Stake Wars IV: Attack of Transactions w/ Illia Polosukhin & Bowen Wang

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

1. 0:01

   awesome so we've had already a few spaces about phase two and kind of protocol road map so um I'll just quickly recap so NE protocol had a kind a multiphase approach to rolling out uh we had uh phase zero main net in 2020 uh which launched with one Shard we had phase one launched in 22 uh that uh Reed to for charts and we having phase two happening kind of um within next few months with uh testing to really

2. 0:40

   introduce uh stateless validation and uh in return allow to have more shards in the network have uh most validators have lower Hardware requirements and uh in increase throughput of each individual sh as well by putting State into memory uh we can dive in into some of the technical Parts but uh maybe good to give some uh context because obviously B and Dennis participated in Stak Force before kind of uh what has been the purpose of Stak War so far uh kind of

3. 1:22

   why is it really important to have such events and uh how they help us uh really build robust Network yeah I mean yeah I can I can start yeah uh I think I've been through all the Stak Wars uh great yeah sorry I yeah I can I can I can start um yeah I uh I been through all the stakeholders that we've ever hosted I still remember the very first iteration uh where you know that was uh uh before mainnet and uh uh we were kind of trying to start a a test net together

4. 2:02

   with a community and on the first day it didn't go very well because we there were like 40 people trying to start the network together and it didn't work um for a variety of reasons and we've come a very long way since then um so historically the Stak Force um mostly have been focused on uh getting more people into the V Community as well as testing the major releases of near uh mostly for sh in so had one before mainnet launch and then after main

5. 2:34

   launch we had Stak Force for um for the uh uh the introduction of Chun only producers as well uh which which also expanded the the valid set um but this time uh it's uh it's a bit different uh with the new upgrade uh on the horizon uh we plan to uh so which would also complete phase two of char making near fully shed uh broad chain Post in terms of state and transaction processing um which is very exciting uh but for the Stak Force itself it's

6. 3:11

   mostly uh focused on testing the implementation um and maybe less so on the uh on the validator side uh so we still have uh validators going to have invite validators to join the testing uh Network and wrongness validator see how how Network performance and and uh also the adjustment to Hardware requirements and things like this um but uh uh the primary focus is on sending a lot of transactions to see how the uh the new design or the new implementation uh

7. 3:45

   Works under onto the load uh and also with the in memory like moving state in memory we should also see uh performance improvements as a result of that uh so should be able to handle uh more transactions as a result um and that would also be part of the uh the program and what we want to test um so yeah I want to pass on to Dennis to talk about it from his perspective yep y we can hear

8. 4:28

   you

9. 6:00

   yeah uh to yeah Denis is exactly right about the the audience right I think uh uh historically um State Wars uh have been focused on validators or like potential validators who want to get into the uh community and become validator near uh but this time uh as St is pointed out is bit different because we're primarily focusing on the uh uh kind of more of the testing of the the new implementation itself through like sending a lot of transactions it's why we named it ATT tack of transactions um

10. 6:37

    so it doesn't you you don't have to be a validator to to join or like you don't even have to want to run a validator node to to join the network it's not uh as as long as you're uh interested in the um uh in near uh development um or like the just want to see the sharting panels uh and actually getting uh implemented uh in practice and and see the um Improvement of ser of near uh then yeah it's a it will be a great opportunity uh for you to uh to join the

11. 7:12

    program uh and send transactions to to test test Network and I actually think it's probably um you know for this type of activity a smart contract developer would be a great uh would be great for uh for the program um because like with any kind of testing uh we want to have like more variety there so we wanted to see people sending a variety of transaction loads uh to help us uncover uh maybe issues that we have not uh seen

12. 7:43

    ourselves or some bug in the code or or things like this all right so to summarize we have the new network um called stateless net that is up and running that that uh you can either join as a validator and uh provide kind of uh like more validator power test out learn how to how to run a validator in the new environment or you can uh you know deploy a smart contract and uh engage it

13. 8:22

    with a lot of transactions uh try you know some regular things like sending uh sending fible tokens or minting nfts or uh any other use case that you want to do and uh really try to squeeze out all the power that this new network will be providing when it goes on Main net um so I think the you can go to new.org Stak Wars it redirects you to uh GitHub which describes kind of different ways

