Hashing It Out
Hashing It Out

Episode 30 · 4 years ago

Hashing It Out #30: CyberMiles - Michael Yuan

ABOUT THIS EPISODE

On this episode, we have the extreme pleasure of interviewing Michael Yuan, Chief Scientist at CyberMiles. CyberMiles is building a blockchain platform targeted for e-commerce with those concerns in mind. Using a fork of Ethereum and the Ethereum toolsets, CyberMiles has an upgraded EVM with support for extended and extensible opcodes and a dPoS consensus system for validators to stake into the network. Their research is pushing the limits of what the existing technology can do, and they have some amazing progress in proving alternative systems for developing blockchain. Michael is a very exceptional scientists in this space, and he and Collin had a great conversation, not just about CyberMiles, but the space in general.

Links

https://www.cybermiles.io https://github.com/cybermileshttps://twitter.com/juntao https://medium.com/@michaelyuan_88928

Work. Welcome to hashing it out, a podcast where we talked to the tech innovators behind blockchain 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 hash to get out. As always, it's me, Colin Cuche Hey. I regret to a form you guys at cory will not be joining us for this particular episode. He is tending to a family emergency, but we wishing the best and I'll be continuing without them. Today. Actually, we have a pretty good, good episode. We've been try to get these guys on for a while. It's Michael Run from cyber miles. So, Michael, tell me a little more about cyber miles and what you guys are doing over there, so our audience has some background, so I don't murder it. Sure. Yeah, high quality. So this is Michael Gran problems cyber miles and coal founder and chip scientist. I cybermouse, and so each ark. What do we want to do is to build application specific blockchain and the which application it's the ECOMMERCE application. So we want to build a public block chain that it's optimized for ECOMMERCE applications. That's what I want to do. Got You. So, when you're looking at a blockchain specifically ECOMMERCE, how do you define ECOMMERCE and what is the actual scope of challenges that are specific to that feel that you think you need to address with a unique blockchain? Right? Yeah, so, you know, I think people have been wanting to use blockchain for ECOMMERCE, thinks the beginning days of, you know, bitcoin. You know, we all remember the guy who, you know, who spent, I forgot, like Tenzero bitcoins to get to pieceas you know, something like that. And then one of the earliest use case a block or Bitcoin is so corolled, right, if we if we remember, you know, it's you know, it's a DOOC marketplace that you know, it's very ecommerce. You know, it's people use bitcoin to buy any sell stuff. Yeah, absolutely right. You know, that's so, I think, you know, using blockchain technology or use scrimptal currency for for ECOMMERCE type of applications. So how we define ECOMMERCE? Is a market place that allows people to buy and sell physical goods or digital goods. So anything that can be sold or Bob is ECOMMERCE and that's pretty much the the the the underlying you use case that people have been wanting to do since they why of cryptocurrency. However, if fast forward to today, we have not seen much success on it. You know, we have seen some success from scade, although it's your eve also it's been shut down, and we're seeing marketplace like open bizarre, who use bitcoins payment message, and today I think most people still use crack vests or Ebay. You know, that's very few people actually use open Bisarre. And if you look at, you know, bitcoin pay as payment method, you know, a lot of companies has, you know, regular ecommerce companies, has tried to accept cryptocurrency on the website. Right. You know, there's a I think overstock is one of the early pioneers, and then, you know, there's somethings like even down, you know, or you know very traditional, you know, computer hardware company, and there's payment gateways for their bit pay and there's bunch of companies like that, you know, who enable traditional ECOMMERCE to accept bitcoin. But things like that has met with, I think mixed success, you know, or I would say not much success, because, you know, as we learn from the guys who spent like Tenzero bitcoins by stuff, the primary purpose of Bitcoin is not be used as payment message. You know, because the plot, the Prix structure is a lot right. You know. So if you look at those things, if you look at you know, if if you like to me if you have ever spent bitcoin to buy stuff, you would always remember the day where the Bitcoin to price bubbled after you bought something and you you come to regret. How come I spend so much money to buy a chair or buy a you know, buy a box, you know, something like that. You would never remember the days when bitcoin press drop after you bought something and you know you got it to achieved, right, you know. So from the personal from the consumer psychology or from personal psychology point we you know, there's a you know, the traditional cook the currency is not very good fit flow for for ECOMMERCE, although the demand is definitely...

...there. So from our point of where you use that, you know, we are trying to build the whole infrastructure that address not only this problem but also the problem with the the the winger takes all or the centralization of the ECOMMERCE marketplace. We have tried to build this on in serum and and what's not successful because of the technical limitations that we're going to talk about later. And so that's, you know, basically prompt us to build a new brock team for structure that is based on serum, is compatible with Stum, but with a lot of optimizations that are ECOMMERCE specespect. That's okay. So that's where you paying my interest. You. We we are sole's. I'm going to probably jump ahead a bit on this one, but you say you want to build a compatible etherium blockchain, meaning it's running off the fork of the etherim protocol. I take it correct? Yes, okay. So why did you go with your own blatching and not a layer to kind of blotching like a plasma chaine? Right, so plasma of SOPs the some of the payment issues, but it's, you know, for us, it's there's there. There are two fundamental innovations that that's that we made from these rum so the protocol at the Protocol are we try to stay compatible with this, but there are two things that we have done that significant different. The first is that we want to jump ahead and become delegated proof of steak, or at least proof of steak today, because it's going to take a long time for us, your arm, for Casper to come along and then shots are right. You know, that's you know, it's for them to balance the interest. Became the mining community, if you end the pures community, is going to take, I think, at least a year and a half or maybe even two years, right, but by building a new block chain, of building new infrastructure, we can just jump ahead and without all this baggage, right, you know, just do a vos chain from day one. You know, that's what we want to do, right, you know, that's that's the first thing that consensus lare. We want to do. You know what this room has always wanted to be doing? You know, that's that is to build, you know, ten second confirmation and a fight of finality in each block so that you don't have every year really and much, much faster profitsing speed and all the layer two network stuff can also be built on this network, because this is the room compatible, right, but with much faster main net. And the second thing we really want to do, the the is actually the technique co issues. We want to do is to improve the assuming virtual machine. You know, the the evm is a touring computer version machine. It's can perform any computing task that human, human beings can come up with. Right even that's what touring complete means. However, it is not optimized for any specific task. So there are a lot of things that going to take so long to computer instrum. It's in it's it's not practical very goodly get down those including, you know, editorying over large amount of data, you know, a large array for instance, or common encryption, public key encryption and decryption stuff. And because those people have measured, you know how long is pick to do, you know, sex Keki stuff on its the conclusion has always been going to blow the whole block as limit. So you know, even if I entire block, that's nothing but to excuse that small contract that incrypt, that encrypt using a public key, it would not be enough. So things like that, while seroretically possible, instrum is not going to be practical. So what we however, in ECOMMERCE marketplace, that we want to build something that can address on many number of ECOMMERCE challenges, including buy and sell personal data. The data marketplace is one of the integral use cases that were on the dress and also the ability to eat to to bring another very interesting, takenical challenge, that the that I didn't mention is to bring offline often data on to the blockchain. For instance, you want to bring things like the delivery confirmation or even, you know, suppliching finance, where people have paid their loans or not. You know, things like that, where you want to integrate with the banking system and what things like that. There are lots of encryption and decryption stuff that has to have an on the chain, including, you know, not only have the public key stop, but also proxy encryption and other, you know, more more weird and more interesting stuff, right. And for those type of work that we can pro those type of workload that we can provide simplus plus place extension to the in certain virtual machine. So a lot of those computation flouts drop down to the native operating system level and can come back in I wouldn't say milliseconds, but the inspect second right, instead of, you know, doing it as a virtual machine level, which going to take much, much,...

