App Chains vs. dApps on shards: Cosmos and NEAR

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

1. 0:20

   I'm here world gotta lock jinx Kenan everyone's here hope you enjoyed dinner last line in the fusion here I don't know but it's about to start so I think you guys ready the max I'm not going to do a long intro this is sending your cosmos total badass this is Gillian from here also throwback and we're gonna hear them talking about some cool stuff of fun so if you see you number one you have some burning questions hold them because about halfway in on the classical audience

2. 1:13

   somebody's got a little swear yeah yeah you sure you know so this looks weird this is so we can record this for later the audience questions so one question if you have a question you want ask someone will passionately swear you can talk into swear it'll record it so just a preview so you can forget that but if someone gives you a weird swear that's like alright let's get started illy do you want to ask the first

3. 1:38

   question you want me to ask the C question go ahead okay so I'm gonna start with one of the most question on everyone's mind which is if you're an animal I think I've always had a natural affinity for hedgehogs so I think that is my spirit animal yeah we worry about you dragon yeah so hey everyone thanks for coming on very busy night I'm sure in San Francisco blockchain week so the topic of this

4. 2:23

   kind of discussion is apt chains versus taps and this is kind of the discussion we've been having for a bit now which spun off from like the idea of you know what's better cosmos or you know near or is near just part of cosmos or any other part of cosmos or cosmos apart in the area or cosmos is near we can figure this out but I mean the kind of where's this is coming from is this idea that we're like if I'm a developer right I'm want to build some application which

5. 3:00

   platform should I choose right really this is kind of a big question that a lot of people are asking and let's just start like what's a good arguments for choosing an app chain or like a cosmos chain cosmos zone okay so I guess in my mind if I guess the first question to ask is like in my mind what is cosmos like how do you how do you know like what is a cosmos chain so the core idea of cosmos is you know building tender

6. 3:32

   mint was a means to an end building the cosmos SDK was a means to an end the end is this thing that we're demoing to either tomorrow or the day after at the defy hackathon called IBC inter blockchain communication and inner ball chain communication is an idea that goes back really to the 1990s where the first were like Mark Miller all the founders of what is now a gorrik we're working under like Xerox and a variety of different other entities

7. 4:05

   startup companies etc on this idea of smart contracts as a synchronous communication between sovereign systems and I found this work you know on the internet I didn't know Marc or any of the Agora people at the time but like it came out of my frustration with the whole sharding conversation like basically in 2016 I'm gonna like we have time I'm just gonna like do the whole story

8. 4:37

   I feel like I have to frame the whole story it like it it has to make sense like why cosmos is the way it is so we'll do the whole story so 2016 I'm sitting across a desk every day I'm doing a like Enterprise bot trans startup sitting across get a desk from Dominic Williams who's working on Definity there's kind of a side project for him at the time and every day at lunch we would argue about charting then I would go on the internet and argue with the Potala can flat about sharding I talked to Jay talked to Jay about

9. 5:09

   tenorman and I got real frustrated by what alex has I think actually branded really nicely the adaptive corruption problem but it was a problem that like really frustrated me in 2016 so I was like real sad about it and I'm like because basically what happens with the adaptive corruption problems if you build a sharded system the whole system if one shard gets taken over by an evil

10. 5:35

    set of validators and they do an invalid state transition and it gets recorded to the rest of the system there's not really any completely sane way to I'd go and like fix the system you kinda have to like stop the entire universe and fix it again so I found this word on the Internet at this like super 90s website called I rights org it's still up it's the most 90s website ever it's full of brilliant ideas and I'm like I don't

11. 6:04

    know how to adapt these ideas to both channels but I think they're really cool and they basically have a framework where if evil validators take over don't know you know if you end up interacting with an evil computer in their system you can actually figure out what that evil computer did and like manage the damage so I got really excited about this I I started talking to Jane Ethan about it that idea turned into cosmos we realized that in order to build this

12. 6:34

    idea you needed consensus and a and a machine for building ball chains shall we spent you know basically the last two issuers building that and then once we built that we could actually implement IBC along the way all of the people whose ideas that this originally was the started a company called a gorrik and we have the privilege of working with the original creators of the idea to deliver

13. 6:57

    IBC and IBC now is the sort of formative thing of Cosmos but what when you say an app chain you see I think we have like two really good canonical examples of them one is a chain actually we have three I would say we have the by Nance chain finance Dex chain probably like the most heavily used highly transact upon value carrying cosmos network cosmos SDK tender bit network in the world we have the terrible chain which is probably the ball chain that is most

14. 7:35

    used for non speculative applications in the world it's used for for payments and you have to ask the question would either of those chains and Libra is kind of also like it's like alt cosmos like take all of my idea all of the ideas I've been working on five years till I get out the door and they like or like will hire 40 Facebook engineers like we make them cool thank you it was free so the so like those are the kinds of things that I think of as when

15. 8:10

    we talk about app chains and I think the best argument for like why app chains will always exist is there's fundamental social reasons why a company like an organization like finance an or like Terra would want total control of the environment but they're operating it they want a blockchain but they want their community they're aligned self-interest to fully control the system and I think that is probably why

16. 8:38

    I think IBC and app chains are a big part of the future cool yeah that's a good story I mean one thing that's like the way I like to portray the kind of difference between apps change versus like building and half on you know a stadium or some other smart platform is that spinning up your chain is actually a lot of work and that's actually the the reason why it makes sense for this

