Hashing It Out
Hashing It Out

Episode 73 · 2 years ago

Hashing It Out #74 - SKALE Labs- Stan Kladko

ABOUT THIS EPISODE

SKALE Labs is seeking to change the world through decentralized cloud solutions. Using EVM-powered blockchains as their primary compute and storage mechanism, they seek to capture the market long-dominated by AWS, Azure, and other centralized solutions. We have Stan Kladko, CTO of SKALE on to discuss how this is supposed to work and explain their vision of a blockchain-driven future.

Links:

Donate to Hashing It Out

Discuss

Entering work. Welcome to hashing it out, a podcast where we talked to the tech innovators behind blocked in infrastructure and decentralized networks. We dive into the weeds to get at why and how people build this technology the problems they face along the way. Come listen and learn from the best in the business so you can join their ranks. Welcome to hashing it out. I'm your host, Colin Cuche. Corey can't make it today. He's in Paris doing some really cool work with status. Today we have Stan clad, co constant, Constantin slip, claud coach Gees. Yeah, you're right, maybe it's a lot easier if I say stand. Huh. Yeah, so from skill APPs. So we'll have them on. We're having them on to talk about what he's doing. And Yeah, maybe you can give us quick intro about about yourself, how you got into the space and what you guys are doing. is at skill. Yeah, so I actually know I'm going to cheaper called yours. So I've been doing physics quite a long time. So I did wake PhD music set Maxicollon Institute in Germany and then I moved to the states. I was it was almos stationally in New Mexico. Then moved to California to Stanford University. I was doing physics. They're my boss. There was a Nobel Prize winner and physics. So by background and my physicist but then when I moved to Silicon Valley it was right exactly this, like startup time. So, as many of my friends, I moved from research to start up. So I joined the startup which was founded by Stanford by Professor Bonnier, a Crypto Guy, and that was a startup which was doing a cryptographic hardware box and the start up at the up raising like fifty billion dollars and we're the upsetting also for fifty million dollars. So didn't really make much money, but I have many friends coming from this time and that's how I started in Crypto. Then after that I started the Crypto lab which was affiliated with American government, National Institute of Stand US technology, and I lab was doing all kinds of crypto certifications, evaluations, nutuation, tasting for many, many products and in silicon willing. So I've been doing this for several years, the scryptal lab, and then actually started my own crypto start up which was in Wi fi space. So we raised money from an investors and that's why, when I got back to u cream, because I needed to hire engineering team, and so hired really in Ukraine and ended up selling the start up, this Spifi start up, to a telecom company in London, in UK. So this fi five footware still operational and actually works in many cities, in New York City, in San Francisco, in London, in many other places around the world. And after this startup I started another startup. We got money for food, basically a partform which was going to simplify big data in Ai, and we got money from to venture capital firms, seeing a ventures and flag a capital build the actual prototype. But when you started setting this big data software, what we found out it's really hard for time to start up to compete with like googles and a double uses and microsofts of this world. So we have a start up trying to sell little put your trying to pilot it. Then you see that people are saying like okay, it's a great prototype, but I am going to be like there in five years...

...or ten years. So having all of these huge companies nowadays, it is pretty hard to compete in sauce. I can. It's personally. Yeah, that's it. Yeah, experience as well. It's tough. Yeah, so I understand. It's really pretty muchly got back to our vestors and those pretty much one way was just to shut this down, like to actually hire, and we had several offers to actually hire the startup. But what are investors said? No, press somesing else, in particular a Mike Vehicles never likes sighting things down. He always if you started some money remaining in the start up, he wants to do something else. And then our other venture capital from Rick Tomson, he really said that you guys have to to take a look at crypto and it was exactly like the moment when Cryptos like wearing skyroll teaching, and I literally like, and Rick said, you know, you can actually create this earcat wanted talk and on his zeroom and stuff, and you can use big data. What they are. And I jump in airplane and got too difficunt three and I didn't even have a ticket. I had a busy guy who actually gave me his brain. So I got into this dificunt three and like the moment I got in there, it wasn't come cool, I realize that that's the place I want to be, even if Theyd Sis and upload giving me any money. It's really like a like a totally new thing and it's really earlier time. So I saw like metallic or jumping on a like a unique worm. People you are little like cheering and celebrating and it was really like, I think like, if you know, there was this summert of laugh in San Francisco from of ower. It was like very interesting feeling of people like that found something interesting. So it's really excited and get back to to San Francisco. And then they started to understand what we can do. Obviously you know, our the ground was the book is spect data, so we could make things faster. And then and then I we were looking for a CEO for for partner and through Sidney Adventures, found track where it out. Also were in on Crypto, and then that were it up like really funding caistry together and looking really well. And Jack has its separate the ground. He's the ground. This informal, but then he was the informath. He actually stumbled on this idea of digital money, using digital money to insentiing inside corporations, distribute resources. So he also got to the crypto space and we got together with Jack and then and then, since it was big data and mean, it was obvious that the idea was scaling the Zodiom and slow. So we decided to look how we can split up as idiom using big data. And then initial idea was to do and exchange it really fast. Isn't feel as a change, and I remember we went to some investors at Harvard and mit and people were saying, yeah, you guys have to do the change because it's very simple. Is a Delta small, so just do a faster change and you can charge like some money every time a transaction goes. So really kind of sort about the change for a while. But then they said, okay, if we can do a change while, why don't we actually you know, great infrastructure for someone else actually can do change. And that's how we insided it to the scale and we got back to the mists and raised the some mimot money. It come itselves in doors. See grown from from the current misters plot it seeing me at winterers and then it's actually how the scale started and I'll feel you would likely I later how it preceded but basically that was what their ins catch it. So could you describe some of the product in more detail than of what you're you're trying to do with this and where who you think who actually be the target user for this? So actually I think it becomes of the product changed a little bit. You know, really it's like what do we call these...

