Jito BAM and the Future of Solana

Transcript

[Music] [Music]

[Music] [Music] [Music] [Music] All right, we're live. Welcome back to

Zerox Research. Uh joined by a couple of the Research Boys and Dan Smith, uh Dan Reverend Smith, Carlos from Research, Ryan, Cash Flows, Connor, uh and Lucas Brutder from Jetto to talk about BAM today. Um before we get into BAM though, how's everyone doing? How are we feeling? Um there's a lot of excitement in the markets right now. So, kind of wanted to gauge where everyone's heads are at. It's a It's a tough day to have a bad day. Um, no, it's a it's a bad day for the Bears. Things have been fun. It's uh I tweeted this about maybe a week ago now and I was like, this is the first time in the last four years where it's truly felt like a bull market in the sense that there's, you know, a dozen plus things they need to be giving 100

plus% attention to. Um, and then today I followed it up with like a couple of those things and now uh JTO has just launched yet another one on on the radar. Um, so yeah, it's that's what that's the big change for me is, you know, I think we've had, you know, significant rally off the bottom long before the last two weeks or so, but it was always kind of like one or two hot narratives or two interesting ideas floating around at the same time. But right now, it's just like everybody is launching a new product or a product upgrade. uh and and it's in feels like innovation left and right on the back of insane rallies in prices. So, uh, really exciting time. Personally, the the 2021 stocks are back. Like, Open Door is back. The Spacks are ripping.

Like, the things that people left for dead. Like, it's I think I think that there's a small portion of the population that picked up their Robin Hood account after a few years and just started smashing by again after seeing Open Door 3X. Like, the names that I'm seeing is is the most interesting part. It's It's definitely a blast from the past. Yeah, it definitely seems like if you didn't as a business, if you didn't actually die within the last few years, um retail is just ripping into anything that um you know, survived and um they're all just sending upwards right now. I'm curious, Lucas, from your POV, like are you guys you're just heads down

building like this all the craziness around, you know, treasury vehicles and whatnot. Is that distracting at all? I know you guys have had some uh we've talked before about positive conversations like at the SEC, for example. So, there's there's certainly some good sides to the like positive news that's been going around. I'm curious where your head's at. Does it feel crazy busy? Yeah, definitely heads down busy. Um, yeah. I don't know if we expected like this craziness to come back pretty soon. I mean, what was it like October through January was just insane on Salana and just like the markets in general and

then things like really cooled down there for a while and we knew it'd come back at some point, but uh it's definitely sooner than we thought. But, uh, it's kind of funny. Uh it's like all the like boomer coins are kind of going up a lot right now. Uh I was on like Coinbase last week is just like Cardano, Hideera, Litecoin. Like it's the familiarity bias, right? It's like to Danny's point earlier about, you know, someone picking up the Robin Hood account again for the first time in maybe a year or two. It's like, oh, I remember this one like smash.

Yeah, Tazos is crushing it. is those tokens like up 200% in a month. I mean, if you're an XRP guy and you were just like averaging, you know, the last couple years, I mean, you're you're having a good couple of months right now. Actually, well over a year, I guess, at this point or nearly a year. So, yeah, I really want to see what XRP did to like gain this cult following. Someone needs to do like a there's like the I remember taking a business class in college and they have like the the case reviews from what was it like the harder Harvard business review or something where they like dive into business and what they did like how is Ripple doing this and or like how did

they do it? It's like pure Tik Tok gorilla marketing though that's my guess. They were pre Tik Tok though, right? That's the new age of their of their marketing campaign. They they were early to the community like cult following like the XRP or like the Ripple Reddit subreddit has been around for a long time. I remember I mean the conversations in there back in 2017 were like yeah this is going to replace all payment rails in the entirety of the world. Um which I think that's still maybe part of the you know the thesis but now it's just it's migrated to Tik Tok. I think they just they somehow snagged some of the like loudest, you know, people out there and got them to kind of spread word of it.

And big TAMs. I think you need a quadrillion sized TAM. Like that's the key. If you can find some TAM in the quadrillions, like you're set. Awesome. Yeah. Got to sell a very very large vision. Totally. Yeah. They've all um I think they've all learned from one another. um you know, you pick out all of your favorites from 2021 and I think that was the that was the big takeaway was to aim really really big um regardless of product. Um but hey, you know, if you bought those lows last year, you're looking quite nice today. Uh we'll see. We'll check in again in another year or two. Um but we had you on today, Lucas. We wanted to

talk Jetto. We wanted to talk BAM. Uh I think maybe just to kick things off for the audience and maybe for ourselves a little bit as well. Um the announcement's fairly new uh dropping this morning around this this BAM. Um I wanted you to just maybe give us a quick rundown overview. What is it? What does it do on Salana? What does it mean for JTO? Um just some high level thoughts and then we'll maybe dive into some some more specific questions as we get into it. Yeah. So yeah, we've been working on BAM since like we started kind of thinking about our system at a high level in like October, November, December. Um the way that people were using it is not kind of

what we originally expected like we kind of just imagined be for MEV. Um, but it turns out like all the Telegram bots and all the frontends and all this stuff started using us and uh so like our you know we found really good product market fit but uh it's not really like what the original tech was built out for. So we kind of started going back to the drawing board. also taking a lot of the lessons that we've learned over the past like year or two from like basically the start of like this like crazy memecoin trading and uh just like trading uh season that we had throughout 2024 and early 2025 kind of started thinking more like if we

