Fireside chat with Sandy Peng, Scroll, and Kartik Talwar, ETHGlobal I Pragma Singapore 2024

[Music] all right welcome back everybody from that break just a few more talks to go before we go into our happy hour and we'll Jump Right In for this next interview I want to bring on Sandy from scroll we're going to talk about a lot of cool things about the evolution of scroll some stuff that helps us understand how we should think about other l2s and a bunch of stuff in between so please give Sandy a big round of applause I think we're just 12 months in to the mainnet launch maybe a couple weeks away yeah that's right so it's been a year congratulations year we launched on the 10th of October here we go to three weeks yep um how has the launch been since and uh if you were to reflect on the last 11 and a half months what would you uh say kind of went out went really well and uh what did not um I think there definitely a few things that um went really well and a few things that could have done better um the the things that went really well when you're looking back is always that nothing went wrong um and everything still works um and so we've been um running the train for a year and there's definitely a ton of learnings and um and and and the minute your product ships the level and the quality of feedback just becomes an order magnitude better because like before that you're just relying on guesstimates and assumptions and um we've definitely fine-tuned our development road map and fine-tuned the way we organize ourselves quite significantly over the last year to optimize for growth and to optimize for um a lot of the overwhelming feedback we got from the community so um so that that's the good part um the bad part is um I think um as an or is like running and uh and people are constantly using your chain um our our response time and our reaction Loop hasn't been as tight as um uh we would have liked um so you identify issues but the time of which you fix it is never good enough because um people on chain demand like realtime response and so if there are like ticket skipping or um any kind of you know like minor issues um we we we hear a lot about it and so there's just been a lot of like you know bits and pieces um that that just keeps us busy but is that you commenting on the fact that if you get a complaint or if you know you have to make some improvement it just takes more time to fix it because you have to wait for a proper protocol upgrade or something else uh protocol upgrade that's one thing and uh the other thing is just how seeing people behave on chain I think we could have anticipated some of the things were just maybe not as sensitive so um yeah that makes a lot of sense sense if if you were to launch mainnet now or looking back at all your learnings from the last year uh would you do anything differently or rather advice for other protocols we trying to launch main net um yeah so um I think um we focused a lot on the things that we thought were essential so um when we think about building an ecosystem we kind of thought about it from first principle and um about okay what are the things things that um are essential um in ethereum and uh in The Wider blockchain space and we started from there so uh we're very blue chip oriented as an L2 so we have um we have base uh sorry we have um a and uh we have um like Unis swap so on and so forth like compound Lio like all very early on in our development and and and and that is good for testing the compatibility aspect of our design because it's a net new chain uh but we probably should have been a little bit more ambitious in terms of focusing on applications that uh leverages our uh unique selling point which is um kind of super fast finality and um but fundamentally for a chain single chain application developer uh this doesn't actually matter that much we had a thesis that people who are super dgen on chain who are making allocation across different l2s would really appreciate this feature and we try to talk to a lot of these kind of onchain like superactive um and and price sensitive users uh but for them they've um the feedback is that they've gotten used to a way of managing their portfolio where they don't actually move from chain to chain because just they they're they're quite set in the way of like you know where where where where chain movement is a little bit painful or creates a little bit of friction uh but how but I think the the time for that um Behavior to change will be when uh finality gets to the subsecond category and then um you know call data between chains and assets and nfts moving between between chains will become near instant and at that point I think there's going to be a a kind of a seismic kind of behavior change where with one wallet one interface you can maneuver all of your assets across the l2s um and then the next problem after that is then like how do you use one wallet interface and maneuver assets across like salana and other l1s and Cosmos Etc so uh I think baby steps but I think we're very close to getting like sub minute proving time we're currently at um like eight minutes uh but you know in three to six months time we're going to get to like under we're going to get to like 10 seconds that's awesome that's a massive improvements yeah um so it so fundamentally for people on the main chain this doesn't make any difference whatsoever unfortunately so we we we look and feel exactly the same as every other chain uh but I think once uh and the and and the costs have come down significantly and I think that's going to be the the kind of the critical moment where um the entire like evm ecosystem