...exactly? I don't want to be interrupt you, but, like, I completely agree. That's a very stute observation that these are more like protocols. Is More like systems, are more like package reasonable kind of like modules that you can actually put into other larger applications or integrated into larger enterprise systems. They have their own value propositions. Up to you to make that value proposition. So it's one of the challenges is actually figure out how you get money from these sort of things. So I'm actually curiously what do you what are you thinking? So well, the start that I think this topic is a seemple idea. Is just is you do is slow, but the great essentially a parallel mythroom where you can upload some of these you transactions and make them quist her. But then, pretty much like every time you go to ABC, your a about the ideal this product market heat and you have to be fine. You'll market usually be so simplistically that you know the market is just from sexual us but basically paying small amounts of money for coffee or whatever, for parking. But now we see that in two thousand money the timing is a little bit different. You have just the small transactions. You have defied sinceralized finance, essentially people crowd sourcing loans and loading money to each other. You also have games and then you have decentralized social networks, initially centralized cloud like centralized Google. So what we see timing wise, is that I don't really think that small payments like just paying for coffee is really important two thousand twenty maybe it will be important two thousand twenty three, but at this moment, you know, many people actually quite successful the check credit cards of case, and you have to really get, you know, critical mess to pay for coffee in Crypto. But what we see a lot is definitely what we see is centralized finance. Basically, I think it can be in general explained by the fact that if you just hold bitcoin, you know basically the ideas that you buy Bitcoin, you you wait for twenty years and then after twenty, twenty years, that coin is going to be super expensive, right, and then you wait but to make but that's only targets a small subset of people, because many people are not actually ready to wait twenty years on that. Again, if you can provide a product that provides people, you know, five percent annual rate, risk free or risk or six percent, and you'll read actually that there were many, many people will jump put that so to do to move to mainstream. I think it's important that we're kind of moving from this idea hold of Hoddle, which is more for like could it doctors and ent users, but the idea that every average drove, you know, can take some money, put it in sound defy application and then regular and then, you know, even despact to seven six percent interest. So that's why I think it's really defies picking up so much and I think they're going to see many, many, many applications of defy this year and it's I think one of the critical applications is going to be people to people lending where, you know, if I if I have a developer, I'm engineer on Kipthup. I'm a super cool developer, but that maybe living in Ukraine right and the create no one can get get any role. So but maybe it since I have my Kid Cup stars and I'm super cool developer, some of the people on Internet can lend me money so I can buy a car or can buy, you know, mortgage whatever. So this defines picking up a lot. And so what scale targets? What we realize is that we need to move that from simple payments to define. But then if you take defined, the next thing is games, the next thing is centralized social networks, sinceralized Google, and then the idea becomes actually centralized clouds. So what what we kind of involved to is that this time if there's just speeding up transactions, but then they realize what we need to do is the essentially centralized...

...competitor to a w us, where every a Wus server becomes a block chain. So it's a very simple mental picture to take a WS or to take Internet and then we take every single Abu server like it way, you know, facebook or whatever, and you two are lit into a block chain and then there's book chains need to interact and then it's all teach compatible. The reason why it's Allas to be teach compatible? Is it your compatible? Seldity compatible, because this only community which exists, that that's the only community which has development tools, programming langulation, which study shows at once a programming language appears, it's very hard to kill it. So Se lenity may be look perfect, but you know, people will involve and it's kind of re evolution. That the way, very similar to Drava scrept. You know dravascript. It's really look today as it was twenty years ago, but it's still called drava script and that's a kind of an evolutionary future that we wet on is that salidity is going to evolve into this language that will be used to create centralized cloud applications. And our proposite scale is it by the end of the year? We want to say that anything you can do on a Wus you will be able to do on scale. So sure, so. So when you see anything you'll be able to do an able us, you'll be able to do on scale. It ws as a whole lot here. I mean it does so all right. How are you going to integrate with stream processing? How are you going to intergrate with s three? Like there's projects like storagees. Try to build decarchialized file storages as three compatible but it's not getting the traction that you think you'd get. How are you going to actually serve content? Because blockchains are terrible, terrible forms of mechanisms for storing data. They're extremely in it hype, like extremely inefficient, hyper redundant databasis. They are not intended for EBS styles storage systems. So what is what do you mean when you say anything you could do on aws you could do in scale? Because that's a that's a very, very, very bold vision that has a lot of steps cut out. This is a great question that should have been several insts. Well, answer is that when we got because are like this here, this aren't guys like inside the scale we built want to be like all other guys. What all other guys do? They just use the UNFORCA, the POKM. Then they thought the sort of slate sell the pookem and not no weed. Basically, so the Reasul the net work is pretty much, you know, the aim in this sense we wanted to be beginning, in the sense when the same that you know, we may we take us twenty, twenty years right and so if you look at this, this lie, look at you, a Musque Musk is actually a great like marketer, because what he did when he created Tesla, he started with Tefla roadster, and then he had test models, Tesla model three, and took him, like I said, literally like almost like twenty years now to get to this point of teslam woods three. But the great marketing part was he was giving people a vision and then they didn't actually who ready to wait? She give him more one essentially right, and he was able to pull this, which was really interesting that he went through the stage. So definitely want to say that we want to move to the future where this is since Alis Google and slies, facebook and the utilized Internet. We're giving people a little trow vision, but the want to to understand what the stages are for this vision, even if it's take us like many years, and I think it will take us when years. There's another observation, which is you look at Google, for instance, one thousand Nineteen Ninety Eight, it was pretty much clear for people in one thousand nine hundred and ninety eight that Google will become a...

