ZKPodcast: Mixnets and privacy technology with Claudia Diaz from Nym

Transcript

[Music] welcome to zero knowledge a podcast where we talk about the latest in zero knowledge research and the decentralized web the show is hosted by me Anna and me Fredrik this week I speak with Claudia Diaz from university of liván and the project NIMH about mixed Nets the history of privacy technology and how there may be some opportunities for new privacy tech to emerge in this post pandemic world but before we start in I want to say thank

you to this week's sponsor aztec aztec is the company behind a super-fast privacy network on ethereum this has been live on main net since January they're also the team behind plunk the new standard in Universal snarks in fact you may recall we actually did an episode on pong just a few months ago so Aztec makes private transactions easy for developers via its privacy SDK this removes the complexities of ZK proofs and note management letting you focus on building your DEP with Aztec network you can build custom private assets as well as shields for ERC 20s the team is now developing a zk zk roll-up or as they've called it a Z K squared roll-up it's an

upgrade for their network and it adds id privacy and improves throughput so as tech invites you to integrate privacy ensured app already through the SDK and Doc's they're all available at aztec protocol comm you can also email hello at aztec protocol comm for API keys this would give you immediate free private maintenance actions you can also follow aztec on Twitter or join their telegram channels for updates about the project I've added all the links in the show notes so thank you again Aztec for supporting the show and now here's my conversation with claudia diaz [Music] so today I'm sitting with Claudia Diaz who's a professor at university of liván

and the chief scientists at NIMH technologies today we're going to be talking a little bit about what NIMH is we're gonna cover mixed Nets and hopefully we'll get a chance to talk about the importance of privacy thinking and privacy technology especially in the light of this kind of current need for things like contact tracing or other tracking techniques during the pandemic that we're currently living through now so yeah welcome to the show Claudia thank you for having me so let's find out a little bit about you and your work what you know what's what's your background what are you doing now you sort of you have two roles so maybe you can explain a little bit about how that came to be yeah so I have an academic

background my degree I'm an engineer my degrees in telecom engineering I did that degree in in in Spain in the University of Vigo and then I moved to live and to start my research already in year 2000 and my topic was actually anonymity in in digital systems so that was year 2000 at the time privacy was not so much on the table as it is today it was before Facebook and Google was only starting actually so yeah things have changed a lot I have seen therefore how you know online systems have evolved and how privacy has not particularly gotten better I would say and in the systems that we have today so my my

topic is mainly trying to design technologies that support privacy that provide people the ability to take advantage of modern technologies to be able to talk to others to access information to do all kinds of nice things without necessarily having to give up all the information in a way that can be exploited by others and what about how did you get involved with NIM what is NIM maybe say what that is because we've never had anyone from that project on the show so named technologies this is a startup that has been working for more than a year and the goal of name is to develop a system that supports

privacy for applications it's actually we are actually building a privacy infrastructure that other applications can can use in order to offer privacy to their users so the the story here is that so as I said I start I've been working for a long time in in privacy research and specifically in anonymous communication system some privacy technologies and about five years ago we started a project a European project called panoramix and panoramix aimed at advancing research in NICs networks by taking up work that had been done already in the last two decades that had not been advancing so much since about 2005 because once we had one store

became like a deployed system a lot of the research efforts went into analyzing tor and improving tor and attacking tor and building confirmations for tor and therefore mixed nuts which is an alternative design kind of fell a little bit out of attention but the issue with tor is that i mean i can talk about that later but the issue with tor is that it's very good for web browsing but it makes some sacrifices and in particular because it doesn't introduce any latency it is not resistant towards very powerful adversaries we can see both ends of the connection while in mixed networks they offer that kind of assistance at the cost of higher latency there's northern revelations of 2011 actually led many to realize the

pervasiveness of traffic analysis by the intelligence agencies and the realization that maybe tor wasn't enough to protect communications towards such a strong adversary so that's a little bit what led to the proposal of this European project the plate was accepted it was called coordinated by Aguilar she is from the University of Edinburgh and it we were a team of about eight or nine groups working in there so I mean we didn't really nice work in that project the outcome was prototypes for mixed as well a lot of research papers with new designs new features and but also deployed systems for different applications a mixed network for voting

