2023 Ethereum threats, ETHLisbon finalists and more - The Daily Gwei Refuel #585 - Ethereum Updates

Transcript

hello everyone welcome to another episode of The Daily gray refuel where he kept latest news in ethereum ecosystem I'm your host and this is the 15th of May 2023 alright everyone let's get into it so an eventful weekend yet again in the world of ethereum obviously well obviously to anyone who was really paying attention over the weekend the finality issues actually ended up coming back I believe 24 hours after the first event that happened and honestly this wasn't a surprise to me I mean I woke up to this again it happened overnight for me and I woke up to it and I wasn't surprised because there hadn't been a fix implemented right the it was kind of the network killed itself but then everyone was like you know why did this actually happen and then the root cause was actually identified I believe an

hour after I I recorded um Friday's refuel and then a fix was put out I believe on on Sunday or yesterday or Saturday to address these issues but the root cause of the issue was actually uh I mean it was prism and and techu I believe is like the the two corporates here but it had to do with the fact that uh prism basically was using up too much uh resources and eventually death spiraling which caused the client to crash and I believe techu was doing this as well and that's why we saw such a huge drop off in participation on the network uh simply because the client was overloading it wasn't it doesn't seem to be malicious it doesn't seem to be an attack doesn't seem like anyone was triggering this it was just an unfortunate kind of I guess um event

that happened and there were a few things that had to go wrong for this to happen and it did and but since then as I said there's been fixes put out by both uh I've got the Tweet here actually by by both prism and teku so if you run those clients definitely update if you haven't yet I think most people have updated because we haven't seen these issues uh come back uh since then but if you want a full breakdown of it Terence has a thread here explaining uh exactly what happened and how there was uh CPU spikes and memory spikes that led to basically this death spiral and I believe there's going to be a more detailed post-mortem published this week now of course this is like the probably perfect and and best example we have so far about why client diversity is so

incredibly important on ethereum if we only had one client this would have been catastrophic for the network the network would have gone offline and we would have been in a much much worse position than what we are are in now and this speaks to I guess ethereum's core value prop if the ethereum network itself like layer one it's called value properties to be a global settlement layer and a global settlement layer has to have redundancy built in right it can't go offline it has to be redundant against both attacks and and bugs that can happen because obviously bugs are always going to happen there is no way 100 foolproof way to avoid bugs and attacks same thing you can protect against attacks you can set up defenses but if there was like a for example a critical bug that gets exploited on the protocol

layer while uh that that could happen there is always a chance that that could happen right so building in redundancy in the form of things like client diversity is incredibly important to through not only the future of ethereum but ethereum's main value prop of being a global sediment layer now of course I've showed this website to you guys a bunch of times before client diversity.org that shows the different uh the different percentage share that each client has I do want to say that if you are running a prism or techu and you want to switch to another client for whatever reason right maybe you think to yourself well uh you know prism and Tech who had this bug in it I want to go to another client that didn't Lighthouse seemed to perform best but the problem is is that Lighthouse is already the

like the majority client here it already has 38 Network share so if you're thinking of changing clients from from uh prism or techu maybe from Prism to techu or one of the other clients out there uh definitely consider one of the minority clients either that being teku Nimbus or especially lodestar don't change to Lighthouse just because it performed best because you're going to give more market share to to Lighthouse and they don't need it right now I think as I said uh previously the consensus client distribution right now looks pretty good but that can very quickly change you know lighthouse could just rock it up to 50 because everyone saw that Lighthouse performed well and maybe they were running prism that wouldn't be an ideal thing you know we don't want the consensus layer looking like what the execution layer looks like right now

in terms of in terms of market share with guests being 61 and the next closest being at about 18 uh but yeah if you are considering switching do keep in mind that load style is actually running an incentive program right now I've covered this incentive program before uh so if you're interested in participating in this there's a blog post here with here with full details on what this entails but yeah if you are if you are a solo steak um that wants to change uh basically who you're staking with or so which client you're using I should say then definitely consider the load style incentive program but if that's not of interest to you you know nimbusoteku are waiting for you there don't go with Lighthouse I mean Lighthouse is an amazing client I'm running Lighthouse right now I'm also running Nimbus but

