Whiteboard Series with NEAR | Ep: 1 Zaki Manian from Cosmos Network |

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

1. 0:03

   hi everyone we have anxiety from cosmos today and we will talk in the preparation for the main net launch which is going to be happening very soon like you will tell more yep we will be discussing the the protocol and we're gonna go into a lot of very deep technical details so super excited to have you here and can you please introduce yourself yeah I am my name is ducky I've been sort of contributing to the cosmos project for almost two years I write now work on sort of research and

2. 0:34

   development and sort of new opportunities for for tender mint and I'll be talking explaining a lot of stuff that probably has not been well communicated before but we're probably is like probably one of our biggest goals over the next couple of months is to really start communicating some of these higher-level concepts of the protocol about how the blockchain is composed together cool awesome and to

3. 0:56

   start off can you I was playing like a very high level overview of what cosmos is yeah the goal of it so the goal of cosmos is to achieve sort of scalability in of decentralized finance sort of secure computation through a network of communicating blockchain so you have many block chains but one ecosystem so what we are launching the maintenance what we call the hub the hub is a very simple blockchain it's designed to just

4. 1:35

   do be basically proof of state consensus and then it's designed to keep account balances for zones so and so the basic concept of the initial vision of Cosmos though I think we're gonna get deeper into is each of these are a bok choy and those blockchains exchange data packets stick with each other and if those data packets contain a token or an asset the hub will say okay there are you know 100 cVTC on Zone

5. 2:29

   B there are 10 CTC and if the packet says if Zone A says send a cbgc to zone B it goes it goes like this and this becomes and then the proof of that state transition being sent to Zone B allows you to have one more CVG city and see what you see is personal Bitcoin well it's like the pegged to the hub security model version of something like that okay so that's the basic vision of what cosmos is and most of what the two last two years have been about is both making

6. 3:16

   the consensus engine robust enough to operate with a bunch of strangers up running the nodes no BFT system in the history of the world has ever been operated like this before and then the second piece was getting the proof of stake model correct so that it was not trivial to manipulate for like things like even in the context of a BFT system you still if you have sort of large cartels you exist in a world where there are various kinds of griefing attacks etc on

7. 3:49

   the incentive layer and no one has ever built an incentive idft system before so that's the other piece that we built and then the sort of third piece to the full vision is completing the state transfer between zones make sense and so since it's proof oak stick there's some negative token there is not only exists from the harbor Orton also exists every will have its own native token in the

8. 4:13

   current design there is we have a we have we see a need for potentially a single token to secure multiple block chains but that is not the common case that is the sort of specialized there are certain circumstances where this makes sense guys but there's a special program specifically for hub yes those are called atoms another thing to remember too to know is there's not like one hub in the cosmos network there are going to be other hubs there's one called iris that already exists and it's going to be launching sort of being

9. 4:50

   parallel with the main that where we anticipate there will be more so there's no centralized hub there is no centralized and so if I want to do a say if I want to send some token from zone II to some zone which is not attached to the same hub it's gonna cool and so one question is the the entirety is that the zones are expected to be some of different block chains so what would be

10. 5:21

    a use case when they would want to launch a block chain and not just an application and an existing block chain so one of the so one of the things that we've sort of fundamentally question in cosmos is the notion that there should that there is like one common pool of security that many applications make sense to draw on it makes you know part of the sort of crypto economic model that we're that is where we are

11. 5:51

    challenging the orthodoxy is the idea that unrelated applications that are not part of sort of a a common ecosystem that share common incentives actually make any sense to host on a single blockchain in a world where pft consensus and BFT computation is scarce then you only have let's say the ethereum blockchain and the etherion blockchain is your only BFT computation resource so then you want to host all of your applications but the idea

12. 6:24

    of demonstrating that BFT proof of snake actually works removes this fundamental limitation around scarcity as long as there are people who are willing to operate a blockchain and are willing to put enough skin in the game to convince an ecosystem that they are incentivized to securely operate that blockchain then it makes sense to have like it would make sense you have many functions thank sense and so another question sports so for CB t see ya at some point for cbgc

