LIVE: HARMONIC & BLOCK BUILDING ON SOLANA | 0xResearch

Transcript

What's up? Oh, I still need to fix this up properly because usually my co-host does it. All right. What's up, guys? Welcome back to Zerx Research. Today, I'm joined by the lovely Carlos and Ben, who is the CEO and co-founder at Harmonic. Ben, thank you for taking the time to chat with us. >> Yeah, appreciate it, guys. Happy to be on. It's like, you know, talk talk. Pretty um pretty excited. >> Very good. Um before we get started, do you want to give a short intro on yourself and on harmonic and then we can get a bit more in depth?

>> Yeah, cool. Yeah. So, for those who don't know me, uh I'm Dan. I've been in this long ecosystem for almost 5 years at this point in time time flies. My background is um you know, all like math, CS, um HFT, stuff like that. um during my time in school in HFT um really really fell in love with DeFi. It's you know like I really believe that you know DeFi is it's you know decentralized finance in general is technology that will will make markets more efficient and you know better better for everyone. Um and then of course you know one thing that division markets need or one thing that capital markets need is infrastructure for internet capital markets and that's

that's our lock. This is this is this kind of we built on a salon. Think of it as um an open block building marketplace for Salana. It's very open and competitive. Right? So for you know background on those that aren't as familiar with uh Salana as a network right is the background of Salana is you have a bunch of validators and a bunch of leaders right you know like these these are the guys that are responsible for basically the entire network right there's a little less than a thousand now about all the world and one of the hardest things they can do is pack blocks right so you know they're they're in charge of ingesting transactions from all over the world sequencing those

transactions broadcasting them out to the network. there's a recent grow increasing demand um for for the outsourcing of this block building right there's there are many validators now that are now saying hey you know I want you know more more sophisticated entities to build my blocks for me and this can be for a variety of reasons right one is application step or application control execution right sequencing applications a specific way to their preferences um another obvious the black builders wants is more revenue, right? Sophisticated block builders are, you know, generally going to be able to pack blocks more

efficiently and earn more revenue that, you know, the default becomes shipped inside the client. And then like the third I think main group of customers for this product, you know, of course are the users of chain, right? you know um there's you know a lot of native sequencing behavior that we've seen going on in slot in the past right you know it's no secret that we struggled with this in the past users they want more control over how their transaction flow is handled right you know I as a user want to say hey you know I want to submit my transaction to you know helious or temporal or you know someone who I know it's not going to sandwich me um and someone who I know who's going to

handle my transaction responsibly it's, you know, in in a trusted manner, right? Um, right now I kind of have to broadcast my transaction. I have a whole network is play this kind of cat and mouse game of, you know, sandwiches, you know, and all this other stuff. So I think the best way to describe harmonic is it's a system that you know not only increases revenue for validators and creates value for applications but also bridges this incentive gap between applications and users. >> That that's an awesome like overview. I think like there there's a lot to unpack here. Like I think people are mostly

familiar with Jetto's block engine today. Like I think one of the main questions people have is like is harmonic competitive uh to go or how is it different uh from go? >> Yeah. So um so Dio's uh traditional product right is their block engine which just kind of like basically it's it's an auction that creates a bunch of bundles right which like instead of old blocks um you know they they kind of send bundles to the values um then their new product is is this right which is full harmonic different because It

creates a competitive marketplace, right? So with like any individual system, um you're basically completely vendor locked into into one builder or harmonic is an open marketplace of block, right? Instead of only receiving blocks from one entity, receive blocks from as many entities as I want, right? And um so that's that's like pretty important right for a few reasons right so one is like in different scenarios different blockers are going to have different edge they're probably going to generate more different scenarios and as a validator I won't do in every scenario um the second is for you know all the

features that applications can use as fonts if you have you know 10 20 30 plus you know these font builders that are all giving on features or applications that are all competing to give users the best execution experience and the best sequencing experience. The end result is probably that we're going to have a much better world or the end result is that the world is going to be much better for users and applications. I think that one of the things that you know I really appreciate about Salana like I have appreciated about Salana over over the last few few years is every single lay of the stack you have a lot of good teams attacking

