Protodanksharding Upgrade | Presten Van Loon, Terence Tsao (Offchain Labs) & Christine Kim (Galaxy)

okay hi feeds will be drastically lower and we should see a lot of or some movement from from the L1 to l2s where things are cheaper and sort of moving towards the scal of L1 being more of a settlement layer than where we're doing most of the transacting yeah so let's talk a little bit more about the fee part because that's really seems like the goal like one of the main results that we want from EIP 4844 from the denune upgrade I have yet to see really any concrete estimations for how how much cheaper exactly we're really going to get it seems like it's the real answer is we don't know it'll get cheaper but maybe between the a very large interval of like 10x to 100x and you know there might be we're pretty confident there's going to be secondary markets for blobs but we don't there hasn't been a ton of analysis on exactly who these markets are going to be run by there seems to be a lot of open-ended questions of what's really going to happen to l2s and the blob fee Market that I I haven't really seen developers really care about or explain um K Terrence can you explain why that is why do we not know what's how fees are going to be impacted by l2s if this is the point of the upgrade yeah that's a great question so also I like your shoes thank you they're very nice this was actually for your pocket just supposed to we jungle themed but yeah anyway shout out shout out to your podcast jungle themed but um so the texbook answer to that question is no one knows right but I can give you my more like math like more like a math estimation that if today you take the arbitrum data you switch from code data to blob data it's about four times cheaper right so you can say at the Baseline everything will be four times cheaper but then again that's the first order effect then there's this second order effect if today it's four times cheaper then more people will use it then it becomes the same clost as before or even more expensive it's almost like the thermos right you put it in your room you set the temperature to 75 degre and then you can go up and down but eventually you will reach the equilibrium and today the equilibrium is this three blobs per slot just if the if today we have three blobs per slot if it's at three blobs or lower than three blobs the price is basically free so anyone can post anything it's basically free right if today I see the price is free it's zero I I mean us us like a hacker or someone that's poor I can just post random data on there I don't care right so you can think of that from that angle someone will always do something so you will always be three blobs right so now the question is okay what happens what's the second order effect and today um old layer 2 combine use about one gab of data per day and that's about 1/4 of the whole EIP 4844 blob space so we have another have to reach before the price increase and I have no doubt that we can easily reach that okay as far as says they are I mean that is a little bit more clarity so I appreciate that but assuming that L2 fees really do come down significantly and become as competitive as say um competitive alt layer ones like salana I don't know like Avalanche etc etc my question is are l2s really the place where we want to push ethereum users to go to now given that the level of decentralization say security say scalability of l2s is really not that different from what you could experience on say salana or these other old l1s that I mentioned and really like the competitive advantage of ethereum's network effect is that you can you're able to to utilize and leverage the full like composable stap ecosystem of ethereum like are you are we really sure that we want to start pushing users to the L2 Now with an upgrade like deun are l2s ready for end users and are you worried at all that as L2 as users move to l2s they notice that the user experience um for this level of decentralizing decentralization and security could perhaps better be better be utilized on a platform like salana uh I'll start with that I you know I would say that an L2 is only uh as secure as its weakest link right um and if you think about a complete L2 a complete L2 benefits entirely from the security of ethereum so even if it completely fell apart you still can exit and and complete you know whatever you were trying to do back onto L1 so the security is still there although like that's kind of still like a a worst case scenario so I mean if it's a true L2 I wouldn't say that you have a less secure environment I would say maybe you have a a aliveness risk or maybe a reliability depending on the L2 right they could go down but ethereum doesn't go down um you know but as far as like security wise like if it's a true L2 it should be inherently just as secure as L1 security so l2s in their current state you would say is competitive against other Al of l1's well the caveat is this has to be a complete L2 right you have to have like some mechanism to prove any fraudulent behavior of a sequencer so you have to have fraud proofs you have to have all these mechanisms and I wouldn't say that all l2s have this I wouldn't even say that any L2 is complete yet uh but they are in different phases of of reaching that uh that level of security anything to add there so I will add so I will maybe like a little bit more controversial point is that if we today if today we don't have this like scalable data on ethereum