# Ethereum Roadmap 2020   Eth1 to Eth2 Transition Devcon5

- Channel: [Devcon](https://streameth.org/devcon)
- Date: 2020-10-09
- Duration: 37:48
- Topics: blockchain, smart contracts, distributed ledger, ethereum
- Watch: https://streameth.org/watch/yt-wfB5R0gu9HA
- YouTube: https://www.youtube.com/watch?v=wfB5R0gu9HA

## Description

Visit the https://archive.devcon.org/ to gain access to the entire library of Devcon talks with the ease of filtering, playlists, personalized suggestions, decentralized access on IPFS and more.
https://archive.devcon.org/archive/watch/5/ethereum-roadmap-2020-eth1-to-eth2-transition

The Ethereum Magicians give an update and a Roadap for Ethereum in 2020 and beyond, looking ahead to the transition to Eth 2.0

Speaker(s): Ethereum Magicians
Skill level: Intermediate
Track: Consensus layer
Keywords: eth2, developers, devex, technical

Follow us: https://twitter.com/efdevcon, https://twitter.com/ethereum
Learn more about devcon: https://www.devcon.org/
Learn more about ethereum: https://ethereum.org/ 

Devcon is the Ethereum conference for developers, researchers, thinkers, and makers. 
Devcon 5 was held in Osaka, Japan on Oct 8 - 11, 2019.
Devcon is organized and presented by the Ethereum Foundation, with the support of our sponsors. To find out more, please visit https://ethereum.foundation/

