# A eulogy to Frankensteins tech stack - James Corbett | Superchain

- Channel: [ETH Belgrade Community](https://streameth.org/eth-belgrade-community)
- Date: 2024-10-07
- Duration: 29:02
- Topics: People & Blogs
- Watch: https://streameth.org/watch/yt-cHYOMPSfkFs
- YouTube: https://www.youtube.com/watch?v=cHYOMPSfkFs

## Description

A eulogy to Frankensteins tech stack - James Corbett | Superchain

## Transcript

[Applause] easy easy well thank you very much James for the great introduction um he's recently invited me to the James Guild which I didn't know was a thing so now we can just take over the world it's great jameses will be everywhere and actually my team is uh Joe Jamie James and Jay as well as a few others but they don't count because they don't begin with Jay so um we'll just start from there so what I really want to speak about today other than some hilarious naming is actually some kind of hilarious expectations we have building in crypto which is that we're going to use Frankenstein's technology stack right we're going to bring all these things together to try and make a crypto dap now I really hope that today is going to resonate with a lot of people who are builders in this space um and you know I really want to talk about some things that maybe we can do and I'd love some U more kind of Q&A at the end uh with any ideas of what we can do to kind of alleviate these kind of key issues so everyone starts building a appap with something really really simple they start with ethereum blockchain RPC and then maybe some simple user experience some JavaScript um that's how it should work right so you're this is this is this is what you want to do you start here and then something happens maybe you need some historical data you need to see something over time or maybe you need a data source from somewhere else maybe you're building a lending market so now you need to bring in an oracle so now you need to go and negotiate with an oracle provider bring those in luckily nowadays there are a lot of Oracle providers this is a really good thing um but this is a real complexity in terms of building something thing is that now you have to integrate yet another protocol just to know the price of something so your little D app that started off with okay I can run my ethereum node at home I can host my UI now I need an oracle provider as well so now I need to go and negotiate with say chain link or Redstone Etc and get them to come onto the chain that I operate on and also pay them Etc which is an off Trin transaction and I need to subscribe and all these things so you end up needing some history you end up needing some context to give your user and to give the user experience so you end up using S like a subgraph um either hosted or on the decentralized network and this is kind of the again something that should be really easy but is not right so RPC is really good at grabbing like one piece of information at one time it's really really bad at grabbing every single block um across time of all the data and actually you'll find that most wallets which have the little RPC provider inside them uh they usually are a free RPC so if you make too many calls your app would just stop right and you saw this actually in defi summer where people were using um very simple tools I don't know if you remember like vat tools to view their positions in all these different chains but it was making so many requests on RPC it would just stop halfway through and you're like this is a horrible user experience and it's way outside of expectations so things like subgraphs do alleviate those issues now something that's becoming increasingly more important in terms of your protocol you know your protocol is growing now right you've got some tvl in your lending Market in this example it's growing how are you going to show to your stakeholders be those token stakeholders or equity and investors in a normal business or even to your users that the platform they're using actually is useful to other people not just them you need to show like the tvl right so things like defi Lama are really good for this but you're probably going to want your own dashboard so you're going to head on over to Dune to uh Flipside crypto Etc you're going to want to then build your dashboard in SQL and this is yet another integration right and for example if uh Dune doesn't support your chain or flips side doesn't support your chain you then need to actually pay them and integrate the data yourself so on a eum this is okay they support ethereum um but building out this dashboard and infrastructure is another thing we need to do as a Dap developer um which is kind of outside of being a developer right you're basically negotiating with suppliers you're now building a business that has three suppliers on top of your decentralized network now what happens in general and the experience that our customers often have is that you know a subgraph is great for like summary information but kind of starts to fall over after a little while especially as the protocol gets older bigger Etc um we've seen that and like pancake swap on binance for example it's basically impossible to get the data for that um so you end up building a custom indexer right but you just wanted to build a lending Market you didn't want to build a custom indexer right this is a very different skill set and if you are say on a chain like polygon you're going to need to deal with block reorganizations which could be like 200 blocks right so reality is going to shift underneath you now I don't know how many people here have tried to build an indexer before they start simple and then they end up in a really horrible Place uh so there's some good hands up there so yeah great well this is good I'm speaking to the right people um and really this was my history in crypto of building custodians trading platforms um Merchant scale wallets the problem wasn't necessarily just in the execution it was in the data that we couldn't get the data right especially for things things like Bitcoin which have very poor support for for data now you started on one chain that was bad enough but now you're on three right so if you're Sushi swap you're on like 16 right how much infrastructure is this right you're now having to hold in your hand manage and things and and you know if you want to become a dow one day now you need to synchronize this all through a dow right some of these things don't take tokens yet although I must say in RPC very excited to see that some of the more decentralized solutions are starting to come to defition like pocket Network and there are others as well um but this kind of complexity is hard these chains have different block numbers maybe their