13. 6:57

    to appearance on a it needs to be locked on the Bitcoin network yeah is it zone a that recognizes the fact that was locked on Bitcoin or is it to have the hub that it's the hub that recognized okay and like that designation of what is so there could be like there could there you know in all likelihood will be many pegged to bitcoins in the hub there could there will be you know Bitcoin take one big balloon peg two big three but there's a likely to be sort of one Bitcoin peg in this ecosystem that the

14. 7:32

    majority of the social consensus and the largest amount of locked up collateral is securing and that is likely to be what most zones entry does but there's no there's no that's a single zone that is decision that is there to sort of represent Bitcoin in the in the system know like creating zones is a permissionless thing creating zones is permissionless thing creating hubs is likely not as permissionless it probably

15. 8:01

    needs more assent on both sides simply because if you connect to a hub and that hub fails and we haven't talked about how everything like how do you actually represent state moving back and forth between James you do we've just form and waved it at this point but this is the fundamental question is what happened like if you have a BTC peg zone and that BT see peg zone is compromised all of the you know all the clutter all the

16. 8:33

    Bitcoin that's stored on the Bitcoin blockchain for that zone suddenly is stolen what happens to all how do you reason about this and the sort of core design of cosmos is how do you actually think about this problem and making it so that the ecosystem as a whole can understand comprehend read how a security failure of one zone affects the whole ecosystem make sense and how however there's another purchase called other means

17. 9:07

    right is the other main part of the use a separate project is it ok so here's what here is what the notion of ether mint is basically there - it actually is there to solve three problems in our ecosystem or three challenges or any business one is so what a base either MIT is the EVM as a cosmic as a module in rst there is really no requirement on these zones that they be written with our software any fast finality bft consensus layer

18. 9:44

    will work as the consensus for a zone or even a central I mean even a centralized database would theoretically work and it would potentially work and the and like you can run any state machine that you'd like well what we've what we think is a great tool so we've built this cosmos SDK to make building hubs and zones easy and the one of the tools that we want to put in the SDK tools chat is the EVM SDK

19. 10:16

    model maybe M SDK module gets a couple of things for us it allows you to what we call hard it basically allows you to read EVMs like read if you're a mistake so you have some blockchain that has some state in it that you want to use like a token distribution having EVM as a module in the SDK it allows you to build a blockchain that starts with that state and then does something interesting with it so

20. 10:49

    this is what we call what we typically call a hard spit the second thing it says what's you use EVM soft so if you take an ecosystem like 0x for instance as I've said it doesn't we don't think it makes any sense to run all every different application that is running on aetherium on one block chain but there is a ecosystem of applications that exist around things like auger or 0x it might make it would make sense for all

21. 11:31

    of those things to coexist on one box it and so what having the EVM as an SDK module would what's the developer do is reuse the software from an from a an integrated interoperable ecosystem that exists on aetherium inside of the cosmos ecosystem and benefit from that right and these are plans to have a zone which starts as a as a point in time for code athina' sue we are not so we have this notion that we are not there are no

22. 12:12

    definitive plans for this because we are going to leave this question up to the governance signaling mechanism on the hub as to what we actually do but one of the challenges that we have in the cosmos ecosystem is that there is no native store of value medium of exchange to the Kosmos ecosystem atoms are intended to be highly illiquid and to the extent that other sticking tokens exist in the cosmos ecosystem which we

23. 12:40

    expect there to be many we expect them to also be highly we think that is the optimal design for a sticking token so we have three strategic initiatives that we are pursuing to help the ecosystem develop a medium of exchange in a store value and one of those ideas is that we would grab the state of aetherium run it as a zones and you would have this new zone down here call that is like it has the

24. 13:10

    etherium jam first this is just a hypothetical day it could be any date ever assirram account state and but the intent is not to run this as a fork of etherium forever because we think this makes no sense what are our point of view is is that the ether that is represented in this state can flow back into the hub as photons and to the extent that etherium is a store of value even a medium of exchange that people use which is it is in today's world we