it right marketing look how spread or look how high spreads are these problems right when you have more teams attacks on them It's it's going to make that work better for everyone, right? Competition keeps people honest. It pushes people forward and you know it forces us to all innovate and to create better products. It's like the place where we can at least see competition basing is is is right. I think uh I think Gio's a good team, right? Um, you know, I think, um, I think that, you know, if they want, and I hope they do, um, go is going to be able to build blocks on, right? And

users that only want blocks from G, like they already have the option to to only take, right? Either by pointing directly to Gus, right? Or, you know, they can do that through Harmonic 2 and say, "Hey, I don't want, right? Um, you know, this is just opening up optionality to validators that want, right? Like if you only want blocks from one entity, like that's fine, right? A lot of a lot of valids want, you know, blocks from a lot of different things. A lot of want to, you know, maximize revenue in every situation, right? A lot of believe that having more folks attacking this player will result in a better experience, more applications for users, right?

One thing you kind of touched upon in your introduction is this idea of like applications being involved in the block building process. Can you expand a bit on how that works and why it would be beneficial for Solana? >> Yeah. Yeah. So when you only have one blocker, right, that block builder is probably not going to be good in every scenario, right? So what I meant by that, right, is you know, every every block kind of has their own specialties, right? Maybe maybe some black holes are good at generating, you know, a lot of revenue in certain scenarios and other black holes are good at generating revenue in

other scenarios. And you know I as a valer want the best block in every scenario right and harmonic allows us to do that pretty seamlessly right because it's um think of it as kind of like a block auction right where everyone submits their blocks right and the block that belongs to all preferences for for example revenue um is selected right and then the other thing is there's kind of an incentators are you If in a completely selfish short-term world incentivized to you know behave extractively right Harmonic bridges this gap in a few ways one is one is

obviously there's a lot of rules that you know you have to follow as a logical system right like you obviously cannot stand people right you obviously um have to follow these AC rules right so applications kind of essentially declare a rules, right? They say, "Hey, I want my transactions to be sequenced a specific way. Violate these rules." They have to abide by these rules and they compete on throughput, right? So, if you have one builder that in the future is printing out 100k a million plus, right? That rocket is probably going to win more,

right? Because in this future world where we have several thriving vertexes on salon as market and all these updates get even more competitive, you want blocks that are cranking out 100k plus million TPS, right? Like onset learning are really great teams, right? These guys I totally believe can scale can scale the network to 1 million plus TPS. We need a fast 1 million TPS builder marketplace that scales with it. We need fasting and TPS that scales with it, right? As well I mean like out of schedulers are good, right? There are there are sophisticated block builders out there that have proven to make them

better and that has just created an immense amount of for like this that again allows validities to receive these blocks in a way that is not harmful to the user application experience. Great. I I have a couple of questions that just to follow up with that like I think in the race announcement you mentioned that uh Sena has a need that is currently not being met like for many applications. Uh one clear example of this is like per on Salena which have a clear structural disadvantage like when it comes to like market maker cancels.

you touched briefly on that, but I was just curious on like how you view GTO BAM and how this plays like into harmonics value proposition. From what you've been describing so far, it seems like you are uh creating a more competitive block building marketplace and giving back like the power or control to the validators and applications to have like the most specific control over their transaction sequencing. I'm just like yeah curious on your views of BAM and how it can help or how it evolves in the SA ecosystem going forward.

>> Yeah. I think I think team I think they've traditionally done done a lot in the past to push the network forward. Um that said, I think there's, you know, also kind of a big hold market here, right? Um you know, there's there's no competition in the block layer, right? Um you know, there's no multiple teams that are all keeping each other honest and validators don't have any optionality, right? So I guess you mentioned a few things, right? One is right. I want us to get to a world where we can, you know, 100k a million plus per TPS, right? We are much more likely to get there and have the best age experience for applications, right? If

we have, you know, several good teams attacking this layer instead of just one, right? I think in you know almost every vertical right we we've seen how competition creates better faster and lower latency markets right again right example of this is these proposals right the cheapest way to buy sold right now is on salon for retail right retail can send their stables to salon they can swap to soul from stables, their average is probably going to be like a few pips

