How we will launch Shelley

hi this is Charles Hoskinson I'd like to make a video tonight to discuss the launch of Shelley this is a very complicated thing yeah it's not like launching a normal cryptocurrency and it's not like launching a normal hard fork the reason being is that we're going from a federated model to a completely decentralized model and that has to be done in a very systematic careful way and so what I wanted to do is make a video and try out my surface here and do a screencast presentation and kind of walk you guys through with a white board a digital whiteboard of exactly how we're gonna do this so let me just clear the canvas and let's talk about it all right so in September of 2017 let's use black I get to be Bob Ross today okay in September of 2017 we released Byron and Byron runs in what's called a federated mode so federated means that there are collection of core consensus nodes and these nodes are basically responsible for running the network meaning they make blocks so since the beginning those nodes have been operated by three entities the Cardinal foundation of ergo and IOH kay and then when people install Daedalus Daedalus would be a full node so it would have a full history of the blockchain and also all the ability to validate the history so nodes would communicate with each other they'd send transactions to each other those transactions would be made into blocks by the federated network and then the full nodes could validate that so there was this beautiful interdependency between the users of the system and the core consensus nodes of the system and technically we could run this system indefinitely if the three entities wanted to keep doing that or we found other people to do it but the whole point of Cardinal is the centralization and one of the things that we've done is we've invested about five years of research into building a class of protocols living in a superset of protocols called proof of stake and basically the idea is that we want to go from a federated world this world here to a dynamic world so this is static and federated and we want to go to a dynamic and decentralized world okay so we had to figure out how do we go from the biron design this is Byron to the Shelly design where we have a dynamic and decentralized consensus algorithm okay so the first step in that was just seeing if we could actually get a group of people who would be willing to run card on outside of that federated set and that was the point of the incentivized test net and that was launched in December of 2019 and basically the itn really is its own cryptocurrency it's peer-to-peer it has staked pools and as delegation it's a full version of Shelly in a sense the difference is though it doesn't have the Biron history and so the idea of the ITN was to get data on some parameters in particular we were looking at things like how much ADA is staked and I think we're approximately around 12 billion here which i think is around 40% of the total supply and then we were also looking at things like how many state pools we were looking at things like Network quality so basically is a network stable is it up when people here are trying to send transactions are they posting and validating quickly are we making blocks and with the expected block schedule these types of things and here's the reality the ITN was an overwhelming success our decentralisation target was instead of having three entities to have 1,000 or as close to a thousand stake pools as we could get it's a pretty big difference three to a thousand and we've seen over a thousand register and I think around 300 or regularly participating making blocks which is a tremendous start tor it basically a test that that has some incentives behind okay so first step was to understand this model and make sure that lots of full nodes exist and we are there thousands and thousands of them exist the next step was to verify that there's lots of people who would be capable of running the network and we verified that with the incentivize test net and then the final stage is to begin a phased transition from the Biron reboot to full Shelli so let's talk about how we're going to do that phase transition because it's currently underway so step one was to get everybody on to byron reboot and built into byron reboot was basically all the bridge code to start the process of upgrading to Shelli so we launched this March 31st of 2020 and built within it is or Boris BFT Oh BFT you're going to see this term a few times okay so what's gonna happen is we're going to go from or Boris BFT Biron but still running in that federated mode this this federated mode up here these guys here okay we're gonna go from o PFT Byron boo there we go and then we're gonna go to basically Shelley upgrade period or phase then we're gonna go from the upgrade phase to Shelley hybrid phase okay and let's walk through what's happening in each of these all right so with Byron otft that's what you guys are installing when you installed Byron reboot and that's gonna be available next week for the main Daedalus client flood clients 3 was just released today basically Byron otft is gonna run for a little while and in parallel we're gonna start releasing a series of shelley haskell test nets and there's really three of them so there's the friends and family shelley Haskell test net so this is internal test net that we're starting to turn on and some of the more prominent state pool operators people who really did a great job here are going to be invited for an alpha test and the goal here is just to verify that everything we think about the node and the command line and these things are working correctly and that's not going to run for a long time that's basically just a burnin before we do the actual node plus command line shelley test net and this is really the the first opportunity for the stateful operators who participated in the itn to start rebuilding and redeploying their infrastructure one of the reasons why we're doing an internal alpha before doing this is that a lot of the people who participate in that curl alpha will help teach the remaining state pool operators and give them advice on how to configure we notice that there's a lot of cross-pollination in the ecosystem a lot of people participate on telegram and so we felt it was really important not to slow down our ability to deliver the product to make sure that we train a group of people