zkStudyClub: PrivacyPost-Quarantine Info Gathering Session

Transcript

okay so yeah I want to say thanks to everybody who's here for the ZK study club where we're gonna be actually looking at privacy post quarantine in a way I want to use this time to explore opportunities and challenges that are going to be facing privacy tech as we move into a post quarantine state I just said this before we started recording but I'll repeat it sort of the goals of this session are to identify what are the projects actually emerging in different countries and potentially States around this it's not purely ZK that we're looking at it could be other privacy systems that are related or maybe maybe there's systems that are

emerging that could use it and aren't yet I think the goal of this is to have a little bit of feedback I think Daniel made a good point here where we can also cover or we can also have as a goal what are the most difficult barriers to have adoption of these systems and yeah and then the other purpose of this is to highlight a little bit who here in the group is working on what so that we can get a bit of a sense of what we could do after this in terms of more study clubs maybe a podcast stuff like that so I know that Henry I was going to kind of ask if you could share a little bit about what you're working on because everyone you know we're sort of aware that you have a project and then we can sort of open the floor to other projects

that people would see maybe the best way to sort of lead into this is by giving a little bit of history so a few weeks ago I put a blog post on the Z cash foundation blog basically saying that contact tracing is sort of one niche sub area of coronavirus response where cryptography is actually potentially

useful and as a result of that post we got in contact with Scott Lee brand and Dana Lewis who have a project called Co a B which is community epidemiology and that's trying to do like decentralized self reporting and symptoms and they had already made contact with another group called kovat watch that is in some way affiliated with Stanford or you know

some people are there who are trying to do a corona virus alerting app and they had already agreed that it made sense to try to have a common protocol so they were working on that and at that point I started helping with that and the result of that was a protocol for doing contact tracing called cem and then turned out that there were all these other applications that were also trying to do this and so the kind of common protocol level was or

the common protocol level work was spun out into a new arc called the temporary contact number coalition TCN and so we have a protocol there's like a you know ODOT for version on it so it's still unstable but it does work and it lets you do basically decentralized contact notifications and it's designed to be extensible so that it's compatible with a bunch of different use cases for instance you could use it to do only authenticated test results that are you know signed by some health authorities

or you could use it to do self-reported symptoms but still be interoperable what like so this is coming out of these from what I understand it's kind of like somewhat independent organizations but is there a is there like an adoption path is there a way for this to actually be used is it it is it going to result in an app or will it be integrated into something yes so there's there's a number of different there's a number of different application developers and the the effort for doing a common protocol is basically has come out of the desire of those app developers to collaborate

so the path to adoption of the protocol is just the path to adoption for all of their constituent applications okay are they are they go ahead sorry Henry out of curiosity so I'll be really doing the self reporting and doing some machine learning on top oh is it just I have a cuff of a fever I'm just reporting to report this and then you have the result you know have the curated data set to actually do something but their toes it's just contact tracing per se security requirements are both are very different that's why I'm asking so I think that

it's better to so traditionally write contact tracing normally refers to like an activity that can be done with various tools right yeah I'm a doctor among other things so yeah I just don't fully a bit of all know I just I know for you just to kind of like to lay out all right so I think then for this system it's really better to think of it as like contact notification rather than contact tracing and I think that there are different possible use cases that

are each valid but that requires setting different user expectations so for instance the co at the project is explicitly just to it just to do self reporting of symptoms not just for coronavirus but also for like flu-like illness okay and and she's focused on doing authenticated test results and so the protocol is aiming to be compatible with both of these use cases even though they have different security requirements as you know yeah so it's a question for you so essentially there are existing so I I

know if somebody you understand furred who actually runs influenza what do you call that watch thing they're like a bunch of people who are doing that I'll be thinking of actually offering them this possibility of reusing whatever we end up building what you end up building so we could cut in here that you do that because it's an ongoing process this is yeah okay I'm not as I have a kind of a peripheral understanding of everything that's going on there yeah because even just trying to stay on top of the developments for the protocols is challenging but my understanding is that the goal is that

it would be nice to be able to to reuse some of this work and I know that for the Coweta team specifically one of the side goals is to have the to be able to build anonymized volunteer volunteer datasets other symptoms that could later be used by people to do official early studies there's some other there's some

other kind of systems that we that there may be other members of the the group here are working on or are aware of in their particular regions sort of want to open them the floor and you're all you're probably muted so you unmute yourself they I know that the European Union or in Commission had had an actual fund that the application was over like a maybe a few weeks ago to explicitly have like funding of projects that are going to do like things related to coronavirus right unnecessarily privacy related but I think they were probably several submissions around that there