17. 9:10

    companies is because they have resources to do this work and in case of Finance like they are like the most of the work is integrating with themselves so yeah and by strategic integrate in the same way with terror most the work is go to market and like the blockchain is just like a real-time gross settlement system for them yeah so yeah if I'm a developer right and I'm trying to spin out a product the amount of work you need to do to actually get from I have the

18. 9:40

    standard like cosmos SDK to this cosmos SDK is accepted everywhere it's in everybody's wallet and it's on exchanges and people can use it that's a lot of work right yeah but maybe IBC makes that easier let's talk about it so that's the kind of flip side of this right if you just deploy an app on helium you kind of get like a lot of it for free in a way because it's you working in the same

19. 10:04

    common environment so kind of the idea I use this pretty much building an app change like building up your own data center or building up your own server rack right it's a lot of work you need you know buy the hardware set it up you know configure hook it all together you get some a lot of like value out of it because you have full control you know you can install some better networking equipment etc or you can go on like launch an app in the cloud and kind of

20. 10:29

    you know get maybe less performance you know a little bit higher price but you can do it today right now from your computer so that's kind of the way I like to look at it and obviously like there is a reason why I like you know Google Facebook would not use Amazon to you know to run all of their computation there because they like way too big and you know at some scale it actually makes sense to set up your own infrastructure and actually do all this work but the good examples like

21. 11:01

    Netflix which kind of grown up with Amazon and even though probably now it would totally make sense for them to have their own infrastructure it is embedded into echo system so much that it actually will be really hard to integrate so I think like we have the discussion before that you said like for some applications like maker it would actually make sense to move out of each idiom and spins their own chain because they actually not leveraging this idiom

22. 11:27

    as much as as they could have or like as well I mean like yeah well they're paying a lot to use a theory em they're not getting additional security theory em they are all well they and things that are functionally competitors to them have become like essentially and leyland and has seve had a great paper about this I think that came up yesterday or this morning I saw it on Twitter this morning about how I effectively all of these applications have become de facto the governance

23. 12:00

    layer of aetherium so okay I want to like let's like try and systematize this this conversation because on the whole I think we probably agree with each other I think here is a great project I think building sharded ball chains is awesome and I'm super excited that it works I just I I think we are working towards like two pieces or many there are many pieces that we have to get right to build a global crypto economy I think Mir is gonna be super happy that IBC

24. 12:35

    exists so they can connect to the finance decks and get liquidity and assets from there to interact with and like bootstrap liquidity under I think that'll be cool and why basically what IBC should ideally allow for the entire fastball Cheney Co system near as a fastball chain cosmos change our fastball chain pulled out it is a fast block chain B 2.0 a fast ball chain all of these fast ball chains you'll be able to instantly be able to bootstrap

25. 13:03

    liquidity in the same way that you can instantly boot sharp liquidity on aetherium if your ERC 20 via IBC because near can implement IBC they can connect to the by Nance Dex's implementation of IBC or the cosmos hubs implementation of IBC and immediately get access to all the exchanges that have already integrated on the cosmos hub etc so when we think about like what the cosmos vision is really enabling it is enabling it the the core goal is to get most of

26. 13:36

    the as many of the benefits of being on one big integrated chain sort of available to this entire ecosystem of fastball chance so I mean let's actually dig deeper how IBC actually enables it right because from the perspective of let's say I'm an exchange right that would be integrating and let's say I mean by net by nan Stax is one example right place yeah I think I think let's talk about the finance decks and like imagine this world because it's like

27. 14:08

    concrete and like like we I think many of the mostly people in this room have like this like have a mental model of like what near live will look like like when yours live there'll be defy applications they'll be gaming applications digital assets of all kinds will be issued on this on this chain it'll be this rich ecosystem and and you know over time though you know near may get probably will get many exchanges that integrate with it but you also will

28. 14:41

    have this finance decks ecosystem the finance text ecosystem might will have many appealing things it'll have on ramps and off ramps in many jurisdictions you'll have like liquidity on it like so all these things will be really attractive features of the finance tax and so you're sitting you know you're you know video game developer you've issued a bunch of assets on near and you want people to be able to sell them they'll definitely be various decentralized

29. 15:12

    centralized NFT exchanges that correct connect directly denier but your see this like possibly giant pool of liquidity to this by Nance decks so there's an IBC connection between the by Nance decks and near you freeze your NF TM near you send you generate an IBC packet it gets sensitive by Nance decks your NF t gets created on the finance decks people can trade it they can collateralize loans with it they can do all kinds of crazy they can go 100

30. 15:41

    X leverage on your crypto kitty and and that's the world that I've been working towards but at the same time IBC right now connects to ten durman chains right so you still need a peg zone which will pretty much for the year all we need is a light client like that's that's canonically all IBC needs it's just nobody else has written down what how a white client works for any of their consensus protocols because for fast change light fines are really hard

31. 16:12

    I mean we we did it sell was doing it I'm sure there's a way to make an ear work I think that the joke was I think James made it everything that start to see has has a life line of code SLO cosmos sure I mean yeah I I don't I I don't know I we were talking a little bit about how nears consensus works and I haven't figured out exactly how the white clients gonna work but I I suspect

32. 16:51

    there's a way quick interjection this is a moderator question which is how light of a light client right this gets thrown around a lot the idea that you need a light client but I mean I could just make an argument oh this note I'm running is a light client right so can you can you define like how light of the client it would have to be yeah this is a really great question that James brings up many times so in proof

