To Solve or Not to Solve, That is The Question!

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:11

   um but this is actually going to be a really exciting panel because um I have I don't know a bunch of different topics that have been boiling at the back of my head about spping uh for for quite a few months and I gave both uh Connor and Hart a grill down when I went to to assemble about um about state of solving and um but I what I tried to call Atomic Bridges and you didn't go with it so

2. 0:36

   that's my first question why didn't you use Atomic Bridge why why' you go for intense Bridge uh intense is the the hype word that I think and and like I think intense emulates atomac City I don't think it is atomic so I think it's maybe slightly more accurate I think intense was the right choice in the end but uh but I don't know I think Atomic is nice because it makes people feel like it's more fast um I would love to hear a little bit about uh Conex um and

3. 1:07

   it's a program that's been around for a very long time and has had this vision of different blockchains for a long time I want to hear how has things evolved as you know the the space has exploded with different rollups how different trans have going to adoption how have you guys sort of adopted your adapted your your prodct way for that yeah so with connect we started like many many years ago with um basically like State Channel Tech and

4. 1:33

   we kind of uh similarly to AC cross like adopted this idea of Opp based bridin and uh considered that as like the Forefront of bridin Technology um after hearing Hearts talk about like settlement layer we're actually in a similar kind of mindset where that is a very important layer for just to be in um and liquidity is kind of just like um another like helpful thing on top on top right but the challenge is that like sourcing liquidity is really difficult

5. 2:06

   um it's much easier in a bull cycle when you have money FL into the system but it's also super capit level efficient so uh right now like that system works well and for end users the experience is really nice but it doesn't work in a system or in in a future thousands like not even a dozen or two dozen it doesn't really scale to that degree so we are and the reason is because you need to have liquidity on every single Ro up to go the fixel the transaction size that's going through yeah not just like between every different chain it's every

6. 2:39

   different asset within the chain so you can see how that just like explodes and the number of different like denominations of currency that you need so we're also kind of like flexing our model to discover what this attendo baseb looks like and actually connects right now is already kind of like an intent based protocol I mean everything kind of falls under intents right now um and we have system of solvers albe it they are um they're they're a very small

7. 3:04

   slice of the solving domain like Solutions are basically confident and what conect does is we have Network solved they do a very small slice of these solutions they just provide like the quity and the paths so we're trying to think of like okay is there a way that we can expand like what we do well here um and that's something that we have like kind of cooking under under the hood so there's probably a little bit more to that okay

8. 3:34

   great um we so we have two you know members representing the solver crew um I will sort of flip the question to you guys right like there's this problem having inventory on all the different chains and a lot of this interoperability stuff is sort of a Inventory management problem you know where do you put your inventory to be able to F fill the commands um how do you guys think about where do you

9. 3:59

   position inventory are you how do you like Source the cost of it and just what is the integration of a new chain actually look like um no that's something that we've been thinking about internally a lot um like as you right said this is not a very scalable model where we can think about holding liquidity on six chains that are that have the most s of flow but then going to a thousand chains maintaining liquidity everywhere is Not Practical

10. 4:22

    and it's not something that we want the status qu to be um so that said our decision to add a new chain is always Reliant of the order flow that's coming in from that but um yeah currently we are in the middle of a few experiments to determine if there is a way to integrate within existing Market maker auctions to perhap Source liquidity on these destination chains so folding liquidity and inventory management isn't

11. 4:46

    something that we focus on as much as we focus on scaping and removing the need of having to hold liquidity on each of these chains and if we can just swap our way to it I think that's what we work towards and that's how we thinking about yeah we we don't know hold the inventory uh we don't really want to take on that risk we just want to abtract all the actions and find the optimal paths we just offset this to market makers I do

12. 5:12

    think maybe six seven months ago solver started envisioning combining market makers R inventory and being a sober for the attend efference now for me it's very clear the distinction solvers are graphing the route finding the optimal paths finding the edges market makers this pried to lot of market makers are plugging into a cash flow they're aggregating other market makers or plugging directly into market makers to get this Edge as well so I don't ever see them combining I think people

13. 5:42

    experimented it 78 months ago and now it's clear separation but what does that interface between the market makers and the software look like this why hash exists because they just aggregate all of it so it's one unified API you can plug into and then they Source the best quotes a lot of market makers have different API so that the just need to integrate every single one and then you've got people like na them also having token lamps the background and so

