Hashing It Out
Hashing It Out

Episode 23 · 4 years ago

Hashing It Out #23: Avalanche - Emin Gün Sirer

ABOUT THIS EPISODE

Revolutionary! On this episode we have the brilliant Dr. Gün Sirer to whet our appetites for Avalanche, a new category of consensus protocol. He distinguishes Avalanche from Paxos/RAFT systems, blockchain, and traditional gossip networks to show how this new algorithm prevents double spends while allowing for high throughput (early tests supporting over 1200 tx/s). We learn about his new coin, Ava, which will test the limits of Avalanche's public consensus on a global scale! The future is exciting!

Andy 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. Episode Twenty Three Hashing it out. As always, I'm Dr Corey Petty and I'm here with Colin Cuche. Say Hello Colin. Hello Col and today's guest is somebody we've been trying to get a hold of for quite a while, Dr Emmon Doon Serier, and we brought you onto the Bitcoin podcast a long time ago and when you've been there a multiple times, but in the first time you often you told us that no one calls you by Inman and everyone just calls you by goon surrier or doctor surreer. Is that still true or I think some of your celebrity over the past couple of years as definitely had you people call calling you inmon for a long time. You just gotten used to it now? No, yeah, I've got start to use it more and more often. So I respond to anything. I'll respond to hey, YO, whatever else. But you know, all my friends call me good. That's my given name. The first name is a formality. You inherit from your grandfather. So, but it's okay and the way I use it actually I can use it to tell apart the people who've met me from the people who have not met me. So some people who haven't met me called me Emin and people who've met me in person called me good. If any emails start with dear em and the spam filter is much more like it's a good, good trick. Everybody knows that. So so we brought you on to talk about a paper that just kind of came out of nowhere on IPFs regarding a new consensus mechanism, and it's been down that your are your distributed systems professor and you've been doing it for a very long time before pitcoin. Can you, can you start to and I think we'd like to do is try and get both a bird's eye view of what this is and what it's implications are, as well as a lot of some of the technical details and white different how it differential itself from other consensus models. Sure, and of course, you're referring to the avalanche paper. That is correct, very exciting paper to hit the air waves. Just to provide the big picture here. So the problem of consensus has been with us for a long time. It's a long standing issue. It's at the core of almost every distributed system that people use. Right. So all of the Google services have exactly the same issue that bitcoin and etherium and other cryptocurrencies face, which is Google has a bunch of servers. It would like to keep them in sync. This is a really big fundamental problem. And also from databases. You have a bunch of database servers, they store some data. You'd like to keep them in sync or in distributed control. You have a bunch of machines. They'd like to take concerted action. They should all take the same action. That, once again, is a consensus problem. So the problem has been with us for at least forty years that I can trace back, and the early attempts were quite naive and Andre were sort of ill founded and there were a series of papers that sort of laid out the field and established the very first approach to this problem back in the late s. So the best known of those papers is Leslie Lamp Ports Byzantine General's paper, and it's a fascinating paper and it laid out the field and it started this style of solutions or family of solutions to the consensus problem that are sound, and I call that family the Lamport liscof family. The two people behind it are Leslie Lamport and Barbara Discop and they both have touring awards for very, very good reasons and they've done a huge, huge service to the community by establishing this area and by going into depth into that particular family of solutions. And all of those solutions rely on all of us figuring out who should keep the data, you know, the set of nodes that that are going to be holding our information, and then coming up with a read protocol, in a right protocol for modifying the data and reading the data. So so you know, these protocols are still in existence. They're used in every permission network,...

...so hipper, lager, Corda are and you name it. You know, of course, Google's lock service. You know, everything that you do ends up touching one of these these classical protocols to for consensus. So that's the one family. It started out in the late s and has been hashed out to death. If you ask me, there are umpteen different variants of these, but they all come down to one fundamental sort of mode of operation. We all have a quorum, let's say. Just if you want a picture in your mind, it would be kind of like a parliament, right. You have, you know, a parliament of let's say, a hundred senators, and then you and I are supposed to to enact laws that everybody should agree on. So then, what should the protocol be? Well, it turns out that there are some that work well and can be proven, and they all op create by US writing to a supermajority of the senators or the nodes, and then reading from some subset of them to ensure that the set you right too, and the set I read from always has a correct node in its intersection, or sometimes a plurality, a majority of correct nodes in its intersection. They all depending on the fault model. So so, okay, so that's sort of that. And and just from the mode of operation, you can see that this approach will fail if you and I don't agree exactly on who the senators are. Okay, well, who exactly is in the room? So if you and I don't know who those people are. Then we can. It's going to be an uphell battle ensuring that you and I have intersections of the right sizing in common. And if we don't have that, then all of the invariants breakdown, the correctness breaks down and the protocols stopped working. So they're very fragile and by their mode of operation they entail the sort of an almost a permissioning step. You, we all have to agree on WHO's going to be in the set of agreeing nodes, who's going to be the senators or who's going to be the nodes in the system, and therefore it's kind of difficult to build something like a cryptocurrency and open, permissionless system based on these protocols. So that's why when the second family hit the scenes, it was a major breakthrough. And so that major breakthrough, of course, was done by one Sotoshi Nakamoto, whose identity were still not very clear on and said, Oh, she was brilliant. Right. He change the problem them in multiple different ways, and I can expand on those if you guys want to, and but the bottom line is he changed the way the protocol works. So instead of US having to know who's going to be in this system a priori. We now have this lottery like system. We have these miners and the minor gets, you know, self elected in some sense. He says, Hey, I solve this really difficult crypto puzzle. You guys have to now bow down to me. I am the leader for the last ten minutes of operation and I will tell you exactly what should have happened in the last ten minutes. That's approximately how the NACOMOTOR consensus protocol works. So that was two thousand and nine. So where are we now? So we did one thousand nine hundred and eighty nine was the first family and it lasted a long time. Two Thousand and nine was the second one, with Nakamoto Consensus. But NKOMOTO has a big problem and the well like has a multiple problems, but the one of the biggest ones, of course, is the energy expenditure. The miners have to be constantly hashing. There is no notion of efficient macomoto. It's by by design. It has to be inefficient. There is no downtime. You the miners cannot be turned off. So so there is that problem. So it's not green, it's not sustainable. And you know, some of the people in the bitcoin circles might say, you, screw the environment, why do we care? You know, let it burn or whatever. I don't subscribe to that, but some people do, even if. So, okay, so let's Day, screw the environment, let the let the Poles melt. You know, it's going to be nicer and Ethnica New York, where I live, if that were to happen. Actually so, even then we would still have a big problem because money leaks out of the system. It has to necessarily leak out of your closed store of value ecosystem because you have to pay those people to burn that electricity, and that electricity has a very real, very tangible cost in the multiple millions of dollars per day. Okay, so, so that's not a very efficient way to build a store of value system. And further of course, there are other problems with Nkomoto, such as the long latency for Bitcoin. It's ten minutes to the first confirmation, sixty minutes to something you can actually take, you know, take to the bank, and of course, the limited throughput, which then leads to limited scale. So that all of that has been hashed out in great depth. And so the Bitcoin community has. You know, it used to be controversial to say that. These days it's not, because even the bitcoin community has realized, Hey, we have to go to layer two solutions to handle the scale problem. So so there are problems with both families. The first family, the classical so landport listoff style, was great, except it's not a good fit.