33. 17:16

    of work you can have these super light white clients and James is super well known for having built them for for many protocols and so you it's actually good enough that you can do slow ball change the Slovak so you can actually do Bitcoin into a theory m1 dot o but the in the world of proof of stake and the world of fast ball chains you need to track additional contextual information basically how the validator set changed

34. 17:47

    how do you know if a new block header is valid so ideally you want to be able to track that with as little information as possible and so the canonical smallest amount of information is just like a zero knowledge proof what we do in and what we fall back to on in like the tender mint world and the and and some of the hot stuff designs hot stuff is kind of another contentious protocol

35. 18:17

    that is vaguely like tender mint is if we can if the validator set hasn't changed very much between the last block you saw and the next block you saw you don't need the intervening blocks and though that that mechanism will not as theoretically perfect as a zero knowledge proof that like all the intervening blocks were correct and exists is we think probably good enough so that's what we mean by how wide awake I go but I guess the interesting question still is like you're still only

36. 18:56

    verify inference for like line the validity of the kind of block based on the whatever civil resistance is you don't verify the validity of state transition yeah I mean for IBC you don't need a baby for IBC you don't need the validate and so this is where the problem was like Charlotte systems comes in is that because we have possibilities that some shards go rogue and obviously

37. 19:21

    like it near have a way to contract it but it takes time and may reverse chain that like that actually requires yeah you have to wait longer for an IBC packet you need to wait longer but also you need some kind of like way of still recovering from that type of activity right and the normal way of recovering it is providing fraud proof which actually tells you that that was loud I think we can probably potentially get this kind of thing to work inside of IBC

38. 19:51

    like we could have the ability like there's nothing in IBC that says that like the sending chain can't just like like declare a packet that it previously sent as invalid it's possible the semantics of that require because what IDC does and this is like one of the experiments that is IBC that like if this works will argue in favor of app chains and if it fails argues in favor of shorter chains is that we can push a

39. 20:24

    lot of the responsibility for validating whether or not until like sort of end users and that end user could be like an exchange that just monitors like the source chain of all of things it could be you know the wallet that does it or a wallet provider or a block Explorer like that there are people out there who can who will like keep an eye on these source chains and be like oh something

40. 20:47

    either terrible has happened there or something that is part of the process but like sort of a Black Swan event like a state reversion has happened and sort of make enough of the economy aware of it that gets priced in yeah I mean that's actually like my main argument against like something can be see is that now you're kind of pushing the security on the user side where as a user I need to be aware of the asset I'm trading on finance tax is actually issued by a cryptic itty chain for

41. 21:16

    example which may or may not be valid at this point right so I need to be like aware and kind of know what's happening what's the source of this and ideally be like looking at hey is that chain actually still alive right maybe like although data's quit and that she installed and this cryptic idiot doesn't actually worth anything yeah it's it's really a interesting open question about like how well markets priced this kind of thing

42. 21:41

    at maturity and we may not know for like ten years but like when you say price right that means you're mostly talking about assets that can be like priced by let's just call it like some market forces right if you're talking about user data or some other information which you know has value to me but no but to nobody else now you cannot use that anymore like you cannot use market pricing as a kind of argument for data

43. 22:09

    validity that way yeah let's think a little bit about that so it's like a good example of a use case though I mean the example we like is like social graphs right it's like let's say I have my social graph which is encrypted you know and I for example when I use it in some other application on another chain and like I need to prove that this is my social like some connections about it so

44. 22:38

    I need the merkel root of that or whatever and I have a journal which prove that you know I know those people I mean I'm just making this up right now but like you actually need to send that Merkel root two thousand for it to kind of validate it and like some history of it and if I do that but the change the other changes appeared like that Merkel root has no value to know anybody else

45. 23:03

    in a way except me but and maybe a contra party right like maybe I'm actually trading my ear I'm selling my data to Facebook for whatever reason and you know the Facebook it has value to me it has value nobody else would like to pay any money for it partially because I'm not gonna give keys to my data to anybody else as well yeah so this is what I would describe as like sort of

46. 23:25

    some variant on the data availability problem which is in cosmos so in a in most sharted block chains and there's some not exactly shorted box chains and also have these properties you or in cosmos quatrains we don't care we don't assume that data is available across wave which has one upside which is it allows sort of private proprietary systems interact with public buttons using the same protocols that we use to do public

47. 23:57

    launch into public law chain interaction which seems to be a useful factor potentially is useful but it has this downside which is you you don't know whether or not you'll like the source chain is going to exist and if the source chain becomes unavailable to you you it's going to have some impact on your ability to use the wider ecosystem and that might end up being really hard

48. 24:21

    to reason about and I fully accepted that bike might be a downside to a cosmos like system it it may not be a problem in practice it may be that like you look at the validator set of a chain and you like okay this is like pretty likely to be to remain available because you know that's all the same people I mean there's a 200 issue alligators in the world right now like I think one of the biggest constraints on growing the

49. 24:53

    validator set is proof of State ball chains have to become a lot more valuable before we can afford to pay significantly more people to run validators like you know we get like cosmos as a practical matter can't even pay a hundred validators right now enough money for them even though like break even yeah I mean that's that's definitely kind of a it's it's a it's a significant challenge in the space so yeah I think I BC like these are the