was I'm gonna say one of them there was this hep PT pan-european privacy-preserving proximity tracing they really like the peas in this one I'll stick the link this this is it's got a website I've started to explore it it was kind of brought up actually in the podcast I went live today we did the interview last week and this was sort of mentioned in it does anyone know anyone working on that or is anyone familiar with this project oh yeah just unmute yourself if you wanna think so I mean I'm not working on that but I have I mean I have been

following this a little bit so as I understand I mean it was some kind of there were some kind of a group of universities working on this and and there is this protocol called VP 3t yeah yeah that somehow resulted out of that but it's not exactly the same as they and this project but somehow like the centralized version of what they wanted to do so apparently there must have been some kind of disagreement internal

disagreement or some kind of different focuses of the people working on this project and so this PE e PP PT is like this craft or some kind of umbrella for different projects but the only project that I actually know is this DP sweetie I think maybe there is some kind of other centralized version of that yeah I actually I can find a link on that [Music] poster so paper recently that actually ripped the a guy from EPL then take too kindly with with the DP 3t technique because I'm on like it's privacy

preserving but the service condom the way said of the service company so we were seen as not strong not strong not having the strongest privacy guarantees but not being as good as it claims to be hmm I think that might be is this what I just shared this analysis of DP 3t it just came out on April I'm a $23.99 yes I said that's it yeah that's actually shared in the group too so that's interesting yeah we actually had a few talks with those professors we had a thread discussion and yeah they it was kind of hard to you know highlight

the the issues with that protocol we even suggested like different one but it seemed to be not leading anywhere was this was this like you were you were meeting with them because you had seen it and he reached out to them or were you in this or were you in the working group we as we we saw it and we wanted to contribute and we wanted to present like an alternative one body didn't manage to we didn't manage to make it go through so what happened in the end we found another professors from ETH and that yeah now we're writing a paper

they're writing a paper and release it soon okay it's a different you're gonna end up working on a different project in a way or yeah yeah it's not perfect like I can give you an overview but yeah but I didn't want I just I just join actually this is what we're trying to kind of do right this part of the I think you missed the the sort of intro part where I mentioned the goals but one of the goals is to start mapping what are the different groups what are the different systems that are being introduced among like that we know of among this smaller group of the ZK study club and this is sort of a starting point for us to discover it so if you are working on one of these systems I

think it's interesting to hear what it is okay but so super short intro my name is Matteo I work at Google but this doesn't have anything to do with my work at Google however if some of you comes up with a super cool scheme or whatever you know given my advantageous position I could bring it to the attention of relevant teams in Google inside Google because as you know Apple and Google are you know collaborating on on implementing content tracing so yeah okay so this other scheme it's it's basic is based more on a symmetric and by the way I'm not a cryptographer I do mobile development so I will say a lot

of incorrect stuff but basically it's about using public key cryptography so I symmetric keys so if I if I if I if I if there are two people in a room people basically they just advertise their public key which is 32 bytes with let elliptic curve cryptography and then if somebody gets infected so they want to tell other people that they're infected they take all the public keys that they've received and they encrypt a message containing also some random randomness it says hey I'm infected and optionally they also include the time of

the day and you know the location and they send it to the server and they always add it with random data so that whenever a person says hey I'm infected this blob of data of these messages hey I'm in fact that all encrypted with with public keys of people it so it has always the same size so that the server cannot infer how many persons did this person in fact so the server doesn't know who this person is talking to and he also doesn't the server also doesn't know how many people yeah this person affected however the people that receive

this feed of infections they attempt to decrypt these things and if they manage to decrypt one of them it means hope I had a close contact so I could be infected and also to prevent inflating fake contacts because you could be stealing someone's public key and advertising it all over the world you want to also send a signature of the and timestamp with with the public key that you broadcast so this is how it would work and there's a paper that's coming out and this is the project is

there a name for this yet or yeah it's the we trace so we we were the winner of a hackathon it's called we up was called we trace let me paste it to the chat there's already a preliminary paper but it's like a really disgusting draft but let me find it we trace so the so the app was we trace dot CH the code is really incomplete but let me face the we traced on ch got it yeah and and the white paper is here this is the draft so that that ignorant people like us without cryptographic knowledge growth