...three and dollar company because they had some money right then. Yeah, the great yeah, I was a gut fitting so. But the reason why I took twenty years because when Google was released, people alread different. A new generation needed to raise and for this new generation Google is going to it is very different than for people who a living in nineteen nineteen. So this, this delay is due to this generational change. Very much the same, I think, for Crypto. You know the generation that will be there twenty years from now, from pro for them Crypto is going to be very different and it's unavoidable. As saying this, twenty years wait. You know the wonderful it doesn't mean that it's will on grow, will grow, but it's a lot grow in a day of Bitcoin. Will Lot be like they're like three years from now it's going to compete with dollar. So that's kind of on it's but then definitely the defferent marketing peach refinantly by the end of the year or we're not going to do everything stable years can do in the but reacts. You know, targeting, but you know some basic things like MVP or major things that people run on the Wus, you will be able to do on scale. But but what your things? We mean this year. Definitely okay, so any type of like web application, mobile application as well as I'll talk about file storage in a moment. We're not targeting video or singing this year, but for file storage actually we actually think that will be able to compete to this drug bucks even this year, because you know when you have when you have blockain, and we've actually extended selidity to be able to store files right, which is store files just in a regular limus breaching system. So when it goes it goes into ad but then goes for medium to Limus. So the storage would be using actually a regular Linux storage. And then our difficult chain is sixteen no team. So when you appload the file, you will actually to get get cooked, obviously to all of the sixteen notes, because block chains everything is a cup here. So then on suppulled the file to scale, it's actually available for Adum and then it's Cutty to all of the sixteen notes on this chain, this book chain. But then, if you think about this, if you go to store a file on it ws, you needs to have content distribution network anyway. So if you have an image, your website has an image, this image usually gets cut it to Australia and America and Europe. Multiple servers? Yes, actually the sixteen old thing. It's what we see what a wot pen and we actually hope that our storage, initially will won't be much more expensive than a the WS and but them are. Again, it's a long term revision. I'm not saying that it's going to happen now, but we really don't want people to run our things on a wers, although we have the skin they will do it. This is very expensive, right. I mean I said, they will rettle soon realize that if they lose on ABS, they will lose lots of money because there's much, much super way to run a stuff. And so I hope what will happen when we start they put will startle the WS. Then I hope they will atlely start moving from a WS to just regular hosting, just for the work questing. And then the long term vision is that there are many things around the world like so cooled dark showers, like enterprises by Surabrs, this which both you know, suppusting companies by sorrowers. This both really the long term. Movies, that is to get cheaper and cheaper. And the good thing about will chain with him from, for instance, the mining. You know, mynors there interested in creating cheaper and cheaper way to wine and they've created less of a novation. It's a long term reviews the whole that this also created a novation for computing, power storage, the fact that they will make it grit. That keeps money, but the cool they can many months. So so you set a lot there and...

I it's a lot to pick apart. So there's there's some immediate things that jump out at me from what you've said. First off, sixteen non network. That's fine. Probably using a class either classical consus, sagorithm, or using some permissionless system with private networks, you know, click whatever for a you know, if you're doing a theorium, it's probably something like that. Or maybe proof authority. Doesn't matter. You're still come to consensus on the data. So that means the data is replicated across all the nodes. That means that if I have one, if I one user in the system which is dumping a tremendous amount of data into the blockchain, that block that is replicated sixteen times. And if we expect users to run their at nodes or put the scale system in you know, let's say not a WS with their own hosting system. That's still massive replication of data. Companies like storage, you've been attacking this problem of data rep of basically data storage using a consensus mechanism like a blockchain. have typically use things like checkens. So basically essentially equivalent to sticking an IPFs address into a smart contract and then that would that IPFs address is self managed and replicated through some replication scheme. What are you doing? Are you actually putting the information in the blockchain or you putting a reference to the information on blockchain and then have a separate system for replication which people can either trust or not trust, or not need to trust because it's content addressed? That's a great question. Action is so so if you look like we'd like to that to like radical Andy different things. One thing, you can put just data into into evm. It's just super explosive. You can literally store of things on the main metal. As a man, that look course, to let dollars through what they are like line line of fixed so that's really super explosive. Not Things like I'd best which Fu it'd like to be a super cheap private that definite type book with cheap and this and disentual life. But I'DFS is not competible with smart contracts. You can't just really cold the poll lead the chromatdfs into Selady to you by just using slidity, cold vo our stuffing the needle. So our storage is probably not willing to be as cheap as a IPFs. So if someone needs a really, really, really cheap storage, probably they will use apfference where the middle? So if you want to create a website on on the blockchain, will probably use us, because in our case the great thing is that the storage goes through Evm, which means there the smart contract can process this. Let me give you an example. We're also working on adding ai into our vum. I'm going to talk about this later, but a simple example is, let's say the social network that runs on the blockchain. So you want to run a social lot work network on the block chain, you come to scale, you pay to scale for a year and you get easert. Medium chain or small chain or large chain? Different chains have different storage. You can keep capt traditional capabilities because the split servers. They actually the large chain you get sixteen fole servers. But for the tiny chain you can get. You get actually sixteen pieces of servers and which piece Su just one, one, one hundred, twenty eight dator campaigner. So so you get this Weik chain. But anyway, so let's say you get a chain from scials, your own chain, your owner, and you can uploid any smart contracts into it that do whatever you want. So let's say your smart contracts do social network. See your social netwal brands in the video and solidity, essentially, and then a user comes and the user extra that P psel, I mean, which what you want to do is really fulfill the Thal the purse thing is skilled for you don't want some of things, your images, the social through,...