I've been considering switching away from from Lighthouse to something else just to help with client diversity here uh so yeah just wanted to give that shout out there on that front uh there was also a thread here from POTUS who breaks down the loss of finality that happened uh over the weekend and on Friday and there's also uh Ben Edgington here who pointed out that the ethereum mainnet experienced its first ever inactivity leak that happened the second time that we lost finality now if you're unsure what this means Ben has a link to his upgrading ethereum book which basically explains this year but essentially the tldr is that when the big and chain is not finalizing it enters a special inactivity leak mode a test has receive no rewards and

non-participating validators receive increasingly large penalties based on their track records this is designed to restore formality in the event of the permanent failure of a large number of validators so it is an inbuilt mechanism it is a really important mechanism and it kicked in here obviously because too many validators were offline for too long the second time around and Ben also describes here about uh uh basically what actually happened in terms of rewards he said lots of people asking how much eth was burned by yesterday's inactivity leak back of the envelope map uh math assuming eight Epoch leak 65 of validators offline about 28 uh eighth were burned uh because uh zero point triple zero six eighth per offline validator however attestation rewards

were zero for the duration so that's 58th not issued as well so I guess it ended up being very slightly bullish for eth because we uh burned more Ethan we didn't issue as much eth as we otherwise would have but generally this doesn't exist for that purpose obviously it doesn't exist to pump eth it exists for the reason that Ben notes in his book is that it needs the ethereum network wants to restore finality as fast as possible and this is one way that it does it now there were some questions around what this would mean in a single slot finality world as far as I know the results of like if something like this happened in a single slot finality world would be the same you know we would have a loss of finality for the period until we had see at least 66 of the validators

uh attesting to blocks but we would restore finality within a single slot or 12 seconds or a single ethereum block rather than uh waiting uh two epochs to restore finality so it would just get restored a lot quicker than than I otherwise would have which is really cool but obviously we want to avoid having a loss of finality at all costs finality is crucial to ethereum proof of stake and that's why there are a kind of mechanisms involved to get finality restored here but glad to know that it wasn't anything more malicious than just uh you know a couple of clients basically having issues where they death spiral a death spiral due to too much uh resource usage here and as I said there are hot fixes out for this with proper

client releases I believe coming as well probably will get released this week that fixes this stuff so if you are still running the old versions uh definitely consider upgrading now on that note as well if you weren't on a resource constrained system uh like if you had a lot of resources say you had like 128 gig of RAM and you had a really high-end processor or CPU you probably saw a spike in usage but it probably didn't kill your your client it was more the resource constrained devices that suffered here in saying that like my device my rock 5B that runs Nimbus validators uh didn't go offline um I think it ended up the second time around missing like four attestations when the second thing happened but it wasn't because it was um going offline or

anything and it doesn't seem like Nimbus was was affected uh by this it was it was techu and prison by the by the looks of things here um but but yeah it just happened on resource constrained systems if you had a lot of resources you probably didn't see that happening uh there but yeah that's that's basically what happened and those are all the fixes and different threads which I'll link in the YouTube description below for you to check out and to also check out Ben edgington's absolutely amazing explanation of the inactivity leak I only read the tldr at the top there is a whole page here with lots of information about what actually happens under the hood when this inactivity leak kicks in so you can go check that out for yourself um I'm gonna yeah last thing here about uh about each staking was a tweet that I

put out basically saying about almost 50 000 validators in are in the eth staking entry queue right now three validators in the exit queue and then I said much death spiral such Mass withdrawal obviously uh there has been no validator withdrawal uh you know Mass exit Mass withdrawal there's been no death spiral it's really just been up only and based on the fact that there is almost 50 000 pending validators there's actually uh if I bring up Beacon chain here for you guys to see there is currently 49 343 pending validators uh waiting to get in and 64 exiting validators and obviously my numbers were based on a different time period but like that's almost a month guys of an entry queue to get into

staking but that's that's crazy to me the the fact that like there the entry queue is now almost a month and I remember I warned about this along long time ago that I believed up uh I I think I said post merge but it's it's kind of happened post withdrawals that we would say a really long entry queue so get in before that happened and I think a lot of people did but a lot of people are still stuck in the queue my mini pools converting them from 16th to eight eighth um uh mini pools and spinning up new new validators there I went into the queue and I I think it took like two weeks or something like that and I was like wow I'd never been in a queue this long before to get into to staking uh but yeah it's almost almost a month now

which is which is pretty crazy there but also on this front uh there has been uh because of the fact that there are so many validities being added to the network right now there has been renewed talks about the evaluator cap or I guess like the um the active validator cap that I've spoken about on the refill before and if you've been seeing those discussions and want to know exactly what that entails vitalik's uh hack MD or notes post I should say here from about seven months ago talks about this in the context of single slot finality so the the title of the post is part towards single slot finality and this covers all the different ways that basically the researchers are thinking about doing both single slot finality and also capping the active validator sets because there will probably need to