...much longer. So you know. So it's so basically we have built based on the certain virtual machine. We have built our own virtual machine. That's that's made that extended in certain direction, right, you know, way that a lot more faster and a lot more capable, but also adding a lot of, you know, op codes and keywords for very specific tasks with that's why I call it's you can or. It's a boxing optimized for e commerce. It's a virtual machine. Is Build this way. It's it is not. It is going to be just as still as these are on the most task. But what you commerce arelated task can be a lot faster. You know, some of the performance test that's we have done is a hundred times to us to tenzero times faster. So it could be, you know, for water, some magnitude faster because it's not performing those tasks in the in the VIRTU machine and box, but drops it to the to the native class layer. So that's if you'd ask me. What's other two? You know, engineering and UNIMATIONS that we have down those other two, you know. So we build bbos and then we build a new orcher machine and yeah, so you've covered a lot of things that just strike a particular chorde with me. So scalabilities, just one problem in the theium in my in my mind I do. I do, like you know the efforts that are being made and I think there's some some incredible stuff being done on the conceptus mechanism side to with with, you know, serenity, otherwise known as Shasper, otherwise known as beacon chain etherium on the layer one side, and some amazing efforts with the plasma implementers and these general state channels to hitdle scaling, and I believe that those are all good general solutions that should apply, if they work, to pretty much everything that goes forward on most networks, although I will say that I think that proof mistake might have an issue and that people need to build value before they implement proof mistake. Otherwise it's kind of an issue for the early adopters. Are Too many, too much power. So I'm interested in the in the consensus mechanism side of things that you're doing, but I think because of the radical nature of what you're doing with the virtual machine, I want to talk a little more about that. First direct during virtual machine was built specifically with certain intents in mind. One of them is at the the execution of code is sandboxed and there's a security around the fact that everything runs in the evm. When you're talking about adding, see extensions to your own custom Vm, what how do you prevents? First of how do you make sure that everybody has the same extensions? And second off, how do you protect those those calculations done on the operating system level? Right, so, you know that's well sinside. That will happen. We would have to make tradeoffs, right, you know, that's you know, that's also why I say you serum would probably never take the approach that we talked because it breaks certain philosophies that they hold dear. You know, that's you know. That's so when explain what we did, right, you know. So. So, in the end of the day, the EVM is written by Crypso, CARCO coursers. You know, those are, you know, written by you know, it is a very small vm and it's not very optimized, but it's it's but it's very tight and it's designed to reach to reach consensus. So it's has no ambiguaty. You know it has. So, for instance, you know, there's there's no two ways to come up with us with the same answer or to cold scol or piece of code on different machines would not produce different results. You know, that's characteristics of the VM. The UM also shows its limitations, like I just said. You know, there's a you know from our point of you you want to build a Databa cotplace or or things related to e commerce. It's really hard to do that with with re spart contracts on UVM. So I think even in your community has realized that and we are also active participant in that. Is the next generation of virtual machines. Even don't right, you know, even though is you should on play whoever assembly and where. Assembly is a much more versatile virtual machine. It's it's support. I think you know all the languages are ilbm supports. You know which sin. There's maybe twenty of those languages include beings of US class, right, and the certain flavor means it's separates out the consensus computation and the general computation, so the consensus computations being still being sandbox and then being being put on chain and then the much of the...

