Decentralised or Distributed for web3 infrastructure? Peter Stewart, Asif Khan and Luis Schliesske

Transcript

hi everyone um we're here to talk about web 3 infrastructure I chose a topic about decentralization versus distribution I'm from anchor head of infrastructure hey guys I'm Luis I'm the co-founder of gelato Network uh hello everyone this is Asif from polygon I take care of the D5 stuff at polygon I'm really excited to be here today yeah so um obviously um within for anchor we do a lot of work to um distribute our Network there's a lot of discussion about what does

decentralization mean in infrastructure obviously to a certain extent we are centralized as an RPC provider we have a website we have a you know we use cloudflare and so on but we do work very hard to distribute our nodes around the world so we're in 37 Metro areas 37 different data centers including web third-party providers as well um and we are always striving to understand what we can do more to be more decentralized be more distributed and so I'm interested to hear what other people's ideas are about [Music]

their efforts to or how important decentralization is to your protocols to your blockchains what efforts you're doing to do that and so on yeah so at gelato we also in the business of building the decentralized backbone back end for web 3 applications whether they be on polygon BNB chain or all of all of these places and we also in an eternal struggle to try to decentralize as much as possible but also being Mindful and sort of realistic about the fact that there's a bunch of very good technology out there that doesn't necessarily lend itself to

decentralization like HTTP servers or whatever but they still have their place in this world right like a website nothing wrong about that not everything has to be decentralized although of course I think one of the points that I see today is still sort of lacking is like ipfs and so on is really cool but it's it's it's not really that production ready I would say it's this people have problems with pinning services and so on all the time but but yeah so what we're focusing on is having this decentralized back-end component so we're working together with with anchor with RPC providers we're working together with networks like like polygon to basically provide a decentralized backend for decentralized applications because

um developers always think okay the smart contract is my back end you can like write a bit of solidity and then just put a front-end a dab on top of it and it works but in reality smart contracts by themselves are not powerful enough to really you know support a lot of like high usage applications there are some differences maybe like uni swap is a very powerful application but that's that's about it everything else really needs a very strong back-end components for things like automation data oracles transaction relaying at scale all of these things account abstraction and and what we're basically trying is to give developers these Technologies like these server Technologies but in a decentralized way and give them access to this decentralized network of data

and transaction execution um with as little centralization as possible and for example what we can do there is you can use Smart contracts to regulate data and validate data on chain and if that is not possible then you can still validate the the data in a network of nodes off chain they have to come to consensus on it yeah but that's basically what we're working on and one of the things that are also still centralized in gelato centralized but distributed is for example our the API Gateway for our relay API right like um there's no really good way I think to decentralize an endpoint like that I mean I'm sure you are researching the centralized rpcs which is super interesting but we

haven't really seen them yet and yeah we also sort of have the same question because yeah giving developers a P2P client simply is not as developer friendly as giving them an HTTP API that that's basically it and and yeah so you know um that's where the centralized chalk points still are but maybe they will remain um maybe yeah would be interesting to hear your points on that uh which I think the way we think about decentralization and services is that you can you can probably verticalize it across uh so on the one end of the spectrum you've got things like the base layer that is polygon blockchain beta gkm which will come live

on 27th uh or our current POS Network and then on top of that there are a lot of infra layers be it uh oracles beat RPC providers and be it user-facing applications beat dabs or even like these front-end applications you have right now like things like zapper debunk so I think as you move down the Spectrum ux matters a lot so maybe things like debunk or zapper they can be very well centralized because they are catering to the end user so I think the ux matters a lot uh maybe then maybe than how their backend is sent

decentralized or distributed because you need like fast access to data there but I think on the other side of the spectrum if you talk about polygon the base layer you've got a decentralized set of validators where you cannot afford level of centralization because economic incentives are involved there and if people will have incentives to be a bad actor they will that's what we have seen in last two years last one year especially so yeah I think it's more about this spectrum and figuring out where you can sort of tolerate some centralization and where you just cannot yeah absolutely no I think finding that balance between decentralization centralization speed

accuracy all of these different trade-offs that you have to make when you're designing these Services uh are critical um I don't think there's a particularly a right answer about this in some of these cases you can you can play it off I mean I I had a lot of conversations yesterday with people who were saying look at the end of the day how many people really care about decentralization on RPC right they just want a fast service a fast accurate service they want one place to go to and they want to be able to get their data they want to send their transactions at the same time we're not sitting down and saying that's the end of the story we've done a good good enough job we work very hard to try and get third-party providers in there are some privacy concerns for us because we're very at

anchor we're very focused on and we've just Rewritten our privacy policy to make it really clear we don't mine your data we don't front run you we don't we don't link your address with your IP address and all of this kind of stuff once we start bringing in third-party providers for our rpcs we run into this issue that we don't control that anymore so there are some trade-offs that we have to make there and some understandings about these things yeah I think um maybe the sort of conceptual mistake that a lot of web3 uh crypto ideologists make uh you know I also started out being like everything has to be decentralized and so on and that's true I think over the

