Joe Andrews - Decentralization or Delegation? Rethinking How We Build Validator Ecosystems

Transcript

Um, hey, welcome everyone. Uh, it's great to see so many stakers here. Um, I love coming to events like this one. Um, it's where kind of the real uh, people behind Ethereum are. So, thanks for having me. Um, my name is Joe. I'm one of the co-founders of Aztec. And before we start, I'll just give a bit of context on what Aztec is in case you guys aren't there. Um the talk today is going to be a bit about our lessons um kind of bootstrapping Aztec staking sets and hopefully will be a good lesson for other people thinking about staking in different ecosystems or building their own uh decentralized validator sets. Um we're going to cover kind of what our view of decentralization actually means.

Um why we think foundation delegation can be a trap. Um and how we think about building different ecosystems. and then we'll have some time at the end for questions. Uh before we start, um what is Aztec? Um and I guess why can we speak about this? We've been building kind of a fully decentralized programmable uh world computer or private world computer. Um the goal of Aztec is it's permissionless from day one. Anyone can stake um and it's kind of a stage two rollup uh on Ethereum mainet. It enables kind of optional privacy at every level of the stack. you can have user privacy, smart contract privacy, um, and even kind of like, uh, execution privacy as you go.

In terms of our progress, um, we launched a decentralized test net uh, earlier this year in May and we had about 10 to 20,000 nodes from all around the world and we've kind of taken lots of learnings from that into our current uh, estate. Uh, the current ignition network is bootstrapping on mainet as we speak. I think we're 80 to 90% of the way there to it launching and the execution environment will launch next year as a stage two rollup um on Ethereum. All right, let's dive in. Um in terms of kind of decentralization, um we've built Aztec to be decentralized from day one and we've thought a lot about kind of uh this question like what does true decentralization actually mean?

I think about it in terms of four categories. Um and each of these is needed uh to actually effectively decentralize a network. Um the first one which is kind of an obvious one is around stake distribution. Um you know how is stake distributed across your validator set? Ethereum's seen this kind of firsthand with like large institutions like Lido or kind of Coinbase owning large amounts of the set. But you need to think about, you know, what threshold um does any one entity have in the set and how many entities need to collude before you break some honesty assumption which could bring down your network. If you're not kind of using Ethereum's consensus and you're using kind of a different form of consensus, it's usually going to be around a 33% threshold or a 66%

threshold before something goes wrong. Um especially if you're using something like um comment BFT. The next point is access. Um so who can participate in block production? Um is it permissionless? Who can participate in staking? Is it permissionless? Who can participate in governance? Is it permissionless? Um a lot of kind of new networks keep this behind closed doors. Um and so access to the validator set is gated. You have to go and you know apply on some form. Um uh you have to apply to a program to get delegation to stake.

And then the last kind of like but maybe not so obvious one is sometimes networks are permissionless but it's1 to5 million to participate. Um and then people are basically priced out and can't afford the stake. Third point I want to talk about is diversity. Um and there's kind of three camps here. There's the location of your stakers. Um where are they located in the world? Uh what client are they running? Ethereum does a great job of this. Um but on kind of smaller networks it's usually one client. Um, and then lastly, kind of who are they? Are they institutions? Are they solo stakers like a lot of people in this room? Um, or are they kind of like validators as a service? And you don't want any one group of those to outweigh. You need all

of them because they all bring different views to the network. The last point is one that's usually forgotten. Um, what we see with most validator ecosystems is other than Ethereum, um, people don't own their stake. They just own the rewards that that stake entitles them to. And this is a big mistake because it means that validators are not economically aligned. They just do what is required to extract those rewards from the ecosystem. They don't actually think about what's needed to make the network successful because they're not invested in its long-term future. If the revenue stream dries up, they'll move to a different ecosystem.

So, this is what I call kind of the foundation delegation trap. And uh it kind of goes a little bit like this. Um, I think foundation delegation came about as a result of like the Biden administration and as a way to kind of bootstrap decentralization uh when we had very um very kind of like strict uh laws around distributing tokens. Now we're in like a new era. Um I think they actually do the opposite and if we're not careful foundation delegation can entrench a small validator kind of monopoly or oligopoly and they prevent um the market finding new operators um who can kind of come in and prove their competency. Um and in the worst case

they delay or permanently prevent decentralization. Um and so you end up with networks kind of stuck in this limbo of temporarily centralized and their kind of uh ecosystem is clinging on to the foundation uh because they don't actually own their stake. I think about it a bit about a bit like um scaffolding like you need to build a building and then you can put scaffolding around it like you can't just like use the scaffolding to support the building. In terms of memes um it's not that all delegation is bad for us. It's the ordering that's really important. Um, so if you kind of try and use foundation delegation to bootstrap a decentralized network, it's probably going to be quite

difficult. It's much better to try and decentralize the network from day one. Decentralize ownership um at the start of the network. That's when it's easiest to do when the network's kind of in its genesis. And then you can use foundation delegation to maintain decentralization over time. You can kind of think about it a bit like uh a tool to combat different centralizing forces. So if you see on a staking dashboard that you're missing validators in South America, great. The foundation can start delegating there to, you know, pull uh the validator count up in South America.