...general computations on the note that you that that you connect to and it may not get on chain. So there's a complex design with the Evenson. That's that's that separate out the consensus computer and the general computation. So that's also the kind of direction that we want to go. But we are because we are modifying off vms. So we are we're taking a less strict or less mathematical approach than they do because, given them, to be frank, I think it's still at least two years away. You know, it's because it has casper and shaffer has be down first and then, you know, because the virtual machine is tied with the consensus mechanism. So it's a long way to get there. But for the for the EVM extension that we did, we are just using CPRUS class. So so to answer your question, there's two the two questions that you have just raised. The first is how do you make sure that everyone has the same extension? The the there is a there's a mechanism our blocked in because our blocking pos it's in fact dedicated to prove up state. So there's a boating mechanism for people to propose what's C plus for extension that they want to put on the block chain and then for the supernotes or for the other notes, all for the we co valuate our notes to evaluate those and we have tools for them to evaluate those and to step out randomness and you know, things like that, and then determine if those extensions can be put on chain. And I've to set majority of the value that don't knows voute. Yes, then this extensions been put on chain and put down to every note on the on the network. So you know. So it's like I said this, some people may consider it, let's optimal, because now you have a committed deciding you know which you know we which software extension can be put on chain. But to to compare to what's what has been done today, you know, if you look at this term, there's there's notion of native contracts and precompared contracts and those are built into the software itself. The updating is the optate cycle of those software is, but it's basically some software itself and it being determined by a small group of core developers. And here we it's also being developed, being determined by, I would say, a relatively small court group of core, you know, elevators. But I've at least to say those validators are the governance, are providing the governance of the block chain. So I would I you know, of course this is based on where I sat right you know, bick but for me it's probably more preferable to have the validators determine what new functionalities goes on the boxing rather than core developers, who basically do not aswer as the community. But you know, people can argue about this all the time. You know, that's you know, but that's a builty of it, right. We got to try different governance mechanisms. So, you know, for us, you know we can. We can update the block chain without forking or stopping or changing the software, even changing the software. We can add in software to the box in this way. So we compared this to, you know, replacing the endine where the airplane still flying. Your an right, you know. That's because you've the passengers vote to say that we need a new anything that we can put on a new engine. So that's how we make sure that everyone has the same set of extensions and then, you know, those extensions themselves. We also only support a limited number of operating systems. So we are like the sum that supports a large variety of different operating systems out there. You know, the suburbus boxing as of today, only have you know, no software for for a small number of denux distributions. That's our limitation, but that's we hope to improve this in the future by building our new tools and building our building out for new platforms. But as of today we only supported a couple flaylors of Linux. We are not, you know, likely sroom that that support. That diversity, a large diver stay of Albert existence of set. So those I don't know use that also feel or too questions. Well, do you? It does. So how you see, I have so many questions that it's very difficult to choose which wants to jump in with. I guess. I guess we'll have to talk about we'll start with the proof of steake, delegated proof of stake. Explain. So we've actually never had anybody on the show with delegated proof of steak as their core consensus protocol, and my knowledge of delegated proof of stake isn't quite as strong, but I did remember reading I think it was called bit chairs and I remember reading some articles on it about a year ago, I want to say. And they have a, I think, a delegated...

...proof of steake consensus mechanism, and what I saw was that they were dealing with issues of our questions of not necessarily should be questions of centralization, and it sounds like you are aware of those issues and don't consider them a particular concern. So, for the benefit of our audience and myself, can you go over delegated proof of stake in your in your system, how it works, why you chose it and, you know, address some of the concerns about keeping the network trustless? Right? Yeah, the would love to do that, you know. So you know from our point of view. I think you know. That's I think this is a probably the consensus. The what is dedicated proof of state. The prop of state is direct imitation of a human political system that is called representative democracy with land suffrage. What does that mean? It means people who own land can elect Congress and then the country is governed by the Congress they elect there. Okay, so in not one person per voked. It is your vote depend on how much land you are but still, the the the voters, do not govern the country directly. It's the all government society directly. It's a representative of the converress that they elect. I think this is you know, this is a political system. That's Rome has has adopted and it's also the early political system in America before there's one person, one worlds, right. You know. So there's so there's a large among the presidents of why this, why this is the case and why this is might be interesting, for for Blochain, because, you know, because if, UN like the human political system, if when a person was there's really no consequence for him individually, right, you know, so there's so you see people voting. You know that. You know. I wouldn't say if you're responsibly, but you know that's they can mote what however they want. But in a in political system where there's Vand sufferage, you know, there's because you have to put your land or your probably online. So if you want all those, those the problemties that put online could be flash. So this slashing mechanism, I think, is critical for any POS system. Now the BBLs system and POS system are, you know, I think the it's probite different level of centralization. You know, it's if you have truely pos system that everyone can walk the end, every content government, then that's then you may argue if it's provide the the the best decentalization. However, in a DPUS system, where you elect your your representative to govern the system for you, then by definitition you have a level of centralization because the elected representatives are centralized. If a just centralized complements. However, in an extreme where everyone is elected representative, you know, so you know, if you elect a large number of elective representatives, then you have sufficiently decentralized right. You know, that's if the number of the representatives is the same as a number of voters, then you are pos. The exactly like PS. However, if you're only erecting one representative, the king or the you know all the dictat it are, then you are very centralized. So Bloss dopos and the pols. The centralization debate is a matter of degree, you know, it's a how decentralized I can make. Might be pos but well, certain. The thing with Pos is that, you know, anybody with the proper amount of stake can decide to be a validator and check the network, and that means anybody could basically say I want to make sure that if the ums honest and they can detect problems themselves, with delegated proof ofstake, it's about the number of people. The the from what I understand it's the amount of currency someone has is is somehow tied to their voting power, because you can't mean in theory, what you could do is just take your, you know, all your currency and have have a minimum, minimum vote amount on several nodes that you run, and then you've just increased in number of votes you have. So it's really useless to try and cap that. What you do instead, as you do vote power proportional to currency owned. Is that correct, and I understanding that right? Right, you know so. So there are too levels, right. You know? Why? Is How much currencies that you have and is ilyse how much currencies other people can vote for you. So you know. Well, different people are different. Dopis a system set out right in our system is a lot like our political system. So the congress person can be poor, he can have no...