It requires permissioning. And the second family, then, Coomoto style, is wonderful itself, except, you know, it has it's not green, it's not sustainable, it's slow to first confirmation, slow to finality and is limited in scale. So avalanche comes in two thousand and eighteen and it's only the third such event in history that I can I can, I can find and and it introduces a new family and at that new family combines the best of both. And how does this new family differentiate itself? I mean, where is the value conor how do you how do you come into agreement by not doing the things you did previously? It's very difficult in my eyes, to come to come up with such a system where it's open and permissionless for joining and leaving. Yet you can be sure that you come to an agreement relatively, if quickly, an officially exactly. I think that's that's why this paper is so interesting. So the core mode of operation for for avalanche is fundamentally different from everything that came before it. And for your audience, I think the easy way to summarize it would be its gossip inspired. So what that means is the mode of operation for this protocol has more in common with gossip style protocols then it does with anything else. The correctness does not stem from the intersections and so on and so forth, and and so so maybe the best way for me to describe it is to just sort of go into it and give people a sense of how it works. For curls a clip. Clarification is this. Is it? Is it? Is it flooding or is it gossip? It's gossip. What's the difference? Well, okay, there is a slight difference between the two. Flooding gossip are very similar. It is not a flooding protocol. It's really inspired more by gossip than by anything else. So for the for the audience, again, flooding protocols are those where you have something to say and you contact all of your neighbors and you tell them and then all of your neighbors contact all of their neighbors and they tell tell them. So it's a wavefront of activity inside the network as this thing disseminates through everybody. And so they are used. For example, they're used in Bitcoin to flood transactions. You want your transactions to get into the hands of the miners and the peer to peer overlay floods that information. But they're used in all sorts of other settings. Flooding protocols are very, very useful. So routing protocols use flooding right. So Internet routing, for example, uses announcements that say things like I can get to CNN in two hops or whatever, I can get to this other destination in five hops. All of that information is flooded. So, in contrast, in gossip protocols it's there are different variants of them, but what you do is you know a bunch of people in the network, you have neighbors as before, but you don't contact all of them. You pick like a small subset and then they pick a small subset and so on and so forth, and and so so avalanche happens to be in the latter category. It doesn't require you to contact all of your neighbors. You you contact only a small subset of them. Okay, great, and I guess that, like the the decision making process of how that message, it's relayed across peer to peer as as it hops, is, I guess, the differentiator between previous gossip. It's like cops gossip inspired protocols. So okay. So in gossip protocols, what you're trying to do is you're trying to get information into the hands of other people. Okay, so that's a sensible thing to want. And what you do, as I say, there's you pick a subset of people that you know and then you tell them and then they tell their friends and so forth. And if your network is, you know, of size, n the dissemination will take log and steps. Right. So why? Because, in ef fact, what you're doing is you're creating a random tree on the fly. So, just for the audience, just you guys should know what they told me right before the show. They said our audience is pretty technically sophisticated. So talk as you would to a normal colleague. So that's what I'm going to do. So but let me, let me slow me down. If I end up saying things. Are using terminology. Yeah, absolutely, but what you're really doing in the gossip protocol is you're inducing a tree, a distribute distribution tree, on the fly. I pick a couple of my neighbors, they pick a couple of their neighbors, etc. Etc. And the log and steps, I'm going to reach everybody, because a tree with with height, log and is going to have to to the log and you know, and that equals n number of nodes that I can reach. So that's all fine and Dandy, but gossip protocols are not consensus protocols. Right. So, in gossip the goal is to disseminate information and in consensus I want to know that, you know that we all will, at the same time, in sum atomic in in controvertible fashion, commit this particular factoid or commit this...

...particular step in a state machine transition. So so that there's a huge step in complexity between gossip, which is kind of like a communication protocol, and consensus, which is an agreement protocol. It's it's going to be much, much, much more complicated to come to consensus than to just disseminate something. And just to be clear, so the differentiator between that is the ordering problem. Is that correct? So you can gossip all these transactions out there, but I mean unless they know what order in which they're executing and you don't really have it truth single state cross the network. Is there any ways? That kind of what differentiating between avalanche and traditional gossip protocol right, okay, yeah, that's a good way to view it. So I think a simpler way to be I'll you're right. That's one way to view it. So the simpler way to view it in my from my perspective, is what happens when there are conflicting messages. So in gossip you will get to find out both of the messages, but you won't know which one to do. You don't you know. Do we attack at dawn, do we not attack at dawn? We have to decide, right, and what did the leader actually say? And that's that's something that gossip protocols cannot answer for you. And in consensus protocols, at the end of the day we will have chosen to attack or not attack and all the correct nodes will either attack end mass or they will back off and not attack again, all in unison. So that's the that's the big difference and you're right, if we could, if everybody had could see the same order, then in effect we would have achieved some kind of decision between the two. But because these are independent machines operating at different speeds, seeing the the messages and different orders, then, and subject to different Byzantine behaviors by participants in the network, they they might come to see or perceive a different order if we just use a simple gossip protocol. So so that's the big difference. Gossip will just give us the info but will not order, it will not determine the true, true chosen decisions, whereas a consensus protocol will decide. Okay, this is what happened. All the two conflicting things you heard. The one that we're going to pick is this one. Okay. So we have consensus protocol, is basically make minimizing the difference between local state and global state, and that has to come through some type of decision making process on the local level. How do you how do you do that? Yeah, that's that's exactly the question. So, as I mentioned, we used to know only two ways of achieving consensus. One was the classical style, the Lamport LESCO family of protocols. The other Wasna Comoto, and then comes avalanche, and I'll tell you how I avalanche works. It's very, very simple and unlike Nakamoto, which takes a while to figure out, and unlike Lamport LESCO, which also takes some reasoning and some math to figure out, avalanche is very, very straightforward. So here is how it works. Imagine that you, me and you know everybody else are in a giant, crowded stadium. Okay, we know roughly who's in the stadium, but we don't know everybody who's in the stadium. And you might actually know that there are your might. You might know a five thousand extra people that I don't know about. I might know you know five thousand xt that you don't know about, and that's that's going to be okay for this protocol. That's one of the magic, awesome features of it. So suppose we're in this really, really large stadium. Spose Maracaibo Stadium in Brazil. Right this is the biggest it gets. I think it's a hundred thousand people. So there we are and and we'd like to come to an agreement and and so these agreement protocols are difficult right. So there will be people in the stadium who might want to trip us. They might they might tell us one thing and to tell our neighbors something else. They might flip and flop. We will call them Byzantine, Byzantine actors, or if you want to think of them as trolls, or that's fine too. So what you want to come to a decision, and here's how avalanche does this. And suppose the decision is we want to pick a call are okay, we want to pick between red versus blue. Okay, that's that's our decision. It's abstract. Think of it as well. Whatever, there's this think of it as confirming a transaction, if you will. But but you know, that's not let's just actually go with the color of analogy. So we're going to pick between red and blue, and imagine that everybody has a red card and a blue in fact, one card with red on the front and blue on the back, and everybody initially puts the cards on their forehead with with their preferred color standing out right. So what's going to happen? Well, the following is going to happen. In the very first case, in step was zero. Let's suppose that we have the reds on the Blues distributed around our stadium equally. So we have a fifty split and there's like a mixed up...