If you launch a new client, you can make a program to kind of incentivize the usage of that client. But if you start with it, what you end up doing is kind of entrenching an existing group of people who um maybe will keep the network centralized for a long time. So our approach um we've been seeking out kind of solar stakers, institutions and node operators all around the world and kind of give them all equal access to the network. Um we don't have foundation delegation for the first year. Um we want to ensure the network bootstraps decentralized. um it needs to kind of launch in a way without the foundation and then over time the foundation can use delegation to make

sure it stays that way. Um no insiders can stake for the first year. Um and only the community can stake and own rewards. So for us this means that the community has pure kind of ownership and control of the network which is really important. Um it means that they're kind of economically aligned with the protocol and they're not just waiting for the next delegation program. Um last point is yeah I think it means that validators can compete on performance um and not just connections. So new validators can spring up and um we're seeing that a lot in the Aztec ecosystem either first-time stakers or new businesses that want to kind of prove

themselves uh in a new chain. So as I said earlier we started with a fully decentralized test net. um we have between 10 and 20,000 nodes running fairly consistently from about May um till now um on every continent that has people on it. So I think no one managed to do it in Antarctica. Uh but we're still we're still waiting on that one. Uh we have low hardware requirements and yeah, as I said before um uh a good group of kind of first-time operators as well. In terms of where we are today, um this is the network kind of bootstrapping on Ethereum. um when it gets to 500 um it will go live. Um so this is right now on Ethereum mainet um it'll be fully decentralized on day one and controlled

by this validator set and yeah no foundation delegation for the first year but after that point the foundation will look at the makeup of the network and start to kind of um delegate where it sees fit. I'll leave you guys with kind of two points that I think uh the ethos of Aztec and kind of how we thought about this problem. Um I think we need to build these networks um that are built to last. Um they need to be decentralized from day one, not just with a promise to do it later. And I think we need to avoid the trap of like uh foundation delegation being kind of like a shortcut to decentralization.

And the second point is yeah, I think it's not a silver bullet to just use foundation delegation to say your network's decentralized. You actually need to decentralize ownership and economic rights to the network. And that's a lot easier to do at the start of the network than it is at the end of the network or when the network's successful when the stake is uh much more valuable. So yeah, TLDDR use foundation delegation programs to maintain a decentralized network and combat centralizing forces. Don't use it to kind of take a shortcut. Um maybe I finished a few minutes early, so we can take some questions and uh thanks uh thanks for having me.

Yeah, we most certainly have time for questions and uh unlike last time, I'll keep an eye on the uh the live stream folks uh to see if there's questions from there as well too. But uh does anybody have a question for Joe? >> I guess I guess I did have one uh for playing around in the Aztec test net here a little bit and I guess it's approaching ignition here. um do you want to talk a little bit about um what happens during that process and and next steps, right? You mentioned in the presentation you're about 80% at your threshold, >> but do you want to just foreshadow a little bit about what happens uh once that threshold is met and how that's different from your mainet launch?

>> Yeah, it's a great question. So, um we've built kind of all of Aztec. Um what's deployed on Ethereum right now is a fully working version of Aztec. The block gas limit is just set at zero. And so uh it's set like that so we can test decentralized block production works as intended on Ethereum mainet. We're very confident it works on testn net. It's been running um uh for a long time on test net but with real economics and kind of you know actually competing for real blob space on ethereum. Um we need to test this uh in in the real now with um deploying it to Ethereum mainet. So the point of the ignition launch is to test that for two to three months on mainet. Uh it's a bit like the beacon chain. Um we'll run that uh until we're

confident that it works and then there'll be an upgrade by the decentralized staking set to basically enable transactions by increasing the gas limit. So that's happening in the background right now. Once we get to 500 uh nodes, the network will start producing blocks. Um and it's kind of at that point it's in the hands of the community. So it's pretty exciting. >> Excellent. Thank you very much. A question back here in the back. This is from uh Tio on on YouTube. He's asking um when's the mainet launch date?

