Ed Felten on Arbitrum DAO, Timeboost, and His Newest Research

Transcript

what a day it's the beginning of ARB verse I mean two years ago people didn't even know what offchain labs was yeah it's the ultimate decentralization one person armed with the truth can win out why are l3s compelling to you and what about orbit makes it a really good home for l3s I think l2s can be more Nimble than ethereum nice hat Ed thanks yeah I I think I need one of those we can talk after yeah okay fantastic I mean what a what a day it's the beginning of RB um I mean two years ago people didn't even know what offchain labs was rollups were completely new and now we've kind of got this like bustling ecosystem of Builders users um how do you feel about that about the growth overall and just

broader Reflections I mean it's incredible when we started this when we started the company in uh 2018 uh it was just like three people in a room um and we had an idea kind of of what we wanted to do but to see it grow into this you know the offchain labs team the arbitrum foundation and then this incredible Community having events like this arbiv verse um yeah I would never have imagined this could happen yeah I think that that the arbitrum ecosystem has gone through like several different uh like Evolutions right it it started as uh you know first like like the arbit from airdrop was a a very notable one for for being a large air drop for ethereum users and that kind of got this like initial Community

bootstrapped then really kind of dominated on defi as well which I still think is very very prevalent gaming and then now it just evolves so much um how do you kind of uh conceptualize what arbitrum is like you've got all these different product SES and what offchain labs kind of is like like how does this come together I mean one thing that's really changed is we started out with a single chain right what arbitrum one um and that was what it was right it was fairly robust it did it did the job but now there's so much more variety so much more diversity in the ecosystem so many applications but also now with orbit uh lots of different chains doing many

different things for gaming or for other uh for other sectors so um you know I think I think what started as just one piece of technology now really is an ecosystem U there's so many teams that are building within the space and that are you know have decided to sort of uh to you know to commit to being in the arbitrum space um and as that happens the community takes on a life of its own that's the thing one of the things that's really exciting to me is the arbitrum Dow sort of achieving uh that escape velocity to where they are uh really um operating as a community and they're being strategic

you have this really sophisticated conversations going on in the Dow about what the direction should be um you know it's not it wasn't a given when we started and launched the Dow that it was going to turn into this um but this is kind of what we imagined or hoped it would be yeah yeah I mean um human coordination at at massive scale is very difficult without that kind of top down uh hierarchal leadership and I think to uh we we saw this really big emergence of dows a couple years ago and they've kind of somehow like fizzled into the back into the into like the background of of crypto today but I think ARB is a great example of of how they still power some of the most important networks in

the space um but are are they've kind of taken this backseat well I think it it differs actually I think uh D some Dows are different right um one of the things we worked really hard when we sort of envisioned and initially uh set up this or this this thing before we sort of launched it and set it free um you know we really thought hard about what we thought would be needed for the Dow to be empowered right um and and that included a bunch of things but um and it some of those things are technical right the way that Dow voting Works where when the dowo votes to do a thing that thing automatically onchain happens right it's

not just a polite request to some multisig to please do something right the Dow has ultimately holds the holds the ultimate control um and that was an important part of it um but um then I think you know it takes a lot of work for people to sort of uh to sort of foster that the growth of that Community um but in some ways it's like uh you know raising a child in the sense that you know they become more independent if you're doing it right they become more and more independent and then they reach the point where not only they're adults but they're actually more capable and impressive than you um which is of course the goal that you should have um yeah um but that does

that doesn't always happen it doesn't happen automatically right right um I think you know from the very beginning even the main that launch of of um of arbitrum one uh you know we put a lot of thought into the fair launch uh concept right this idea that we really wanted to foster a variety of different applications not just like pick one winner in each area but uh but to say you know we want to have a big tent we want to welcome every everyone in we want to provide support and a technology that is easier to adopt so then you get a broader community and I think that sustains over time yeah so I understand that uh you're quite deep into a lot of the research that goes into arbitrum itself arbitrum one orbit stylus multi-m

kind of the whole lot and so compounding off this uh Community oriented discussion offchain Labs itself is also pushing forward a lot of research uh over the last several years years and looking ahead what are some of the most important research breakthroughs or initiatives that you guys are are pioneering today uh to to push the the future of Harish forward I mean I think I think the things that are sort of next on the agenda for the Dow to look at or the Dow has been looking at are bold and time boost okay bold is the NextGen uh proving protocol that allows uh allows permissionless validation um which is really kind of a remarkable thing that

any one user anywhere in the world with just a single regular laptop can force the chain to stay correct and can force it to make progress so even if everyone else in the world is against you if you with your laptop can be can be the hero who makes arbitrum continue to operate um and that's kind of incredible from the standpoint of decentralization right it's the ultimate in decentralization one person armed with the truth uh can win out um so that's bold um and then time boost is a mechanism for uh that does a couple things uh it's a slight change to the way sequencing works but the bottom line is that it reduces the amount of spam transactions on the Chain um and then it also allows the Dow to