could do more uh we have a a lot of smart engineers on our team how can we make Salana better how can we make it more efficient how can we make it transparent and uh the output of all that was basically the creation of BAM. So uh I think I wrote their white paper for it in like December and started kind of pitching it internally and uh basically it uh it took some inspiration from flashbots builderet which is similar idea basically running block building inside of trusted execution environments in uh kind of added our own kind of like twist to it some features that people commonly request. So, uh,

essentially it's like sequencing the transactions inside of a TE. For those that aren't familiar with TEES, they're kind of like a secret vault. So, you can't see what's going in or out of them. Uh, you can't peak peak what's going on inside. And so, uh, this provides a lot of privacy. Um, they also offer verifiability, too. So, u, at some point in the future later this year, the BAM source code will be public. So you'll be able to go on like the BAM repo, see like, oh, this uh this code is, you know, this hash, and then you can go to the BAM node and check to make sure that it's the same hash so you know what the code's running. And the goal of

that is like get market makers and people like trusting Salana infra a lot more. Um, there's been a lot of games that are kind of being played on Salana and we want to uh do what we can to try to like mitigate those as much as possible. Um, so yeah, that's like kind of BAM at a high level. I'm sure we'll dive into more details and I could keep yapping, but uh figure that's a good stopping point for now. Yeah, know that that's a super helpful uh context layer. If you're let's say I'm just like, you know, the average retail guy uh trading on Jupiter, like what does this mean for me? Uh what does it mean for my transaction? And and then ultimately just my experience on Salana.

Yeah. So yeah, hopefully your transaction will like land faster and uh you know we're spending a lot of time at the networking layer to make that happen. So uh your transaction will get processed faster. Um and then I think a lot of the sandwiching will be mitigated as well. So there's a lot of privacy that's involved in this. So uh basically you'll send it to BAM, it will sequence it and then send it to the validator and then the BAM node creates a uh like an audit trail of what the sequencing was.

So then you'll be able to like make sure that things are kind of working as expected. Um a big part of this is just plugins, right? And applications being able to like custom logic. Like could you walk us through some of the examples that you're excited about? like web developers could do with plugins. Yeah. So, I think this is really cool and this as far as I know like no other network has this or they some networks might be like opinionated and kind of enshrine certain things where this is kind of like the goal of this is to be agnostic. But, um I think one really cool one is the just in time oracle updates. So, uh,

like, uh, we talk about this in the blog, but Pith has like 1,700, 18800 price feeds or something like that. They can't send all those on chain like every slot. Like, I want to trade gold or I want to trade Apple or I want to trade Bitcoin. Like, they can't just keep uploading all these oracles on chain. That would take a lot of money. It take up a lot of compute. Um, so one really cool idea that we have is uh just in time oracle updates. So basically um when a user sends a transaction that tries to read from an Oracle account uh the BAM node can basically stick the Oracle update in front of their transaction.

Um yeah, I think there's like this negative connotation around uh when people think of like front running, they think of like negative impact uh for users, but I think this is like one of the cool use cases where it actually can help give users essentially the most accurate price. Um so I think this is like a really cool thing. Obviously the application controlled execution or uh also known as ass um that will be really cool. I think that uh there's some applications that want a little more control over the way that their transactions are sequenced and right now they're essentially stuck with like the priority fee sequencing that Salana

does. Um but also like there's two there's two schedulers and agave that you can choose from. Fire dancer has like three or four. So you don't really know what the like market structure is going to be on your application. So application controlled execution the goal there is to uh if you want to if you want to provide like a speed boost to makers or a speed bump for takers or uh you want to put like your Oracle updates uh like just in time stuff like that like it gives uh applications the option to do that. So um I think that'll be cool too. Um there's a bunch of other stuff that

people are going to build with this. looking forward to see what people build. But uh you don't applications don't have to build this. Uh like your application will continue working fine. The goal is like if you want to uh if you don't want to build in this that's fine. Uh you can keep you know we'll have a uh default scheduler that will do like the priority fee sorting. You'll get the verifiability and trans transparency from that. And then we're still trying to figure out like how do the plugins kind of slot into that. Um, so yeah, this is like a community project. We want to get people involved and there's like some questions to think through on how all this is implemented, how to uh like how do you avoid bad

plugins from being added, stuff like that. So uh yeah, we need a lot of help on this. And the goal is to like get the community to work on this together. couple things I want to tie together there. So in the you know how does this help me as a user example you mentioned the ability to sort of mitigate or prevent uh front running and then that ties into uh the Oracle update example you gave as well right where you can actually use that ability uh to to benefit the user. Uh so my question is like where does the trust shift to? Right. So there's the BAM nodes and then the BAM validators. One, can you just kind of like draw circles around what each of those two groups are? Uh but then two, like who do I actually need to trust to execute that in a way that is

beneficial to me and not uh you know hurting me or harming me. Yeah. So there's the BAM nodes uh on like day one. Those will just be run by GTO labs. So that uh you know I think we're trusted infrastructure provider. We want to make sure we get the network stood up that we can iterate fast. Shortly after um there will likely be a permission operator set for those. Um we can't tees are pretty good. They're not perfect. Um, I think they'll continue to get better, but uh, like, you know, you can't have people like running this in a computer lab trying to like glitch the memory and stuff like that. Um, so we

need to make sure that there's a some security assumptions around like who's running this and everything. Um, luckily it's like these things aren't going to be holding a lot of money. They're just like signatures and things like that. So uh if it does if there is some security thing that happens it's not the end of the world. Uh there won't be a lot of money lost. Um but yeah there's like the BAM node and then there's a BAM validator client. So uh there's some trust assumptions with the validator client but um BAM essentially creates an audit trail of what was forwarded to the validator and the sequencing. So it's like I sent transactions ABC to Dan at this slot and