and and the new one will include all these other additional things that I that I told you about that's actually that was another point I did want to touch on this kind of Google and Apple announcement that they were doing this they were doing something I wondered if anyone had any insight into what they're actually doing and if there is actually any any privacy technology included in that or if it's just they're rolling out updates so that you can push your information like that see like I was kind of trying to go through the articles that I've seen to see if I could find something but I haven't yet even if I had some even if I had some insights of course I wouldn't be able to

share them but it's all public luckily so Google and Apple they share the spec and the cryptographic protocol online so I can paste it in the chat and it's actually the most simple worse protocol you've heard out of all of the one that you what is there I mean it's just this idea broadcasting and it's it's it's susceptible to close contact fake close contact inflation if you want to call it so basically you know if if I take the idea of a person with my antenna in the main station and then I have an another antenna in another city in Switzerland I

can just rebroadcast this ID and I can basically kind of like fake close contacts and and and block our country's economy because everybody will have to quarantine themselves because they got the notification oh you you're probably infected so it's based on that do you have if you can share that that's awesome and also if you can share any of the rebuttals or any of the analysis that would be interesting to under on the plus side away from Krypton what I got up for me is that um and I'm not mobile developer native applications accessing Bluetooth in the background is non-trivial and this update gives gives

us that for free so Wow Google and Apple do what they want it allows anyone else Lubin and concentrate enough to leverage not really so I'm an iOS developer so I looked at the API and and it doesn't really open up background more identities now it doesn't really open up Bluetooth more than it is now for for app developers they they instead have a API that allows you to start this contract tracing but the protocol and all of these things you can't change anything there you just have to you have methods to get the IDS

that you received and you know like really basic stuff but it's not that you can build a new protocol you have you're bound to use with the to to work with the protocol by Apple I'm finding the iOS classes and and crypto scheme that they are proposing and testing facing it into the chat now cool is there any other projects that have been have crossed any of your radars that you want to mention here and if not I actually have actually sure yeah actually I just remembered that some people at Cali bra we are we're using the protocol that actually was submitted to Zika proof

called proofs of liability to do some kind of work on proving Corona a em it's not contact tracing it's actually when when you're claiming a certain number of infected people in an inner region you want to prove into your knowledge certain properties of that claiming something I think that's one thing we're gonna use it for so it's a bit of a different use of cryptography for coronavirus but it's still quite interesting and so one event that might be of interest to the people in this group which I'm gonna paste a link to is there's a very rush organized seminar organized through MIT called impact that

is going to happen tomorrow online and so if anyone is interested that might be a thing to consider going to or you know sitting in on so there's there's discussion of like different perspectives on contact tracing and and the last panel is going to have a discussion of the different cryptographic some of the different cryptographic approaches that have been

proposed so far nice that sounds very relevant so Henry I see from the link he sent in the last panel that Ron Kennedy is talking and apparently he's got this private automated contact tracing I talk to it does that mean that he's also working on something or is he part yeah there's two there's two schemes called pact I don't know exactly what is going on there but yes colonel cool Anna did you add to the least one by ever him the presentation that I sent I guess

doesn't exactly cover it but I didn't what did one eat that yeah okay so let me find the link the exact link and I'll send it sure that sounds good so about these pacts I was reading the one from the coast east because apparently there's like a East Coast and West Coast pacts and the East one is quite similar to the one from the European consortium there are very few differences the West Coast one I don't know and I think that it was just a coincidence of names but

I'm not really sure okay I think we can continue to I mean I think we can continue to like bring it up any projects that we're thinking of but I did wonder like I had some just some other questions and I don't know if you have answers to these questions but these are just sort of thing that I was thinking about in the systems that we've listed so far maybe we can focus on their like but the contact tracing that Henry is working on and this hep PT are there actually any is there any zero knowledge technology in there or are those something else and if

there's something else what are they usually at least in the TCM protocol that I have been helping with yeah um the the only zero knowledge component comes in the form of a signature just very advanced but essentially in both this protocol and the the AG protocol and some others you have a system where you're basically you know you broadcast

random identifiers on Bluetooth and then at some point later you decide that you want to send a notification so you create some reports about the identifiers that you broadcast and then upload those to a server everyone can download them and compare them against what they observed and one idea that you can use to make this more bandwidth efficient is you can derive lots of ephemeral identifiers from some symmetric secret and then only upload

that sort of seed material and in the TCM protocol what we do is we also bind the generation of all of the ephemeral IDs to the verification key of a signature scheme and when you submit a report you produce a signature and the way that you can think about this working is in the in the case where you just have some seed and you do some symmetric crypto to derive all of the ephemeral identifiers or the temporary