time stamps don't line up right they've got different data different Oracle Solutions maybe you need to implement multiple because optimism has a different Oracle than ethereum so you know now you're increasing your smart contract complexity I think the point I'm trying to make here is that your complexity is always increasing right it's not decreasing and the scale of this where you need so many users to earn money for your protocol but you also have to pay kind of huge fees on the other side for something that started as well I can just deploy a smart contract for a few hundred dollars and build my dap has turned into I need a business I need a COO I need like Finance person you know there's a whole whole host of things you need to build around you a whole lot of infrastructure be it people infrastructure or or physical infrastructure um in order to support your application and to make it usable to provide good user experience right because if we want to get crypto used more we need to get that ux kind of stuff down and right now daps take 20 seconds the load which is which is just mental so a real life derivative exchange you know kind of looks like this where on the left you can kind of see RPC custom index a dashboard that's all connecting to this chain with your smart contract and then kind of the Oracle that's like pushing into the chain on the other side and so you have to have these like manage this relationship between all these different things now a real common problem is this happens a lot in derivatives markets is the Oracle is used as What's called the liquidation price which is not the price you see on your screen the price you see on your screen could be from somewhere else and often is from somewhere else so there's been many examples where one price was here the Oracle price was here you put in your trade and you immediately liquidated especially if you're a High um High multiplier like 200 times leverage or whatever which you know dgs do do these things fairly regularly so that mismatch in data that poor expectation setting is a really serious problem right imagine if you were making like if you were a large financial institution making a trade and your trade settled at a different and unexpected rate which was not Rel relative to the market like you are going to be not happy right and you're probably going to start suing people in crypto it's just like oh it was a technical issue or yeah sometimes they get a bit out of sync it's like okay probably not the future of Finance we're building if that's the case so there's a serious disadvantages to this kind of setup right so we got increased time to market right we started really simple and now we've had to recruit more and more people more more and more complexity more and more infrastructure right we've had to make multiple negotiations with lots of different partners we've had to talk to humans on this decentralized network so now there's a human factor as well in this right you need to be on telegram all the time to talk to people which is annoying and after all of this we got bad user experience right because we've got inconsistent data from different sources this is a real serious problem and you know I think it's one that we're hoping to address in some ways but I think we need to look at how we look how we lay kind of Frankenstein monsters text act to rest so I think there's some key principles that we as infrastructure providers can take to basically help projects instead of kind of getting in their way or gatekeeping access to data or you know having to make you sign giant deals in order to get anything Etc so one of those ways is to make it permissionless so permissionless is a really key concept which ethereum itself Bitcoin itself Etc keeps if you have the token you can use the network we can't stop you right this is a key principle most of the infrastructure does not adhere to this principle um and in fact some infrastructure projects their token doesn't have any relation to the actual infrastructure they provide and actually that infrastructure is paid for in dollars in contracts off chain so you know you could question maybe what the purp of those tokens actually are but having a token gate only for your network to have like an Unstoppable Network that's per se providing RPC um which you can pay for in token this is a way that we can alleviate some of these problems we can economically align and we can use some kind of economic principles to build our daps instead of kind of corporate standard business development type ways the one is verifiability so the provence of data is something that's really lacking so when you actually use adap often if they're using a custom indexer or even if they're pulling from RPC your computer is not verifying that data right unless you're using a like client which you know is an option people are trying to run Ram like clients into the web browser which is super cool I fully support things like Helios initiatives and things but 99% of users they're using Etc the data they see on their screen is from a trusted party you're not operating trustless right so again a key principle of crypto verifiability of the data you're seeing the transactions you're seeing Etc now I'll add a really good example which you know I do as well which is really terrible is that we use ether scan to check ether scan is a source of truth that is not Unstoppable that is not decentralized it's a great public service and they've done amazingly well but I don't want to rely on them for data I don't want to rely on them for all of crypto's data right not to mention they're kind of expensive for chains some of those deals are eye watering so I mentioned already but you really can't rely on infrastructure that can be taken down so when you're selecting your key infrastructure providers your RPC providers Etc you need to look at where their risk are right do they have a centralized low balancer that you need to access all this data from right even if they have a token right super chain today you need to trust super chain Labs you need to trust me one day you need to not trust me and I won't even be able to take down super chain right this is the aim it's very difficult to build things this way right but instead it's going to be really difficult for dap developers to build anything decentralized so you kind of have to pick your your battles Unstoppable systems Unstoppable data this is a key tenant of how we need to build in the future we can't just kind of keep pushing things to be centralized and products like Unis swap they recently released a great new user experience which involve them removing any semblance of decentralization they removed the graph they removed