who could then retrain other people so we're gonna do a forensic family then we're gonna do a Shelly test net in Haskell and that's a node with CLI and then finally we have something called a balance check and basically what the balance check is is it's everything it's gonna be the node the wallet Dedalus and it's going to contain the byron and the shelly consolidated history so everything together so effectively that looks a lot like what Cardno is supposed to be when Shelly is turned on and the primary validation of this is going to be that the ITN rewards plus the maintenances are consolidated because what happened is that some people participated in the ITM approximately 12 billion ada give or take a state in tens of billions of dollars worth of rewards been paid out for that test so those have to be consolidated with the existing and remaining maintenances and so that's what the balance check is all about when the ballot check is done this also is going to be end of life for ITN so the ICANN will be shut off at least are part of it it's already decentralized enough if it wanted to to run for longer because it's practically a cryptocurrency in itself but anyway no more rewards will be honored at that point that consolidation is going to happen after we've reached the end of life so balanced check will be the last haskell test net and basically the point there is to make sure on the consumer side that all the rewards are consolidated and also this is going to be where the exchanges will be given a full listing experience so basically that means that all the libraries and all the things that we've been planning for address dia will be done ready to go and if they integrate against this then they get some certainty that when the time comes for Shelley main net that all their integrations will not be broken though they'll be forward compatible from that day okay so after the balance check runs for just a little bit then what's going to occur is we're going to enter the Shelley upgrade phase so what that means is that if you go to the card on a wallet websites and card on Oh dot org and all these other places and those are being rewritten in anticipation of the launch of Shelley or Daedalus wallet and then other providers for example providers like Lloyd providers like your ROI basically what you'll be able to do is download a new wallet and that wallet will actually be Shelley you'll have all the logic for staking for delegation in the Daedalus user interface you'll actually see the state pull registration Center all of these things will actually be there but will still be running in Biron meaning that we'll be in a federated consensus and it's going to be running Oh PFT okay so there's going to be Shelley clients we're going to replace all of the core consensus nodes with these Shelley clients and then you the consumer can download the new Shelley wallet and that's going to run for a little bit to give a nice upgrade window because the exchanges have to upgrade and we also need make sure that people upgrade in the network and then we're going to trigger a heart Fork and once that heart Fork is triggered all the Bayern relays are shut off and that means that no Byron code is going to run anymore Byron's done so if any exchange is still left on Byron the wallets would be locked until they upgrade and at that point no more Byron blocks are going to be made and that means we've then entered the Shelley hybrid phase so let's talk about that this is the last phase so we have this thing called a deeper ammeter in our system and d4 decentralization and basically when we've entered the hybrid phase Shelley is going to have blocks made in two ways that Federation that we mentioned up here is still gonna make some blocks but not get any rewards so they're making blocks for free almost there come on okay great so this Federation is still going to make some blocks and it's going to make it with Oh PFT however because everybody has Shelley clients now and everybody can understand Shelley clients stake pools can start registering and operating and what happens is that when we break an epoch down so that's a period of time where blocks are made some of these slots are going to be made by the state pools and some of these slots are going to be made by those core consensus notes and that's controlled by that D parameter and it's ranging from 0 to 1 so in the beginning it starts at 1 which means 100 percent of them are made by the core knows then epoch by epoch what will happen is the will gradually decrement so it gets smaller and smaller and that means more and more of the slots will be made by the state pools that are registered so there are no limits to the state fools that can register that K parameter of course we'd it to be a thousand but technically you could see five thousand register there's going to be competition there's going to be delegation and regardless of what the D value is set one hundred percent of state pool rewards will go to the state pools operating on that so we're going to do some video series and lectures explaining exactly how these work and these triggers work but basically D is in either-or situation so one of two things is going to happen either we hit some metrics and those metrics are gonna be things like if we go up here the amount of ada state metrics like network quality and other things that we were measuring with the incentivized test net and if we hit a collection of metrics that we feel really good about we say okay it's so decentralized we can just shut this off completely and let 100 percent of every epoch be made by the sync rules or some amount of time regardless of network quality so there's no way we sign up for an indefinite commitment basically this is just a phased way like training wheels effectively to gradually open up block by block to centralization to the system and it's a control valve because while we can move deep down we technically can also move D up in the event of some catastrophic event a bug or that the network parameters don't look we're like what they're going to look so giving some numbers epochs they're gonna range between four to five days we're still working out the final parameterization of how long we want to do an epic because if preyas works a little differently i and if for example were you decrement d by point ten per epoch then this would be about two months an example or we could do it faster like 0.