14. 8:59

    to participate and how to join yeah I mean I I yeah I I can talk a bit more about the details of the program given that we were just talking about the high levels uh before and the you know target audience and and I think I'm pretty sure there are people in the audience right now want to know uh more about uh the details of the program um so right now it's divided into two stage uh so the program is uh uh is designed to run for two months from the beginning

15. 9:30

    of February to the end of March um and it's divided into two stag the first stage is basically the entirety of February and the second stage is the 30 of March uh so in the first stage is mostly um PO for us to set up the network and uh making sure it's operational um and also seeing the the issues by just running the network itself um and also for people to come in and uh uh poke around uh in the network and then uh play with it to see what

16. 10:02

    kind of um issues or bugs you can find uh and during this stage um the activities uh that uh uh would um uh that would be eligible for reward is to submit uh bug Bounty reports uh sorry to submit bug reports uh on GitHub um and that would greatly help us uh making sure that our implementation is uh secure and stable um and then in the in the second stage where uh that starts in in in March we'll have both the um external validators joining this testing

17. 10:37

    network uh along along with s and also um rewarding people for actually sending um um a ton of transactions um and uh how it calate the we would calculate for the transaction uh reward is uh it's detailed on the reward section of the program on GitHub uh can you can take a look um and yeah as as I said again the focus will be uh on trying to find Problems by uh playing around with network sending transactions

18. 11:10

    to uncover uh issues that we have not previously seen cool so if you have any questions about that you can ask them a bit later but uh now maybe let's switch gears and talk about the changes uh this introduces to validators and uh how we expect this affects um chunk producers as well as the validators and shift of the role in phase two yeah so I can definitely talk a bit more about that so the one of as a result of this U big change of stus

19. 11:53

    validation ARG the largest uh change in the design and the implementation since we launch main net um we're um we're switching or like we're uh reimagining the the row of like validators so previously or currently we have blog producers uh on on the network there are hundred of them uh that are responsible for producing blocks and actually uh they track all the sharts to today and then we have chunk only producers that

20. 12:24

    supposed to produce chunk for a specific shart um so there are a maximum of 300 of them on the network today but you know in practice it's not fully utilized um so in the new setup um the rowes are I would say bit switched around uh so there is uh the trunk proposers uh or like you can you call them trunk producers well but call them trunk proposers just to avoid uh the confusion in the naming which the chunk only

21. 12:57

    producer we have today uh so the the trunk proposers are the type of Auditors who would operate more expensive Hardware um they will still uh check one Shard um but they will be responsible for producing the chunks for that Shard applying the uh that the chunk uh including transactions and receipts as well and then generate the state witness required for other vors to um to execute the chunk uh and then uh the reason why

22. 13:28

    they would require ire uh more expensive Hardware is that in the design they actually hold the entire state in memory and this is where um we leverage the uh the Shing design of near uh so that the state of entire network shed into different shards so that we can say uh the the size of each Shard is bounded by some constant let's say hypothetically 50 gigaby and then um the the validator or sorry the chunk proposer for The Shard would just be able to load that

23. 14:00

    entire STI in memory and this result would have to operate um machine with more RAM and that's um that's most of the change and potentially higher Network bands as well due to the need to distribute uh stess sorry St Witnesses and then the the other type is the U uh validators the sters validators uh whose uh job is to verify the state transition uh proposed by the chunk proposers are correct and they do so by verifying the execution against the

24. 14:33

    state witness provided by the chunk proposers um and as a result they they would uh not they would not need to wrun as expensive Hardware as on people due today uh mostly on the side of uh SSS and um the requirement on that should be dramatically reduced um so that's the kind of the overview of uh this change and um yeah as I'll pass it to Dennis to talk about this from his perspective hello again yeah well um totally excited about see that new

25. 15:13

    technology coming along and um of course from being a validator we we do not really care much about you know the how expensive hard we can run it means so long we we can secure the network we we provide is a requirement requir Hardware but saying that if you are aiming for like thousands and thousands of validators uh this is of course becoming crucial point where like a validator can run from Raspberry Pi or like you know from PC or something when when valid