I expect them to execute in that order and so you'll be able to go back like users applications validators will be able to go back and look to see uh if the validator followed through on that. Salana doesn't have commit reveal. I don't think it is like super IBRL compatible. So uh there will be some trust assumptions there but there will be very good attribution. The the challenge that we run into today and uh you guys are I think you're pretty in the weeds on like the ghost sandwich me data is like it's really hard to trace down what's happening and where these things are coming from and like does is

there a leak in the transaction supply chain? Does a validator have a side channel? um it's just really hard to track that down. But with BAM, it'll be super obvious on what's going on because it will essentially know what is supposed to go on the block. Okay, so the BAM node will basically you know have this receipt that says here are the transactions that I gave the validator and then let's say it was like one two three and if the validator executes them you know 132 uh that'll just that mismatch will be super obvious and that's basically uh the the way we can identify you know which validators are misbehaving relative to the fact that we are trusting the the BAM nodes uh but they are running te so we know like there's at least some layer of security there

yeah and like you don't just you can't just trust that people say that they're running the TEU. So, there's like the attestations need to be built out and the logging and stuff like that. That's stuff that we're working on pretty soon here. Um, but yeah, like there will be a way to get the attestations and make sure that the the BAM nodes running what it says it's running. Maybe going back to um like sort of feature unlocks for users or for apps, I wanted to pull a question from chat here. uh I think maybe somewhat relevant relevant after this upgrade has there's been a lot of talk about prioritizing cancels for like per exchanges on Salana for example as being an issue.

Is this something where you know perexes might be able to leverage this you know BAM as a solution to that to that problem today? Yeah, for sure. There's a few like ways that the applications can do this inside BAM and we're not necessarily making uh we're not trying to make like wide decisions for every application like this application needs to prioritize cancels but the features are there so apps can do it if they want. So yeah, I think that'll be cool because um there's kind of it's really hard on these like general purpose L1s that

don't have some form of controlled sequencing for the order books to quote super tightly. And when you have tight quotes, people get better prices. You can trade more volume essentially buy leg uh larger clips without um eating that much price impact. But uh it's really hard to do that because you just have people that are they're just waiting for one sale quote like one transaction to not land or uh for like whoever is market making to be slow one time and they can just pay a super high fee and uh arbitrage it and make a bunch of money and the market maker loses money. So yeah, I think looking forward

to see applications kind of experiment with that and see if they can like try to control their market structure a little more. But there's still all the other like stuff going on that uh you know applications will continue to function as normal if they're not opted into this. And I don't think like every application needs to build in here like there's kind of a small subset of applications that make sense in here. Right. So, not not every application is going to touch this. Like there's clearly there there are use cases for oracles and and clo and um prioritizing cancels. Um but like if you were to have some predictions around market structure

given what you guys have built, I'm curious there like who do you think who do you see leaning into this the most? Um, and perhaps what are some risks? Like what are some surprises of like prioritizing cancels that we might run into? Yeah. Um, I think we'll see like I think Drift tweeted something about leveraging this. There's a few others in the announcement and then there's some that are behind the scenes that are pretty excited about it uh that I can't really say yet. um they'll say something when the time's right, but um yeah, I think they'll be building some features in here just to give users better outcomes. The goal is to like create a lot of value, give

users good outcome, provide a lot of transparency, and um I don't know, I'm not like I'm definitely not a uh I don't have like trifi experience. I've kind of picked up on what other people have said and like thought about a little bit myself. The goal like I don't know what will happen with market structure on these apps. There's a question of like how much of a speed bump should you have for takers? Some people question like should you even do the speed bump at all? Should it just be pure FIFO? that's like pretty hard in these systems just because of uh like you got to pay fees and there's only you you have a supply

and demand curve uh that you need to hit. But um yeah, I I just hope to see that people can like iterate on it fast. So uh you know, maybe they start out with 40 millisecond speed bump for takers or maybe they prioritize oracles uh for users or you know, just being able to add more flexibility I think will be good. And I'm sure some of them will make mistakes or they'll tune it. The goal is to iterate fast. Like we just have to iterate fast on this. So with something like a speed bump or a prioritized uh cancel, right? That's like saying this transaction should be executed with like this set of rules around it. How does that actually um kind of get in plugged into the pipeline

of the transaction execution? Right? Because if you do, you know, uh, you know, the priority fee ordering and then you have to inject the, uh, maybe plug in set of rules on top of that, like how does that actually work in practice? That's a good question. Um, we're still trying to flush it out. There's like we want it to be robust and we want uh, we want to provide like some predictability for these apps as well and like pricing and things like that. So, um, I think we'll be opening up a forum soon and we'll start to throw down how we think about this and then there's a few people that behind the scenes have volunteered to help us think this through. Um, so we