25. 13:56

    think that there might be a shelling point around treating photons which are a mapping of that state into cosmos as a medium of exchange in a store about you then eventually this dies like the hub validators run this for six months and then turn off okay make sense but so let's say after Jen first it works out and then someone Sam who sold here's the theorem immediate from the theorem chain you can use it as photons right yes yeah

26. 14:29

    you probably wouldn't want to do this want to pick a past state rather than a future state when you are deciding but this is again a decision that is up to validate your governance it is also a crazy experiment and no one knows if it will work but in essence anyone can punch a zone which is Sanford with you yeah then and now the question is when I was in Prague there was all the people working there saying I'm a cosmos Bobby Baker yes right so they participating in test nights right yes ultimately as a

27. 15:02

    cosmos validator in the own domain that yes am i someone who is operating the hub or am i someone who is also operating some zones so being a cosmos validator is simply the expertise to run the software for any part of this ecosystem and for a zone or hub to be successful they must come up with a crypto economic model under which the validators who operate the software receive enough compensation that they want to operate this thing so we have a crypto economic

28. 15:36

    model for our validators in the hub which we think makes sense and we think there's a clear value proposition for why you would want to be an evaluator on the hub every zone will have to discover their own economic model that leads there to be validators on it otherwise no one would ever send their money to your zone because they would be like who is running it and they can do arbitrary bad things to me Thanks

29. 16:00

    and then in the in the bright future of cosmos what do you think is gonna be the number of zones thousands thousands and then validators how they expected to run like clusters of expensive hardware and operate multiple zone so they expect a single cosmos holy data to be operating like one of few zones we expect that so my perception is and I've been working on you know creating this validator community for over a year and there's maybe two hundred people who operate validators credibly in the world right

30. 16:33

    now if we are going to have thousands of zones which I think there is a market for and we only have hundreds of validator operators with credibility this implies that most of those validator operators will be running many many zones to secure them there is probably some future in which the population of validate or operators expands dramatically but I think that future is is more is like at least five to ten years away make sense

31. 17:05

    and so the consensus that is ran on every especially at least on the hybrid Thunderman just Enderman but it does not requirement for zones to run tender yes but it is by far the easiest thing to do to just run tender mint on your zone make sense and in the hub when we run tender so as of today nobody ever end EFT in any sufficiently popular blockchain yes so at this point for example we don't know how elaborate

32. 17:32

    is that more than one third of people will go offline for example sure right so but we do have data because we do have a year of running test nests where you know the the one third going offline is within some window of time because if you are offline you know if you are individually offline we unbond you and remove you from the validator set without so it's not like so the question

33. 17:58

    has always been given a removal time on the orderable of of a day or or a week you've been typically running with about a day of removal time will people sort of fall off the network and the network halt and what we've been i've been really pleased to see because in the test net everybody is running our tests mint is basically just like burning money like I've burnt a lot of money or

34. 18:23

    of my own money on this these test debts people have well everybody in our community has burned money and no one has any incentive to like be what rigorously monitoring have full-time employees like when pager Duty goes off because your nodes are down like immediately respond to it and so the question is would we have sustained aliveness on these networks and we have had essentially almost nine months of sustained aliveness moving from test net to test net on these networks which

35. 18:53

    gives me an enormous amount of confidence that the hub will actually stay up and what was they have a statistic in terms of how what was the largest percentage that was simultaneously applying I don't have data on that I could probably we probably let go through records and figure it out but we haven't been keeping data on it I mean mostly like most of you what we would have noticed it we went we notice halts

36. 19:17

    and we have definitely had periods of time or like all the validators are like where where like close to halting and a bunch of validators or like yelling in each other in chat rooms or texting each other being like dude you've got to bring your validator back online but it has been fascinating that like this is a truly decentralized system the test net that's running right now is a 35 have have has 30 535 validators on it we don't run any of them they organize the

37. 19:44

    entire process themselves this is the and so the name with 135 validators there's no something right so tender mint is run by all the brothers yes and our locators known in advance know-how but tender mean doesn't tend them in depend on knowing how many people are poor so there's a bonding process in which you add a new validator it says so there's some initial validator set that starts the network and so basically our