and some mixed network for messaging so when we were working on this project in the last year European projects they asked you to to come up with some sort of what they call valorisation which means exploitation migration exploitation which means that okay you have done this wonderful research how does that impact the market now you know how does that have an impact in the real world and I mean we we had actually some some exploitation strategies already we had sa P was one of the partners they were exploiting some of the results the voting systems were being deployed and for the messaging mixed net that we had built we had this idea okay what if we

you know if we try to hook this up with some sort of blockchain system in order to make it into an infrastructure for applicate for applications that can use it to anonymize their communications and initially this was just some idea that was put on the table but then we got really really encouraging feedback from the reviewers in the European Commission and they said we think this is a great idea you should definitely go for it and that's how Harry who's the CEO of NIMH and it was one of the researchers working on this started working on it and then some others from the project have also joined this effort so it actually came out of our European research project cool what was that called what was the research project the

original research project called he was quote panoramix okay panoramix and you were looking at this mix net phenomenon when like at what point does it start like you you mentioned you kind of like we're thinking of putting it together with a blockchain but is is NIMH a protocol block like is it a cryptocurrency or is it a protocol that lives on top of block chains like how does that interact it's it's complicated that's right I'll try to summarize it so I mean what we're trying to build is an infrastructure for privacy as I said right and there are two main features we are trying to provoke to provide to applications that use this infrastructure one of them is metadata protection at the network layer

and for that we use some mixed nuts that basically D links input messages from messages such that it's not possible to know who's communicating with whom got it so that's that's one piece of it and then we complement it with with a system that enables anonymous authorizations so basically imagine you're going to access a system and you need to maybe pay to access that system okay so the problem is that in many unless you use like an anonymous currency in many payment systems you will have you will be anonymize yourself even if you're anonymous at the network layer right so we wanted to integrate it together now why not use something like C cash because Z cash is really geared towards monetary transactions okay so you can

you know you can transact money from one account to the other and you can do it in a very private way now because we are framing it more as a right to use payment might only be one of the kind of things you have to present in order to access a system just to give you some examples for example if I want to access adult content I might have to prove them also of age that I'm an adult I know if I want to buy alcohol I might have to prove that I'm an adult or I might need to if I if I am accessing a system that is let's say an anonymous forum for anonymous alcoholics I might need to prove that I'm actually a person that belongs to this group so

our our system our voucher system payment is one of the of the kind of authorization elements you can show that you paid the right amount and this this voucher is redeemable by their service taking it but you can also model it's very modular so you can attach to this voucher proofs of other identity attributes that might be relevant to make a decision of whether you're given a service or not or not only whether you're giving a service or not but maybe there are different price categories depending on on some identity attributes if you have a student discounts I'm a student then I would have to pay the discounted amount together with a proof

of me being a student and this kind of flexibility is not possible in C cache because it's not built for that it's just built for transactions I'm sorry I think I'm still not clear though so at what I understand right now is like it's a set of rules or like logic for how to have a private communication of some sort like a transaction or this proving of an age or possibly transact like moving of funds but what is the thing like is it is it a protocol set of rules is it a piece of software is it a blockchain it's a complex system that contains many pieces okay it does contain a blockchain okay

because the blockchain is the where we keep the ledger to keep count of transactions of which tokens have been converted to credentials and back and we also use it as a channel to to broadcast network information that all client should have like keys of the relevant mixes that are active network parameters and configuration okay so that is the blockchain kind of is the broadcast channel secure block a broadcast channel that everybody can access in order to understand who's part of this network and how they communicate with this with that part but basically I mean from the perspective of an application like what can they do with this infrastructure I mean basically you would have kind of a little plugin or translator both at the

client side another server side and at the client side that piece of software would be in charge of taking the application packages imagine the application is a chat application or is like blockchain you know Bitcoin broadcasting transaction application whatever it is so it would take care of taking the application messages converting them to the mix net messages and sending them through and at the other side basically converting them back to the application so it's basically just a transport layer there is anonymous okay if on top of that you want to also have payments then the the client would first have to deposit token in an account get a credential for that

