All Core Devs - Execution #82

Transcript

[Music] [Music] [Music]

[Music] [Music] [Music] [Applause]

[Music] [Music]

[Music] [Music] [Music]

[Music] [Music] [Music] [Music]

[Music] [Music] [Music] [Music] alright hello everyone and welcome to

meeting number 82 of the etherium core developer call i'm hudson i will be your host and organizer today and today is going to be a little different because we're not talking about a ton of subjects it's generally gonna be two subjects and if we have time James Hancock has proposed a few other ones including stuff with a difficulty bomb if I'm not mistaken so the current plan is for this to be a two-hour call the first hour is going to be dedicated to the BLS curve signature stuff that's we need to get ready for earth 2.0 and the second hour if it takes an hour is gonna be on collecting technical updates

on Prague community descent on Prague POW and next steps for Prague pal so the purpose of the Prague powd discussion isn't as much to necessarily declare it's alive or it's dead that's gonna be up to the core developers per the EIP process obviously and we don't really have a good process for going back on something if it's accepted because that hasn't happened very often but we'll be getting to that later that's generally something I guess the core developers are going to have to decide and others after some discussion on process but the main reasons are because there's been Prague updates as far as some updates based on the least authority and Bob rau audits

that are reflected and new versions of Prague pal there's been some I guess the order B exploits that have been found in Prague path since then that are being addressed and need to be understood we also want to talk about the community descent a Prague pal that's gonna have two people from the pro side and two people from the anti side and then Ben DeFrancisco talking about his compromise idea if you go to the agenda you can see who's representing both sides and then finally next steps for Prague pal if we get to a point where we can move on and figure out the next steps from a tech whole or community perspective that's another thing we're going to be doing so

that's the summary of how the prowl discussions gonna go I know everyone's here for the BLS curb discussion though so we can go ahead and start that one thanks everyone for tuning in and I'm gonna pass this over to the hard fork coordinator James Hancock to run the BLS curb discussion for now and he can tap me in if needed take it away James so for some context on this the the VIP centric model we focused around one and EIP is ready then to ship it and to just got decide the schedule at that point and coming up is the phase zero and and the deposit contract for phase zero of

eath two and that uses BLS and and so that gave us some kind of orientation to try to get something to work for them as well as address some of the concerns in EIP 1962 of having a very large attack surface that was that came up in the last few calls so there's been a lot of work since on an on a working group around pre compiles for elliptic curve cryptography and you want to kind of go over that Alex Alex or Alex yeah well depending who would want to start who's

another Alex could lead it yeah whoever from your okay well I mean I can start in principle so right now well first of all there were three pull requests into the EIP repository one for BLS 1281 curve which is what assuming 2.0 guys want another one was for what's called a sexy curve and as one of the flavor pls twelve and the third one was for a cure which allows efficient aggregation of the

proofs or on the sexy curve so kind of like one layer of recursion those are very similar but after some recent sorts and some comments from italic into one of the DRS I think I will change suspects they will be even simpler and binary interface will be 32 byte aligned like everything which exists right now even while we don't need this alignment it's just easier for developers to cope the second part that will change there will be an extension because for sick Bailey signatures one needs to have functionality of being able to hash into the curve point which wasn't in the

original proposal and was never part of the AIP 1962 but it's not so difficult to add this form so in principle it will require like one more week of work and then it should be kind of final in terms of this back and there will exists exists at least one a permutation last [Music] did you say that yeah I'm just wondering because the agenda doesn't contain the links but except which number is the one that is required for further ado intro I think it was Sam's like 2537 I put them

in the in the chat so the original draft of this PR was largely based on kind of legacy of IP 1962 following the same budget but since this will be terse specific we can like it's no longer possible to like it will be difficult to make subgroup checks optimal subgroup checks optional but instead we can use much more aggressive musicians to offset the cost of the subgroup check so I will update its SPR and other corresponding cars with this information

and the binary trees but in principle there is very small difference between those is they just for now at least three different curves and so difficulty of implementing them is already negligible because they all exist eyes are in separate implementations like in sexy is a part of 1962 and some standalone libraries so in principles would be good to just accept would be good to just accept all three of them together from the eve to side Alex Stokes you have any on how this fits in

to Martin's question sorry I guess had mr. Martin's question was this just like general why do we want this the wine or you can you can say it again well no my question was only exactly which which eat number it was that was directly related to Easter oh and the answer was all right goodbye 37 so I guess my follow-up question would be so 2537 does that fully cover all the needs for Trudeau and it's it is it does it add anything unnecessary or is there an

unnecessary complexity which is not actually needed I mean could there be a [Music] ilex please answer I can accept it but oh I just want to say for eath too so if anything will need to add another pre-compiled and as part of this like hashed occur functionality but that's for like some more sophisticated applications like if you were to do a like client as a smart contract a lot of the stuff for pushing for so far is just to have this one pre-compile was a 25-30 step which could enhance the security of the deposit contract and

that was part of the one I should I should say I should say usability like the deposit contract is secure I had a quick question was there a meeting in Paris in the last 48 hours about this I'm just curious like what like if there was a meeting and what the goal of that meeting was as far as I know at least from the developers chat because I wasn't embarrassed personally it was canceled teachers caregivers or like people were leaving Paris earlier so it was cancelled yeah that's right don't happen I'm Alex so you have the three pr's and it'll take a week to put in the

is the hash fungus catch the curb and yeah it will just take a week to implement the current like there is an IETF draft how to do this and I will just take said draft and implemented and it's what people so I would I would make a suggestion here that I mean the BLS 3:81 curve or whatever it's called it's obviously there's a particular need for that and the use case I don't personally see any reason to roll out three heaps

and three new free coupons and three new curves in one go and I know that Alex Plaza wants it or metal apps but I would I would like to hear the opinion of for example V Tarek and Alex Stokes who are perhaps a bit longer history in ethereum and cryptography the curbs that are being proposed other than BLS 1281 one is the BLS 12 3 d7 377 and the other is the curve that goes inside a pls call 377 so the reason

why people are interested in these curves is basically because they wants to do recursive snark so that you can do a yet AAS an arc over one field that and it efficiently proves over the arcs that are in the other field there and this is valuable for a bunch of applications so like for example if he wants to use as your knowledge proofs for both privacy yet scalability and then in most setups you're going to need one layer of recursive starting because the users provide the first level start can dead whoever's doing the kind of witness aggregation that does a yes the ARCA

Hall of the U of the participants darks and that's how you get your scalability back the now the question I guess is is adding these two extra curves and if necessary for achieving this objective so like one way in which it would not be necessary for example as if we invent say yeah efficient range checks which would allow us to efficiently do arithmetic over one field inside of another field or potentially some some other and of totally different construction where for example something that either gives you bulletproof Saudis

on the inside or you or even something else entirely so we have enough current and technology like indefinite adding support for things like you know won't be lost well 377 in particular is going to make and building me of those kinds two layers dark applications easier but and I personally am aware of some research developments other Korat on the horizon that could also allow for recursive darks without but using it using the special curves so if we really do not include these but these special curves that it would be the I don't

think it would be enough logged long-term fatal to aetherium sake utility has a way for I'm making it as a way for building recursive Starck applications but it couldn't the way thanks for some amount of time so it's untreated off between Magus making things easier for developers and horses and the added complexity of supporting news that's three curves instead of one I should add to with Alex words that there already exists constructions which allowed was privacy and scalability in this form just more tinkering with the

same things like snarks over here last well three 377 as any just make a start like proof over the base field of be less 12 377 because he has enough roots of UDT but it's just an audition so in principle you could do this already but an amount of work which would require for people to actually make it work is enormous compared to what people can already start to use with what sexy has made and in addition just difference between two TLS 12 curves is completely negligible as was demonstrated by 1962 and what's both usually called the outer

curve for this construction was also made in more than one different code bases so I don't think it will be necessary to support them because there is no streaking bias occurs specific and as soon as they are final and follows the specs and this apart of say IP 1962 work as there was a like there was already a lot apart it defies abilities it will not be a work to support will be only it will even know be a work to made the initial imputation because they already exists more lessons you should take salmon integrated to the code then supports and after we also have Kobe on the call who joins been working with the cryptography

I wonder if you have some thoughts on this as well is actually the sex tape construction itself which already allows using doing some sort of primates work contracts in Starks which is something that is very interesting for for various applications and even specifically the applications that mission in their paper and I want to sympathize with what what

Alex said that the amount of work that would be needed to use the current constructions for either private smart contracts or scalability is is very small if you use this pair of curves and all the other constructions are currently working progress so either you starts or fractal or things like that these are things that are not available for developers today and it's not very clear when they would be available and how efficient they

would be for practical use cases so I would be happy to have this kind of thing included but I would also support it and said that it wouldn't be fatal to long term ethereal or of recursive starts but it would be really nice for developers to have ability and to just clarify but that last part you talked about was that concerning the two eeap's that have nothing to do with e - yes for the 142 I just supported many pluses right I have two questions

so one being that I would like to hear what Louise Goodman has to say and the other one would be I just say both things right away if the if someone has I mean I recall when we added Blake and six months the engine process x''k actually tested the ABI and noticed that it was kind of cumbersome to deal with and that you have to shuffle things around if you want to do it repeatedly and things like that has yes I'm just wondering if anyone has done it sounds to check that this format is kind of sane and fits the use cases that someone might have on it well in principle for

as a part of 1962 which has much more complex API I have implemented the solidity contract to do the calls which was this one I mean for this set of three bigger files because they were ABI will be updated to be searched by the lines it should be much super can I just react on top of all of the comments so far so Martin you asked me to give my opinion so far so my concern was specific to 1962 as I believe that the scope of the VIP was too broad and very

would be very hard to test and and also back then I felt that the the like the scrutiny the dcpip have been through going through from the rest of the committee meaning people like Kobe or Zach or the people from other recruit ography teams was with liking regarding specific curve I won't I don't have any concern I mean from like the role cryptography in them to not talking from the implementation perspective having specific curve from my perspective shouldn't be too much of an issue when and I just want to comment on one thing about what Kobe said but

fracture Historics being not variable to public today that's genuinely true I mean some people and I like scan knows knows about it because he's working on on starting a new historic equivalent I guess we read shift but I'm not opposing any or source sort of technology to come in because something else exists somewhere so my point would be as of if we know what's curve we're having if we know we have a lot of scrutiny I would be feel a lot more comfortable with with with VIPs so yeah from my perspective as I see it we have obvious is something

similarly well something that feels like a deadline for the BLS 12 381 I as far as I can understand the other two dozens so I would propose that all our efi but we we prioritize it as well 381 and then we'll see if the other ones make it in to the same artwork or perhaps a later one and I would probably aim for a later one for the other ones but now it might be

trivial to to add two more once we do it first one is there any formal definitions a deadline for this because from my perspective there is no difference in time when both for example 12 381 and 12 377 will be finished because they are just not different yeah James do you know about that or anyone else more into the tirado roll-ups schedule so I have just another question on the the 3d one so if we wants to make any two light clients the thing that's that you mentioned that's not in there yet has to do with points

to hashing and I'm gonna guess that's like the swu map or whatever it is at a point to hatch construction well for now I was aiming for as dvw construction which is universal or always respects rights I mean the parts that will not be a part of the precompile it's like the first step in yeah my ETA draft which is cashing into the field element but this can be done by sis marketer so I don't want to touch this ground because maybe people will want to change the hash function later and I don't want the cart code it but the expensive part which is kind of making the film element it is a

curve point which is expensive but non-trivial it will be as a partisan some thoughts on the schedule does Alex do you the hash the curve is that going to be going in with the it's part of the EIP with the BLS pre-compiled yeah I mean I need to update the PR for this part yeah just for Less clear what should you and it's not Alex our for also the 3d a PCO proposed and appearing your misgiving is appearing check or hit appearing do we get the GT element in it no for the same reason of kind of like backwards

compatibility with what's currently available for diem25 for it's just a check you don't get an answer in the form of GT element and I know there was a kind of request for this but then it would will it will inflate the same set of precompiled actually later uses GT element so it's not part of the current okay that means it's okay I mean it's too it's too bad as like it would be nice to have it but I understand you're going back to Barnes thoughts on scheduling that the BLS pre-compile has

a pressing need to get in and having that be the focus for what will go in around Berlin makes sense for the other ones although I know that they take a similar amount of work to get ready it would be nice to have sort of to battle test this the scenario with just the one pre-compile and then if there was a group that that had showed up that had a pressing need that says we're working on this thing and this is why and we have this cut and we and we've looked at the AVI and and then sanity checked it and those things as well on the other pre compiles that would be random or supporting that it should go into the

first four well otherwise I like the idea of having Berlin include just the one the OS and then having the upcoming Forks include the other pre compiles no my question was more like by like

