AMA with dAppLion | EPF5 Office Hours

all right good to go awesome yeah thank you so much for coming everyone especially uh lion here thank you so much for joining man it's an honor to have you here um and uh yeah so we have another office hours with AMA in past two weeks uh three weeks we've been doing presentations for uh from uh you follows and now we are back to the uh sessions with mentors where we host AMA where we host these uh there where you can uh chat with a mentor where with some somebody from the CFS and uh get to know them uh get to learn some technical uh details of their work but also about them as a person learn about their background and ask them anything so I hope you prepared some cool questions for de liion because we have very cool guest today um de liion is um uh uh great mentor to have here because he's been as right now deployment Works in Lighthouse uh before he has been leading the development of load star and uh generally he's been contributing to the research around the consensus layer he's also DJ so we have many things uh uh uh he's definitely one of the best people to talk about um talk about the theum consensus layer or etherum development in general yeah um yeah thank you so much for being here man it's honor to have you here I have light on right here physically next to me so it feels weird that I talk to the camera and also to him but yeah thanks for joining man thanks for having me yeah awesome and uh yeah before we get to dma before we get to um uh the questions from the audience I wanted to introduce you a bit uh get get uh get uh to know you a little your background your journey so maybe if you can first tell us about yourself how did you end up working on theum brought you to the core development for sure so yeah you mute in the same room it's a bit chotic um so I started in 2018 I randomly found uh the people that started D notes and if you know that it's kind of an old project um start by Jordy viina and Griff green two legends and from there yeah it sucked me in I did a bunch of contributions to things like Strong City Aragon it's mostly like a layer uh and then the beon CH became a thing on early 2020 and then the time I joined load star which is the JavaScript client cont client at the time the Bon chain hadn't even started and it took some time for the client to become production ready but uh read that and then as I Got Confidence with the specs and everything I try to contribute as much as possible to eips research and other things that have uh more wide ranging impact and as of January this year I switched to Lous which is the r implementation also consensus so yeah I'll be happy to help I'm very knowledgeable with consensus um consensus P2P s um future research things if you have execution questions I can try to help not my expertise but yeah should yeah whatever should it yeah awesome uh yeah that's where dep lion comes from from the dep node right um so yeah that's it's a really cool background um yeah so and now about about present actually what uh so and you've been yeah you changed the team uh big betrayal there came to the lighthouse so yeah what are what are what are you working on these days can you uh can you tell us more about what you are focusing on uh currently with what sort of work yeah late last year as I was joining Lighthouse um I was contributing on the main features of the Electra hard fork and so yeah that's Max CD mainly but that thing kind of got wrapped up uh research wise so implementation was taken care by other people these days I'm Focus uh I would say 80% on sampling specifically pure uh which is the first type in scalability that hopefully will Le Electra if nothing goes too wrong um it's kind of something very new that requires a lot of testing um yeah it's kind of new frontier for us so it's it's we hit we're hitting more Ro books than expected so lots of debugging um yeah besides that just we have a uh maintenance and just taking care of the clients now that we have some time for the fork because yeah you want to reduce technical depth before it catches up to you so that's mostly it yeah awesome yeah thanks for sharing that and um yeah before we jump to the AMA questions I have at least one more to get some Bigg ground on you um I'm wondering what were some most challenging things in your developer past when working whether Lighthouse load star or something else in the uh in the core development which you've been uh what youve been working on something that will challenge you something that uh uh maybe you um um uh you uh you had to learn from uh or yeah maybe something exciting that that you find is highlighted in your past yeah can you can you share something uh you've been working on before here yeah so loar one of the main themes that we found through these years is that we are kind of fighting the language it's um if you want to be a highly performance system application written in JavaScript it's kind of a bit complicated so you have to do lots of STS uh so finding all those little things that we did was kind of very exciting at the same time um but difficult and always feel like we have to catch up because the the problem with the pon chain is that the state just keeps increasing like that is not tomorrow so that's mean more data more man withd more state everything gets more complicated so you always have to catch up and keep doing uh optimization just to survive on the network uh on on Lighthouse um they have quite mature networking layer but still uh it's such a complicated piece of technology that uh any issue that we find debugging it it's it takes so much time so I spent a lot of time figuring out how can we have like more visibility on what's going on uh so that we can find bags in one hour instead of two days and we can do that now which is exciting