## Transcript

we're about to kick out the eternal one /t theorem to transition and I think that Danny is the right person who will introduce you the topic yeah I'm Danny I work at the theatre foundation on a lot a to stuff the question of what the existing theorem chain does and is in relationship to the theorem 2.0 has been kind of an open question over the past year as the etherion 2.0 protocol has emerged as people have begun to work on it for a while it was maybe be rolled in as a smart contract within one shard that was also floated maybe it just exists in parallel forever and we let it die both of those one of those have seemed kind of complex and the challenges were not so clear how to solve them and then the other it just letting it exist forever kind of didn't feel like the the humanist approach as Casey likes to put it it didn't know all of these humans in this room work on the etherion protocol so it's been a priority to figure out a better way to slot eath one into eath two there's a lot of conversations this week around execution execution environments and it turns out this like very highly abstracted execution notion of execution might allow us to more easily slot eath one into eath two as its own execution environment but it's a lot of things to figure out what that actually looks like and then more importantly even if you do that we now have this we essentially we have all these developer interfaces on how we interact with the existing theorem chain and one of the big upcoming challenges is how do we take all these developer interfaces and once we've made this prior to making this migration how do we kind of rework the internals of these exposed very similar if not the same interfaces such that the developer experience is fluid and continuous through that migration it's definitely the this is kind of the lay of the land of the problems it's the technical problem of how how eath one becomes any if during that transition what we what things we might choose change there's been talks about changing the Merkel tree structure how certain things like block difficulty and things like that are translated some of these old op codes and things might change do you open up the sharded universe to this de and and expose new things to theorem one within a theorem to a number number of other challenges like that so it's kind of the technical challenges of that migration and the developer layer challenges of masking the complexity and kind of like shifting the entire community over is that the kind of stuff that we wanted to talk about today okay cool way is going to introduce yourself it's so it shall we found some foundation research thing and I've been working for joining for a while and now it's two okay hi I'm Alex Stokes I work on Trinity which is the Python client sponsored by the foundation focused on 82 I've also been looking into some of the research around the finality gadget which would be one stuff where we start to kind of merge the two systems looking towards a longer term transition I go by my online tech presidenta and I've been picking up more and more of the testing work for a firm to just respect across in the world yes hi I'm Adriene Sutton I work for Pegasus a part of consensus and I've been working on our each one client which is basic and also our Artemis client which is - hey I'm well I'm in quote have been doing phase to research and been working on various execution environments and pieces like that hi everyone yeah tech coming from status research where in this context working on a theorem one and then the theorem to client both of incoming hello my name is Trent I work with white block which is a network protocol testing platform hi my name is call me theorem research and I'm mostly working on migrating validators over to e3 at the moment Oh Kiki I'm not a researcher I'm kind of just like an angel investor and I also work for circle my name is Alex and I work in the Russia forest intake Association right now and we are looking at the theorem from the compliance point of view and developing this software for the national standard cryptography hey everyone my name is Victor I'm over at bison trails we help people run nodes and so we currently support one point ax extremely excited for 2.0 and validate hi my name is Joseph I am an engineer at go packs Korea's crypto currency exchange I'm John I'm an advisor to go packs hi I'm Justin Drake research at the firm foundation mostly working on Eve - I'm Felix again I worked on eath one so Jason Carver working with the etherion foundation on the Trinity client more on the eath one side Vitalik you probably know about me because I already introduced myself on the previous panel and one other component of this migration that I wanted to mention is that ease and execution environments which is the notion of state instead execution in e2 are stateless so there are depending on the e you'd have to have some notion of state in a separate piece of software or something like that so the theorem one clients then likely become that interface to the eath one he e in terms of the state and provide our state and sync and things like that so there's also this component of how death parity another mind nimbus things like that that piece of software integrates into this new stack to provide services to the eath 1e e I'll just start with a very basic question what can regular users expect from the transition how would they be impacted in any sense in the short term not and unless you personally choose to take your 32 ether and stick it into the deposit contract and start running an Easter node and become a validator and and if you do then you'll be able to start experiencing the wonderful world of ease to early and if you do not then if you do not then nothing will happen eventually if there will be some form of eath ones to eat to transition and with the kind of a road map as exami suggested of eath one transitioning into being in e2 execution environment it's theoretically possible to write the software so that you experience like very close to no disruption at all like though the two kind of the worst things that might that might happen one is that the chain might have to like halt for some period of time to make this transition go smoothly to allow the merkel roots to be recomputed according to some more efficient tree structure and so on and the second thing is as I mentioned in the previous panel gas costs for op codes that touch state are going to have to be increased massively both for you through compatibility reasons and for kind of other scalable a scalability reasons as well and so some applications might have to and have changed the way that they're implemented in order to contain and to be optimal so in that and I I mention it explicit at the beginning the challenge and a lot of these challenges cannot necessarily be tackled today the research around phase two although is progressing quite rapidly especially compared to where we were say six months ago we're not at the point where we're interfacing where we're actually dealing with kind of like the user level and developer level interfaces into these things but when we do cross that road say early next year it's going to become extremely valuable and important for the maintainer zuv the developer interfaces to web three libraries the the developer all the developer tooling to begin to collaborate to start