if that, right? Which is in order amount to lower the fees that you pay on any drug exchange. That's pretty cool, right? This is this is the result of open and fair competition. >> Yeah. I mean I just for context, Ben and I spoke at permissionless this year and he had this view that prop amms can actually be a superior model to order books to provide like the deepest uh liquidity and tighter spreads in Solana DeFi. Then like I would love if you could like walk us through through your thesis basically on like propm as a superior mechanism to sorry yeah propm

has a superior mechanism to order books like uh for defi systems. >> Yeah. Yeah. So look like the key intuition right is that clubs were built for a world where everyone is colllocated and where latency is zero. A decentralized network means validators are spread all over the world, right? You know, like what's your, you know, average ping like over the open internet like to and from Europe and Tokyo maybe like 220 milliseconds or so, right? And then, you know, over really fast infrastructure like double zero, right? You know, even though you're going to have direct fiber lines between, you know, say Europe and Tokyo, you're bottlenecked at the speed of light, right? The speed of light is too slow, right? um you know like 140 millisecond

roundtrip time between Tokyo and Europe right like that is going to cause trouble for you know makers that are trying to coordinate across both regions right you know I I as a maker need to post orders to you know both this validator in Europe and this other validator in Tokyo propm solve this problem by running a lot of their logic on chain right for example right one thing that you want to do as a market maker is you want to react um to orders that hit you, right? When when your logic is running on chain, you can react instantly, right? Because onchain is as real time as you can get, right?

However, when you are on a collab, right, first I have to read that data from the network, right? So, it gets broadcasted out, fanned out through turbine, right? Then I got to submit it all the way back to the block builder/v validator and I it's going to be pretty delayed, right? So, you know, I think this trend of makers and markets running more and more and more on chain is really exciting, right? And I think that Salana is the best chain for this to happen, right? Like the throughput that we've seen on Salana is insane, right? And we're really only scratching the surface, right? Like you may have seen, you know, just like a few months ago, um Katie on our team put out like 100k TPS block or just like something like incredible, right? And like this is

again just scratching the surface, right? like having these prop AMMs, you know, blasting so many Oracle updates out to the network, right? You know, like this this like wouldn't be possible on a lower throughput chain, right? And Harmonic is also going to be pushing throughput even further, right? Because, you know, now I'll have a whole bunch of block builders, you know, each with their own specialties, right? And, you know, each with really cool tech. And these guys are going to be competing to pack the most Oracle updates and to fit the most Oracle updates into the blocks, right? Because they earn more fees, right? And like as a result throughput is going to go up right and then TPS will be able to scale right again on fire dancer I completely believe that these guys are going to push TPS you

know and network capacity into 1 million plus TPS in the future right we've made extremely good progress right and again there is demand for a builder market that scales with that right we need to push Salana into you know 100k a million plus TPS territory right? All of these Oracle updates like if you just look at the chain like you know like Sulfi, Humidify, you know, Tacera like all these guys are they're blasting so many oracle updates to the chain, right? And like again there's there's naturally demand for a high throughput block backing a high throughput blocking system to accommodate this demand.

>> Yeah, for sure. I mean, go ahead, Buck. >> I we've kind of danced around this a bit. I think both you and Carlos are a lot more clued in on it. Can you just explain to me like I'm very very stupid how um blocks are built such that they're more efficient and how a block builder can compete with another block builder. Um >> yeah, so it's it's a really really hard problem, right? So you know like let's say you just have a bag, you know, of just say like a 100,000 transactions, right? Lots of them are garbage and you know invalid and worthless. Lots of them have hidden value, right? And then a whole bunch of them conflict with each

other. Others don't conflict with each other, right? You know, different transactions take different amounts of time to execute. They all touch different state, right? It's extremely complex, right? And so when you have a really complex problem, you know, it requires a very elaborate and intricate solution, right? So like this is like in the CS world, a problem we call NP hard, right? Because there's not really any, you know, polomial time, you know, known optimal solution for this problem. So instead you have to make a really good approximation, right? And you know different different block builders are