want to make sure that we design something reasonable. Um, so I guess like I'm okay saying that we don't necessarily have all the answers right now, but the goal is to work with others to get those answers. like we don't want to be uh in charge of making the all the the uh decisions on this. Okay. So that would you would you kind of would is it fair to say like that's probably one of the bigger unsolved questions at the moment then? Yeah, for sure. I think it uh if you have like one or two plugins then it's like a maybe a little more easier to reason about. if you start to get a lot, it's more difficult and then depending on what the plugins are doing, it gets a

little trickier and then you have like the composability between plugins. Um, so uh I think there's like a lot of these questions come up in application specific sequencing not in this particular instantiation of it but like the MCP instantiation and other instantiations is like how do apps compose with each other? Um, so yeah, that's something we're trying to figure out. That's definitely a big question. Uh, there will probably be a few like really simple ones just to prove the idea out at first and then as we think through this, we'll start to uh get a lot more going.

Got it. Maybe to maybe to zoom out a little bit. I know uh our our loyal friend Picacio has been taking some questions from the the research telegram group. Um one one ask here was around how would we think about BAM in the context of like Salana network extensions given you know the ability to have like custom execution logic here. Is this is there like a trade-off here or is are we thinking more like this is almost in place of um what traditionally was thought of as going down a a network extension route.

The goal is definitely to get it in place as going down the network extension route. I would like everything to happen on Salon L1. it's just way easier as like an app builder to do that. Um I think that this enables one of it, one piece of it and then there's also just getting the one challenge with doing it all in L1 is you still have a lot of other applications that are running. So like you might get a little form a little controller with the sequencing but there's still contention with like the global block space on Salana but Anza Fire Dancer and other core engineers have been spending a lot of time trying to get the compute units up. So I think we're at 50 million now

there'll be a bump to 60 pretty soon. I think it was just activated on test net. So that's a uh 20% increase. And then hopefully later this year we'll see an increase to 100 million compute units and I think that will just keep going and going. So uh um at some point we'll just keep throwing enough more and more compute at it. And hopefully this is something we can spend some time on too is like figuring out what we can do either in the core codebase or the protocol or BAM to make the compute units larger. But uh that should be less of an issue over time.

So there's like some I'm seeing a copy pasta of a follow-up question here, but it's like it seems like core reasoning here. You know, apps are basically are not necessarily forced off of uh Salana mainet anymore to go and build custom logic for whatever they want to build, right? I think we've seen even some Salana teams, right, exploring um you know, building their own kind of opinionated design um for for some of those reasons, right? to get some specific execution or other other pieces of the pie in there. But it seems like this is really an attempt to kind of mitigate that but also like kind of recapture all that back to Salana essentially.

Yeah. And I think that will be successful. Um we've been talking to some of those apps and uh yeah, I think maybe later this year they'll there will be some that will have some exciting news here. But uh yeah, the goal is to get everyone on Lawan, keep the composability and make Salana the main place. So in the phase roll out that you guys have laid out, you know, one of the uh core objectives is is getting staked uh on boarded. So from the perspective of the validator, like will I make more money by participating in this on day one? Uh what does that look like from for me?

Uh yeah, so the goal is to have a like pretty good base scheduler running on day one. So uh we'll we'll have validators join mainet when we're ready for it. There's still like some tweaking and stuff that's going on behind the scenes to make it performant. Um but yeah, I think it'll be like pretty competitive with the current stuff. the plugins uh those will take there's kind of like a network effect thing with the plugins similar to the goto block engine like we have a lot of experience kind of dealing with this where like when we were at five or 10% of stake it was really hard to get people to send bundles to us just because the other 90%

of the time it doesn't work. So uh there's kind of this like snowball of plugins. So, uh, hopefully we can get like a few like really simple ones, uh, built out to kind of prove out the idea and, um, that will continue to snowball as the amount of stake grows that runs this thing. So, uh, yeah, I think there's like we talked to a lot of validators. Some of them were announced in the, uh, the tweet. Um, I don't have it pulled up right now, but there's been overwhelmingly large uh like positive reception to this, I think, from a few angles. One is like the MEV side of things, like making it super transparent

about what's going on. Um, unifying the execution so that like it's just more efficient inside the validator client. And then, uh, there's a lot of excitement around the plugins too. And like I think uh there's some validators that want to build stuff inside here that they can see being value add to. So uh there's a lot of excitement from validators and I think we'll have testn net soonish and then mainet shortly after and the uh client has been audited and the bug fixes are being uh fixed right now.