is going to come together and maybe just so I personally understand this better the proving time is not the same as finality here right this is just the settlement on layer one but they are still noticing what 2 seconds for confirmation and that's finality or is finality still the settlement like what's the finality time now versus the finality time right now is about like 10 10 minutes I see so which is the actual eight minutes plus okay understood and that becomes when it becomes UPC yeah then it's actually beneficial because you have way better guarantees about the money correct yeah amazing um so that seems like a when you kind of thought about this as part of the launch you you said well it may not actually be as relevant to people but then um you kind of notice that only post launch from feedback um you also kind of started doing a lot of programs to kind of grow the ecosystem usage and uh was a loyalty program that you introduced um scroll sessions and um how kind of did you come up with that setup what was that that incentive program and uh what has been sort of the outcomes there um I think um we we we started the incentive system to try and be kind of like a GPS to direct capital to where um it's needed to get the ecosystem jump start uh previously we've been running the chain for about 6 months and we had about 200 mil tvl is on chain and um and that's completely like un steered super organic and everyone just like does what they want in the theme park but one consistent feedback we got from uh players was that I got a lot of pain saying like oh where should I go where do you want me to park capital and um and I didn't want to be a a kind of a manual traffic controller um and so instead of telling people oh um please be an LP here please like you know put the lending here and assessing their risk profile and explaining like you know how much this is needed versus another uh we thought of like trying out an automated way to do this and uh and it's been a very uh important learning process so when we started um blast kind of achieved tremendous success in terms of um capturing capital and attention with their points program and uh and and I think there are definitely a lot of learnings there where um the gold and the regular points um kind of uh left a lot of teams feeling um kind of unpre underappreciated so on and so forth and um and at the point of which those points ended it was basically the end of a project because it meant that there was no way um Capital would go somewhere that is not directed from the uh from the centralized front end and um and I think we we we somewhat like basically we Tred to be as credibly neutral as possible so by first like incentivizing the asset end so um and then the asset can still go wherever they want and then we started trying to direct assets to certain pools uh that were like you know that were so-called like more beneficial for the ecosystem but mainly from like a capital velocity perspective so um you know LP like for example e Pairs and important asset pairs on a DEX is much more valuable to an ecosystem and it generates much more kind of usefulness uh quote unquote um than anywhere else uh but um I think I think the high level learning from from running the program so far is that a uh people respond to incentives a lot um but do they stay around after the incentives are done uh so I think um we don't know that um and uh I think we plan on continuously applying incentives until we're uh and and and and and and I think that there needs to be a path for incentives to slowly ramp down so like all like web 2 or like internet companies uses incentives to onboard new users until a point where a certain type of user behavior is pretty set and uh so we will continue running incentives uh but we will be trying out different things based on the learnings that we've done so far um another learning we've had is that uh people are not as like economic driven like purely as we had hoped we we wanted to direct you know a lot more Capital to Dex pools and so we kept increasing the incentive like multiplier uh but the market is just not super sensitive to that people can see that your um your your your marks which is our version of points goes up quite quickly like exponentially and um and then but but that that that doesn't move the needle unfortunately um and then another um interesting point is that um people um um people have like existing habits so like there are um like uh lending users who are just here for the lending and they wouldn't even consider anything else and there are like blue CHP type of users who have very stringent kind of risk uh tolerance requirements and regardless of what the incentive is um they don't care they just have one type of behavior on chain and for those type of users um the the the the the the the uh how you apply incentive actually matters less and we've been kind of sneakily trying like you know fine-tune it if someone is like not super reprentative and um and and we're getting more and more experimental and getting more and more focused so two weeks ago we launched a uh a program with um with ethery cash which is a uh onchain credit card that allows a onchain d user Park their capital on a uh Park ether by eth on a or also normal eth and rap eth and so on and so forth and borrow stable coin against it and then swipe at any like Visa um accepted Outlets so that you can live your your full Financial life on chain in a very very Dent way um and uh and we've designed the um behavior in a way where um we're hoping that this can set forth a um Behavior change so from previous studies the assumption is