Alex Vlasov: what is the deadline for Berlin

before yourself so it's like by what times the BLS 1281 should be finished by itself the there isn't a it is when it's ready when the when the EIP is ready to go and we hope that that is that that lines up with the schedule for 8th - so maybe Alex the other Alex as part of eath to

schedule how does that fit in well yeah I think all of this is just kind of when it's ready the thing that we would do here is again just to enhance these ability of the deposit contracts it would basically let us make another spare contract that says hey is this deposit valid before someone sends efj to the actual deposit contract but either way like the intention is already to have other places in the stack that like do these checks right so it's it's not that it's like the end of the world if the timelines don't align and also - I don't know if anyone wants to commit to a beacon chain launch timeline right now Alex I have a question for you I

heard that the BLS the the if to curve the BLS we're talking about right now if part of the II IETF like process right now maybe being stored standardized are we planning to you the insider pre-compiled are pre like a known and somewhat battle-tested library or is it something that is going to be built custom for the for for for the fork well in Crystal for now my reference implementations were based on a current work for 1962 in the sense that just it's all already built we just need to change the API and make it pure specific

alternatively developers could use that cash library for BLS 1281 or current or other Easter pointer libraries they just say maybe not all of them already have hash to curve function everything else they already have + - you will need to make an API there will be no difference between them in terms of the results if they are all valid the only thing can be is a difference in performance because I can only guarantee something for the code which we made for 1962 so that this is I mean where is the

performance is good and it aligned between the implementations for everything outside which will be up to the particular note developers to users there is a code basic just cannot use this guarantee but I can give the guarantee that there exists as an implementation which is possible Thanks Vitalik did you have a question in chat or did that get answered but I think Alex just answered it in chat that's reasonable now just the reason why I'm

asking is that if the goal of the 381 is to make it possible to make it easy to client inside of each one then that hash the correct thing is going to be necessary now is that I will fast you expect to use that hash to have to care to find a boy to have to find a hash well the current I if I remember the road from IETF draft it's not based on a sample and reject so it should be just the same price every time I cannot give you an exact number because I didn't page market okay okay because it would be I mean and the consequences of

something that just would blow up with either risk or is it like impossible basically well I mean an example for this one you can actually make it recompile but it's useless as actually is the optional part of the current realist 1281 pre-compiled which is pointing compression which kind of was on the table initially but then by just looking those numbers which you get from pointing from the call to such vector file it's much simpler to just send decompressed points as a part of the data then try to use the decompression especially for g2 points so it's why it's Elsa but most likely going to be

ignored but for caching take care if it's more like know where like there is no option we Isaac use it just to get a functionality and even if it will be expensive which you don't expect there is no other option or just people will not use this functionality if they have some other way to get around us okay and the last point I we have I saw an X I talked a lot about why white box fuzzing you elaborate on this what what kind of guarantee should bring in like what time line can we expected and also we concerned the security a three VIP

you're pushing right now or the concerning on me 1962 as defined previously well more or less what's the point Alexi I wanted to discuss it's like whether you care if medic as well it's in a sense if you take two point three cool points you add them up together and let me try to take this point and I'll go do some results inside I mean such basics check basic checks are usually a part of the work when you implement the pre-compile you when you just implement scary observations and if any of those fails then you will never end up in actually implementing the firing operation which like the most difficult one but for now in the code base which

is in part of the 1962 server checks that what could be called white box they just checked basic arithmetic properties by definition that if you take point and it's an engaged point you get the point of infinity as a result of addition and same for by linearity for paying check see if it can ever be called white box fuzzy planets yes but for me it's more like if you are using sports or if it doesn't and if it doesn't you just don't you will never end up actually getting sobriety self-aware thank you this has been this

is really helpful to get through I wanted to have go through some administrative parts of this just kind of in order the two five three seven two five three nine and two five four one Martin talked about being comfortable moving those into EFI I wanted to hear from the other clients if that is also their position I don't believe Daniels here but it's Tim could you speak on that from Basu yeah I think for Basu I would agree that moving all three to efi makes sense and having you focus on the first one and given that like he said that I was not here I I think for

us committing to just that one and and looking at the implementation in the client before kind of committing to all three first specific date is probably better okay and never mind I think Thomas is gonna be late oh no he's here I'm only call that somehow is the question is it about the AP that they are discussing or the repeated question it's the three the three II IPS for the elliptic curves that we've been discussing we want to move them to EFI but I wanted to have all clients to ever change today because we were working on

if - it means that we already have the the library for BLS connected so how is this thing the idea that whatever happens we should be ready without any problems because we already have all the bindings - the curves they the BLS curves and all of them are supported in for anything so there's no opposition to moving those three VIPs to eligible for inclusion um I don't want you make any statement here or whether it's a right decision or not so mm-hmm I think that it's up to address order to know exactly whether this is ready I think that the only risks are if there are some changes to how exactly they'll

be used in a tea room - and if our api's will be actually offering the perfect response to do that needs so if that's is fine then ya know position from us okay and then from open aetherium is there a representative here to weigh in on these I think art ins here are you are you on mute oh yeah but I am NOT going to comment on this one so no it's a no Martin you're still okay with this

from guessing yes but I suspect that I mean it's good that we say we make the bill as well 381 the priority because I believe that there is still quite a lot of work to do on the actual leaps and the specifications and then there are the implementations so yeah I think with with focus on the phone and I mean being in the EFI means that once all the steps are done we may schedule it so this is not like finding ourselves to getting all three to just want to note that and

yeah I checked with Peter that I'm not operating on behalf of the gap team without your approval no I don't feel free to do it I just I wanted to have the motion of moving those CFI first and I also agree with the focusing on the BLS pre-compiled in particular and having that be what is getting ready and asks that would help for scheduling the other ones would be having a team that's implementing or shows a need for it that would provide a timeline in the same urgency that the BLS one has but I don't I don't that's not something I

know today that's just a feedback to the proposers that would be helpful when it comes to scheduling it so I would I'd say we that gets accepted to those our move to efi without if there isn't any strong objections good can I just have one thing I think those kind of the adding of curve will happen more we the first one being VLS but even the to my benefit from the process that's going to be defined for her for the first one could we treat this one as sort of a test ground for how we accept new crypto into the network which would mean that we probably require some crypto people

come in defending VIP coming in and be willing to reply to integration issues during the course it's like maybe attending like several calls just to make sure it and then no question comes up during during the mission yeah

James Hancock: that sounds great

you know and that was sort that's what I'm hoping to get out of this and yeah I guess two more comments one is like yeah the tube pls I think well I well we were having this call Ben and Alex you know made some comments on Twitter that there's still some stuff that needs to be finalized so like just

being mindful that you know there's gonna be a lot of iterations on on that and then the second comment is just for does that mean I guess what does it mean for AIP 1962 do we just move that out of AFI I don't know what do people think about that or we just leave it in efi do we ever put it into a if I I thought it was but maybe maybe I'm wrong maybe it's still just a draft I always remember it was put into the efi even back in some slack its yeah was kind of the requirements that it should be alternative implementations and a lot of fuzzy testing which kind of was

certified but still was not enough so we can either stay there and if it's um biggest for now most of these new curves at least work as a transition curves who filled me from my site will be based on universal code base of 1962 so more or less I think people will understand it's not that difficult actually satisfy also all the requirements from 1962 and eventually can be included or maybe Easter will come first and then we'll give asthma and okay no problem to just have it have everything one Martin did you have a comment well I'm just

confused I was kind of under the impression that if we roll out these kinds of recompiles one at a time doesn't that kind of defecate 1960 well it kind of if we stay at today's state been more or less a set of course which people want like which people can find an application and even though three PRS four separate curves actually not even close to the set of curves which people want for Marchman for you and via their application because for example it does not include the public cursive snark cycle if it ever be usable in the future but why not

so I mean this would require to include more markers if some application will be found herself or for example if someone will want to have some like a freaking power for Peterson caching essential access which was kind of all covered in 11962 so it kind of the minimal viable set of the curves which people will want and ideally I would even include two more as a separate curves first step one kind of I I read there's less objection to overtime adding more curves as people

want them especially as we're we fine-tune the process of how how to do that from an administrative and a technical perspective then it would be to do it was more or less a comment that it's much easier to support one precompile slash implementation then kind of give developers a choice to include a lot of separate curves which potentially much more much larger number of separate implementations which can actually give them much more problems because the implementations will be much more different problems from an implementation complexity or a security

well more security or maybe a performance I mean for BLS 1281 there exists let's say five implementations at least if not more and then in principle they are all valid they're all for the same curve plus - maybe performance differences some internal stuff more or less aggressive optimizations for some parts and it gives just developers a lot of choice which implementation to include into the node and then and it can lead to you the problems that even while in most of the cases this works they still can be a case when it doesn't

work between so many different implementations should not be a problem I hope but just may happen while for 1962 it was more or less three complete implementations which people could just pick one and they're all consistence between each other and then you never would have to kind of deviate from them or if you want you can just abrogate aspire but the spec was just spec was in the forms that you just do these steps and never deviate from it ends and it's consistent and it's she was made this way so that I'm if I'm understanding this right Martin's the

preference of from Martin is that the possible usability difficulty isn't worth the possible security vulnerability for having such a large surface for things to go wrong and that when when weighing those to the Cortes would favor more pre compiles over time as they're needed verses so I also think that this is something that might be very important here so it's kind of good if there are alternative implementations made by like a different set of people

for some of the recent files and I think this is something that with a design choice where basically we add individual three compiles for all kinds of functionality it's easy to replace one free compile with a different implementation but not yet the other one where's what it's like crypto VM you just you have the VM and it works the way it works and you can't actually like you know I don't know implement like specialize handling for something so in the past we've flown like pretty safe having like many impersonations of different the IPS and it has uncovered bugs in like some implementation so for 1962 we would basically source all of the implementations like you know I think as far as I know all the

implementations one for 1962 were made by the same group of people so I mean more or less I explained it every previous call but it's very to some institutions we're done by me and there was a separate one done by a completely separate person but it more or less my point was that by having a lot of independent P compiles and which potentially would allow a lot of independent implementations it's potentially linear work in maintaining awesome or switching them or testing like cross-like cross testing between each other which was already done as a part

of 1962 for their current conditions so it's more or less for developers to choose like was there one risk outweighs another or not but it just is the more choices which are possible as a larger model discrepancies between different implementations I mean it's great that we can basically always fall back on the 1962 implementation now so if there is gonna be like in all future work to add like more sort of like programming functionality we can always just basically reuse the 1962 implementation to get it like very quickly and then still you know like if there is a need for us to I don't know implement it in a different way or we just want to try out

you know like a different limitation we can easily do it for just this one recompile and this is more what I was saying is that I feel like this is kind of a good a good thing I mean - well I mean it all started with a formal state of 1962 so whether it's like painting delay it's infinitely staying in eligible for inclusion or whatever it just you can place it in a any fork in any form I would say it makes sense to have it be have the BLS precompile supersede this one but have the implementation tied to it or like the I'm not saying that correctly but having

these this new set of EIP superseded 1962 and some like an official capacity so that there is a expectation around it yes it sounds like a right move okay so then we'll move those three GIPS in the EF I will note that we're focusing on the first one BLS as part of being Berlin and that because of the its usefulness for eighth - and as a way to prove or test the process for doing

these kind of pre-compile ads for cryptography then also that 1962 I'm hope I'm saying that number right is superseded by these changes and I think that's good Hudson awesome thanks so much just a moment I need to see is Matt Luongo on the call well again before we go into the prog pal stuff the Danno isn't here I just gave in