long term as much should be a decentralized as possible but uh it shouldn't stop you from Shipping something quickly most important thing is that we get more users in at a quick rate and um you can't you know ask users to use a website that is down all the time or slow or something right they will not come to a three if that's the case so so I think there's totally fine to to make some shortcuts as long as you're not you know violating the core tenets of crypto which is you know don't custody the user assets um don't don't have one guy or one company be able to rock everybody and so on so so as long as this is still true which you can regulate in most of the

cases via Smart contract you know if if you're frontenders still centralized if your API is still centralized or something that's fine you're a business uh you you have every sense of for it to be available at all times to users and but then again right we want to be an open global economy and there are a bunch of people I don't know maybe in Iran or so that are good guys and girls and they want to build a reliable application and they don't want to be olfac censored or something right so it is important I think in the long run but you shouldn't try to shoehorn decentralizations in something that isn't ready yet so I think we should all on our roadmaps have decentralization plans for everything we should invest in

decentralization where it's not here yet but we should still ship with whatever works today and then progress there with time yep for sure totally agreed I think what matters here I think totally agreed that early on maybe you can you can tolerate some trade-offs because you want to offer services to people you want more people to come in you want more users and maybe in short term if that happens maybe at the expense or decentralization I think that's a valid take but I think you need to have a very very clear road map as to how you're going to go about things in future so let's say if you if you if you think

about polygon uh when we started our Network the POS chain which has its own set of validators so it does not borrow exactly from ethereum layer 1 security but I think back then we had a very clear path that do we want to ship a product up front we want to get users in we want to offer dapps but at the same time we invested a lot on the ZK capabilities and that is why we're gonna now ship the GK UVM minute beta on 27th and it's been a lot of hard work behind the scenes and it's been a lot to sticking to this idea that we want secure decentralized uh infra for

the Developers I think even if you talk about GK VM right now there would be a centralized sequencer because we just released mainnet but I think there's a very very clear path as to how we want to then decentralize the sequencer as well so there's something called proof of efficiency which will use Matic token and which will then enable uh independent permissionless sequencer Runners to come into the network to participate uh to become sequencer validators in a way you can say sure I think for sure you can tolerate like

centralization in some manner for some short amount of time but you need to have a very very clear path as to how things will look in future yep no I think that's a big challenge for ZK I mean essentially I actually started in the crypto space back in 2018 with Horizon which is a clone of zcash and um yeah that was more decentralized but not on the validator side this was all gpus and and so on like that but um as we as I've worked with anchor it's it's more it's definitely about that it's a challenge that all of the sequencing all of it it's all off chain it's all happening somewhere else on somebody else's computer so to an extent the Privacy is compromised uh

potentially compromised um and so definitely decentralizing that is is a very important feature I think for moving forward with that stuff yeah and I think the the panel title talked about distribution versus decentralization so uh maybe a word on that so I think distribution is really a question of modern software deployment architecture and really you want at a certain scale um if you have a ton of usage um decent Lots you want your system to be distributed because then only then can you Ensure High availability across the globe uptime all these things

um and and yeah so so that's I think really important and many companies web 2 companies they're all distributed server companies and in some manner they're also I think decentralized sort of internally and they use decentralized architecture sometimes P2P I'm sure Amazon AWS uses P2P and so on right so so you you you you can actually if you zoom in you can be a centralized company and still leverage distributed and decentralized Technologies to great effect but in web 3 the decentralization question so distribution is not a question like it's it's yet do it or die pretty much at some scale um and what three decentralization means not just one company having the power

right not one company can go out of business not one company can shut down your application or servers it's really about that and and there I think we can you know all get inspired by the ethereum protocol by the Bitcoin protocol and at the end of the day what we're trying to build here is not like companies that control a network like fake networks right [Music] but what we actually want to build is self-sufficient digital economies and that are decentralized so we have these amazing digital assets that are digitally transferable and everything so now what we need to do is we need to match for example in gelato the demand from application Developers for server

power with the server power Distributors and and so on and all of this should be matched in a self-sufficient on-chain protocol where the exchange or payment and so on is all encoded so that eventually as gelato we can also step away we can go out of business or whatever it is or turn crazy and that network will still be self-sufficient it will still have an economy that supports a certain number amount of node operators so that any one of these can go out of business and you as an application developer can still sleep well at night knowing that your app will be served uh yeah so I think that's like a the big difference here as compared to to decentralization in web 2 where it's really yes it's decentralized technology but one company makes the decisions here it's about there's a protocol that

regulates this Marketplace and it's always there you can join it permissionlessly you can exit it and it still works the economics are right so so yeah one company or one country I think is also the the other question um certainly for for anchor we as I said we have seven eight different providers we don't use cloud at all in our RPC business we only use bare metal we are distributed in that sense we have lots of data centers all over the place one of the reasons obviously to move away from using AWS as we had done previously was about price but the other one was about the ability to not be censored in that way so if AWS gets the Mandate or they're being told no more

crypto what happens then right so um I don't know if that's a feasible situation a real situation that will ever happen but who knows I mean regulations have been coming down on various crypto companies and anything could change from day to day I mean we're still a US company we still have to abide by U.S law but um at least from the infrastructure perspective we are distributed yeah uh I think you know we have been talking speaking a lot about the infra side of things which and specifically like our PC and blockchain layer but I think would love to talk more on the Oracle side as well

