PQ Devnet Updates

Transcript

Okay. Uh hi everyone. Um so my name is O or an um so I'm from team from one of the client for for the process so far. Um so today I'll just be giving updates on overall like um progress on implementing the death nets for post quantum signers. Um so basic recaps um some objectives of why we are doing the dev nets. And so to prove signature viability um validate you know multi client interops validate resource requirements and also see where the specification tests are. Um some secondary objectives um basically you know creating usable infrastructure so that we can test more things on top of this um establish some common metrics

especially you know across clients and you know start to develop de uh coordination protocols. Um why that's so early? Um so this is basically like you know the the first kind of testing ground for coordination and also integration of um you know multiple components um you know researchers and implementers coming together um client teams both existing and new ones and also um being able to you know try to start testing and implementing multiple specifications uh components. Um so yeah if you you can say that it's basically you know like kind of like first integration test of consensus working together as a as a complete system.

Um some of the philosophy that we're trying to to to achieve here is you know instead of one complete death net at the end of you know all this progress is we try to do like many incomplete uh death nets. So basically you know to paralyze implementation with specifications uh create a tighter feedback loop between implementers and and researchers um discover spec gaps and also you know hopefully all of this helps us accelerate to to production readiness sooner than later. Um participating today and over the course of the next two three days um we have multiple clients here uh client teams here. So there's reim who are implementing rust implementation uh ze doing uh

implementation in zig there's clean uh implementing in c++ and also doing a lot of p2p research uh there's also lantern doing c implementation collibri uh doing light client research lambda class cryptography and prism obviously u doing the weekend client in go so hopefully this is going to be really productive having um you know thoughts and opinions from both researchers and client teams and both you know exist the client teams and also like the new ones. Um so just a bit of recap of like death net zero. So this is still in progress.

Hopefully we can kind of like call it complete somehow. Uh over the next three days. Um so what we have done so far um it's a lot of bootstrapping work and to be honest like really big disclaimer is that there is no PQ signatures in place in def0 yet. Um most [snorts] of it is bootstrapping you know the clients and and all the infrastructure needed. Uh we have a new specs framework that's in the repo. I'll share the links later on. But yeah, thanks to Fidipe and and Thomas uh especially Tom I don't know how you managed to fit you know creating a new specs framework along with all the work that you're doing with you know lean VM and security and all that. Um I guess he's probably like an AGI somehow behind behind the [laughter] human there. Um we

also on top of that framework we have more or less complete the client specs for for net zero. So that's like minimum like PDP >> [clears throat] >> layer. Um there's also consensus and yeah like I mentioned here um the basic consensus that we are actually building on top is actually based on 3SF mini. So hopefully we are also experimenting not just the the postquantum signature part but also like other kind of components that we are um trying to aim for for the consensus or the as well. So in this case it's through SF mini and so far I think we pretty much enjoyed um working on that. Um there are some stuff that we are uh doing in progress. So mostly infrastructure stuff um you know test infrastructure metrics uh multi client interop hopefully we can we can work on

that uh over the next two three days. Uh we are going to be working on fixed validators for now. So there's like no deposits no withdrawals you know just like pre pre-generated you know uh validator keys and so on. Um starting off maybe with with probably like three first. We are aiming for 10 and hopefully if that that works and all goes well we can try to increase that you know to to maybe 100 or more. Uh and again disclaimer no PQ signatures yet for de zero uh hopefully completing within this week. Um and then there's definit one soon. So this is where the the BT postquantum signature uh work happens. So we are hoping to test PQ signature generation of course signing and verify uh verifying for for blocks

and votes. um no signature aggregation yet in defet one and hopefully we can complete this and um demo this or announce about this completion of devet one with pq signatures um by def connect so that's that's so pretty uh aggressive time right there um so some of the goals for net one um so yeah block and a test uh testation signing uh using the hash uh repo that that created uh hopefully we are aiming for full block pipeline within 4 seconds and that's basically one uh so so those are broken down to um yeah so so four 4 second blocks justified and finalized you know for let's say like 24 hours or

more and within each of the block you know block signing time we are aiming for target um less than one target is like five 500 milliseconds signature verification is also the same less than one second um I going out yeah attestations uh attestation signing. Um that's maybe that's not a different I'm not sure. Yeah. But anyway, like so so signing and verification under under one second. Um and in in terms of overall timeline progress, so our kind of like initial discussion about you know create let's create a devet actually started in June as part of a beam day in ECC in in cans.