14. 6:08

    forth and does does that allow you to have the inventory on the target chain that you're looking for they have the inventory and then you can like add C data at the end so you can append let's say steps or prepend what you'd like to do with the Infantry so you don't need to have exposure on any trains you just build call data as to this so I know thecross has also its own sort of liquidity on the target destination

15. 6:30

    um how does that model for differ yeah so this is where thinking about Inventory management actually matters a lot um and so like Eddie's Point here is very real if we have uh a thousand rollups uh with uh 10 assets each it's just that's an impossible Inventory management problem if it takes you 7 days to get your assets on off chance um if it takes you 10 minutes to get your assets on off those chains and that's relatively affordable maybe that's a Sol problem right and so maybe

16. 7:10

    we get there in like CK land ZK roll up land um we're not there yet so okay what to do first first of all do you really need 10 assets I'm actually of the opinion and I'm going to see where you B goes here that the things you need to move between chains are three or four things um mainly and maybe it ends up being like stable coin and something like that so that helps but then the next question is okay well um why don't we separate the risk

17. 7:44

    um there is some segment of of people out there that are passive liquidity providers who would be very happy to provide liquidity to a solver Network in order for them to do their solving and Market making um and I look at just from like a risk risk uh preference perspective um a solar like propeller heads wants are really high they're they're actively using Capital they demand a really high return on their capital and that's their business model

18. 8:14

    um there's other guys like me personally I'm not doing that I'll happily earn 15% on my e or something like that and so then now there's like a a trade here can we design uh something of like uh a cross chain single asset lending Market uh that will lend liquidity to solvers so they can do their solving business charge them a small fee um but basically provides them with the ability to uh

19. 8:44

    take repayments on the Chain where they want um and just to finish this so across basically does this so you can think of like across as LP pools but like what those LP pools are really doing is there like a convenience function to our solvers to allow them to take their repayments where they want money and the thing that's actually kind of cool and a kind of unknown feature is uh a relayer we call them relayer so a

20. 9:10

    solver relayer on a cross um can fill a user going from arbitrum to optimism but actually take the repayment on like base or on ethereum on a third chain um and then we we price that dynamically so those pricing changes based on like utilization of that but but this is under the assumption solvers generate revenue and they generate large amounts of fees so the percentage they need to pay should be very very small like the whole purpose of the competitions particularly like

21. 9:43

    housewall you're some sers are even losing money to get trcks so having this mechanism paying interest the amount must be very very SM so um I don't have all the economics in front of me but you stealing my fire my last question was who the hell makes money in all this but you know we can skip right there so so I open put a little bit more context right which is like solvers have a very difficult job they don't have inventory they you know they're trying to provide a route and it's a highly competitive game they're

22. 10:16

    quoting against a bunch of other solvers that do like you know searching of routes and it's you're trading off like engineering mindset like Talent of being able to PR these algorithms and compute power that put behind these routs to be able to get margins very small margins on the efficiency of your trade so how do you yeah how do you make money off of this how do you like actually you know settle these orders that are coming in from uh from these Bridges while still uh you know running a business I mean

23. 10:48

    just to respond directly just on across here are are real lers make money it's it's like um so a lot of our gas costs are really low and we really do have a very big advantage over our competition on our gas costs which means that we can charge small amounts and they're they're small for Lending assets but that equates to like real abis or the LPS that pass and Rel layers are still able

24. 11:19

    to earn enough that it's profitable the push back is this what if someone fors acros and then reduces the fee that like gives to solvers so that it gets routed through these aggregators well we're going to do that too we're going to end up running like auctions so solers compete on price rather than compete on spe and so like onit's piece from Frontier I actually hadn't really framed it this way we're and across today it is

25. 11:43

    solver speed competition with basically the way it works is um when a user Bridges across suggests a fee um that the the the bridge user agrees to um and then sers raise to earn that fee and so there's some CH suggest fee that the bridge user the bridge user will then sign the orders here's the fee I'm willing to pay right so I go to like a a bridge aggregator I request a quote

26. 12:08

    acost gives me a quote and says oh look it's still cheaper than all the other Bridges but that's the only intend bridge and I want to take this y but let's say there's a future where there's multiple different intent which then you guys will be competing with the other ones on the feed that you charge well we're going to do that too so we're run an auction right so people will come and