...money, but you people who have money want to stake him, they can elect him into Congress. You know what I mean. So so you know. So it's a matter of campaigning for the for the community, so that the people that token holders themselves may not necessarily want to be at Sadi data or that all delegates, right, they can just elect someone to act on their behalf. So that that brings up a question that I have is what would it take to own the network from a currency standpoint? So let's say how much, how much currency would some many have to own? How much voting power would someone have to control in order to sway the network in their favor? Right? So, so that actually to two questions. Well, actually, this is a complicated question. I A prive three answers. But also, is that not too much? You know? So the first is, you know, whether you know, if someone has a controls a large amount of currency or someone controls a large amount of voting power, can he cheat on the network? You know, basically school the network to be spayor yes, you know, that's a that's also, you know, one of the things that we that we repeatedly see, which which I won't name names, are certain projects that have thepos what is the foundation controls? Ninety percent of the course, the token supply. You know, then it's essentially the foundation going to do whatever they want, right, because, you know, because they have they have all the voting powers centially, and that's but, but that, but that's also the case. That's also not the case for a lot of those, for a lot of the dpos networks, for instance, was, you know, the foundation controls less than ten percent of the understanding on coins, right you know. So, so you know there's no major players that have so much voting power that are going to screw it, right you know? So that's the first styles. And the second answer is in the protocol. So different people have different mechanism to prevent the aggregation of votes and to prevent, you know, bad behaviors that are going to happen on the network. So in our protocol that we have a cap of top percent vote. So you know, when you have voting power you get benefit from the network, you know, in the form of you know, say, you know the network has certain inflation building so if you are so you have validator, you can get part of the inflation can benefits, right you know. So that's, you know, essentially how most of the DBUS network set up. And we have a cap of top percent. So if you go over top percent you get no additional benefits. So in that case the economic incentive is a if goes strongly against people staking you over top percent. So you find my large total holder and you are put a tissue and wants to be a nit air note and you already attract twelve percent of the community's votes. There's no reason for me to stake you anymore because the Polka I stank. You would not get in it any benefit, but you know so, because you would not own any traditional ECLASIA. So I would chose to stake someone I also have a smaller stake, but in a kind of like. So this is not the same as elected officials, where they get, you know, they get one vote. Type the deal, because technically, let's just say, some bitcoin whale says I'm going to own the cyber miles network. Okay, I'm going to buy, you know, let's just say over fifty percent of the currency. It's probably more like two thirds. Is what you need. But considering in the courtesies probably valued kind of low at the moment because you know, it's new, then they would go there will be something you could just totally do as it could go. I'm going to buy, you know, a significant portion of the currency, or me and my friends are going to buy a significant portion of the currency. These kind of people also have major compute power at their figure tips. So how do you stop a whale from throwing up, you know, let's just say five voters, five up, sort five, five delegates to represent the the network, out of out of like let's just say nine, okay. And so now they control a majority vote, even though, you know, just by virtue of their curretty and even and they're also staking in their own votes, they have each one of the twelve percent. So they'll control sixty percent of the vote. How do you, how do you stop that kind of thing from happening? Because in the in the market place, we have regulation, and so this is this is not a, you know, a pure like government model and that a government person, you know, a congressman, can...

...only go and have one ya or Nay. This is a this is a proportional model where certain congressman's have congressman and more weight. And it's doesn't sound like there's a strong cap on the number of congressmen available. So I could spread out my currency and throw up my own delegates, but furthermore, my friends who are aligned in my interest can do the same and we would have a sort of a trust situation, you know, like we have antitrust laws in so I'm speaking of, you know, business trust. We have anti trust laws in our system to stop certain industries from talking to each other. For insist, if you want to speak with the CEOS of the Tier One wireless carriers, they can't be in the same in room without the STC being present. So what can you do to prevent somebody from creating a sort of brain or economic trust from manipulating your network? Right, so, you know, that's I think. What do you have? Just describe the civil attack or you know, or you know, you and you can have you know. You know. That's people team together to try to come control the network. You know, people team together to control country as a while. You know. So, is there a mathematical way that we can prevent this to happening? Now? There's not. So, but there are very good solutions, however. You know why? Because because all though the validator notes propose the blocks and builds the block and and and produce the blocks, the all the work that they have done is out there in open, right, you know. So people can all the other notes, Ph notes on the networks. They're not validators and they get they do not get paid for, you know, for some information. You can still see and check and independently check what the validators has done. So if the validators decide to cheat, they add their own illegal transactions into the into the system. So, for instance, transferring money without the private key and you know, things like that, without proper signature. Can they do it if they control the voting hire? Yes, they can do it, but once they do it everybody else on the outside would be able to see it and the network would lose public trust at that point and then all the coins goes to zero. Because if the network social factors to sort of keep the network honest, rather than here the economic factors. So I could see where that that might work in practice. I can also see where somebody could obfuscate that information in practice, meaning that it's not them but it's their cousin's best friend or it's somebody who hosts them a bunch of money. Well, you'll be the validator for this network. Blah. Well, we're going to take all the shares out, you're going to have to run the hardware, and that's okay. I could see a lot of ways that people can in social and so in the real world, like manipulate the system so that people don't even know it's the same person. How do you detect that as a user on the network? Well, yeah, that's so, you know. So I think this is the check I just describe does not depend on with I knows a wizard. Those validator the same person, all related by by something, by blood, all buy some financial interest. Is that if they want to schools the network, I have to do something illegal, meaning to have to do something mathematically legal, right, you know, they would have to transfer money out of Acom with all the signature, and this can be done by them because they have the voting power to do it. They say this is a legitimate transaction and has to be gloting a black up. The cant do that. However, once they do that, everybody else on the network also sees it. Yeah, but that's that's the underlying problem. That doesn't team. You know, the general problem is really and problem of ordering of transactions more than it is anything else, meaning that you can order transactions in your favor and not do anything illegal and still the the network wouldn't know any better, because the whole purpose of validators is to choose an order and elect a leader to choose that order of the transactions and how they're processed so they don't have to necessarily do anything illegal to also still game the network. Well, you know so. So I guess we're at the whey describing different levels of the packs, right, you know. So there's say the most the most severe type of the attack is to easy money for this out, but you know that's that will be really illegal. But you've got talking about other most subtle things. I think you know. One of the things that's you know, Talb about open, transparent network. He's have everything is network can be seen by everybody else on the...

