Do Based Rollups Solve Everything? | Justin Drake

from my perspective like once you appreciate base sequencing and your bass pild there's there's there's no going back uh I've I've I haven't I don't know of a single person who who has gone the other direction it's it's kind of up only up to the right in terms of the the the converts welcome to Bist where we explore the frontier of Internet money and internet finance and today on Bank list we explore the frontier of based rollups today on the show we have

ethereum researcher Justin Drake to give us the reasoning why he thinks the future of ethereum is based so what are based rollups how do they differ from normal Roll-Ups why the ethereum layer 1 is the best sequencer for ethereum's layer 2 Roll-Ups how does it fix ethereum's fragmentation and what does this all mean for ether the asset there's already a handful of teams publicly working on the based Roll-Ups future Tao puffer Spire but also Justin alludes to at least a few of ethereum's biggest rollups already planning to go based in the not

so distant future so with this episode today and perhaps a few more soon we are front running all of that by getting you downloaded on based rollups so when ethereum inevitably does go based you'll have already seen it coming I think uh both David and I you'll find in this episode are very bullish on the concept of based rollups because it really solves a lot of the challenges that ethereum is facing with its its current rollup Centric road map so this is the most comprehensive episode I think we've ever done on on based roll-ups and Justin toward the end it said that he thinks that uh base rollups have really

reached escape velocity I know some people listening to that are like what they just came on the scene But realize a lot of work has been going on behind the scenes to sort of like propagate the idea and design architecture for based rollups I think it's getting ready to take off so this is maybe a few months in advance of of front running but 2025 could be a big year for Bas rollup stav and I want to talk to you about all of that in the debrief that we do right after this episode if you are a bankless citizen you're on bankless premium that is an extra podcast no commercials it's

got the bonus content David and I put in every single week you have that in your feed that's the rule that's the uh episode that we're about to record where we give our thoughts on the episode It's called The debrief so go check that out as well speaking of uh extra content Justin is giving a talk at the bankless summit so if you are going to Devcon and want to hear Justin talk uh the banket Summit is happening the day after Devcon it's like TED talks for ethereum I've hand selected some of the best speakers in ethereum and charge them with the responsibility of producing the most interesting talk possible I asked Justin

if he could leak anything about his talk at the end of the episode he doesn't he doesn't give me too much but he did say in a chat that it's painfully bullish so I'm I'm excited to see uh what all of that means so let's go ahead and get right into this episode all about bass rols with jus and Drake but first a moment to talk about some of these fantastic sponsors bankless Nation we are once again here with Justin Drake which means that we are going to learn something today Justin Drake of course is a Reacher at the ethereum foundation he has taught us many things on the bank list podcast uh and today we are hoping to learn about based rollups Justin

welcome back to bankless hi David and Ryan thanks for having me so before we get into the idea of Base rollups first talking about rollups generally how would you say the rollup Centric road map uh is going the ethereum effort defined and then moved towards the rollup Centric road map in 2019 so 5 years ago uh 5 years later how would you evaluate our progress our Direction in the vision that ethereum set out for itself in 2019 yeah I think we're doing exceptionally well on the certain Dimensions one of them is that there is

a lot of experimentation and a lot of teams that have uh taken it to to build an L2 or rollup or or validium another thing that I'd say we've been successful at is dramatically reducing the fees and so in some sense uh scaling feels like a Sol problem you know we have a whole road map to increased the amount of blob space uh and we've seen tremendous Pro progress with the fraud proofs and with the uh validity proofs and so uh from from the perspective of of scaling and

having you know relatively secure uh rollups I feel like we're we're doing exceptionally well what I think that we're starting to transition into is basically the interrupt phase of the rollup uh Centric road map where we realize we acknowledge that we have a fragmentation problem we have a a a silo silis problem and we' like to bring back some of the network effects that previously existed at layer one and that we have lost the network effects at

layer one were unsustainable uh because we had anti Network effects uh namely transaction fees right it's not sustainable to have a $100 uh transaction fee and that's something that we've solved but now we have another anti Network effects which are I'd say is the centralized sequences which prevent this uh this optimal form of of composability which is which I call synchronous composability when you say Network effects Justin what do you

mean by Network effects and why are they important so Network effects is basically when you have a network and every additional node brings in more synergies than just its raw weight so you can think of it is just 1 + 1 equals 3 you have um positive sum games you have win-win situations and one example of of network effect uh would be you know shared liquidity every rollup kind

