Zero Knowledge Community Group Call #7 – NEAR DevHub October 27th, 2023

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

1. 0:08

   all right well hello everybody um and thank you for joining us for this uh zero knowledge Community call for near here on devhub uh my name is Maxwell I'm a devrell here at devhub and I'm excited to get to talk about a few topics in ZK and the community today um just a few reminders about the format of these Community calls these are sort of just open-ended calls for the community people can propose agendas people can come talk about the projects that they're working on or tools that they're improving or just general things

2. 0:42

   about the community and the developments that they're working on are seeing and uh today we've got these two topics we're talking about the Privacy use case on Mir and then we have some updates from uh Opa which I believe was formerly known as hide your cash and um with that I guess I'll just pass it along to Pedro are you going to start us off then okay uh sounds great uh guys I don't have a presentation today it's basically just talking but I can forward

3. 1:16

   you guys a couple links after so regarding Opa formerly Hydra cache which went through a rebranding uh due to some concern I would say regarding what is the message being passed by a a service called hyro cache so our version on year went through the rebranding and it's still the same smart contracts and everything but we're running on a now different front end and different address if you guys try hyra cach is going to redirect you to opec's new

4. 1:54

   website and we are also finishing a new version of it on kadina which is is the first topic and it's and we like to call it an upgrade of the version we have on year and how is how does that work uh so first things first it is the same basic concept where you can do Anonymous transactions but there uh what we did is you're not limited to a ticket so like tornado cach or former Hydra cach other mixers where you you could do transfers

5. 2:32

   with one 10 100 near e or usdc now you can have a it's basically a private wallet where you can have any balance and perform any kind of transfers that you might want to uh we you have a a model based on UT txo so kind of join them together and burn them in order to build new uo and that way you can transfer any amount of any asset that you might want to uh we also made it available available so that you could have private nfts and do

6. 3:13

   private nft transfers the it should be going we already have a a version for tests online but you would have to deploy your own front end and also there is some upgrades that were necessary on cadina that we had to they specifically they use packed which is their programming language and we had to update it due to not having basic infrastructure for ZK so there's still no testnet with

7. 3:47

   those upgrades for us to to test it on testnet once it is available I can send you guys but our code remains uh open if anyone is interested on on seeing the the mainly I think what is cool about it is the having the ability to perform those transfers with variables amount variable amounts and now uh I don't know if you guys have any actually questions regarding it more if they're like too deep into Tech I might not be able to answer them right now but I can address it

8. 4:20

   later if not uh I'll talk about the following topics yeah I'm not sure but I think we can jump to the uh ZK ceremonies okay so now uh more okay Phil I was just gonna ask I me just out for industry learnings you mentioned the fact that um you know you were kind of rebranding and switching a little bit for you know because of concerns about the regulatory environment is there something about the

9. 4:56

   design of the um system now that mitigates that or okay so OPAC or former Hydra cach we already had a uh a layer before you use the app where you had to join an an allow list uh we use Hy one as a provider so anyone that tries to interact with the protocol first what you have to do is join the allow list and when you join the allow list we are very actually verifying happy one

10. 5:30

    if your wallet is listed at as a elic as involved in any listed activities but also while using the application let's say you performed a hack and really quickly deposited into Opa there is a chance that if your account is listed on happy one as an account related with elicit activities you would be block listed in your funds would be locked locked until of course if that happens by mistake we

11. 6:06

    you we would talk with happy one and address the issue but this is right now this is our safety mechanism so you got a bit you have the hockey one part but you also have a bit of Game Theory there where if you're a hacker whomever it is you don't want to risk losing your funds so this is our but that's what we have at the moment Phil we also don't have like large transactions our largest ticket ticket value is for 100 year or $100 which are

12. 6:44

    almost the same yeah so unless you're hacking someone for like two three grand more than that it's going to be a lot of trouble to move a couple millions and it's more of a we like to think of it as a proof of concept rather rather than a final version of a privacy product that could be used in scale because there are a lot of regulatory issues that also concern us and we're based in Brazil so it's a bit easier like we're not in the

13. 7:16

    US uh regulation here kind of there's no actually no regulation regarding crypto except for taxes and taxes they just dropped this week so we're relatively safe uh like developing proofs of proofs of concept with ZK products and privacy products here but at the same time we know that I mean crypto or web Tre is kind of countless right so if we're developing something to be used or mass adopt we

14. 7:49

    have to have that in mind thank you and now uh regarding the ceremonies and the proofs uh those are two new Concepts that we've been in studying we've we've been getting a lot of feedback uh first I'll talk about the proofs because I think it's more related with the first stop but we've been getting a lot of feedback especially from users in places where they have not that great of an access to Internet or maybe they're mobile they're using it to they using a an app they use ZK through