uh because if you if you talk about d5b lending markets pop markets their oracles are a big critical piece of the whole D5 stack and I think the last one year saw a lot of perp volume happening on chain so be it GMX which has done like almost 100 Mil 100 Mil gains has done like 25 billion polygons in Futures has done like almost 20 bill on polygon but if you think about oracles and their centralization uh in a way if you talk about order flow auctions the whole story that started around 1993 when in these brokerage firms started selling

order flow to these market makers I think if you think about Oracles in a way oracles are the way to realize order flow auctions and in a way a sort of an MAV can happen on an oracle layer so let's say let's say you have an oracle provider there's there's some bunch of data providers and they also happen to be market makers who are let's say trading on cexs now if they are providing prices to the Oracle Network in a way they control the price that is existing on chain and in a way they can then front run the Perpetual transactions that are

happening on chain on Oracle level so I think chain Link's been doing really well uh but there are some trade-offs there as well because on some of these price feeds the heartbeat or the deviation may be too conservative so for some of them it may be like 0.5 percent heartbeat uh 0.5 deviation something like that and some of these newer oracles uh take data directly from these providers who may be market makers so I think like as a whole D5 Community uh Builders we've got to think about Oracle decentralization and this whole Oracle extractable value in a way

because I think this will be like a key thing because now perps have significant volume on chain which I think was not happening like two years ago or even one year ago so would love to know your thoughts on the Oracle side of things as well yeah I totally agree with that I think one of the unfortunate situations that D5 finds itself in that is always a bit um hushed over like it's not really addressed that much it's becoming more now is this you know vast Monopoly that chain link has I think they got there deservedly right they they have amazing technology they push really hard for this uh but they have they bear some

resemblance to a sort of 90s Microsoft you know super aggressive sort of Monopoly company and I think that's very bad for D5 right because at the moment if you look under you know what what diva is under the rock knob pun intended um then then you'll see that a lot of it is is really just all uh reliant on chaining and and I think we need to see a lot more Innovation there are a lot more competition uh and uh and hopefully some new Oracle designs that are also much faster that's needed for a lot of applications I talked to the the guys from Redstone recently it's a I think a new project um there are some cool designs like oracleless design it's also an idea

right like lending markets built on Universe 3 like time swap but they usually I think have some restrictions in the sense that they have to be sort of fixed time and that doesn't really translate well into our world um but but these also access to oracle-less Solutions are amazing I mean um uni swap for example okay they have their own top oracles but but that's that's a a very cool on-chain thing that that you know is really decentralized um but yeah I totally agree I think we need to see a lot more Innovation there on this front and not just from one company and and I'm we will we will see that in the coming years and D5 will be better afterwards okay so I think we're heading up to our

last four minutes um do you guys want to talk about any future plans for any of your products that you're working on to towards decentralization distribution Etc or anything like that yeah so um at gelato we right now have uh three services in our Network that we offer um one of them is automate that's the original one that we sort of invented we were like the first smart contract automation protocol um and there it's already pretty decentralized because it relies on on-chain logic so as an application you have you know full control over the data flow in your application on chain and actually as of recently we finally removed a centralized technology rapid

mq and swapped it for lib P2P at decentralized technology and and we're able to onboard the first node operators to our Network so it wasn't just us running the gelato nodes anymore it has been now you know staking facilities for example and we're talking to many others so at the moment gelato node operators are in a private beta if you are a node operator please contact us and our long-term Vision there is to by 2024 2025 be a completely open permissionless market so that anyone can spin up until not similar to what you can do in ethereum just have to stake some gel and off you go and and for relaying it's similar it's also using the same network so as we add more node operators it will become more decentralized um but there you always have

decentralized choke points like an HTTP Gateway like as you know from the RPC business um and and last year end of last year we shipped our web free functions and that's our own sort of first foray into I guess what you could call oracles where we're now allowing off-chain data apis and so on to also be used again adding node operators adding staking it's coming and just working at it being patient but never losing sight of the target yeah for sure I think for us like the biggest Focus area right now is definitely rgk UVM Network which will be coming live on 27th for transactions everything and just making sure that

this networks runs smoothly is the best execution layer for D5 nfts gaming uh yep yeah I think that's I think it's interesting to think about um protection potentially reducing the need for some of these decentralization efforts by protecting the transactions as they go through and things like this so that it's not so that the the potential for censorship and tracking and so on like that is less yeah uh certainly with it within uh anchor we're working we did actually at one point release a third-party node providing service uh we're working on this heart as a project it got Sidelines but we are starting that back up again and that's certainly one of the focuses for us is to get third parties individuals we will

focus initially on uh you know a vetted group of professional people or professional companies that are doing RPC um and do that and also we're working on version three of our Network which is fundamentally without releasing any more details uh much more decentralized than the current system that we have with cloudflare and so on like that cool well I think we're at time thanks everybody and uh have a good time thank you thank you everyone take care