of adds to the liquidity pool and now ethereum as a whole can become this this defi Hub uh with the best-in-class execution and whatnot but right now we're in a position where liquidity is fragmented and so that particular Network effect has started to break down a little bit so um David asked the question like how how's the rollup Centric road map going and and you talked about some of its strengths you also mentioned toward the end some of its you know weaknesses maybe uh shortterm or or medium-term or long-term I'm not sure what your perspective is

but could we dwell a bit more on some of the I guess you know trade-offs or some of the the weak points some of the areas that ethereum needs to to figure out so one is fragmentation maybe and a loss of some Network effect are there any others that that pop into mind here associated with the def fragmentation there's a little bit also of mimetic confusion and uh also a loss of of unity maybe at the at the social layer there's you know competitive forces uh there's maybe you

know on aligned incentives and things like that but uh nothing that fundamentally cannot be solved so you know I I remain optimistic uh and you know interruptibility is is is a big word under which there's lots of any different constituent Parts um so one of the low hanging fruit for example is just having basic shared standards you know for at the at the wallet level the equivalent of erc20 but in a cross rollup uh uh uh model the idea of having

you know shared governance this is something that's being experimented on shared sequencing shared stock aggregation shared data compression and and blob packing uh shed deposits uh the idea of having shed fundability all of these things are being worked on in in parallel and directionally we're we're making tremendous progress uh to towards fixing this this fragmentation when you say mimetic confusion at the social layer a few things that come to mind for me and I just want to sort of name them

us you like explicitly to to make sure we get kind of our our problem statement right on the the sort of the social layer side of things like one mimetic confusion area to me seems to be this idea that l2s are Paras I well are they parasitic uh like that's one area of of question which didn't exist prior to the rollup Centric road map there was like only one kind of uh ethereum shared state so it's very obvious that every app that deployed on ethereum was like netg good for ethereum the social layer would sort of hands clap cheer it on so

maybe that's one area another area of mimetic confusion I would say is around ether the asset and this question of okay what are the value acral mechanisms of in this rollup Centric World we've seen fee Revenue go down at least in the in the short term you know post uh Proto Shing uh and the deployment there and so now there's question of well what about this ultrasound money thing you know the there's not as much fee burn at the uh base layer for ether the asset so these

are two areas of mimetic confusion that I've seen this idea that l2s are parasitic this idea that uh there's no value acral uh from a cash flow statement for ether the asset is is that what you were explicitly referring to uh when you said mimetic confusion or if not that what else so I think you summarized the perspective very well of the the E holder who's a little confused but there's other entities that are also confused as a developer you know I used to you know have have it very very easy for me I knew that launching on ethereum

just meant deploying on layer one but now I have to choose across a whole ecosystem and it's a little bit more confusing and similar thing as a as a user now make I need to make an explicit Choice as to which ecosystem I I need to to uh put my assets on and then you know there's like 20 different flavors of epher with every single deployment of U of A rollup so I think we're we're in a position of uh Max awkwardness in the

puberty phase and hopefully from now onwards it's uh it's downhill I want to zoom out a little bit and kind of look at the progress of ethereum in in a sense of order of operations uh we Once Upon a Time had the very close-knit nucleus of the ethereum layer one and in 2019 after we kind of solved plasma which turned into rollups with this idea of fraud proofs uh we had these rollup teams that extended outwards from ethereum like these uh just like you know these vertical Integrations these

uh these things that would grow away from the ethereum layer one you know optimism arbitrum uh polygon all of these networks that would scale scale away like you would move your assets from the ethereum layer one you know away to these to these layer twos and this is what would create the fragmentation uh and now that you you illustrate Us in being in the interoperability phase of the rollup Centric road map now it's like we're filling in the space that has been created as these rollups have extended outwards from ethereum uh and there's I

think there's an order of operations right like first we needed to create the fragmentation problem before we could actually solve the fragmentation uh problem I don't I don't know if like we could have even gone back in time in the ethereum road map and tried to solve this fragmentation problem before uh before it was even created maybe you can kind of just talk about the the benefits of the fact that we first created all of these different rollups uh and now we're kind of just like filling in the Gap how do you feel about the order of operations of of how ethereum progress

has gone yeah I very much agree with you that uh it's a it's a logical progression uh I mean the could have been alternative uh path and alternative ways for for a FM to tackle things one way would have been to go down the the Solana route um obviously you know we don't want to be taking these kinds of of of shortcuts um another thing that we could have done is to scale ethereum layer one using um basically randomly sampled committees which you can think