but it took a lot of time to understand all the interactions and things like seeing has an insane amount of edge CES like it looks simple on on Surface but there is so much interaction between different things that makes it super complicated so you need to have kind of a very good knowledge of a lot of different uh F yeah okay uh you have to have a good knowledge of a lot of different pieces of code base to really understand how they fit together otherwise you will not understand these Ed cases so that that's kind of um very challenging yeah that's interesting it's definitely change of environment coming from uh coming from uh typescript where you need to fight with the language itself to like complexity rest which can be efficient but comes with other complex problems yeah folks uh if you any questions feel free to raise your hand free to uh prepare a question for lion I have I have few of them but I want to give uh give you guys space so we have Kata yeah hello de line can you hear me yeah yes I can hear you yeah yeah my project is about pure dust specifically P dust metrics so I wonder um how developers and research team um go with the results of testing for example for currently it's like a testing period and um how do you how do you deal with the metrics how do you deal with the results uh do you use graphon dashboards or any other analytic tools um yes so grafana and so if you have good metrics prome and grafana all you need that's it it's just basically the most important thing is coming up with metrics that are important uh to whatever you want to the back and always Mets are a tool to answer questions so for example right now what I'm working on is we are running these death nets for p locally and the network just dies after two days and we don't know why just we have no idea so um I'm trying like right now I'm trying to figure out what metrics do I have to add to ask the questions I want to answer which is basically why did it die uh so if we if you don't have enough visibility then so it's it's trying to come up so for example um what we discovered is that the the nodes suddenly they ban each other and they disconnect and that's that's why the network dies so from there you want to ask why they did disconnect uh when they started to disconnect so you maybe metrics um like uh so whenever you disconnect the P it's because the score goes down so you have these events of down scoring you want to have metrics for those and then for each one of those imagine that they are sending bad data so you want to ask why did they say send the bad data you can have metrics for that so it's kind of um thinking about what type of for instance you want to answer and then creating metrics for them and then having some dashboard that you can consume the metrics in an efficient way to Just DE back yeah the one reason I found recently is the uh data dis uh over overflow like yeah it fills up after two days and at least my cortosis is falling down okay uh thank you so much it was a great question um when the next one is George hi um thank you very much for joining us today uh so the first question I want to ask is uh because I read quite a lot of article written by you and uh uh you you care about the the solo trade a solo Stak quite a lot uh and together with you and others uh um so I think you you also care about the economics in terms of the the healthy of the network not only about the scalability so my question is uh with the given chance of uh raising for for example risk taking um protocols uh some of the decentralization some of the centralization of the um uh risk um could happen and how we going to prevent I want to take some opinion uh from you uh uh anything you are working on or how you thought about this kind of centralization risk uh so that's question uh one another one Mara if you don't mind I ask a general yeah I don't have good memory so let me answer question one and then you ask question okay yeah um okay yeah okay so what if I understand what what you are asking is uh risk taking is is a risk a risk to the set because it can give an economic incentives and economic advantage to uh either entities that are concentrated or entities that have good connections uh such coin b or uh Etc which and that is true um I think the the end game for that that's a very controversial topic is uh targeting so it's a combination of me burn and issuance targeting that way we can so the protocol can capture these anomalies like uh risk taking and absorb them so you don't have this variance between Sol of takers and concentrated actors uh that that hopefully should be enough uh but it would mean that the the role of producing blogs which is what have most variant would be uh abstract away from um from the current set of validators who will only attest but that's what's really important for uh digitalization uh but yeah to get that it will take some time because we need to figure out uh M bur which is a which has the as the dependency some form of epbs and we need to have inclusion list main sense of resistance uh yeah and what what's a bit scary is if there is a reaking protocol that takes into account Stiers sorry a testers or something then that's going to be a bit complicated to absorb into the protocol but it's a bit far H like I think there is a lot of hype with angle layer uh that still have to be proven to be because I think as of now there is absolutely no yield that you can get from so we's see hopefully we have enough time to figure that out so that's basically one side the other side is uh ANP I have been working on with Tony which is uh 7716 uh and that one will give a slight economic Edge to Sol of takers the way it's going to work is it will alter the way that uh in activity penalties are calculated so if you are a concentrated actor and have