38. 20:14

    process or starting a network is we have an account state people submit bonding transactions which are signatures under those account state and then tender it processes that into a the initial validators and then you know two thirds plus one must upgrade to the new software run the paying and then a bonus and so for bonding transactions to get through the present set of validators needs to approve it right yes

39. 21:06

    do they have any incentive not to do so um so censorship is probably the most interesting thing to think about at this stage of the maturity of the system like what is going like all right block proposal censorship pre-commit censorship these are all things that we have thought a lot about now you need two thirds plus one of the network or you need one third of the network to be concluding in order for censorship to occur and no one has really ever tested

40. 21:38

    this and like how do networks bullet behave under the presence of censorship this is the purpose of game of stakes game of snakes is a test of what does censorship look like on a live network make sense and so here so the bonding transaction needs to be included into the block right yeah so effectively if the first block producer if the current block producer wants to censor it they

41. 22:03

    can propose a block without the yeah transaction but then the next block producer also needs to exclude it needs also needs to it or you can have a cabal of two-thirds of one-third plus one that is like a cartel 1/3 plus one that is watching all of the boxes and essentially rejects any balk proposal that doesn't meet their reasons I see Thanks but these attacks exist but if you have more than one third plus one you have bigger problems right look ah my my

42. 22:36

    Quentin what I think is cool about tenorman is the and has like turned out to be one of the most valuable things and I really credit jafer for doing this is one of the things that Jay focused on the design of tenorman is that actions are attributable you there's always a signature associated with the decision on the network so and an absence of a signature in the in the case of sort of

43. 23:07

    non duct so putting together the evidence of censorship and like detecting censorship is possible in this system and ultimately what we believe block chains are for is therefore automating social consensus so if there is censorship on a network social consensus has broken down about what that network is supposed to be doing and then the question is how do we resolve that and I think over the next week we're going to see this in

44. 23:39

    real life because I I'm in game of stakes we're currently aware of people planning censorship attacks we are aware of people planning Forks to respond to those censorship attacks and I think what we're going to see is what this actually looks like playing out in real life which whereby which censorship occurs on the network and then the social and then people attribute that that censorship is occurring they show you the block proposals that have failed the signatures on those block proposals

45. 24:10

    and that in pre boats and pre commits to those Park proposals that have failed and that gives people enough information to detect the failure of the social consensus on the network and move to a new social consensus and so as an example of that so let's say there's a cartel of 1/3 plus one yes that one to censor out a specific funding transaction yes right so and so at some point there's a look proposal with that transaction yeah all of them signed it yes there is no it feels to me that there's no evidence that they

46. 24:45

    actually did not sign it right the absence of a signature is not there but you do have the signatures of everybody who was not part of the cartel right but but could it be that it's the everybody else who censors the signatures from the cartel - for example - the intention will slash them for um for colluding i am i think that there's these questions about how people convince each other who is at fault are going to be super interesting and ultimately rise to the

47. 25:14

    level of a social process and there will be some end like there will be like when there is disagreement about this but what we will have is we will have a signed block proposal and signatures on that block proposal at the time from the time and people will be able to go in and say okay this is but like this is the grand experiment this is the experimental thing about running bft consensus if it works and it actually can't and peep and

48. 25:45

    at a social level when these kind of censorship events occur people can can eventually come to some sort of consensus about whether or not censorship occurred this will be but we what I'm skeptical of is automating this slashing for um there is acting there is automatics there is a small automatic slash for downtime and in game of stakes we are experimenting with all of this stuff being a lot easier to do than it is in the normal world so in game of stakes you know you don't have to put

49. 26:24

    down capital to join the network just head SunNet so if you get all your friends to join it can be part of your cartel you joined gamma sticks you have a large cartel Indiana stakes in the same way for you also we are also tuning the slashing parameters so that censoring people and getting them slashed is a lot more effective on game of stakes than it would be in the real world because we want because we want to see these attacks I mean once you start understanding what do they actually and

50. 26:56

    you think these No so let's say there's no automatic slashing for for not signing a book yep is there then some sort of manual way to size those people later or they will be just punished or do you work and losing their melee I mean participate the new social consensus that spawns some new blockchain that is you know hub prime can instantiate those accounts of everyone however they want it's ultimately just like the fundamental