50. 25:26

    these are the challenges that we won't really start to understand until the system is live and has been out there for a while so I'm really glad that like other system designs are also evolving in parallel and I think short-term like a big use case is gonna be like bootstrapping liquidity bootstrapping initial applications on all of these different fast chain design variants so we can quickly so like it's not it's not so you know you don't have as hard a

51. 25:59

    problem of bootstrapping an entire economy as these as we try out new technology yeah I mean I agree that that's definitely like given cosmos is already here and you know whizzed on like whiskey no zones launching that's definitely interesting I think like looking forward right the question is where the balance will lean toward like where would the new commerce new developers in the echo system come in and how do we like guide them from like well if you're building this kind of

52. 26:29

    application you should choose this type of platform that's assuming there is a platform that is fast chain it scalable and actually has enough focus the creative partnerships already set up as well yeah I mean I think you are definitely it's so what I expect is that both will exist I think larger organizations large communities pre-existing communities will almost always choose to build their own sovereign chain rather than jump on some existing chain just because they

53. 27:02

    want that sort of full integration of the stack we'll see well you want to push back against that I mean the question I have and this is like a pricing question as well right because like if you're launching let's say I'm a community I mean like we were just talking that there's like some communities from reddit that actually spun off as their own website and their own like it's not

54. 27:25

    a blockchain at all they just literally need a way to monetize the community and they wanted sovereignty and they weren't they wanted sovereignty on features but like they don't really care like where's a you know financial system come sure I mean they'd not using blockchain right now you know what I'm saying yeah yeah they're just using databases the block sensor just databases yeah but the question there is mostly like how much they're willing to pay right because

55. 27:50

    you're saying cosmos hub right now is not made breaking even if we're talking about something even smaller that like has a smaller community then it's a bit harder to break even for validators to care and then you have smaller set of validators which are by the way very correlated with every other well like well data set so at the end you're kind of getting the same security as like from from an individual you only mean on the

56. 28:15

    whole like one high value blockchain to sort of anchor a set of validators into existence and then I mean this is mostly like this is our hypothesis but I think it does describe the world as it is which is mostly what has happened is like okay so there's like there's been enough like cosmos is profitable enough to bootstrap a you know 20 30 40 validators into existence those validators can make money to sort of pay the bills and keep the lights on by running cosmos validators hopefully with launching IBC they'll make some

57. 28:53

    more money not just from atoms going up in price but actually like transaction activity securing other chains like actually interesting crypto economics and then they use that as kind of their base and then they go out and secure all the like go out and like talk to all of other chains and you know find other things secure and find incremental value it's not like like one of the things that's I think interesting about the proof of

58. 29:17

    stake world is it's not clear that every chain fully pays for their own security it may be that like all of proof of stake collectively sort of subsidizes the security of all of proof of stake now there's like one stake security pool and as because reputation effects are so large that like effectively you can be a much less secure chain but like let's say because you know poly chain labs or occlusion or somebody else who has like

59. 29:49

    a lot of reputational skin in the game decides to validate on your like we make know like occlusion makes almost no money validating on Terra we do it because it is an interesting project in the cosmos ecosystem we think we can probably upsell other services to the Terra ecosystem there are a variety of reasons to do it but like on an operational day-to-day basis we make where we have lost money on the Terra et

60. 30:14

    Ossa but we still run Terra on our production ecosystem and like the Terra ecosystem benefits from the security that it could of like the full occlusion system she's saying if somebody comes in and tries to bribe you for Tara specifically probably not gonna do it they're like there had been oracle attacks on the Vitara or on the tarot stability oracle i haven't done a lot of analysis i'm like how those happens though so but like the for what validator infrastructure they do run on inclusions

61. 30:47

    production infrastructure I mean no Oracle is a separate story because there's I think any validator can though can app can invite contribute to the Oracle though so it may have been in some validators involved in the attack we also the the largest double signing slashing event on cosmos so far on the Terra blockchain you know like a fifty thousand dollar or slashing event so this is a good point to sort of turn it

62. 31:10

    over to the audience I think actually the first question I'll ask has our moderator power which is kind of changing gears like what was I'm really curious about sort of looking inside I mean I know a lot of the stuff that goes on inside of near and but but this is to both of you either way which is looking inside of your organization what are like some of the mess ups or things that

63. 31:33

    you wish you would have done differently or things that you are like like oh moments that you learned a lot from that were like you know formative for for how you move forward um I guess the thing that has been honestly a real challenge for cosmos is that transition from you know what like launching a system like that that launch moment - you know we're six ish months out from launch and you

64. 32:03

    know while we were working towards launching we had like seeing your purpose we knew how to prioritize resources it was obvious like what needed to be done next like we had we had we like we agreed on a launch roadmap that we like essentially like followed in like April 2018 and that was like that took us all the way basically to March of 2019 like we had a year of just like this is the thing on the list that's done we'll do the next thing that's done I think it is really hard as

65. 32:33

    a ecosystem like once you get past launch to figure out what to prioritize what's more important what's less important should we be making 10 Dermot faster or higher throughput should we focus on more features that are appropriate for should we do like lines and like honestly at least we have IBC together at the door so that gives a some amount of focus and right pretty soon we'll

66. 32:58

    have IBC out of the door and then will become even less clear what is the obvious next solution and maybe like what happens you know so like part of our strategy has been like to spin up you know help spin up other teams that are beyond kinderman and that has worked ok but yeah it's it's it is a little bit of a this is a bit like the next phase the phase that comes as hard as launch is the phase after launch also super hard got stuff to look forward to yeah that's

