You Probably Don't Need Your Own Chain: Evidence From Dissolving One at Scale

Transcript

[music] Thanks for the warm introduction. Um, cool. So, uh, what I want to share today, I want to talk about one, uh, major experience that we had recently and to have a discussion about app chains, uh, versus, uh, the alternative. [snorts] But before that I think like it's better to share one big number 92% that is what we paid on our network uh of the general emission for validation for actually security that's a big number that's a good reason to get out

of uh an appgen on its own before that a bit about about myself. My name is Ail. I'm the CTO of Lava Network. Um almost 10 years in the space, mostly involved with engineering um co positions. Um mostly curious about um um infra staking um u me uh RPC account abraction, D5 protocols and so on. whatever involved with like engineering before that web two like as everybody else and overall 20 years of tech experience including Microsoft and Facebook and one step before I'm getting into uh

the topic itself I just want to make sure we are all on the same page because I'm going to share my experience about Lava but it's definitely not only us it's not just a a specific use case it's a trend we see that on the last two years app chains getting out of their uh ecosystem, getting out of their um um chain and picking other alternatives or just killing the app. So first in order to get the position that you're migrating you first need to get or to select an app chain that mostly happened on some specific time and um and and maybe it's worth like

sharing what was it back then or what was the atmosphere when projects protocols and so on got into that solution. So it's around 2016 maybe 17 18 the it was a lot of back then and the atmosphere was super positive everybody was [snorts] um consider themselves as going to blow up and have a huge traffic and a very successful product but that means that yes they are deserve an app chain um by the way it's not only that it's not just like a feeling of that I'm going to take over the world and be like super successful They actually had some very good reasons back then that for us we may consider them as uh very uh obvious

block uh block space was an issue cost was like a major thing speed was like was very slow and and and on L1 you weren't able to get much and of course anything that relates to strong community tools and even L2s were not there like not much alternatives. It's whether go with an L1 or build something on your own. [snorts] But then of course so the the the selection back then for to go for an app chain was very valid makes a lot of sense. Many did that. But then reality come.

We saw very fa very fast that um security is become an issue. uh protocols or app chains actually didn't get to the size they want to be. And that makes uh um an apps that were like small size midsize chains in a way that they don't have enough stake not much validators and actually being in a position where the impact can be like uh um very dangerous and on some point the you can see even now app chains that very few validators can can damage or can make some impact or to rule the the

chain very easily. Here this is like the numbers of validators that get like third of the the overall staking. So on some cases it can be like a couple of them that can get involved with like um very massive amount and make some not necessarily make some trouble but can take over some uh decisions and elements that relates to that app chain. And of course most important when there is no profit there is no validators if uh so it gives you two options whether to sponsor that to pay emissions to validators in order to keep them on some cases it's not even enough or on the other scale you get them leave you

get them to leave and we see that on the last year like massive amount of validators were uh actually um set to jail and didn't come back on Cosmos ecosystem. Next, so of course security layer is something that you mostly take into consideration, but it's not only that. It's not only having um a layer of security to handle and the efforts that comes with that. By the way, like having a security layer on your own also mean maintaining that you have a community, you have validators, you need to take care of them, you need to get into engagement with them. And that's also an effort that as a team you need to have.

But it's it's not only that. There's only there's also cost on the de development side. Imagine having starting from the bottom having like even a very fancy protocol like unis swap or a that has couple of like few thousands of lines of code as as a a smart contract comparing to build your own chain on cosmos comparing to have like couple of modules involved in that and also it means to maintains to maintain the the client itself that even if you're not the one to build it you are the one to make sure that is on the right place you need to fix some in some cases and you need to be on top of it on any minute.

Besides that talking about developer development efforts, we also have other part that is the the ecosystem of your chain. You need wallets, you need to have explorers, you need to have dev tools and so on. And it means some of them you do need to build on your own. That is an effort of course but on some cases you also need to be dependent on others in Cosmos ecosystem by by the way now uh we see uh a rapidly uh grow of wallets leaving the ecosystem we only left with few and that's not a very good sign.

What else? Um so again talking about dependency and having the ability to deal with things on your own when it comes to maintaining a client that you not the one to build it that means you need to rely on others as well. If an if something happened and you have a community of validators that need to run this client, you also depends on them to adopt the new version when it comes to uh having a fix or or like a super emergency um um bug that you need to make. By the way, like on that case comparing to a a simple transaction submitting onchain, making it on your

own, make sure that is there and you actually own the entire process. no need anyone else to be involved and no need to get into others live uh deployment of your own version and as I start uh mentioning it's also an effort not even uh outside of emergency if you need to have your own deployment and every deployment involved with not just deploying uh uh a transactions on chain and deploy a new contract it's involved with the entire community making a decision with everyone to make sure that they adapted and make sure that you also uh backward compatible because if some of them it's

take time and it's a progress and sometimes can take a week or even more and on that case if they haven't got caught up you make yourself in a position when you have an extra risk risk or an extra um aspect of taking care of when you have when you're actually in between two versions of validators in some cases it can be even more than that you three or four versions to handle running in parallel. Some of them decided to not go to the next one and you need to take every aspect of it into consideration.