meaning that meaning that if we provide scalable data then more users might move to later too because it's cheaper if we don't do that then there's other da Solution that's coming right people will eventually move out of ethereum anyway to use the other da solution but then the Downs side of that is that well you're not settle on ethereum anymore and you're posting da something else then the security is worse right and then and then as someone that's working on ethereum space I will hate for them to take our lunch basically right so that's something that we have to do something right and whether like like whether how to do it like what to do it like when to do it that's another question wow ethereum competing also as a DA ler that's another really big topic that you know I don't think we have time to get into but it is a point that everyone should take note of and perhaps we should do another panel on okay let's also talk about testing for Denon a little bit because I got to say Terren prism was in the hot seat the prism team the client team was in the hot seat for many of these calls um so many client teams on the execution layer and consens layer were pushing you know let's get these test Nets up let's um move forward with scheduling um the test net upgrade for denune on a certain date and you were one of the voices sometimes the only voice asking for more time more testing more review um let's see how the me boost infrastructure is working and I think in in many instances um there were periods where where other client teams were kind of suggesting that your team the prism team was the reason that you know the the upgrade was being delayed um were you feeling very pressured about you know trying to get all of the code changes that are required for denune in in a timely manner are you concerned at all about the testing process having been too rushed given um given just like the amount of of kind of of of how how the prism team was a little bit in the minority this time around for for um testing timeline and and opinions around the testing timeline yeah yeah definitely so this is a great question so I think prism team was actually the first team along with Lighthouse that had this 44 for interrupt to start with right and then between them we have found like many issues with the spec within the code itself and then we kind of iterate on those issues right and the question was just like I think this is just expectation right just say to say we have been on menet since speak challenge for about four years now so would you arue the bigin challenge was rush or would you arue the B challenge was right on time right and to be fair when the B launch us we probably have like 90% confidence that this is going to work but we have to ship something right so we had 90% confidence then at the merge we probably had 95% confidence that it was going to work but then it went really well right but then as the pro as the protocol alifi as the cliental aify I argue that it's becomes very hard to make changes and that is for the good thing right like today would you ride an airplane if someone tells you it works 99.9% of the time probably not right you want to you want something that's perfect as close to perfect as possible so we kind of had this mindset when we when like we kind of had this mindset Bak in our brain as get towards the Manet that like we're still finding bugs uh the relay is not ready the and then the M boost is not ready and two weeks test time is not enough and typically we do way longer test night time but this time we only to do two weeks for test n to us that was very Rush right I guess it's just managing expectation if today you say okay well if we're okay with 90% ready then let's ship it then sure we're fine but we're kind of striving for that 99.99% by Perfection yeah we used to move fast and break things I think before the beacon chain launch you can look at some of our early test notes kind of Crash and Burn um and now being in a position where we're one of the two most popular clients we can't afford to mess it up um and our testing internally we just we're really really looking for the bugs and making sure they're ironed out uh cuz when prism has an issue ethereum will have an issue and we're really just making sure that's not going to happen I 100% agree with this logic but it doesn't seem like the logic that other client teams are adopting like prism might have a 95% are we 95% sure but it seems like perhaps other client teams have a little bit of a lower bar and I'm curious to know like as the complexity of ethereum grows why is it that there seems to be increased pressure to get upgrades out faster if we're talking about denune the timeline for denune especially the last three test Nets there was even talk about you know instead of having the last three public test Nets two weeks apart just like shortening it to one the next upgrade Electra there's a ton of pressure to get it done before the end of the year if ethereum is getting bulkier more complex difficult to test why is it that we the testing process and and it seems as though the The General kind of feel is to to do more upgrades faster like basically start changing ethereum like move fast and break things um yeah I I'll start with that um you know if if we were more of a minority client we could probably YOLO a little bit more like CU if if 5% of the network goes offline well things are not going to be too impacted um and to the point about people