26. 15:47

    becomes um a point where it doesn't have to be be like 100% up time where you can like bring it and on on and off like then then yes of course the the hardware becomes really relevant and you really need to look at uh you know making it more affordable to run for like ordinary people like you know with the Technologies out there like you know with the scripting and everything anybody can install the validation in no time um but like right now with a

27. 16:20

    limited set I don't think it's very important from valid point of view to run like on cheap Harbor we do not mind to run on most expensive also I think um there would be like some questions in terms of like a problem itself probably leave it to later yeah I think Dennis mentioned a good point about the decentralization of the network indeed uh with this change um there is kind of like a bigger maybe

28. 16:50

    shift in like how the uh the validators are or like how we think about uh the the validators set right so before is basically um like every validator is essentially like equal to another validator in terms of like their role in the network uh but now with the stus validation change there is this asymmetry introduced uh so there is this um chunk proposers essentially uh a a few of them in the network that would uh operate more expensive Hardware um and

29. 17:23

    responsible for producing the chunks and the state uh witness as well uh but then there's a ideally a lot of the uh stus validators uh who can potentially operate on on cheap Hardware on cheaper Hardware um and uh they would be responsible for verifying the uh State transition proposed by the uh chunk proposers um and the important thing here is that the chunk proposers while there are a few of them and they operate more expensive Hardware they actually

30. 17:56

    will not be able to do anything malicious because security of the of the network is dependent on those stess validators actually validating the uh State transition using the state witness um and as a result I think this is a good tradeoff to to accept uh and uh is not only U providing more potential for decentralization of the network uh but also it's more future proof right as we as we can see that zero knowledge

31. 18:25

    Technologies maturing very rapidly today and we expect that uh maybe in the next few years we will be able to see an actual integration of Z knowledge proofs into the uh into layer one protocols uh in that point we the steros validation architecture uh can mostly remain the same except that um uh instead of validating the uh a chunk against State witness uh they can validate uh just one zero knowledge proof um and uh make make sure that the uh the chunk that they

32. 18:57

    verify is valid yeah so maybe to add and contrast this to other Solutions right that we see we have ethereum with rollups where each rollup runs a sequencer and then they settle on some data availability uh they put the kind of transaction to data availability and then after that uh settle execution on the ethereum in the way uh the chunk proposers are a decentralized sequencers there's you know multiple of them per each Shard uh making sure you have you

33. 19:38

    know you're robust from halts of any individual uh kind of chall producer you have the data availability integrated directly into the protocol after they produce uh chunks and then you have the kind of validation settlement done with stateless validation uh in a way you know making it a realistic rollups um so you have kind of very similar design space right I mean there's other blockchains like polot uh that are also using similar but with

34. 20:18

    kind of usual you know near focus on how do we make the user experience as straightforward as chain abstracted as possible uh kind the idea here is to minimize like to you know not have any changes to how users or developers are interacting with this while being able to add you know both performance throughput and also uh being able to add you know more validators to the network uh as the requirements get lower uh I think also Dennis mentioned a very

35. 20:52

    interesting thought that as we have more of these validators which are not uh which now can validate uh any Shard because they don't need to have the state of the chard uh we can potentially also reduce requirements for that for their online uh uh up time and in turn we can have more validators for are kind of more uh providing security to the network but at the same time are not online all the time and so can have like

36. 21:27

    kind of a longer tale of home validators or uh folks s running you know potentially on their that stop that make you know that doesn't work that that's not online from time to time I think that's all kind of in further iterations of this but uh this definitely design is kind of pushing in that direction to make it uh larg larger network with more validators who kind of can plug in and plug out uh

37. 21:59

    without needing to maintain the state of the chain which is the current Uh current design yeah um that's definitely something we can explore further in in the future and and indeed with the you know thec becoming more mature like this question of like how much stake we actually need to secure a network right because at the point like the uh the fundamental security model the proof of stake blockchain becomes also an

38. 22:31

    interesting point of discussion because like everyone like you can in your browser can verify the uh validity of the of the blockchain and um maybe at some point we don't need like a it they we don't need necessary them to be recognized on the protocol levels validators uh like it's more like it's also about the uh kind of the U in a way consensus or like social consensus outside of the uh the protocol itself that can be achieved through like white