capture uh a bunch of the value of me that the that their chain is is generating got it so on the on the Bold front the first point the core breakthrough there with this honest minority uh assumption is is effectively that as long as there's one good actor that is that can hold the the the information about the arbitrum one blockchain uh on a very easy CPU device MacBook Pro then then the Integrity of arbitr one itself can be uh sustained exactly right you can protect the Integrity of the chain and you can stop the and you can and you can uh cause the chain to actually keep making progress right all by yourself okay and then on the on the time boost proposal this was

this was more of a of a breakthrough uh from the previous way that l2s specifically arbit one and others were sequenc kind of a shift from that first come first serve model now to almost what is like proposer Builder Separation on the L2 but not exactly yeah I think I mean I think it's a better fit for l2's needs I mean l2's and arbitr you know is one of the pioneers of this uh do the sequencing or block building differently from L1 in a way that is more protective of users right right protects you against front running and sandwiching uh for example um and is also very fast right much faster than L1 uh you have two you have uh 250 millisecond block time and that can go down to at least

100 milliseconds there are orbit chains that do that so I think arbitrum sequencer has been great um in a lot of ways uh but the one thing that it didn't do is uh capture uh capture Meb value um and um and that's something something that uh time boost lets you do um and so that's available to the Dow I know that you know delegates have been looking carefully at this there's a lot of discourse around it and the other thing about sequencing that's currently um uh you know we've been talking about is a project that we've been doing jointly with thepro systems to decentralize the arbitrum sequencer so that is basically take the the sequencer that you know and love plus time boost and then make a

decentralized version of it so that you don't have to rely on a on one party uh to ensure that the chain can make progress efficiently so this sing so this single sequencer model which I I prefer that term over centralized sequencer model as well this the single sequencer model is highly effective for um transaction ordering and for uh the ability to to basically have very fast pre-confirmation and extremely low latency transactions moving into the next stage of arbitrum requires would you say more censorship resistance or what would you say is like the big I know that that the you know anytime that I tweet that I'm like when will when will these l2s ever decentralize their

sequence or it always just blows up because people are always like kind of pushing it but I'm not sure they understand like the core properties needed right so yeah so I mean some of sometimes when people complain about sequencer being centralized it stems from not understanding what the sequencer does right but based on what you're saying here that's not you um the uh right because they think that the sequencer is responsible for ensuring that the chain is correct and safe but that's not we would never trust uh a single entity to do that right the sequencer job is just to report which transactions arrived and in which order and to do that accurately right um it does mean though that the sequencer has

the power if it wants to censor you um it can slow down your in the inclusion of your transactions quite a bit um and so can somehow discriminate it can let its friends transactions in faster or other stuff like that right if it misbehaves and right now the model is there's one sequence or it's chosen based on Community Trust and pledges not to not to cheat in those ways right so but what you would prefer to do um I think is to have a committee of sequencers and then say unless there's some pretty big collusion among them to cheat you have a guarantee that this Collective of sequencers will behave like an honest single sequencer right

and so from a user standpoint you still send your transaction in it gets sequenced pretty fast and uh you still have good user experience but it's more robust because there's no one party or even like small group of parties who who can cheat or put their thumb on the scale in the sequencing yeah yeah and I think that the community sentiment it's good to have the community kind of back stop of always like pushing teams to uh move to the next stages of the centralization whether being L2 beat stages or just like the like like the next iterations of of kind of like what the best alignment with the core ethos of crypto why we're building here today and pushing that forward for teams exactly right so decentralization you

know if you're not if you're not trying to decentralize then why are you on a blockchain at all right um because um I mean that's the big virtue that you get with a with a blockchain and you know the arbitrum stack and from one have been decentralizing progressively but really the two pieces of decentralization that are left I think Technic at the technical level are um completely permissionless validation which bold does um and then decentralizing the sequencer so you get this protection against discrimination by the sequencer got it on to so tying back to to the to the Justin Drake beam

chain L1 improvements at the consensus part um a part of this was was technical debt of ethereum and we talked and and through the presentation we you know we can understand that through the work from proof of work to proof of stake there there is technical debt accured uh and this also applies to rollups um so like I'm just asking you objectively here like do you think arbitrum has technical debt that needs to be addressed or is there a route for uh L2 to address their technical debt uh periodically to kind of improve as there there could be more l one improvements I mean I think any software project has technical debt anyone who tells you otherwise is has has either is lying or has never been in this business right um