51. 27:30

    question of whether or not uninitialized a vaca as legitimacy comes down to whether or not the operators and the wider economy perceive those people as legit and so forget starting in your social consensus is a card for ya make sense and in the pub how how often are the blokes created um so we have been running with a five-second delay on ball creation for from basically the reason

52. 28:03

    was back at the beginning of the year we did not have a five-second delay on ball creation and this resulted in the blockchain growing very large very quickly and so we so right now I'm just looking at the numbers on our test net for a hundred and thirty-five validators we have six point five second box and we have a five-second delay in there so it seems like one to two seconds on main net for a blockchain is for Pretender

53. 28:35

    man running with a global a global validators that seems key but tender mint itself if I remember how tournament works the the time that it allocates for people and for pre-commit is hard-coded tried it the the time at section is a parameter of but if the people it happens before the timeout um then they are forced to pre-commit without waiting until the time is over I mean it actually waits to the end of the time I see and how long is time out there I

54. 29:05

    have to look at what we're doing right now my face is on the order of like a second tour it's like 500 millisecond by millisecond say so and so that means that within 500 milliseconds they so after people these 500 milliseconds for quicker means right yeah so if valid lady validators are spread out around the world yeah if isn't the latency between within China and United States when you fragrance is around 200 milliseconds they would not be able to

55. 29:31

    sue to do the precariat right to be completely honest this is not a thing that we have fully empirically understood exactly what's going on the goal of ten durmand has never a dis stage in its maturity has never been transaction throughput like like transaction throughput metrics are things that like there are multiple constant factor improvements over current Endermen that are achievable

56. 30:02

    like you can make a pre vote on block n plus one a pre commit on block and why there are so that would take out a significant amount of latency in the system what we are we are what we are mostly observing is along the system setups that our validators have chosen to run so far this set of parameters is working well and the parameters are not in consensus even so people can adjust them and thereby just what they end up being a

57. 30:41

    sort of an emergent property of what the operators choose to run so the operators you know if the operators wanted like sort of pull down these parameters and have a more geographically concentrated hub they can do that if they want to increase the parameters and have a larger validator set it in a more geographically dispersed up they can also do that this is just sort of essentially these are like emergent print constraint free parameters but like there's also an enormous amount of

58. 31:06

    engineering work that we have not done which is just like make 10 Dermot fast um our goal has like for this phase has been make tender mint robust enough that a bunch of people who are not something who you know are basically professional or a hobbyist validators but not necessarily like distributed systems engineers can actually operate this thing with strangers over the Internet I see awesome and then I guess from here

59. 31:36

    we can move to the most exciting part the IBC yep so IBC my understanding about ABC that's a way to move yeah let's go yeah so I'd be seized a way to move assets between sounds right yep and so from perspective of overcome yep so the hub doesn't care what sort of consensus the zone is running it does need to be able to submit a like client proof of that consensus see hub that the hub is able to understand and and can people

60. 32:16

    deploy sort of custom code into the hub for validating so in theory yes in MVP yeah like apparently polka dot plans to do this I think that is a great idea I don't think it is a kind of thing that anyone should deploy to a blockchain in like the next year make sense like what I did like infinite flexibility and the ability to like chuck and like jumping between protocol code and between mike the code that's in the runtime and the

61. 32:54

    code that's provided on chain or just make the complexity of these systems like greater and we have so little experience and knowledge running these systems that like we prioritize simplicity so if there was if somebody out like right now the only BFT protocol that meets our needs that even exists in the world is tender mint so it's like what are we building to but like in the event that someone builds something else that is good like that would be a thing

62. 33:25

    that would be a probably a great idea like I think the progress so like what are the what are the components of IDC so one is like client purse it must be efficient given some root of trust like so you need some root of trust in an efficient way of getting to current state to prove your own state like hypothetical blockchain will be snap sitting to latest books so usually when you've sent an IBC packet you send so entender mint

63. 34:05

    for instance kinderman is a very efficient like my persistent you send the blocks so you send the blocks that prove validator transition so there's some number of blocks that are depending on how dynamic the validators that was that are that showed that the validator set transitioned and then the current block height and then a miracle proof of your IVC so for as long as two thirds of validators are honest yep like