Um and then in July uh we had our first interrupt call. Uh we green team and seam team shared their views of what they think that net should look like and what how should we work towards that definite um we had our lean spec first comments I think that's also by Toma or or Philip I'm not so sure first comment in July as well in August we uh started specifications for net zero um and we also started some you know early implementations both for for V and seam and we had uh clean uh joining uh a bit later uh in September then we sort of like completed our DevNet zero spec imple implementations and now we're into October. So October is mostly going to

be about like in infrastructure and trying to complete DevNet zero. um November like I mentioned defet one completion with PQ signatures and then further on uh that we also want to discuss during the next few days is you know what's next for for the dev net after devet one right so it could be signature aggregation you know some sort of like P2P improvements uh rainbow staking that could also be uh that's also a candidate there um so lots of uh lots to be discussed and hopefully like this is not only part of just the client teams are discussing this uh we are also hoping to discuss with you all the all the researchers see what's the kind of like the most mature or like what's the most um like valuable things that we could we could try to to prototype on.

Um so so those are our overall like all um on the DevNet progress. Um now specifically for the interop that we are going to be discussing the next two three days um objectives you know reflect and improve. So basically reviewing defin net zero in our process see you know what what's what's been slowing us down what can we improve what can we accelerate more um identifying pain points and yeah basically unblock remaining works that we need to do for dev net zero and then the second group of of objectives are basically planning forward for for the future dev nets right so kind of like come and reflect on the north star is what what do we want to make an impact on form based on these like dev implementations

We [clears throat] want to specify definite one and beyond on what we prototyping and establishing some metrics and more importantly the targets that we should achieve by doing these def. Um a quick summary of outcomes. So hopefully in the next three days uh we can uh successfully demo net zero in one form or another. um we have all our lessons learned from death zero um documented and hopefully like start to discuss how we can improve that for devet one and then the third one is yeah as much specs for net one and beyond um some sessions I think you can find out in the schedules but there's also a day two afternoon and day three

afternoon that's still pretty much like fluid so we can decide on the spot what works best for us to to achieve the three outcomes that I mentioned uh before um so those are are open to discuss question that we can basically discuss or or or um decide on the spot on what we want to work on. Um some of the big questions that we kind of like maybe want to want to discuss about is you know are we lean yet? Are we are we that mean you know the promise of like singular consensus science are we actually getting towards that or are we just you know um basically getting retaining all the complexities. Um is the pace of doing net zero right? Can we do something to improve that? Um can we contribute more to overall the effort um and what should be in dev two and three

and so on. Uh these are some of the questions. So there there could be more um some of the aspects that we uh would be discussing about ranges from you know specifications client implementations. So we can talk about you know cross confident interactions for example between you know PPP and and and postquantum signatures is that viable um metrics and targets drop testing um definite infrastructures cross team coordination or maybe to some point to some level as well public communication how much should we actually communicate our work to the public get people excited you know and get more people to to to be aligned with with with what we do as well both in terms of definite and also in terms of achieving job overall.

um some of the workshops that we'll probably find time to work on first thing first is definitely you know definite one specifications um you know what what do we actually need in terms of in terms of the P2P um standards signature types and so on um there's also work on the testing infrastructure so so far I think we have been focusing on cl implementing our own client and now it's the point where we start to have to integrate it with with multiple clients and actually test together and try to get finalization across clients. So cross client um testing is is important and something that is probably useful to be like in person to discuss about this. And the third one is yeah basically DevNet launch right trying to hash out the the

final details that are remaining for for DevNet zero and try to to get it to work within the next three days. Um so yeah that that's this overview and if anyone is interested to learn more or see our progress uh on the devet implementations there are meeting notes in the lean ethereum repo uh pmpq inter all the lean spec including all the the researchers are there but you can also see in the leanspec repo there's also client folder and you can see there like all the networking all the containers specs that we are using for net zero and beyond is going to be in there as well.

There's also a separate read repo for lead metrics. So that's I think that's maintained by Katya and basically uh tracking you know all the metrics that we want to we want to track across multiple clients. And then there's lean road mapap website um which contains all the you know different tracks and all the papers and all the progress uh that you can easily follow up on. Um even though like maybe you're not it's not that useful to you because you're on on top of everything but lead.art part is probably the easiest website that you can share with others to to let them know and and learn more about being uh be progress um client repositories. So right now in terms of the dev net0ero we have three clients participating. So um the repos are are all there. So those are on

GitHub and yeah so that's basically an overview of the definite effort. So uh let's hopefully it's going to be a really uh yeah uh crucial uh three days in Cambridge. [applause] >> Does devet one require the post of signature aggregation? >> Yeah. So, so we are intentionally leaving that out, but I guess that's also what what we want to discuss as well. I don't know what uh in terms of I think from from client's perspective we're not so sure if >> um the the the direction is charged for us to implement but if that's the case

then we can also try to to figure that out. But for now I think we are we are uh we are assuming that um just single signatures are are already going to be um enough for us to try def one and if you want to do aggregation that probably is best to net two but if it's ready in time then we can shorten the the development time for deed [clears throat] there was The metric you showed that instead of something like it takes 24 hours for something to finalize.

>> No no no no. So that's our dev net to to call you know devet zero successful we want to run and finalize the chain for let's say like for a number of hours. So you know 24 hours seems like a good number to say that the dev work successful. >> Ah okay makes sense. Thank you. Okay, fantastic. Thank you so much.