...people and you know, half of them have reds facing up and half of them have blues facing up. Our goal is to do some magic and at the end of that magic, we want we want people to pick one of them. So here is how avalanche is. Magic works. Okay, it works as follows. Every note asks just a very small number of other notes what color they have on their forehead. That's it. Okay, so I contact let's say five other people in this ginormous stadium. I just pick five and I say, you know, what color do you have on your forehead? And back come the answers. And and the answers are going to be something like red, red blue, red blue and and, let's say and. So if that were to happen, though, the majority looks to me like they're picking red. All I do is as a correct node, based on the answers I'm hearing from the audience, based on the fact that everybody seems to be going towards read, I add my weight to read. So I flip my own card to read. That's all. That is as simple as this protocol gets. In fact, that's as simple as any consensus protocol gets. So so what does this accomplish? Well, it creates a Meta stable environment. Right. It creates an environment where the protocol is highly unlikely to remain in this in this by valence state. Okay, so that by Valan state is the state where it could either pick red or blue, that fifty division. That's by Valent it could go either way. And so this, this process of trying to everybody trying to put their weight on the same side, creates a sort of an inherently unbalanced situation, what we call a metastable situation. So metastable, Meta stability, is great. That's exactly what you want in in a consensus protocol. You don't want the nodes waffling and flipping and flopping in the middle of the road. You want them pushing their weight towards one decision or another. And so an avalanche. What's going to happen? Okay, well, so so we started fifty. The were Fiftyzero people with red on their foreheads and Fiftyzero with blue, and and everybody asked their five chosen friends what color they had. And at the end of the first round something interesting will happen. Now, what is that? But we're not going to achieve this decision in one round. Then not everybody talks to everybody and and it's impossible to achieve a proper decision. But but at the end of the first round, okay, what's going to happen is really interesting. We will have oversampled either the reds or the blues. So in my example, I ended up over sampling the reds. And you know, we have fifty one we have a hundred thousand people doing this in tandem. And at the end of the first round, you know it, depending on how the random die tosses come out, depending on who they which in the random friends they chose, we will end up with a situation that's that's going to be most likely. Fifty one percent read to forty nine percent blue, or vice versa. I don't know which way it's going to go, but the chances that it's going to remain in balance our miniscue right. So that's that's incredibly unlikely. Much more likely is the transition towards one of the left edge or the right edge, like an oversampling and overrepresentation of reds or blues. And suppose after the first round we veered towards red. Okay, so then what's going to happen after the second round? Well, the chances that we're going to go towards the red is much, much, much higher after a second round of this, the exact same process. This there were fifty one percent reads in the audience. So the chances of converting people to read are much higher, fifty one percent, than converting somebody to blue, which is forty nine percent. So so that process is going to self amplify. So at the end of the second round we would expect, you know, making numbers up here. They're not, you know, I'm rounding things up, but the chances are that we will find ourselves in a fifty three, hundred and forty seven split. We will have gone towards read even more and we will get faster and faster towards that destination. And then the third round, of course, we go more deeply into red territory until we reach a point that we call the point of no return, and past that point everybody will commit that's safe to commit. You know that the entire stadium will be the same color. So the math in the avalanche paper shows that for very low periods of four, very low numbers of rounds, so on the order of thirteen or so, you can achieve consensus in very large networks. So, going back to our example, I think what was Marcaivo Stadium? Hundred thousand people. I think the number is something like fifteen or so. So after fifteen rounds of this, all you're doing is you're sampling five friends of yours. Okay, that's all you're doing. So at the end of the productcol you will not have spoken to everybody in the stadium. To the contrary,...

...you will have spoken to five times fifteen. What's the number? That's seven, eventy five, I think. So you will have spoken to only seventy five people. And yet you'll look around the stadium and this is the magic. Everybody will be holding up the same color card against their for it. And that's why I think it's so cool that this thing operates in a fundamentally different fashion. It's much more efficient in the way it works and and it gives you very, very strong guarantees with a fine tuning knob under the hand of the under the thumb of the system designer. You can change how many rounds you go for. You can change how many people you sample to ensure that you know whatever level of whatever probability of reversion you're comfortable with. This is equivalent. If you want a comparison, in Bitcoin you have the number of confirmations, and the higher that number is, you know it's six versus you know, five versus six versus seven. The longer you wait, the better it is in terms of what the likelihood is of it getting undone. There are similar controls in avalanche as well, so it's okay. The goal. Yeah, that sounds really actually kind of reminds me of a brown eye blue eye problem. It's a logic puzzle that's actually been featured on x case Exceed Kitty, whatever it is, and it sounds a lot similar to that kind of that kind of problem set and that solution and it is neat. But my question is that works great with red versus blue, but what if there's eighteen colors and and you know some of them qualify and some of them don't, and like if you have a stating full very many different pieces of conflicting information, number of rounds it would take to come to consensus, it seems would significantly increase. But also there's possibility for maybe chaotic states where everything kind of half the audience flips and have the audience doesn't look, you know, our I guess eventually it would kind of come to consensus, but it would just take significantly longer. There's multiple pieces of conflicting informations. That correct, not really. So the number of outstanding competing proposals is not really Jermaine to to how long the protocol runs. It doesn't matter. So imagine the same, same situation. Okay, so imagine that we had three colored cards, red blue, yellow and and instead of a fifty split, I don't know what you want to have. Let's say you want to have a one third split for each color. That's fine too. Maybe actually you just just take the following example. Imagine three different decisions. Red, blue, yellow. Let me illustrate with the with the with a very simple to see example. Imagine that yellows are one percent. Okay, in one round they will be wiped out right there will be no more yellows. So so that's that's kind of interesting. And if you start out in the one third one third, one third scenario, which is the worst case for three colors. Well then quickly one of them will will dominate. You right, it's the reds will veer over the one third boundary and and then they will again go into that same metastable process and and we'll Vier, you know, in that that direction. So in fact, the presence of more things to choose from makes it strictly easier for the protocol to decide, because the the other correct nodes get divided and and therefore the the self reinforcing stuff works even faster, for for the case of more, more, more colors. But you wi, you write about one call and let me, let me mention what you are right about. There is a there. You have a very insightful thing that that I don't want to lose, which is it is possible for this protocol. So, okay, it is not possible for this protocol to have a safety violation. Okay, so that's one of the things that's proven in the avalanche paper. What it says is, if you pick red, the chances that I will pick blue are minuscule, so small that I don't need to worry about them. It's in fact they are so small that they can be made smaller than the chances of all of the oxygen atoms in this room just migrating to the other side, leaving me to suffocate just in pure nitrogen on the side of that can happen. It is possible, right, we live in a probabilistic universe. These atoms are colliding with each other, some of them go left, some of them go right, and they could easily actually do this, not easily, but they could do this. This is a possibility. It's just incredibly unlikely. We choose not to worry about this and I as far as I know, nobody has died of a night reigion and random roll it. Probabilities work. Yeah, so, so so for that reason, so, okay, so so. Safety violations in avalanche are incredibly unlikely. But but there is one scenario, which is if you have a...

...sizeable attacker. Okay, so you have an attacker that's commanding, let's say twenty percent of the nodes in the network, then, unless you manage to break out of that initial fifty split, he can keep and and and he knows exactly what he's doing and he can see everything inside the network. He knows exactly how people are proceeding. Then he can keep you in that zone before you start to fall off on the edges. So the protocol, this particular protocol, does not guarantee liveness for for decisions between conflicting transactions. So that's a mouthful and I can expand what that means. But for normal mortals what it means is if somebody's trying to double spend, okay, and and he launches an attack with his double spends at the same time. So so that is I issue two transactions, one paying Alice, one paying Bob, by also at the network level, using a lot of resources, I launch a large attack that requires me to see what's going on. Then the paper very nicely says, look, in that case the attacker can keep these, both of these decisions from making progress. So we neither pay alice nor we pay Bob. And and then but then the paper says, look, this is of no consequence because of this other construction, that that decisions of this kind that are stuck in Limo Limbo have absolutely no effect on other decisions. They're all independent. So if I ever do this to you or to Alice and Bob, let's say, then you guys paying each other is completely unaffected. So so it's a very nice thing. So you know, at a higher level we need but impractical. Let me just complete this thought. Let me just complete the start out. You just just kind of a cute, cute thing to mention. You know, in every distributed system you're going to have some badness, like there's some impossibilities that are always going to have to be there, and these protocols like both Nakamoto and classical, you know, there's always like some badness that you need to deal with, like classical breaks down when you have bad actors over a certain threshold that come out. has other issues in this one in avalanche. All of the badness, all of the sort of the you can't get rid of this, you got to pay some price at some point, is all collected into a tiny ball and that ball is placed squarely on the shoulders of anybody who's trying to cheat the system. So if you're trying to cheat the system, then the system doesn't guarantee anything for you, and so that's good, because you're trying to cheat the system in the first place. It's not going to get the guarantee anything for you, but for everybody else who wants to to assure that their transaction got committed. It provides a strong safety guarantee and provides a strong lightness guarantee. So kind of positive situation where that might in reality become an issue. Maybe I'm wrong about this. Let's say. Let's say a coalition of the nation states want to embargo a smaller nations state. They collects together and redirect all their all their other Hash Parsons, all their all their act is a thing of the past. Yeah, it's a redirect all the all the network traffic and basically break the head. Look, this is this is from now on. We're going to we're going to block any transaction tries to go through the network from this particular set of Ip ranges, and we're going to do this on the state level. And what we're going to do is, basically we control twenty perfess of the Hash and the calculation power. I don't know what you call transactional confer validation power on the decisions. Let's go that decision power, decision power on the network. So we're going to constantly flip from red to blue every time they try to confirm, and so nobody on the network can ever validate their information. Is that not a potential like network threat for this particular protocol? Yes, that's that is a threat. So let's go through what can happen in response to that threat. So, just like the fifty one percent attack, is is a potential attack on the Comoto, I think so. Okay, so let's you know, of course a Valanche is going to have a number of attacks possible against it. So let's go over it once again. So twenty percent is a rather small number. So eighty percent can easily pull pull ahead of of people who are who are trying to do so. Case, let's just back up. Okay, let's back up and let's start from scratch. So the if you have a network where the majority of the nodes are committed to the proper functioning of the coin. Okay, so just like, just like in Bitcoin, you assume that fifty one...