a bunch of servers that go up and down at the same time because you have all of them in Amazon or all of them in the same data center or something then you get penalized more so on average if you are an independent actor you would make uh more attested Revenue thanks to that it's not g it's you're not going to be like making double or triple but it's maybe like 10% or 20% so and that can help even out a bit more the economics and make like Source more sustainable yeah so thank you the reason I'm asking this is because I'm working on maxv and we just speak uh in Discord a few hours ago uh so that's relating to second question um because the the the keyword I remember from that conversation is to go deeper so uh in your experience um by working on EAS um is any kind of challenge time you've been facing uh and how you overcome overcome uh the challenge or the difficulty uh uh in your in your study or in your work here with E Yeah so sorry are you asking about like the same question about the calization pleasure of maxv noly about no no is about a general question uh uh to uh share some kind of uh uh learning experience uh how you um uh go through uh some of the challenges you are working on and then uh uh you find out the the solutions uh how you overcome uh uh those kind of difficult period yeah or or how to go through the uh you know um to resolve the learning curve uh from your side yeah okay got it um I guess for maxv I have a special case because I I design it together with Francesco and mik so they did most of and they so um if basically it was me well actually Mike started maxv creditor to him then mik joined then I joined and then franccesco joined uh but basically the three of them they are uh really smart and have really good idea so whenever we had some problem we just posted on the chat and someone came up the solution within a few days and if someone didn't understand they would just take time to explain so I would say people is the best way to unlock uh you in that situation so if you want to learn about maxv and you get stuck just either ask me or just find someone else that wants to learn Max with you and you guys can like P it out and try to figure out that at least that would be my cheat yeah cool thank you yeah awesome thank you so much it was great uh we have done yeah hey lion so uh still in the context of somehow of uh the before the my uh fellow before uh my project it's uh about adding delegations in protocol which is yeah not trivial you you reviewed Po's uh PR of delegating validators back in 2021 but that was actually just transferring access balance to another validator but you from what I saw in the pr you were uh involved in the in the review review of that uh PR do you have you know in in general and in the context of staking and uh rainbow Stak framework and now orbit ssf you know all these new uh researches that research papers that basically come and shape the the future of what's coming do you have do you have an opinion of of how or if we should add if delegations in protocol I mean I started with with rainbow staking framework but now it's a bit you know L put aside for other means to rotate or to select or to diminish the validator set selection including including using maxb and consolidation incentives and stuff like that so in this context I know high level we cannot enter in detail now but I'm very curious of your of your input yeah I think um so about delegation we we should be pragmatic and accept the fact that delegation exist today in eum through entities like Li and it's just a terrible delegation it's absolutely terrible so accepting that reality protocol yeah well and it not not not only is off protocol but the security factor that you get is absolutely disgusting like the people running notes have very little at stake compared to the amount of value that they are handling um yeah it has all these negative externalities it's it's pretty pretty bad so acknowledging this reality then we may want to consider what can we do like from the start there has not been any delegation mechanism uh first for Simplicity and second because we didn't want to kind of enshrine a specific way of doing delegation and the may I ask I'm I'm sorry to interrupt but I've heard this position before and I've asked around and I didn't find an answer why did you know research at beginning didn't want to okay Simplicity I understand we aim for Simplicity but what was the reason if you can tell me one reason why there was no wish for enshrining I mean there was there was a wh but the problem is that there was no design it's kind of Now with uh epvs everyone wants to get rid of MVB boost but there is no there is no design right now that everyone feels um really sure about so ins rining something that's not uh long-term sustainable is even more costlier than not doing anything so right right you have to consider at the beginning the most important thing we have to do is do the merge that's the most important that has an insane amount of benefit and we have to do it SE so many years we had a design so delaying the merge to wait for the I think would be a mistake so that's why it wasn't it was not made if there had been a solid design that was future proof and secure and blah blah blah everything then it would have been considered but that never showed up right right and second question if possible more a bit technical um you mentioned the growth of State The Continuous growth of state do you see something like stable container serialization something that we need to to get to because inevitable even with our projects we get to the point where we have to add new uh uh new lines in the containers or we need to modify existing containers add containers and that gets to growth of so stable container any any thoughts is it something that