...network. The things that are added that are dead, like you said, is to make it official, used to say this is this is the way that we think this should be done. Right. You know, this is our conference has because you know, there's there's x number of validators and two sorts of them. Said, you know, we have but we, you know, we to do it. They and have the same results, you know, especially for smart contact and you know, sings like that. But everyone else on the outside can also compute to themselves and it's some although there's no mechanism in a network for you to say to dispute the results from the validators, the Vody, that is say final. However, in a larger social contact you can all you can always say those value that are start to cheat and for different reasons. Maybe they have a natural interest or maybe they've just been bribe. Right, maybe there's some. You know, we talk about Bitcoin. Wells they don't. They do not even have to buy those coins. They can just the promise to buy the validators. Right, you know how to prevent something like that? You know, that's you know, it's a it's a they would have to do something that the community disagree. If the if the outside coming, if it sufficient people on the outside of community disagree with what the validated, then you know, I think the network would lose value. And the network loose value. There's a direct consequence to the net, to the validators, because the validators holds most of the coins. Right, you know they are not for most financial interest. Not to say this happen. So you know so, but what I'm describing is a social experiment. You know, it's a when we look at Bitcoin, ten years ago, we did not know that people going to join mining stuff, attacking it, because there's you can't. People might make money from fifty one percent attacks, then mining right, but turns up mining safe, you know, at least up on to your now, you know, as far as we know, should be to clean your mine. Is Safe. You know so. So I think for for Peo ass whole or Dopos potball forward there that could be a lot of experimentations and, you know, things like fib you know, that's we are one of the systems that we are trying to build that. That's that we think has a has a good chance of being safe. Right. But is it perfect to say I can never gattee that, like you said? You know, there's there's a hundred times that you can think, you know, that's that's that's that people might attactor. But then I say people won't not HAP to because they're going to lose money. And you say all that. You know the money they lose. It's not going to be the money that they get. You know, that's you know, that's but that's I think that's why we need more blockchains than any term. You know, we need more. You know, a situation where you have a lot of block teams that a lot of experimentations and they can exchange in that you can some format, but that's that may be a discussing one another day, but that's because we, you know a lot of our audience, are in the space trying to build on atherium and are also trying to build alternatives. So I see your point about how etherium doesn't quite fit. Then the market doesn't a niche fit both with, you know, the current state of the consensus mechanism and the and the EVM. Why not just why? Why not just build your your system as a plasma chain, though I still don't see a solid I see a need for a layer two solution like yours, but my question is, why not back it with something that's trust less, meaning that people have an exit strategy in the event that they find something wrong, so they don't want to devalue it because it's being backed by something super strong like etherium? Current, early implementations of plasma are available and that you know, they'll probably round it out sooner rather than later. I think there's been some articles on coin does saying that plasma is dead, but I will guarantee you that's not the case. Plasma's very much alive. It's it's actually thriving. They're doing some really cool stuff. OF PLASMA CASH and just regular plasma contracts. There's there's very interesting ways of staking into separate chains. Why not join that effort and make cyber miles a high throughput dpos backed e ORC twenty or your see whatever token, to to enable the trustless mechanism that etherium has to back the cyber miles protocol and also maybe enable other people to fork your stuff and put put up their own sort of cyber cyber miles variant. Yeah, you know that that's you know, that's that's actually something that we are we're actively looking too, you know, because, you know, because it's were open source, right. You know, so someone want to use our software and build a side change, and you should that.

They can definitely do that. However, I would also say, you know, that's the plasma is really, really new. You know. I think plasma mvpeaces came out like a couple months ago and there's lots of issues, resolve issues and as far as I know, there's no less than four, five, you know, competing plasma implementations, each addressing a different problem. Although I saw cows talk, they account for I was in the room and that's you know, that's that's peas motion is that there's a you know, we are we have finally done the long time, but at the top of the here and we can see the summaries. You know, as much as I want to believe that, I think, you know, it's a probably were optimistic because you know, and we're also addressing some of the most easiest to use cases for Pasma as right now. You know, some Evm, you know there's no virtual machine consplization. There lots of things that I think plasma has had a way to go. And also, I would also like to point out what we have done is actually sharting. You know, it's a it's a plasma in the side chain, but sharting is another way to, you know, increase easier throughput right. You know, that's what we have done. Is What we call logical shouting, because if there's ISS compatible block chains for each particular application, verticles, then you can share your workloads this way, because games going to be on blocking, a e commerce on boxing, be, you know, financial blox, see Iot on boxing, for you know, that's you. You can actually share your work, workloade like this, as long as there's a way for those blog teams to exchange tokens and because they're all in some compatible, it's, it's it might might be possible to connect it to big to the beaconching like this way, right, you know. So I think there's there's lots of interesting and unresolved issues. But we can't wait, you know, we can't. You know, you know, our project started until months ago. When plassma happens and exist by, you know, back then, not even the idea, right, you know. That's we can't wait for plasma to be fully maturized and then build our stop on public right, you know. But we are open source projects that we are actually, you know, exploring it. Is it possible for our virtual machine to be wrong as a PASTMA, as plasma side change? So what do you have? What do you have? Just discussed this very much our mind as well. You know, how do we play with with an ecosystem that's larger community as be of fight? You know, all do we play the charted all do we play in Pasma? You know, Bob, you saw, but we need to see something and we come sit here and the at the and it just the wait, right, you know. That's you know, that's we used probate yards we twin the cult and and it just have a traded on sense. You know, that's not what we want to do. We want to be actively building stuff, you Ain, for the means that we have to build at all ourselves and maybe considering the grain later. I completely agree with that approach. It's a business answer. It makes perfect sense to me. Build your build your own thing and when the time comes to the technology matures and you can integrate, be open to the idea. That's all. That's all I really need to know. Is because a lot there's a lot of side projects out there where people are doing their own thing and I feel like, you know, they're you know, there's got to be. I have a personal philosophy there will be one blockchain to rule them all. There has to be. There has to be one single source of truth. We cannot have a mishmash of different truth mechanisms out there, because one will eventually be the category king, one will eventually be just the de facto standard and it's just an evolutionary process. It takes time, it takes it takes maturity of the technology, it takes, you know, it takes adoption and all these things are slowly happening and slowly coming together. But I agree if you need to build something to to be ready for that moment, do it now and just go for it. So I totally get that Um and also I want to go back to the virtual machine stuff. So we could talk about DPOS and it's kind of controversial, flaws all day, but I think that's been, you know, to use the episode in the podcast title, Hashed out quite a bit already elsewhere. I think. I think that DPA sounds like a wise choice for you for your needs right now, and I do think that it would actually be more interesting when it's integrated with like an etherium blockchain. But the the thing that's really the most interesting to me is still the virtual machine that you've got going on. So you've mentioned that the virtual machine was built by a budge of crhotographers with cryptography and mine, and because of that it's had a lot of flaws.