67. 33:30

    something to look forward to and make a note for sure I think on our side I mean I think there's a kind of hardest pivot of of the sort that we did was switching pretty much totally how we do charting that like pretty much at some point we realized that the design pretty much everybody else's uses right now like right now and kind of was what you know helium has been doing before is very

68. 34:00

    much like affected by like adaptive adversarial corruption it's affected by data availability issues in and else alike very hard to implement and that was kind of like we actually implemented it I like being part of it and as we were like trying to figure out what is the all possible thing that can go wrong with this we're like well like just figuring out how to like get it to

69. 34:26

    actually like production state would take so much time and at the same time we know holes it's known issues who is like how to break it so and as I mentioned if it breaks it's really not obvious how to put it back - yeah so that was like let's just say we do we did like somebody did break it there was like no clear way to reverse it so that's where pretty much spend some time in specific Ballack spend a lot of time thinking like is there a better way to do sharding in

70. 34:54

    general and then we just kind of set down by whiteboard and cut down all those things making it simpler and simpler and simpler until we actually like made design very simple such that a it's easy to implement beats easy to understand and see itself solves all its problems and that was like a pretty transformative moment and obviously pushed down a little bit from what we expected to lunch I also forced me to implement tender

71. 35:20

    mint with with our runtime just so we have something running for some of the developers we have well I mean I I would say that you know for instance with IBC we have we what we decided so like you know for most many of the pieces of the cosmos stack have been rewrite at rewritten him once or twice or three times and a lot of the reason why we think that was necessary is we didn't do a lot of design work upfront so one of

72. 35:52

    the things that we really focused on with IBC is do all the dish like literally four months of design work in like that open community process where like it was mostly tender mint Enderman interchain foundation and gorrik folks but there were other contributors as well who would come along and like do a do a design process with all of us and hopefully the plan is that that allows us to deliver IBC in a faster way when

73. 36:23

    now that we've shifted into development so far going pretty well and we'll find out you know we still have you know we had this demo we have security audits we have a somewhat unclear idea of what watching looks like so there's still a lot of work to be done cool all right yeah I want to turn it over so does anyone have a question I can pass you the magical cube I've I feel awesome listening you guys but I feel like the very very first question was never answered like if I'm a

74. 36:54

    developer today why do I use cosmos water use near can near work on cosmos what am I supposed to develop where what Chane do I choose so if you're a big company then you can go with cosmos I mean smallest startups have had some success I mean but it's certainly not it's there is no admittedly no real career path for like a solo or two person to be person team to end-to-end launch a cosmos you know how she and I would say like if you're a smaller startup depend it really depends on what

75. 37:33

    your goals are because if you really want to deliver a product if you really want to get in front of the users you should go and build on something that provides you all the infrastructure the same was like if you're building a startup and really what you're trying to do is get your product in firms users you should go in AWS or in Google Cloud if you're trying to build you know robotics farm or something or if you're trying to build like some cool stuff

76. 37:56

    that we've seen like cosmos is something called the cosh network which is why one of the founders of firebase or actually I think the founder of firebase and CEO before he sold it to Google of firebase and he built this whole control plane for like arbitrary kubernetes workers on top of the cosmos SDK it's it's insanely complicated or if you are a big company but you only have a small number of developers and but you have like a very clear product vision like tera did like tera I think probably had like three or

77. 38:31

    four developers even though they're like you know they've raised lots of money they're part of like a large corporate entity like network in Korea it allowed them they like it just like get a fully integrated system out the door like really fast cuz you know there is a ledger app and validator infrastructure and hsm integration and a web vol it that you can fork and all of that stuff like that all those pieces actually are

78. 38:56

    there so it does it does accelerate your time to market what I get what I what where people are obviously frustrated in the cosmos ecosystem is that's like I have some simple idea that I want to try I'm gonna try some financial and like primitive some variant on staking some weight I want to just change things up in maybe a small way or experiment with something new like cosmos does not currently offer any

79. 39:28

    real path to this now you know probably potentially months or years down the road we will have we intend to have things like you know cross-chain validation we're like you can just pay some cosmic hub validators to validate your chain sort of in protocol and so like a single developer could sort of negotiate all of that in protocol and and have a chain come into existence so I imagine things like that get become possible in two or three years but you know near might just be like completely

80. 40:00

    awesome by then and maybe people aren't as excited about that so there's two sides to this does that answer your question all right cool anyone else got questions in the back we'd be awesome now I mean alex has a bunch of code to write so it's got a quick question primarily on IBC so let's just stick with like an e rc 20 token if I'm trying to send it to like two different block chains integrated with IBC let's say by Nance Dex and somewhere else so I need

81. 40:33

    to also build the original smart contract on both of those other chains in order to mint it on those new chains now so basically the way so this is called ICS 20 yeah for whatever reason the it is the token transfer standard and basically so IBC is designed to have to be between heterogeneous systems so like you might have a minting contract that is in a minting and freezing contract that is on solidity on one chain and runs on the EVM and you might

82. 41:15

    have go coded the cosmos SDK on the other chain that implements the bits like comparable business logic but if the ICS 20 standard is implemented on both sides there will be a freezing and minting contract on one side and there will be some module that implements freezing and minting on the other side so you should be able to send your you know token whichever chain you're on to that implementation and