15. 8:23

    a mobile device and they have a lot of trouble building those proofs on a mobile device at the same time if we do that the proof generation on a server we might be compromising user uh privacy so we've been studying with uh we've been studying a solution using te like trusted execution environments uh there's there's Intel sjx but there's more specifically Fallen Network I don't know if you guys ever heard

16. 9:02

    ofala but you can execute verifiable privacy private computation inside of F and also decentralized you could do that just by using AWS or Asia but F brings the also the value of being decentralized and we're developing now a proof of concept to build those proofs on on Fall Network first we'll actually try on on aure and later on there so that you kind of break the proof generation barrier so you have better hardware for your users that might be trying to use it on mobile or don't have

17. 9:42

    access to as I mentioned good internet providers or anything like that so we visualized that from the small user base that we have right now this is one of the largest issues towards adoption right because un fortunately there's still a lot of requirement to generate proofs so that is the next step we're going towards OPAC as AO and we could use it for OPAC tickets on year we could use it on kadina but more than that anyone can we

18. 10:15

    we really want to build it open source so that anyone can use it for their own ZK product and the ceremonies they come in the same way where we've also had uh once we when when we built the app we had to make a choice we were basically uh either using graph 16 or plon and plon we could reutilize a ceremony that was already used but plon the proof generation time takes way longer and that sacrifices ux and some other times

19. 10:50

    that even blocks some some users from using it like we had that that issue with Hydra cache where some users got in in contact with us and they said I just I have an old Motorola device and I just cannot generate the the proof there and I cannot withdraw my funds and that's something that we had no idea that we would bump into so using gra 16 is something that we want to do but the

20. 11:13

    ceremony is just a huge pain to do because it takes a lot of effort with community and takes a long time and every time you want to deploy a new version of your app you have to with r Ross 16 which is faster for Pro gener you have to so that is a great question because from our perspective like five to 10 seconds is what we've discussed with a couple players as a like good industry

21. 11:44

    standard but I've talked with the guys from privacy scaling Solutions on ethereum and like their goal is to hit the 5sec mark like 5c for proof generation is good uh but I think sorry question please yeah uh is this uh in browser or like mobile proof generation or is this like server s optimized so we would like to and then again this is a go for the future but I think five seconds for any kind of user would be great like I'm I run a Mac M1 and got

22. 12:28

    decent memory so it's faster generate proofs here but mobile device that is another story and but we know that there are some users that they don't want to generate the proof anywhere else they want to do that on their own device so there are multiple angles that you kind of have to attack here right you have with we feel like using graph 16 is good because makes it faster for everyone but also having the option to generate those

23. 12:57

    proofs on on a server like a private trusted server it also improves but for the ceremonies itself uh what we want we're developing a proof of concept right now uh with f Network which is to generate this the ceremony for graph 16 inside of Falla in a te environment so we could use like actual actual random numbers from there but we can have also external inputs different order of inputs but to build a ceremony

24. 13:36

    there so that it can be built faster I won't say that it's safer but it's also i' say that it's in the same level of security as a regular ceremony but that those are the two steps that we're taking towards improving um I would say user experience as a whole but not just user experience but also developer experience uh because if you want to build Ceremonies for a ZK app you could do it way faster if the proof of concept

25. 14:08

    Works which is it's not ready yet so I don't have like the something to bring you guys to show you like oh it's working or we might have issues here or there but this is the direction we're heading to uh not just looking at the app itself for the final users but how can we also develop backend stuff to to improve the experience as a whole and those are my updates yeah uh thank you so much U from

26. 14:39

    the perspective of proof generation uh yeah you said like you use your Mac uh have you tested uh like browser proof generation or like uh from the Mac itself because it's it quite differs a lot uh sorry Anon would you mind repeating your question like my's neighbor is being a bit loud here no worries yeah so you said you have tested like proof generation your uh MacBook uh and my question was like is it refering

27. 15:14

    for uh in browser proof generation or like on Mac at all oh no it was in browser proof generation like right now if you use uh opad on year it's going to the proof on your browser we don't have any server side proof generation but what actually one thing that we're trying to reproduce right now and we're not being able to we're trying to find devices like especially mobile devices where we just cannot generate the proof because some users said that they just can't generate the proof in

28. 15:54

    any ways and like I I have an iPhone I tried with the Samsung device but which runs Android and I could not uh like not generate the roof it takes a bit of time but it generates but we we're kind of stuck like trying to find some old mobile device from someone in order to to to actually reproduce the mistake that they're having the year that they're having uh I think uh like I haven't any troubles with proof generation in web I