James Hancock: keeping everyone updated for what is being considered for Berlin or what could be possible so you have some time Hudson to warm up the two three one

five simple subroutines for the EVM you have any quick updates

Greg on that there's but pardon to three one five the simple subroutines for the EVM is there any quick updates from that like a 60 seconds or less we're just moving along implementing code going over I don't pictures been up a whole lot of motion since the last meeting okay and Anna who's taking up more of my time Dano isn't here to talk about two four five six which is the

time based upgrades and then I made a post about updating the difficulty bomb this morning to five one five but we don't need to go on that today that's just updates for the community hey Hudson there you go all right awesome thanks so much for running that James great discussion everyone so just to reiterate joining for the BOS stuff that is very helpful for it yeah thank you all so much there's a lot of people here to support BLS so and the discussion so thank you all so we're going to talk about prague pow now the reason this came up was there's been a lot of community discussion around it and again this call or this portion of the

discussion is to go over three main things technical updates on prague how community approval /descent of prague pow and next steps for prague pal so this is going to have some representatives from the pro prague house side and the anti prague pal side for the pro prague past side we have christie lee mine Han and Michael Carter aka bits be trippin on the anti Prague pass side we have Martin Kopelman and then Matt are you here Matt Longo okay maybe not able to make it no problem and then representing the compromise ideas Ben DeFrancisco which we'll go over so to start things out I

wanted to just do really quick introductions of the five people I just named or roughly four people I just named so let's start with Christie answer to about who you are if you want to give a representation of an or you want to say that you represent anything in particular that stuff sure if death helps one of the proposers of VIP 105 7 and my work is in both software and hardware and I am currently representing if Devils today great I'll be referring to Michael as B BT so B BT if you want

to jump in yeah the guys thanks for the the time yeah so my name is Mike Carter I've hosted B BT a channel that's been dedicated on mining since 2013 so one of the earlier ones in the space and I'll be representing which effectively is the community that is prog pal on GPU mining ok Martin Kopelman sorry if I'm saying that last name incorrectly I'm I think I just many voices well but have been against propel I can't really say that I

represent this community because it's kind of there is no no organization of that community but but other than that well I'm one of the cofounders of and also the theorem space since almost the beginning and recently was known as we are also starting to take over the development of theory on but ok no that's perfectly relevant in my opinion and then beaten Francisco yeah I'm Ben I run a small software consultancy called

scope list scope lift we're focused on crypto I'm a software engineer myself I actually don't have any strong opinions on Prague POW which is kind of why I put forth the compromise proposal but was posted on earth magicians just view it as a possible way forward for the community is not there were a number of people who had proposed similar things all I did was to kind of formalize it into into a post with some specifics around it okay thank you and even though I just introduced everyone who I invited to represent each side that doesn't mean that everyone else in the hall isn't allowed to speak or

anything like that you can definitely ask questions have comments things like that anybody else who is in the call Oh Matt what's up no problem just give a quick intro like one or two sentences of who you are which side you're representing and if you have any like company representation or things like that about me I got involved with Bitcoin 2013-2014 I started working on a dream stuff in early 2017 on a project called keep I think what's interesting we do a bunch of applied crypto so while we're not doing quarry protocol work I do have a

fair amount of knowledge I have an engineering background but I'm also a CEO so I'm not as in it as I used to be and I'm representing kind of the maybe let's not push prog pal through side I don't want to call it the anti prog house side because I feel like I can be convinced just not today yeah yes awesome thanks Matt all right so just to get started here let's get technical updates on Prague tile the things I want to particularly get covered are since the two audits were released I don't believe we've had a core developer call update from if def else not that they were asked and didn't

come or anything like that but on some of the things reflected in the audit and then additionally there was an exploit that was released a couple days ago I think by someone in the community and there was some confusion over that as well as some conflation between that and the things outlined in the least authority and Bob RAL audits that the etherium cat herders got going so I'm guessing Christie he'll probably cover this but is there anyone else who volunteers along with Christie who would have the technical capacity to comment on this it's all you Christie sorry about that

sure no problem so in the least authority order which I guess we'll go through first there were two suggestions one was actually to increase the number of dag items from - five six - five twelve oh sorry a data set items generated and that we will have a pull request hopefully by the end of today Andre Allen from Jay has been working on that piece in particular there was some work done by PO L on how that would affect light verification as well and we have some paths to potentially optimize

more on the like verifications side the other concern that least Authority had was in relation to our catch hack implementation specifically the padding around our attack and let me just pull up the audit here so for architect finalization the value of in the catch hack specification it clearly states how you're allowed to use cat check we don't add any extra padding in no way that affects the cryptographic properties of cat check so the padding

in catch actually just exists to make sure your different variants don't produce the same results but we don't ever actually care about the result so how we handled cat jack is well within the scope of official cat jack we've reached out to the team that worked on cat check actually and proposed it though we haven't heard back from them just to give a look over it and then the other suggestions on least authority were to create additional documentation which is being worked on and the other two which is to establish a security framework for evaluating ASIC resistance as well as monitoring hardware industry advances is really well is not within

the scope of the proposed EIP though I do agree that these would be very helpful such a questions to have and shall I go through the recent proposed exploit by handle at kick X on Twitter yeah that'd be perfect and also if you can my personal thing that I want to hear about is how that's different than the light evaluation attack that's referenced and or not attack I guess but light valuation what's it there's a word for it but basically the that was in reference and Bob browse on it light evaluation method yeah that's it yep so for the well let me go through kick

X's exploit first which I think would be very helpful for the broader community so about a day ago and or two days ago an anonymous user called kick X put up a proposed repo called get called pro exploit specifically this exploit said that if you have 64 bits in the seed function you would be able to do a reverse guess on what your on what your proposed notes is going to be and then you would be able to mangle the extra data field you would be able to mangle the extra data field to be able to

basically mangle the header and produce extra guesses this is actually the Bitcoin this is the exact same thing in Bitcoin where we have over AC boost it's important to understand that kicks method is not related to the light of evaluation I'll go through that so kicks discovery relies on three basic assumptions number one it requires you to have your own custom node implementation that is defers from the current public mining infrastructure your own pool and quite a lot of these custom-built Asics number two it requires you to have each blocks header hash modifiable along with providing the extra data field something again

provided by mining infrastructure today and you need to be able to generate enough ket Jack to bypass those memory accesses within the block time of aetherium which is currently today between 12 and 14 seconds depending on depending on the network state so this is an interesting thing that least Authority myself and none of us actually thought about given that we did not think you could mangle the extra data field in in etherium like you can in Bitcoin however our proposed fix for this is actually to ensure that the seed

mix right now oh sorry our proposed fix is to have the hash mix consume all 256 bits of the seed produced by that initial cap track round while having the final catch Shack consume only 64 bits of the seed so that ensures we have 256 bits of security throughout all stages of the hash which is ideally a continuous 256 bit of security every step of the way this should be a really minimal fix so it shouldn't have any effect on hashing performance Andreya long tranche a and genome share have also been working on

an alternate solution that instead would not need to increase the to on increase the hash mix to 256 bits and instead would ensure that both the first round and lost catch hack would not be static and they would continuously require other outputs other inputs and so we've put the proposed fix up on Gator we've also put we've got a full request on the FFL github repository for people to look in view we've reached out to solar designer for comment we've also reached out to the other people that have worked on profile throughout its lifecycle to also give sort of a review now for how the

light case at alike evaluation method works this is specifically let me just pull off the least authority ordered it's only feasible if you're able to generate enough data set items on a tiny ASIC connected to a bunch of RAM and according to leased authority as long as we ensure that our data set items are above 256 this would not be possible and on a hardware level it requires a hundred megabytes of SRAM that you can access quite fast and very few cycles and the ASIC specifically has to be

built just to generate these data set items over and over again so their proposed fixes upgrade to 512 - different - different attacks one requires custom mining pool infrastructure custom ASIC design and to basically brute force the search space and one requires to generate a bunch of dag items ok was there a question I heard someone get off mic I just wanted to say I participated more as moral support through their work on the exploit and I was quite impressed just jump right on it quickly found the problem quickly found a fix and I've

reviewed the PR it's quite straightforward quite clean and shouldn't really cause any troubles with performance so it's not quite happy with it and you're refering to the PR from the FFL steam or the PR from on the on kicks exploit so that would be the first one she mentioned not not to like got it okay other people have comments or questions okay any other further technical stuff anybody go ahead Martin yes so one thing that has been talked about a bit but I

don't think it has been discussed in that forum or in a similar forum it's the fact that in April apparently the dog is going to go over four gigabytes which will cause a drop in hash power because a lot of the ìiî as I understand it's assic a minor III asacs will not be able to cope with that and it might be even Chris knows better but might be some around for the percent that will be lost in that time I just wanted that to be raised and maybe maybe it effects

this discussion today I want to add on top of this I was about to ask about same question people I feel also by the community opposition that people imagining that mining any theorem exactly the same and mining on Bitcoin and I just want to point out and Christy will probably be able to confirm any firm but for my understanding is in Bitcoin you've to move it mean it comes to die the mining who just had keep going verses in the film will have at some point to cliff where the asset would just drop out of that backside name right you and I would like I would love to have Christie opinion the topic

and see if that's a good or bad reason not to you to go to break up I can definitely highlight on that so let me tackle your first Louis so the dog was really designed to ensure that there was a continual pressing need to phase out infrastructure with the hope of gently transitioning to proof of stake obviously things take time and that isn't really how it worked out in practice I do think keeping the diag as it is is important I do think that we are at the stage where most of the GPU infrastructure in aetherium is slowly

phasing out regardless just because there's better equipment on the market so most farms have slowly been upgrading BBT can elaborate that on that more in a second because he's closer to the mining infrastructure today the III was specifically designed with four gigabyte of low profile ddr3 so it's not GPU vram it uses a different kind of memory controller we have not been able to confirm whether you could do a firmware fix to actually ensure that this device could stay on the network until October the only the only confirmation we have that it will drop off on April was an email statement