going to have different approximations. They're going to have different strategies, right? And in certain scenarios, different strategies are probably going to be better than others, right? You know, I think that if you have such a complicated bag of transactions that touch so many different accounts, touch so much different state, you know, some have hidden fees, some don't. And the and like the other thing is that some transactions might have more value to certain block builders than than others, right? Um, and so you kind of throw, you know, all this really complicated stuff together, you get this really, really hard problem, right? And this is this is a problem that is definitely much better tackled by dozens of block builders than just one, right? The hardest problems in the world, right, are solved by, you

know, generally several teams all around the world, you know, other than just one, right? You know, like for example, think like all of this AI stuff that's going on, right? like you have a few big AI labs, you know, that are kind of rising up, right? And, you know, each each kind of has their own niches and each kind of has their own specialties, right? You know, I don't think any, you know, certain chatbot or, you know, anything like that is always going to be the best in every single scenario. And like again, like the other thing is this fierce competition between these labs, you know, has again pushed the technology forward in ways that I don't really think would have been possible otherwise.

Carlos, do you want to go ahead now? >> Yeah, sure. Like I have a question regarding like Sena's evolution. Like it seems like block building to some extent like rises because of Sena's inefficiencies at the microsstructure level. And so it seems like increasingly over the past few months uh Salena microstructure has favored priority fees as opposed to jet tapes or other types of uh me inclusion mechanisms if that makes sense. So if you look at the real economic value chart for instance, you see uh like the divide between uh priority fees and jet

tips, right? Like a few months ago, jet tips represented most of that whereas if you see like the evolution over time like the market is increasingly preferring priority fees with priority fees taking a higher share of that uh if that makes sense. So kind of my question to you is like >> as Sena becomes more efficient like why do we need these block builders like why are priority fees not uh sufficient enough? it seems like ME uh providers or blog builders are increasingly are increasingly competing for a much smaller share of the pie, right? So there's so there's a few things here, right? And so like one like you know as even you know as priority fees become

you know more and more important right you still want to maximize your priority fees and that still requires sifting through hundreds of thousands of garbage transactions which requires extremely high throughput efficient infrastructure right like try doing a 100k even a million like ED25519 like sig verifies like a second right you know like thank god we have very smart teams like fire dancer that have you know done a ton of work at this field and you know are really working to scale this throughput like up up to these levels right but then even then right I need to I need to determine whether or not it's spam I need to determine whether or not it's executable right you know the average out of the box scheduler is not the best at this

problem and this has been proven right you know we we've written custom schedulers you know I'm sure other teams have written customulers that do better than the baseline that do better than the default as institutional adoption continues to increase on Salana, right? You know, shout out for like the first actual Salana ETF that we have, right? You know, I think the Bitwise ETF went live like just two weeks ago or something, right? And with this increasing institutional adoption, the there's going to be a natural demand for yield, right? And being able to sift through, you know, all these garbage transactions, being able to pack them in the order that is optimal is going to become increasingly important, right? Um because you mentioned that, you know,

priority fees have fallen off, right? You know, I believe we need to grow the pie, right? And we're going to do this through a few ways, right? You know, like one is obviously like Harmonic has integrated ACE. This is going to create value for applications. This is going to enable us to build in the future a million TPS per deck on Salana, right? A lot of the infrastructure for this already exists, right? You know, we are we are just a piece of that puzzle, right? You know, we are we're going to implement ACE, right? We're going to create a builder market, you know, of builders that are all competing to create the best experience for these burpexes and their users, right? And the result right is that the user experience is going to be better right more and

more users are going to come into Salana. Salana is going to be the best place to trade perks. I think it's already the best place to trade spot for retail, right? I cannot get better prices on soul anywhere else as as a retail trader, right? And soon I think that'll be true for institutional traders too, right? As demand continues to increase, market makers are going to provide more and more more and more and more liquidity, right? So the mindset here has to be growing the pie, right? Yes, we can maximize revenue for validators, but we'll do it in a way that is sustainable to the network and that pushes the network forward and that meets the increasing throughput, low latency, and sequencing demands of applications and their users.