...percent is honest. In a similar context here we're going to assume that the majority of the nodes are. In fact's assume a supermajority just to make it simple for the discussion here. Let's see a supermajority that are committed to to seeing seeing transactions through it. If the person doing the spending is not trying to cheat the system, these people have no chance. Okay, so that if it's me and I'm spend and giving money only to Alice, then there is nothing they can do to stop it. So this is an interesting result. It's nontrivial, but but it is, you know, it's it's it's very it's very, very, very cool. So how is this possible? A very straightfoin, in a somewhat straightforward fashion. If I'm not issuing double spends, I'm doing virtuous transactions, I'm actually just genuinely paying Alice, then that transaction will disseminate to all, let's say eighty percent good people, and eighty percent will say, well, look, I see no reason to reject this transaction. It's signed by you know whoever and and it should get to his destination. There's absolutely nothing that that twenty percent can do to counteract a group as large as as the eighty percent of the network saying this should go go forth. So so that case is pretty straightforward. Now the second case is gets to be esoteric, complicated scenarios, but again they will come down to two scenarios that have no consequence for anyone except the attacker. So if you have a group that's dedicated to to sort of creating a liveness problem, they can only do that if there is a come, there's a conflicting set of transactions. If I'm paying Alice and Bob at the same time, they can hold it in limbo and that is not that is in Lookul. Let's actually contrast this with bitcoin. In bitcoin, a minor will pick one of those two. Okay, I'll get resolved. But in avalanche somebody could pick. Could somebody could launch an attack where they maintain the network in perfect balance between Alice and Bob for a long time. Maybe. Okay, let's I'm giving the attacker a lot of power in this discussion and I'm I'm going to get to that in a second, because real attackers will not have this much powered. But but the main point here is okay, so for, you know, for an intellectually advanced audience, we have to we have to just be clear that. You know, this protocol is different. If there are conflicting transactions, it doesn't guarantee liveness. So it differs from the Comoto. But there should never be conflicting transactions as long as the issuing person is honest. Why the heck do I have the same utxo spent both to Ellis and to Bob? Why is it that I'm sending the same coins to two different people? So I should never have done that. And if I do do that, then the attacker can hold us in limbo. And if they do, then then they do. And when? What's the consequence of that? The consequence of that is felt only by me. It doesn't hold up block production. There is no such notion of a linear blockchain and avalanche it creates a directed, a sicklic graft. Yeah, I was one of the I wanted to point that out it because it's it seems as though in the case of avalanche, to type of consensus protocol that runs in parallel like a director, is aasic the graft would. And so you have the resolution of denial of service for single transactions, which doesn't withhold the denial. It doesn't have a denimal service on the entire network like you would with a block producing blockchain. Right like if you if you can deny a single double spend, it could potentially have problems. Were consequences that spread across all of the transactions the network. But with when you have a resolution that's so fine grained on the message level, those who are trying to attack the system only have denial of service issues or, like you know, there are issues with people coming to considers on that transaction wall of not having an effect on anything else. So good quality or good behaving transactions get through, but having to care about what other people are doing to deny service. That's exactly right. The system will just route around it. In some sense. Right, it's creating this graph and suddenly the graph will have a node that splits into two. And as long as the attacker is launching that attack keeping it maintained, then the system might actually be in limbo and might be unable to decide. And this again just like you said, Corey, the rest of the graph will grow around those notes and the system will proceed at full capacity and the guys in limbo will just remain in limbo as long as the attack is in progress. And if the attack or ever slows down for whatever reason, the system will actually pick one of the competing transactions and will actually resolve it. So that's not a problem. And and of course there is another open issue which the paper doesn't actually advance enough, in my opinion, and but you know, everybody...

...who's looked into it, including my group, in depth, says that there's lots of extra work to be done and lots of interesting things to examine in the following scenario. Well, what exactly is an attacker capable of any way in avalanche? So the avalanche paper says, let's let's assume an omniscient, sort of anessa style adversary, and adversary that can see what everyone else is seeing, and adversary that can look at who communicated with whom, who told what to whoever else, and so forth. That's not that's not actually realistic. Even an essay is actually an essay. Can Tap wires, but they they won't necessarily know who what the communication is. We're going to use and to end encryption in avalanche. So so so that the paper is just giving a lot of powers to to this crazy adversary, this this mythical adversary, a realistic adversary is not going to know exactly what's going on on each node it's with. Might have a statistical idea because it itself can poll, but it won't know exactly what's happening. So as a result, we're in a very interesting situation. I think it's you know, even with large, large numbers of nodes, adversaries might get overtaken. So in any case, these are all esoteric, academic sort of, you know what, diversions from the main point, which is this thing is is is bulletproof in terms of its safety guarantees and and it's liveness guarantees are also bulletproof as long as nobody's misbehaving, and now we are. We're kind of like we're going into depth on what happens when people misbehave with attackers that are trying to sort of do things at the same time. So it's a very esoteric point. I didn't want to get the audience confused about this. Almost every avalanche discussion kind of devolves into this because for a distributed systems geek, this is these corner cases are the interesting ones, the main ones, like the main usage of this network is going to be I pay Alice and Alice waits for my my transaction to pervade through the network and she pulls a whole bunch of times and at the end of that she is convinced that everybody has seen this transaction and nobody will revert and that's the end. So it's a fascinating very, very quick process. So so the the protocol in practice ends up achieving finality in a matter of a second to a second and a half for really large networks and and it has throughputs on the order of tens of thousands of transactions per second. So it's really an exciting area. It can do things that other people cannot do and my group is currently busy building a new coin on top of the avalanche, the core avalanche consensus for Lookal and and and there are fascinating things one can do on top that one could only dream of with other other coins. I had a I had a question about I feel like a lot of this is in decent terms of the corner cases we talked about adversarial settings. Is the choice and how you randomly sample your own friends to ask questions like you have to assume that that sampling is done well because you have a large enough like coalition that can basically lie to everyone else. But if you can sample appropriately randomly, then I feel like that it mitigates a good a lot of the power that a that I guess adversary would have Right. You should use a good random number generator. The the interesting thing, though, is you don't. You know. It's the random number generations up to you and you know it's I don't know. It's it's fairly straightforward in how it would work, right you you just pick any sort of any good random number generator. The one that comes with your os these days is good enough, and and then you sample according to that and that I'll get you the the diversity that you need. So it just has to be unpredictable to the attacker. Avalanche, I think, is a really appropriate name for how how fast is gross because like it take, the decision making process is very exponential. Right, exactly exactly. You start going slowly away from the midpoint and then suddenly you just avalanche down like in a giant mass. That's exactly right. So I'm still stuck on Educationes and I just got one more to get out. I know there's probably some people already addressed, but there's a backdoor and Cisco Routers and they you know, somebody ploits that and winds up being able to manipulate the network so that transactions that are posted. Aren't the transactions that were intended? With block changes like an audit trail, you kind of sign things and have intent as part of the network. Yeah, but this doesn't seem like that's the case. You can't broadcast actual intense got actually proving that you intended to do this one and only this one thing. Is that actually an issue, or am I just coming up with something that's kind of too wacky for for reality? It sounds a little confusing to me. So how would...