expecting things to move a little faster I mean we've been hearing this for a while when merge like that was a meme for a long time like we almost got a little bit like come on give us a break we're going as fast as we can here we're only human um and yeah you have a great point about complexity right like now we have two layers we have to manage and there are a lot of clients and there's a lot of different combination of things that could happen and there are a lot of sophisticated uh ideas about like attacking PTP so we just really really wanted to be robust and liveness uh it's really important um you know I just say I feel like in the in these instances it's better to take a little bit more time because you know when you look back like no one's complaining that the merge took too long we're all like what about the merge it's old news you know so let's just have a little bit more confidence as close as 100% we can be like every time we do a release we say do we have the highest confidence we can have on this and if not let's take a little bit more time to look at the data a little bit more time to analyze to soak this to try to break it so we can catch the bug before for it's out there and actually impacting users I can talk more on the research angle right I think this is like a future and the bug because like social coordination is very expensive on ethereum level right it's like we like to argue a lot about nothing typically no we don't but like between the researchers the implementers and different and and and and just in different forums right but then we can also speed things up it's very easy we can say okay this is a topic we made decision w't just just one week from now right and that's it right but that kind of just becomes like a company right and one of ethereum's strength is this decentralization governance is this raug governance but the D of that is takes a very long time right versus like you go to another approach which which is more top down like company style we can make decision much faster right so it's kind of the trade-off here and ethereum kind of prefer the later approach okay very quickly because I have another big question related to this um but very quickly percentage wise right now how confident are you that dankun is going to go off without a hitch there's not going to be any problems and issues March 13th okay that's a tough one um I would say we feel really really good as close to 100% as we can be I mean we've had a lot of testing and even making changes as uh as much as a couple weeks ago to just say you know what if we do this we may not need it but just having it it's going to be a little bit better uh so I'm feeling really really confident if it was tomorrow I would be sleeping tonight very well I'd be excited um I'm not nervous at all very similar to the merge teren said 99.9% I think and that's I feel really good about this next upgrade it it I'm just excited for it to happen I feel good about it so there are also a lot of dependency right we can't be 100% but ethereum right now if you look at the stack there's multiple layer right there's C client there's EO client there's me boost there's relayer there's Builder there's sear that's blah blah blah right then today if one today if the one of the real doesn't work then we're going of screw it if today one of the Builder doesn't work we also kind of screw right so you kind of have this dependency as well which is kind of beautiful I guess but then again there's there are a lot of surface for risk and we're trying to reduce the surface that's fair there are an increasing number of dependencies um since the merge for all these other softwares and protocols that need to be upgraded um as you upgrade your execution layer client and that does introduce more risks okay um okay so one final big question um unfortunately we're not going to get time to talk about Electra like I thought um but we'll have another panel for that another time um big question about the future road map of ethereum so the merge was supposed to be the last phase of development and now with Proto sharting I don't know if you guys know this but Proto D sharding is Proto it is like one stepping stone to full dank sharding so there's more upgrades that need to happen to even get the full vision of what we're talking about to reduce fees for l2s um and aside from that there's just a ton of other development items veral Max EB inclusion list ens shrin PBS it seems like a NeverEnding list of upgrades that need to happen on ethereum and I'm wondering if this worries you I'm wondering if you guys think that this constant pattern of more upgrades is actually a major centralization risk for ethereum the fact that ethereum can upgrade so easily so so quickly Noe operators don't ever kind of hardfork away from ethereum and the release that the client teams are putting out um this seems like a norm and and a and a vision for ethereum that's not going to be changing anytime soon um How concerned are you guys about the increasing complexity that comes with all these upgrades the increasing centralization and resources that comes with having to upgrade like twice every year um my short answer is software is never finished um an ethereum Court protocol for example will be continually making even if you said it's feature complete and we're not add any more features there still maintenance stuff to do uh and this year in electure like the prism team's point of view on scoping it is let's pay