Um, and you mentioned ACE a few times and specifically with regard to per, um, is this something to do with like not being able to prioritize cancel transactions? Is that why we think um, >> yeah, so look, I mean I I have my own opinions on, you know, like what's the best way to build a DEX, right? But like I think overall I think applications should be able to experiment with anything that they want, right? I think that if they want to do cancel prioritization, they should be able to do it. I think that if they, you know, want to do more of like a FIFA type thing, they should be able to do it, right? Like I think overall, we should just give applications maximum optionality, right? And allow them to experiment with different microsructures

and allow them to experiment with different, you know, sequencing rules and application designs, right? So, um, for those that want to prioritize cancels on harmonic, they they will be able to do it, right? And like also to that point, I think one of the other things that's driven Salana forward so quickly on spot, right, is that all of these market makers, dexes, application developers, etc. have all been able to experiment with new market structures extremely quickly, right? Like the C5 world is bogged down in regulation.

There's a very high barrier to entry, right? like you know I can't just make a prop amm and like ship it to NY right I like I just can't do that right um and like you know I also maybe maybe don't need to right because on a centralized exchange latency is zero right um however like again in DeFi world right I can just ship a prop in like a week right I can ship a new DEX really really quickly right and just being able to permissionlessly iterate on application design and just to permissionlessly just publish new apps and acrew users to those apps and just to immediately compete on liquidity with all the other

great market makers on Salana is pretty cool right and I think that's a huge reason why Salana's winning on spot fills for you know retail trades on majors today >> yeah that that makes a ton of sense to me like I feel you're one of the persons who understands uh prop amm most like better than most people. Like I have a question. It may be stupid, but do you think prop amms could potentially enable per like instead of just using Yeah, go ahead.

>> Yeah, I do. I do. Right. So, um I mean look like per right like to preface, right? There's a lot of different components you have to deal with, right? Matching engine is just one, right? And I absolutely think you can implement a matching engine that involves props, right? There's many ways to do it. I would absolutely love to build this like you know in in the future, right? you know, like our team was, you know, jamming on on uh designs for like perexes a while back just, you know, really just for fun, right? We were kind of thinking to ourselves, how do we make this better? And, you know, like one of the one of the things that we came up with was, yeah, kind of kind of like a prop amn based matching engine, but there's also a lot of other

hard things that you have to do, right? I think risk is extremely hard to get right. Right. Like I think when there's a huge market move, right, all of the responsibility falls on you and your risk engine to make sure none of your users go underwater and it's a really really hard problem, right? Um and so like I think that you know propm will definitely make mashing better on perks. Um but I think that alone is not sufficient, right? Like you also need a really good risk engine. Um, but just in terms of like getting the best fills on Salana, I do think a prop mm based approach would be better than anything that exists today for the reason of like that latency for makers to react to

trades being zero. This is a dumb question, but do you do you think there's any issue with Oracle reliance to sexes when you're running a propm like >> so I mean like different assets have price discovery in like different places right so like for example for an asset like you know Bitcoin right a lot of it like the majority of the trading still happens on Binance right now we we will change this in the future right um but for now the majority of the trading happens on Binance, right? So for traders and for like sophisticated market makers, this is important, right?

Because these guys are constantly trying to predict where the price is going to go, right? And the majority of that signal value for where is the price going to go, right, originates on Binance, right? Um now this will change, right? Like as we continue to push throughput higher, right? you know, with harmonic and with fire dancer and with all the perf improvements that Agave has been making, right? Like again, like we're gonna get more and more transactions on Salana and I think you more and more spot volume will move to Salana. Um, but yeah, like I think it's kind of the responsibility of each individual propm to create their own signals, right? And to create their own oracle pipelines, right? Like I cannot just use a pick oracle and be competitive, right? best mark makers in

the world have sophisticated models that they essentially predict short-term sometimes even long-term price movements right so each propm is probably going to have their in-house kind of like oracle updator right one of the things that I think will be interesting is when we start seeing price discovery for assets on chain right like when price discovery for assets is happening on chain and the majority of volume for assets is on chain you might have people like running actual machine learning models on chain and like actually running a lot of these sophisticated models on chain which I'm really really excited about um because you know you're going to enable all all of this sophisticated modeling but also with Salana's extremely high throughput and composability