...you? Well, how would you signal intent in Bitcoin? Like if you show a transaction, you know that that's the transaction in Bitcoin, right. There is no there is no notion of well, I'm doing this payment, but I didn't really intend it in bitcoin. So so why should there be a similar you know what, why should I have latch do something that the Cooin doesn't do? Guess you're right. Okay, so cool. So what is the progress you're making on actually building out this coin? Ah, so that's a great question. So consensus protocols are fantastic building blocks, but they are not by themselves a coin. Right, the coin system actually requires a gazillion other decisions and and a lot of exciting new architecture on top. So at the moment I'm building a new coin. It's called Alva, and we created a company to sort of take over the the development of this and and we're sort of proceeding quite fast down the path of building a fantastic new platform. I'm really excited about this and let me try to give you the high level bits. So the AVA coin that we're building is based on avalanche and it takes advantage of the avalanche consensus protocols, protocol and it then, as a result, inherits all of the good features of avalanche. So what does that mean? Our time to finality is less than two seconds, so that's pretty good starting point. So within two seconds you have as much assurance as one would after I think the math. What did we do? I think it's on the order of fifty confirmations and bitcoin or something like that. It's the chances of it being reverted after two seconds are our minuscule. So that's that's our starting point. So that's finality. Our throughput is on the thousands to tens of thousands of transactions per second, depending on the complexity of the transaction, and that's the throughput. It's when we're green, so if there's nothing to do, then we can just sit there without consuming any energy. There is no notion of constant energy expenditure. We're not going to consume the entire energy input of Denmark just so people have some ideas, so they can visualize it. CRYPTOCURRENCY mining takes approximately two nuclear power plants and higher net output. So I'm sure everybody has driven at some point next to a giant nuclear power plant. These are really, really big, very impressive constructions, and two of them, you know, imagine all those cooling towers and all that stuff. Two of them are being used worldwide at the moment for confirming cryptocurrencies. That's way too much, way, way too much, and it's that energy could be used for anything else. If you can't do anything, smelt aluminum, you will love at least accomplish something. So so we can achieve the same level of security as existing cryptocurrencies, or in fact higher than that, and and not require a tiny fraction of the electricity that's going into cryptocurrency. So that's those are the cool features of that we inherit from avalanche itself, but there are other things we're building on top that are really exciting and I want to just get get through them very, very quickly. So number one, avalanche is not going to be a single coin platform. Yes, there will be a single coin, Ava, that will feel the whole system. That's true, but in conjunction there will be the ability to create any number of coins on top. That's going to be pretty exciting. It's so most people are somehow in this one coin on chain mentality, and and so we thought critically about there's absolutely no reason why we can't support multiple different coins. So you want to start a new coin that represents real estate transactions, that's have unique properties, you know you can merge them, if they're next to each other and so forth, then then then you can do that. Or you know, you want to represent gold on the chain, on the graph in this case, absolutely, you know. And Gold has other unique properties, like if it's in the same vault and same bar, maybe you could combine it and that. So all of those kinds of special features we could we could support. In addition, we will be able to support, and we are able to support even now, multiple scripting languages. So there's going to be a base scripting language that is simple and safe and secure that anybody can count on. But you could introduce a corey coin with its own unique scripting features. or You like, you know, let's say you like Mon Arrows on a nimity features and you like the Bitcoin, you know, whatever the Bitcoin scripting language, you could combine them. Absolutely easy to do in our language. So so those are nice. Let's see what else can you do? There are a lot of a lot of other things one can...

...do on top the coin itself. The parameters of the coins itself, of the coins themselves, are subject to governance. So one interesting thing you can do with Alva that you can't with other coins is use the the underlying consensus mechanism to pull for social consensus. And so what does that mean? So you know, in the design of every coin there will be certain numbers, certain parameters, that are absolutely crucial and that are hard to guess for the system designer. So let's take, for example, Setoshi Nakamoto. He ended up one of the parameters was how many coins shall I have outstanding? Twenty one million. Okay, that's easy, and that's that's okay. Like once you pick any number, it's as good as any other, as long as it stays constant. That's fine. And then he picked the minting schedule. So every four years there's a healvening, right and and so initially every block gave you fifty, fifty BTC. Then it was the case that it gave you twenty five. These days you make twelve and a half. Pretty soon we're going to go down by another factor of two. So how did he pick that right? So was it a good choice that that particular curve? And at times it was. You know, it ran behind demand right, which was actually good for the coin holders. The coin kind of you know, it kind of did this mooning thing where, you know, there's high demand, not enough coins, the price goes to the moon. That's okay, but at times it's too much like for a theorem. Right now, a theorem is minting way more than it should. It's rewarding those miners who are not doing all that much work for you know, with too many coins. And so so that's a problem. They had the problem in the theory and now they're trying to make a whole bunch of changes to try to change this. Well, an avalanche. You can do the following. You can say I want to push a special transaction onto the network. And this is a transaction that says, let's change the minting rate from, you know, let's say one percent a year to compensate for loss, to zero percent and if enough people agree with you, that is absolutely the kind of decision that you can make. That that that decision will be confirmed through the same avalancheing technology, the same avalanching method, and and that will be the new law of the land. And, you know, or the converse as possible to it could well be that if you have just zero reward, then you won't have enough participants in the system. Then you might want to increase it and then once again, that will require a special transaction and you know, maybe you put it up to two percent and and try to get more notes to stake and more notes to jump in the in the fray, and so so it's kind of interesting and, in fact, more than kind of interesting. It frees us from having to make many of the decisions that other people end up making that are arbitrary and are then proven wrong by just what happens in the real world. Eos is suffering from a bunch of decisions like this. They rampricing is all kind of messed up. You know, it's subject to gaming and so on and so forth. So or etherium has a bunch of other issues with it, for example the cost per per instruction. And etherium was picked essentially on on one laptop, I gather right. So people just measured how long it took to perform different operations and then they they they put different numbers, you know, different numbers, and based on how long it took on that laptop. Well, we can actually have its subject to governance so that people can have a meaningful pick here. So so these are pretty interesting. There are many other things we're planning to do. We of course want to support smart contracts and we have a very interesting technique for doing so. And at the end of the day, Alva will be a platform, will be a payments rail and a contracting rail that is much faster than anything known, much, much, much more decentralized than anything known and and way faster and more green than everything else in existence. So I foresee a time not very long from now where we end up seeing these coins consolidate. will end up seeing a whole lot of coins just throw in the towel, and these technologies are very hard to kill. They probably won't die, but a lot of communities will either pull their money out or will change their underlying technology, because what's the point of trying to maintain, you know, food bar coins of Whoo, bar watching right. It's it's much more sensible to move on to a better platform and and there will be techniques for actually migrating your utex so so it or your accounts balances too, of and by doing so you'll get a huge, huge boost. So may try and macabs of hair. BITCOIN consensus works because it takes advantage of some unique characteristics of the underlying data structure or the blockchain itself. There is no such requirement for avalanche.