39. 23:00

    adoption of their knowledge Crews um yeah that's a lot of interesting stuff to be explored all right so maybe uh Switching gears again um I saw a question around what are the rewards for the program so maybe I mean you already mentioned about kind of the the uh schedule and how it runs uh would be good yeah kind of what is the uh what do people expect from this but also from Dennis perspective kind of if there are any comments or thoughts about

40. 23:38

    the program and then we can open up for questions from the from the audience yeah yeah so uh yeah again like the I would uh uh refer people to to kind of the rewards section of the uh GitHub uh statefor GitHub um and probably I should post it in the in in the comment as well but roughly the idea is that um we're rewarding people for like the the issues that they find uh during the the program and uh

41. 24:10

    obviously you know not every problem or every bug is is equal like some are more critical than others uh so there's this um kind of assessment on how critical a uh the the issues that people find is uh and um on how complete the report is and there's some multipliers uh that will assign to those things and then uh and then there is a this Baseline reward of 150 year and um so we multiply by the multiplier and then decide on how how

42. 24:45

    much reward we uh we give to people uh and then for the um for the part of uh traffic generation or sending a lot of transactions uh to the network uh it's it's all also like a similar concept there's just um so people would post a summary of like the type of transactions that they're sending um and uh we would assess the uh like how uh how important that type of traffic is or like how

43. 25:15

    novel it is uh and also see like the impact on the network and then there's a multipli sign and then there's Spacey reward of 500 year uh so we multiply them together and uh give uh the that would uh determine the final reward um so yeah that's kind of the the idea um and yeah because it's somewhat complex so I would say like please refer to the actual document for for more details um and

44. 25:43

    yeah yeah yeah yeah yeah I posted as an answer here to near Ukraine Guild tweets but yeah Dennis go ahead um okay yes uh that's very very interesting top topic and um we know there is a lot of validators outside of U new yor system who runs other chains who already trilled and would like to join n system because all of the validators trying to expand their portfolio and um some of them never never never looked at near some of them had before uh and a lot more now as I

45. 26:24

    can see a lot more looking at near and right now all of the most of the questions like the top one question of this how our stake words for and help them to join the main validators I mean the words are words in like they have a monetary value but for validator is more important to have a seat on the main net uh where like some Foundation or some other NTC or someone else can can provide with their minimum Stak so they can start valid agent so this

46. 27:05

    basically leads to the question is the main net set is going to be expanded in the future or not I think this is top number one question I have been seeing talking to to other validators and other network about St was four yeah that's sure go ahead yeah and another question is basically more technical I don't know if people would like to hear that it's just basically about the transactions and stuff so I know there there there are ways to send a lot of transactions from one single

47. 27:39

    machine there's also ways to send it from you know various sources but the question really is you interested in like volume so let's say one single machine or or it's going to have to be like volume from all all around the world it's all different kind of fish terms of uh complexity of setting up these kind of uh setups so if you need just one or two or five machines to send a transaction is one thing if you need like hundreds or thousand machines send your traffic that's another kind of fish yeah

48. 28:13

    so yeah I think both questions are very interesting uh so for the first question about uh you know the Val set and how to become validator on here um so I mean yes definitely as we get to more sh there is potential to have more validators on the network but what I will say is that today like we're nowhere near like utilizing the full capacity or like full capacity of the the data set right now so as I just check we have 2 11 vors uh on Main at

49. 28:47

    today and I think the the maximum is 400 uh so definitely there's room to to have more uh Val there um even without changing anything right so uh that's still like quite a few validators uh if they want to join the network they can um and I think the minimum uh stres code to become validator today is about 25,000 year um I mean yeah I think it's obviously subjective on like whether

50. 29:17

    that's high or low um but yeah I think the um I mean I'm not sure about the uh incentivization program to to delegate uh stake for um for like newcommerce uh I I think uh you know I don't think near Foundation plan to operate anything like this uh in the in the near future but maybe IO can talk more about that part well I think in general even at previous take Wars we we try to uh instead of near Foundation being kind of this operator and delegator which generally you know has a

51. 29:57

    lot of criticism on itself has um kind of additional um you know operational overhead overhead uh we prefer to use Liquid staking providers and and really work with them the benefit of near liquid staking providers compared for example to ethereum is because of kind of Highly programmable staking uh they are not actually a centralizing power but are decentralizing power uh I think both Li and St near are taking across you know