to define these interfaces and kind of like realign these interfaces to plug into eath - so early mid next year I think forming some sort of group that of people that do work on these toolings to kind of get get ahead of the curve because it's gonna be important for migration I had a question on the shifts to proof of stake are there any specific points in time or they're really events that are important for this transition in terms of like maintaining the security of the network if it becomes more secure than secure at some point and what the parameters are you know so the parameters that probably matter are the kind of the events that you need to care about one is the time that the deposit contract gets published and so if you're a validator you will be able to kind of pre join the other is the time when the chain launches and basically if you're a valid air you should be sure that you have a client running at that point and when the chain launches that you'll have a proof state chain running the next major milestone after that would be the phase one switch and the phase one switch is important because that's the time when shard chains will turn on and so that's the time when especially if you have a lot of ether your computational requirements will increase like because your computational requirements will basically be proportional to the amount of ether you have the more you have the more value doors once you have and some of the more shard blocks you will be called to make a cross link on and then after that there will continue to be Forks but kind of the general way the chain works will continue to be fairly similar in and then after that of course these one two if to merge these also which haven't been fully figured out yet so in that roadmap we still don't have a way to exit from being a navigator that's that's a good point this is something I missed is one or two not fully this not fully decided if will become transferable the kind of default roadmap is that it will become trance but it will become a transferable and more and more people just wants to have their eath on the east suicide I mean if desired there's also of course the alternative of like making an exploit that eats you back to eat one bridge which of course I can absolutely be absolutely be done but it's something that thick is extra work as well so yeah I'm thinking of the scenario where sudden where you buy in in phase zero and suddenly you end up with lots of more responsibilities in Phase one that you realize that you cannot can no longer support so first of all you can muster oh and you can just have your eats a sit there and what when transfers are repla mented then you could also just sell your ether to some to some well or just anyone there is probably going to be a bunch of like basically if the thing is right that if the price of kind of eighth on the east suicide kind of goes below the price I'll be fine on the east one side by even a tiny amount then everyone who would have joined by joining through the normal process would instead wants to join by by just treating the wreath one for a little bit more to and then joining directly from the ether side so there's this kind of natural demands pressure that you would be able to turn to or kind of trade into and you would in order to convert your Eve to back into if one side if you want to and if this natural pressure is too weak and practice then you know it's possible that the beacon Eve will go lower but the bigger the price discrepancy the more of an incentive it is for us as a community to build a to a bridge so in the way it's kind of self-healing because the possibility of the bridge will reduce the gap the suggestion was to commit to doing the bridge and then kick the can down the curb for as long as possible I so some of the like peripheral peripheral work that we can do to unify these systems and kind of give value to the systems out of out of e to anything on I think the decision on whether to actually do this work is probably contingent upon where we're at with eath to add any points so I I mean we can commit to if we if we need to do it we'll do it but more importantly I think kind of assessing it's always like it would detract it would it would take work and take effort and take you know focus away from shipping eath to the more that we kind of income encumber the two systems together and we have made a commitments because like once we have the native integration that is a two-way bridge I guess the question is do we have something in between which is a high latency to a bridge as opposed to a really fast native integration a hypothetical question if it's if it's gonna cause a contentious hard fork on each one to make a bridge from east to back to ease one is is it still something that that you would so okay listen are you here I'm thinking is that okay ease to happens and something really bad happens and a bunch of extra ether gets created and then this kind of price discrepancy happens and there's a demand to make this bridge to return the path to the s chain so the nice thing about phase zero and one is that it allows us to architect the system this complex in many respects system in a production context but without having users yet so if some underflow happened and we had 20 billion Heath created we'd likely fix it the because it's an early system there's not that many users you know users are validators and that's one of the that's one of the reasons that transfers aren't enabled necessarily in phase zero but there's a class of books where it's not explicitly a problem right now you can't have this kind of misaligned incentives or something and there can be either created and some people say oh this is totally fine this is part of the protocol but other people say no this is a serious problem and it's difficult to tell the difference between the two not in a way that would generate a bunch of ether there potentially slash like may like if half of the validators that got slashed you could see that being very contentious right if it was some sort of like bug in the software or things like that but 10 times the supply of ether being created it is like explicitly not an intended portion of the protocol so but III hear you the more that you encumber the systems the more that it becomes more difficult to for coming over one or the other or the more that pull it politics over here might affect politics over there when you're thinking about designing a bridge how much of your consideration is around the work that it will take to implement it versus the like social construct of like being bought in and having no way to leave I my personal opinion is that like uterus paribus I think like having a social like having it beats you way is somewhat better than not having it be two-way the reason we're not into like enthusiastically going toward implementing the bridge on day one is because it would Jim Utley is a lot of a lot of work and if it and in the happy case that's work that we expects to never need but I know people's people's opinions on this my on this might differ right and for better or worse each one takes a lot of time and effort to fork and get things in and so the longer that we can operate in eath to very rapidly and can iterate quickly the quicker we can get things out and kind of get to the vision the sooner that we choose to tightly couple these systems the harder it