The Shelley Hybrid Era and the d parameter

hey everybody today we're doing a video to talk about the shelley hybrid phase and the deep parameter and all that other good stuff welcome to the shelley era i hope you guys are enjoying it as much as i'm enjoying it we're still kind of working our way through stuff so huzzah and hallelujah all right so we are in the shelley hybrid phase i made the prior whiteboard video about this [Music] i believe maybe mayish junish but we're here now and this began july 29th with the hard fork okay and we have this concept of d the d parameter and right now it's set to one and it's kind of a wall that gets slowly but surely taken down so the network is fully decentralized when d equals zero so what d represents is basically the set of blocks the proportion of blocks made by stake pool operators not the obft protocol so right now the network is operating in a static and federated mode so the same set of obft pools are making blocks for no profit no no money they just make blocks and this is the way that we ran during the byron era so now that the shelly hard fork is done all the logic is in place to turn on orvorus pros which is what shelley is running under and that's accomplished by setting d something less than one and that tells you the percentage of blocks that are made by stake pool operators the proportion of them so there are really three phases to this okay okay so and they're determined by hang on a second let me clean that up there are three phases and they're determined by the d value okay so the phase between d is one in point five the majority of the blocks are made by obft with a minority made by peros between 0.5 and 0 the majority are for else and the minority v are t and when we hit d equals zero these obft nodes i turn them off they never run again huzzah pretty amazing okay so this time period here is really a major validation that the network is healthy the stake pool operators are actually making blocks and i will see an increase in participation so you're going to see more spos the rate of spos is going to increase the amount of stake being staked is going to kind of hit some sort of threshold so active stake will start going up and probably get asymptotic to some level maybe 45 50 percent see a big kick in the beginning and then as exchanges get liquidity more people come hopefully we get more than 50 but most networks sit between 25 and 60 as a baseline you never get to a hundred because there's cold storage lost coins people don't wanna participate regulatory reasons in certain countries and other things there's friction in that so you have more stake pool operators you have more of this and there are other metrics to follow and you can kind of aggregate these into a concept of network health okay all right so this is kind of a shaky time where everybody's learning what to do and moving in and we'll have an easier time than most because we had the itn and the pioneer test nets with haskell so we should be able to get through this phase in short order the book end of it is that there's another hard fork plan by the way we don't call them hard forks anymore let me hide that we actually call them hfc events ha ha because we have the hard four combinator and we can run both sets of ledger rules so it's something unique to cardano we don't have to do hard forks anymore so we have an hfc event planned right around here and that's going to start implementing the gogan style features so native assets and pluto's foundations and other things and during this time frame we have another thread that's running and that thread is all about peer-to-peer so marcian and his team is starting to turn on the peer-to-peer governor that was pre-installed in shelley and they're doing performance profiling and policy to load up that just turn on all the peer-peer capabilities so as we're moving from d equals 1 to d 0.05 kind of during the majority obft phase we're turning on peer-to-peer and getting ready for that hfc event to turn on some gogan related features then post that as we're moving into the majority prairies phase there's still a minority on the obft and so there's some disaster recovery there and the majority of the network is now being run by the state pool operators and the bookend here is a little bit around d equals zero a a little bit beyond because it'll take some time to prime this we're going to replace the update system with a fully decentralized update system the update there's that dyslexia coming in d update system and voltaire so the bookend at the end of this phase is basically having a voltaire hfc so we take all the lessons that we learned from voltaire running as a prototype and the dc funds and the improvement and voting and higher participation there and basically we roll them all up and then put them into a hard fork to fully integrate the decentralized update system and the treasury system into cardano and this is really the book end of the cardano 2020 project basically everything we did from 2015 to 2020. at that point the system is fully decentralized 100 percent of the blocks are made by pros the update system is fully in the hands of the community and cardano has smart contracts and native assets and it's it's running along and it's a beautiful beautiful thing okay now there's an open question of how long should this take okay so to get through all of this you could you have to put some sort of drum beat that gets you there and we'll call this alpha and it's a constant decay parameter and what we're going to do is we're going to have a blog post on friday the 14th of this month that's going to set an alpha value alpha equals something and basically what that means is every epoch d is decremented a minimum of that amount okay so by knowing alpha you'll know the maximum amount of time it's going to take to get to d equals zero now we also have this concept of network health and you can kind of look at it something like green something like amber and something like red so if you have some really nasty conditions going on you can stop the constant k if you're in the green the constant decay just continues and maybe you slow it down a little bit amber so we'll try to come up with a nice parameter for everybody to see here as an aggregate of the amount of active stake and the amount of stake pool operators and if they're making more than 90 percent of their blocks and so forth but right now it looks like everything is going to be green people are very excited there's a lot of competency in the core group so we're going to set a constant delay decay parameter friday the 14th and that will give us a good time horizon for when d equals zero so a straw man parameter would be something like .0025 and if we did that that'd be about 200 days it would take to get to d equals zero perhaps faster but this is the minimum constant delay meaning that it every epic will decrement at least by that much the better the network conditions the more confidence everybody has that things are working in the right direction there's an acceleration of that decay network conditions are for some reason really nasty and things aren't looking good especially in this phase maybe we pause the decay for a bit but we would like to have a constant decay parameter so every five days you see that occur so we're going to write a nice rollup of a blog post that talks about all this and has pretty pictures and kind of explains what network health is all about and explains how we're bookending these two hfc events and so forth that will give a little bit more clarity but i hope this helps a lot about explaining the phases in the hybrid phase you have the majority obft phase that we're just about to enter we have the majority perils phase and then beyond full decentralization at this point 100 percent of everything is run by the spos in the community including even the update system for cardona we need the update system to ensure that we can deliver the hard four combinator events in order and once those are delivered the system is self-sustaining completely okay so that's the hybrid phase now we had a lot of discussions this week about the d parameter and whether we should lower the d parameter next zip buck or not yesterday we released a communication that said business as usual we're going to lower it after some more conversations today specifically about the exchanges we've had some anomalies and issues with certain exchanges in particular there's some emergent bugs that have come from how they're using our software in virtual environments that have made it really difficult for us to replicate those bugs on our side but they're apparent on their systems so what the back end team did and the benchmarking team did is we created a custom version of cardano for those exchange operators to get additional logs so that we can figure out what's going on in their environments that prevent them from migrating this is about three or four exchanges of the 15.