83. 41:45

    immediately get it in the form that it is in the other implication and like things that we had to do in order to like preserve like that vision of compatibility are like for instance cosmos SDK integers are 256 bits like aetherium integers which everyone hates but like they were like there's so much fighting about this because like we really don't need anything more than 64-bit integers but we made them through

84. 42:11

    we made them 256 bits to make sure that we could support you know usually $20 denominations which are 128-bit does that make sense yeah I guess so either way like if you're trying to send a maker token and a die from etherium to cosmos what did you need to separate contracts on cosmos to track ownership of the cosmos maker token and the cosmos dokdo no so like the car the ICS 20 implementation on let's say the cosmos hub knows how to

85. 42:49

    handle tokens coming from arbitrary ball chains so it basically says you don't have to implement a new contract it just says oh on ICS 20 messages come in I do this now there's a maker token that originated on aetherium on my chain same thing would die and so typically you know in out in the ICS 20 ecosystem like all of the complex business logic about when do you issue and stuff like that all has to will reside on the source chain so you'll have arbitrary things but then you just implement this uniform

86. 43:23

    ICS 20 packet and then any other chain that like understands ICS 20 can just like it know that the token is frozen on the source chain and instantiate a movable token on the destination chain so you can only move kind of let's just say n ft or or like a fungible token so that's ICS 20 and then there's another ICS that I don't remember 4 NF T's and no I don't think I forget what they would picked

87. 43:52

    and then but like you also like so then you have and like we imagine there will be many of these other application layer protocols like another one that's being worked on right now is cross chain accounts so accounts that sit between ball chains so like you have like count on one ball chain that is essentially controlled by the validator set of another ball chain that's an example of

88. 44:19

    like not token transfer use case cross change taking all that stuff and then you have what gorrik has been working on which is sort of generic code and message in JavaScript transfer system that also will live on top of IBC so we have we have all of those and I'm kind of hoping you know over the next months to see the developer community come up with lots of new application layer protocols I mean we just have it easy we

89. 44:51

    just know that it all in one chain in the same protocol so sure you can call anything get anything and you know it don't work so yeah so I mean you just have like a function call interface between there like R is exposed either both on the same chain and multiple chains hi yeah I have so many questions have you considered that G of incubator like IBC in an incubator so I guess the

90. 45:28

    big thing that like I really feel about like so if such a thing existed I would want it to be a Dell is basically what I would say I think the one of the biggest like I'm hugely critical of like ecosystem funds and and like incubators from the project team like all of your judgment about like what is good and what is that is totally corrupted the your ability to a capital is completely like

91. 46:01

    misaligned with incentives you're bad at judging things so and it only gets harder so like I would say that you know we're six months into having cosmos be live and it's kind of been completely obvious like what we should put resources into and I think every month that goes by and more of the like sort of general investment investing community starts to like believe in our ideas it becomes harder and harder to I figure out like what's under-resourced what's what's good what's bad so if

92. 46:33

    there isn't some sort of like novel investing like decision-making structure on top of this I'm I have always had never really believed in like that kind of a mechanism yes I mean like from our perspective the way we think is like this space right now is missing really like successful companies and like oh yeah and like part of it is like when we go and talk with our friends outside of crypto and like say hey oh you have friends outside of crypto yeah you're

93. 47:05

    lucky man well I don't have any threads etc and we're trying to convince them to quit their job and you know start building in crypto so we have more friends in cryptography see that's what happened to be exact it's all my friends who were not encrypted to be in crypto we didn't exhaust that pool yeah the kind of the feedback there is like hey we've been building and been successful in you know doing the up to like so special has a friend who build patreon

94. 47:29

    for example and like they have you know there's huge like huge amount of tools you can build really quickly execute amount of users you can on board and there's business models you can use and here in blockchain like we're kind of lacking even if we have ideas how to solve some problems we still lack in examples that people only can like oh yes and like firebase actually is like interesting example where that when they

95. 47:54

    launched they had like few example apps which people like oh like I don't care about the ABS of the law that they have but like it's a good example of the use case that I want to use in bread no like that's how I actually started using firebase myself was like oh I gotta care what that is like it was interesting experience that they tried and like okay now I'm gonna go build something on myself so I feel

96. 48:19

    really torn right now it's like so like it was obvious that until we had like scalable fast proof of State Bar trans it was gonna be really hard to build anything that anyone would want to build with like so it was like you know it basically lost like three years of my life to make sure those came into existence and now they exist so now what's the question is is the you know and like we struggle with this internally inside cosmos and they argue about this like a wall and this is you

97. 48:48

    know part of the the having a live system but like trying to pick direction which is there's one strategy which is go all in on like base layer tooling you know build a mobile sdk build a JavaScript SDK like get IDE integration well I just make all of those tools come into existence but all of that still assumes that someone else is going to come along and actually build the application the alternative strategy is to go all-in and building an end-to-end

98. 49:23

    application yourself and maybe hive off any of the tools that you had to build along the way but like focus on like one good solid use case as an organization as a company certainly both approaches sort of exists inside cosmos we don't know which one to pick but I think I think almost every project is going to face us amor I wanna I think I mean the kind of incubator idea right is kind of doing both is and and I agree that it's totally biased and like that when you

99. 49:59

    need to pretty much when you're giving money to somebody that's where that judgment is totally off for sure so yes we're like you can but there's so much other help you can provide a project being a system right was a knowledge given access to developer as you can prioritize their tech support request you can build libraries they need based on like what they do and we've done some of that and like there's an application also at the defy a coupon called micro tech that is like very much