27. 12:24

    be like okay now soers is going to compete on price ideally we want them to compete on price and speed we talk about later but they'll compete on price for the best price and now I think that price should compete down to a zero profit margin right like you're supposed to squeeze people but zero profit doesn't mean you make no money it means it's like this the economic concept of

28. 12:47

    zero profit means after you pay your server like your actual costs and your oper like your uh your Edge your IP your your internet like what's left over you don't get to make excess profits and frankly from a bridge or interop perspective that's the way it like should be um and I think that's cool right competition's great user gets a great fill um and sers shouldn't be losing money I don't think that that makes any sense here um but that it

29. 13:18

    should be a hyper competitive game yeah and you know what we SE in all these markets that Doe 10 competition is you have a few members that end up emerging maybe like you know two three different different solvers that either have the scale the economies of scale or the inventory to be able to compete um and the prices was still very efficient you could you know you can still call it a competitive market even though it's an

30. 13:40

    alistic Market um it is still efficient at providing really good quotes to uh to to users I want to I want to flip it back to to KX a little bit and maybe you know we touched on a few things we touched on um Crossing uh sort of coins from one chain to another versus the major we talked about solver competition and profitability um and yeah I wonder what's you know what's your guys respect so yeah actually if you just look at the volume of transactions going through our

31. 14:12

    group specifically it's mostly like uh weighted toward the stats and the major assets right so in a lot of cases it doesn't make sense for us to go out and get put in for of these like share coins um and I think that that actually might go towards uh like supporting and really optimizing that as you place more for commity to use um and I actually I I do have a question for the solvers that people who are building

32. 14:43

    the solvers because I it sounds like to me that cost is not even only mod equation or like it's also like building trust so like the more you know the more solutions you bu the more users you you get R to you the the more you get later so even if you're taking um if if you're making negative revenues like it could still be beneficient right like is that part of the predatory pricing that's what that's but I doubt like that works because in auctions all of isn't Ticky

33. 15:14

    so even if you are you know subsidizing uh cost right now to win off no doesn't guarantee that you win the next auction right it's not like user acquisition I mean even when you acquire user to sell at risk of them leaving right it's acquire more in 10 memb that you can be a Sol on and more people to consume your API so if you're playing the game that you lose money in every single trade and you subsidize it when the token rewards

34. 15:38

    which a lot of are doing you get more order FL from different athletes and then you just play this game who can subsidize the most who may not exist in sense ofs so there is a benefit of doing it so on the Soler side you know what do you think the competition involved into what do you think the The Edge is long I think the edge is um not routing I think that's fine we can do yeah we can all comp pricing who has the most price optimization but if we want intent to succeed it's more about what actions

35. 16:13

    can you do in different CHS so intens me abstract all of this logic way and merely what we're doing right now with the tens swaping bridge and I think Sol out in the future would be how much exposure do you have on different chains for different actions of D5 or nft or whatever it may and then offering L to 10s yeah I would definitely agree and I think there is some Edge that you can build on the kind of features that you offer for example one of the things that we about to launch in the coule of days

36. 16:44

    is uh this intent based API where a can just plug into one API it kind of abstracts a v gas for the user approvals that kind of one and you don't need to implement 47 for it so now we find creative way they're offering different feat try toing usually that's that needs better you experience so are are sers always sort of competing for orderflow for access to orderflow differentiated orderflow that others will have access to like we we specialize in Def flight rad being able to R here balance witha whatever it may

37. 17:17

    be as where we specialize we do the swaps and but other people are p in us at SS so that's our is focus on there's only one other C SE so when that does this and but we get volume you know we dois J got s x we got all that osis P so when it when it comes to this orderflow right and sort of capturing order flow there's always sort of this incentive to say like or I'm going to aggregate all the other aggregators like I'm just going to go one one level above and then I'm going you know going to have more Integrations and I'm going to

38. 17:52

    be able to like propagate more um where does that end I mean if we do Define a standard for let's say intense and let's make let's go simple and say we're just doing cross-chain limit orders right now like we just doing swaps and if there was a St the way I mentally think of it is it's it's not complicated it's like an order ticket the user has an order ticket they're like input asset amount output asset amount minimum amount right