64. 34:37

    that proof could not be and there's two thirds of signatures if actually yes on that and if they're dishonest submitting evidence of conflicting blocks in this block light would terminate the connection but then they still if the zone is corrupted yep what can they say thank you will do is they can completely fabricated state right yes and then send the proof that's what the areas that the destination zone is taking right so

65. 35:07

    sorry if you if you okay so like let's okay what's this comes down to like sort of the core set of diet of like understandings of of how given a system like this how does like how it how do you reason about corruption okay so you have some hub let us assume for the sake of argument then the hub is not corruptable like that corrupting the hub is hard this is the the the route the reason for

66. 35:48

    that assumption is the sort of core reason why hubs are a scarce resource if they're corrupted they affect the integrity of a very large ecosystem and therefore they so everybody has an incentive run full nodes who are participating in the ecosystem there are many parties and this is why Forks and heartland upgrade should be rare but so you have and then you exam so you have let's say so I'd say this is evil zone

67. 36:23

    door actually so the question has been a sort of fundamental question here is soo evil zone is doing evil things there they could withhold box they kid fabricates fabricate say transitions and the good zone does none of these things in the hub doesn't and the evil zone sends Evo Krypto kitties from the evil zone to the hub to the good zone and the good zone is like a decks of and people exchange

68. 37:24

    their cats for money so the whole point of cosmos is is that when you see an evil cat on an evil zone it's actually labeled as what was the origin and then what is the asset name and all the honest participants in the protocol will maintain this relationship so when you see this the good zone anyone who's exchanged in the goods arm is gonna be evil cat will see evil cat from evil

69. 37:59

    zone and you know they will also see like good zones liquidity's okay so you've good times opening and people are exchanging evil cats for good zones of credit each other our expectation is is that if you're withholding blocks and you're fabricating state transitions essentially eventually the market will be like why is there this blockchain where no one can get the box and why you

70. 38:45

    know no one really trusts the zone operators who are they and no one can obtain a fullness like I tried to run a full node and the information that the evil zone is evil will be socialized and everyone who is a counterparty of things under the evil navel Spit namespace will eventually be a will be like I will value these things as zero right but by that time someone might have lost and

71. 39:06

    they will catch right sure someone could have lost some some like you know good zone wicked tokens in exchange for evil cat and the you know if if you know there is a possibility that the good zone could upgrade their state to try and make these people whole yeah or it could not your probably easier just not to because this is a blockchain and terrible things okay what my whole reason why I think this is an acceptable

72. 39:41

    design of a system is because let's say that instead of having a good zone and a hub and an evil zone this was all one giant state machine and had exactly the same security properties as everyone else the reason why I think this is fundamentally an acceptable design is this set of reasoning where you have to reason about the state machine that originated your asset it's still there as long as that state machine can be

73. 40:12

    upgraded or has an admin or bike really as long as there are any of the hooks that like fundamentally are necessary for software as we currently build it - like sustain itself over time you still have to think about is the state machine that originated by asset and every asset some level of risk associated with it which is that the state machine that manages it could be corrupted right so I have two questions here yes the first one so in this example with the evil cat

74. 40:43

    yeah let's say that evil cat did not originate in people's own let's say that it originated in on a theorem yeah there was a regional cryptic eating yep and you sound sort of way to to mark it on a theory on yep it appears on the heart it went to the someone would say it's awake an eighth cat now originates comes in okay so each cat goes in the hub the eath cat then goes to it gets shreddin trait goes to evil zone so now

75. 41:15

    you have like each cat on this on the Sun now what could evil zone do with this eat the cabinet it can steal it from its owner it can be like no this is ours or we transferred it to anyone we want cuz it they have arbitrary control over the state machine they can so that's the main thing that they could do they could go off long so like my eath cat is now frozen in this zone that has died but what they can't do as long as

76. 41:47

    the hub is honest is is make 2/8 cats right so effectively whomever is that person or that entity labels on that person the eth cat that that was their responsibility to assess the security of the able chain so if you if you are using evil zone you should have some opinion about the security of you myself and so the second question is it feels to me that the so if let's say I'm