we're going to have yes so just to clarify stable container addresses an important issue but it's it's different from growing the state so when I talk about growing the state I mean in terms of of size in terms of B size so that means that clients need to have more d space more R more bandwidth what stable containers address is the the list of fields actually growing and that breaks uh the merization of of the state so if you want to make proofs your breaks so for staking pools or any sort of onchain protocol that wants to prove that data belongs to some Roots they need to have governance or something and with stable container they could just hand Cod it and forget about it forever which is a nice property so yes we should do sa con the question as always is when because the the next hard Fork is also very full has a lot of things so it will either come this fork or the next one yes we will do it I think we should do it probably thank you yeah very well we have another question from you K yeah it's about P again so sorry so now we have um I can see uh my blob transaction in the Explorer I can see all the data and what will happen after puras updates so uh could I see my um transaction data in the Explorer for example uh or on the commitment how it will work yeah it's a good point um if so after P the main feature is that we will have sharing so if you want to you will not have all the data that means that you will not be able to retrieve blot transactions if your note has some sharding coefficient greater than one you would need to either run a what we called a super noode which has all the data or you would have to go to third party providers like blob explorers or something like that to retrieve uh your transaction so yeah I think we haven't yet but maybe if you have some sharing then the not will be like 404 we don't have the data sorry yeah so uh yeah for example e scan it will just run super note and have all the data right yes yeah basically the expectation is that all Explorers uh applications that need High degree of security like exchanges high level applic like application running uh I don't know high value things uh rollups basically anyone that has a lot of stake in the network should run a super node out of selfish because that that would give them the 100% assurance that that block is correct so they can act on it it will be more for yeah it will be more for like uh sers or people that want to run nodes at home uh that they don't have to have all the data if they don't want to yeah thank you yeah I cannot wait to run my super archive note um yeah uh thanks thank you guys um we have a question from hangang yeah hi so uh I'm currently working uh on improvements uh grading uh implementation in uh P implementation in grin uh but uh like you said in like you mentioned in the previous question uh like there are some a lot of question lot of problem with the networking including like a sying problem uh so my question is uh would like to not uh asking about that because you uh have been mentioned this already uh so my question is regarding on your uh proposal on the last uh HD C on the beon blocks uh by rank ver uh version three so could you explain more about that and how it is different from the previous version yeah thank you okay just to give more context uh what hangan is referring to is the a proposal to add a new RPC on the consensus b2p so when on the classic spec when you want to sync you issue a request that's called blogs by range and you tell your peer hey give me blocks between slot 1,000 and 1,000 then and that the other paer would just retrieve these blocks on the database and give them to you because you are requesting only one slot uh you can only request blogs that are canonical uh so what happened in Lighthouse uh I was working on sync and making it more resilient to unstable network situations and some case that we don't handle now is if you have a long Fork so imagine you have some old block that's finalized and then one fork with one 10,000 blocks and then there is a second Fork that has also 1,000 blocks and you want to sync this this Fork you cannot do that with this method you need to you need to find a PE that has that that chain as canonical and then request it from it because otherwise the current RPC doesn't have enough expressivity to ask for that chain and also you don't know if the pier is going to be switching because it could do that it could have a reor and switch from one network to from one chain to the other and then you are requesting something and you get something completely different and then you could be like hey why are you giving me garbage blocks you s these blocks and then he gives you these blocks and then you don't SC each other just a m so what these new uh LPC does is you ask for a range of slots but also you ask from which Fork you want them that way you have the assur that the P understands what you want and you can downcore it if it gives you garbage and it also allows you to sync this side for from everyone visic thank you um so yeah that that's the motivation I think that's that's all yeah yeah thank you thank you okay thank you so much um yeah folks feel free to get your hands up um I have one more one more question about pias actually um one drama I wanted to touch on maybe you saw the Tweet by Peter recently uh accusing the CL or the researchers in general of uh destroying home stakers Etc so the thing is that with pias we are able to parse more blob data but uh we still have the limitation of the E which needs to like handle Big Blocks so uh how do we solve this like is this can you can you shed more lights into like Peter's words like whether this is true whether uh whether there are some other limitations with the uh the the uh you haven't seen a tweet by pet you know what I