It's literally you have this many choices. Everyone's signals what they want. The correct choice, the one that has the most out of that a subsets should avalanche to the correct answer despite other people trying to stop it. So it doesn't seem it's beholden to any given data structure outside of a Dag. So you can have that paralyzation of decision making. Is that it? There are there's something I'm missing. That's correct. What you said. This correct. The DAG is not the DAG for avalanche is a performance optimization. It's very different from the dags scene and other currencies, just by the way, like the way the DAG is used in Iota and the way the Dag is used in avalanche are completely different. It's a very subtle difference, but it's a crucial difference. So and and, instead of the Dag in avalanche, we could have used any other data structure that gives us concurrency. So we could have used the multiset, for example, and would have been just as fine. Except multi sets are a little hard to reason about. People don't really understand them and you know, the moment you say multi set gets confusing, whereas dags are kind of Nice to visualize. I don't know. So. So, yeah, we're going with Dagon AL, Alva and and so. And dags are very, very nice because they give us that concurrency. You and I could be working on the, let's say, north side of the Dag, while Corey is working on the south side, and you know, and other people are working on the west side or whatever. So the DAG itself essentially creates innate sharding. Right. You can just ignore some portion of it and grow some other portion of it. So charge work really well and an embarrassing Pari. Embarrassingly parallel situation where one worker doesn't have to care about what the other workers around it are doing in regards to the decision that they're making. How do you how does this work? And a situation where that's not the case and you need to know information that's dispersed across the network or you have the Ke of race conditions on those types of decisions. Yeah, of course, we thought quite a bit about this. So one interesting thing is. Let's see. So in typical blockchain protocols you have to know the entire history of the chain. That's how that the calls in this universe, in the in the Avalanche Universe, you only care about what we call the live edge. Okay, anything that's in the past, that has been spent, you could actually prune if you want it to so. So we have the option in our nodes of pruning state and and ignoring those sort of the path through the the genesis vertex. So if you are strapped for space, you could just ignore it and you have all the same guarantees, all the same security for the live edge, even without having to check it all the way back to the the genesis vortex and for sharding. If you decide that you want to keep only half the graph, then you are perfectly capable of doing so, and and and so in. In fact, if you decide, you know, to be a lightweight client that cares only about its own notes, that's perfectly reasonable as well. So how do you think? That's that's the thing that's come of throw me off about this as a how do you join in? UN You know. How do you join this? That work like I'm a note. I want to hop on. You don't need to know the entire history of everything, but how do you know the live edge? How do you know what edges are important? How do you know what's pendent? And how do you know the information you're pulling from the network is correct? So you're set up with an initial state that is right, right, calling. So there the idea is very similar. Or the attack that you're mentioning is very similar, or in fact is identical to what's known as an eclipse attack. Right. So I come into bitcoin. How do I know that the tail I got is the actual tale? How do I know that some idiot did not intercept my connection to the network and feed me a different chain than then the one that the rest of the world has fed me. And in fact, in Bitcoin there are the way they get around this is they have about nine or ten seed nodes, and so you can neck to one of the seeds and the seeds themselves give you, give you the different tails that they know of and hopefully none of them lie to you, and then you find the hardest, longest tail end and that then defines the the blockchain for Bitcoin. And it's going to be, of course, a very difficult, difficult process to try to attack this for bitcoin because you know you'd have to go to all of those seeds and somehow intercept what they do. But but let's not fool ourselves. It's just nine or ten seeds that they have right. That's what it comes down to. In in Alva it's somewhat similar reasoning, but with much larger numbers. All you have to do to join the network is find one node that knows of other correct nodes. You got to find just one correct node. The joint process is simply of going to a bunch of seeds and saying, Hey, you know, I know you are part of the network. Who Do you know? Who's...

...part of the network and then walking down that path. And if you walk down that path and you're going to find all the other notes that speak the same that was the same language, and you will find all of the other nodes that have the appropriate staking amounts, that have participated in network and you'll be able to vet the current set that you're talking to. Okay, but that nodes informing you of other nodes. So you're also depending on that node as being an honest actor. At some point, crack seed like come am I misunderstanding that? So you say you find one. No, you're like, Hey, gimme, gimme other notes. Right, yeah, had notes you any number of other bad notes. They could eclipse you, right, they could create a potenken village for you, and you think, oh look, there are like millions of nodes out here, but except they're all the same guy with the two horns and the trident in his hand. Right. So, so that's that's the situation you want to avoid, and it's easy to detect those situations because as long as as long as you can find one good guy who could connect you, connect you to the rest of the network. And once you connect to the rest of the network. You'll notice that there is a dag and as part of that Dag there exist a whole set of correct nose that have been vetted and have staked and so forth. And you know what, there is another Dag on the side that has nothing in common, or very little incommon, presumably. Well, nothing at the live edge that's in common with this particular set of nodes. And and it turns out that all of the guys, what the horns on the tridents and so forth, the red guys, that devils, those devilish people, are all on some side Dag. And and that's not a difficult process to resolve. Ah. So, yeah, that's something else that kind of like caught me about things like I ooda is like you could have part of the network, in short, based on its own value proposition, meaning that they're they diverged at some point. It didn't seem like there is any way to really you reunite in a solid way. A you know it. Let's let you. You gave the example earlier. You know you could only focus on one portion of the network as opposed to the other portion of the network. You know your carry about nor side and I'm care carrying about the south side. Now double spend would, of course, require that I communicate with the other side. But what if the rules of the protocol itself for getting divergent? Is that something that we have to worry about? That's your tie. You're touching up on two different things there. Let me let me try to pick them apart. The first last to do with iota. I'm I'm a little afraid of saying anything about iota because because one of the designers of that system misunderstands everything and then then gets up atty and goes crazy and sends that's not that even this good dumb is good, but come from beyond. A misunderstands things on occasion and and then, and then, when he misunderstands things, then the entire entire iota army comes out after me, and you know so. So let's be clear about IYOTA's situation. So what I said so far about IYOTA is I like IYOTA. The Dag idea, you know, is a good idea and and so there are many other systems that use dags and it's a good idea. In IYOTA's case. What I think you're referring to is the problems they've had with the side tangle they have. They have this this thing on the side that some notes seem to believe in and seem to issue transactions out of, and they are convinced that this is some kind of a long sustained attack. And and I'm not. I'm wondering and I've always been wondering this for like in a month's on end now, but I don't know how to evaluate it. I suspect that Iota might just have a buggy client somewhere. I don't I can't imagine that somebody's spending all that time and energy creating a side tangle. So so I think it's not really it's not really a sort of fundamental whatever it is. I think that's sort of a different thing than what they think it is. And and iota is an interesting case. It's it's correctness is still on unknown in my book and in the book of many other researchers. So so I think I'm going to leave my Yota at that. The way we use the DAG is fundamentally different and and I think I kind of would need to sit down with CFB for ten minutes to describe how it's different. To him so that he understands that, oh, yeah, these guys are doing something in lava that is not at all iota. Like Iota makes a very simple it just uses the DAG structure to encode the votes, whereas Alva does not, and that that makes a big difference in the final outcome. So, but the second thing you asked about was what happens when there's a divergence in this sets of rules chosen by the participants?...