...to go through fust of the press, smart woe chart. It's what would that use this ai? Two more, if less. You find the image and say, okay, this image doesn't paid, inessing offensive. And then and then being, which is stored Goo Goos intredum with you a virtual machine and the store inside is you do a virtual machine in Linux. Okay, point this is this is a whole lot. I need to send in reject right there, because you will go on. I get the feeling this needs to be more of a conversation and I think you have some said some really cool things there, but I also have some skepticism about them. So running AI and a smart contract, that is a your I'm assuming you're not running you're not using this typical gas operations. You're not counting, you know, counting pennies over. How many operations is particular smart contract does? But one thing we need to focus on is that for these smart contracts to work, every single note in the network must run that same ai and they must come to the same conclusion on that Ai. So so that is a really bold claim to say you're running AI and smart contracts. Is it's a very difficult thing to not only get right, pick have have it. Everybody comes to get sensus on what true looks like in these these kind of systems which sometimes, by the way, ai uses random permutation which we can't leverage in a smart contract. So what do you in when you say smart contracts are using AI? Yeah, this is great, which I told you that actually before. prevoctive cannot into Crypto, which you would do the AI, so using which you create this ai simplification platform which was based on Google tensor pool. And so we actually did before that they are pround the top top models from Google and understood many things about Ai. So I actually this is as instant on the STA got two pieces of way learning and prediction learning. When you you can need to have like silience of photos or images whatever, and then you train your neural network and it's a huge, huge reputationally intense saying, and then you'll probably won't be able to do it anyway because they'll have the images or the sounds Google has. So the way things go on all days is that Google, of face will actually trained neural networks and all other people, like the small guys, they just use this new networks. You know, sometimes they tweak them money. So so the prediction step, when neural network is is trained and then the prediction is this like an apple or cat or whatever. The file did. Very important point to understand is that the prediction step, or with typical neural nettop of a top neural net will coming from Google facebook, is not so much computationally more intense than a crypto operation already in evm. So essentially the AM nowas like eight different cryptogorisms and we just extend the V also with a different algorisms and then run a proximbody to take the same computational power as, for instance, you know whatever signaty, digital signature, totally good. So you're basically using free trained APIS and treating stick in the man right better, totally. You mentioned that as more important thing, which is crucial. How to make this a too ministic, because lots of the lots of new let works use randomness. And also the floating point computation may be different for different computers. So you can file one computer floating point looks one way and another computer works another way. Doesn't really matter usually for people, but you know, for Blockchain, one big difference it is a catastrophe. So you need to be this thing to be fully to menistic, and that's what we are. But we now were taken Google tenser flow. Make it it, making it too menistic, and there are two things that we need to do. First of all, random number generators. We need to make sure they all start with the same seed whatever, so they actually too menistic. And then also the floating point. You can do to toomunistic floating point, for instance by competing Google tends are flow into web assembly, because we assembly is is too many. So we're looking on this. It's it's a pretty complex thing. What is doable, and that's why we're not releasing in this this spot for the first version of our network,...

...because it has some doable, what tedious things, but they it's will under way and sometimes this year we're going to release this. I think she's going to be pretty much who will tells our whole prediction. Part that, as you mentioned, very important, that it's going to be too menistic. Okay, but what does? Okay, so that's that makes perfect sense. I obviously the the smart contracts themselves don't have any flooding point operations. So you'll have some your own custom like floating put libraries and solidity, Yada, Yada, add it. That's also no problem. What is what are people actually using this system? How are people actually going to interact with these AI? In other words, when people, when I think ai on the blockchain, typically the model that people are bit pitching, which is different than your model, I'll fully fully, you know, say, is that the actual smart contracts are going to be able to do some of the learning part the the you know, but you're actually using premade models. So what how do you see people using these these neural networks that are literally built into the already trained in in this system? How do you see people actually using this in the real world and how does that impact your defy, I'm not defy, your a cloud vision on. So, like, let's say I have an image and I want to upload it. I know you mentioned that before. How would that actually benefit me as a user to have these AI little robots in smart contracts that I can actually like, submit my data to and then they would give me an answer from it's a great question, and then it actually they would be. When this talk that this entirely. I think was literally thinking cream. We just wanted to an AI too. It's really excited about then realize we can. The unions don't know how to do it and I'm really have a market because you wait. So we sat it. Okay, let's just add the simplest thing, the prediction. We added this and then we're looking for like killer application for this and we found the social network thing. One of the first killer applications, the social networking, basically analyzing images or not hold if you want to do a decentralized facebook, making sure that images on the wed so we came up with this and that was already exciting because, if you think about this, if you don't have this feature, it's very hard for you to create a centralized social network, because facebook constantly utilizes images on up load. So we can cross to the social network killer up. But then then are some other things which are all, like research, very interesting. For instance, how can one do stable currency based on some neural network? Can can you create because now you can have this neural network which is pretty powerful? Can you can? Can you create a stable currency which would stabilize itself better because it uses Ali? Also interesting things like putting face recognition, orthentication, the things, and also a kind of Fun Financial Ali. But I think at this point we actually another thing that other people are trying with our EIDEA is copyright protection. Or let's say you have a network which pays for music, but you want to pay for original music and not for someone stealing someone else's music. So probus applications for prediction. We hope that most of the stuff is going to be done by our developers during hotcatoons, because many people find different things. Most of the things, I think at the moment we are around images image classification, things like that. But that's how we see the doom the social networks. They is kind of the keyler thing. Then they defy all kind of way algorithms. I'm going to put this well, it's a little trowth in the future. But also have the privacy. Will be when I air, the private compute computation. So because will have private computation, they will be some possibilities to do...