39. 18:23

    um and like input chain and outut chain right and then they just sign that and if we had something like that as a standard and the user signs that and then it gets thrown into prob all of these systems right but it get but like that is what's going to be the best execution for the user um it will be ridiculously competitive at the solver there for just those types of crute swaps which like is where the end State should be I don't think I think we shouldas that and then like there isn't

40. 18:56

    this kind of like agregation of all these fragmented different apis it's just a here's the user order ticket let's go fill it um and then I think the question goes curious to hear think is then how do we extend that from just a cross chain limit order to something maybe a little bit more interesting um too do you envision these let's say you create the standard and it imp plugs into all intent networks do you envision

41. 19:20

    these intent networks also subsidizing and their need of token to also get the hard FL to get more Surplus so the solers fing this right now right on the rewards they get certain protocols do you see 10 networks doing this cuz any also about this whole layer of those networks also competing and the solvers are not going to plug into every single Network because if you integrate with every single solver Network as a solver

42. 19:44

    or int Network it takes you weeks right CU every protocol upgrades and so forth so you're going to get these buckets of solware supporting certain networks and they inves 10 yeah rewarding so the need of Tok so the way I see it and I actually think Conor there's probably a lot to dig into with video on this so I maybe we do this more please please we have 10 minutes left we have to fill it

43. 20:08

    so let's go well the way I see this like let's go back to this mental model this like order ticket that the user signed um and they literally like sign with their private um so where does that order ticket go I imagine it's um going into different RFQ systems at the at the top layer and so like you can swap back to an RFQ system and like uh crossw an RFQ system and then increasingly I think

44. 20:32

    applications themselves have an RFQ system um and how shared they are open question whatever but um and and at each of those RFQ systems there is an opportunity to uh take a cut for uh for the owner of that owner flow uh so like unol X charge the fee to now um from a solver perspective every one of those RFQ systems is going to want every solver possible listening to one and if you have a standard order ticket it should be zero integration work for your

45. 21:05

    solvers to listen to all those systems okay so that's like your top layer and then your middle layer all the solers are listing systems now competing and they're kind of in ends up needing to be like a standard for how these RFQ systems operate that's where I'm not totally sure that will evolve like I could see the different there being differences there and how they work but I don't know that's what and then the bottom layer right is in this Crossing world at least the way I see it these

46. 21:34

    user assets are esro um until the intent is fulfilled and verified is fulfilled so then there's another need for a settlement ler and I think there can be competition at that s layer there can be different ways of doing that esro and then verification of the intent and that's where like across's vision is we want to just be a really really efficient secure competitive settlement

47. 22:00

    layer that does have some economies of scale um and so to answer your question like so this if there is a standard order ticket um perhaps we have fairly standardized auctions at the RFQ level and then we have a sort of standard for what the settlement layer looks like and there's kind of competition could you expand a little bit on what competition of the settlement look like yeah so okay so for

48. 22:25

    example um Unis swap X doesn't we we talk them pretty CL because their design is quite similar they're great yeah yeah sure we love them we we do um but their their design is line their white paper um has uh user assets getting escro on the origin chain on per like each each trade has its own escro basically um and then they get filled on destination chain and then they get optimistically

49. 22:53

    repa on or CH okay so in this first of all there's an inventory management problem they can't get repaid on anywhere other than the origin chair um and you also then have like gas feeds uh of uh repaying the solver um for each transfer what it cross does is we kind of escro them in like a pool concept a kind of a cross chain pool um and that allows us two efficiencies one we can repay uh the solver on any CH Network and two we batch repay that which means

50. 23:28

    we have gas cost just like one uh transfer to the solver every Epoch every period um and so when we talking about this these are two different settlement layers the way I look at it and uh I think that there should be competition I think frankly the user should be able to pick their settlement layer um or be able maybe have a list of settlement layers they are okay with for maybe different trust assumptions and then the

51. 23:53

    solver can go and pick the S layer that works best for them um I'm almost feuring already like the solver goes from just routing to like launching liquidity pools now they're like doing this job of routing these messages through all these RFQ systems and settlement layers are offchain so like the solving job just like goes being just onchain but also being with all these offchain API systems like I was about to ask mentioned this idea of a