and one of the things that's brilliant about technical debt the analogy to debt right is that um it kind of makes clear that it is like debt right it it's not the right policy to never have any debt yeah it can be paid off it can be paid yeah and that's the thing you can't get in over your head number one number two you're going to have to pay it off eventually and number three until you pay it off you're going to you know you'll be hampered by uh by interest interest which is really kind of the drag that technical debt puts on your ability to do everything right um and so yeah it's a process of managing technical debt in the codebase and managing technical debt in the protocol I think l2s can be more Nimble than

ethereum um and it's that's appropriate right ethereum is um has more at stake um and it is the foundation of security for you know all the l2s and l3s and so it ought to be really cautious and move very carefully um l2s can lean forward a little bit more um because governance Works differently in l2s I think also you can go more quickly right L1 has um L1 can only change by sort of the emergence of a very strong social social consensus because every time you do a hard fork in ethereum they you know you need to be confident that virtually everyone will will follow the for right

whereas in a an L2 you can have a dow vote you decide you implement you move on it's done right right um and it's important to doe diligence right but you have the benefit of actually reaching a decision you don't have to debate until you know one side uh decides they um uh decides to stop fighting right okay so with that lens in mind of the of the L1 kind of uh slowly but potentially changing in the future this like like one of the big reasons that we discussed over the last 24 hours is um is this is this interoperability between l2s and so arbitrum arbitrum orbit has this kind of native interrupt that that's that's

there within all the orbit l3s and l2s similarly there are other rollup clusters as as well how how important is it for the orbit cluster to you to be able to seamlessly interoperate between other clusters in the ethereum space and like how do you think that's going to happen because I'm I have this kind of thesis that we we'll see more like coordinated block building or like shared sequencing within clusters in these enshrined um zones and then more like intent based systems or like transport uh layers between so like what is your mental model of like the importance of that and how do you think orbit cluster will interact with the others yeah so I tend not to think of

orbit as a single cluster it's kind of more of a constellation okay um I think it's important um I mean there are different ways of managing a group of chains right one way is to say we're going to have unified governance we're going to have unified infrastructure um and we're going to operate um almost like sort of different branches of the same organization right all these different chains and that's one way you can do it there are some advantages in the ability to interoperate from doing that but I think that once the once sort of the research on Cross chain interupt gets more mature that Advantage is really going to dissipate and the other model which is more like uh kind of more like the

internet where um you have these separately managed chains and they're each making their own decisions about when to upgrade and how to configure and of this and they can still communicate and interoperate with each other pretty well um and if you can get to that endgame then you can have good interop even across chains that are very different that are managed differently that use different technology stacks and so on it'll take us a while to get there as a community partly because we I think still don't know what are the best technical ways to manage that um although that's you know that's one of the biggest areas of research we have going on at offchain Labs is to understand try to get a really crisp idea of uh what we think is the best way

to manage Crush chain interrupt but I think you'll have something more like the internet right where you don't have to have one Central it's not like corporate it right it's more like everyone manages their own stuff and uh you figure out the Technologies you need to make it work together of efficiently and smoothly safe is moving the world GDP onchain with the introduction of safet safe is taking this to the next step very soon check out n.s. Global to learn more about how safe is moving the world's GDP on chain welcome aboard on our journey into the singularity this is the endgame road map for fra and fraxel the new L2 out of the fra Eco system

that is totally based on fractal scaling fractal is a modular L2 blockchain built by fra with modular and fractal scaling in mind fractal scaling is based on the mandal BR set of fractal images and it explains how l2s become l3s and the entire app chain life cycle goes through its Evolution the frax ecosystem is made of some of the top builders in the space and has stemmed from its OG days of the frax stable coin it is now in the LST space the LRT space and this fraxel L2 joins the existing Suite of fra products

including Frack Swap and fra lend you can see fra fairy the bridge to go from any chain arbitrum optimism or main net into the fra main net Bridge your tokens and bring your FRA assets into the fractal ecosystem join today to get the best out of fraxo you are still early join today take part in our Expedition into the singularity Redstone modular oracles are available on 50 plus chains trusted by over 100 clients and over 6 billion in total value secured visit redstone. Finance to find out more and

would you say for arbitrum one of those Technologies has been this recent breakthrough of the fast withdrawals how does that kind of tie into the picture yeah I mean fast withdrawals is a feature that's available for uh chains that use the any trust technology right and it lets you do a withdrawal from say from an L3 chain down to arbitrum 1 in about 15 minutes um so even though you have a even though optimistic rollups have traditionally had a 7-Day delay and ZK rollups you know hours of delay typically uh to get to finality um because of without getting too deep into the weeds because of the way any trust works you have available this option to safely confirm um uh or settle