from bit main support that made its way into the media stating it would no longer work in April and asking people if they would like to purchase new Asics there is another device currently on the network the a-10 this is uses GDD are six memory it has a different kind of memory controller and we have confirmed within a silicon that that will be hashing until october but beyond that it will fall off and they have also started collecting orders for a four point five gigabyte version of the device my concern with this is twofold one depending on the network how much of how

many Asics are on the network aetherium could suffer quite a sizable drop in net hash all at once which would not be ideal for our block time and it would pay play a little bit of havoc all night I do hope it would recover fast and I can talk a little bit about our models on the a3 in a moment - I am concerned that this infrastructure that it continues to be rebuilt and put into the f 1.0 ecosystem is slowly pushing out GPUs and it's designed for very short-term games so a 4 to a 4.5 gigabyte upgrade which is only roughly 8 months of mining time so those are my

two concerns on that front and also the fact that no one right now unless you're inside of China is able to get access to in the silicones updated machines or as well as it means updated machines BBT are you able to elaborate a little bit on the mining the infrastructure you see today and whether your thoughts on the DAC can I ask a question before going to the money mr. structure I try to ask you to make itto chat that as was mentioned in an audit it just requires around 100 megabytes of fast static like SRAM to make such DAC bypass attack feasible but sandwich there was a

conclusions at this number of turn like this amount of memory is not available right now but so far it doesn't look that it's not available from a technology perspective because it will just require to have four billion transistors and this amount is negligible compared to everything which is available in modern processors sorry is a least authority light evaluation method yeah I think so yeah so you already went over it the I mean right now when you has mentioned that it should be required to have a around hundred megabytes of fast low latency memory on cheaper to do such bypass mm-hm

so like and this amount doesn't look like doesn't look right and even if you increase it tenfold the amount of transistors which would be required to oh man such a sick as a Elsa doesn't look enormous and it still would be possible for someone who is really interested in doing this to actually make such a hardware or maybe is there any other argument what would potentially stop someone from doing this mm-hmm happy to elaborate on that so SRAM is quite expensive it's not just about the transistor cost it's also which company makes it and which company makes it and also what is the current

supply time and supply chain so SRAM is about 20x the cost per bit as DRAM for these larger memory sizes and it's also possible to get this kind of SRAM in a lot of the higher-end FPGAs today where it says you RAM or what gram we have different types of RAM that we can configure as well the biggest concern right there is it is not as accessible as you would think number one due to those supply chains SRAM is mostly used in other kinds of configurations they have preference number two it's not just about having the memory it's about having the ability to do this at very very low voltage and a very low power efficiency because

currently SRAM is very power hungry it's quite power-hungry so it's a little bit of balance on the hardware side and it's also much more feasible for an ASIC manufacturer to just ignore the SRAM right now and just go straight for the gig of the gddr5 or g6 given that that is a much more cheaper cost-effective solution and it's easier to build and maintain my concern specifically is that given that that kind of amount of SRAM is available in specific types of FPGAs but not accessible to the broader community due to the lack of supply due to the lack of due to the high price

that it would sort of centralized and mining in that space and that that SRAM specifically is not available to all kinds of hardware so it's not available in your gpus which would further widen that gap between your GPU miners and your ASIC miners again the issue isn't that 86 are bad or GPUs are good when it comes to hardware it's just about supply chains Asics have very very narrow supply chains that are hit harder by more effects there hit harder by you know we've had the recent kovat that's affected the supply chain quite heavily many ASIC manufacturers had to cease doing any sort of operations or business

for about a month so there's effects there that do affect it whereas your GPU and your CPU companies are much more cushioned to these kinds of effects well my concern was about another strategy how to implement it and it's not it was actually about doing the custom ASIC chip which has all this SS ROM on it's not from external supplier has actually never ever assured that SRAM is available as a as a separate chip I mean I thought about SRAM being implemented the same way as it's in a modern CPUs where basically all the cache lines are just SRAM pens O's already in quite

large amount while the number of transistors still is like 4 or 6 it's not that different transistors per cell per bit so in principle if you would want to manufacture it as the current process levels it will not be humans that expensive because the difficulty of the processors is if side chips is much below the difficulty of even painting for from 10 years ago are you able to cite your sources on price I mean I'm not no no I mean that I mean from my perspective if it's possible both in principle to pull this and use a

manufacturing and I think people who would use a who did some estimates for custom chips for vdf the the price that wasn't that high they even maybe whilst they have smaller number transistors the process of manufacture was most recent ones but I think they had very reasonable amounts for sounds like a 1 million per thousand and something or even more chips at sounds like 18 for 12 or 18 or 20 nanometers so it also depends on the scale if someone would ever want to make it and produce it there is like it's possible from technology perspective and it may put a

higher price tag but it wouldn't immediately rule it out as economically inefficient it I agree that it is technically possible today this is where I've pushed back on the least authority ordered on that point I do disagree that it wouldn't be economically viable because the cost of your etherium ASIC isn't actually just that one little SRAM with 100 megabyte so that doesn't get you like a net hash ASIC that that's not how it would work you would still need quite a large compute core and you'd still need lots and lots of memory this would just allow you to do a light evaluation method on the dag itself so you would still need to store the dag

somewhere you're still going to want to paralyze it you still have the cost of the silicon for the actual chips and you still have to compete with you know the SRAM which a few of the fabs are able to do that on do that process but you still have to compete with that space not saying it's not feasible at all however I did want to just tie boxes to another six minutes or so but keep going Christy no I just wanted to state that however the working with SRAM isn't something that many you know basic manufacturers have the experience with we only have right now two that are really prominent in the etherium mining space which is

bit Maine and which is I know silicon in a silicon has traditionally favored working with with their GDD are six because they control that memory phi bit main may very very much be able to do through this kind of SRAM ASIC but that again isn't ideally what we want or what we hope for it would give them and their machines a competitive advantage over everyone else and you'd have a choke to supply on those machines the other end I mean the vdf a six you quoted some price numbers there that's not necessarily efficient when it comes to mining and then you had a yet a thing for BBT right

Christy you were saying a few minutes ago yes so when it when it comes to the dag and actually um what let's have an open discussion about the dag is it a good thing or is that it is it a bad thing so two years ago I believe that uh strongly that the dag should be kept in place because we want that gentle transition over to Casper FFG obviously things have changed quite possibly in two years from my standpoint of my vantage point most of the etherium mining structure has been on four gigabyte GPUs but miners have steadily upgraded to six gigabyte GPUs as the surge of other kinds of applications have come into place and

just because of you know they've are a wide they want to upgrade their infrastructure BBT how is it looking on your vantage point and what are your thoughts on decreasing or increasing the dag as it were well a majority of the farms these days and the folks that I talked to I mean the current standard right now is the 16 60 TI it's one of the best cost effective GPUs out there to use and that's at six gigabyte so the natural progression mind you some of this technology that we're talking about is now going on five plus years old with RX for 70s for 80s in addition the specialized mining GPUs you know a handful those are all eight gig

because of that they were referenced off the 1070 with eight gig so you know I wouldn't want to disrupt the current integration that has been done with prog pal to then have a science project you know being that it's tied very closely to at a point of implementation it would be at where the current block is the block predicates the dag sighs right now is the way it's reference any kind of science project going through trying to readjust that would I would say not be as fruitful versus where the current mining scene right now is from an upgrade standpoint na'vi being on the AMD side the seven nanometer stuff is

all eight plus gigabyte eight to sixteen gigabyte so honestly I think it's a moot point I did bring in reference and a few the last video is saying that technically it would be possible but you know it comes down to at the end of the day what's what's the police pain from the implementation standpoint I think leaving it the way it is would be still accepted obviously from the mining scene from the state of the upgrades that we're at now two years ago this cut was this discussion probably would've been a little different but now that we're there I think most people are expecting the four gig to fall off in October okay do we have we're gonna wrap this up in just a couple minutes the technical part

of this but based on what BBT just said is there anyone in the call who has comments or questions on that yeah I was just going to comment so basically with the dad going above four gigabytes and obsoleting a lot of the older GPUs Rockwell doesn't do anything to fix that correct correct it does not affect the Dagg size at all okay other people yeah just sorry I'm a little confused this doesn't mean I thought I heard Martin swen say that the e3 a six were going offline and the majority of the hash power was going to be a result of a six going offline

could somebody comment on whether the majority of the expected hash power going offline is GPUs or a six yeah um so this is in relation to a report that we've a white paper we've been working on where we have a posterior model that's actually able to take a specific note signal attributed with the e3 s based upon coin metrics previous research and actually give that a give that a percentage number of what it would be for the network and the report is currently in peer review so a lot of people have been helping with the

wording and the structure we hope to get that out on the weekend happy to distribute it wide widely as long as people don't look us too badly for our spelling mistakes they look good not based upon our assumptions we but we believe that and based upon the signals of the ìiî we do believe that 40% of the network or roughly 39.2% 41.1% belongs to on the signal the ìiî generates when I say signal I mean that specifically the ìiî fixes one bit of the announced at zero for every single solution that they generate and this we theorize is we've seen this in Asics before where

it's a tracking pattern so while an ASIC manufacturer may know how many devices they've sold they don't actually understand what happened they don't have any data when it leaves their warehouse and goes to their customers so they're able to track it through pool they're able to track the distribution and they're able to make estimates of when those devices will go become end-of-life and when new machines need to phase out we've seen this in some Bitcoin devices before and based upon coin metrics research they found out that the signal perfectly lines up with the release the announcement date of the ìiî the first shipping date of the ìiî and it extends back a year period into April 2017 so just minutes back in English 40% of the

hash power estimated from your report is going offline and all of that is Asics the signal that we are tracking which we believe is tied to the e3 is 40% of the network energy is Asics tied to the ethernet is a core gigabit ASIC correct crap okay awesome okay

Hudson Jameson: let's go ahead and wrap up the technical discussion to go really quickly into this let's first jump into just a couple minutes of Greg you had a comment on the process and I wanted to

make sure you were hurt on that since you've been vocal in the comments of the agenda on that

so if you want to just give your view real quick before we jump into community perspective the Benz compromise the other things