and then send that credential in those packets to the other side the other side would be able to verify that this is a valid credential and accept it and then later on redeem it and get the money okay does this are independent pieces like some applications might not charge to the user yeah and then they would not be using this payment system they would just be using the transport layer is that is the blockchain that you have underneath this is it a like is there a token is it like yeah I'm just curious how this cut like the company the project like their nodes running this blockchain as well or is it somewhat like there's a blockchain roughly but it's like proof of I don't know like it I'm curious here like is this like a

blockchain in the traditional sense that we've been looking at which is like very decentralized or is it more just like there's a blockchain to keep records of the privacy part so we the blockchain we are using thundermans actually at the moment so it's not that we're not developing our own I see our own component so we're using a proof of stake blockchain well not really actually so it's again it's complicated we don't seem to fit into this neat categories yeah a proof of work proof of a stake we do require what we call operators to stake so operators are the validators the evaluators are the nodes that that maintain the blockchain mm-hmm

so it like in like many other projects so you basically have these notes kind of my signing transactions and block cementing the blockchain so this is one operator the other operators are more the next maps so these are relays that shadow traffic right so the idea is that operators to become operators of this network they will have to put stake now the stake is not like in proof of state systems where you're paid more if you have more stake and you're paid less if you have let's take we actually pay operators on the basis of the work okay I do so we used to take a little bit as I guarantee as a deposit I see sure that they have incentives to behave well because if they misbehave if they drop

packets if they are not signing transactions if they are not performing well then they will lose the stake right and then the rewards are given on the basis of of work so for validators they have to be active and signed transactions and this is visible because the the signatures that the validators issue are visible to everyone so everybody can see if a validator is doing work or not for the mixes is more complicated and we actually have a novel solution to to measure whether mixes are dropping packets or not whether they are providing the quality of service that they and the bandwidth that they actually promised if they do then they receive rewards that are proportional to

the work they performed but this is not useless work this is not just doing hashes for no good reason it's actually the work they are doing is shuffling packets and sending packets that are from users so they are doing work that is useful to somebody but they will be paid based on that so the stake is more like so if you think of carrot-and-stick the validation is the stick and then the the act like the work the useful work is the carrot yeah our payment for useful work is the carrot interesting yeah it makes sense that it does I mean we've always like on the show we've always been sort of skeptical of what is it useful proof of work does that not create other problems like have you already explored what the challenges of this type of model are so the thing is

that is that these this work that the the routers are doing is not to maintain the blockchain yeah okay so it's separate yeah they the work they're doing is to provide a service and we want to pay them for providing a service for others they say okay so basically they are doing a job for others that others benefit from and the economic logic is that those who do work for others get paid and those who benefit from the work paid for it got it okay so I I think I have a bit of a sense of what the NIM blockchain looks like and I and I have a bit of a sense of these different parts now maybe we could explore one of these parts that we haven't really covered that much on the show which is the mix net now we've had we've had a group like

called tornado cash they came on their mixer like a mixer application that lives on top of aetherium using zero knowledge proof s-- now I don't think we've actually covered mixed Nets properly and so I think it might be good to understand what is this mixed net that's been now like now I think we understand how it how it's been operated but like what is it okay so the closest analogy to a mixed nut for those who never heard of it is a store so a mixed nuts is an overlay network so that means that in between two notes to two of these nodes are connected through the normal internet and communications that are routed through the mixed nuts go

router via multiple hops in this network right in the kind of mixes we use there are many types of mixed nuts the kind of mixed nuts that we use is what is called a decryption mixed nuts in the decryption mixed nuts packets are encrypted like an onion encryption so basically in layers you first in create for the last one and then you encrypt that with the previous the predecessor and then you increment the predecessor all the way to the first proxy that will receive this message and then you send the message each of them peels one layer encryption and forwards to the next and each of them is able to see the predecessor and the successor but they have no visibility of who was you know before or after so it's like so each