all dependencies they they now have their own centralized API that isn't even open source so although Unis swop the contracts sure it's open source it's on chain V3 V2 is it easy to use but the reality of actually trading on Unis swap which often involves hopping between multiple pools evaluating different chains Etc that's something that is going to be difficult without some kind of API whether that be 1 in unisoft own API or some other third party um there are some interesting takes on this as well like intents and things which would definitely think a decentralized intent system could be very interesting to solve some of these issues see I've got a fan there for decentralized intents not centralized intense I do want to differentiate between those two things soon TM okay um yeah so un software systems are really key um we need to kind of stop this recentralization of things in the name of like oh user experience and things when user experiences isn't even getting that much better so what we do very briefly is we're building this verified and transparent data to try and address this but we don't want to just be on our own we want to have everyone else help and build more decentralized applications we don't want to kind of be the only ones trying to build data infrastructure or infrastructure for supporting daps you know we need way more stuff than this we need user experience UI Frameworks ways to submit on RPC and decentralized ways no there's a ton of ton of stuff that needs to be done that's not just the data at Super chain that can be used everywhere else so you know we feel that we've able to squeeze a lot of the uh complexity into one place that's verifiable um but you know there's a lot of other complimentary services that need to be around this for this all to work really really well um so yeah follow me and message me and we'll fix crypto together [Applause] so I think do we have some time for questions cool so any questions put your hands up please tell me why I'm wrong I think there's a gentleman up there oh yeah I can see you now so you mentioned the idea of Unstoppable data it would appear to me that a lot of the data now with subgraph and indexing happens offchain because it's very difficult to process all of this onchain uh what approach do you think could allow this to become Unstoppable would there still be offchain elements or what do you see sure I think that's a really good question um you know there's a few approaches so you can use zero knowledge to verify state from other chains for example so that's quite um like a a thin slice so you can probably read like one or two properties like that um but you could have a smart contract I think there's a uh Herodotus I think is one of these Solutions um so zero knowledge I think can help especially where you are trying to do things within smart contracts which you know need a small amount of data small amount of processing power obviously there's an offchain component to that in terms of building the proof um unless that's natively done within a chain with something like Risk zero but there's a lot of like Tech buzzword that would have to come together to kind of build a very comprehensive solution um we've chosen to do something a bit differently where it's not on ethereum chain we're adding a data infrastructure layer so another layer unfortunately but at least we're replacing some of them um but we have standard consensus so we have a deterministic data platform um which we built inh house from scratch um which lets you verify that without zero knowledge and things so it's like way cheaper to run on commodity hard Hardware that's our approach but I don't think it's the only one I think zero knowledge has got a lot of legs especially where you don't want to know the underlying data you just need the proof um things like fuel Network are particularly interesting where they're like building utxo sets and stuff but they don't know the inputs they just know the hashes um so there definitely some legs on on the Zero knowledge side as well any other questions ah James with the best name yo thank you for a great session there James just a quick one there you mentioned with Unis swap kind of going to their sort of uh closed code um you know pivot pivot shall we say um how what do you think the risks of the space of more players in the space doing that um and how can we kind of counteract that happening obviously super chain is at on this kind of crusade to do it but um I just worry that with big players in the game doing this like how like what does that mean for the space really moving forward yeah so I think that the uh economic incentive is huge because if you have a fee switch in your UI as Unis swap does um you're going to want to be moving as many people to your user experience as possible so I think that if I were Unis swap I'd be doing exactly what they're doing based on their shareholders and who they they ultimately are responsible to um I think that the there's two sides of this one is the fact there is competition in terms of like uh defi by llama has a swap aggregator so an aggregator of aggregators so you can like compare like cow Swap and um Unis swaps own routing and stuff which is which is really good so that helps reduce the fees but I think the repercussions are basically there will always be a fee and that fee will be erased to the bottom to some extent but because there's still a moat in terms of the data um you know like 1 in for example the advantage they have is the biggest data warehouse of everything including all their historical transactions and intentions and the actual resulting trades right so that's actually a useful data set that they have providing access to that data will help commoditize routing and other more complex uh operations for the user to do um so yeah I think it's like not the worst thing in the world what's happening with Unis swap um but I think that it's definitely the wrong thing because obviously that's such a big brand and things that you know we're now pushing more people towards it a centralized experience um with a fairly large Mo especially without it being open source which to be honest I was pretty surprised that maybe they'll actually open source it at some point but um right now it's that old one that's that's there which is dependent on the graph which is not really maintained anymore yeah because so V2s depend on the graph still and then V3 is is just moted like that is that right uh no so Uno's new API support all of the previous versions V2 and V3 um but the subgraphs I don't think that they're maintaining anymore so their old like open source API which uses subgraphs I mean it will be it'll be a bit quite a bit slower and a bit