...trading or EI Algorithms for trading the block chain. That's that's another application when we see I really interesting. Yeah, but that actually has a lot of problems associated with it. They're just completely separate, like front running. What is it called them? Yeah, I can flash front fresh flash relations just racked everything like a decentralized exchange, like they're flashboys, two point of papers, pretty, pretty hard to contend with. So do you think your system can actually contend with that in any way or what do you what are your thoughts on on some of the like transaction reordering and just the consensus sensibility in general. And they defy like de sercialized exchanges. Like do you think that your system can actually do something around that or no? Yeah, this really interesting question and I can tell the story is so our our consensus using is using a cryptographic algorithm which is called called threshold signatures. It's really interesting. Our consensus are consensuss are you gonna signature this since you were today? So yeah, okay, cool, yeah, that's I remember him like actually researching this. Remember the time when I was taken a course from him, when she published the paper and at the moment quick the paper was totally people said, you know, who would be hell would ever use the thing? No one was using it literally for twenty years old. It's really my sematics. It's fun to say that this bill assignatures. You know, the paper uses research from a French guy and they while Andrea, while while transformation, they die. Actually in there was a war, like a simple the first world war. The draw money and the guy wouded the draft and they they caught him and they put him into jail and he had literally nothing to do in the jail. So he was like researching in group Seri is some totally you know things that had totally no practical importance for anyone, and he came up with this wild clearing and then no one would use it like four years and then when they are like, I think, like almost like eighty years after use this while playing for the ls and then bls wasn't used. Like what one years it's to see how my sematics, you know, it's a very different thing. Text Year arst. But anyways. So, so, yes, we're using lots of the US in our consensus, because our consensuss probably my semmatically probo consensus, where the first ever synchron us, and probably pubo consensus from the beginning to the end. But this is easy. Finish our synchronous. It's increase increnous. Wow, so ho or a synchronous are partially a secrets. Like what do we talk about? We are fully a free, fullest incromus consensus. So you can, you can, but meaning that you can start our system for a week and then when you connect the network cables it will start looking as nothing. There's no time on. Yeah. So so what is the signatures? Is another thing we just called fearful encryption. And and then you just want to use it somewhere and go like trying to find a use for it and we realize at some point the stumbled on the fact that fiftful encryption can actually put anly solved the front funding problem. So actually now have a method which put an is sold from funding and we reprove the release it by the end of the year. The idea is that it's very interesting is that when you submit the transaction to what's felt called encryption, means is that if I encrypt to a set of sixteen guys, level of them need to get together to decrypt. So so when I said with the transaction is it's encrypted by me and the guys. The transaction goes into blockchain encrypted. So by the time it's committed to the blockchain the consensus happen. So the block is created because become sensus inside the board. The transaction side the block is encrypted and then once the transaction, who was in the...

...blockchain is already in the blockchain. This all of sixteen guys, eleven, come together and decrypted and then go and dip. It's pushed to reveal. But the point is that at the moment you decrypted transaction, it's already in the blood. So it's impossible to front run it literally because it's already in the blockchain. So we hope that when we release this front running protection it's really the thing. It just looks magically in the sense that people just can code and it don't even need to care. And then this, this, and Justin crepit transaction when they send it, and it's, I would say, so almost all front running problems. So what's the reason time? So what actually decryption? Maybe didn't didn't get can a part. Okay. Well, what point is the world able to actually you do the concerts on this if it's an encrypted transaction of blog? Yeah, but I give you an example. Let's see. Let's say. Let's say you know a transaction is a is a page in a book, right, and a book is is a book, right. So people send their pages and then create a book out of it. The point is it? When you send your page to the book, to the to the nom right, the page is on encrypteds. People can from front it because they see what's on your pain right. So let's say in stay and offer saving just plain text page. You send them encrypted page. So the mode gets this encrypted page. Another have a woman, but the book is composed out of encrypted pages. Now, but the book is tot already on the blotching now, Idum, and when the deal time comes for the ideal to process the transactions. The dum takes this book and decrypts it, one page after another, and processes it. And every decryption step involves elel because they have sixteen note bloching. Every decryption requires elevel of this guys to come. So we're separating. Essentially been using the fact that the separating in time, the time where the block is created, the time when the block is added to the blockchain and the time when the dum processes that. This book. You can see idim support. Right. I want to tell people about idiom and field up. It's like it's a Bot, that it's transaction. It's like pokem one. You know, it's transactions. So creational block chain and running on Idm a different things. You can create like a blotchet of our thousand block and dam can run late on it. It's a Dal. The processing of smart contracts is a separate most moments of time. So so that that's that allows us to increase and the blook chain and decrypt at the very last woman. Essentially we just doing the decryption be heal. So it's the visual question. So every virtual machine is the ability to grypt every transaction isn't that equivalent to, like, does that matter? Like? So that's that's the problem having right here. If you add an encrypted thing to the block, but block, but every node, even if they could collude, have to conclude in some way to form the key to actually post to decrypt, doesn't that mean that it's equivalent to just being able to decrypt when they added to the block? I can't they just do that right off the bat? Like, what's actually protecting them from decrypting it until later? That's what I'm not understanding right. What protects them is eleven of they have supposed. Most of them was supposed to be good. So most of the modes won't decrete, won't participate in decryption procedure until they sure that this, this cornfection, is already in the book. So we had a lot of sixteen guys. All, the last sixteen guys, eleven a good guys, I wot it, and the eleven guys actually, you know, good people. So we get this page is right, oh the scrimble and CRYP, we create boots, sold on them right, and we...