>> Uh so the ignition like the first kind of point is happening right now. Um and uh we expect it will complete this week I think. Um uh it depends uh when when enough community members stake. Um the we're targeting kind of early early February maybe March for turning on transactions. The only kind of missing pieces are completing audits uh for some of the execution environment code. All the stuff on mainet is currently audited, but the actual execution environment for transactions is a lot of complex cryptography and so we're still auditing that ahead of mainet um uh in roughly February.

>> Thanks. >> Uh hi. Yeah, thank you for your talk. Uh I had a question which is about like uh the node operators that are writing uh uh part of the Aztec sequencer like what what is how do you balance like decentralization that the protocol obviously wants with like the performance like the quality of service that you want to get for like users like how do you generally think about that? >> Yeah, it's a really good question. So we're a little bit fortunate in terms of being in L2. So Ethereum has a problem where like every validator has to execute every transaction. um in our model that's not quite the case because we rely on Ethereum um to kind of uh verify the the computation with a zero

knowledge proof and so the performance requirements are a bit less as as a result for validators but there is still attention in the Aztec network. it's around bandwidth. Um, and so we we have to just pick a threshold that we think is suitable to attract home stakers um and get that side of uh the network um but also not jeopardize the network's goals. At the moment it's eight cores uh roughly and 25 megabits up and down uh connection um and so it will scale I think with MOS law after that and we can achieve reasonable throughputs with that. Um so that's how we we've got there.

>> Thank you. I think there was one more at the front maybe or Hello. Uh uh I have to make a statement. I'm a lazy homestaker. Lazy enough so that I never use things without like helper tools. That's why I had to develop uh CSM and Lido. Uh no the question uh I've been trying to get my hands dirty with uh ad tech testnet >> and even though my friends send me like a ton of videos on how to set it alongside Ethereum node again I'm a lazy homestaker. Do you guys have any plans on integrating to Eve Docker Dab node software or wherever else where a lazy homestaker like me can easily run this

ad stack stack nearby my Ethereum node. >> Yes. Uh I don't know if Cohen's in here but Cohen's our validator lead. either out there or maybe somewhere at the back. Uh we have a DAT node package. Um and I believe the OAL team is also looking at integrating it into their stack. Um but yeah, there's a DAT node package which should be one click for you. So if you got a DAT node, you're good to go. >> Yeah, I'm laughing at Dimma's question there because I too am a lazy homestaker and had the same experience. But we have one more question up here in the front. >> Hi there. Thanks. uh when you're talking about uh foundational delegation that sort of thing >> my understanding is >> um and maybe this is more for proof of stake L1's um

>> like part of the challenge is that you want decentralization but you also want enough stake there that a single whale can't just easily take over the network. >> Yeah. um how do you balance that and you know is that a a challenge? >> Yeah, it definitely is. I think like this is kind of why you have to do it at the start because if you you know wait for two years and your your network's successful and it's now worth a lot of money. It's very hard for people to acquire stake at that point. Uh but if you start this way um is a lot fairer access. Um there's still an issue with kind of civils and people pretending they're more than one person. Uh we use

kind of ZK passport which is a kind of proof of um passporthood to kind of verify that on the test net and try and make sure that like one person is not five entities. Um so there are still problems but it is easier if you do it at the start when the token has no price effectively or it's lowest price. >> Okay. So it's not entirely permissionless like there is a a limit per person. um in the so there's a there's like a sale happening at the moment there's a limit per person beyond that it's completely permissionless um but at the very start there's a yeah you just have to prove you're not on a sanctions list >> okay and and so if I understand right then um it's still like say a whale could take

over but um you know the network's not worth that much so you know who who cares >> while while it's still early delays. >> Yeah, there's a there's a normal like exit delays and entry delays. So, you would once the network's bootstrapped, you'll be able to see that um happening and the community can respond. So, yeah, it's it could happen the same way as Ethereum, but it's you can respond in the same ways. >> You have one time for one last question on the far side of the room here. >> Um on the ZK passport uh question, how was it developed? Did you guys develop it? Was it developed independently?

How's the documentation coming? Because last time I used it, I remember you couldn't like granularly share like see what info that you're sharing. I heard that it's different now. Um but it has the documentation for that and has and it is it open source? >> Yeah, it's open source. Um so it's an ecosystem team. So we gave a grant to fund it. It's powered by the language we wrote uh Noir. So they wrote their circuits inside Noir, but it's an independent team that that built it. Um, I think they used it as well for Dev Connect. So, if you're from Latin America, you got a free ticket by proving you're uh kind of uh from one of the Latin American countries. So, it's it's pretty granular now. And inside the app, you can see what it's asking you to prove. So, are you above 18? Uh, you

want a sanctions list? Those sorts of things. Um, so yeah, it's it's really granular. >> Excellent. Um, well, the round of applause for Joe, please. >> Thanks, everyone.