mean okay then we are going to play telephone so I don't know what he said but I can I can just bring up other criticisms that people have said and not TI to Peters specifically so one one issue that is is true is when when we are Charing uh you don't have to download all the data to a test so as a as a ster you have mainly two roles you you basic vot on what's what's canonical that's a testing but you also have to propose blocks uh sharing affects only a testing but it doesn't affect proposing because the one proposal have to uh create all the data and send it uh and that's an issue because you need to have a lot of um bandwidth to upload all the data otherwise your blog will not become canonical and that that's true that's that's an issue but we have a solution and you can so actually today what some Builders are doing like flashboard is they just refuse to include any BL transactions whatsoever which is kind of stupid because they should be able to do it they are serving big players but they are kind of greedy and just decide because it's not worth it block transactions and paying off and just ignore them completely so solo sters could do that if they want to like if this becomes a problem they could just um ignore blood transactions altoe uh and a better solution is uh when you include a BL transaction and compute all the data what likely happens is that that block transaction is already on everyone's men pool so what you can do is uh tell the El to have some special block production rules where you will only include in a block a block transaction that has um a high confidence that it has already been propagated in the whole network so for example wait for a few seconds that you have seen a transaction to consider for inclusion what that would mean is everyone else now instead of waiting for you to upload the data they will just go to the men poool retrieve the block transactions and recompute the the data columns that you need for PS and there could be even alteristic notes that they do that uh always and they just broadcast the data columns uh to everyone so you as a solo speaker you only have to decide which SL transactions to include but you don't have to propagate the data you just tell the network hey this is my selection and then someone else will uh calculate the columns for you and upload them to the network uh so Lighthouse is already implemented this for the net uh so it there is no this recomposition but at least you can consider blocks for head faster but we could totally do that for uh p and I think it I don't see why it shouldn't work interesting so there are other trade-offs that uh uh solar seers can do um yeah thank you so much lion uh yeah sorry couldn't hear me there um guys do we have any more questions for lion can you can you raise your hands uh yeah we have ch go ahead about uh um like uh other nodes just using the blob side cards that they have in their meol already uh like in light of like one DP D wouldn't encoding have to be done and kcg uh kg proofs have to be calculated like a single entity uh and then distributed with uh like you know like you have to calculate for each cell and then distribute that seed that through the entire network right so it becomes kind of like the computation itself becomes kind of centralized where proposer has to do that or the Builder has to do that to see the entire network with the S okay if if I understood the question you are asking um I'm not sure if it's a problem of who will do it or if it's an altruistic problem of who will actually do the like spend time calculating the kcg proofs and the polom extensions is that correct uh yes like uh because so my my it's not a so it's a doubt per say where uh like when that has to be done by a single person you cannot uh anymore use the side cards from your me pool directly right because they have to be encoded and someone has to commit to the encodings right or so anyone can compute the encoding as long as you choose the the right l transactions so if if you get the block and the block say that I'm using this block transaction and this other block transaction anyone can compute the extensions and anyone can compute the proofs and they will always be the same so if you use a builder and you commit to a specific header the Builder can compute the extensions and the proofs for you and similarly with the trick I explain now uh if you have the blob SS from the execution client on your local machine and you know which transactions are the right ones that go into the block you can also basically anyone can uh compute the like the data column side cards or the BL side cards oh okay okay yeah that answers my question yeah wait someone someone asked on the chat about she said solo Can't Ignore blob anymore if Els end up being valid for n plus1 inaction that is true uh but I think so far the El design does not include blood transactions which is kind of annoying but that's what it is but it's important to understand that ethereum wants to move to a future where producing blocks would be really really expensive so that's why in order to realize full D Shing potential with 32 megab block we have to have some form of epbs so that's kind of the end game anyway go ahead D just wondering because you also mentioned that there's no design that for epbs that has you know like a big like like a large consensus around it you know putus has this uh uh PR open uh 7723 32 uh how do you see that design uh especially intersecting with current El inclusion list uh design and and going towards uh uh sharding uh are we going to be you know to have to renounce the Builder uh uh roll all together and just do APS I mean just you know auction the the right and and don't actually involve Builders anymore I don't know um okay it's maybe a not very well formulated question