contact numbers when you publish your seed you are proving knowledge of the of material that was used to generate all of the temporary contact numbers but doing so in like the most trivial kind of proof of knowledge where you just reveal what the witness is and so this means that then the clients for this to be secure either the client or the server has to implement stateful replay protection because otherwise someone can include numbers that were previously reported in their own report and so having this having a signature and

binding the generation of all of the TCMS to the verification key of a signature scheme lets you have an additional sort of zero knowledge of knowledge involved even though it's just in the simplest form of signature and that prevents sort of resubmission and lets you basically it's not necessarily important for all of the clients to verify these signatures but if the server is verifying the signatures on report submission and that lets you basically have the server stateless Li verify that only the person who actually generated the relevant

numbers is the person who is submitting them okay that was what you were trying to protect for yeah it's just that it's correct so we actually wrote up on a list of properties for ideal functionality and trust assumptions which I'll just paste into the chat and I'm actually gonna I'm gonna record everything that we have because there's a lot of good links here and I'll share that after so at the bottom of that section there's a list of properties and ideally the protocol should have the following properties and we we split it up into source integrity and broadcast integrity so source

integrity is the property that users can't send reports to users that they did not come in contact with or send reports on behalf of other users and having this signature so if you do the the symmetric crypto derivation procedure as in the AG protocol and others you get source integrity only until the first report and adding a signature gives you source integrity always and that means that you don't have to worry about reflux no in the PE PE p PT PT 1 is there or the DP 3 t is there any journal auction

that is what systems are they playing with so I think DP 3t is the actual protocol and the effort that has had like sort of more substantial work put into it okay the kind of I it's unclear there's some kind of umbrella organization I'm not sure it from my vantage point it feels like the DP 3t effort specifically have done a lot of work and I'm not sure about and what's in that does anyone know what's in it what do you mean how

the protocol functions like what kinds of systems are being used in there that we might be familiar with I mean it is a femoral ID so yeah is it similar then to what you're doing Henry I think it's a bit worse than what Henry proposed so I think Henry's solution probably is better so yeah they also have these these secret seed that generates all these where is it where they derive all these ephemeral ids from and then they broadcast them and then at the end they upload the seed that was used to to generate this ephemeral ids to the server so that people can then you know download that

and yeah basically check like read arrival use ephemeral IDs and check whether they they have any any stored ephemeral ID that they received it's music now am i confusing this with the one that actually requires a TPM to prevent like a false reporting one or which is the same thing they mentioned DBMS in their paper I think as a future sort of additional layer of protection I have been not sure and another point that I would make about DP 3t is that they had they did a first iteration of their protocol and I

they have also done a second iteration that I think uses bloom filters in some kind of a way okay actually they're actually using cool filters who what are they called Maps doors yes but I have not at all I have been kind of swamped so I have not yet come up to speed on the latest version I saw that earliest version is it the one by Sergey Watney in the records with yours name whatever his name is analysis of dp3 yeah I just posted in there yeah I had a look at that one here

but anyways the reason why I am really thinking the this dp3 method like I think the biggest limitation is the is this attack that I was talking about it about before which is inflating inflation of contacts let's say as I said before you put up an antenna you receive all these tokens you immediately relay it in another city or you know in other parts where you have an antenna or if you're a group that wants to attack this everyone has an app that rebroadcasts all of this and it's basically generating all these close contacts that don't exist and you could attack a country by making this system

ineffective so the contact relay attack or I don't know how we want to call this this is the the major problem with with that solution in my opinion there is some there are some proposals to mitigate that there was a print of the is er my list of Pietra that basically the idea is that you include some information about the location and time when would be the interesting thing I guess yeah you could have a window then the deep layer talks are very timely about that you are sending your you're a female ID and at the moment

where the other person like receives it then the other person needs to check whether the location and time that are hashed somehow in information that they correspond to the one that they shoot so actually you're then you don't need to send your location and times to their to the server that's at any points if I'm correct because I'm I'm not sure I understood how that would work because if you just send a hash so so basically you you sends the information with the time and location and then you send some hash that includes the time and location and then the person that receives that that will check against its own time and

location and see that's that it's actually correct at the time of receiving the the familial ID so so you can actually check that's that's it has the correct information in it so that then you may not like replay you getting an ID and then sending it to some somewhere far from from you yeah and then will receive it there we'll check that actually that's not consistent with the location I was about to say even this attack is known in iff as well so