some of this Tech debt down and kind of simplify things and then we'll go back to the more complex ideas in the following Fork um Terence what do you think I'm not concerned at all I mean ethereum works as of today right if today say if everyone goes away etherum will keep working we have great client diversity we set up a really good foundation right now we just keep adding features on top that we wouldn't have right and then I think we have accomplished what we wanted to accomplish which was the merge and now we have some sort of data sharding right and we can just keep building on top but yes there's a lot to do but I don't feel like we're pressured to do this like tomorrow B and so you guys aren't worried that one of these upgrades is going to break ethereum go ahead um no because there's client diversity that's the only reason I don't worry about it there's enough like people at different geographic location there's different software stack and stuff like that if one of the bug goes done unless it's like a really crazy spec bug that's open source which everyone can view then like most of the clients will still survive yeah same answer for me it would have to be uh something fundamentally wrong with the specification not really in particular client because of the client diversity model the multi client model right well thank you guys thank you guys so much for chatting with me about the danoun upgrade I hope this was helpful for people in the audience too um if you're a node operator please update your your nodes before March 13th um and yeah thanks for thanks for listening thank you yeah all right so good news we still have some time so we can dig a little bit more and and jump on other topics I think your mic is off but it will be on sorry guys yeah we we the timer we I thought we were going overtime but I guess we have more time so maybe we do actually have time to talk about Electra a little bit um what yeah so okay maybe we should talk about Electra let's talk about Electra because we talked about Denon talked about no please don't go um let's talk about electric because we just talked about denune um one of the most controversial things that have been brought up for Electra is obviously reducing staking rewards would you agree or would you say that there's other more controversial say that one is I feel very controversial I don't know if it's the most controversial but it when you're looking at that reducing the validator rewards right what is the problem we're trying to solve is it's this growth of validators right we're trying to say let's disincentivize it a little bit to maintain this growth um I think the controversial part is that this won't fix it at all I mean it might help a little bit but it's you're still going to have the same problems because you still have other incentives you have block fees rewards rakings other kind of things you I I would even go to say that if you set block reward or uh validator rewards to near zero you're still going to have growth at least a little bit maybe not on like uh uh the institutional side people are not going to be very interested in 0% or 1% apy but uh like at at home Staker um when I saw that I was like whatever I'm still going to stake like or and if I get more eth I'm going to put more in there it's there's still yield I'm still going to do it this is true is this about the insurance curve yeah this is about the reducing oh man this is okay so I agree I think we cannot leave in a world where 100% e is stake I everyone agrees with that I think 25 right now the insurance curve is Target 25% my opinion is that's probably too low I think there is a middle ground that where other everyone's happy I don't know what the number is maybe it's 50% or something or something like that but I do agree that like we need to do something okay so this is the my only worry is that say today we ship something at 2024 and we don't touch the insurance curve okay and then the next hard for say delay like always supposed to be 2025 but we delay until 2026 and that's basically two years of doing nothing right and at that point the wor case scenario we could be looking at 60% of each stake right and that's GNA be painful to fix between when you look at 60% each stake and then you have to kind of like make people add it down to 50% or you're happy with 60% so this kind of weird timing but I do think 20 25% is probably a little bit too aggressive okay but question about that why is it that the staking rate being high like say 100% is bad if if the staking is not done through like if you had enshrined liquid staking or if you had enshrined I don't know reaking yeah isn't the reason why like a high staking rate isn't the reason why it's bad is because most of that staking is happening through like a smart contract so the smart contract risk is bad but if you were to address the smart contract cont ract risk would a high staking rate be fine I think that's probably fine as long as it doesn't go through like Ste or r e because now your majority of is control through Lial governance right but then if today you have some inshine staking which I haven't really heard like a really good proposal there I think that would take much much much longer to enshrine like a staking onto etherum but maybe yes but like you said maybe that's okay yeah it just seems impossible to me to stop Innovation on the smart contract level without thinking of a way to enshrine these these um these uh procedures these like