>> one um kind of unrelated followup in terms of harmonic when you think of like who your main customer is is it the application is it the block builder or is it like >> yeah so I'd I'd say ultimately right like it's the users of the chain right like the users of the users of the chain are the ultimate customers of every single application right they're the customers of every validator right these are the people that are paying fees to both the applications and the validators right like to be clear I think the applications are you know also a really really important customer and I think the validators are also a really important customer right um like I think overall our main priority for us is making sure that users have the best onchain experience and like often times that also means you know making sure

applications have the best onchain experience. You know, one of the things that we did as we were building Nomi, right, is, you know, we fiercely, you know, um, communicated with all these applications, we're constantly asking them what they want and we're implementing those features for them. And this has just resulted in, you know, adoption beyond what I could have even imagined, right? And, you know, for harmonic, we're going to do the same thing. You know, I think we're going to have the best customer service for users and for applications. I think that you know really once we launch this thing and gain significant market share like a lot of my job is just going to be talk to users talk to applications improve the product talk to users talk to applications improve the product right and for validators too right as like

validators are the ones that are actually running this thing it's you know our job to bridge this incentive gap again between the validators applications and the users right so you know we're also going to be fiercely communicating with validators asking them what they want asking them what they want to see and improving the product, right? You know, our our job is to give the apps, users, and validators what they need, but in a way that is sustainable for the network in a way that will push throughput, you know, to 1 million plus TPS or ACE included and like again in a way that will result in the best, you know, application and user experience.

>> Last question, Carson, and I'm going to let you. Um, so Salana had a few very interesting governance proposals over the past few years. Um, is there any that you were very very against that you were like? >> Yeah. Yeah, there's there's um there are uh there are definitely some that that I was against, right? There was there was one called um there was one called pro a long long time ago. This was like back in like 2022 that this was being pushed. Uh it's like paid application rebatable account write fees. A long long complicated name. um long long complicated name for a long long over complicated SIMD. Um the the idea of this SIMD was to give applications more

control which I think is really really good. I think that's a you know really big thing that harmonic is doing right. I think we're giving applications more control right the problem is that the way that the SIMD wanted to give applications more control made the protocol really really over complicated. It would have delayed timelines for fire dancer a lot. it probably would have delayed timelines for Agave and it just overall introduced a lot of you know overhead and accounting to the protocol which is already pretty complicated right like I think that we should continue to simplify the protocol as much as we possibly can standardize it document it right create the best foundation to iterate right so you know like the SIMDs I tend to like are SIMDs

that simplify the protocol a lot like for example we've seen some SIMDs some SIMDs to reduce certain constraints on transactions being processed. That's great. There's a recent SIMD that puts priority fees and compute units in the header of the transaction. So now I don't have to parse through a 100 transactions just to figure out what is the priority fee of this transaction or what is it worth, right? Like the kind of SIMDs I tend to like and that we generally kind of tend to like at Temple are, you know, SIMDs that push throughput forward, SIMDs that simplify the protocol, right? Now this isn't true in every case, right? Like I think for you know things like you know MCP right and you know features that you know you could argue add complexity right like I'm not saying that they necessarily do

but like some people might argue that they do right like I think there's tradeoffs that should be discussed with with changes like this. >> Yeah that that totally makes a ton of sense. Um you mentioned a lot of upgrades like that are coming to Sena that can potentially increase throughput by 100x by 10x like you mentioned fire dancer I don't know if you mentioned Alpenlow but one >> protocol that you haven't mentioned yet which I am curious if you see any impacts on is double zero and how you think it plays out into this future.