Greg Colvin: I don't think that's relevant until after we've heard from these people

okay

Hudson Jameson: sounds good to me will reorder that then so let's go ahead and get to the part where we talk about community approval and dissent so how this has worked before was in January of 2019 Prague POW was put an accepted state and that was an indication from the core developers that

it would be included in a future hard fork it would move to final state once it is implemented on the network using a hard fork network upgrade whatever you want to call it and since then there's been the Istanbul hard fork and muir glacier that did not include prague POW and then recently it was brought up that because prague POW is an accepted state it would need to be included in a future hard fork so since then there has been dissent from the community some of them feeling like that this was kind of sprung upon them etc from what I've gathered so I kind of wanted to hear from the the anti I guess we're just

gonna call it anti prog path side what your perspective on this dissent is and how you would kind of formulate it as a summary and we can start with martin if you have that opinion then if not we'll go to matt sure sure so i I would start with why I personally was was and am against that then kind of the community perspective and kind of my perspective on the governance process so yeah well I was from the very beginning against it and vocally against I believe and it was never a technical opposition it was just kind of I never technical issues it was a proposal I just thought the proposal themselves the

goals of the proposal are nothing what what I wanted basically so so my view on things and a core promise of aetherium is reducing a platform risks so whatever whatever people want to build on top of ethereal or around aetherium they can do it and they can have extremely high guarantees that what they built will not be made worse by the platform making an arbitrary change so I personally never had any investment with a six or whatsoever I'm not invested in any form but if I would have I could

totally see that I years ago would have made a decision to build an ASIC for a theory I mean I would definitely not have felt that I would do something legit kind of wrong or illegal so if now the platform kind of sites to kick me off I would totally yeah feel it screwed so in my view there can be exceptions or there can be situations where well hiriam has to has to make a change that affects users negatively if it's if it's if it's extremely clear advantage for aetherium or if it's if it's a situation

where the the survival of the network is at stake but in my view as long as those things are not the case then those type of changes that yeah badly effects some some people should not be made and with with proper it's not even clear to me why ASIC resistance is in kind of why that would be a good thing so there are a bunch of arguments that Asics are in a way even closer to prove of stake the thing that we all want in the long term because AC kind of requires us capital upfront investment so there are those arguments

then there is this argument that it's not even clear whether prop pal would guarantee it so personally do one basic resistant segments not here does prop crowd chief Isaac resistant and and well this recent exploits it was published yesterday shows well it's quite likely that it is was possible until recently to to build an ASIC even under profile and it would make the situation much worse because now well if proper would have been implemented the way it was proposed until two days ago the incentives if you can then manage to build an ASIC to not

publish it and kind of to have it quite kind of this monopoly that we are afraid of are much higher because you would never want to share it because you would know if you if you share the or if it becomes public that Nasik exists here IAM would would would fork again and basically this guy who released this this report said something like what was his words of course don't know him but but he was tweeting something like i finally decided to to release it or something like that so clearly indicating that he or she was aware of it for a long time and to me it seems like the only reason why he now released

it was because they were thinking well profile is that anyhow and now i can now i can say it so to me it seems quite likely that they had it in the a mini ver time box this just because we show on time so okay so so and yeah those are some of my personal reasons to be against but it's the current state that doesn't matter that doesn't or that's not the critical thing at the current state the critical thing is that from where I sit the majority of the community is against so so there was this letter and just to name a few projects where are the founders or many of the of the

people behind the project Eragon synthetic status none of your eggs make our Yunus World Order all of them have kind of come out against and I'm currently not aware of a single project building on hiriam that that that says yeah well we are we one profile so from where I am sitting the majority is against it and even if it's even if that would not miss majority and just kind of a large minority even that would be for me enough to say well we can't do prop out there's not consensus to do it okay yeah so to me the conclusion is the another question is is

the government process broken or something like that and to me it's all perfectly fine so to me and the criticism of a theorem was sometimes small committee is unilaterally making decisions and my view kind of says the small committee exists call we are now on and it usually makes a decision people in community very much trust sales decisions and in 94 they have always kind of accepted it and to me it's kind of now a very positive thing that we actually see well there is not a blind trust but there is still this kind of this is the second step that after after this committee kind of makes it suggestion that there's still this

community approval and kind of so far it always happens and I happened and I expect it to happen most of the time in the future but it's kind of it once in a while stuff is rejected then I think it should just be rejected we should move on okay what I'm gonna do here just so that we can get through everyone's speaking is I'm gonna have Matt kind of add on to that to wrap up the not completely wrap up but to wrap up this portion of the anti side to give an overview and then I'll get to Christy and DBT on their perspective on why it should go in and any kind of quick rebuttal they have on that so Matt do you want to kind of add on to that with any of the thoughts that maybe were missed sure thanks let's see yes oh I have a

slightly different take I'm actually not immediately technically against prong cow what I'm against is this kind of governance mishap of what felt like a closed issue to the community I'd actually be enclosed what feels like something that is I know it doesn't feel like it to you guys can work on this for years but moving way too quickly and I do not think that the upside of prog cow which to me is unclear is worth the downside of a community split so that's really the crux of my argument outside of outside of tech and outside of whether or not we

think basic resistance is a viable thing to have I think that this is really showing us the cracks in governance I don't blame any depth for being frustrated that there are random Twitter mobs that appear but these things happen because there's not a great place to have these conversations because many people know you don't feel heard and because frankly I mean a lot of us who do pay attention to these calls and who have been on these calls in the past we're still surprised that product how is still thing and still moving forward

and still a topical discussion so I guess what I see here is a lot of parallels to what I saw on big point in 2016 and 2017 I saw a group of entrenched developers who were doing their best and trying to serve a project I saw a bunch of unhappy companies who wanted something that may or may not have been better for the chain and as well as a split with miners and what I I guess what I don't want to see is you know this will fall back to how crypto governance always falls back which is voice our exit right now we have community members who are trying to voice we're being told that this isn't

the right platform this isn't right everywhere we try to talk there's not a right place we're being integrated and cold I mean you know just like influencers I guess there's an assumption that none of us are technical or not in my case of professional engineer there is so there's just this real split and I'd like to see us step back from that and reconsider whether the social split is worth pushing an uncertain technical improvement or if we want to go the other way which is exit I do know that there are at least two teams who are ready to support clients that do not

include Cracow I do think that it shouldn't go that way and then I guess my last point is you know in in the past on a chain like Bitcoin a lot of these you know we had segue to X and we had a bunch of communities try to come together and and sort of like force a hard-working Bitcoin didn't work and for Bitcoin that was great they're an etherium we have a ton of incredibly stateful projects that I believe will actually be the ones to decide the majority chain I believe that tether maker any security tokens all of these folks have a huge

economic incentive to sit on the right chain and if they decide that is that is a no product how chain it's it's it's going to be painful so personally I think that we should avoid the issue and I think that we should either work on a much smaller tweak if we believe that a stick resistance needs to happen it could even be something that happened semi-annually or spend the next year actually educating the community consider rebranding prog Cal and convince the stakeholders that feel like they've been cut out of this process that this is the right move and maybe we can kind of heal the

risk so that's my thing okay we'll start with a BB T and get your perspective and then Christi and whoever else on the call wants to speak

yeah no that sounds good I want to kind of actually echo some of what you were just saying there at the end Matt I think at the end of the day it comes down to a health of a network and I think as I would say an outsider from the technological side of it initially being very skeptical of it two years ago doing my own testing on this on this particular algorithm trying to deep dive onto some of the other you know ancillary effects that would happen with changing you know from eat ash at

the end of the day it comes down to the community understanding you know from a platform side what's worth the risk and being in this space and being in mining since 2011 seeing the different phases of you know GPUs a six FPGAs it comes down to you know the stability of the network and what's best for the transition when things go - you - oh and everything in my space is telling me from my experience that if you create if there's an environment that gets created that limits the exposure the the risk of an infraction or a split is it's far greater and so I mean I look at it from

a participation standpoint the ability to participate on this network or not nothing more nothing less taking a more you know conditioned approach that involves the community members they've said what they're concerned with from a stability standpoint this being kind of dark for a little bit and then just being reintroduced I mean even for myself not tracking you know I thought it was essentially dead in the water and then we talked that hey it's going to go forward I I understand the perspective and I respect the the concern that the community is talking about I think on the other hand also we have a situation where yes have a high confidence that a majority

of the network are not a majority let us say 40 percent at the end of the day in April we're going to know the impact of the Asics that were on there at that point in time if there was a decision to move forward with any fork when you're looking at one community ironically that's the time to do it because now you don't have the the infiltration of this other you know a 6fp jeez anybody else that's involved in this at that time right you don't have that contention of one hardware versus another of splitting I would say that we take the next month and really breaking down the concerns for maker and everybody else that's on there from a participant a developer on

this network to understand at the end of the day as we're still holding proof of work it is still eath one oh that a rebranding of making something like prog pal plus all the fixes that we're talking about the the kick discovery all in inclusive of an e - 2.0 and you know continue to pound on it and make sure that it's the right thing for everybody well you need to get everybody on the same page bottom line this thing is not going to go forward if you have a majority of your contributing ecosystem on this platform against it but I want to make sure that everybody has a clear and concise understanding of what happens if you've centralized are

allowed the centralization due to the supply chains there's nothing against Asics I have nothing personally against a six I've said this for years on my videos but it is the fact of the matter of a limiting the participation and then you're gonna have a huge economic incentive to not you know move to something like e 200 time bucks this one yeah I mean just bottom line I mean I'll you know produce my findings everything that I have related to this subject I'll put it on a medium we can have a more civil discussion but I I kind of echo Matt's piece with regards to let's make sure we can answer some of the the growing concerns but

I think for the health of ethereum and for the ability for individuals to be able to participate in this space as they've done since 2015 that understand the vision of 2.0 that it's it's a healthier narrative to move towards the ability to participate now all it yeah let's have Christy take it over

Hudson Jameson: and then Ben can discuss real quick what he's written up and kind of get the feeling of the room among the participants in the core devs what we're feeling here so Christy go ahead I want to start by stating that prompt Pao was really developed due to the ecosystem that was around in 2018

there are numerous CIP proposals around asic resistance when the III had first been linked by a leaked via CNBC out an analyst and we were a little frustrated by each proposal because they were short-term band-aids on a much longer-term problem most mining algorithms have not been correctly designed for hardware and thus coin networks often get decimated by these devices and then they lose their their community and a lot of their value theorem was a different beast in 2018 in 2020 we have a significant portion of

gaps that are now dependent on a stable Network and the last thing we want to do is actually have a chain split my biggest concern is ensuring that we're at a time right now where new investment is going into the 1.0 chain based upon my vantage point in the community and I do some work on some research projects right now 2.0 related we know that 1.0 needs to be as secure as it can possibly for the foreseeable future I don't see it suddenly disappearing in six months and so I would like to see a very secure 1.0 chain I would also like to see in

the 1.0 chain be something that all sorts of Isis and all participants Tinka news given that mining is very much a user onboarding experience into cryptocurrency and there have been multiple videos and content produced in the last two weeks with miners telling their stories of how they discovered aetherium mining often acts as a gateway drug for that and so really the intention of Cracow and agreed Matt we should have just called it at hash 2.0 um really the intention is number one to strengthen at hashes an algorithm to help ensure that the original goals

stated in the yellow paper are met and to help ensure that the playing field is level between a fixed function mining ASIC and a GPU ASIC and so that eventually if people do build mining Asics it does not push off your GPU community it does not it does not Gateway its it doesn't block miners coming into the ecosystem and new participants coming in and earning other and ensure that you know it remains an onboarding experience for the ecosystem I will state that right now prog POW is in its proposal period it needs to go

through it needs to go through implementation and testing off their implementation it needs more eyes on it and I think the last two years a lot of the a lot of the people that have worked on front POW have simply gotten burned out because as solar designers summarized at best India's e cash grant when he was working on reviewing Pro pal he said that he's losing motivation because if you don't know whether your work will be seen or adopted you slowly lose a lot of motivation to keep spending time on it I do think if a path has put forward where we say hey we want to strengthen the 1.0

chain we can get a lot more community feedback we can get a lot more people working to solving this problem and we can get a lot more eyes on the proposals as well and nothing is ever perfect and finally I'm I do not believe that that prowl probe POW should be treated like any other technical upgrade but it needs to be done in a way where all participants in the ecosystem are comfortable and understand what the repercussions may be and how it works and again echoing that statement we have not got done a very good job on educating people on the mining community

as a whole and how and mining really does strengthen an ecosystem for a coin okay and then finally Ben let's get your just quick summary on what your proposal was and by the way we started real late at least ten minutes late I think so this probably will go a little over the top of the hour okay great I'll try to be quick on the proposal and motivation behind the proposal as I mentioned a while back in my intro I really don't have a strong opinion about plug file I can see both sides of the issue and I think there's an argument to be made on