100. 50:29

     like there's no formal incubator but we were definitely incubated that application using a combination of resources from externally the originators of the project and our internal resources now I've been doing that for projects building or nas yeah I I think I've been seeing it and I'm excited about it yeah and a big part of it actually helps us a lot to improve our own like developer experience and user experience because it kind of is immediate feedback loop from their users and their

101. 51:00

     developers - yeah especially with your developers and you're just building tooling it is easy to sort of wander off and boletes well I mean I think that's what consensus was able to fill none of their applications were really that successful but there was some benefit to sort of co-locating application and tooling development like the tools that are the parts that have succeeded and actually have adoption basically none of the application succeeded but maybe there

102. 51:33

     wasn't some benefit having a feedback loop so if I want to go about implementing IV C let's say for example on near where do I start there is a rust implementation of IBC that is being like some of the pieces of IBC that's under the interchange c't hub which would probably be pretty helpful for building one that targets near you can probably build all of the pieces of IBC targeting just the near BM you don't

103. 52:05

     have to make any base layer changes and then the other part which is the hard part is how to build a white client for their consensus protocol which may require some more help yeah we were working on that because we're actually building a bridge to a stadium and we already fully filled like figured out how to verify serum headers in like very nice cheap way nice and now we trying to figure that out the other way we're cool I'm fine like our headers pretty much on

104. 52:38

     the helium which is way harder yeah III think there's lots of people on the sort of cosmos research team that would love to talk to you about it we're very we're frankly very interested in helping build white clients for other protocols we've really not seen any other research efforts that are we just had a presentation today from the chain save team on Heath to like clients which went well but it was also very confusing and

105. 53:10

     yeah I would say that you know white clients were a thing that we thought were easy and then over time realized was and then we tried to do it and we realized they were hard and then we spent a lot of time writing specs and now we have implementations of those specs and hopefully that's good enough I I do think the approach that the sello team has has been following is also pretty rigorous but I we desperately need like my clients are the basic building block of interoperability so if

106. 53:51

     we we've got to figure them out for a lot of different protocols we got a question over here as well you guys have been talking a lot about slow and fast block chains what makes you feel like the new generation of ball chains is fast enough or significantly fast enough that it's gonna be different than what you're calling slow I mean from our perspective like we're building a shard a block chain but more specifically

107. 54:15

     building dynamically sharted block chain which means that over time as usage grows we actually add more capacity to the network and resize it add more shards which really means that like obvious I mean there's like theoretical limit on network etc and it is quadratic starting at the end so a lot of asterisks but like it is actually expands with demand so it doesn't have like this idea of like slowing down

108. 54:42

     or doesn't have this idea that at some capacity now you kind of eyes a bidding or waiting for transaction to settle and on the other side of it as proof of stakes in general are faster because you know who is the next validators are and you don't actually need to wait for block propagation you can just send them to directly you can pipeline a lot of things kind of align everything such that every single thing happening as fast as possible right for example like

109. 55:07

     we were just talking about the finality gadgets like the tender mint is finalizes every block but that requires two rounds of communication you can actually pipeline them and that and have like two blocks in advance already produced when are you finalizing two blocks before crazy Chinese people actually implemented this and then ran it out of money but their github is still there I mean and that's how Libre

110. 55:33

     works also the implemented pipeline consensus I would say there's a couple of things in my mind that distinguish fast and slow and I mostly think that like once you are like well also I'll describe what sort of gets you into the category I call fast and then I can I will argue that once you're in the fast category being faster probably doesn't really matter that much at infinity

111. 56:04

     there's certain applications and certain use cases where it does and there will definitely be specialization you know they're going to be used cases where like the insane throughput of like something like Solanas is appealing there are gonna be applications where you know a slower tender mint style chain is kind of is is appropriate but okay so one of the big things that differentiates slow and fast ball chains

112. 56:30

     there's like two characteristics in my mind one is synchronously safe versus asynchronously safe consensus so Bitcoin is slow because of its consensus Nakamoto consensus is only safe if you can assume that all of the parties in the you like that care about your block chain will be able to receive a block within a certain amount of time so and that time Bitcoin has parametrized it's 10 minutes

113. 57:00

     aetherium has parameterised it as i don't know some number of seconds 12 seconds yeah Sirians parameters are query far more aggressive and probably represent the upper limit they also using uncle's as a way to kind of circumvent some was a game theory that comes in yeah well like that doesn't get you around like pure delay yeah so whereas a protocol like tender mint the validator set can finalize blocks as

114. 57:37

     fast as they want as fast as they can and you are not going to end up on a fork of the network or like seeing inconsistent state because you're receiving blocks slower than that and at some point we'll have state sync enabled by Nan's already did it but under weaker is security assumptions and you won't even have to have the burden if you're not a validator node of seeing all the blocks and you'll have this rough basically the same security guarantees so from that like the sort of

115. 58:13

     optimal you know given infinite time and infinite money what tenorman would actually look like is if you were running a full node you'd never you wouldn't actually sync all of the blocks you'd only sync state from blocks that you care about you'd ignore the transaction data and and like the only things that would ever see all of the block data are the validators and that can be very fast and so that's fast one of the and like you might have caught on

116. 58:37

     to a little bit and let my explanation is one of the consequences of that a state becomes a lot cheaper state for aetherium state for Bitcoin is really expensive because that state has to be globally propagated to everyone who cares across a global network and because state is expensive things like white clients for other chains highly stateful applications like gaming etc become whereas there's a there