Yeah, look, I mean anyone that has ever, you know, done, you know, HF or, you know, done like extremely high throughput over the public internet knows that the public internet leaves a lot to be desired, right? So, you know, I I I had a lot of experience with this, you know, during my time at HFT. We had a lot of experience with this building. You send stuff over the public internet, the jitters crazy. Sometimes your packets just randomly get dropped, right? Sometimes it just takes a super weird route. Like sometimes I want to send a packet from you know maybe Amsterdam to Frankfurt and you know like it'll route me through you know the US for some reason right that public internet has a lot of quirks right we need to I mean look we need the IBRL right we need to increase we need to

increase bandwidth and we need to reduce latency and I think zero is a fantastic effort at doing this I think they're a really really great team Andrew Matteo and Austin you know are all really really smart and I think they've built a phenomenal team and you know I think I think that the work they've done um is really really important for you know pushing Salana again to this like 1 million plus DPS and we would love to and we'd love to figure out a way to you know put put Armonic on zero right like we've done a lot of testing and they they are they are improving they are improving a lot of the routes right and the bandwidth continues to go up right and I think they're going continue to iterate on their product and I think

look like the end result is going to be we increase bandwidth and reduce latency. >> Very good. I um kind of jumping a little bit back you very diplomatically said you care about the applications the users and the validators which I think is the correct stance to take in public but these three have incentive misalignments um I imagine. >> Yep. Correct. Yeah. I know they do they do right and it's like again like as I mentioned like initially it's our job to solve this misalignment right like it's our job to be that intermediary between the applications the users and the validators and figure something out that works for everyone right so and like for one right like this this involves ace right so like instead of validators

sandwiching everyone this is bad we don't want that we're not going to allow that on harmonic right just to be very clear no sandwiching on this thing right um validators that want to do it, they're not going to do it on harmonic, right? And they're actually going to end up losing out on revenue, right? Because we're giving applications the opportunity to control where their transactions go to control where their flow goes, right? In in a harmonic world, you know, applications get to choose where their transactions go, right? So, if they don't want to send it out to, you know, validators that might sandwich them, they don't have to. Right? Now, like, you're kind of just stucking your transaction out to the open network.

you know, hopefully there's no one that'll sandwich you, right? But like we're building tooling for users to avoid sandwiching, right? We've already built a lot of this tooling with with um NOMI as well, right? You know, we've we've detected hundreds of sandwiching validators and you know, we've all blacklisted them from Nomi and the end result is a lot of them just shut down, right? And you know, as we give users more and more tools for them to be able to control their transactions and to be able to control who sees them, right? sandwiching is gonna plummet, right? The second right is you know like this like this a stuff right like the pi is not fixed right like this max extract sequence applications extracts or in a way that's maximally extractive is short

term the applications they'll leave they'll form app chains they'll go to other chains where you know there's nobody sequencing them in malicious manner right like this is like again really really bad for validators in in the in the long term And as such, we're not going to allow this. And for those validators that want short-term profits, right, I think I think there's something to be said about what the correct a fee is, right? You know, I think that validators definitely deserve some consideration for their hard work in sequencing these applications the way these applications want and generally deserve to be sequenced, right? So, you know, by by taking this a right, we create value for the applications.

Validators still earn money. we grow the whole pie. >> Yeah, that makes sense. Um, one thing that I'd like to understand a bit better is Harmonics business model. Uh, it seems like you are improving things for the user, but that also implies as you mentioned like you're sacrificing revenue to some extent by allowing user to have like better fields and whatnot. Like do you take uh do you charge a take rate uh for transaction inclusion? like h how does harmonic like essentially uh generate revenue and become a profitable business like while advocating for the users, >> right? Yeah. So, it's so it's just fees,

right? So, you know, like there'll be fees on tips, ACE, and possibly priority fees, right? We're still working out what the correct number is for this. Um, but like what I can say is we're going to grow the pie, right? Like if we take more fees than the validators would otherwise generate, they're they're not going to use the product, right? you know, we don't we don't expect validators to use the product if we just brought them blind, right? You know, instead, we're going to create lots of value. We're going to grow the pie, right? And in creating this value, we're going to have more and more applications come to Salana, more and more users paying fees to Salana. The end result is the pie grows for everyone, right? And even in the short term, right, there's going to be immediate revenue from applications that say, "Hey, you know, I

have certain rules for how I want you, the validator, to sequence these transactions." Since you know in exchange for these guarantee in in exchange for these guarantees I am willing to pay you the validator and you know harmonic the system that is facilitating this some consideration for that hard work. >> Yeah that's really helpful. I guess I have one more question which is aside from uh propm and perexis what are you excited to see people build on Solana like like that we haven't seen before or do better >> yeah so I mean first of all I think that uh I think that everything will be better and that everything will continue to get better you know as we again push