be a cap eventually for a number of different reasons it's not that the beacon chain can't handle a large number of validators it's more about like what is the actual security benefit of um you know more validators being active on the chain versus what is the risk of the chain becoming unstable or insecure for whatever reason and there are different and there are different kind of like trade-offs of each of these approaches which you can read in this ProStar now this post I don't know if there has been an updated version of this but I would wager that pretty soon we're gonna see a lot more talk about this from researchers from developers it might even be something that makes it into the upgrade after Cancun like single slot finality and a validator cap because if you do napkin math on this and you you

say I mean this is extrapolating out the things yeah but let's just say that there's there's going to be constant 50 000 validators in the queue I don't think that's going to be the case but let's just say it was and uh we want to cap the active eat stake at 30 million each well that's not too far away we're we're at I believe almost 20 million each uh the staked by a beacon chain uh has here is just the effective balances so that's basically all that's not including the pending validators it's just including active validators so I believe if you include uh pending validators as well there is uh uh almost 20 million staked eight I believe I've seen those those kind of numbers floated around there so 30 million not that far away right and if we can continue this

pace here we'll reach that in under a year so if we want to cap it at 30 million it's probably going to be a lot better to do it before we reach that number and it's probably going to feel like less of a rug for stakers if you do that before you reach that number because there is going to be debates around the staking APR because there are as I said there are multiple different ways that researchers are thinking about doing a validator cap but it depending on how they do it it could impact uh staking APR and let's say that you're a Staker and then the staking APR is cut by like one third and I'm talking consensus layer rewards it wouldn't affect the execution layer Awards but let's say on the consensus layer side uh all of a sudden you've got one third

less rewards than you signed up for we may see a kind of mass Exodus then right front from staking I don't think so I think there's a lot of different considerations when thinking through the the implications of that but I would rather see uh just even a draft of an EIP go live sometime this year that discusses capping validators in the context of single slot finality because I believe it is it is critically important for us to get that ironed out now also I think that there is a bad narrative forming around that as well where people think that the validator cap would mean that no more eth could be staked that's not how it would work it would be a soft cap when that what that means is that let's just say the cap is set at 30 million eat steak now the cap

would actually be set at the number of validators but for ease of the argument let's just say 30 million e-staked is the cap now that is a soft cap which means more eighth can be staked it just means that only 30 million each stake or 30 million each worth of validators it would be active uh and actively attesting and producing blocks at any given time so there could still be like 60 million each staked but only half of that would be actively uh validating and earning rewards and in an ideal scenario the reward curve would still be the same where everyone is still earning the same consensus layer rewards but then obviously it becomes a bit different with execution layer Awards but again a lot of the execution layer awards are luck based like you know if you propose a block that has a high uh payout a lot

of the time it's just luck right and and that's why you look at these things over the long term and and kind of like average out the APR instead of just taking the APR from like a week and extrapolating that out because that's not going to give you the full picture because maybe you get really lucky one week and you propose like a really high reward block maybe you get like one eighth worth of rewards and then the next week you get like 0.1 each that's not you know that's not a good way to look at it so you need to average it out across like a long time period such as a year to get an accurate view of execution lab rewards but consensus layer rewards are easier to to calculate because it's just based on the issuance curve but at the same time there are

other things that impact that as well such as what we saw at the finality issues right the inactivity leak such as potential slashing slash such as are you going offline right if you're if you're offline uh for like a week you have to you you'll net out to zero if you are online for a week after that but that's a week worth of rewards that you kind of missed out on right you have to make up for over over time but you never really make up for it you know you're kind of always chasing it so and then obviously our tester uh or validator Effectiveness affects the rewards you get as well so there are different things that play into it but just generally consensus layer rewards are a lot more stable and a lot easier to kind of map out because you know the issuance curve than

execution layer rewards are because execution layer rewards are highly dependent on the ethereum fee Market not just because of the test but also Mev if the fee Market is is at like five to ten way then Mev opportunities are going to be a lot less than than other one what they are at 50 Gray because they this but you know that's obviously a sign of network Demand Being lower than at 50 Gray and then your rewards are going to be lower because of of that there so yeah that that's the rewards thing in a nutshell but as I said uh if there is a cap put in place which I think there will be eventually it'll be a soft cap it won't mean that no one can stake Ethan anymore it just means that uh the APR will will definitely be lower because I mean at 60 million each stake