of as as sharding and so here the the safety of each shard is not uh done through fraud proofs or through validity proofs it's done through naive re execution uh from from committees I guess what we did is that we took a a very pure path where we said that we don't like honest majority assumptions and we know how to do these things without an honest majority assumption and so we kind of opened it up to to to the broader Market to build these rollups that are um you know in some

sense fundamentally fundamentally more robust because they only rely on on pure cryptography or you know these these cryptocon games which are for proofs you know reflecting on the past and you know the different the different strategy that we could have taken I guess one advantage of this this this shotting approach that we didn't take would have been that we we would have had more homogeneity across the the different rollups but the downside is that we would have had way way less um uh experimentation and I'm actually quite

pleased that we don't go down the the the Shing route because there's a a new way to do execution Shing which is so so much more powerful so first of all it uses uh SN ZK rollups to give us to remove the honest majority assumption that we knew how to do but now the technology has caught up that we have line of sight towards having D data layer one but the other interesting aspect is that we can take this uh ZK

evm and and create a pre-compile within the evm to programmatically deploy instances of the evm so with the the vision that I that I described that we didn't go down we would have had a fixed number of of shards so you know something like 64 or, 24 and that is kind of an awkward position to be it's just an artificial number that we had to pick whereas with this pre-compile we can have as many as we want and then the other aspect of programmability is that

we allow room for creativity so we can for example have um zkm with different sequences with different governance tokens with different fee mechanisms everything around the virtual machine is fully programmable I'm wondering Justin if you have an opinion on um the following question uh in addition to kind of fragment fragmenting the ethereum network because we have all of these different rollups uh We've also kind of fragmented defi apps right there's many different implementations of Unis swap across all these different networks there's many different

implementations of a across all of these different networks and uh defi is something that is strictly better when everything is kind of contained in the same environment uh composability and liquidity Capital it's just more efficient in fact when everything is in the same spot so in a future of ethereum that we're hoping for do you see that there are fewer implementations of these a Unis swaps uh of these applications uh and is that kind of something that we should be uh striving for or hope to be

striving for is that like a success scenario when maybe instead of like 27 different uh implementations of of Unis swap across all these different chains there's actually just like maybe just one maybe there's just one imp implementation of Unis swap or all the the liquidity has aggregated maybe there's just one major implementation of a is that a goal is that something to Aspire to or is that maybe something too prescriptive I mean I'm personally a believer of the app chain fees and I think it does make sense once the barrier to entry to deploying an app

chain is low enough that there is you know one uni chain one a chain etc etc the issue that we have right now is that each individual app chain comes with overhead so for example if it's a ZK app chain each one of these ZK app chains need to verify a ZK proof and if you have a thousand uh ZK app chains then you need to verify a thousand proofs every once in a while and that's just very very expensive so we need infrastructure to for example aggregate

the the SN proof so you have one SN proof to verify a thousand app chains and it's a similar thing with the blobs the blobs are like fairly large chunks of data and it's possible that a single app chain only consumes you know part of it let's say one quarter and so you want to take all of these different app chains and like do the blob packing together so you have optimal utilization of of of resources um a similar thing to block packing is this idea of data compression the more data that you're compressing the better the compression

ratio and so what's happening right now is that each individual rollup is doing its own little isolated compression when really what we should be doing and this is a you know a network effect it's a win-win thing where we take all of the data across all of the rollups all of the app chains and we do one F swoop uh data compression Justin you were talking uh earlier about you like um your you're happy you're satisfied that we took the rollup Centric uh road map because ZK has made pre-compiled on the L1 possible

and so we have more uh potential for you like rolling out ZK based like VMS and incorporating that into the layer one I want to isolate that for our conversation because I don't think that is the based rollup conversation right this is something that we're talking about in the medium to long-term future and is something that is tangental to based rollups am I correct in that that's exactly right uh it is orthogonal to based rollups and it's called native rollups so when you look at the uh an L2

there's three different important criteria criteria number one is the the data aspect do you put data on chain or do you put it offchain the former is called a rollup the latter is called a validium and then the other layer you can look at is the sequencing layer like how is the ordering of transactions done and here you have base rollups which use the L1 for sequencing and you have non-base rollups uh that don't use the L1 uh for sequencing and then the final