is going to be rapidly iterate and get out these phases of production so definitely a trade-off there I think the other aspect that I quite like and you know at risk of disagreeing with italic is there's a real incentive that people are excited about it too and and so I want to get in and stake and there's then a real incentive to actually make it work and go all the way and get main it over to it if that's the only way you can then really use that if you just putting in so kind of having it locked up is is nice in some ways it makes it a little harder to sell people on the idea of becoming a validator but once you manage to get them over that hump it's kind of downhill in terms of getting them to support moving forward which is nice and if we want to go fold or die then we can change the deposit contracts to instead of locking the youth forever it becomes accessible accessible by the Dow hacker on January 1st 2027 just given liquidity alone can't we expect a liquidity discount on the e to ether and if you're one to one burning your eighth one ether and then creating one e - yeah I mean a decision is made there and it's based off of how much you're expecting to generate in new e - as being a validator but granted it's a very complex right so you might you might expect some sort of discount for some time horizon but if your time horizon is longer than that with which the systems are merged and unified then you don't really need to necessarily you don't externalize that that discount for the time so I do expect early validators to be hobbyists enthusiasts long-term holders and I expect them I expect them to get probably higher return than when these systems are unified and there's more institutional players and a lot more people kind of willing to accept a different risk analysis and so honestly I hope that it's a boon for hobbyists I hope that it's a boon for people that are like deeply invested not just finding it financially but like invested in the technology and believers in technology when what are the right conditions for the eath one folks look at using the finality gadget like implementing the finality kitchen what was the question what are the right conditions for us to start talking about implementing the finality gadget back from the beacon change maintain conditions technical conditions or there's probably two things one is one is the technical and two is [Music] understanding like better understanding the time horizon of phase say phase two if phase two is a hundred years from now we should implement the finality gadget if it's two months from now we probably shouldn't and then somewhere in between there's there's a line it doesn't have to be like we as in everyone right like a finality gadget could like the Roo I think one of the reasons Jason's asking is that is doing the Python client and Python is slower than other language is that executing history and so if you got a fitting it finality gadget you will need to check like five hours of history which is really nice you know yeah yeah I think you definitely be a not being I know optional and of white client type thing that you can do question so going back to the question about I guess they have the first two phases before or like a with regards to building the I guess either one-way or two-way bridge between eighth one and two the apologies if this is orthogonal but within the risk analysis have you guys considered let's say some centralized exchange just basically say create or allowing and facilitating and encouraging their customers basically deposit their balances into the deposit contract and basically creating this like weird futures 2.0 like market before there's even like on chain liquidity and like if you guys considered at or is that just completely orthogonal to the rollout I guess like actually one question with that is that like if there was a kind of futures market for if two then like why would anyone I guess like voluntarily participated in that construction because I give that if the price is if the price is lower than one then you'd see why people would want to buy but if the price is lower than one then like why would you even want to kind of move your eath on to be suicide and lock it up unless you're also going to be a roof Staker I guess like at some point I think our transfers on the easier side are gonna be enabled and so like up price between the two sides is going to exist and like I do think there will be I I do think there will be a lot of pressure to kind of push that price code to be very close or equal to the one level any other questions can you please come over to the front because we have cables and we can like have there been any considerations of like test implementations for some of that we're testing the the ecosystems whether they're like a deterministic test net or possibly like a like a confrontational test net where you know actors are incentivized to try to disrupt the network as much as possible or just any any ideas around that front these things are certainly on the horizon and after the Interop workshop where we're kind of doing initial interoperability testing that is definitely the next phase although some private more scaled out networks we're probably on the immediate horizon before we open it up but I'd certainly like to do some adversarial stuff I've had a couple conversations with some guys at cosmos to hear about their experience over there for better or worse we have a lot of clients so some of the complexities of doing that and managing the complexities of many implementations of software in these public test nuts gonna be interesting in equivalent to what we did with Olympic in 2015 with incentives to dose the hell out of the thing would be great hi my name's dan from corn base I didn't hey I'm curious what your what your mind set is around leveraging exchanges and other entities to migrate 'if 1.02 to point out in terms of adoption I guess like the first question is kind of what role exchanges would have right because like the default process for moving eath over to become a validator doesn't really require any kind of special third parties I mean I guess like in the case where there are if two people that wants to migrate back to the eath one side then there's these one people that wants to a deposit then you could kind of act as a matchmaker between those groups and create an easy interface for that yeah I certainly expect when transfers are enabled on the beacon chain for exchanges to kind of step up and provide liquidity options to validators and kind of expect you all to do that are you gonna help organize that I just expect us to I've talked to people at coinbase so cool [Music] but I don't know kind of explaining especially when transfers come into existence having literature maybe focus for exchanges to better understand what's going on would it help enabling exchanges to enable our users so yeah sorry can we expect eat to eat to be traded as a separate token yeah well that's what I'm saying like it's an expensive lean exchange decision like we could have exchanges this okay yes okay sorry I didn't mean to be like blunt but yeah if in exchange were to like similar to I guess BCH for the fork for that like yeah okay yeah thank you for your answer yeah I mean the thing to think about there is ultimately whenever you've got anything that's a deliverable currency you have to be able to deliver