that if you used a new card for three months continuously then this is this card is very likely to stick with you as your primary card for a lifetime even in web too just I mean like just that's the incentive that they give you for the points bonuses correct correct so uh so so the first three months is going to be highly incentivized for every net new user and uh they're going to receive like you know a bunch of different kind of points and incentives and um and and and and and and and and I think that that is something that you would pay for in terms of user acquisition for the defi core um anyways that makes a lot of sense is there um any other group that's targeted Beyond defi users or kind of how do you think about other categories of uh yeah yeah yeah so um for any category that is outside of the defi core I think the incentive mechanism is a lot more untested and it needs a bit more um like an experimental approach um so we are um we've been we're running an experiment with um a project called Huma uh it's kind of like rwa offchain yield of like 10 to 15% depending on the time you lock up and the money goes towards this um kind of a supply chain Finance like very tradun type of yield source and with also a variable onchain component to it so an aggregate is a very uh like attractive y Source overall um but like people uh the the current onchain user set who are super Deen uh did not react super well to this kind of setup even though the total like numerical yield economic value of yeah exactly exactly yeah so so so but this has been incredibly popular when uh we go to like family offices or people who are net new to the space who doesn't understand where like you know Dex yield comes from who who doesn't understand like the a looping strategy like so it's just like you know it's a new with new products you need to like find new audiences and um and we learning a lot of these lessons um on the ground and then uh the the last project I want to give a shout out to that is going to be launching on scroll very soon um is also a semi semi rwa project um the uh the the the the the mechanism and the smart contracts are on chain so it's trackable and um it it essentially um leverages the um the the onchain pool of capital but the yield Source comes from the price difference between the different usdt and uh local currency markets on P2P exchanges in Africa and so the base yield will get to about like 8 to 10% um which is very solid and then there's like variable yield based on like whatever the token launch might be in the future that sounds really awesome useful like this is super practical correct yeah so they struggle to raise funding um you know otherwise and if this can speak to the onchain Deen Community who are you know all moving around like it's like a perfect example of diversification if you're betting entirely on corre on chin yeld correct which I still don't know where that comes from yeah so we're we're learning we're learning and we're iterating but um it's a process and obviously we wish it could be faster but you know um we we're doing this with the Community are there any other themes that you've noticed on what's kind of popular at scroll like are we seeing more gaming are we seeing very specific ZK applications are people using the ZK VMS uniquely like what else is uh trending um I think um we there are like in terms of technical tools there's a lot of ZK Innovation um there uh we have a ZK wasm um kind of L3 that has launched on scroll and um it basically offers an evm environment and like faster and cheaper than salana and we've had a L3 order book deck um that is very kind of price competitive compared to something like you know close to like kind of an L1 like hyper liquid for example and so there's a lot of these like really cool product with Innovative Tech and applied ZK um but that that you know to get commercial success and get like visibility and build a community all of these things takes a really long time so I think the type of builders on scroll are generally like super strong like Tech Builders and they're leveraging the design space that's opened up by the ZK evm and also all the ZK tools are available and um what is somewhat missing is kind of that like that business um acument of trying to find a niche and trying to find the user and and and there's like maybe a little bit of a gap there for for some of these it's just really hard to be really into deep Tech and also into design and front end and acquisition yeah it's a it's a it's a well-round their Sport and um and we're trying to think of ways to get more people from different backgrounds to be attracted to scroll um I think people come to scroll because uh you know they they resonate what we're doing the fact that we're building cool things and we're uh we're building something that that's original and but these type of people who understands our valueable proposition are not necessarily like the best artist or the best like meme Lord and uh so I think the next stage of growth for us that we need to kind of build a stronger brand a more distinguishable brand and appeal to a wider set of Builders so that is the the next step of growth but maybe the lesson there is that you know we should have thought about this earlier it's uh that that list keeps on getting bigger if uh if you go down that route of uh retrospectives when you're trying to assemble a plane yes um what have been some things that have either been changes that have happened in protocol upgrades since the main net or things that have kind of come up as hey this would definitely make things better for the future development