Definitely huge effort and obviously something that takes some some resources out of your team. So um next step so we realize that definitely it's an effort. We know by now the the market is more mature and know that like ideas are fine but efficiency is more important. If you have a product, if you have a protocol that you want to run uh for the long term, you want to have something that is efficient, something that you can handle properly and something that maintenance won't be a huge challenge. Of course, as much as you can, but definitely you want to get that decreased as as much as you can. So, where to go?

Two main options. There are couple of more but there are two main options. one is an L2 rollup and another one can be a smart contract based protocol and I focus on those two not just because this is my opinion because it's actually what the market or what the industry proof as the valid solution for who moves from app chains towards something else and and it's not two or options it's just like a scale when we talk about an app chain toward moving towards an L2 and then smart contract based protocol it means a matter of how much control you want to have how involved uh you want to be in the uh the lower level elements

and that's the makes the decision where to go with smart contract or an L2 with an upchain you actually control everything you have full control you can change any element you can make any dec any decision and change any feature or any element in the protocol that you feel that need to be something else on L2 on that uh on that case you move towards less control but more secure you now relay on the L1 the Ethereum uh um um lay security layer that gets you like way more you not need to you don't need to handle more any more validators but you do need to to deal with other aspects like sequencers uh RPC nodes and so on that's still some kind of an

effort but way less than deal with validators or you cut some some of the challenges that you have and but still preserve a couple of them. And the last part or the last step that is on the other side of the scale when you talk about um uh controlling or maintaining or managing your chain or your protocol, it's smart contracts where you don't need to deploy anything beside your contract. Uh the rest is probably the offchain aspect that you have anyhow in other places and the rest will be handled by someone else probably um like bigger teams uh more diverse um uh like with a way more resources

resources that you have as a protocol to handle situations bugs and so on. So you actually get a better service, a better product, a better uh layer of security uh with minimal effort and again of course the cost is um you do pay some cost on it. But still when an upchain is still an option or where where is the time when is the time when you feel like upchain can be a good a good fit and I think on those uh tails should we should also have like a question mark because it it's supposed to be theoretically an issue like speeds and cost and and so on but most of them in fact in reality are

not anymore like I don't see too many of examples I can say that they cannot find a solution right now that fit fit with their speeds and cost that is uh that they need to go into their own app chain. Okay. And on that case it's make it very limited. And also by the way the same goes with uh with features as we saw like on we saw earlier we started very minimal on features anything back then was considered as custom but now we see almost everything available uh including account obstruction all kind of an ERC and EIP to support uh DAPs and so on. So it minimize the ability to find something that you can say about it

that is worth having your own app chain. Let's uh now let's jump a bit into uh our case study into lava and see the the impact over there. What we have before how it's going to how it's going to change. By the way we are still running the the migration. We are still working on that. We made a decision as a community uh to to do it already but the process itself haven't been done yet. Maybe next year we can discuss like what are the impacts and what works well and what's not. But on lava's case you can see how the how we decrease dramatically very important elements some of them involved with a lot of resources being saved uh including validators like you

don't have validators anymore. We have almost 70 now. We deal with nothing. We cut the costs because we don't need to pay emissions anymore. No, no one to pay. Like we only left with the rewards that that goes to who serves RPC calls. And it goes as well also with versions when you had when we had like 10 uh versions a year more or less. Now we uh that each one of them involved with huge effort like takes long and also involved with like um the the team it's completely focused on that now deploying a smart contract become way easier solution and some of the cases even even saved because uh they were related to uh

the client uh fixes some some cosmos related uh upgrades that made not by us and so on. So we actually only deal with our own deployment. Um and yes we also decrease complexity having like not having a staking enroll. We have two staking roles. So now we only have one the one that's actually providing services. So that's also like an extra complexity that we put out that makes it more simple and generally the feeling is that you get the focus uh uh solution aiming for your target business and don't need to handle many aspect that definitely unrelated to what you're doing

and still generally upchains and cosmos specifically has some aspects that we do uh see valuable um that you don't at on L2s, L1's protocols and so on. For example, time based triggers that don't exist or or some elements that relates to IBC. But to be honest, that's if you're not getting it by the uh by the chain itself, it doesn't mean you're not able to provide it or to have some kind of solution to deal with that. So we we do need to have uh some to put some extra effort in order to have it but still it's something that relatively to what we have saved considered as

minimal. Let's talk about takeaways or what we see as um as most way to to measure whether you want to go into an app chain or not. So I think like the most important part that is not mentioned over here is how custom you want to have your uh your DAP how deep you want to get into the process itself and do you even need it because if you don't there's no reason to go into an app chain and you better start with a um like as a default as a smart contract based protocol and if you see that you're not getting this value that you're expecting over there you should move towards L2 to and the final step