it to the right place so if theif two and eath one is separate you can't transfer it back then anything doing deliverable trading absolutely has to trade it to as a separate thing to the each one if it's non-deliverable you can get more fancy and people in fine it's like to get fancy but I won't try and speculate on that well I will say one thing about if you look at existing like shafts and OTC markets like there's already like well-established processing islands were how people trade these illiquid like promises essentially and so it's not out of the question that the same thing would happen for if to as long as some liquidity providers or changes were supporting it it's the reason why I asked this question is because I feel like you know from like regular human standpoint it would make sense if you know like the transition would be kind of seamless but it's clearly not like it is like you know like it's it's it's also an investment opportunity and you know so there are certain issues there which I feel like it would be maybe that would be one reason to like go for the bridge like right away so we wouldn't have this problem even like from it just from a user standpoint like they would be really clear that like you know we would never get into the situation where you kind of like gamble with like eath 1 vs e 2 or on an exchange so when instead of right away another option right there is at the moment that transfers are enabled so anytime at the moment there's a liquidity an option on 8th to there being a fungibility option and that's that's interesting I think that um there's many factors that play here but looking historically coins like Zillow or like what die is about to do there's a lot of a lot of things going on here it's hard to predict but looking at previous coins that have migrated to other blog chains that support their old blocking as well their new blockchain such as silica and and I may set precedent for what can happen here or what can't happen here something that's really interesting coming out of this conversation is at least the people that are speaking it seems like the biggest concern is value and liquidity and all the questions around the token stuff yeah yeah ok is that not if that representative of everyone in the room who cares about value you know you can care about valuable lives there's a probably other someone said no yeah I'm Barry and I don't think that it's represented about the group I think that the people who were sitting closer towards the front self selected as as like having having some kind of strong motivation to interact in this conversation bigger bags strongest motivation a very strong motivation is a financial motivation except for me I just said here I will say something that I'm very curious to hear is how much of the work that is being done for each one point X is repeatable so if we're like pushing back state rent and it'll get implemented let's say like a year or two from now like is it an easy drag and drop onto 2.0 or will we have to rehash a lot of the same issues it's a the East 2.0 kind of core design is designed around and if state minimalism in sort of state is a kind of somewhat higher level a level concept than consensus but like basically all of the stateless client work that's happening on the one point X side is also simultaneously work that prepares if one point X for and if being a transplanted into being an executor into being an execution environment inside of each to I can also say that from a technical point of view like a lot of the work that we've been doing is kind of like shared in-between like you know the clients I mean after all we we do share like for example we are making like the goal based eat one client and there is like sharing of libraries going on with the east to implementation so it's not the these projects aren't completely separate when it comes to the actual tech it's more like they are conceptually different of course and they do implement different protocols and there are different products in a way but the software is there is a lot of sharing going on and the same can be said about Trinity I guess because it has both so it does share a lot of code tacking on to that question and like between these 1 & 2 we're doing the same thing but changing everything so we're changing the consensus we're changing the networking protocol we're changing the execution environment and we're changing the cryptography what are we hold constant [Laughter] it is my smile I mean the fact that current applications will continue to work through the Train really transition kind of isaiah's like modulo concerns about opcode repricing z' I think is one major thing I have a question about as clients so we have what seven clients in development for e - would there be a efficiency to be gained if efficiencies to be gained if we had less client teams and more work being done on getting to the end and also a second part of that is if there are seven clients the performance of the blockchain is going to be held back by the slowest one so that increases the chance that is not at full performance right so if you could take all of the people working on all the clients and put them on half the clients maybe so these are independent companies and entities that you can't necessarily just do that with that's just the cat herding problem that said there's probably diminishing returns after five but we've had kind of an incredible amount of technical sharing and idea of sharing and I think it's actually a been a boon to the process in many respects with the upcoming tests nets and move to production I do have some concerns about managing complexity and managing user expectations on what which software to run right it's hard to know it's hard to choose and we will have to speak to the second question metrics minimum set of criteria that to be production-ready so if you can't handle a state transition with 4 million validators and under this amount of time then like you don't get the check in the box and you're probably not ready for production and that'll be very clear to users and so much better to have three great clients than seven that maybe no all of them agree right and we probably will have three great clients and maybe more but we will certainly have three great clients so I think we end up in the same spot last question please I was just going to add to that that one of the problems we've seen eath one is that gap is so popular that we're gradually kind of converging and a lot of that gas pricing is based on guest performance which is good and gets a great client ID you know but one of the advantages of having so many clients initially in e2 is it is it helps us avoid that centralization so we do are much more likely to stay with a good broad range of clients with a broad range of approaches to solving problems yeah I'm really hoping that we don't have a client with over 80% of the network I think it's really valuable to the security that network and I we will see how many actually launched with a zero as production-ready I'm not I'm not sure we saw the problem with each one that that a minority client managed to slow down the network until well in each one we also saw that a minority net client managed to speed up the network right like one that I did in a particular instance yeah yes in a particular very important instance than what otherwise have killed us you