I want to say not flaws is not the right word. Let's just say hardlined features that I particularly find difficult to deal with, one of them being that register sizes in the virtual machine are fixed at two hundred and fifty six bits, which is a large space, and also it means that you anytime you want to store anything, I mean anything, you're consuming two hundred and fifty six bits right there off the bat, which means that any sort of storage onto chain, if you want to store a single character to flag something, that's two hundre or fifty six bits. Right, I should be seven bit, but no, it's two hundred and fifty six bits. You you know things like that. The lack of SEP like encryption functionality in such an encryption oriented system seems peculiar to me in that I'd like to do some elliptical curve work inside by virtual machine, but I can't. So you know that that's interesting to me. And another one is I don't really need turning completeness. I need verification. I need formal verification and a lot of my contracts. So why would I need a full turn complete system which can't be form of formally verified, when all I need for like ninety, I'd say, a solid ninety eight, ninety nine percent of all contracts out there just need basic, non turn complete procedural kind of like almost like database procedures kind of work. What? What is your virtual machine doing with regard to any of that, if anything? And what are the does what does the design philosophy around your upcodes in your your virtual machine design? Right? So, yeah, that's I'm you brought it up. You know. So there's, you know, for the EVM, there's we can make fundamental changes and then we can make your particular rbal changes for the like the the size of the fundamental you know, Dair structure. We tend to not to make those changes for now because that breaks compatibility. You know, we want so they did the applications so that the Barticle to be able to run our version. We should we not change so. So we basically left those most of the things there. However, we made some fundamental changes. So, for instance, you know, when you talk about touring complete, one of the issues that's I think I would even say the majority of smart contract issues are integer overflow. You know, is that, you know, when people explored because you know there's a trouble, has a safe mass library and sometimes people don't use it, and then that's create opportunity for hackers to, you know, to intentionally call those contracts and and and add all multiply very large numbers and overflow those numbers and the the the overflow is an interesting feature. That's that's that's it's part of the touring complete right, you know, because people can always argue maybe the programmers intended to overflow, it's some kind of feature. However, we say, you know, like ninety nine, point nine and Nin nine percent of the time it's but you know, that's so. We would rather shut down that feature. So, in terms of ALP code, we could. We we introduced a new out code, new data have called safe integer. At safe integer level in the virtual machines, automatically wrapped around by the safe mass library, you know, and also e. When you without the safe mass, if the virtual machine detects overflow, it's a bought the execution. Its execution just to fails. It would not overflow and print pretend the PROA. The program is still running correctly. So you know. So. So I'm just giving you an example of the fundamental data types that we are trying to modify their behavior or change their things, although you could say that's by by not allowing, you know, integer overflow, that we are breaking compatibility with the sum because some applications that will behave certain ways on you term would not behave the same way our word of machine. But that's a risk we want to take, you know, because we think this this behavior change ninety nine percent nine and ninety nine times, is about that we help to fix. Right. Well, do you make it an explicit type, like a safe in type or safe you and type, or do you go ahead and just make all new and safe mass? We did both, you know. So it's we have a safe integer type. That's if you declare that type and use that type, it's it's all operation around. It is being wrapped in safe mass. But we also have at the virtual machine level, at the round time level, we we monitor those you ins and the...

...if, the if, the if. They used to two hundred and fifty five, two hundred and fifty six overflows, the virtual machine stops and so error. You know that. Yeah, so, yeah, the course I was saying asked is if, if you know the you went on the virtual machine overflows on something that isn't declared safe, then it would still be compatible, because you know, that's just how solidity is. But if you only did it for the safe version, then that would that would also. That would. That would if you did it for the safe version and the regular version, it almost seems like there's no point in having the safe version. And Yeah, well, there are still something subtle different. That's that's you know, that's are probably would getting do here. But, but, but it all right. You know, it's if the safe vergine safe and as a regular version is on safe, then we can still claim compatibility. We can do it that way. But we just the so many existing contracts out there that I'll say, if that using regular version right. You know, that's we thought it might be better just to fix on that as well, if you sup. So that's but that's a that's a toy that we make. You know, that's as you know, all those choices are hard because you know, they all have trade offs. You know, that's definitely, definitely incompatibilities a difficult thing to stick to. Stick your guns on that. So what about having basically encryption functions with that be relegated to the the sea extensions? Yes, so, so, so that idea comes from, you know, you know, Early Java Day. So you know that's you know, if you like, Oh, the guy like me, you know that's such fun. Joba for many years, as you all know, the Jena, the Java native in Nevers. You know, because the origin and because original Java was very sandboxed. So if you want to access anything like a device driver, like the sound card or the video card or whatever, you know that you need the device driver. That rhythm of Java and you know that's a Jaen Life. But that's so we have something called the laby in line that we have. We have a special op code. That's can you know? That can wrap around C plus plus applications in that. So we have, you know, you know our say encryption. You know that's a basic PKA, encryptions and all that stuff. Reading said plus us that can be directly called from the virtual machine. By the way, this is not a really a new approach. This is exactly what sum does with their native contracts. But, like I said, you know, the difference between this and native contract is this can be added at any time by a vote, but the native contract will have to be down by software. UPDATS circles, right, you know. So so there's there are lots of those eat. A lot of those things are not, I wouldn't say that radically different technology, or they are, you know, just the other that we are so smart that other people haven't thought about it's not. People have thought about it. We just took a different approach. You know, that's and then there's one of the thing I want to talk about on the virtual machine is at the work as a language. About what we have integration, which GPOS. What does that mean? It means, you know, because one of the key use cases in being in ECOMMERCE is that we need our coles, we need to bring optain data on chain. You know, so for INS like delivery confirmation, bank accounts, you know, things of that nature, right with the product, physical product is being sold or be received. You know, things like that. In the I've been saying, oh the world, or in the world of re serum, the article is a vergect at the problem. That's you would have to provide centralized trust or you would play some kind of cryptual economic games. You know, even alarm clock, and you Sam is very hard that you would have a community help people who would be in sentilized by some kind of synderized incentive mechanism that would send certain signal to the blockchain at certain time. You know, that's that's that's there's project call easyum alarm clock, right, you know, so for on regularly serum the the cost of of having our callsizing is very high. But for us that we have at the virtual machine. Let what we have. You know, you know, in a in a in a regular SORTITA, there's a there's a you can specify some function as owner only, you know, that's only the owner of the contract can operate on those. We have a new key to or codevaled the only meaning, the only current value. The dators can operate on those contracts and then we can't. We trying to extend it so that's only the majority of the valdidators can won't to perform certain functions. That allows us to bring often data onund change for validators, because we made an assumption that the validators provide governments and they are trustworthy, because if they are not trustworthy, then all the stuff on the boxing's trustversy, right. You know, so he yeah, you know, you've we assume those are good guys. The then we can't rely on them to bring real word information like...