the consensus layer APR is pretty low and then obviously the more uh validators there are the more the execution layer rewards get spread out because the less chance you have of proposing a block per validator uh and then it takes longer and longer so you know with the current amount of validators on the network I think it's like every three to four months uh you have a pretty good chance uh or a pretty high chance of proposing a block per validator if you double that then that doubles the thing right that's at six to eight months now of proposing a block so your execution layer rewards also go down so just overall the API goes down the more ethi staked but in saying that the whole logic behind it is that okay more E30 staked and locked up and not not doing anything and he's actually um

you know uh is actually uh been bought to be mistaked the more demand don't eat the asset the more people view it as a story of value the better the eighth USD price is the more valuable it becomes and then it just kind of works it out like that but you know it depends obviously Plateau at one point where people are gonna be back it's not worth to stake it at this certain APR but then you throw other things into the mix like restaking and other incentives that can be layered on top of ethereum staking it gets crazy from there so there are a lot of unknowns uh in terms of what exactly it's going to look like but we can assume things based on economic incentives we can see things as they as they might play out in the future and make some educated guesses around it but

do know that uh it's almost certain that there will be a cap implemented at some point I think it will be sooner rather than later just because I think the the researchers in the devs and the community really at large wants to do it before we get to the point where we would have kept it anyway to avoid any unnecessary drama or controversy around this because we don't want to cause controversy in the ethereum governance process that's something that we want to avoid because even a small amount of controversy around something can stop something from going forward and for example if there is an EIP that's really popular but like say I don't know I thought my head 80 of people want it and 20 of people don't that 20 of people have a really loud voice and it means

that it's more likely to to not go in than going and that's why we say we need a super majority we need like 95 plus percent we need almost everyone to want this upgrade to go through and I think we have a better chance of that happening if we do it sooner rather than later and so as I said keep we're going to keep an eye on this I'm going to keep you guys updated on this as I see more stuff coming out about it but if you want to see one of the best summaries or still the best summary of the thinking around this uh you can check out vitalik's post which I'll link in the YouTube description below all right so super fears had a great tweet uh that I think was very relevant in the context of client diversity and the finality issues that I was talking

about he said it's cool to know about client diversity but it's really a 2022 threat the real threats in 2023 are any entity who controls more than 22 of validators and be careful they'll play games to hide this and block builder centralization for Mev rewards so I fully agree with superfuse series that client diversity it's not a solved thing nothing I don't think you could ever consider it solved but it's in a much much better place than it was last year and it's not something of a critical threat anymore I believe obviously the execution layout client diversity needs to get better but I think it will over time um and I think that we're not at that critical level because gas doesn't have over 66 of the network right now so it's it's actually um you know not at that not critical level

um but in terms of entities controlling more than 22 of validators well right now there is one uh kind of uh I guess like group of entities that do that call old lighter now I think what superfizz is referencing here with uh uh uh you know someone like lato trying to play games to hide this is that Lido and I guess myself as well and a few others out there will say that yes lighter has you know what is it 33 or almost 33 of validators on the network but they distribute it out to 30 different entities while that is true and that is better than just like one single monolithic entity like like a centralized exchange having that percentage Lido as an ecosystem is still a monolith because Ste the the staking

token the stake the LST the liquid staking token you get is still just lighter right there's not it's not like you get an individually different um LST depending on the entity that you're staked with so it is a monolith it is a cohesive kind of hive mind ecosystem even though it distributes the stake out to different entities and even if even though it does distribute it out to different entities it still shouldn't have as high of market share as it has now lighter withdrawals are going live I believe today actually the 15th of May yeah they said they'd be going live today so it's going to be interesting to see what that looks like I'm not expecting a mass Exodus or anything from Lido I have said it before and I'll say it again I think lydo's market share is going to be taken away by other staking

providers just offering better Services over time rather than people Mass unstaking from Lido and going some you know to an existing service or going somewhere else I think it's going to be a slow burn it's going to be a slow-moving ship but I I don't think that lighter is going to maintain such high market share given that the opportunity for creating staking products and especially getting whales involved and getting those that massive amounts of capital into your own products is so large especially because there's going to be whales that are currently staking with Lido that would probably prefer to stake with more of a white glove tailored solution that I don't think lighter is going to be able to offer if they want to be a decentralized staking provider right if they want to go towards a more

community-owned provider so there's going to be a bit of stuff happening there but I fully agree with super fees um that you know no entity issue control over I mean 22 I I I think I know how he got to this number but I I think that the number itself isn't the critical portion here I think what what what's most critical is that we don't just have one validator that has so much more market share than others so for example if there was oh sorry I should say entity is not validator so for example if there were two entities that had you know 22 market share each that to me is a lot better than one entity having you know 33 even though those two entities equal 44 just together I still think it's better than just one entity having

that in saying that it depends on what the entities are if the entities are known to collaborate with each other while there is obviously a risk of collusion there if the entities are known not to collaborate with each other and like they're totally separate uh so for example Lido and rockapool right I don't think that the lighter validators and the rockable validators would decide to collude so there there is a lot of social differences here you have to consider as well in saying that I don't like I ideally I don't want to see any provider above even 20 uh because I think that's that's quite High uh and I think that we can eventually get them all below that but it's going to require competition it's going to require our patience and it's going to require a lot of work on the other staging Service