will be an app chain. We do see like I think like one of the important thing is like do you want to handle those things? Do you have value of customization um and other elements that relates to the process itself of of building the block and and and setting up what will be um uh the way or the the the the governments to do that. And I guess like um and of course uh this is kind of it. I think like uh the general idea is to for us on this process was to get into a position um that you build

something okay and you can run fast because everybody knows that you want to get like more into execution. You want to get things done as soon as possible. You have a road map. You have a lot of res you have a lot of things to handle and on that case resources are limited. The idea behind making such a migration is not just saving cost or making it more pro profitable or or or attractive is about to allow the team to focus on the right things on on what we uh expert in and where we want to go. So the impact the overall impact will be shorten the processes getting the road map earlier and make things work the way

we want it as soon as possible. Thank you. >> Okay. Okay. More more more more applause. More applause. What a great talk. What a great talk. The the thing I love about uh Cosmos communities is that you guys you have this I the most pronounced democracy. I would say you can make such a decision like let's just move our chain. Well, that's that's cool. Let's go straight to questions. So, the first question that we Oh, Jesus Christ. I pressed the wrong button.

We'll skip that for now. Anyway, here's the question. So, let's go with this one. How do you convince validators and the community to vote to dissolve their own chain? Yeah, it's actually like huge question because you have validators that actually that's what they do like to earn a living that that's what they're like making the profit out of. So you're actually saying to them on some point or some aspect that you you're fired. You you fired the validators, right? No other way to say that. But the approach should be like way different because validators still are token holders. They still care about the product. They still care about what you're doing. And on Lavas case, it's

also like a dual staking. So staking means validation, but there's like staking means also serving an RPC calls on the system. So they hold two titles. Now they're going to hold only a single title that hopefully it should be like rewarding enough but the effort should be like minimal for them. >> Okay. Yeah, absolutely. Makes sense. Uh I figured out how to pull from the downstairs the question that was there. So the next one is what was the main thing the main thing you lost dissolving your own chain? Was it all roses or do you maybe want something back?

>> So, it's it's still that like a sense of control when you own the thing. It's definitely uh like feels better. Of course, you know, you know what you what's going on, you handle uh stuff and you can actually uh be the one to point any aspect. But I think the effort of getting this uh sense of control is something that like if you mature enough and if you're experienced enough you know that is actually uh this is something you need to give away. So it's very hard to let go on on control and definitely uh tons of code that uh that people wrote and now it's like going uh into a going aside but still that's the

right decision to make. >> Okay. Yeah. Well, having control is is probably a good thing from a sense but also a big responsibility. Let's move on to the next question. And the next one is back then having an application was also related to a protocol/dow mid-regulatory regime in US and uh also oh my god due to changes down during I think we will cut it to back then it was about protocol and the dowo but what if the law change again >> well um so like on on that aspect we actually didn't change didn't change anything.

Governments keeps going the same but on a different platform. So it's a technical aspect that was changed but we keep most of the rules there and the approach remains the same. So so from the legal aspect there's not much drama over there and actually uh it keeps like going as as it was before. >> Okay, agree. Uh moving up next, this is a technical question and I personally love technical questions. How do you keep the network servant traffic while you migrate underneath it? >> Yeah. So this week we I had actually three talks related to this topic. One is about it's this like the why. We also had a a discussion about uh the how that

is actually uh super important. And the last one was about the security aspect. That's also a major issue that relates to that about like the migration. This is like a whole event besides planning the the the the final state the migration itself become a product on its own. So you need to make sure that you know how to run two things in parallel. There is no uh issues with that. You need to make sure that you have a collaboration with all the validators and all the community and on somehow there still it's you're still going to have some gap between those uh uh chains connecting two unrel um synced uh uh systems that you going to that you need to find a way later to uh to make up and

make sure that who was like who was impacted on that uh time window get what he deserve. So we just need to do that like very carefully and make sure that you everything is well well designed and planned and go step by step by the plan and and hopefully and hope for the best. >> Okay, the final technical question and uh we are wrapping it up on Cosmos. A lot of the data is delivered by validators not oracles. Did you find it difficult to deliver offchain data after you transition your chain to layer 2?

[snorts] So we first we are not migrating to L2 we're migrating to a smart contract uh uh protocol. So that's makes the the the questions a bit different um about oracles and validators and so on. So different it's actually different solutions uh about difficulties. It's it gets you some tri some challenges but on some aspect you get some benefits because validators dealing with data it's still an issue hopefully to get that um uh out of their control and find some valid solution can be even like again saving some responsibilities getting someone that is expert in this field to do that and that makes it like a super positive um change.

>> Great. Okay, let's give another huge round of applause to Afir. Thank you so much for the talk and we will be gradually transitioning to the next talk which starts in 5 minutes. So stay there or go grab a coffee very fast. Do not forget that the QR code on the screen is the place where you ask questions about the talk that is going on right now. Uh so do not lose your chance as you listen to the talk and you have a question, go ask it right now. Go vote for the other.