you can actually have this attack happen it just historically they're Israelis in this call so they will explain it it's publicly documented attack on AFS so you could probably do you know Windows as I'm like very tight window then you can actually have a loop and then you can't back so now would be the easiest solution to this exist impossible but it needs to be to happen in like very efficient yeah to close by and yeah but how much data would it add to the advertisement packet don't think it's discussed because I don't think that the

proposal actually that at least if this is the same you print that came at quite recently I skimmed it only briefly but I don't believe that they actually considered the physical constraints at all yeah because we are pretty physically constrained to 16 bytes that's the advertisement data that the Bluetooth specification supports so whenever you're broadcasting something you're broadcasting you know this isn't that and then you have these 16 bytes where you can put something inside there and I still didn't quite understand how that would worry about it I also

understood about another solution which is not ours that would mitigate that attack but if you wanted to say something else before I wanted to make one comment about the earlier paper the 2020s 399 I found that kind of frustrating to read because it was a parts of it read sort of like you know when someone is doing like a vulnerability analysis and they kind of like throw in the entire kitchen sink of everything that they could think of and it was kind of hard to parse out which of the problems that were identified

were actually problems I'm like well one which ones were just sort of complaints so for instance one of the the problems that was supposedly identified is like oh well someone can impersonate the server and like you know show up at some broadcast numbers to them and then send them a report while they impersonate the server and then like oh this would be addressed if the server were is authenticated and it feels kind of like well you know the people who are writing up this protocol like maybe they didn't include explicitly the sentence that like the server uses HTTPS because that's just kind of like a baseline

assumption and and it felt kind of I was a bit frustrated that the I did not feel that the attacks were really all at the same level of practicality and severity Antonio just added another link in the chat here that is another kind of attack paper on deeply deeply attack interesting and that just came out two days ago I think I want to there's some people in the chat who haven't spoken and I just want to open up the floor to

them if anyone else wants to sort of weigh in about what we're talking about I mean for anyone who's new newly joined we've been covering a lot of the different systems that are out there and talking about some of the pros and cons trying to explore a little bit what's in them and obviously we can't cover everything we're just like a small group so we're just kind of pooling together some of the info we have I know one of the other topics that Daniel you wanted to cover was about the barrier to getting some of these things adopted but yeah I wanted to kind of open the floor to everybody else here if anyone wants to mention anything if there's systems that they've heard of that we haven't mentioned yet or if there's if there's other ideas around those that we have or

if you want to talk about this adoption question I just want to throw it out there ah okay so Antonio actually is having a bit of trouble with his mic but this is the paper and I think yeah I think we kind of work we're all exploring that paper now and Antonio so thanks for sharing that I'm sorry that your mic isn't working and you're all muted so if you do want to say something you have to unmute yourself just FYI I was just

taking a look at the paper that Antonio posted so so I'm particularly interested in how this technology gets implemented especially in less more fractured health ecosystems because I'm I'm from Nigeria and I looking at it and it's very interesting when I think of the MIT reality of implementing it it's like this hospitals don't have and not gonna run a server or they or they just really like it it's like it's very good but like them they're the most fragile economies and how does how does them contract

trace it as it stands today lend itself to to the developing world because our problems are very very rudimentary only finite immersed in Merstham cases no fight no fight things work because I'm thinking man Antonia said that more someone mentioned that when a poem when we did get this upgrades they will be on like the latest Apple phones for people who use feature films like chaos or yes it stuff that's really go in there so how do how do this Publius implementations does anyone know anything that's working I work in in developing world yes I could

so yeah I probably would put my hand up so I was born in Kerala and I went to med school there so I know what's happening there they actually got it is it's part of the reason I wanted to get this calls her there's a lot of controversy going on there's a US company that was hired to analyze the data set so essentially they got the cell data you know every device has an MC and then you connect your authentication protocol actually revealing some information you can collect on that and that's been actually given to somebody else and this is a trade off question and they decided they will actually do that I mean you know at that moment in time they thought that is

the best possible decision maybe I don't know I mean this is like a very hard problem that we have we have a bunch of solutions which are very pretty good at least in the part of being pretty good but getting it implemented in places where a is a resource-constrained and B it's pretty hard for the people in that part of the world to do anything it's going to be tricky because the people who are involved if we had a solution today I have a channel to the decision makers I was on the phone with them yesterday so if you have a solid solution I am happy it Connect who to the government the state government in