29. 16:26

    just had like the troubles uh with uh the time limit for example when your pro like not your but when proof generation uh takes like I don't know like more than 10 to two minutes or three minutes for example if it's large proof uh the browser can stop this process and like uh like the stop this worker uh and uh like my following question have you considered like for the uh for the

30. 16:57

    desktops uh I mean like laptops uh have you considered like web gpus no we have not yeah because I I've seen uh several one team is working on them trying like to leverage web gpus uh for proof Generation Um because yeah I have my my previous question should be have you uh uh made the parallelization proof like proof generation in parallel or or it's done like in in in one way uh you mean for OPAC tickets like what we have right now or for the proof

31. 17:42

    so for the proof system it's literally there's like no code at all it's just a concept uh and for the tickets it's I might be wrong here because I was not the one hands on on it but I think that we don't have anything running on Parallel okay but with that yeah there is uh like the ways that you just divide the processes like you pre-process the data and you can actually you divide the gross 16 Pro

32. 18:17

    generation in separate tasks and you uh prove them like run them in parallel and then you're able like to agregate them and final actually it's like division of the gross 16 improve generation uh why it's important because you're able like to have a separate web workers in browser uh that will have a dedicated uh virtual CPUs on the laptop for example and they will not share only one like worker okay

33. 18:49

    yeah it makes sense uh I will I will get back to you and and I'm also curious to to know if you have the name of the projects that you've seen because maybe I can bring it to my team so that they can uh uh look to it one yeah well like I think uh yeah let me find and I'll send uh here in in the group yeah we've seen that one that is really good is uh rayun or Railway and rail gun they they have really fast proof generation not everything is open

34. 19:27

    source there uh but we know that they have like different circuits depending on what you're actually trying to do they have more more than a a single circuit so that sometimes you can do things faster using a smaller circuit but since their code is not all open source uh we could not like completely find out but they have like really good proof generation time what just Al one one quick question what's the input what's the cryptography

35. 19:59

    of the inputs to the proof is it ecdsa or Bill uh I'll have to get back to you on that one okay just curious I I just dropped the link for D web gpus oh thank you I will save it here yeah but I'm not sure that uh like on which stage they are now so I'm not sure whether like they're close to any like demo concept or it's like just the research because it truly will depend also not uh

36. 20:33

    like I'm not sure whether it will depend on devices but uh because um the gpus itself are quite different I mean on the even on the laptops uh and so on uh and that's why I'm not sure how to how to properly they will be utilized yeah one thing that ask a question regarding he said that the rail gun guys have a fast generation proof like uh what kind of uh in which environment is it like on laptop or on mobile

37. 21:10

    device uh so one of our devs he tried on mobile device and desktop and he's the one that brought the feedback that it's really fast to use uh I know that is faster than what we have for OPAC right now but our proof generation I think it would take 15 to 30 seconds on desktops so it's not it's not good good at all or or approve generation time but I specifically haven't tried uh rayun or rayway I've tried a couple of protocols

38. 21:41

    like HHO uh which is they on polygon and BSC they they have a good uh good uh Pro generation time but they've migrated their smart contract right now so I haven't tried the new smart contract only the old one but as I mentioned Railway unfortunately it's not all open source so that we can see what's going on there but with can you can you can you pin me on telegram yeah maybe oh

39. 22:26

    will but from the uh Anon a bit on what you were talking about the the proofs what we what we see from users is like for small transactions people might be doing that with their mobile but like if you look at tornado cash uh transaction history like they have the largest amount of transactions are on their 100 ethereum contract which is like a huge amount of money and we assume that people are doing that on a uh desktop

40. 22:57

    and if if you're transferring 100 e you probably have like a decent computer that can process like any kind of proofs and it's not something that you're doing like on a daily basis or at least I assume that people are not just moving Le here and there like on a daily basis so it's okay to take more time but for like small scale applications like oh I don't know I want to buy a movie ticket

41. 23:22

    let's say it's a movie ticket but it's with a private account and it C like5 to $10 which is a small amount of money compared to the 100 that is something that people will probably use their mobile devices for and that that is where we see feed for like a proof generating system not as much for larger transactions and Mike uh I'll try to find you at the group right now I already sent you a message in telegram oh great yeah regarding the uh

42. 23:59

    tornado I'm not sure whether even like uh it's desktop proving I think it could be also done like on remote desktops for like remote servers and so on like as a automatic like trading boots so uh yeah I think like in general torado currently is not a good example yeah yeah not at all like I think this week or last week there was like another hack and I think was on salanta and it happened and really fast they already transferred their their funds to tornado cash I think it's like