117. 59:07

     there's a tenorman based chain loom that at least over the summer has like a 200 gigabyte state and it's like mostly trucking along just fine like hopefully the like we made some improvements to our myrtle trees that like make those less io intensive but in general like you can have these like very steep feel very large block chains when you have a synchronous consensus space you're gonna save consensus that you can't have in the slow Nakamoto regime so that's why I

118. 59:39

     think the emergence of these traps block chains like fundamentally transform like what the use cases are for block chains but is going to be at least 10 years if not longer before we have confidence in these systems as base layer money cool we have time for one last question from the audience any where you guys raising your hands are just scratching okay anyone good no okay that's one year

119. 1:00:18

     so the idea right now so we do have actually a white client it's just not that light that's that's the main issue so we actually have stay sync it works like if you're running out right now you will not need to validate every single block and in the Charlotte system in the first place you cannot validate single block every single block that's a kind of whole assumption so we do have light

120. 1:00:41

     client we just validate headers the way we elect and select validators is somewhat complicated because we working around a lot of kind of proof stake in current problems that capital efficiency of proof of stake as you know price lock shades is very clear so you actually need to adjust the price per like without loudly receipt so like you see name has 32 ether which is like an arbitrary number taken from I'm assuming sky but I hope there's some math yeah I

121. 1:01:14

     think it may be 40 to minus 10 that's probably what happened but like we actually have adjustable price to adjust able kind of stake size that you need to be a validator and that requires actually doing a pretty much like a sort in shuffle right that is like in salinity expensive because you're trying to transport the deformation and you just have the last validator set sign the next validator set and do that you know how many times I have argued that

122. 1:01:45

     this is the worst idea ever because if I'm a validator now and you know this somebody kicked me out because I don't have the steak anymore why in the hell I would sign it ever well instead I should sign me myself right and like if you have a majority I mean but that just that's just holding the chain it is foliage I mean like I agree that this is an option if there is a better option

123. 1:02:11

     I would prefer using that I think like relying on this assumption that you have like to search carrying over from ebook to ebook because that's kind of what you say is there is lesson to search no no no that's not what I'm saying I mean I mean it's less than if there's more than a half that witch that got kicked out that half should collude and sign themselves back into the next block and

124. 1:02:33

     control the lie but that only tricks like clients yes and that's it's laughable in the main changes are still bond yes so yeah so you if you're adding like slashing conditions from other chains that's true yeah no no no it's not Sascha conditions from other chains like the signet I mean the same really getting into the we yeah okay but yeah I got your point yeah so so there may be

125. 1:02:58

     some week we have a great write-up on all of this on our the one important thing that we want to add is what such thing called skid block and this is the idea that you actually building a skip list of blogs so you don't need to verify so reproducing right now one block a second and this is a lot blocks so you actually like want to skip a bunch of blocks when you like traversing

126. 1:03:21

     a chain so if you just produce the skip blocks within an epoch that's fine because it's a model data set and you kind of can just keep ebooks yeah so so I'm sure there's like some more ideas how to how to get it to to state where it actually works and the main problem again is like the prices of helium gas on signature purification and things like that and our our desire to have more validators as like network expenses

127. 1:03:52

     we mostly give it up on the idea of like that's what we have picks yes I know that's that's why I was asking like how are you gonna do it for as a change because yeah well I think for like fast ball chains it it should be fine each one dot oh is this weird like it's not a fast ball chain but it has some like interesting program ability properties you think so the do see don't seem to be like options for doing like low trust

128. 1:04:21

     but it's tough you know look how complicated TVs DTC is yeah so yeah yeah TV DC's but yeah stay tuned for a video we actually have his mad bum and James about tv DCL on near I guess an are YouTube or their YouTube so yeah I want to wrap up on something that like a lot of these conversations go super deep in the weeds and so I want to pull back because these conversations happen in like rooms but they don't often happen

129. 1:04:53

     on the stage which is like and and try to keep it super short because we're gonna go overtime by like two minutes so what is your like 100-year vision like like the sci-fi future where blockchain wins like you have to think about that at some time like maybe when you're going to bed or in the shower or whatever do you central our spaceship market there you go like global mass participation in political systems that's your end goal

130. 1:05:18

     that's the end goal but you I don't think you can have global money until you have global participation cool so let's say it's like One World Order are you in the Illuminati yes all right you guys heard it here Ellie do you wanna say anything to that are you good I mean I think like what's really interesting is as we go especially if we go interplanetary house like I mean governance for sure economics how these things will actually be evolving you definitely need IBC if you could because

131. 1:05:49

     you're going to cosmos well you just can't be synchronous between you know I shot the earth shards and the marsh are they're just I mean it's all a synchronous man that's yeah cool all right thanks everyone for coming that's a wrap I do want to make one quick announcement which is that near steak wars is starting on Monday so we're actually kicking it off that's our incentivize test net kind of like cosmos

132. 1:06:15

     is game of steaks right yeah yeah mistakes so yeah we learned a lot from from them and talking his lucky of course so yeah if you guys want to participate test out running a node even if you don't plan on being a validator it's gonna help us a great deal and it's incentivized which means at the end you can get tokens on maintenance so check it out it's near protocol calm slash

133. 1:06:32

     steak wars or near ai slash steak wars so thanks very much you guys been an awesome audience that you want to give them a round of applause it's been awesome