in different, I would say fields as engineers or as executives like yourself as CTOs who are saying now, okay, you know what, like maybe the time is for me to go and work, you know, in a team working on something blockchain. So like any, any like kind of playbook that they need to follow so they can shift from whatever they are in today to the blockchain.

Mehmet: Um, 

Srinivasan: yeah. Um, I think blockchain, people who have, you [00:25:00] know, you need to have a very good understanding of how distributed systems work. Uh, that's the first step to decentralization, and sometimes people get confused between distribution and decentralization. Uh. You know, one of the things that I, uh, that we used to say, uh, in, in Amazon is that the, the fundamental rule in any, uh.

Srinivasan: Distributed system is that it'll break. It's just a matter of when. So having that mindset that something is always gonna fail, and being able, being resilient for failure and thinking in that mode is a, is a very good way of, uh, getting into the mindset. And blockchain just takes it one level deeper, uh, from distributed systems.

Srinivasan: So now you have, uh. Uh, you know, so I, I would say that anybody who wants to work in blockchain, they should. [00:26:00] Be very comfortable working in distributed systems. Uh, cryptography is another area that, you know, if you have to be interested in, in cryptography, uh, to, to, it all depends on at what level of the blockchain.

Srinivasan: You know, I, before I got to short dm, I was always on the other side. I'm, I was the user of blockchain technologies and that requires a very different skillset. You know, you need to be able, you need to like, uh, you know, um. Like, uh, if you're writing smart contracts, right? So these smart contracts are typically, the good ones are immutable.

Srinivasan: So what that means is that you have to write, you have to write code that you cannot change. And then, uh, so you have to be super careful to write code that can either heal itself or can, you know, it can, uh, it, it should essentially run pretty much. You know, by, by itself without any tweaking or any updates.

Srinivasan: And that's really hard to do. So you have to be [00:27:00] a, you have to be a good software engineer and you have to be able to think, uh, to, to be in that space. Uh, to work in that space. Of course, there are ways in which you can, you can theoretically modify smart contracts, but, uh, you know, the, uh, the I idea is still the same, is that it's like not.

Srinivasan: In many of the applications, you don't want somebody writing a defi application that can be modified, uh, later on. Um, so anybody who wants to move into this blockchain space, you know, uh, I would say you, you have to be comfortable with thinking of distributed systems that can fail at any time. And that's, I I, I think, and, and your programming mode has to switch to adapt to that.

Srinivasan: You cannot assume that whatever you write will just work it. What happens if it fail? You have to be able to think that. And yeah, 

Mehmet: and I think same applies from, I mean, leadership skills perspective, like, uh, they need to understand [00:28:00] the technology and they need also to. In addition to have all the traits of whatever it takes to become, you know, a CTO in a, uh, an edge technology.

Mehmet: Uh, still, you know, like the, the conversation really was, you know, amazing. Anything that, you know, I didn't touch on, maybe you want to highlight, uh, you know, as kind of final things as we wrap up and also where people can get in touch and learn more about what you're currently doing. 

Srinivasan: Oh, okay. Yeah, that's a good question.

Srinivasan: Um, so, um, I, not to be shilling for Sham, but that's a sham org is a good place for people to go and learn more about Sha diem. Sham is, so the, the thought I won't leave you with is that. Sham is an incredible, uh, idea. It's, it's it's technology that it's, it's solving some really hard problems in not just distributed computing, but also in, in, in blockchain, uh, [00:29:00] computing.

Srinivasan: And, um, it give, Shahim provides a way for a. Uh, a blockchain to not only be, uh, you know, uh, secure and scalable, but also be decentralized at the same time. So it solves the tri, the blockchain tri. At the same time, keeping the gas costs low and predictable. So I think that makes it a very compelling solution for a lot of people working in defi or in, you know, anything that, that they, that they think a layer on blockchain can be useful for.

Srinivasan: Um, I'm also gonna caveat it by saying that blockchain is not a solution for all problem. You know, if you, if you're just looking for. Just a shopping cart, you know, or a shopping application, maybe a centralized application just stack would be just fine. But in, in cases where decentralization is important, I think shady can play a huge, huge role in that.

Srinivasan: And I think [00:30:00] that's the takeaway 

Mehmet: for Great. Yeah. Great. Thank you for sharing this. And you know, of course I will put the link, uh. Um, in the show notes and in the description, uh, you know, any, any social media where people can also maybe follow you or, uh, get in touch. Yeah. 

Srinivasan: We are also on Discord. Uh, we are on, uh, X Twitter.

Srinivasan: Um, okay. So, yes, absolutely. And e and if they go to the shadi.org, gfa, uh, you know, uh, site, they can see all the social media, uh, coverage that we have. 

Mehmet: Great. Great. I will, I will make sure that I'll put all these things there. So thank you very much for, for your time today and, um, you know, for the audience, as I said, you'll find the links in the show notes and this is how I add usually my episodes.

Mehmet: This is for the audience guys, if you just. Discover this podcast. Thank you for passing by. I hope you enjoyed it. If you did, so please give us aamp up and share it with your friends and colleagues. And if you're one of the people who keeps supporting us, sending us their, um, you know, [00:31:00] encouragement and suggestions, thank you for doing so.

Mehmet: And thank you for making us, you know, trending in the top 200 charts across multiple countries this year and making the CTO show with Mead as one of the top 40 most send business podcasts in Dubai. Thank you very much for tuning again, and we'll meet again very soon. Thank you. Bye-bye. 

Srinivasan: Thank you