L3 results to L2 much faster and the way that caches out for users is is this fast withdrawals uh so that's that's a piece of it but that's 15 minutes right from a user experience standpoint 15 minutes is way better than 7 days for sure but you know we we aspire to better than that right um you know you want to get to the point where the user doesn't get bored waiting for the for this to happen right um and you know the question how fast you can make it can you have you know sub 1 second cross-chain interop that is trustless and doesn't require the chains to be under unified management I think that's possible um but you know we have some distance to go to get there but that's

like for me at least to me that's the goal of interrupt research uh right we don't want to compromise there's sort of these three things right we want independent management we want it to be super fast and cheap and we want it to be trustless right and I don't think that's a trilemma I think you can actually get all three and that you know that is where we are working towards as a as a space is yeah it's where we want to get to but you know but we don't know we don't know all the details of how to get there I think I I kind of feel like we can see it from here but we don't know how to get there yet I feel like that is applicable to a lot of different research areas in crypto today it feels like it's like just out of arms length

you know it's it's getting closer than ever which is is something to be optimistic about um I also understand that you are maybe as a contrarian take but maybe not is that you are bullish l3s or at least you are extremely interested in the development of l3s where other potentially more monolithic chains will yell at ethereum in the ma system as bro do you need more layers L3 L4 L5 L6 right so why why are l3s compelling to you and what about orbit makes it a really good home for l3s I think there's a couple reasons why I'm interested in l3s one is the market is interested in l3s right there are a lot of l3s they do a lot of traffic right l3s do pretty close to half of all the

gas all the traffic in the ethereum stack is is measured by gas um and that is um right and that's that's a lot right and I that surprised me somewhat I thought we'd see some healthy L3 traffic but not as much as we have so that's one reason right the market seems to want it um and you can kind of understand why I think if you dig into the details around things like cost of operating a chain right the big advantage of l3s is that you get to settle to L2 instead of L1 and that helps you two ways one one of your big costs in operating One of These Chains is the cost of settlement which means you're buying gas from your parent

chain yeah right and you'd much rather buy L2 gas than buy all one gas arum one gas is like going to like Thailand and on your mod overs is like you're like truck in the states yeah or like flying first class yeah on a Jet right yeah um so that's a big plus and then you know also l2s if you're talking about an L3 on top of arbitrum 1 which is like the most popular place for l3s to home um you know arbitrum 1 has features like stylus that let you do things say in Cross chain bridging or in verifying ZK proofs that you couldn't do on L1 yeah um and so for all those reasons you know there are advantages to having your parent chain be be arbitrum one um or in general in L2 um you know there's some

drawbacks to being an L3 as well but I think people find those must find those advantages compelling yeah and so you know I think you know you could ask Will there be an L4 I don't I mean I l3s give you enough throughput no l3s give you yeah I think l3s give you enough and sort of getting off ethereum which is just a different animal than the l2s uh as your parent chain uh gives you most of that benefit so as we kind of come to a close here Ed um you you hinted at some of the interrup research that you've been doing um what other research has been particularly fascinating to you whether inside of arbit ecosystem or outside

what are you seeing that it that you you think has a lot of potential to push the space forward yeah um so we''ve been been we've been focusing a lot on what you you can do to make the speed of individual nodes faster right there's a bunch of ideas out there in the ecosystem and some of them well formed some of them more uh sort of more nent um but sort of really looking inside what happens inside a node and how can we make that faster number one and two make it safe so that uh we don't like you know in the same way that you can't just like increase the gas limit infinitely because uh then nodes can't keep up right how can you squeeze more

performance out of your chain while still making sure that that that home node operator uh doesn't uh is able to keep up um and so I think there's a ton of fascinating stuff and as a person who's been interested in computer systems research kind of what's inside the guts of a of the way computers work uh you know for my like my whole career um I find that really interesting um so I mean I think that's important sounds like ZK ZK is ALS partly there um I think it's an interesting area yeah um I think ZK has value but is overrated in terms of like having a chain use all ZK proving um because I think uh you know

Swit if we were to switch from optimistic to ZK I think our users wouldn't even notice right except well they would notice the gas fees were higher because we had to pay for the cost approving right so ZK proving your chain doesn't really pay off much in terms of user experience um that said but it is on the road map in some capacity it is in the road map yeah yeah because the reason not to do it yet has to do with cost right right and so as that cost comes down um then uh you know and we understand how we would or will incorporate ZK proving into the arbitrum stack but the other interesting thing is ZK proving is not just for like proving all of the computation of your chain

it's really important tool in your toolbox as you're designing protocols and using ZK proving more locally in lots of different places in the protocol I think there's huge advantages there and our team is kind of going going nuts on um identifying opportunities to use ZK to do interesting things awesome glad to hear it and Ed thank you for the time I hope you've had a good week so far in in Bangkok it's been great uh been great to be in Bangkok and Devcon is always a ton a ton of fun a lot of Builders no fluff a lot of a lot lot of Builders a lot of fun serious conversations and also just a good time yeah it's a it's the Gathering of the tribe thanks saided thanks