Right, that's a big question. We've seen that question come up in other context. So, for example, I don't know what's a good well, you know in Bitcoin case, right, what happens when when your notions of what should be valid are different from mine? So those are cases when forks develop right and in fragile systems, systems that are subject to lots and lots of pre chosen parameters, every single parameters a divergence point. Every single parameter is a fork inducing change point. So you want to change the block size, Bam you have this enormous debate, bike shutting up the Wazoo. People who used to be best friends are suddenly blocking each other, people are going on brigades, people are hiring troll squads, people are organizing dragon dens or whatever, and it's just a big mess. And and fundamentally that was because your protocol really relied on a huge Tu Pul and it turned out in this case the very first item in that too, pull the block size, created a divergence. There are many other things to disagree upon and at this rate, bitcoin cash is going to figure figure those out for us. So they seem to disagree on block size, they seem to disagree on features and so forth. One of the Nice things we want to do and avalanche is to reduce the number of such divergence points, and technolog there are technological tricks for doing this, where people pay the costs associated with a change only if they are willing to go along with it. And that's perfectly okay for me to support things that you don't support and yet to be able to come to consensus on decisions pertaining to the to the same. So that's something that we can do, we believe, in a large, large majority of the cases. But there will of course be cases where, you know people, some people want to go this way and some other people want to go in the mutually exclusive direction, and so in those cases there's no choice but to create a new network. I don't there is nothing anyone can do about that. So that's just it's just a fact of life that one has to cope with. So you're going to let market forces decide, of course, so, as with anything in fact, but we're going to we want to let the market forces decide in the network. So if people want to want to have really heavyweight, very very heavyweight smart contracts and not everybody wants to carry them, then at the moment you're out of luck, right you you look at the theoreum and you say I want to deploy this thing, and then they'll just say no, right, there isn't enough gas. You can pay for this contract and you're done. But that's the end for you. Etheroreum is not your platform anymore. So that's not the same Portava. That is not going to be the same, same mechanism. So that's that's the interesting part. We won't have to turn people away. So I know we're getting towards towards of very we know we're getting towards the end of the show, but did mention spark contract? So port just being one of your goals, one of the ways that a theoryum maintains consistencies, that everybody on the network execute every every transaction, call in this particular case that that isn't necessarily going to be true correct. Not Everybody in the network is going to be executing everybody else's code or scripts, I guess. So my question is, how do you ensure that things like state management from Vm, like Ebm or you awesome, are correct across the network? Is it adhere to the same principles as simple transactions and value exchange, or is there something, some other protections that you can implement, such as bounties like true bit could offer that kind of inherent to the protocol? What's a lovely question called and that was a that was incredibly insightful and and I do want to answer it, except I'm afraid it's going to take another five to ten minutes and and also it will give away a lot of the cool stuff that we have in the works. So, with your permission, I just want to say this. We have a paper coming up in the next two months or so. We call it the coin dynamics paper, the of a coin dynamics Papor and and that's going to be the place where where this particular issue is addressed. It's a very insightful question. It's exactly the right question, which is, how do you agree, or must you necessarily have everybody execute everything in tandem, etherium style? And and if you don't, then how not right? How do you not ensure how do you not require that? And so so let me let me take that offline, if you would, and put in a forward reference. There will be a coin dynamics paper. I'm really, really proud of that paper. It's, you know, in everything I've done in my career I've typically brought this sort of a you know, like this fresh look. Right. So I don't know, people get sort of olsified and how they think about coins. You know, one coin, one scripting language, one chain,...

...you know, and you know this is how they should look. You should have your own vm. It should look like this and that and and I think my approach has been to sort of question even those basics. And so we did that with all a coin and it's really, really fresh. It does almost everything differently and and yet it gives you a very familiar environment at the end of the day. So if we do everything right, which you know, we believe we have this coin dynamics paper will end up describing a universe where youth can take your solidity applications, things you've done an etherium, and port them directly to avalanche, bite by bite. Just take the same code, run it on avalanche and and it magically works in this new setting, but with far better guarantees and and you know what, you can also have more than one crypto kitties application at the same time, which is a fascinating thing to be able to support terms of performance. So so I'm really excited and so so, with your permission, let me not answer the question and just say in two months or so there will be a coin dynamics paper that does exactly go into the great gory details of the technicals required to handle that. That question. Yeah, I actually like to have you back on the show at some point to discuss some of your progress. So yeah, that'd be that be great thing to kind of bring up at that time too. I have another I have another question here that I that we've we've been we've touched on a few times and celarily, but have it gone into the details of it, and that is all all open permission. As block chains that has some type of value associated with them have some type of incentive mechanism built in through steak. With with, you know, Bitcoin, it's upfront staking of real world energy at the chance of solving the correct cryptographic puzzle to than submit a block. But proof of steak, consiensus mechanisms. You have like on chain asset staking. So then allow us to instantives, enters, incentives. What type of thing are you using in terms of like the incentive mechanisms for people to do the right thing within avocoin? Oh, great question. Also, so you know, it takes a little while to get into to get the right foundation, and then all the good questions come out at the very, very end. So okay. So in Alva we're going to be using steak with minting. So the way this is going to work is so to participate in the network, people have to have some skin in the game. I don't want everybody, every Joe Schmo with his you know, fake sock puppets, try to participate and and potentially interfere with the network. We want the participants to have some some economic incentive. Anybody can join the network to listen, but to have decision power you have to show that you have committed some resources to the network. So that that is going to happen in the form of steak. If you wanted to have a similar, similar sort of proof of skin in the game and Bitcoin, you would have bought minors. In this case, you tie down some resources. There's a steaking period. So steak is going to be the staking amount is. People ask me how much is it going to take to stake? At the moment we have we have a name for it. It's called capital Delta. Okay, so nothing has been set in stone yet. It's just a Greek letter at the moment. And so if you can show capital Delta, you're part of the network. And so with that you you get to be consulted on every decision, on every decision, not necessarily every every decision, but you know everything, you hear of you can be pulled for so and you play a role in determining how it goes right. So so that's nice. And and the tie down period for your capital Delta coins is some other thing that people asking me about. It's going to be Tou okay, so it's another Greek letter. Just so I think I know approximately how long Tel is going to be. I think I want it to be around about about a week. So you're not going to tie it down for four months. This is not a fragile protocol that that has to have stability and commitment and you know for for four months or whatever. It's just going to be one week, so you know, or around one week. So so you commit for one week some amount of resources and at the end of that that period, you are guaranteed to get your resources back. You could have, you know, you could have attacked the system, you could have done all sorts of things, but but we are not going to take away your coins for any reason. So systems, proof of stake, systems that people are using with classical consensus protocols necessarily end up having to punish misbehaviors. So if you look at Casper, it has this notion of slashing conditions and if you ever what we call Ecuval Kake, that is, you say something to one person, say it's opposite to the other person, if you ever do that, well that's...

...really bad behavior in those systems and they have to make sure that you don't do that and therefore they have to punish you by taking your coins away, which is really a problem. And why is the problem? Because you can have software bugs. We were guaranteed to have them. There are infinitely many bugs out there right so you can have software bugs that calls you to suddenly lose a bunch of your coins, and and that is really sort of disconcerting. I would not want to build a new financial system on such a fragile or so now I like technologies where I can go to sleep at night knowing that nothing bad is going to happen to me or things that I set aside. So the coins should be safe from tampering or from protocol. You know US or use orping by protocol. So you'll get your coins back at the end of the TAB period, plus a slight reward, and that's how minting is going to be done. The amount of that reward is an annualized rate of return known as our and or row, I think, is what it is in any case. So all these Greek letters are unbound at the moment. We're going to determine what their default values ought to be. When we initially roll the coin out, they'll be frozen for some time and then after that they will be subject to governance. When I say subject to governance, I don't mean arbitrary changes are possible. So, in fact, I actually don't want or don't like systems where the system can jump from any different from one configuration to a completely different configuration all at the same time. That is not that is not quiescent in my mind, that is not that's chaotic system. Yeah, it's Ye definition. Yeah, financial systems, you want them to be somewhat I don't know what the opposite of chaotic is. I don't know if if I want to use the word orderly. But in this context, do you want them to move with some inertia? So, and you also want to understand the limits. So every one of these these limits, these parameters, capital, Delta, tow etc. They can, they are subject to change, but they should be subject to only so much change, part time period, so that people who don't like the new decision have time to exit in an orderly fashion. So it has to be a nice system to use. I suspect the feature combinations we have is just just right to sort of give rise to a pretty nice system, very very predictable, very safe from sort of a somebody who's got money invested in it for perspective, harmonized, if and the world of like like surface optimization. You say smooth and been well behaved right, smooth. Smooth is a good term. Yes, we want we want to avoid discontinuous behaviors when it comes to two critical parameters like this. So anyway. So, yeah, that's some menting is going to be done. If you stake, you get some money at the end of every every taw period and you get to restake if you like, or or take the money out or whatever you want to do with it. All Right, I think let's let's start to wrap on that one. That has been that's been a very insightful discussion on how Alva will work and now avalanche works, as it's under linkage to suspechimism. Is there any type of question or comment you'd like to, like dud like to put forth to our audience that we need get to ask you about now? I think this is great where it's living through a very exciting period in cryptocurrency evolution. I think we ended up having what a one thousand eighteen hundred different coins. Most of these are worthless weird things. Some of them are securities which are different, there's there, there's something else. But the sort of the new infrastructure is that people are proposing all but one of them, all but AVA. Are are coming from, you know, either the the classical setting or the the NCOMOTO setting. And and suddenly we have a new option. And people who sort of pushed forth the necomotive idea they ended up pushing forth a lot of narratives around it about how the network decides, how the network routes around this and that, and in many cases what they said wasn't exactly right. Right. So we've seen this time and time again. Bitcoin does not achieve social consensus right. It doesn't find social consensus. It does not. It's not geared for it. It's it's it's in fact, can easily break apart if there's a fracture at that level. And there are many other things that that are sort of pushed forth in sorts of narrative form but aren't really there in technical form. So the funny thing is for AVA, they're there like they Alva will find the social consensus, if there is one to be found. It will technically do so and and there are a bunch of other things that that it can do. That are in the sort of the same vein of you know, this got said about Bitcoin, but bitcoin didn't really do it, and here are something that that might. And so so I'm excited.