52. 24:17

    standard a single unified standard um but if there is value capture to be had at that ler we can probably expect multiple standards competing at the top Air to and then it's just the entire that every single place is an auction it's a the maybe but this is where I think the solers have power too right or like they don't have exactly pricing power but everybody wants the maximum um

53. 24:42

    the maximum offchain liquidity possible there is like that is a positive SU game where everybody like we we all want all the solvers solving every Network so you we we sort of talked about the liquidity fragmentation problem right and like does it turn out that the solution is just to take all the liquidity off chain I mean I I'm some people don't like this but I am totally cool with it because there's permissionless entry to this so one of the things I think people

54. 25:15

    should AC critique is if this all just becomes like hfts um in like traditional Finance that's problematic because you don't have permissionless entry groups Market you have people that just have regulatory barriers and they this is a spicy C okay I like this well but like I this is my take where provided you have you tried to sign up a Sol to one inch but this I think needs to change

55. 25:40

    and I think it will change I think they just we haven't figured out how to do like decentralized auction properly yet I'm not sure if it will change their incentive is a token utility and a narrative that is so nwor there two different parts if it was if if they really wanted true intent water they would have more than five Sol two to I think I I think competition wins like so

56. 26:11

    so I I think that if you can build a better like let's call it a better L inch a better uh consumer facing application and it's better because there is maximum solver competition you will win so so you know I'm sort of saying this tongue and cheep because this like permission nature of being able to to per to allow solvers to receive messages is a nonsol problem um and in many ways it's both the way that you can sort of create

57. 26:41

    these order tool auctions that are more um sort of um P off style models right by saying hey you're not competing over the individual order in the right to fill it but instead you're competing over a seat at the table to be able to see access to all the Overflow okay so that's like a gatekeeping aspect it's not completely permissionless but you're running an auction to be able to access

58. 27:04

    to the flow it's called a p for um but there's another aspect that's sort of an issue that comes in designing these types of auctions which is privacy like how much do you want to leak about the users's intents um or the user's you know desires trades whatever to a public permission set of actors um there's also some downsides on efficiency that come from doing that I think Vlad is right your intentions will be used against you

59. 27:30

    this thing is very clear I mean c swop is fly open so people assume only solers can see the requests every single person can go and see this you can even assume what the solvers are proposing in the auction and in front so this is already kind of open as such being able to execute and also propose RS that's permission but the viewing of it all is open right now and I I can imagine some

60. 27:53

    people who are using users intentions against in 2 weeks we are going to drop some GC research talk just that what what def find we two weeks so like you know it's funny guys cuz like I I used to trade g r bons like a long time ago at go sax and it is an RFQ system like my job was pricing RFQ trades where uh we're etiquette like half of the rules were enforced by etiquette like and so that's not a very good enforcement technique here well social BS are everything well crypto is osed by

61. 28:34

    social BS come on exactly it's a whole other panel right um but um uh I I don't for modest size orders I think seeing that order for is fine I think if I think personally and I want to find some economic research but sure really large trades yeah you want privacy on that but modd trades I think if you have strict and clear rules about how that overflow should be handled that is where you're going to get like efficiency and execution gains and the information

62. 29:06

    leakage of like I want to sell 10 e for usdc it doesn't matter I mean the solers in theory could be very malicious here but is new man we see the order request before inst solers generally as far as I've seen have a good moral compass I am as far as I've seen who know in the next couple of months and they're not using this maliciously because they know other solvers will be able to figure this out and will call them out this happened with some solvers that were playing a

63. 29:38

    little bit malicious on other networks and yeah they just got outed for it so I think there's also that layer of morality and being exposed for what you're doing cuz it's not that you're an Anon seure there's teams behind us there's faces There's real companies I wonder to expected that honestly all right um we're running out of time I wonder any final words about solving in 2024 it's not just swaps it's not just

64. 30:09

    swaps can be all actions on any chain that's what ts are meant to be this baby expand oh to SWS I think that's a 2026 problem we're still just swapping if do swaps today then you you're the best I'm interested in seeing how solvers can actually collaborate in this space like not just on understand but also because like you're all operating on information based Sy like everybody

65. 30:39

    has their own specific information in the back pocket and there are solutions that can't be solved by any single one by itself right so in aggregate like composite Solutions can be possible so I'm interested in see that SP all right big round of applause thank you very much