but I hope you understand uh where I'm going yeah to be honest I haven't catch up with the EIP specifics uh I was familiar with the design as of five months ago so I'm not sure if something has changed um if it's the same that was before I have some it just on impl it just leaves it just leaves relays uh aside yeah but they're still Builders only their Stakes they are staked but they are still it's also hard like you cannot it's very hard to do a design that completely eliminates relays because someone if there are enough incentive someone can do some off protocol solution that will always be possible okay that there will be the obviously uh you can choose to use relays uh uh even with the the current epbs but there is an option to not use relays and have this in protocol exchange of of uh of payload so as far as I understand epbs allows to do an A trtic change between a proposer and a builder through the protocol but you could always do that off protocol I think it's it's going to be really hard to design a system that makes that unfeasible or at least it will always be feasible but maybe we could make it uh not as much economically viable so that most of the block production flow happens through the CL that we want right right okay so and and that will go well with sharding I mean as with a fundamentals I would say so I think it's orthogonal okay I I think we'll have to rewatch The the recording of this session just to take some info okay so an example would be uh EIP 155 n yeah and then flashboards came along and they did all of these private mle where you could submit transactions with zero gas um so with e19 we were forcing everyone to at least burn that fee so that that would be an example of a protocol mechanism that defeats a specific usage of these of protocol channels yeah word would be a a way to capture this revenue and uh make the off Channel not as much competitive but it's it's still going to be complicated like May word is a bit of a pain ass because the block rewards are so unstable uh that are harder to capture than than transaction fees okay thank you for the exploration yeah awesome thank you guys uh yeah K again yeah do we have time for questions yeah uh my question is um when working specifically on some part of the protocol uh do you need to to follow uh all the other parts for example you're working on the heras do you need to follow all other updates like BBS Etc how deep do you need to understand it uh yeah no there there is absolutely no need uh to follow all the things I think we Thrive as a human species with uh specialization if everyone had to do everything and know everything we wouldn't be very inefficient like so there is value in specializing you can I don't I don't know how many hours you have in a day but I am not able to keep up with everything and I I used to try to do that or at least choose maybe two or three itas that are interesting to you and just keep up uh if you can provide value to them then it makes sense but otherwise you should be wory because if you try to do everything maybe you end up doing nothing um so like an example would be epbs I used to follow that very very closely but at some point um the other require more time so I had to let it go and now I'm not up to up to up to date but if in the future um I don't know I see I can contribute then I would try to catch up but maybe I would have to drop something else that sounds good because yeah when you see all the specs it's a bit scary thank you yeah the spe the specialization is very important in such a complex system as ethereum so that's why we encourage you guys to choose a project focus on one thing during the cohort uh do we have any more questions anyone uh many people haven't haven't asked anything and you can also you can also type in the chat if you are you cannot talk for some reason so yeah let me ask um let me ask a question uh Dan uh yeah um what is your most controversial take in ether you do you have uh do you have any anything that researchers or devs would find controversial something that's uhbe recently discussed or overall any controversial takes for us you want to mute yeah so something that's kind of a hard field to to swallow is that the the solo Staker will only Trend downwards from here in my opinion unless something fundamental changes in ethereum the economics of scale are just too weak so with the new issuance debate what people get really angry about is that um we want to make it the change that will diminish the economics of Sol staking but the point that we want to bring up is that the no action so what we have today will also do the same but even worse and solar sing has always had this I don't know like altruistic premium which people don't like that but it's true like most of the people that run nodes at home they have strong beliefs um that that matter and that makes them make these decisions that it are not perfectly economically irrational is the same way we we fight for the things we want we sometimes take action that are not perfectly economically rational because we believe in something or we want to do something good uh so I think that most of the like the most of the so sters have already been onard that and we will have more people that have the same characteristics of this like altruistic profile um that makes it makes them do this decision and because they have this premium if we alter the economies a little bit it they will still be here so that's why I'm contributing to EIP 7716 to make them more economically viable so that we kind of offset the changes that we have to do in the future but yeah um if if you just want money you just put your money on a state on a on on on an exchange and just fake that's just cheaper for you uh it's true that there is like CH risk you can get screw and if you solo