...don't add the good, good, good guys. Would are how agreement is when we create books out of this preaches to boot decree because because they have right. So we decree create books and then one one inch shoal us is the same book, and then it talked to each other. We know it's in the book chain. Now they say, okay, guys, we all now know that it's already in the book. Right, yeah, you can now change the pages at the book. It's fixed. Now you let me love the book to each other through the even of books class again and says, guys, no, you don't know that this page is. Won't get the book. They can won't be from running anymore. Now, please decrepe participe point in the decryption ceremony and you level of US each want data, a piece of decryption, and this pieces are grooved together and get the COMON decryption to the decrease signature soliciate. In that case you would have ten out of sixteen, is as required. If only five of them are our business, I wouldn't come. Is Malicious, then then they can't decrypt. The problem I have those that it does incentivize people to do this off chain, which means that the evidence trill for this kind of front running wouldn't even exist and there would be no way to really detect whether or not they're actually doing this kind of behavior. Am I misunderstanding that? So in this the system, they cannot be front running and unless you live on, unless those are basic security assumptions, I understand. Okay, coolite. That's why. No, I get it. Okay, cool. So your security assumptions is that some you trust up to some end number of participants and what whatever. Yeah, that makes perfect sciment. Okay. And if that's your security, because assumption in in your public with it, that's that's fine. I mean, if people are without, then that's that's okay. You can even get their search certified authorities involved here, so they have some base level of Kyc a aml going on. So they're actually like there's but there's there's there's a lot of oversight, maybe regulation. Of course, this does kind of undermine the very foundational tenets that defy seems to have. In my I mean it seems like it. Am I wrong in that, or is that is that is like, because now you're injecting very centralized authorities into the actual, like security of your network, because you can't just touch, you know, ten random strangers to not be malicious. You can't because, like poloniacs, could easily own ten of those nodes and you wouldn't know it because they've given one to their uncle, once of their their stock broker, one to there, you know, to their lawyer, and they're all working with their software that has something in front of it that actually does this sort of like analysis ahead of time. Am I missunderstanding this? Yeah, it's a great question and I actually cover this until the more engineer picture that we see in this industry, because now we are the brokers. I can wait now. And we howstant you booked old fans of money daters, you know, people who are sitting out this funny dator shorts, and I think that we have about the keep you any dat us in our program so we know who the people are. WOULD THAT BE? But it eating and this entire business, the money dation business, is like skyloking this sallium dollars. Stay it now, and this year's going to be crazy because everyone is doing the staking. What we see is in this staking environment. We see that well, at least in two thousand twenty. What's going to happen is that when the that looks like ours and some other Netflix come up and change to most of the validators are going to be easier a large, medium size, you know. I think most of the capital is going to be going to the validators and it also see lots of geographic separation because naturally you know if you want a particular country, if you want a friends, you probably good a validate in France because of taxes, personal relationship, and it seems to be very similar like, you know, regular business. If you are really the seller or contractor, you...

...can know your customers and your customers are your kind of treasure. So everyone has influenced. So we see this influence by separation of the world on many gets, through geographic basis, but also on industry basis, whatever, and the less important. The point is very important is that I want to say that almost every network is going to see have the same validators in the sense that you change too many dators and scale, many dators are going to be the same people, because if I'm an INS in friends, you know, and I validate echge to, I would probably want to call sale to my customers and say, Guy, guys, you know, they also want to validate. Is To you to scale. So naturally all of the SIVALIDATES. For guys, are gonna be validating a set of brand, brand networks. So for us at scale is going to be like is a tie or or survived in a sense, that is everyone realidate scale, you know, or no one what it did scale. So it's going to be like this, like we did to sell our you know, set a Microsoft, you know, Drewni pers score. If you are getting to this least, then everyone does you and then if you start up. So so that's what they're going to see. And so the point I would like to make, very interesting point, is that what we see is that almost every network the probably have a similar level of security, because the same validator holidates it. Teach is going to be validate scale and also not. A very interesting point is that the total stake will better and the total reputation little matter, because if you walidate scale and you walidated teach and let's say you get bribed and people know that you screwed scale right, people will undelegate, take money, you will remove money for for a tige to. So your reputation as a valid dator is not going to be like scale would chage to bad is going to be cross cross boundaries of the network. That's why what's going to happen as at fast that scills ship and the official, because this guy's a pretty idate. US also validate each to and you hope to kind of take you back on their reputation for atiage to blets grow us to you. Right. So there's really interesting than Nadix are going on and nothing I would to see is that the NY be through. Some twenty is going to be very, very deferent people compared to to winners, hope, tossals and seventy. It's so, what are you from? Pipelet Gotcha. So okay. So let's let's transition a little bit into what more of the cloud competing architecture kind of looks like. So you're storing in blockchain. You're actually like story literally in the blockchain. So that means that all the all the content is part of the part of the block Hash. I assume. Is that correct? So, like you can some every block is of a variable length size and when you'd commit a block to the blockchain, the image is at the let's say the image. You're sticking images and text posts for decentialized facebook. Let's just use that example. I make a post. I post a meme and a posts of Haha, this is funny and it says it's from me and it which I have some address and has an image and then it has the text. It's all that would be stored at the blockchain. How does this scale to you know, use your company? How does this scale when we do stuff like this at large like I'm not sure I see you quite the scaling solution at this point, meaning that by using the blockchain as a means of storage, redundancy is is handled through the blockchain, but it's ridiculously run. It redundant. Do you do anything like chunk up the data or partitioner responsibility of storage amongst nodes, or is it fully, fully stored across all nodes in the network? Is there any way to prune this data or reasonably reas it so that you don't have to perpetually keep it in a decentralized storage format, because that would be kind of growing quite quick as you gain adoption, as I think. I yeah, I don't expeney how it actually works. So you...