77. 42:18

    trusting either ethereal so let's say theorem and court must have the same security from my perspective sure the cosmos cop for instance yes but this particular zone has different security right so then if I keep using a stadium it seems to me that I only need to be accounting for the risks of a theorem chain being corrupted yes or or theorem hard-working and accepting a fork which

78. 42:47

    contradicts my personal values yep as the one yes right which for many what's happened in the past then what while in cosmos it feels like I also need to be accounting for the fact that a small a smaller set of operators has colluded has colluded and this becomes people yes so what is your bit like in the absence of a notion of cost that seems like a negative trade-off but one imagines that

79. 43:19

    you know like mayest I sort of back in the envelope calculations like rough guesses is that computation in the more securing the ones that are relatively well trusted like even if you move your Creek Creek well you want to trade your crypto kitty on a theory am you're gonna be paying tens or hundreds of thousands of gas for to use a Dex you're going to you're gonna these conformation times will be you know on the order of minutes

80. 43:51

    if not longer and that will be the status quo and when you are on cosmos you will be you know with today's software getting you know one to two second absolute finality times your gas costs will be at least a thousand times cheaper like you will you will be so this is why ultimately this crypto economic model competes against the the etherium crypto economic model and then the question will be where it you know for what set of users is the decrease of cost because right now what you have on

81. 44:30

    on on aetherium is you have two choices you have choice a which is I can run on very expensive very difficult to corrupt software on aetherium or I can go to a completely centralized system in which the admin of that system has arbitrary control what cosmos is basically building is a middle ground where there are validators the validators can't harm you and that validator pool is not enormous it is not

82. 45:03

    completely infeasible to corrupt but it is difficult to corrupt and your costs are much cheaper and one imagines that there will be there will be some more validators W zones with more you know more state and so and potentially in those zones with gas prices will be high and so when you're saying that gas prices when your thousand times cheaper yeah is it coming from the fact that I'm many zones or from the fact that there is no the cost is lower because you know you don't need to card hardware to to

83. 45:34

    compute hashes um there are two reasons one is simply there is less like the gas costs on etherium must pay for like etherium like a theory and gas costs are in a competitive marketplace where this zone ecosystem is a competitive marketplace for gas where people are looking or where the market is trying to choose between security and cost and is continuously pricing these things and so

84. 46:04

    if your zone is charging more than the market will bear for security a competing zone will show up and compete with you and the market will have to judge this um and like one of the things that is like a fundamental question but like once you have bft consensus that can be run in a public setting I think emerges naturally is whether or not the market can price security accurately and it may take five years it may take ten years for markets to like and there will be horrible disasters of

85. 46:36

    zone takeovers um in that meantime but my thesis is is that this architecture is is in some sense natural and as long as eft exists in this architecture will exist too and the market will eventually find an equilibrium of understanding how to price censorship resistance and security versus the value of the computation make sense and another question I have is so you mentioned that today majority of the validators the

86. 47:07

    some sort of ass-like reduce core channel yeah no well I mean there are many channels I see there there's the validator community has like alright like one of the reasons why we have to ensure censorship and resistance and integrity around the hub if we just said like this is the Blessed slack that all validators then the admin on that slack would essentially be running the hub we

87. 47:33

    would be running about and obviously we've wanted to like avoid this case as much as possible so we've thoughtfully encouraged the validators to have many channels by which they communicate and many channels in which they form social consensus um and that we don't manage that social consensus in any real way but then if there's exists a channel where more than 50% of validators are

88. 47:59

    well they they're yeah then doesn't that create an opportunity for someone to use social engineering now what they would convince all of them to do something absolute just like hard-working into something absolutely is definitely a risk and this is why being an operator of any of these PFT systems that we're building is what requires a sort of operational security mindset sort of deep suspicion about the world like deep

89. 48:28

    suspicion about what you're seeing and it has been very interesting seeing that evolve over the time like we've definitely seen social engineering attacks in these channels before like we will see them again like yes as long as validators are communicating with each other those communication channels are exploitable our main form of resilience 2 is to encourage as many of them as