aspect is the virtual machine the execution if you are exactly evm equivalent because you're using this pre-compile which is basically ethereum exposing the evm to the application layer with this pre-compile then you are a native rollup as opposed to being being a a custom roll up with a different virtual machine than the exact copy of the evm I mean we should do an entire episode separate episode at some point in the future probably on on Native rollups because it sounds like

that is kind of the almost the final evolution of bringing some of this execution back to the uh layer one validators the reason we're not talking about it for the bulk of today's podcast is because it's a it's a more distant future right think of that as the the medium to long term whereas base rule UPS my understanding Justin that's the the here and now like we're ready for base rollups essentially uh at this moment this is not a distant road map sort of EIP in the future that we're going to incorporate in some hard Fork though there might be some upgrades

necessary to fully optimized based rollups but is it right to think of based Roll-Ups as maybe the next evolution of Roll-Ups in some sense but the final evolution of Roll-Ups and I don't even know if this is the end game but maybe I'll just say the end game for uh rollups is uh is native rollup technology is that how you think about it yeah so the very nice thing about base rollups is that they they don't require hard fork and so you have dozens of teams now that are racing to go uh build all the infrastructure and deploy them uh to say that they're like ready

for prime time might be a a little bit of an exaggeration in the sense that we only have like one base roll up in production at Tao and you know there's adjacent infrastructure that that is being built out but I do agree with you that you know the the progress in base laps is definitely happening and I think 2025 and 2026 should be like very big years for for for for base sequencing as for an native rollups in order to get to that you know very exciting endgame we

need ZK EVMS but not only that we need realtime proven ZK EVMS and the the good news is that we're we're definitely getting there we know in theory how to parallelize snog proving so that you can get the latency down effectively as low as you want and uh historically you know we've had snu ification at the at the of the evm where a single block would take multiple hours tens of hours and then it got down to you know single digit hours

and then you tens of minutes single digit minutes and now we're at a point where it takes tens of seconds to prove a layo a layo one evm block and I wouldn't be surprised if in 2025 we we have like single digit second proving latency for for the layer 1 evm oh very exciting you can expect bankless Nation at like a future episode I'm sure on Native rollups when that um you it comes closer to fruition but now now let's get to maybe the main episode topic and and the bulk of where we want to spend our

time because all of what we just said is sort of a context it's a it's a preamble and maybe uh to your point Justin we are not ready for prime time for based rollups but but let's maybe call this the on of based rollups because we are off zero we have at least one deployed in production and you think and I think we share this belief that that 2025 is going to be a big year for these things we call based rollups I I feel somewhat that we should start with a definition but um before we get there I just want to say the promise of based rollups

there's somebody on Twitter who made the comment that um hey if we had based rups when we started you know rolling out the rollup Centric road map in you know 2020 2021 then we wouldn't have the fragmentation problems or value of cruel questions that you were mentioning these mimetic problems in in the social air if we just started with based rollups and it's almost like it's a shame we couldn't just start with based Roll-Ups to begin with because the promise of these things as far as I understand it and this is again the mimetic narrative that we'll have to dig into the validity of this

that it solves fragmentation you have shared liquidity across all of the based rollups and indeed also with the layer one so isn't that nice we don't have the fragmentation problem and two it also solves the value acral problem to eth because we are using the validators uh as the sequencers and so we get to preserve some of the security around you know censorship resistance uh at a deeper level than maybe a traditional rollup and also eth validators get some fees from this transaction a greater share of the fees from this transaction so if if we just

say those words it almost sounds like oh base rollup solves everything and so coming into the the episode can we title this episode based rollups solve everything Justin give us a definition for for based rollups and you like and tell us the promise here so um just to you know to set expectations base rollups don't solve everything um they do try and Tackle like one specifically of the stack which is the this the sequencing layer um and basically when you you build a rollup you have a choice as to what is the sequencing mechanism

who are the actors who will determine the order of trans actions now since the the dawn of blockchains you know Bitcoin 15 years ago we've had essentially just one thing which is called base sequencing where the L1 is doing the sequencing the validators the consensus participants uh whether it's miners whether it's validators all doing the the the sequencing and if you look at every other blockchain you know whether it's you know Solana you know BNB you

tesos xrp you name it they all have base sequencing and for you know the vast majority of the existence of ethereum everything was just base sequence that is the the natural normal default you know canonical way of doing uh sequencing but then something happened uh a few years ago where we saw the emergence of these centralized sequences specifically in the context of the the rollup Centric uh road map and I think I