...have the block chain and let's say you upload an image, right, so the image, it initially goes into the block, but then block the transfer transaction bok is got included good transactions, right. So then IDM is is it? What's a popular one? And this pokemon processes one transaction of the time. It's IT and changes its state. So when ivm actually eats a transaction in the block, this image goes from the block to the dum storage, through the EVM to the limix separating system. So now it's stored not in the block but also in the Vum. But the book is not process it becomes a historic block. So the image now is both in the dum and also in the historic boor. But if we can prove through the historic books, after a week or so, a book at checkpoint, so we can. We can the old blocks one they've been has been processed by EVM. We don't have to keep them, we can just prove them. So this sexual explosure of storage, you mean blow bags, doesn't actually happen because we've actually proven old blocks and the just, you know, essentially change the genasis point of the block chain because after a week we don't really need the blocks. Don't actually store the old blocks or songs this. That's actually solved the problem. You mean, and then actually kind of like essentially sixteen limits servers, each of them storing in which so it's not so much more where it's almost though, throwing on a single server. But, as I mentioned, if you would actually do a good website, if you will use the CD on anyway, in CD of you cut the image the multiple throvers anyway. So you could say, okay, just it's a CD and in fact whetually whole that our chain can be used as a controm distribution network, because essentially you've kind of the sixteen node, you know, chain and one of them know what's hopefully in Australia, long in America. Hopefully they don't hold around in the same a queer videos, but if it really seen, is actually really cool. I like. I like that idea, like the idea of using a blockchain is or some something central system is as a replacement for existing CDNS, although I don't know if there's enough demand for it. It is very lightweight solution to that kind of problem. My question is is, like you keep saying sixteen nodes and I'm kind of wondering, is that a limitation of first off, what could sets us? Algorithm are you using on your chains, because you're saying using a fully asynchronous network, which blockchain itself has synchrony assumptions worked into it. So I'm making it literally inherently synchronous. You say, though, however, using a fully asynchronous system. That makes me, that piques my interest, like I really want to know what that looks like. For's all, like, what algorithm are you using to be fully asynchronous and how do you get around the blockchain synchrony assumption where each block depends on the previous block in order to build actual chain? It's a great question. So so it's a I'll take a bit longer to explain because it's the very kind of fundamental thing. So so when you start that I'll only started to do this big data thing and blockchain. The thought that we could do it really easily initially and we just try to use regular big data algorithms. But what we getting is that anything we tried like just simply we pult find a security whole and then we found read some other people's wide papers and we found security wholes in them. We realized that there was not there was like the papers, like every paper, use their own language. You couldn't understand it. And we when we try to go deeper and deeper, we found more and more security hold. So so we decided to step back and understand how can we make it secure, because at some point we got so depressed we we had said, okay, we need to understand how, for ourselves proved that is a secure what the mathematics of it? And we got back and what we did. We said, okay, because it's because it's a very simple problem. Let's be a complex problem. Let's approach this as...

...a car. You know, car is a very complex system. You know has many things, but if you are able to logically split the car into carburetor, engine and wheels and brakes, if you if others that logical split, then every single part is much less complex in engine and then you can and then becomes modular because you can complay cards. You can say, okay, B M has better engine, but you know ourcidents has better breaks whatever. So we decided, because it's complex, the thing they blood chain consensus, we need to build a logical model at a high level and then if we would build this coherent logical model, we can prove that if every card is secure and then if they at a higher level interact properly if we can prove the entire system is secure. So we build this logical model and we have now the seventh step, or it step, logical model where we know that if every step is secure, then been done. A thing is going to be secure now for every step. Then we went to to basically click computer science, to academia and found the best algorithm that we could that was probably secure for this particular part. And then, and then they build our stuffs. What's that? First of all, is modular and then it's probably secure from the beginning to the end. Not because the proved this, but because we have this model where we can prove that if every part is secure, we it's does thing is secure and every part is secure. The academics prove. Now what this start doing this with your list. Interesting thing is that we can now take any other consensus and we put into this model and when the map with it, we can easily a point security problems because we can look at the consensus and say, wow, here the car, this car is a huge engine in but this car doesn't care any breaks. And they say, okay, it's insecure. So so what we're not saying is it not claiming our consensuss the best? We're claiming that our consensus is depressedest, probably secure consensus. They really first a consensuss but they are insecure. We don't know if any secure consensus expressive and ours now the one of the critical parts of this consensus is a mathematical algorithm which is called synchronous binary consensus, also called the desert in generals problem. It's a problem where general steels to walls counter attact the city from the left of from the right. They need to agree. So for this bizent ten generals problem, which took the most, the best currently available thing from academia. It's a consensus which has been developed by is by a team at University of Nouns and friends. So we're using this consensus and it this consensus usually using bliss signatures. So basically the idea is that we don't really develop any know this. A startup doesn't have time to do academic research. We just stole pretty much all of the things from a canare but now we know about many other guys. We know that they can huge security calls. We don't actually announce this because we don't want to make don't want to make too many you know, people would like us but to consensus, of which I really put don't been saying about this. I just want to say to that I really bad hat graph and a lunge this. They have tremendous security holds, but as if that's okay, we don't have to get it out them. Okay, all right, so that's fantastic. I don't know. I don't know what the actual name of the Algorithm using. We have a pot post. I can I can see you about the block posts, because it's hard deserted, goes into detail. Okay, Gotcha. Okay, cool, no problem. So what are you most excited about in the space? What are you looking forward to seeing coming out of the decentralized space, and what do you see being the future of where all of this is going? Yeah, that's a great question. I...