abilities but you're right it is a very complex upgrade but speaking to um you know addressing the growth of the validator set another very important code change is maximum effective balance why in the world is that not a higher priority for Electra I'm confused because I thought that could literally break like the networking layer of ethereum um and the the kind of Band-Aid solution we have in denune reminder on denune we've got a a Max turn limit of of eight so WooHoo for that um but apparently you know that's not going to stop the growth really all that much especially if that that maximum turn limit keeps getting keeps getting hit so why is not Max EB the top of the list for Electra from from the client uh developer perspective Max CB is one of the more invasive changes in client implementation has a lot of complexity we can do it in Electra our the the argument we're making is if you want to ship something this year we're not going to do Max CB um and we're suggesting uh as I said earlier let's let's simplify things and pay some of this Tech debt and then start bringing in the complex changes because it's hard to do both at the same time and it's also hard to ship and upgrade with a lot of different uh changes because there are a lot of things to test and you know there could be gaps perhaps it's just it's just harder to deal with um I do think Max CB is a really great idea um I think that's something that people have have been wanting for a while right like the this concept that you deposit 32 eth and you're earning yield but you're only earning yield on the original principle and not the um the the yield that you've made um that's just driven more and more adoption towards liquid staking where people are like well if I just put that in there it's just compounding over and over again uh and for the average user they don't have many manyi multiples of 32 e so that's a ra a an Avenue they've taken um but I think it just boils down to complexity and shipping on time because I can imagine we say yeah let's do it and then we're hearing you know when's electric going to ship stop delaying it let's go let's go let's go but it's if it's really complex it's it's going to be hard to do that I think the problem today is that okay so before I say today the merch there is a headline then withraw there is a headline and there is um Proto sh 444 was the headline right but if you look at Electra or Petra like what is the headline there is this is the first time in etherum history where we have a hard for but we don't have a headline right but because we don't have a headline then people started like wondering what Future do We ship or what timeline do we base on right so so now we have to ask our question do we determine the future based on the timeline or do we determine the timeline based on the future right and and literally no one has answered that today I think that's that's the first thing we need to decide as a community right and then from our end we have always promoted that we want the timeline to be fixed that's 2024 and based on the timeline we decide the future but then like say if today we do the timeline is on 2025 then yes we can shoot met EB we we're very happy with M EB but we just don't have a timeline in mind okay that's a very good point but already developers seem to be thinking that the headline feature for Electra is inclusion List It seems as though inclusion lists are the big feature that's going to be going into the next upgrade and I'm a little bit confused because I don't even think inclusion lists are going to be that helpful for censorship resistance on ethereum um which we can get into but I I personally am not confident that Electra is going to happen before the end of the year and so I'm just I'm wondering if if your team is again in the minority about pushing for a timeline and a type of upgrade that is focused on reducing technical debt but decisions are being made on these calls on these ACD calls where we're including upgrades like inclusion lists that are complex and are not necessarily reducing technical debt but increasing complexity I mean are you is it what what are your thoughts Teran yeah so like so I think the whole idea with this timeline is that we really wanted to sh veral so veral was supposed to be this next speed thing right and then the question is okay verle will be delayed for 2025 that is Osaka right now do we do something in the middle or not so if we want to do something in the middle then that's then that middle something has to be small enough such that it has to be the midpoint between now and then Osaka right just if you make it too close to Osaka then Osaka will be delayed and then and then and then the verle will be delayed so we're pushing veral apart so that was kind of like the basically like how everything was decided but then again like I'm just only person speaking like yeah I'm just speaking like there's so many opinions out there like at the end of the day it's going to be like Community decision and I do hope we reach to the decision very soon anything to add there Preston nope okay yeah um well okay the other thing about Electra but also broadly about um about the way that these upgrades and especially the denune up grade um other parts of it a big feature that we're talking about that we talked about that developers talked about on today's call on today's developer call was account abstraction and um basically code changes that are going to impact the end user