It's a new era, I think, in the evolution of these currencies. Will see a love of a year of consolidation. Next year will also see other people roll out proof of steak coins and and I think in that time frame we're going to hopefully have of a come out and and show the world what it's like to have finality that's on the order of a second or two and what it's like to have really, really, really high performance on the chain, what on the Dag I should say, as well as a sort of features that that really, you know, usurp everything else. So that really sort of supersede everything else that's been out there. So I'm really excited, me too. Oh there's one other thing, I'll guys are homes. One other thing. Okay, so this is important. I should really I should really nail this. So so we have this thing, right, of a coin, and it's going to be. It has the potential to be incredibly decentralized. Right. And what do I mean by that? If I look at Bitcoin, I see nineteen miners and mining pools. Okay, that's the level of decentralization. You know. Sure we can get into details of how many, you know, like those mining pools are really actually, you know, more people, Blah, blah, blah, and you know, and that becomes an esoteric discussion of its kind, which is, you know, yeah, they could really break apart and reform, etc. But maybe not. You know, they it's not clear that they would actually be able to detect malfeasance by the pool operator. We've seen this time and time again the bitcoin gold attacks just happened, where people attack the currency and none of the participants, none of the miners, even knew that they were being used as part of an attack. They are just not geared for it. But in any case, let's see that. To them, you know, maybe it's more, but it's not that much more. Okay, so maybe, you know there are a hundred entities out there. Maybe they're a thousand, but you know, that's that's what we're we top about at if I look at e Theoryum, I see eleven miners and mining pools defining the entire chain. If I look at Eos, for all. You know, all of the people make fun of yours for a bunch of reasons, but at the end of the day they have twenty one entities that are producing blocks. So they in many ways have more decentralization and others. So suddenly we have AVA, and AVA has the potential to have to give an equal voice to tens of thousands, to millions of participants. So anybody can stake and anybody can then participate and and anybody can have as much voice as Gehan Wu. So you know, Gehan is is great at building hardware. He's wondering. Is a brilliant guy. I've met him. He's fantastic, very well spoken guy, and and and and he's great at building these miners that go and compute these hashes. But when it comes to serializing transactions, he is just like everybody else. Okay, he puts one transaction before another. That's all he does as a minor. All of that hashing is where he spends all his energy, but at the end of the day, his contribution to the chain is as simple as what I just said. He just serializes transactions like everyone else, and I too could do that and you too could do that. And in fact I believe you two should do that and if you did, that would create a much better world where everyone is participating and defining what happened, not leaving it up to a specialized minor class. That specialized minor class creates a lot of issues for coins. I mentioned the value leak from from stores of value, but there are other problems as well, like why are there, why is there even a different class of entity in my network? Why aren't we all aligned? So so, leaving all that aside, Alva has the potential to give a voice to anybody and to have people participate in a system of unprecedented decentralization levels. So you can have millions of participants in this thing easily, and that's a fascinating world that I hope will be able to explore. Intuitively, it feels like the avalanche grows and efficiency, the more you have that leve lucky centralization, coming to consensus from the disparate, distributed local state people who have their own opinion on what's going on, is better. The further defective. The more you have that initial set right that the larger number in is, the quicker you get to the final consensus. If there seems to be like. Intuitively speaking, especially with like, the more options you have, you should also come to you should also avalanche to the correct decision quicker. I'm not sure. Well, technically, the the there is a parameter, the number of rounds, and it does grow as you have you have more nose, but it doesn't grow linearly. In fact it grows way way sublinears grows as log okay, so so it's going to you know, if you have tenzero people, it might be, you know, eleven rounds. And if you I forget the numbers now, if you have a hundred thousand, it's like fifteen Ishie, if I'm not mistaken.

Man, now that's too much. Maybe it's thirteen. I don't know, I forget, but you know, around seventeen rounds is the largest number that I've seen. If you I think that's for tens of millions, if I'm not mistaken. So yes, it grows very slowly and there is great value in having more people participate in the system. It's they're not necessarily adding to the speed. I don't think they make it faster, but they do make it much more resilient. That's right. More of a bus yeah, exactly. It's they push back against anybody who might acquire wealth and try to attack the system. It does sound like that increases network traffic quite a bit. Now they the load of an of an individual net network traffic, maybe put the load of an individual two rounds. Doesn't corrow with the size of the network. It does not. That's right, so don't it's it's you end up. You end up sampling some number of people, like five to ten, some number of rounds. So as the network grows, you know, suppose the number of number of people you sample is going to be the same. And if you go from eleven rounds to thirteen rounds, you know that's only whatever it is, it's going to be. Sorry, it's going to be twenty extra packets for you. So that's not much at all because you don't have closed loops to them. The network should it should come to goosensus. That's the thing is for you. I'm thinking on a global scale. The number, the amount of network chapter kind of increases. It's really a good mean it's probably not that big deal. I'm just kind of like, kind of curious. No, no, that let's let's getize it, because it's important. So calling it's so for classical consensus protocols. Okay, so if you were to use Lamport list of style protocols like what's used in Hyper Ledger, I'm going to try to make a decision in that in those protocols, and in every case people I speak to could be lying. So then what I have to do is ask everybody and then they have to ask everybody themselves. So if everybody has has to ask everybody else, you get you get stuck with an ND squared protocol. So and squared it seems like. Okay, well, it's just a square. Well, so bad. Squares grow really fast. So yeah, if you have a most few as a hundred, a hundred validators, then suddenly you have, what is it, a hundred times a hundreds, tenzero packets flying out there and it's just, you know, it's just a lot of communication in avalanche. I think we just did the math a second ago. If you have, if you have, you know, about a hundredzero participants, you ended up needing about seventy five messages for a decision. That's it's and if you have more like it's going to grow a little bit. But but that growth is is is proportional to the log of the size of the network. So in aggregate, so the number that's and squared for classical is all of N for for avalanche. So the total number of packers that the network must bear is just all of then. And the energy required for let's say a million messages is is a tiny fraction of of what you know, say, Bitcoin or etherium consume per second. So it's a very efficient protocol. Yeah, and if there were a morse law for new growth, and I don't know if there is, I would guarantee that's lower than that. So that yes, of course, yeah, and there is. Yes, yes, there is a more slow equivalent for network speeds. So yeah, all right, that's that's before we did avage other tangent, which we could easily do. Let's let's wrap this R up. So thank you, Dr Current Serier, for coming on the show and and helping US figure out how this works and what's coming into the future for our audience. If you enjoyed this, please claric light, subscribe, tell your friends, share it on twitter. You can find us on all platform spotify, the BITCOIN podcastcom etc. Etc. Etc. So thank you. Thank you for having me this is another thing,.

In-Stream Audio Search

NEW

Search across all episodes within this podcast

Episodes (127)