...think. You know, I think I did some pome twitter in that I think people who started Crypto, they were many of them there libertarians, I think many of them, and I like Libertarians and I sympathize to libertarians, but I'm kind of part of the second generation' Open Source Guide and my view is that we that crypto is something that can be used to provide a business model open source. So so I'm excited about how open source, Crypto and share the common is going to get together and the same in the future. What we'll be able to do will be able to take some of the hugest monsters of today, Google and Facebook, and to turn them and run them on on the centralized cloud and turn them into much better, better applications. I can get give an example about Google, several examples. One example about Google is that, you know, we're doing this Bifi start up and other stuff. Was Great, it was a free thing. We were doing this Wi fi management like public wi fi. We would the free priplet people who could count like this one access points and like run for free and other stuff. Was the only thing in the world that was actually doing this. But Google would still put us on the short page. So you would go to Google and they would like, you know, enter like free Wi fi management software, and it was just create like two pages of crap and the short page grow us and people actually wanted us and then they had to pay five dollars to click to get to the first page. So Google has this incredible problem, a conflict where they should have been putting us on the first page because people wanted us, but they didn't, putting us on the short page because they wanted to make money. And but what's it? Realized this, realized that Google is one of the hugest breaks on the renovation of this world and it makes not life, not that life, so much more difficult. So I've been trying myself to avoid Google at all means. So what what I do? Usually, if I need to search for something, I just stop and I say, okay, well, where do I find a way? It's best answer the vote to Cora, to go to stay overflow, to a vote to, you know, Wikipedia, the world, to give hub, and I just buy pass Google and then go to a particular side and and the search and usually it's much better experience. So, you know, I can putetor of Google. It may be as simple as as a discentralize think, which has sends you to the search box of a different application, of a different female side. So I truly believe that I discentralize book, book is is possible now because Google is just becoming so bad, you know, the change their Ui, they do. Wouldn't possible to cheat to differentiate as from real results. So Google even work. I've been looking to Hubert drivers and they are like here in Ukraine, for instance, day like they hate your work because Hubert takes thirty percent of their money. And you know if it's a driver. You pay for games, you pray for your car. I come like appreciation, and then you know if they if you, but take thirty percent of your real reading for you as much, much more than certy percent in troubles of the of the money you take. So this, until at Luberg, could be just an algorism. Opens Furs algorism, the truns on blow chain and people get hundred percent of their money. There's no real and so I really who. I think that this is something which is going to start happening this year and it's going to change thing tremendously. Also believe in centralized financing law, Lads. So I think this year is probably going to be laws and centralized fund as Ruful Games, because games are the foot thing with any technology. You know just when apple released muckinthorth, the first thing your games, because gamers are like like to be excited. So Games defy. But then I think the centralized search engines and centralized share the common the applications and I think it will provide really much better for...

ECOMOMY. I don't think it will be revolution, but they will be involution. Many of the the dicentralized communities like going to be more commonly effective they have corporations of today. So I'm very hopeful that. But you will start seeing this and what I want to see is that it's changing. Protodyne cryptocurrency is actually not so important as the actual distance really decentralized application. People. Currency is the foundation for decentralized applications. It's like kind of been goal. Is just it sounds of the decentralized cloud will use to create this new decentralized Europe. Yeah, I totally agree and I'm a definitely aligned with that vision. I do think that, you know, your vision of cloud in blotching is obtainable in some form. I think a lot of these services don't need to be centralized or can be done in a different way that will be a little more democratic and a little more open and transparent to in in the cases we want openness and transparency, which is in all the cases. So I'm really excited to see projects like yours pop up so we could start testing these theories and working on these ideas. So is there any question? I should ask that that I didn't and how can people read? So yeah, I can. I think the prison I would like you to ask is hope. Can people help us and sell the whole people can help us. The funny thing is that now this so much, that so many ways to make money on the things, because detroize networks. You know, this year's crazy. You know we have money from WEC's. You know, other people have much more money than us. So everyone is giving out money for people doing good things and we and it's lots of money. I think we'll have like prices of up to tenzero dollars for hacking us or finding security holes. We're about to run this incitiize testment where people can tinker and find any problems. The problems can be smull problems. You can find a problems like the humentation. You can find a box. Doesn't help to be. It's not only about like being looks super duper like cryptography. You can find an UI problem, which you know that's right, run while crestions, anything like that and and actually it's very little. Few people owing about this. So if you want to make money just to you, if you have might twitter, you can just subscribe to the skill. Twitter will be announcing the program so the problem will go through Gitcoin, which is the website that people can make money by helping this. Another thing which is called Hacker one very nice. Think Hackers Unite and somehow it try to file to fix things. But also even the sudes programs will being for just developers, open source developers, to help our project. It's output that is fully open source. Anything everything is on GITHUB. You get several reasons why you want to contribute. One reason you want to just put some fun stuff on your raisume. So if you just, you know, help us create documentation or fix fix a small bug, you can put the stuff on your Rasium and people will hire you better. Or if you want to contribute more, we can get care act our tokens. So if you become a little more invested with our project, we have some budget, you know, to help outside developers. So it really, really interested in people actually contributing to us. And also, if you are idea developer, if you really want to do a startup, you know, just real life facebooks. Until I stuber, we do have some connections this pcs. We can refer to vcs and you know, if you are excited about decentralized clout and really want to do something, will really be happy to help you. What we see is that too many people are just going to Hakatons and just, you know, not so seriously doing things. We really love people if they want to do something important. You know they really want to do a startup. You want to do a startup and but your life for some period of your time, you know, on this thing, will be...

...really happy to help. Thank you, stay and that was awesome. I really appreciate it. So, okay, everybody, check out scale. Amazing product and working on so. But he's got attle. It's got a really cop solid vision. So yeah, thanks, thank you. Thank you so much.

In-Stream Audio Search

NEW

Search across all episodes within this podcast

Episodes (118)