like are there any kind of things on on the layer to itself that you are improving or have kind of been introduced yeah uh we've been developing a number of things um since mayet so first of all uh We've designed key store to uh unify smart contract accounts across chains and uh we have um L1 eslo to read from L1 quickly allowing like better interop and uh the fees on scroll has reduced 3 to 10x with proof aggregation and um and and we're we're launching a bunch of other features to kind of serve a multi-chain future and uh and better ux in general so since May that we've done four upgrades and um the the lore which um uh we we've kind of like hidden in there is that uh each of the upgrade is named after like a well-known mathematician from the past and uh with like the acronym starting from like ABCD so we got 22 more upgrades to go and then we're done right the the blockchain's complete well think of something maybe different um how do you think think about a lot of these uh new zkv Ms coming up we've kind of seen a pretty big evolution of uh people taking the risk uh risk five um VMS and uh introducing the an optimistic rollups or you know just more application specific ones do you kind of think those are interesting or applicable do you feel like the more those things kind of come up they end up sort of not making or it kind of dilutes the case of having a distinction between a ZK VM or ZK rollup versus optimistic rollup like what's kind of your lens on this um I think depending on the time frame of which to answer this question um in the long run every chain will be a ZK chain um we're pretty certain of that um future um in the short run um I think there is a uh there is a difference uh so so in the short run I would say the time to build a general purpose and a holistic chain team that time has come and passed um I think there will always be um Niche and incremental improvements so for example um faster provs or faster throughput Solutions or more optimized noes there's always going to be these things that comes up and that can be uh and there will be really strong teams that focuses on the execution of each of these verticals but the time to build a holistic and well-rounded L2 team has passed a well-rounded the team it means like having a ZK function an infra function a protocol function and having the expertise um around you know 360 Degrees to know what is relevant and make the right choices I think like that kind of insight and knowledge is what will enable a chain to make the right decisions going forward we have about maybe three to five years where there will be constantly different competitors and like different Niche improvements and different incremental improvements the performance being like the the TPS game we played five years ago correct yeah yeah yeah so there's going to be a variety of that coming up in the next 3 to 5 years like paralyzation or something like you know but fundamentally um as long as everything's open source and you can plug into the existing stack um I think the main difference is just how well you make these decisions and how well um the core team can like absorb these new innovation so so maybe kind of if you continue that line of thinking what do you think five years from now or even three years from now the world looks like because you're effectively saying everything becomes table sticks cuz now everybody can copy every performance enhancements into their setup does that mean best branding wins does that mean there's going to be some Mo and defensibility on what's getting more traction continues to get traction or are you worried about now other competitors for for scroll um I think for the next period of this um so-called kind of infa arm raise the the thing um that most people don't like to talk about is a bit like iffy or not iffy but like a bit air faery is branding like branding is the only sustainable mode that I can see when you look at all the major defi protocols which is like the big battle of the last cycle um I would say like a and Unis swap they exist just as Brands um there are so many different like iterations and like slight improvements on them but no one can beat the brand mode and so whichever L2 can create a authentic and sustainable and memorable brand I think we'll do really well and uh right now honestly I don't think any um L2 is even close to getting there yet are there uh ways you can get there we talking about what celebrity endorsements and uh and know a lot of ads on F1 cars is is this uh is this how we get to better branding um or just big the better products on the chain is kind of what gets you there um I think one analogy I've been trying to use like increasingly to like speak to um you know audience outside of the the the nerdy techy Corp is that um chain is going to be um more and more commoditized uh there are some like product decisions that needs to be made right in order for to stay at the Forefront but essentially it's going to become like trainers and but whether you buy Nike versus Adidas versus something else it's about what um what you really resonate with as a person or as a developer and um and and and that brand needs to be authentic it needs to come down to like storytelling and in terms of how you once you found that story how you amplify it you know there could be various methodologies like you know eth Global for example thank you it's the perfect time to uh to conclude here I I'll take the compliment um but that was an amazing conversation and um I am looking forward to the next few years of scroll being the dominant brand in LT so please give Sandy a big round of Applause [Music]