and also dap developers so in denune actually there's a couple of upgrades that are going to improve you know the smart contract developer experience like transient storage that has been championed for several years um the deprecation of self-destruct um these are always minor code changes that are a little bit hard to get into upgrades and kind of tacked on have you guys thought a little bit about prioritizing like the smart contract developer experience for certain upgrades like I've always seen the main feature being the merge like um you know 4844 and it seems as though as ethereum focuses more on being a DA layer da layer these code changes that are focused on making ethereum a good execution layer are starting to really the the wind in their sales is is is lessening and I worry that you know things like eof improving the solidity language improving the evm all of these code changes that are more focused on users on the developers on ethereum they're not going to to they're they're not going to headline an upgrade is is that concerning to you or do you guys want developers to leave ethereum is that is that really the goal here I mean definitely don't want developers to leave ethereum and I hear I hear you on these some of these pain points here in my uh personal opinion some of the things you listed are on the tech dead side of things like uh State expiry and um deprecating self-destruct I think we should you know really be focusing on these things but it's also hard because we have a lot of like all l1s that are like oh work ship the feature faster and they have the advantage of not having this Tech debt with them um so it's kind of this like balance of how do we move fast and have new features and stay competitive but also like take care of the mess that we're leaving behind from like maybe just uh poor design or things that are no longer relevant anymore so I think like Kev prent I almostly on consensus layer so we cannot speak so much on the execution side but from the consensus side we actually have been trying to make the developer easier for example um we are exposing the beacon route onto the evm that helps IG layer with the reaking helps rocket pool and then for the for the Electra we also have this extion layer Trigg withraw that also helps IG layer and staking pool so we're trying to make the deps easier but there's not much we can do because we mostly focus on the consensus layer yeah H yeah I mean I feel like in the upgrade there's competing visions of what ethereum should be like there's certain upgrades certain co- changes that will really help keep the developer dap ecosystem on ethereum strong and others that will push you know encourage users and daps to move to other protocols like the l2s um but you know ethereum's trying to do both compete on both maintain both improve on some um and I can see why there's there's lots of lots of competing priorities um okay in the last couple seconds that we have assuming that Electra happens in 2025 um not saying that it will but just potentially if it happens in 2025 what is like the main code change that you would like to see in Electra uh well if if we're confident that it's not happening this year happen 2025 um and they were're happy with it happening anytime in the year I think we can open up the scope a little bit and reconsider at least from our point of view you said we're kind of in the minority of saying let's keep it small but if if it's not going to be in the in the midterm uh before Osaka then let's go ahead and do these things like Max EB inclusion list epbs is really exciting as well I know that terance is a lot uh input on that um you know all of them well if we're talking about we have nearly two years from today or say conservatively 18 months I think that's a little bit easier to try to get more stuff and reevaluate the scope uh I think that the scope that we initially said was as small as possible right keep the in there so we can ship something this year like November December but if we're saying well let's do November December of the following year then that's an entire year of development so I would say that we could try to get those things in there because they are all nice features and we do want them uh but as teren said it's like a little bit of a timing thing like do you want do you want it all or do you want it now like you can't have both do you have one terance or are you also similarly thinking like if it's not going to happen in 2024 then why not just do it like all of them and delay veral even further yeah I mean this is also like has to do with the El side planning with vertical planning right I mean ideally yes if you do everything in 2025 we can squeeze in a lot more things because when you have two hard for it usually means longer time than just one hour four because you have to upgrade three test Nets and the man there right and then there's about two three weeks between each test net so those time are kind of just Idol I guess it's not Idol but you can focus on other things but yes it this I guess you can squeeze we seeing a lot more features with just one half for all right I okay yeah well thank you guys for sticking around for the second half of this talk on Electra um we hope that was helpful and thank you guys for for sticking around and thank you Terren and Preston for the Insight on the next ethereum the next next ethereum upgrade yeah thank you thank you yes let's hear it for our great panelist thank you Christine and