Okay. So like as a validator there is a CL specific client that I have to run and is that basically how I can see uh what the BAM nodes are serving me in ter in terms of transactions because I can imagine if I was just running like you like vanilla agave or whatever it may be uh like how would I actually be able to get those transactions. So there there is a specific yeah so we uh took the geo salana client and added some new code to it. So uh it will connect to a bam node. um it will like receive bundles and transactions from the BAM node. It executes them all in a conflict FIFO manner. So uh it'll just kind of uh essentially just FIFO from what it gets from BAM and then BAM's kind of doing the actual like priority fee sequencing and plug-in

stuff. Um so yeah there's a BAM node and or a BAM validator client that you'll need to run and then uh you know it's pretty simple right now but uh right now I think another cool thing will be like right now goes like seven six or seven block engines we want to get to uh we can't have that happening with BAM like we need to distribute this thing way out so there will the u permission operator set people will be able to run hardware in way more regions eventually instead of like when you run the goto client today you're like I want to connect to

the Amsterdam goto node or I want to connect to this but when you have a lot of operators running it you're like do I connect to Helas do I connect to Triton do I connect to Jetto like there's a lot of questions there so the client will just kind of discover the closest BAM node to it and make sure that station matches and then connect to it so Yeah, hopefully we'll get very large global distribution of this system, not just uh the the regions that Jetto is running in. Okay. And the other question that kind of comes to my mind that's in this neighborhood is um let's say I'm just like an arbitrageer and I see an opportunity right you know between sex arb like what what does that look like today right former in the

current uh setup as exists right I just send a bundle bundle over the goto API um my transaction gets processed and executed but what does that change at all or or what does that look like in in going forward yeah so uh there's a few ways to send transactions right now And it's actually kind of complicated for users. Um, so you have like JTO, you have the normal uh I guess you have JTO, you have like stakeway QoS, you have non-stakeway QoS, you have all the RPC providers, you have companies that wrap JTO and stakeway QoS and all RPC providers. It's really confusing. So, uh, ideally this thing is just like super transparent. you just

send the transaction to the network and uh it gets processed. So um it will just appear like transparently on the network. You won't even know it's there. And does this like deprecate the need for tips going forward then and everything will just be running priority fee sorted? Um there will be there's no like deprecation right now but that's uh you know something might change there in the future.

Okay. So going forward makes sense. Yeah. Interesting. with uh with plugins. I could imagine the applications will build them them themselves and I could also imagine like a world where like people specialized in just building market microstructure plugins like how do you see that evolving? Yeah, I think that's a good point uh or a good question. I think like we don't want you don't really want people developing plugins for applications that they don't own. That's just kind of weird. So, uh there will most likely be uh like the application developer for instance, uh

like uh trying to think of one like Drift like I think Drift will probably build it themselves for the Drift market or uh maybe they contract it out and they like kind of approve it for that specific application. But there's I think there's a lot of other like generic tools and things to build inside BAM that maybe aren't necessarily related to one particular application. So uh like one is uh one that I just like thought of randomly um is uh like having more granular timeouts on your transactions. So right now you can have

like um transactions are valid for 150 blocks. You can do some magic to make it granular for one block by like choosing a super old block hash which is what some traders do. But uh maybe there's a way to make it even more granular by using data inside BAM. So uh you can make an order last for like 20 milliseconds or 50 milliseconds. Um, so there's like other stuff like that where like you can probably just enable it for every application and it's okay.

What about the um what about the language on um on sandwiching? It seems like it's going to be much harder or impossible to do that. And so one of you could just explain that and then as a followup why hasn't then that been implemented like in any other blockchain ecosystem because I feel like every like sandwiching is a problem everywhere. Yeah. Uh sandwiching is really tricky. I think it'll this will mitigate a lot of it. it's going to be super hard to get rid of 100% of it um just because it's a constantly evolving game and it just ends up being cat mouse. But BAM gives a few tools to combat this. One is that

it's running inside the TES so like the the person that re like the BAM node that receives a transaction can't actually see what's going on there. And then two is it creates an audit trail. So um basically the attribution will be super clear on what happened. So it's like hey uh Dan sent this like radium swap to the BAM node that got forwarded to the validator and then you can see uh if the validator is like adding transactions around that. So um it will just provide a lot more transparency on what's going on. But let's say like let's say uh we see that a validator was misbehaving right whoever may be

um what's the what's the recourse right so we like can positively ID this we have with strong confidence that hey this validator is front running like then what happens yeah um I think there's a few routes here something we're figuring out there's pros and cons to everything um don't take any of this as like the path forward here but I guess just spitballing like ones you can like kick them out of the ban system. Um, two is like uh you can require them to deposit collateral to get connected the and then they get slashed. The downside of that is that um that increases the barrier a lot for new validators. So that's not the best route. Uh maybe you disconnect

them for a short period of time. Maybe you can uh there's a lot of tools like uh there's like the Jetto don't front Triton one release one that's like it will uh prevent uh transactions from landing on sandwiching validators or it'll it'll automatically fail the transaction if it lands on a sandwiching validator. So they can't take advantage of it because it just fails. So maybe there's like a way for maybe you leave them connected to the BAM system, but you can provide users the ability to control where transactions go. Um, and that's just like built into BAM. Um, so yeah, I think we're like this is another thing that we're still brainstorming on and like look forward to getting

people's feedback on because there's a few options here, but it it just makes it way easier to detect today. Um, I know we're not alone in this, but like a lot of the transaction providers, it's like, "Hey, this transaction got sandwiched. What happened?" It's really hard to know like what happened. Was it like uh cuz like we don't know if they sent it to other transaction providers, like if they send it, did they send it to JTO and like providers ABC? Did one of those providers leak it? Was it a validator side channel? It's just really tricky to understand what's going on.