90. 48:51

    possible but yes this is a like what we are creating with the main launch of our main net and with with with game mistake he's a bit of a like of a Hunger Games style a competition in the world of who can run systems like this in a trustworthy manner and know I think that the you know the the practice of the last year of running these systems intestine is enormously valuable and but we're going to see where does this come from

91. 49:23

    and in many ways of like the last 10 years of operating a cryptocurrency exchange has have tested like many of the same has like produced a population of people who had the operational security chops to have hotkeys on the internet in the same way this in world of validators that is being constructed is going is creating a world of specialists who have the skills to operate a system where you are running BFT consensus and you have the consensus keys live on the internet yes

92. 49:56

    that was all the questions I had about cosmos okay anything important we didn't care what cover um I guess to me I like I said I think the main thing that is gonna be really important for for the cosmos team for tender men for us to communicate to the world is how to think about this mom ultimately the world is not going to think or care very much about the low level proof format but the world every user of cosmos is going to have to have

93. 50:29

    a mental model of what does it mean when I see this evil zone such evil kitty or good zone liquid token you know in another zone and one of the fundamental question and right we are clearly in a very immature place of being able to reason about that I think we're even when you look at like exchanges operating on aetherium contracts how much auditing of an IRC 20 contract do they do before they list it many not

94. 50:58

    that much and you know many of these ERC 20 contracts have admins in them that can arbitrarily mutate state and you know I think exchange operators are just in the very beginning of monitoring that and so if you're an exchange or any other economic entity on this network you know when you are when you are you know receiving tokens you are going to have to have some way of reasoning about is the hub that I'm connected to trustworthy and secure and like are

95. 51:30

    these validators that are staked into this actually operating the system in the same way if I accept if I start being a economic entity whether I'm a retail business or an X or a cryptocurrency exchange or anything else that first starts to emerge in this ecosystem and I start accepting tokens that originated from Goods own do I believe that do I run a full note on goods on better on a

96. 51:57

    full note on good son in fact if you are an economic entity and you are a counterparty to any of these things you should be running a full note of everything you are a counterparty to and the very last question is so you're saying that a lot of that is at least that the realities is going to be more of an experiment so his question was going to be I'm baking frequently in the early days yes or what I'm hoping is is the system that we've

97. 52:28

    laid out can be more or less in place by sort of middle like by q2 of 2019 then hopefully at that point we can leave the hub alone and do most of the innovation in like we can have like okay we have the hub at the hub is reasonably stay about what we like hard spoon every all the atom distributions and do crazy experiments of hub and crazy experiment hub we do we upgrade it like you know all the goddamn time and it's like more

98. 53:02

    or less a glorified test net that it's interconnected to the hub it's like one step removed from test nets and this is where we try all the crazy new ideas like like mutable objects that can be transmitted between zones instead of just simple things like crypto kiddies like alternative event formats so that you can implement like other more complex business flows like there's lots

99. 53:23

    - there's lots of experimentation and the whole point of cosmos and the reason I've been like I've been working on this for like essentially the last three years is that right now when you want to experiment with a blockchain like like we've all seen the cycle of like new technology being deployed on aetherium it's like new technology idea is invented code is written six months of politics happen maybe a hard fork that

100. 53:56

     upgrades that protocol this is like we are never going to get to a system that the global economy can work on using that as the way of doing things what we need is the ability to launch new block chains have them handle real assets in small amounts see if they work see if these big new ideas make any sense if they work at scale if any of the economy has any appetite for them and then move them up the ecosystem into the hubs and

101. 54:24

     so I guess what I'm really excited about is once you have these primitives it unlocks the building blocks to build everything else it basically it makes it very easy like if you are playing around with like actually getting the shared security thing done you plug into this ecosystem very easily and very early like any beacon chain design plugs into this system trivially um and you can just interoperate with it trivially and it's very easy and you can get to play

102. 54:54

     with all of your technology and so you can incrementally deploy pieces of sharding within the security model like there's there's lots of innovation the finally gets unlocked once you have these pieces and they work oh yeah so thanks a lot for my pony