more difficult to use we' we've tried it internally it's it's not too much fun to run um all right thanks James okay so um my question would be how much of this because a large majority of what has to be accomplished is sort of data indexing and data processing so how much of this can be reliably pushed to the edge to the client that's a good question I think like with browsers now it's starting to get more fun so I'm actually really really excited about modern Technologies like web GPU um which is a way of doing compute on gpus in web browsers now you may think why on Earth you telling me about that that's completely irrelevant to crypto well telling you because let's take a very simple example like um the total volume traded on a swap if you have that data in the browser which now browsers can store things on disk you can use gpus to very quickly do maths so you're able to calculate like add up all the volume for a thing pretty much instantly even on a mobile phone a mobile phone being a much more restrictive environment so for me I think these aggregations can start happening on device which is really really really exciting because it's very easy to do verifiability at a data layer it's very hard to do it at an aggregation and computation layer that's what zero knowledge does right so that's like the really hard stuff that we deliberately avoided because we'd be a zero knowledge company and wouldn't do much else right because it's a huge effort so we deliberately went a different way and then yeah we're going to be very heavy on the client side actually um so yeah I think that's one of the ways that we can help but we have to do that efficiently we have to do that in the right way so things like web GPU or mobile especially is going to be a necessity to uh do these kind of calculations quickly and provide that Snappy user experience that that is expected okay thanks I will try not to relate to the previous uh panel that you were on today uh but you said that you want to make data Unstoppable and that one day even you will not be able to stop it now we're seeing on one side as uh your namesake James mentioned uh Unis swap is doing on one side right they're centralizing um how are you approaching this because you are not only uh providing the tools as far as I understand but you're also leading by examples so how does that work and how you're going to make super chain Unstoppable and permission all these different things that you mentioned all right yeah well thank you I'll give you your $50 later um effectively we've built something that takes blockchain data from node makes it verifiable through we have our own database basically produce files if you run it I run it we produce the same data um that basically is a very easy thing to decentralize because you can just hash those files so we could just use any ethereum type Network have the hashes of the files that we made share those come to consensus on those okay five people said the unit swap V2 toolbox for this block range was this hash does everyone agree yes you li go away right very simple consensus mechanisms you know all the same kind of security concerns as you'd have with standard POS um and then yeah built into the into that Network so we're going to start slowly so we've basically proven that our product is actually useful to people um for example working with Telos who one of the other sponsors here like helping out some of their adaps helping out some more derivatives exchanges on fuel Network for example um getting kind of high quality data from both theorical side and the order book side so for us delivering user requirements is the number one priority and then able we're basically then able to decentralize this back end behind because we have this light client and that light client is what basically keeps the verifiability um in and we have some little tricks that just to share a little bit of alpha um is that we are releasing something even before we build our Network that will add verifiability to us our data so we can turn it off but we can't lie um and also we get punished for turning off too uh by the so there's going to be like effectively we call them Guardians who are able to help secure our Network before we have a network even um because we think that's really important to keep ourselves honest you dou it on James dou it on um yeah so I guess a little bit of a contrarian take but uh as you know I'm old and salty and I really like my Bitcoin my oh there it is my understanding is uh you guys have started doing some cool stuff with runes and I just I just love to uh sorry I just love to get you get your take on uh uh if you can if you can expand a bit more on some of the stuff you're doing with Bitcoin yeah sure thing so um bitcoin's extremely underserved in terms of data um and it's been now made infinitely worse by inscriptions ordinals runes Etc um but luckily myself and my co-founder some of the First Data stuff we built in crypto before super chain about five six years ago was actually in Bitcoin um so as these things have become more popular we've also had some great Partners uh with Moonstruck who are building some runes data some runes terminals some explorers Etc you know we've able to to provide Bitcoin data with a similar level of verifiability to the rest of our data um so with runes for example and inscriptions they all kind of have similar problems so uh in EVMS you have Smart contracts so if if I try to send you $50 it will just fail on chain if I don't have 50 usdc so whereas on bitcoin I can just say hi transferred you $50 of this inscription and there's no verification at all uh you have to tr trust the centralized service because all the verification for ordinals inscriptions and runes is effectively an offchain standard although I wouldn't even call it a standard today it's still evolving um and that's a nice way of saying it's a bit random how it's implemented uh like for example we actually have to make a silly decision which is if you minted a thousand of an inscription and you put a comma in it is it valid or not because it's not a number anymore according to a computer but obviously a human being will look at that and go yeah of course it's a thousand but it doesn't PA as a number so we have to like and we have to check other people's implementations to see you know which what's the truth here basically and come to some kind of consensus around that um so yeah that's that's kind of what we're up to with runes so we're releasing basically inscriptions ordinals and runes data um from from Bitcoin uh pretty soon so anyone else for any other questions I can't see you very well so okay well thank you all very much for your time and attention appreciate it