...credex receipts and you know, things like that. They can provide the service to the box chain by by, you know, by a feeding this data into the INSO smart pocket contract. Essentially, as validators can act as a natural articles in that's case. So that's also, you know, one of the things that we get for your commerce. Then, you know, Honderstand this can be arguable both ways. Can Be good or bad. You know, that's your to centralize, you know, but but that's that's that's the decision that we may you know, that's the approach we cook at this moment. It makes sense it. We went through that kind of it makes sense why you went, why you went with this approach. But yeah, now, and I think it's really great, because you want to iterate fast over your machine as well, meaning that you're seems to be like the DPOS is a almost mechanism for driving innovation with your with your group, meaning that if you want to add more functionality to your system, if you ask is a lot easier to get yet done with, then then just regular troop of work and steak. Is it really out there yet? So yeah, no, I think that that makes perfect sense to me. The question I have now is what are the tools look like for this chain, meaning that wallets, you know, mining validator nodes, you just regular client nodes, since you're trying to get your confirmation times down, you said below ten seconds. That could increase the size of the blockchain significantly. So do you have a light client? What is the tooling look like for the end user? Right, so, we don't have a light client right now. You know, it's just, you know, there are a lot of development effort. However, we have a pretty because we had used them compatible. So we are standing on the shorters of giant type, right, you know. So, so, so we can fall all out of the open source project that we did, you know, by adopting this and open source product a far would need. So, instance, we have a you know, we have a new programming language. Is that's the based on. So that it is so we can. So we modified it a Solty, that compiler, and you know, that's what we call a language, lady. You know, it just we solt it the more liquid than solid, right. You know. So it's as we have LETDC that can compile it, and then we integrate that into remix. So we have our own fall off Remix, the of the web based ID for for solidity, and then we have our own version for Fox, version of Meta masks, so that to set is complete. You know. So you can have a Meta mass down several miles and remix on side of miles, have them work together to deploy a contract and executive contract written in the language that we provide, you know, meaning for it today and exixtension. That's awesome. An it's a coupable like existing libraries like webstere or it's. It's your in JS. Right. You know? So not? Not. Well, yes, are you certain, Jess? You know, for the certain functions? Yes, so, for Webstery, we have a website, cmke, because we added a lot of functions and transaction types to the blocking. So all the ACERTAIN block transacting types we already support. But for dpos related stuff, for instance, how do you elect delegate? How do you you know, how do you like the radity? How do you declare? How do you slash? You know, things like that. There's a large number of transactions that we added to that library. So we quite a website, CMT dojs. So that's it's very unimaginative name. You know, that's the terrible names. But but those are, but those are but all those projects traits are lineage to veris parts of the term, right. You know. That's we took those, we fox them and then we make it work on our boxing and then we try to find a way to contribute back, you know, for the certain project. Right. So, so we talked about webstery, remacs, Meta masks, and then we have our ARTC services. It's sort of like a Inferra, you know. That's those service providers. Are you sure them? And then we have for the consumers, we have a mobile wallet called the CMT Wallet where, you know, it works with all our our MINNET and the Itas lists, all the list of the MT, which is a symbol of all token and also the tokens that issue a block chained. Also have a naming services, a little like S, but it's for the for the token names, right. And for that Wallet There's an ftck that you can develop applicate. Is that we can develop apps that directly make is sort of like mathematic integration on the broad side, but this is on the inside of the mobile APP. You can develop apps that that's the wallet to pay for your gas and you know, things like that. So you can you can write edge temp message demel that directly inside the inside of the wallet. So there's the point I'm trying to make. Is there a variety of different tools. That's people are already familiable...