Parts to take that market share uh away so uh and the second Point here about block builder centralization for Mev rewards I think this is less of a of a threat uh just because there are a lot of solutions already in the works for this obviously we have in protocol PBS there is things like Mev burn um there is a lot of research about around Mev already and there are a lot of there's a lot of work being done on it uh but the current state of Mev with having like that sidecar software with Mev boost obviously isn't ideal but I think that block Builder centralization generally has uh I mean I think it has been solved at the research level I think that we just need to implement it and see how it plays out but generally I don't consider it as big of a risk as I

do uh these entities controlling large amounts of validators so so yeah but anyway good of super fears to bring these up very important points uh so thank you to him for that all right so the ethereum.org website has a new section and Page dedicated to user research and user experience in web 3. well and Jacob here says while this is just one page the hope is to improve and grow over time and also you can contribute so this is pretty cool I think the the focus on just design UI and ux in crypto generally is not great I think that there isn't that there should be a lot more focus on it we should be building products that are much easier to use than what they are right now it is getting better over time

obviously it's a lot better than what it was even a couple years ago but still I I still think the user experience generally is not user friendly at all uh for for anyone new to the ecosystem they have to learn all this complicated jar again all these complicated things a lot of them use metamask which is one of the least friendly interfaces for new people um so yeah we can do better there uh not just with wallets of course but also with just the various different interfaces for interacting with the D5 apps for example and things like that so yeah you can go check out this page for yourself if you're interested in that and you can also contribute to make the page better all right so Brian Armstrong has announced or coinbase has announced the formation of of coinbase's global

advisory council with a bunch of different people in it such as uh Pat Toomey here who is a uh a former U.S senator uh there is Tim Ryan here who uh I don't know if all of these people are politicians but they are definitely involved in politics you've got John azalone here uh you've got uh Sean Patrick Maloney here and also uh uh Chris lehane so I believe that all these people are involved in the Regulatory and just politics side of things in the US so they've joined coinbase's Global advisory Council obviously to help coinbase with all things policy and regulation I'm very happy to see this I think that it's about time that the Empire Strikes Back at uh the SEC so to

speak even though the SEC is technically the empire in that analogy right um but but like you get you get the kind of like analogy that I it's about time we strike back against this big evil that has just been towering over crypto for for so long now and just been very unfriendly to us and I think coinbase is really the one leading that charge right now uh and Brian is definitely the I guess like best person to be the face of it because he obviously is the CEO of coinbase and he is uh you know obviously an American citizen code base is US company and uh he's the co-founder and CEO so I think he's doing a lot of great work on that front not just with this Global advisory Council but everything that he's been doing especially over the last year or so but you can go check

this out for yourself there's a blog post TR I'll link it in the YouTube description below all right finally here eth Global has announced the 10 Heath Global Lisbon finalist so they're a uh there are a bunch of amazing projects here that you can check out from something called credential Corgi to crypto Bureau to chain GPT all of these are different projects uh use uh kind of uh I guess a different Primitives like uh zero knowledge proofs for example uh AI tools like chat GPT and other such tools like that so you can check out the full thread here for a breakdown of each of these projects uh which I'll link in the YouTube description below but it's great to see all the energy actually coming out of um eth Lisbon here with global Lisbon

I've been seeing it on Twitter seems to be a great kind of Builder focused conference uh it would have been awesome if I was able to go but traveling uh to to Europe is uh is definitely not uh not an easy uh travel for me from from Australia here uh but it's been great to just see everyone come together and and build and and basically uh essentially uh create new products for the ethereum ecosystem and things that we may not have seen before so yeah if you're interested in checking out all the projects here all the finalists I'll link this thread in the YouTube description below for you to do so uh but on that note that's going to be it for today so thank everyone for listening and watching be sure to subscribe to the channel if you haven't

yet give it a thumbs up subscribe to newsletter join the Disco Channel and I'll catch you all tomorrow thanks everyone