And so with this, it's like way way way more transparent. And then there's a question of like what do you do if something happens? And that's something we're still figuring out. Yeah, that makes sense. And I guess that that also triggered a follow- on question there is like okay, if the if the recourse is being kicked out of the BAM network, um like what actually happens if I'm a let's say you know all 90 plus% of stake is running uh the uh a BAM validator and I'm in that 10% group. like where do I get my transactions from if everything is moving over the BAM network?

Uh so the BAM node just appears transparently on the network. So like it works similar to the way that kind of the the way that the relayer receives transactions today where you just tell the network like hey send these transactions to this IP and this port instead of my own. So if that is the route that ends up being taken, not saying it is, um like the kicking people out, but uh just kind of thinking it through. Um if that happens, then it would just kind of you would just receive transactions to the validator directly.

Got it. Okay, that makes sense. Yeah. But uh yeah, I mean the the base tech is like almost done. Um there's definitely a lot of like other questions that we need to answer and uh hopefully you guys will help us. Hopefully those people listening will help us and uh there's a there will be ecosystem committee that will kind of be uh involved in like trying to make these decisions. So um we have a lot a lot of good people to help answer these tricky questions.

I guess it's like this is like day one. Uh we're it's like it's cool to announce it. It's a little scary, but uh also it's like all right, we're just getting started. Like there's a lot of lot of work to do. So I'm sure you guys, you know, chatted with the ecosystem participants, whether it be validators, core devs, everything in between. Um for the last months while you were building this and like spitballing ideas and trying to just sharpen the iron and whatnot. Uh, what was the strongest push back against this idea in that process? Was like, you know, I don't I don't necessarily need to know who it's from, but like what was like the thing that maybe made you rethink if this was a good idea or that you really had to is is still like an open question around um the execution of

of BAM? Yeah. Um, yeah, releasing new products, especially something crazy like this, uh, can be a little scary sometimes. Obviously, it's like pretty wellreceived for the most part. You can't make everyone happy. Um, I would say there's one push back is like do applications actually need ACE or the application controlled execution or application specific sequencing. Um, can we just make keep making the network faster and faster, add less jitter, get the jitter down as low as we can, and just have it be FIFO or just have it be like super fast. Um, I would say for

like if you talk to a lot of applications, they want a little more granularity over this. I mean this is why like why did Hyperlid launch its own chain instead of coming onto Salana or why are people launching these L2s that uh they have a little more opinionated sequencing. Um so yeah I would say that like uh it's good I think it will be good to give applications some control over this. Um, so I would say that's kind of like one point where like we can still make BAM faster and faster for everything that's not running inside BAM. Uh, we can continue to do all those things, but for the applications that

want to opt into something fancier, the tools are there to do it. Um, so you kind of get like the best of both worlds. Um, I would say another thing too is like is this just GTO labs going to be running this? And uh, no. Like the goal is to decentralize this and get other operators running it. So that's like pretty important to us. I think uh, at the beginning of this year I wrote a blog post that like we're trying to expand our open source initiatives. We're trying to create new approaches of transaction ordering, privacy, all this stuff. And like I don't want to like just say those things and not do it like we actually are going to do it.

Love that blog post by the way. We need more uh another one coming out later today. Okay. We love that. Um can you walk on that point about decentralization and getting broader market adoption of of this um tech you announced it today obviously still you know b fully putting the uh you know the dotting the eyes and crossing the tees so to speak. Um but it's still closed source. Talk about the addition. uh just walk us through the the thought process there of announcing it is closed source but like mentioning you're going to make it closed source in the future like what's the justification there?

Uh we will make it open source in the future. And then like but why like why not today I guess is is the question. Yeah. Um no one's running it today. So I don't think there's like a fast need to do it today. And it lets us iterate a little faster without people like pointing out all these things. the code's still changing a decent amount from like the networking stack. Uh the APIs are going to change um slightly. So um I don't necessarily we need to open source it pretty soon because like it's going to be difficult for plugins to be written if it's not open source, but uh we just want to keep moving. We

don't want to do anything to slow us down on the development. So for now it's closed source. It'll be open source in the future. Got it. Makes sense. And then if I'm keeping the bear hat on here just for the sake of uh conversation, uh it sounds like the surface area for malicious activity uh really does lie in the plugins and if like I build a malicious plugin, it gets approved like how does that actually work in practice? Let's say I you know I wrote my plugin and it's like to sandwich people. Um, do I need the all the nodes to adopt it or is it only like like how does that actually work of saying taking plug-in from idea into production?

Yeah, this is also something we're thinking about too. Uh, there's a few there's a few routes to go down here. Um, one is that you permission plugins. um that isn't the best in my opinion just because it uh like who gets to choose them. Ideally, it's like totally permissionless, but you run into the issue of like people that are uh building malicious plugins or ideally all the transactions stay with inside the TE or if they need to go out, they go out to a different TE that you know is like rock solid and they're not just

running like this meool proxy thing. So, um yeah, we're trying to think through this. Ideally, it's permissionless. Maybe there's a way that like apps can or plugins can like request permissions and some permissions are automatically granted, others are a little more strict. Um, these are a lot of the same issue like I guess we're not necessarily alone in this. Um, the obviously this is like pretty heavily inspired by some pieces of builder and there's some posts on builder that are pretty related to this in uh it's on the builderet forum.