each side I think a lot of the kind of considerations are unsettled or open research questions in terms of what the impact of Asics are on the long term help the network on the transition to East 2.0 and I really think it's kind of a toss-up and I think that's why this split has occurred if it was a clear if there was a clear answer one way or the other I think the community would come to consensus around that and since there's not I think we've we've seen this kind of divide given that this divide exists though and that people on each side are very convinced I don't think the path forward anymore is to simply expect that one side is going to convince the other at least not in the short term or at

least without new information becoming available in terms of you know effect on the network or impact or new Asics being there are things along those lines so given that I think that the way forward is it sounds like everybody I think it's clear everybody wants to avoid a chain split and I think it's also clear that everybody wants to have you know have the best outcome for aetherium the most the best security for the network and the clearest path to a transition to theory in 2.0 so given those kind of those common grounds even though we disagree on some of the details on on what the best path forward is there and

we're unlikely to convince each other and compromise solution where no one is particularly happy but we we focus in on those shared goals seems like the best best possible path forward given that I'm the proposal that I put forward is basically one where plug POW should be fully implemented and tested across all the major clients but that it should not be included as part of the Berlin or any other future of great hard work instead the client should be implemented such that there's a command-line switch which enables Prague pal to be enabled by the operator of the node at a specific block height specified by that operator in addition a Prague POW enabled hard fork

of the ROPS and test network should be created and maintained using that switch so we would have a a full-blown test net if your iam test net using Prague pal that all clients would have to support and function properly with henceforth and then the final point here is that a strong selling point should be encouraged in the etherion community that if there was ever clear evidence of an ASIC led attack on the network that aetherium stakeholders would coordinate to activate an emergency hard fork at an agreed upon block ID using the switch that was implemented so the reasoning here is that again because we share those goals to secure and secure the network and keep it decentralized and transition successfully to a 3m 2.0 we focus on achieving that and using Prague

file and recognizing all the work that's gone into Prague file and its goals but instead of using it as the active mining algorithm on the network which right now is is possibly risky and doesn't have clear upside to the point that several people have made instead we use it more as a kind of sort of Damocles or a sort of insurance policy so essentially if if a sick miners do attempt to four at the curium 2.0 or if there is any other kind of clear attack on the network we have a very very ready option to hard fork the network quickly and with the coordination of the community and

coalescing around the idea that that would be necessary now a couple points of folks have brought up weaknesses in this proposal one is what is that clear signal how do we recognize when a so-called attack is happening on the network and when a split needs to occur I think that it would be valuable to have more conversation around this that's that I don't think there is any way to fully lay out exact criteria ahead of time for what that would entail if there was we could just encode that you know we could write the code to do that ourselves it is almost by definition a value judgment and so in some sense this doesn't completely end the debate around

plug file it attempts to push off ever needing to use it on the network by by re by adjusting the incentives of you know would be ASIC attackers but it does leave open the door to the possibility that continued debate and discussion will occur around if a hard fork should occur due to what is perceived by some as an attack but perhaps as others as not I don't think that there's any way around this at the end of the day we're either going to worry they're going to have a chain split or we're going to coalesce as a community around around a clear need to do one thing or the other

and so what this does is you know hopefully avoids that future ever taking place but at least gives us a path forward right now that recognises the work of prog pal attempts to leverage it in a way that better secures the network and the future of the network and and then allows us to continue the process the messy process of decentralized governance down the road so that's the summary and I'm happy to take any questions

Martin Holst Swende: yeah I have a comment on that sure so one thing that I think it's an underlying assumption with this compromise model is that in the event of an activation of proko on a short notice

there will be GPU hash power standing by ready to take over the mining and I think that may not be true I think they will be slower to actually re reestablish themselves for a theory mining if some time goes by

Ben DiFrancesco: I think that's a fair point and an open question I don't you know I mean I guess the the worst case is that the difficulty drops very low and so the reward you know that's a clear incentive for GPU miners to to come online and come back onto the

chain obviously there would have to be some degree of coordination around this anyway so I imagine that large GPU mining pools would be aware of it ahead of time but we wouldn't know until it happens again the intention of the proposal is that this option would never have to be used but that it would be used in a case of a clear emergency where the the downside risk on to the network due to an attack of some kind was so clear that these kinds of risks in terms of no available GPU hash power were worth taking as opposed to dealing with the status quo

margin which is concerned

Martin Köppelmann: coming from so my current understanding is that currently the

assumption is that right now today 60% our GPUs and 40% are a six and it's quite likely that actually the Asics will oh well won't work in two months anymore so we will then have a significantly higher share then of the GPU so where does the concern come from that there are no GPUs available

well so

Martin Holst Swende: I guess the concern is not in two months time but more like maybe a year from now if they're I mean obviously the s3 will have to be replaced and there will be new developments happening

and right I would assume that the lawyer

portion becomes a six and she I mean that's a bring us

Martin Köppelmann: that brings me to another point that I was always saying hey we should focus on crew stake and we should focus on starting to communicate a transition period to proof of stake where mining rewards kind of are decreased over over time because that alone would prevent people from now doing large investments in in six or eight gigabyte basics so as long as now there is somewhat here passed it let's say wasn't the next one or one and a half years or at least a transition or the decrease of mining rewards starts

towards boom stake and then I think that's a much more effective way to mitigate its basic dominance

Greg Colvin: do we have a solid engineering plan for that transition it could be two years it could be four years we don't know

Martin Köppelmann: well the way we should spend time on that one and

I'm

Artem Vorotnikov: actually the same could be said about implementing proc Pao is an option for the main net because the proposals weakest point is that it already seems that all the developers of all the clients are willing to actually

implement these readiness for the switch because there are a lot of other AIP is they need attention right now and I'm not sure if profile is the most important to the network

Ben DiFrancesco: I just want to weigh in briefly there I mean my understanding so as a software engineer you know I'm not deep deeply embedded in the development of any of the clients eath 1.0 or or 2.0 but just as a as a software engineer I don't think it's I don't think you can just assumed that the development time is fungible like that just not how it works so I think that's a little bit of a of a non sequitur or not as critical as a point

has so many things

Hudson Jameson: and I

believe that Prague pout is coated and go it's pretty it's pretty far along it would I think we got estimates before that it would take about a month for the current Prague POW to be coated is that can anyone remember if that's right it well for every client

every clients liked at like get it in their tests join Gangnam or whatever other tests Network context the gas PR is forty lines so it's actually as far as clients go it's an it's a trivial change to make sixteen copal bucks is that it's food a little

change to

Hudson Jameson: the ones we have currently yeah

Kristy-Leigh Minehan: as bugs are discovered it takes some it takes roughly as you soar with our response to this one that takes roughly six hours and it's a lot of careful testing yes they're trivial to fix again Prague POW is actually not that many lines of code it's just tweaks to the yet hash algorithm and modern Sunday is probably the I mean that's probably the best person to talk about work required to put it into a client since he did the he implemented and guess Martin are you

able to come in on how long that roughly could you

Martin Holst Swende: I think that did it in a couple of days a week but I was also basing it off your the Golan implantation let you guys had

Kristy-Leigh Minehan: which was not ideal it was pretty bad that was my work sorry

yeah bubbly P and Andre still but that it rusts and I think they were more from from ground up especially obviously I just wanted

Martin Köppelmann: to remind people that like Hartford is not just and it's clearly not just a technical process so hard for basically means creating a narrative around it kind of convincing roughly

20,000 people at the same time to convince them that this is the best thing to do right now and in that sense every hard for costs kind of community and attend attention and yeah I said many times I think this attention should be spent on on many other year piece like gas change in the gas pricing switch to prove steak and that's where as a scarce resources community attention

but

Kristy-Leigh Minehan: correct me if I'm wrong but for instance in Istanbul or Berlin on those you know bundle di PS that and become technical upgrades which is

technically a hard fork

yeah

Hudson Jameson: yeah you're defining it correctly in my opinion

Louis Guthmann: can I just add a question for for part of the discussion coming from the the drop we're gonna see in a month I have something and and and think that I mean I'm not sure on the topic and just kind of worried in the future if Prague Pat and get accepted or if we drop AIESEC resistance in the following case we are going to see according to two Christie's numbers forty percent of the hash power and because of the way hashing works in

aetherium it means that somewhere in the future we're going probably going to kick out the entire gpus economy from our ecosystem and we are going to be dependent on ASIC maker to keep patterning making ESIC and this my because and because compared econ actual position to Bitcoin as you get D pre-k did after a certain amount of time we aren't we putting the network at risk that at some point the private I of if is not interesting enough or those makers for those constructors to make those machine and who go have the network goo I mean stop

sorry

Martin Köppelmann: so your first concern is soon 40% will drop and then your next concern is or we

will only a basic wait the 40% drop is

any six dropping out so it would mean more one can basic now

Hudson Jameson: the other ones GPUs are dropping out

- sure but my concern

Louis Guthmann: my concern can express on in April we're going to see a three by Beakman a three dropping which means according to Christie's daily numbers 40% of network if we go this way and I don't know if Christie can give us GPU values I would like today what the video is remaining if we leave it that way it means that probably we're going to kick out all the remaining GPUs by

the wait

Martin Köppelmann: wait wait wait so 40% drop out of some Asics will first of all mean higher

profitability for GPU so it will mean kind of GPUs can longer run or more of them run so that will first of all be a good thing for GPU miners

Louis Guthmann: great but my point if not there is if we are living mean it means that other Isaac is going to be also more profitable so they're going to produce more and advise at some point being GPUs because they're less efficient will be not as profitable and running in at burning and Isaac could you agree with me on this one

I mean

Martin Köppelmann: I really don't see why we wouldn't have

energy use ignore their

Felix Lange: so the main reason why ASIC

resistance might be important to aetherium is because we do not want to become 100% dependent on Asics to do all the mining and as far as I understand profile is mostly that like by leveling the playing field between GPUs and Asics in a certain way it means that basically like we will basically continue to have GPU based mining on a theory if in the long run GPU my name becomes way less efficient and mining with an ASIC because there is keep getting better and basically we'll be a bit screwed because people won't employ the GPUs on

etherium anymore somewhere I'm just

crazy switch the previous take I mean that's like I just like there's someone working on the switch at typical state yeah nevermind but Martin can

Louis Guthmann: just comment on one point you say about moving to prefer state as of I am as I'm working on if 21x in my day to day if 1x should be a contingency plan for if anything goes wrong man if 2.0 and we should think about if one day today independently of if 2.0 in my opinion at least because we are going to do any change to make a transition happen but in the meantime if one 1x should still exist until proven wrong

until proven that we don't need it anymore my concern I'm expressing is if we are killing the GPU economy by not moving to a different algorithm I don't care about rock pop then aren't we going to be dependent on the supply chain of small number of ASIC makers which can end at the position of Bitcoin because ASIC go obsolete technically shouldn't we either Stubbs with growth of the dag or move to another algorithm that's basically my question