throughput forward right and we enable new and we enable new completely design space for these applications Um, look, I think on the I think on the spot front, right, just spot trading, right? I think that propmms were really just scratching the surface, right? You know, I would love I would love to see systems that essentially iterate on this running logic on chain idea, right? And take market structure to the next level, right? You know, I'm hearing that, you know, maybe maybe some people are working on things like this since, you know, these things aren't going to be your standard, you know, spotexes or perplexes. You know, these are going to take, you know, um this kind of prop and

market structure and this idea of running logical on chain and these these systems are going to iterate on that logic and take it to the next level in ways that, you know, result in better fills than like you could have imagined otherwise. I think I'm really excited to see a lot of the innovation we've seen on the router layer, right? I think that all this onchain routing stuff is really really good for users. I think these users are getting demonstrabably better fills, right? Um I think I think Oracle infrastructure is another one where I think we've seen a lot of great products or a lot of great progress, right? And there are a lot of great products, right? And I think that as Oracle infrastructure continues to improve and become more resilient, right? You know, we're going to enable more and more capital inflows on chain, right? Like I

think borrow lens are going to continue to get a lot better. The markets are already getting a lot more efficient, right? like TDL is near near an all-time high right um you know revenue is you know also has been significantly increasing right I think that um yeah I think I think borrow lens are really great um I think I think I'm excited to see more more like Dpin stuff right you you know I think that um look I think demand for energy has massively been increasing because of this AI boom I think there's a pretty big Dpin opportunity there. Um, yeah, look, I mean, I think I think the things I'm most excited for are probably the things that I can't envision, right? Um, that's

that's that's usually how it is, right? You know, I think overall, let's let's give people the tools to innovate and to iterate. And I think that we're all going to be pretty surprised at how good this network and its products can be. >> I'm also very excited about Deepen. I'm just hoping that we somehow manage to figure out a way to tie how it works together with a token. Feels like it hasn't been figured out yet, but that's what the future is for. I'm sure somebody will figure something out. Um, >> incentive design is really hard.

>> Yeah, very easy to talk from behind the camera like me and say, "Oh, why haven't you guys figured it out?" But I can't. >> It's a it's it's a very hard problem, but I can say I think we've made good progress. Um, Ben, I guess before we wrap things up, I um I want to ask you if there's anything that people should be doing. Um, what should they be reading? What should they be visiting to learn more about harmonic? What do you want people to do? >> So, yeah. So, we're gonna we're going to publish all the docs there very soon.

So, stay tuned for that. Um, you know, I think I think podcasts like this are also a great place to go. You know, I think I think you'll be seeing more of us in public, you know, as we continue to iterate on the product and it continues to gain adoption and we continue to create new features, you know, um any validators, they can reach out to us directly, you know, on Twitter, Slack, whatever it may be, right? Um I think I think Twitter's a good start. I mean, we we have great relationships with with all the validators. They all know how to reach out to us directly, you know? We also have great uh relationships with all the top applications, right? Um but for now, I'd say reach out, you know, on Twitter or Slack or, you know, by email, right?

You know, we're um hello at harmonic.gg if, you know, anyone wants to reach out or get in touch and andor has any kind of proposal. You know, I think I think we're once we drop once we drop the docs, we may do something like a more public venue like Discord. It's still it's still being debated. Kind of just depends on what everyone wants at the end of the day. But yeah, I mean overall we've seen a lot of excitement across the board from validators and applications alike. And you know, we're really excited to show them what we can do.

>> Perfect. Um Carlos, anything you want to end off with? I realize I don't let you talk a lot at the end. >> No, no worries. I mean, incredible email. Helloarmonic.gg. >> Get in touch. I mean, anyone that has any interesting proposal or, you know, wants to get in touch or chat, you know, anything, you know, we're we're around. Just shoot us a DM on Twitter or, yeah, I mean, send a send an email to the email. We'll uh we'll be pretty responsive. >> Perfect. Ben, again, thank you so much for coming on. >> Appreciate you guys having me on. Thank you Ben.