52. 30:31

    hundreds if not uh 200 validators and so uh I would expect that like as Network you know continues to grow we will see more and more Stak shifting to liquid staking and you know they are supporting new validators as they come online so I would work with the you know metap pool and linear uh to kind of get your initial delegation from from those Pockets uh and you know as far as I know they have you know brought in new validators on board recently uh kind of

53. 31:04

    through shifting sell their delegations to them I think uh maybe relevant question is right now there's a minimum of 20,000 near that's required to become a validator and so uh if this change kind of uh will also lead to us reducing that limit that may be a good question to Bowen um but I agree that even right now there is available slots so it's more about working with uh the liquid staking providers to get uh some of the initial

54. 31:35

    delegation to hit that 20,000 uh limit yeah uh yeah I think definitely that limit uh sorry that limit could be reduced but I mean it is going to be a relatively involved protocal change so for Contex that limit was introduced so that um uh the so that the probability of any validator like today not producing any Chunk in an Epoch would be extremely low so like essentially like very like pretty much like not going to happen in practice right so because today like every

55. 32:13

    validator is going to produce some chunk or like act as a chunk producer in some uh capacity uh so they are expected to produce um at least one Chunk in in uh in an epoch which is uh 43,000 blocks um and otherwise you will not get reward but I mean if you become a valer and not get reward and then get kicked out that's definitely not ideal right so that's why this minimum threshold was uh introduced um but indeed with this shift uh or which this change to stus

56. 32:46

    validation that's no no longer the case the chunk production is taken care of by the uh chunk proposers and the stus vors actually get rewarded by endorsing uh those chunks um and indeed uh because of that we we no longer need this original threshold that's dependent on like them producing like one one Chun in during the the entirety of epoch um but again I said uh Tech I mean conceptually yes we can we can uh get rid of the threshold but in practice um changing this will be

57. 33:21

    a relatively involved protocol change uh so uh we do expect that uh we would need to take some time to make the change uh okay I think we can uh talk about um uh the second part of the question that uh that Dennis uh was talking about um yeah sorry Dennis can you remind us the the second question yeah yeah sure it's about the um stage one where you want to see a lot of actions right question really how how do you want to see them like just to I know it's all in new document but just to clarification

58. 34:04

    what is in a docum yeah yeah uh I I don't think the this question you ask is actually in the in the in the docs um so yes great question I I I I mean I don't think it we particularly care about like where exactly people send a transaction from that's uh less important whether you send it from one machine or send it from a lot of different machines um I think it's more about the the nature of the transactions right so what type of

59. 34:29

    transaction are they are they um you know transfers or like cost to Smart contracts if they are cost to to Smart contract what kind of smart contract are they calling and things like this and it's and also the volume obviously matters right so if you send one transaction the day probably doesn't really help um but where exactly those transaction uh come from like in terms of geographic locations we don't really

60. 34:55

    care okay I'm assuming we want to see both we want some people who will like spin up you know quote unquote bot Network that will try to you know overwhelm overwhelm it AC from the globe and then we'll it's interesting to see also just kind of massive traffic from you know single servers because we actually see both of the traffic patterns right now uh in real applications right we have metat transaction rers who are routing traffic from a single server or from a single CL

61. 35:26

    like small cluster of servers uh they're using you know hundreds of keys to send uh non non intersecting transactions um but we also have you know many of the non-custodial users transacting from their wallets kind of across the globe uh and doing that uh as well so we kind of see both of this traffic patterns and I would would be interesting to see them in testing as well and at high capacity and uh seeing how how Network performs with both of the use cases and especially yeah making

62. 36:04

    sure that you know between nons kind of RPC routing all of the pieces are working in unison and so if if anyone interested in investigating that I mean that that's like an area where I'm sure there's something to to be found and and reported as a team for extra uh bu bu points okay thank you so I would U suggest all the contestants to get to know new system better read the documentation understand the smart contracts and try all the different transactions types you know create deploy smart contract delete smart

63. 36:44

    contract execute smart contract methods uh make swap use ref Finance test if there is already available or not for for doing some complex transactions and um you know so be a winner yeah sounds good um thanks Dennis uh should we move to questions from the audience