Louis is actually correct and to talk specifically to the four gigabytes that are about to knock off the current III S

on the network your best time right now or our best time is to get these fixes in proposed attached 2.0 and change before more capital investment is invested into creating those a six because the incentive is there when nothing is done and they will be more a six will be created and we'll have a situation where we're gonna be retaking about this with some of the the I wouldn't even call it anti prog power just for hanging off people's capital hardware right because I mean people are making investments where you have a natural attrition right now about to occur where you're gonna have a majority of GPUs on the network not having a six on the network but the longer we wait we're

gonna have a six on the network and then we're going to be right back in that circular argument of well we don't want to reduce these people's capital investment so that's kind of the I I hear the argument of saying well GPUs are gonna become more profitable it's not about that the incentives are built in and the equilibrium gets created around power cost op X around either a six or gpus the bottom line is is we have we have a known event that is going to occur here in the next month where you're gonna have a majority of one particular capital investment that's gonna fall off the network so I think the discussion should be you know if we're gonna move forward with having some algorithm adjustment we shouldn't

start moving into that to not 2d incentivize that activity of hey we're moving forward with an algorithm change and it's probably not going to be economically incentive to do this particular algorithm with the ASIC be it attached 1.0 but attached 2.0 here's the spec in standard we're moving to if you're gonna make any capital investment you should start planning for that and now you have a level playing field and you have ASIC makers that can start making particular pieces of hardware for attach to point out with it what's that comment I mean the same thing three times I just like to jump in here state some opinions some of which just seems so obvious it's like kind of gut wrenching that I even have to first of all pro-park prog how people it's a

bunch of profit-seeking miners lobbying the court a political committee to get what they want all right full stop that's what this is second thing as a proof of stake maximalist I'm very excited by the prospect of miners becoming sneakers but I have absolutely no sympathy for miners of the GP or ASIC variety in the long term the intention has always been to show them the door I don't really see why they feel entitled to a say in governance the the core devs that have so far been trigger-happy with prog pal have not accurately taken into account the risk of bugs i no longer believe they are to be trusted especially given the exploit that was revealed recently only when you know the

person thought the proc POW wasn't gonna move forward anybody who thinks that code change is trivial is not taking this seriously another point is last year when I was responding to Prague pad the claim from the core devs was it's just me right me personally I mean like me and a handful of Twitter people this is not what they would categorize as contentious but now there are far more people outspoken against it including the the CEOs of both synthetics and compound which are like the second and third largest defy adapts respectively you know not not to mention complements a comment that not a single DAP team has come out pro prog now and it's like what have you guys been doing for a year if you haven't

been able to convince a single DAP team to come out and be like yes we would you know we want this right and then further it seems like really silly to risk a chain split in the short term by by implementing prog pal in order to avoid the risk of a chain split later from Asics when we transition from sake it's like you know sort of madness

Hudson Jameson: I think for your third point there's no one's suggesting that we do it anytime soon there's no one's suggesting that in fact people are more suggesting more compromise toward figuring things out

Ameen Soleimani: right one more point is that I actually support them's compromise despite having

him locked on Twitter I don't know probably said something pissed me off while ago but I support the compromise to keep Prague POW in our back pocket and use it eventually because for me you know the assuming you respective of really the 40% or whatever III is going offline like that only actually makes this argument stronger is is that there's no good reason to do POC POW right now all right like that there's really no good reason why you should do it right now because the main argument for putting it in place is really that it'll it'll with the ROI calculations of asic miners right if it's not worth it for me to deploy an ASIC right now because I believe that prog POW is going to go in and say

months or nine months or one year then it's as good as doing it today so anybody who is Pro Prochnow should have the opinion of let's do this you know in six months or nine months or 12 months because like that's actually the best for network security but because they are profit interested miners they're saying no let's get this done as soon as possible right there they wanted to get this done last year before they even realize that there is a bug in it it's like all of these you know all of these positions not necessarily the people but the positions are sort of you know insane if you actually try to align them with the stated goals but make a ton of sense if you just assume these are you know profit interest interesting people

Louis Guthmann: in that case would you agree on freezing

the

deck sighs

Ameen Soleimani: I don't know I don't give it too much thought

Artem Vorotnikov: if I may I have a question is anyone from the POW camp aside from you know from those who read Hughes go with your implementation are they willing to fund development of other clients for example open etherium because i mean a lot of talk about what other clients

do I mean yeah

Hudson Jameson: it's already it's already in Picardy any

slash up in aetherium just to be clear

Artem Vorotnikov: yes but if any bucks who's going to fix that

the open aetherium team and lessees

Martin Köppelmann: will do that we can do it yeah there's it's not a technical problem into funding

program so

Tim Beiko: one problem I think comes up a lot is like yes you know working on pronto is not mutually exclusive which working on two but one thing it is mutually exclusive on is like you know time on these calls significant like mental energy from a lot of people in the community and and I guess if you know we spend another year trying to debate this and coming

put each hash geo or prog pal 1.1 or whatever and I think that's one thing we want to try and do better and personally I think this is why I almost be more in favor of something like Dan's proposal because at least it gets us to commit to something fairly quickly and and kind of you know how can we reduce like just the argument time so that we don't spend half the Chordettes call for the next year discussing

Prague town

Felix Lange: is in favor of the flag right so can't we all just agree to like you know implement the flag

and then

Matt Luongo: the other part where we actually come up with some sort of

agreement or when we think that we should trigger the flag otherwise we're just going to make it really really really really really easy for anyone organise a

chance yeah yeah that sounds go ahead Martin yeah

Martin Köppelmann: for the record I I also think I'm kind of having it available kind of in case of an icicle texture that there's nothing that hurts was that but implementing is as a flag definitely makes a likelihood of a bit higher because there are definitely entities including exchanges colonics and that have an incentive in in spirits and kind of a new

corny

Felix Lange: so in that case it's basically going to remain in the exact same state it is in right now basically we have implementations which

are kind of ready to deploy there might be you know some some more work that could go into that but we basically have you know like in most lines there is an implementation that could like you know include it at anytime behind the flag or not to you know like make proper work yeah we are already at this stage where we basically have it just it's just not you know like we just don't officially have it

so I just wanted to ask that if we

Péter Szilágyi: so just merging some version of profile into every client and then waiting for some event to trigger it I think that's a bit dangerous because you

I mean what's the guarantee that program still works in five years time I just had a number when if somebody decides to trigger it so if we want to go down that path then I would probably possibly maybe suggest running a test met with profile just so that the code is actually live somewhere otherwise it's it will just a placate somebody go just accidentally refactor something break it and nobody will notice until but do we actually want to use it

Ben DiFrancesco: that's part of my proposal and that's specifically why it's part of my proposal so I think we should have a fork of the robson test net that implements plug pile and that all fine that tests not moving forward I I have

so there's a lot of talk about frog pal not frog pal do Prague fell but something that I haven't we haven't talked about and I'd like to hear from the anti prog team is we there the eath hash vulnerability which is the light hash evaluation method wasn't known and it is now known and to that camp is that something that's worth addressing like is having a patched that vulnerability as part of 'if hash something that they're interested in without actually separate from the issue of

prog pal and then we need to hear from the pro side to just as a rebuttal for that but yeah go ahead anti so so the

Trent Van Epps: the key first step here is that this this was released

if I'm correct with the Babri audit last year and it's kind of just sat there some people knew about it and we're aware of it but it hasn't been communicated in any depth to the general public why they should care about this why this is a significant concern so if we are going to move forward with an e - patch that would be the first step I don't think we're anywhere near that and it's gonna see seem like kind of a bait and switch at last minute well actually there's a there's an eighth hash Wallner ability that has to be

fetched

James Hancock: well I've been I've been trying to have this conversation but it gets very smashed down by prog pal or and then a whole bunch of things so that's why I like to specifically address this

Hudson Jameson: yeah what's

the what's the like chain of that becoming an issue I suppose if anyone can comment on that from a technical level or has it already been implemented

James Hancock: and the willingness of the anti prog power team to if that is that a valuable proof-of-work change because I can hear too from them that it isn't that prog Pao isn't a valuable change but is patching eat hash is vulnerability

a value

Matt Luongo: I just want to be clear I don't necessarily think prog though isn't a valuable change it's just I'm weighing how valuable it is relative to a community split how valuable it is relative to the education campaign it needs to happen and the additional diligence that needs to happen before we do it so you know I do think that it's a

lot more palatable to to have a tweak de cash I'm not proud as from with my engineering hat on that's something like that would be easier than maybe something that we think is better and more forward-looking but so anyway I don't know if I would support it obviously details matter a lot but I'm not against smaller heat past weeks to fix

balls what are the only James

Trent Van Epps: James the only point I would make is this light evaluation method correct me if I'm wrong it basically removes the memory hardness or allows it allows Asics to bypass the memory hardness of

eat ash which means there can be a sixth on the network which a lot of the anti prog pal side isn't necessarily opposed to already set correct

James Hancock: this is a very different brand of Asics and then and like a 10x more powerful ASIC is very different than a 1.2 or 1.5 or a 2 X's which is what we've seen so is that a concern like or I guess I mean I just really like to hear from Martin a proof-of-work change that patches the vulnerability to eat ash is a palatable thing for him and in the community he

represents

hunger presented he

Martin Köppelmann: community and I can't make an opinion about something I have I devoted and I have I don't have an opinion

budget

Hudson Jameson: so we'll have to ask the broader community through like eath magicians and actually a lot of what we're talking about today will probably culminate on like benzene Egyptians post and other posts that'll pop up so I highly encourage people to jump on there and avoid trolls because the youth magicians forum is generally avoided by trolls because it's dense so yeah so at this point we've gone well over time so to wrap it up we don't have a process in place for putting a AIP on hold for the

accepted state that it's in right now or reverting it I suppose I don't think we have a process for that that's super clear so because of that we need to discuss that process today and what the core developers want to do be as despite anything the community wants to do the EIP process currently dictates that the core developers decide what state court a IP is in so obviously they need to take perspective from the community the community does have a say in this from a collaborative perspective but they don't change the status of the EIP so is there anyone with an appetite to try to pragmatically change the state of the EIP or pragmatically not change the

state of the EIP

Greg Colvin: all I would say about if no pro or con just that it's in the accepted state if we want to move it to another state we need a consensus of the Chordettes to do that

Ameen Soleimani: I'll do it

I mean you're not a core dev no you're not yes oh my that's a question yeah sure just because so the EAP is in an accepted state is there anything in the EIP process as it stands right now that says a time frame in which and accepted the IP has to move into a hard fork there's no so one option here