It's like, you know, they they want to run searchers inside a builderet, but uh searchers might need access to external data, but you don't just want them like streaming out all the transactions out of builderet to like sandwich people. So, uh yeah, there's there's some shared kind of questions on like how do you do this permissionlessly? Um, so yeah, I think it'll be like let's uh do it a little more permissioned at first and then uh on when the tech gets better, try to per make it permissionless.

But yeah, like the goal is to not the goal is to improve market structure, create a lot of value, give users the best outcome, not have users get screwed. Um, zooming out like and if you look at the Salena D5 landscape, it seems that uh plugins are most useful like for dexes and per uh in general. Do you expect like over the past few months there's been a trend of prop AMMs popping up and they've been able to provide like better quotes uh for like highly liquid assets like the Soul USD pair. Yeah.

Do you expect like applications like Drift to be able to provide better quotes now like with plugins than prop AMMs and phasing them out like in the future? Um, that's a good question. To be honest, I I don't know a ton about the propm. I don't I know there's like a decent amount of liquidity being routed to them on the majors like Sole SDC. Obviously, there's like Soulfi. I think there's a few more like Goonfi and Ober and some others. Very serious industry.

Yeah. Um I think we'll see what happens. It gives them more tools. Um and maybe we'll see some of those propms build some stuff in here as well. Um, so yeah, we'll kind of see what happens, but um, I don't know if I have any strong opinions on like one type of market winning over another. I mean, the the the gold standards definitely clobs. Um, hopefully we can get there. I think we'll be able to get there.

Yeah, that makes sense. And the other thing is like today I think the Jetto Dow takes a 6% take rate on geotips. like could could you walk us through the economics of this for the DAO like with BAM going forward? Yeah, that's a good point. So, uh another part of the announcement too was that we want to align incentives. So there will be a JIP soon where uh essentially all the revenue from the block engine that's going to go lababs today will go to the DAO and then all the BAM fees will go that like JO earns its share of uh will go to the DAO as well. So today it's like 3% labs 3% protocol. the protocol side, it's got

like a few slices to the pie, but um after that JIP, we're gonna send the 3% that goes from labs to the DAO. And so uh this will align the uh it'll make sure that the token holders are receiving value from this. And I think that's uh that's pretty important for us. Like we've heard a lot about this. Um, a lot of it started out with unis swap, but uh, if you if you talk to other investors, there's always that question of like what is what's the equity versus token split. A lot of it is just like regulatory hangover, I think. Um, in most cases, I would say, but uh, that is like clearing up a

little bit. So, um, there's more flexibility there for sure. So from my perspective, like I'm often lost in the like the technical weeds and I like simplify everything to TPS and people hate me for it, but I still do it because it's all I have. Um what is what does near-term performance improvement look like? And then like perhaps after BAM is up and running and all the kinks are done and the C the CU limit is where you want it to be by end of year and perhaps you know in the context of Fire Dancer and Double Zero like what does that TPS arc look like?

Oh man. Um I need I just need it in a simple bundled number man. That's that's all I need. Yeah, I would say like hopefully we'll be able to increase the TPS by the end of the year like 3x even though and I guess there's some nuance here that I want to talk about real quick. I know you asked for one number but um I'm going to give you more uh because I think it's interesting. Uh there's a lot of like gas golfing going on in Salana right now which we saw a lot of on Ethereum. So in addition to like BAM and improvements in the the block size and all this stuff happening

uh like one there's a there's a higher efficient there's a more efficient token program being deployed to upgrade the current tok program on Salana which will increase it'll save like 10% of CUS on the network. So that'll increase block size. There's a lot of gas golfing to uh make the network faster. So like yeah there's a lot of improvements offchain in the networking and all the other stuff but uh there's a lot of cool stuff happening onchain where people are making their programs more efficient and that will uh you know maybe we'll see triple TPS pretty soon here love TPS very bullish TPS I'm the only one left

Dan did you have something to say I think you're muted you No, no, I just I appreciate the the perspective there on Yeah, it's like we get bigger blocks and more efficient transactions. Uh so you're kind of working in both directions simultaneously, which is nice. I want to um maybe go back to this idea of you know maybe app teams or just teams in general not looking to build their own chain um or like other network extension or L2 solution from the ground up to support you know maybe the limitations of an app on Salana today I know Carlos you've written about this actually um some of the teams looking to do like opinionated designs on like SVM kind of based chain architectures

Um, I don't know. I'd be curious on everyone's thoughts like the number he just threw out there, like if you 3xtps and then you also provide some of this custom um, execution capability via BAM. Like, do we think that's enough? Do we see I mean, I know Lucas, you mentioned like maybe there'll be some teams um, that we'll hear some good news from uh, coming, you know, coming back in a concerted effort on chain with some improvements here. But um, I'm curious everyone else like everyone's thoughts broadly here. Is this uh kind of a come back to soul moment?