43. 24:42

    a different type of user even that we're looking for even though they're like really good us use cases they're they're still the they're fair share of unfortunate uh illicit activities but one thing that I've been thinking a lot uh I don't know if you guys got into this to discussion topic with uh anyone else but we were asked a lot about uh Hydra cache or OPAC and the NDC voting because apparently the users that

44. 25:18

    interacted with with opat they were block listed from voting on NDC and we kind of have our our view of it we we know like we don't have any we don't even have volume for listed activities but at the same time it my favor or not favor uh a couple of uh candidates by using Hydra cache Opa to fund new wallets and everything but what is your guys' view on it from my perspective like uh the

45. 25:55

    private voting should be uh like as a feature but like from another perspective I'm not sure whether like users will will be willing to pay like for for such Services I mean from the perspective that they're trying like to hide uh like uh who's voting for whom uh it's all time the question where like where and when to charge users because it's definitely not the case when they come and pay some it

46. 26:28

    should be some like more complex uh idea of of the value generation I think gar had like some also ideas uh about the NDC yeah hi guys yeah I do have some ideas I'm actually on my way home so I won't be able to speak too much but yeah sorry sorry about that okay okay no worries yeah so me and and Jo which works with me I he he joined a couple of previous calls we were sketching just having some ideas about like how to build a private voting system and we

47. 27:20

    realized that it's definitely not as simple as we thought I mean if you only want to have like private voting itself it's easy but the second part that we feel like is a big challenge is Computing those votes privately you know like how do you add up all the votes without showing it to anyone and only at the end of the voting process you disclose uh the total amount of votes and you also prove that the sum

48. 27:51

    is correct like that itself is a big challenge to to tackle there's an article uh that the guys from Mina wrote about it but it uses Mina like we really were thinking on how someone could build something like that on year and it is not as simple as we thought it would be yeah is that is that our is that our blog post about talking using elgl with s j yeah I I haven't you mean the the Mina one yeah yeah yeah I it's really private voting is

49. 28:31

    really complex yeah I was gonna say something like Alama with pet or some sort of Peterson commitment you can do this it doesn't even need to be a Zer knowledge proof well Peterson commitment is kind of but yes yeah the zero knowledge came in for there two things one is privacy preserving proof of Eligibility to vote and then I think the other was for proof of corness of the tally yeah but it it it is a definitely like if it was just to hide the the Vols

50. 29:15

    itself it's s relatively simple like we have what we have with like uh OPAC right now it could be used to but the second part is but it's a nice challenge to tackle and I I feel like it's it's probably the next step that NDC will take on voting Alo have done a bunch of um onchain voting they have a repo for how they do it if you want to look for it or I can try and find it yeah I would love to okay we also have a repo with

51. 29:51

    private voting uh it's pretty old one I think we we did it like year and a half ago uh the initial idea was that you basically deposit tokens on a smart contract and provide a private instruction with like option A or option b and uh on the execution date uh the smart contract transfer your voting to the uh to the um original uh W in smart contract uh but with the instruction like is it like option A or option b uh so you basically don't need to uh

52. 30:29

    reinvent the whole uh voting process you just did this like just before the you you have an execution just before the uh voting ends on the uh Target smart contract if you need we can we can take find it on I think it should be public also send a link to the last year hackaton submission for the private voting yeah in my opinion growth 16 yes that's nice guys let me ask you uh the first topic with the Privacy use cases

53. 31:33

    on your uh are we waiting for anyone to join from Zer pool or yeah I'm not sure understood they're not joining yeah yeah so uh yeah if we don't have any other topics and nobody wants like to trace and to talk I think uh we are all done and yeah yeah and we canot end up yeah voting is like too big of a topic maybe yeah yes attention yeah let me ask you guys uh will will any of you be at

54. 32:15

    newk Clan this year since most of you are in Europe so it's closer yeah I'm planning to oh that's nice probably not but I would like to but yeah family issues I'll be um on a booth there so come say hi we are unveiling some things definitely uh me Joel and Rafael uh will'll be joining from Brazil cool we went last year and we're like oh we got to go back there so we kind of decided like on a last minute but we'll

55. 32:53

    be there to chat what what project were you with because you said you were on a booth right yes which which project or which I work in pagod but um okay we will be unveiling a bunch of stuff I've been working on so um it'll be cool if if anyone is interested to come say hi very cool yeah I knew the protocol team had something in the works but I've only heard like the name so anyway I'm

56. 33:44

    excited to see what's been going on all right well if there's nothing else um I suppose we can wrap up here right and um yeah Che guys yeah thanks for the community yeah thank you so much all right thanks everybody everyone always nice you have a good one bye bye