Lane Rettig: one option here is what we call a pocket veto at least in

the US government which is kind of just say it's an accepted state but we never we choose never to move it out of that state

right that's good until we form the EIP process that would be something I'd like

Hudson Jameson: from a process perspective that that sounds palatable to me

honestly we could we could do something like the that we did this morning with earlier this morning with BLS pre-compile where they're working on a new proposal and have that be a new VIP and we can have prog pals be I forget what the word is but there's a word for replacing one AIP with it superseded superseded so we could supersede prog pal with the updated version that Christy's using in

conjunction with Ben's proposal yeah that would be Eve hash to or something like that it would not be a problem and then the deployment as far as Ben's as that colas is and the unknown questions are answered have that be part of that yeah and then as far as the point on wasting time in the core dev meeting on the subject we just need better offline coordination from both a community governance perspective that I know a lot of people have been talking about lately maybe having a community call that's similar to the all core dev called that lends legitimacy and make sure that you know the community is heard not to blame the core devs for this they have a ton

on their plate they have a lot of pressure and they can't be on Twitter all day like a lot of us can so it's not a situation where they're you know like purposefully ignoring any one side in my opinion I think that what's happened here is miscommunication not enough community involvement from both the community perspective and from us including the community so that can be fixed in the future and I think we're gonna find over the next few months and over the next year that that can be amended and once that is amended we can better answer some of these unknowns with forums and other areas where we can discuss this better any comments on that

or ideas not necessarily to solve the whole thing but like is this something that we can be an agreement on from both sides I have a quick comment about the Ibiza first of all Prague pal the actually IP is still in draft ii it's never been accepted it was tempted to be accepted and based on the audit we never moved it past that accepted after the audit reaction but a further meeting oh yeah no that's that's because the champions kept not being champions on it they had to keep dropping out and so no one changed the status but bits be tripping is going to change the status on that to accept it

it's it's a procedural issue more than not being accepted yep how do I become a core dev um we'll talk about that offline okay so I think we'll wrap up the call I think we've made a lot of progress today and yeah is there any final comments or things like of that nature before we jump off and we can say the next meeting is in two weeks on the 20th so what right now it's an accepted state and nothing has been changed on that there's been a suggestion to make this what would be considered in the

u.s. a pocket veto and then to supersede it once another improved version of algorithm comes out such as ET hash - or other things that are amended and explored and memory hard etc so there's been some suggestions but no further movement on the EIP today so it remains an excepted status correct go ahead Martin yeah I was gonna say that I think for these old court up calls we've all always been kind of courteous to each other and I think we have a certain implicit code of conduct which I think I mean is violated kind of heavily today and I just put it out

there because I don't think that's okay for these calls I completely agree and I'll be talking to a mean after this and probably putting in more things to prevent this in the future yeah I'd love to know exactly which parts of the court code of conduct I violated because I thought I was showing enormous amounts of restraint sure we could talk about that after if we as we don't have any more time today I just wanted to thank you guys for having the discussion here I know that this may be feels like repurposing the call but I know a lot of us appreciate it some um I just have a few questions to ask so things that would really help me understand what it is that the prompt ouside needs to do is guidelines on what

education would be helpful to the broader community as a whole and feel free to like throw it out there and Twitter or an S Magicians I would like us to have some form of discussion about do we even need to the whole point of changing at hash or changing powers all once again on the thought of do we want a six on the network are we a sick resistant or not and it stems back to that principle of the yellow paper and so I think that really has to be the question that we answer collectively as a community before we go down and do any further investment or work so it would be really helpful for me to understand

what kind of things would you like to see from the pro pro team what kind of educational materials would be helpful and how else can we be of assistance for people that have questions or want to understand what the value of strengthening or sorry leveling the playing field on a mining algorithm maybe I can when two one flips this question and ask what else do you want to see from the community to finally decide to not further kind of talk about proper and and and try to push it yeah I also want that is I think you also maybe underestimate the community a bit I

think the last weeks have been helping a lot to educate the community and I think they also have been signing the petition I think they are well aware of all arguments that have been made so I think it comes really down the question do we want to be a singer assistant or not and I think those that aside the petition they are willing to to take this adventure risk of this so I am worried a bit that this is yeah this can go on for very long time and it could be a waste of time VIP 2 5 3 8 or whichever the EIP was the the the letter um that you're referring to did miss a lot of the arguments you

know it's not about throwing one it's not about favoring one group over the other it's about leveling and I'm just wanting to understand since I'm the pushback is that we haven't produced the right there hasn't been enough education what are there a clarifying that part Christi what Trent was saying that there hasn't been education on the motor ability of E - as the light valuation it method and so that should be something that should be addressed so cut it that that being and I I don't know if everyone on the list that sign that petition is aware of the risk that that has or if there is a risk or if they care about it that's the risk is only there if you we're like we're past time

to rehash into it thanks for yeah but thanks for thanks for the stuff so far that's been set on it I think there has been some discussion on it but there definitely needs to be more and I think that just to summarize where we're at for those tuning in we're not where we started on at the beginning of this call I don't believe I think that there is agreement that prog Pao is not going in anytime soon and by that definition in my opinion that means within the next three to six months I do think that I heard no such agreement no I said I think there's a wait I say agreement very good maybe I didn't hear you oh maybe I did say agreement sorry I what I

meant was like except it's on track it's accepted it's not that we decided on a foreign work that's going in Greg that's what I meant we've not going into a hot folk anytime soon no we have not decided which heart for okay no no we have we are we it is this it's not going in the schedule anytime soon I will also say that like it's only didn't schedule is very different and we have a concern ever mind okay so yeah there's their elbowed is an accepted state there are some ideas to make it kind of a pocket veto type thing there are some

other ideas to not pursue it right the second and it sounded like from the pro prog house side which obviously doesn't represent everybody that they were okay with improving and further looking into something that isn't like specifically prog Pao as it is today so actually you know what this is going over time but as far as the core Deb's go is that what where do you think we've come to do you think we've come to anything Martin especially in the talent Peter Felix Thomas so I think basically

Felix Lange: we've come to the conclusion that the community really really really doesn't want it and for me personally this is a

very strong sign regardless of so there is a technical argument to be made for portal there is a social argument to be made against profile it's even if if we cannot reach consensus on all sides about what the change should be it cannot go in it's just a fact I think like I am not willing to put in the change with so much community resistance regardless of

like where that resistance comes from and that is outside of the process technically but I it is it might be you know like process be process but you know with this much pushback I think it's more important to listen to the

pushback and to you know follow the process read it religiously yeah but it sounds so it sounds like Christy

Hudson Jameson: and BBT have said actually this is summarized well in the group chat except the existing EIP 10:57 prog cow gets deprecated for adjustments of the kik fix+ adjustments included zero point nine point four the this transitions to et ash 2.0 is a new e IP that has yet to be written and then can be superseded that's what that has been suggested do we have any core devs that are saying that's a good idea cuz I was kind of getting that impression earlier but I don't want to be speaking out of term or saying you know it is

what it isn't I am NOT the authoritative kinds of things actually like I wouldn't so but

what

Felix Lange: I would say is that it is it sounds to me like a good idea to like rethink it at least after the revelations with like these two attacks they definitely have to be addressed on a technical level before we can move forward with any new algorithm right it's just it's just like that these things have to be taken care of anyway so if that new new ERP is going to be called et hash too and maybe it's almost better

I don't know

Hudson Jameson: yeah I mean like it that's not entirely the point of it just sounding better but I would say I think the core

developers need to speak up more personally because otherwise this is gonna get rehashed and you all are the ones who are deciding if this goes in on a technical level

wait yeah okay so on attack from a technical point of view I think we have implementations of this there has been a test net a lot of the research has been done profile is not called is not it's not actually causing anyone to do any sort of like technical work in the etherium clients right now all the work is basically I think on Kristy's and the bells aside so they are more or less actively working on it but the I think the client developers are not spending any time right now implementing anything related to talk power so this I don't want anyone to get

away with the impression that we are you know like prioritizing working for power over like anything else we are not okay any other comments so

Tomasz Stańczak: maybe I'll give the summary from that so it's quite clear but or at least I'm standing and so one question it comes from the Bronco proposers is not about the proper itself but about the notion of basic resistance in generally I'm in favor of basic resistance as for Prague now definitely we want to deprecate the eat that was not addressing the recently discovered issues with Bronco you should be basically they not properly fixed

version and then wait for the next series of reviews and also I would encourage anybody who wants to push Prague route to just spawn the fork or new test net based on Proctor so people can actually play with it we've the existing client implementations which I believe are go it cerium and already nadir mines does not have at the moment support for Proctor however I believe that the proctor supporting nevermind can be added without much effort on a short notice in case the decision is made the fact that we do not have the proper support does not come from day and for the

reason that we do not support it and just because we believe that technically it's not a big effort but we've seen so far that it was unlikely for the product out to be implemented on it and in a very short timeframe so you are not realizing it thanks

Hudson Jameson: thank you anyone from Gath one a comment before we go or open aetherium actually that would also be a

Martin Köppelmann: yeah I mean from well open aetherium I guess I would come and we would try to develop for the needs of the etherium community from where we said the needs of the Syrian community are not proper but other things so we would develop other things

Hudson Jameson: okay Jeff I feel we all don't have one that's fine

yeah I'm not really sure what yeah I kind of yeah III don't think I can take

Péter Szilágyi: I can pick a size at this point honestly my if the community really doesn't want to do this I must feel except we don't really want to shove it down anyone's throat however I kind of do have stand on the point that if we figure as a school dance that leads whatever that profile may be useful in the future and I still think

we should pursue just getting the implantation done properly so either we should say that profile is dead we don't care what come a sixth go for it in that case we can draw profile and be done with it but then let's accept and let's publicly say that yes we know a six will become and we don't mind and the more the better or if we say that we don't want a six long term and if it's causing issues then we should definitely stop up against them then I do think that we should at least take the code seriously and make sure that it works so that if if there's a need then there's a way to to actually do something meaningful so

that's kind of my personal opinion on this but I don't want to speak in the name of the team because I know that Martin he might have a bit different opinion so I think that the decision to like whether or not ASIC resistance is important for it areum this is a purely social decision and it's something that can be well it's not okay I know some people will disagree but it's definitely there are lots of social factors in that decision so it would definitely help to have you know more opinions and a sequence

distance from everyone and that includes the much much wider community

Hudson Jameson: okay anyone else have final stuff before we go

ahead Ben let's see Ben first yeah go ahead Ben Oh appreciate that thank you um I just wanted to make one quick comment on I think for the last number of minutes here on this call or at least at some point during the call we were kind of falling back into the trap of trying to convince each other and either two sides trying to convince the other side and I really I just feel as if we're past that point and no one is likely there's there some portion on each side of the debate that is very

unlikely to be convinced by any just argument barring some kind of new information or change to the state of the network and so I really think that to move forward as a community we just have to accept that even if it may be frustrating on some of on some sides except that there are two sides that are unlikely to be convinced without some dramatic new information and then consider paths forward that makes sense given that leave that as the baseline and I say that not just with regards to Prague House specifically but with to what's just been being discussed over the last few minutes which is that we have this kind of on what the right word is maybe a philosophical issue about whether or not we want basics on the network and I think part of the reason

as I mentioned earlier part of the reason why there is no clear while there are two sides on this issue is because there there aren't clear answers to this I mean the whole space is only 10 years old we've only seen a few major networks evolve like Bitcoin that have become completely ASIC dominated we don't really know what the long-term game theory is and it's completely changed by a bunch of specific details with regards to aetherium the dag the transition to youth 2.0 so we're not going to solve that problem we're never going to come to a consensus on that unless again there's dramatic new information and I think we just need to take that as a given and move forward knowing that we're not going to convince each other on that point

Hudson Jameson: okay Tim

Tim Beiko: yeah I was just gonna say and it's kind of echoes of

what you just said Ben like I agree that there's like some fundamental kind of nautical differences it's probably kind of pointless to try and resolve them and I think maybe one like concrete thing that can be done it's a kind of signal that like there is a move forward is if there is like a new proposal like I would really kind of favor having that be like a separate EEP just and you know like it might sound like superficial but I think having like a a separate EEP that's kind of started to do some of this that has like a potentially different name I don't have a strong opinion on there but like yeah it seems like like we discussed a lot of like really interesting changes and and that

seem to have some buy-in from from a lot of people during the call and and I don't think adding those on top of the current prop Pao proposal like social we make makes a lot of sense

Hudson Jameson: okay let's wrap it up it's already over gonna apologize for the time going over and also for earlier when I was trying to do a recap of everything I don't think that's my role I don't think I need to say what's been decided or what hasn't been decided necessarily every call I think everyone's heard everything today and can come up with their own conclusions just like the heath community should do with this rough consensus model I should just be here for process primarily and for organization of the calls so yeah great

talk everybody thank you all so much for being here really it's been super important I feel like and I think they're all come fruitfulness from this things will be fruitful from here on out hopefully and we'll see you in two weeks thanks everybody

thank you [Music] you [Music] [Music]

[Music] [Music] [Music] [Music] you

[Music] [Applause] [Music]

Speaker names adapted from ethereum/pm, CC BY-SA 3.0.