now other exchanges like binance and recently today bitfinex obviously don't have these issues and have liquidity again but other exchanges do and some of those exchanges do have a lot of data found at them so given that we had the presentation bug where one-third of the stake pools were not being properly displayed given that a large amount of ada is sitting on exchanges still and people want to get it out but there's going to be a few more days added to that transition for people and given that we've had a few other bugs along the way that have reduced competitiveness for state pool operators we are now in a situation where i think it's prudent that we delay the decrementing of d to one more epoch so instead of having it occur next week it'll be the following epic just so that we have a more egalitarian participation i understand that this is a bit unpopular and that some people may be upset about this other people will be very happy about it it's a controversial decision but ultimately i have to make decisions like this for the good of the ecosystem and we have always when faced with hard decisions aired on the side of the community and aired on the side of fairness and it's just not fair if one-third of the stake pools were not properly visualized it's not fair when a big chunk of aid is still sitting in exchanges and people are waiting for liquidity to participate to start the system without them we don't want to leave some people behind so we've already made considerable improvements we resolved the daedalus bug we have made a lot of improvements in certain things so that most of the exchanges are now on the path for full liquidity and we've increased some of the software logging and monitoring for those who are still having some difficulty transitioning it's a very difficult thing to get all your exchanges online principally because there's more than 15 of them and they all have different infrastructure languages standards different environments some are very transparent others are not in some cases some exchanges they just ask random questions and we ask well what's the context we'd love to help you and they don't reply back other exchanges have been incredibly open and communicative with us and they have been willing to share exhaustive log details and infrastructure information which gave us the ability to help them directly so there's a spectrum of collaboration and cooperation there and where and when we can we do and unfortunately because some of these setups are very exotic they tend to introduce exotic bugs or exotic things that we can't replicate ourselves without more context so we can't fix them because we can't replicate them so that's just the teething pains when you go through a major transition like this and we redesigned the software so that after they get integrated we don't have to go through these types of teething pains again that was the point of the addressee project but we are where we're at and the reality is we have to give it a little bit more time to see all those guys go green no one's no one's going to take a break or go sleep until that gets done so we'll make sure that that gets done there's also the issue of uroi a large amount of aid is still in that ecosystem and sebastian and his team have been working overtime to get full uri support so you're always back online for byron and shelley support is coming and hardware wallet support is also coming on their side we've finally gotten the legend treasurer into the fold and firmware updates have been issued the users of eight of light are able to to use their ledger and tresor there and looks like uroi will be soon for that capability as well so a little bit more time we're not going to decrement the this window we'll do it the next epoch and then we're going to publish that blog post on february 14th which will contain all the details we'll set the constant decay parameter so that you guys can see roughly when d is gonna equal zero we're gonna be in the majority obft phase for a little while but not a huge amount of time and then it'll be bookend with an hfc event which will go ahead and get gogen out of the way at least part of gogan and then we'll do another hfc event at the end of the decrementing so around or a little bit past t equals zero to pull in all the voltaire capabilities at that point disaster recovery is no longer possible so whenever you build a high assurance system and you build a system on completely new technology you always build layers of fail safes into that system just in case so the d parameter technically can move in either direction it can go down or it can go up the only reason it would go up and it wouldn't go from 0.5 for example to 0.51 it go from point five back to one the only reason it would go up is if there was a catastrophic flaw in the design or bug of orborus or more of a completely new protocol and we weren't crazy enough to just throw it over the fence and say good luck everybody we built a backup system just in case and that's why obft exists so we kind of march our way through it in a systematic way to get to d equals zero but we do have a disaster recovery mechanism in place until d equals zero and the update system is decentralized now once that's removed the system has to be sustainable there is no fail-safe there is no backup system it's it's basically completely decentralized at that point but this is how we build software and this is how sensible reasonable people build software who have a moral responsibility to make sure that the things that they do are right so you do it systematically and you do it with an increasingly larger and larger set of people if you look at this layout we go from here an operating model that could run indefinitely and we're all comfortable with to increasingly more uncomfortable territory where suddenly more and more of the network is made by brows to a point where none of the network is made by opft nodes now along the way we're going to discover all kinds of things performance issues slight bugs here and there configuration issues hardening of the software to prevent vectors of attack perhaps improvements to the network protocol that need to be made for certain things that are reoccurring optimizations here and there and as we walk that journey together it means that when we get to this point we have no uncertainty we feel very confident in the things that we've done and we believe that the things we've done are going to stand the test of time and that is process is what allowed us to even get to d equals one to begin with now what's so cool about this is in the broader scope of things this is actually not a lot of time as i mentioned 200 days if we're at .025 as an example that means in the february to march time frame we'd be looking at for d equals zero i'm also going to point out that whether d is one or zero this doesn't substantively impact the profits of staking the same amount of money is paid whether d is .