I attribute it to you know three for different reasons extremely pragmatic reasons uh for those different uh rollups one of them is that they could launch extremely quickly and it was a very very straightforward as to what needed to be done just spin up a server and that server is responsible for all of the sequencing but there's other things that that uh centralized sequencers provide one of them is that they are a training wheel for security so a lot of the

virtual machines um of the rollups are secured by fraud proofs or validity proofs and these are extremely complex pieces of of code you know potentially thousands and thousands of lines of code and it just takes one critical bug to jeopardize the whole rollup so you could imagine like draining all of arbitrum in one F swop that would be absolutely terrible and so what the uh rollups uh

sequences are doing is making sure that the blocks that that end up being produced are created using you default software as opposed to software that is specifically crafted to go exploit a vulnerability right and so as a black hat even if I identify a bug in for example the arbitr and fra proof I'm not able to exploit it because I'm not able to convince the shed sequencer to go build a block that will go exploit this bug the other aspect um the other service I

guess you could say that centralized sequencers uh provide is meev protection for for users so traditionally You' send your transaction in a public mol you gossip it and now you're exposed to front running and sandwiching and that's now that's a terrible user experience with centralized sequencers wallets can directly connect to the centralized sequencer it's an end to end encrypted connection so the only entity that is able to see the transactions

before they land on chain is the centralized sequencer and that entity is kind of trusted to not front run and not uh sandwiched uh their users and so that leads to a great user experience and then another final service that the centralized sequencers provide is that of pre-confirmation or extremely low latency uh user experience you know on the order of I two seconds for for the the super uh the super chain or even much faster in the context of arbitrum

on the order of 250 milliseconds and you can think of the centralized sequencer as being a cheat code AS has you know taking some sort of a massive shortcut uh which is you know very strategically meaningful to do because it's very pragmatic but it's not long-term sustainable uh and there's all sorts of reasons uh for that uh one is that it's not credibly neutral and we'll talk about that and that means that it makes it very very difficult to convince

you know your competitors to join forces in in Shar sequencing like would you know the optimism super chain want to share a sequencer with you you the offchain labs and and arbitrum it's very very difficult to pick one server for all all of ethereum um another issue is that it is a regulatory uh central point of of failure um it's also um a a vector for

accidental liveness failures this is something that we've seen in the past where rollups just randomly go offline for for sometimes several hours and you know that is not what we want to uh base the the the future of Finance uh on and then finally there's a very very big you know trust component to Central Iz sequencers we haven't seen it yet but we could imagine for example a centralized sequencer getting hacked and then you know millions of dollars of front running happening uh because the the

centralized sequencer has been has been compromised and then another aspect is that because centralized sequencing fundamentally leverages trust as the technology uh it's unscalable and so if I'm a random Joe you know I'm a 16-year-old uh you know programmer in my basement and I want to go deploy a a roll La well I don't have any trust and so uh I can't go spin up my own centralized sequencer and so I think what the ecosystem has realized is that

really we want to go down this route of decentralized sequencing that is the I guess the the end game the proper way of of of of do things the the long-term sustainable path but not only that uh we also want sh sequencing and this goes back to the this idea of network effects and Universal synchronous composability in order to achieve this this shared sequencing in an ethereum wide capacity you know going above and beyond the

Clusters that are being formed with arbitrum orbit and the super chain and and the the the ZK sync elastic chain we need credible neutrality and so I think we're going to spend a lot of time in the rest of the episode talking about one thing credible neutrality I want to um keep on unpacking the sequencing thing because I think you alluded when we asked what what's the definition of a base rollup you alluded to the fact that like well rollups base rollups are the same as all other rollups except for this one thing that's different which is how we sequence things that is that is the difference between base rollups and

like arbitrum optimism is that the it's only the sequencing layer uh which is critically different in a base rollup versus the Roll-Ups that that we know today um but I do want to P put a pin on the typical rollups and how they sequence things the the centralized sequencer because I think in the crypto industry we are all kind of like allergic to this word centralization and sometimes overly so uh decentralization that's not decentralized that's not decentralized sometimes it's used as like a cudgel uh in ways that it that it shouldn't be uh leaning into like what

ethereum does and what ethereum provides its rollups the whole point of the rollup Centric road map is that we can actually have centralized layer twos because of the checks and balances that the layer one provides its layer twos so like the layer twos are in a state of can't be evil uh even when they have centralized components and that's actually has been articulated uh in the E theum rollup Centric road map that that's actually strength of rollups in their current form like arbitrum can be allowed to have a centralized sequencer