Yeah, I guess um if you look like at the SBM landscape like if you for instance the the ellipses labs team is building atlas and the reason for that is that they need opinionated ordering of transactions. You also have like a fogo. You also have like for instance the SATA team deploy their own like network extension with bullet. So we've seen overall like this trend over the past 18 months of some Solana teams like deploying um solutions outside of the L1 because of the restrictions or technical limitations of the L1 and it seems that with BAM there's like this this opens up like the landscape for

these applications to take place on the L1 and I think it speaks volumes to like just um Solana develop ers doing everything in their power just to keep everything in the in the main like L1. I don't know what you guys think. Yeah. I mean, I think that's the play, right? Um it's like I I you know, Hyperlid's the hands down beloved app of the cycle. I think it's really hard to to contest that. Um but, you know, we've seen a lot of changeover in the per landscape from time to time. Um, I think it'd be tough to dethrone Hyperliquid from its just market momentum right now.

Um, they just hit, you know, a new new all-time high in OI, new all-time high in revenue, new all-time high in per volume. Um, so I think that's just worth noting in in the conversation, but it doesn't mean that things can't improve for applications built on Salana. Um, and there's a lot of experimentation you can do with this, right? Maybe it is the speed bump. uh maybe it is just the broader prioritization of cancels. Uh so I I think you can start to get some of that experimentation in and iteration that you couldn't previously get on Solana mainet. Um so I I think that's ultimately what matters right and it's also interesting because we saw things work in the other direction where given the constraints yet the significant amount of activity uh you actually saw

builders kind of play within that sandbox if you will. You know, Carlos, you mentioned the prop AMM meta. Um, yeah, they're absolutely dominating volume on majors and large assets more broadly. Like the Trump memecoin, for example, most of the volume is moving over these prop AMMs now. Um, so it's just fascinating to kind of see that dynamic shift. So, it's like if you broaden the scope of the sandbox in in this analogy, um, what else will you see? I think it's a it's a really exciting question. You have something Ryan going mute on

and off over and over. Sorry about that. Technical difficulties on my end. Um I don't know anything. One thing we haven't talked about, this is more broad. This is outside of BAM stuff, but like I'm just curious. I don't know. What do you think about LST markets today on Salana? Are you asking me? Yeah yeah yeah sorry. Um, yeah, I think um it's everyone's launching LSTs now. Um, I think everyone wants a different kind of delegation strategy. a lot of it has flowed into the sex LSTs.

Uh so those TVLs are super high, but if you actually look at like where those are being used on chain, there's basically like no usage of those on chain. Uh we're definitely thinking a lot about actual network effects of LSTs and so I think one of those is the liquidity of them. Can I get in and out with large size onchain and offchain? So, you saw like the there's uh OKX uh Gemini Bullish uh that listed Jetto Soul recently. Uh there's like the Falcon X stuff. Um oh, I guess I mentioned I think I mentioned

OKX. Um there's the Falcon X stuff. So, like uh you know, I want to make a trade. I use I normally use soul as collateral like why not use goto soul. Um, so, uh, yeah, we think a lot about those like actual network effects and liquidity and, uh, that kind of sets us apart from some of the other ones, I think. When Judo Soul ETF, uh, we'll see. I don't know. I can't really comment on anything there other than like some of the public stuff we've been doing. So uh you know going to the SEC to the crypto task force and uh I

think we were mentioned in some of the ETF issue u filings or liquid staking tokens are mentioned in general. So uh definitely thinking about it and spending time there. But that's all I'll say about that. Lucas, I'm wondering um I we talked bam we talked back to LSTs as well. I'm curious like you know in the road map for BAM there's this goal to get to an open source like kind of more permissionless decentralized setup there at some point here. Um, how like when you get to that point in the future, what does like where and how does JTO continue to kind of differentiate within

like the ecosystem going forward? Because I guess once you get to this like more decentralized um setup, you know, your role particularly within that um system is, you know, less impactful, less less centralized by nature. So, I'm curious if if you guys have your sight set even further down the road then. Yeah. Um, we'll continue to invest a lot of resources into it, whether it's uh making BAM faster like we want it to be as fast or faster than anything else that exists supporting people with like developer docs and SDKs and stuff like that. And then um yeah, we are always I love building products. Everyone loves

Edgito loves building products. We do a lot of zero to one stuff. This is like another large zero to one thing and so we'll continue making new products that are uh Salana based and uh continue iterating there. But yeah, I think a lot of it is uh hopefully there's a lot of plugins and like things that people built that are useful here and that uh uh that it provides a lot of value for people even if it's decentralized. Awesome. Love that. Well, I think uh with that forward looking excitement uh I think we've kind of wrapped we've covered all possible angles. I know

Lucas, you mentioned we should expect maybe another blog post coming later today tomorrow. Um, yeah, hopefully within the next 24 hours. Got a lot of uh stuff going on today, but trying to get it out. Awesome. Well, I I won't rush you anymore on that, but certainly for the audience out there, keep your eyes out for GTO or Lucas' or whoever's um next blog post, more details on BAM. Um, and thanks for coming on today, Lucas, and chatting all things BAM. think uh this was a great great conversation. Glad we were able to kind of dig into the weeds a little bit. Um we were sort of all over the board, but you know, I think everybody got their questions answered, so it worked out. And uh thank you Dan Carlos Ryan for coming on of course um

as usual as always. Um for everyone else out there, we will see you again next time. We will be live again on Friday, so we will catch you then. Um but until then, it's uh until next time. Awesome. Cheers, Lucas. See you, Lucas. Thanks man. Thanks guys. [Music] Data.

[Music] [Music] [Music] [Music] [Music]