Kerala since the state of Kerala it's kind of an anomaly of the whole of India if you have like pretty interesting healthcare statistic and we have a lot more doctors for a thousand a population they about the whe limit I mean double is your requirement and we had the first set of go with infection imported from Wuhan and it very few fatalities but it's an exception even their privacy is something that we we had to trade off so if you have a project I will happily do the intro it almost feels it feels often like as as I explore these and maybe everybody feels this way where it's like there's there's

some really good potential solutions coming up coming up but they're not quite at the point where you could start to put them on tiny devices or like non smartphones and even on smartphones they might not like good up-to-date phones it might also not work there and so we're stuck with very with with systems that are definitely privacy imperfect good to give an annotation Carla would you say privacy is what trade enough in that kind of situation because here is is a critic like privacy and decentralization

their higher virtues well if it goes pretty places like an in Lagos there's social determinants and people do not care about quarantine because I'm day each day to day so every day today go out to the daily stuff so like maybe maybe it's an encroaching out or scope creep on this conversation why is that when we effort to build holistic technology solutions the social determinants or [Music] to put this in context the whole of

India when they had locked down they were like millions of people on the street so there is no construct of social distancing even happening and the other thing we should recognize it is starvation kills 100% code virus has a fatality in a person that's right even if you firstly you have a probability of getting infected and the second thing there's a fatality which could be 0.75 even less than 0.3 to whatever single digit so this is a challenge like a lot of those things are trade-offs so I don't know an open question for everybody I think that in way is the like that's definitely a topic that we

are sort of talking around in the study Club like wanting to talk about it so it's such a big one and it's like it's very subjective I think there's there's also a lot of emotions right now about that like some of those topics because it's like it's sort of the utilitarian good versus the individual good and it gets really complicated and emotional you can you can feel that with some some people and rightly so like it makes sense that it is but Mary oh you want to say something yeah I mean am I right to assume that I mean after Snowden and so on I mean I guess it's fair to say that

I guess phone companies Apple Google ad networks secret services they basically have all the information already right they're just not using it or if they say that they don't want to acknowledge it right it's like once they use it there's AK Norris right I guess if you if we don't applaud I mean I care a lot about privacy but if we say we don't care about food EC then can we establish that like a trivial solution would be to just make use of all this data that's actually already there you just marry oh the way I think about it is like it's you know it's countries acknowledging practices so I wrote a paper in 2011 describing some of the challenges and

the challenge was like nobody wants to acknowledge this nobody in any of the various groups okay would even if you meet them and you talk to them and they know you really well they would never across existence of any of these things and I'm pretty sure you all have similar experiences I've at least met the UK equivalent because my PhD mate works there and I've met a few senior people they would never acknowledge any testings everybody knows that's the truth right but they don't won't acknowledges they don't want to release the data set so I mean they don't even need to get to that level that the cell level info is good enough for us if they could get the MC and just track the whole thing out you don't need to go into a lot of detail but you know

even that they don't want to be revealing I don't know Daniel I was wondering if since that was one of the points that you had sort of brought up this barrier to adoption and since what you guys get it what you work on is getting enterprise to adopt some of this stuff so wondering if you have any thoughts on it I don't know if you're still there I hope you are you're muted right now yeah it's a great point so I mean I think it's a bit different in the sense that this is really adoption not in the context of enterprise it's really about consumers adopting this right so it's a bit different but I think I have

a few kind of concerns here I guess the first one is really to make sure that what is being adopted is actually secure right I mean this is I guess a typical concern in cryptography like I just saw that the European Union published the European Data Protection Board right what the Commission that took out the gdpr took out a a letter saying that they would support something more like dp3 ET right and we were just discussing that it's maybe not the right way to go as a protocol for privacy enhancing contact tracing so this is one of the first concerns or challenges that I see another one is actually in the lines of

I asked earlier like if we see that there are so many different protocols how do we make sure that you know we're either that they are compatible that they're interoperable or that we can actually make them like unify them into one right I mean Henry's working amazingly on this is the end to make sure that a few different projects came together and built one unified protocol but we were saying that there is still other different protocols out there that are getting traction and then of course there is the final question of I think on each point two trees in the beginning of whether we're doing self reporting or we're doing on actual tested tested and basically it means that in in the two

different settings we need two different kind of protocols for them that are like basically how do you get people to react from the different settings so that's like it's two different concerns yeah I was actually about to wrap up as well I want to say thanks to everybody for joining us and I think like I I don't think this is the conclusion I think this is the entry point for some of us for some of you I know you've already been working on it but thanks for everybody for coming together yeah and we'll continue it's for organize you

[Music]