...and they can just you know, I think the learning cars is not very hard. You know so, because all the certain stuff come around on side our mass and then you thinks for the things that that we added. You know what I love about that is that, okay, said against your fork of Theium, you're using existing tools which people are already familiar with. Your experimenting with new consensus mechanisms, you're experimenting with extending the functionality of the virtual machine. I really see fiber miles as being a very interesting project for extending the capabilities of the etherium network, because a lot of what you're doing could technically applied to the theorium. If it works right. They could look at what you're doing and go, that's amazing, we're going to adopt this into the main, main net. They've done it before. That's not something oncommon. I mean was it Zcash had the ZK snarks algorithm. They went, oh, that's a good idea and when ahead and dropped it into the the etherium Vertu, you know, the the etherium virtual machine. So, like there's a lot of there's a lot of features that you're experimenting with because you have the freedom to do that and I know that you're trying to build a network all that stand alone. But again, like I feel like there's going to be one category king ultimately, and I really do like the research you're doing because it's going to contribute to the entire ecosystem and because you're basing your work off of Atherium, you've basically become an rd branch of atherium. So I really hope a lot of people hear this and decide to contribute to your project, to get some of the features that they think are interesting onto your project in and in doing so, can demonstrate, with a lower barrier to entry, what it would take to do that on the main net and what that would look like and what works and what doesn't work. But on the other hand, I also know that you you have a business that you you know, business models around this where you want your chain to be successful. So I really do hope that you can integrate with other etherium networks as well, because I do see that being kind of the next next phase of of the evolution. is going to be layer two, right, yeah, you know so. So for us, if we ended up being that, you can converse the Shard Offen, you cirum, then then our maschions compete. Accomplished right, you know, because we want ECOMMERCE, something they can to the point on our chain, whether it's uniform off our side chain or in the form up Shard, that's all fine with us. It's just, you know, that we have to start somewhere. You know, we have to build album, we have to go our own we have to put the first to spect right. You know, if the community comes to the consensus and said, you know, that's the assume, BECO network going to be the torus provider and everybody else being a shot, that's perfect fie with us that we want to be in part of that world. If that's you know, that's to echo what you said, you know, that's we want to be the pioneer united space, not to experimenting with a lot of stuff, you know, especially it's the things that related to the ECOMMERCE and to bring those business down to the blockchain and and and the you know, three years or four years from now, where how we play into the ecosystem? We hope that we are where we would be active and the good player in the usure room ecosystem. That's you know, that's we have been doing open source for many years. You know my share, my personality. I personally have been doing open so for many years. So you know, that's you know, that's the community is our biggest ass that we want to be successful together. That's why I want to say yeah, yeah, now, that's great. So how did how did the name come up? How did you name your name? Your products are your chain cyber miles. Would where that come from? Right about? So a lot of people when they heard they thought it's a, it's a, it's a, it's an airline, multi program being a boxing, you know, but it's not. It's it's because we have a we have an ECOMMERCE company called Five Miles in Dallas. It's a, it's a, it's it is actually a very large consumer to consumer ECOMMERCE company and it's you can think of like crack list, but it's for low coal. So it's all in five miles and it's actually one of the biggest merchandise that be moved by that time for a used cars. I think time house is one of the largest use copulars in the United States at this moment. You know. So it's has you know, fifteen million working users. We're just on the network. So you know, that's a but that's the route. However, that's how we e commerce. Five miles, you know, everything being boy in five miles, and that's we really want to build a...

...decentralized see to see network from firehouse. That's how we're original vision. But that's a you know. You know, on the roads there we find that we are ready need to build up on infrastructure to go there. So we called it infrastructor cybe our mouse. You know, it just the mile makes it look like. I gotta say, you know, yeah, I came in a little bit skeptical because it did sound like airline miles, and it is kind of sound like something like that. I was like, what is this, what is this actually about? And then I also hear people building their own blockchain. I'm like another one. But the infrastructure, in the technology behind what you're doing is so sound in the way that it it's it's definitely amazing research that needs to be done and I love the sandbox environment that's ow several miles provides for doing that. I really, I really am enthusiastic about this project. I think I think you definitely dealted me quite a bit because it's just, it's it's just, it just makes sense the way you're doing things and I really do see that it could play a larger roll down the line. I'm in the etherorem ecosystem, you know, as far as you know, the virtual machine research and eventually like integration of different consensus protocols into the main network or you know, just a lot of the stuff seems to make perfect sense to me and will contribute to the larger picture of what we're trying to build with these decentralized systems. So yeah, Major, Major Kudos to you guys. This is not a small undertaking. Building something like this is is a very ambitious project. So I'm kind of curious actually, what's the size of your team? How are you? How are you guys doing so much work so quickly? Well, yeah, that's you know, ask any good open source project. We don't have over logical we have maybe, I think tolf developers were working on this whole time. And you know, so that's a maybe six up, some one working on the virtual machine and six of them working on the block team. But then again, you know what, I would say that we stand on the showed of giants because, you know, we benefit tremendous if from other people's work in this in this community. So, for instance, we benefit commendous from tenement. They provide bft conference as today, and rebuild the ppus on top of that. And we definitely benefit to from a lot of workings. You don't have to, and also the people you do the community will have to a lot of the touring likes of bet a, mask remix and all those works. That's you know, that save US tremendous amount of time. You know, but that's what I consider are the true spirit of owing source right, you build stuff and you know, what you wish is that other people can use that stuff, other people can build on top up at the which I think we're good player in this, in this and you know that is a fact that we that that we can't experiment these all out of those ideas with, you know, with other people. softwarezing testifies, you know, the power of the community in this in the space. Yeah, no, I think it's great what you're what you're doing. I I'm impressed that so much work has been done in twelve months, even though you're standing on the back is giants. It's no small task to just take their code and make it work for a whole new system. So, you know, mad kudos to you. Is there anything, anything we should asked or anything you want to tell our audience before we wrap this up? Well, I think that's pretty much yet. That's so, that's I really appreciate your questions in the comments. You know, that's you know, lots of work ahead of us and we want to be part of the the success. Like you said, you know, when the technology is ready, we want to be there. We want to be the wives who have the most ready solutions off there, right, you know. So that's that's what we're going to do. Yeah, so how can they find cyber miles and how can they find your repose? And how can they how can they find you, Michael? Right, you know, so it's all. Our website is cyber miles thought I all, you know, cyber miles thought Ale and my if you've got to reach me, so email you can say Michael at side of miles thought aisle. And you know, all give out up a documentation and, you know, get up a repositories, just gipp, githubcom snash, cider mouse and all this stuff or old sauce. All our code open sauce. We have a no close sauce software. So everything's open sauce. Everything is out there in the open and you know to get on through long and no down man that come up round of data. To do anything. I just go to the website and and go home there. That's awesome. That's awesome. I love what you're doing. Keep it up. I hope to see more innovation coming out of your group. So yeah, and as always, you can find me at Colin Cuche Shay, that's Cel N. see us see on twitter, and Corey,...

...who is unfortunately unable to attend this this interview, is at Core Petty. That's C R petty on twitter and you can find us on hashing it out that stream. Thanks. Michael's really great. Great learning about someper mouse. All right, thank you.

In-Stream Audio Search

NEW

Search across all episodes within this podcast

Episodes (134)