stake you have full sovereignty of your money and your decisions but that's very hard to quantify and I I don't think the average Joe that a thing and no no one knows how to put an number so it's hard to put it on the backet of economical rational decision but yeah I don't know like people when I say this alistic fact that that's what it is and it's a beautiful thing I don't know I love it um yeah thanks thanks lion um we have a question in the um in the comments uh how do you guys prevent any attacks and do such a robust testing on new features thatum is not being hacked till date just well it's been hacked for a bit there uh just the contracts though um yeah we have a we have a whole thing on testing with Mario and uh and Pari and uh we had a we had a bunch of sessions from that but do you have something you would like to say about uh uh testing the robustness like before the merge the test thing ramped up so much and I don't know if you have any anything to sh there I it has been hacked yes there was the there was an incident um sometime ago I think someone from trle or like some team did that sent a transaction that split the network like created a consensus back uh that was really serious uh but thankfully it affected a minority clients uh that was uh the sorry that was the nethermind when nethermind fored off in 2021 or uh I think it was after London right yeah yeah this one it was optimism not trle was the optimistic guys who play the network uh also the Bon chain we have had uh there was an open pack on the four choice that was really really serious you could reor oneir of every Epoch continuously and just mess with the network really bad uh but thankfully no one find it before us so we kind of silently patch it for a few months uh we were just lucky that no one expl it so how do we prevent any attacks we cannot prevent all the attacks but as as Mario was saying um we just do we do everything so we have testing uh lots of people reviewing all the code in every client all the clients have a lot of very very experienced people uh and they are able to C this back we have a lot of smart people looking at the specs doing spec tests we have a lot of fing uh both for like the specs there is this really interesting project called uh fossil for the forers to basically V and create this I don't know one billion space uh we also F the networking layer the civilization protocols uh we have differential fing also on DM also fing which has got a lot of BS uh the what's what's the name for the when we have this um project that we pay money if you find a p the back bage we have really really good back bage with I think up to a million for back which is nice so yeah just try to have as much colage as possible test this test and if something happens hopefully fix it before someone explo it okay thanks a lot here it's it's always good talk for testing um Josh you have a question uh yeah so I have seen some recent conversation happening around uh ethereum research happening in the p in the public in a way that's you know challenging for a lot of people to understand especially if you don't have a lot of the you know in speak or the the sort of like you know all the acronyms and all the different like little kinds of ways that researchers are talking to one another um and so I'm ious like how do you feel that like the working in open ethos affects the public perception of ethereum um I think I am I'm I'm sorry so working on the open personally I love it I think it's awesome it's also very convenient because if you forget your laptop you can join the meeting from your friend's laptop and you don't need any credentials or anything it's kind of very convenient and and you there is always a YouTube video people talking about it like to be honest for most of the calls my I do not attend unless I have to and then I watch Like Christine's summary or someone else's summary if they talk about something that is is of Interest then I watch the replay so it's it's convenient uh and you have to understand that things will be complicated and that's that's one of the beuty of Open Source you not everyone has to understand it but as long as someone that you trust can understand it then we are more secure so imagine that signal is open source and they try to publish an update that's malicious you will probably not know about it but there is some Swiss organization that cares about freedom of the press and they look over what signal is doing they will find it out then raise the alarm then there will be a public abar and probably that this thing would be uh roll back so that's kind of the same uh and I think we don't need everyone to understand it but we just need enough people to understand it to have us accountable to what we do I agree with that but what I guess I'm kind of asking is that you know like there's a lot of conceptual stuff that gets discussed on like crypto Twitter or on warcast or you know these sorts of things that are very very new Concepts that somebody who maybe just dabbles a little bit in in blockchain stuff or you know is a dean and and aping into a bunch of of weird like sees um you know some of this conversation that might be just very very idealistic or or you know just in a very fledgling phase and and you know might take that information and say this is what ethereum is doing because one guy said it on crypto Twitter or whatever you know so uh like you know the open source part of of it I agree is is a great is a great thing I'm curious more about sort of the research stuff that isn't necessarily stuff that has been agreed upon going into the protocol or um you know is still sort of trying to be worked out uh conceptually like what sorts of uh effects that that might have on uh on the public perception of the actual whole community and ecosystem