each each jump is another kind of encryption yeah well you the sender encrypts in decryption mixed nuts as I say there are different types in decryption mixed nuts the sender if I if I want to send a message to you and this message is going via three intermediaries I will perform four encryptions I see I will first link read with your key so first the recipient then the last mix then the middle name mix then the first next and then I send my package to the first mix and they decrypt they declare and finds fine who is next send it to the next they decrypt they find who is next and sent and so on right interesting so that that that's

what you mean by okay so do you actually do all of the encryptions in this at the start and then send it out and it's decrypted either what's another mix net model then this is the decrypt mix net yeah so the other side called they are called re encryption mixed nuts and in those you basically I mean there again multiple flavors the simplest flavor is that you have one public key for the full network so you encrypt your message with this public key and they use our crypto system that has some amorphic properties so basically the input the mix that receives my message and many other messages will rerun demise the encryption instead of

performing a decryption they just rerun the mais it ok so they rerun demise that such that when others see inputs and outputs nobody who doesn't have the decryption key is able to see the correspondence and you actually can do the nice things about this this the every encryption mix-ins is that you can do proofs of shuffles so you can show that you didn't add or delete any message that the outputs is a permutation of the inputs so you can do this multiple times and then you need to have some sort of decryption at the end and that decryption at the end is typically a threshold decryption otherwise you would have a single point of failure now of course it's a different system in the sense that it's much more computationally intensive

because you have general edge proofs at every step and also if the decryption is if that one decryption key becomes compromised they can be anonymous everything because the every mix has just rerun demised so they can just record the messages that were initially sent and the kryptos right I see but these are used for voting usually because voting is not so latency sensitive okay and it's extremely integrity sensitive you can't you want to be sure that all votes are counted that no vote was removed or added so this is why they can't afford to wait longer in order to have this shuffle proofs and show that indeed nothing is missing so this is those are the two types of mix net that

you just described now my first my first sort of point was like how does this then compare to something like a mixer so the mixers it's usually I mean tokens you send tokens into a mixer they're shuffled together with other people's tokens basically and then they come out the other side and you're supposed to be able to sort of in a way hide that you own those particular tokens like they're sort of insulting the concept okay so first the concept the concept of a mix it's an abstract concept okay so the concept of a mix is some entity some server and some entity receives a set of messages equal-sized perform some

transformation on this messages cryptographic transformation and then outpost and outputs those message in a in a different order this is a mix and this is the same okay for for a mix in a mix net and for a mixture of bitcoins and what is different is the level of abstraction because the mixer for bitcoins these messages it is taking our Bitcoin transactions and in the mixer for traffic packets the input they are taking is IP the process of transforming the messages reordering the messages and outputting the messages in a way that you cannot link input to output

that's the abstract concept is exactly the same the difference is that the the mix net will be operating on network packets I see okay so so it's actually a generic infrastructure because it's agnostic to the content of this package to the application that this package belongs to it just sees you know a data packet cryptographic transforming its reorder it with respect to the flow of packets you're getting and that's it while the mixer is basically taking this concept and applying it to the application layer for this you know monetary transactions and then the unit is no longer a data packets now it's a Bitcoin transaction going back to what NIM is though so NIM like would you

would you be able to use nimman to as a mixer if you wanted to like it could you take Bitcoin and send it in one side of this somehow and have like untraceable tokens come out the other side and I realize there might be some exchanges along the way but like is is that a use case for it so there's different ways in which you can use it you to make it work with the Bitcoin network you can use it simply as a network layer and the protection you would be getting in that case is that it is not possible for the people of serving your computer to know that you're broadcasting a Bitcoin transaction you could be doing anything else that is accessible through this mix net let's assume we have a mix net that can serve many applications

that the users you know are doing different things in there so you would be broadcasting your Bitcoin transaction and nobody would be able to tell that you're broadcasting a transaction and those receiving the transaction they are not able to see your IP address and they are not able to figure out who is sending this transaction so it would help sort of with the fear to peer broadcast level of of transactions to kind of conceal the IP addresses of who is participating and who is sending transaction this is kind of the simplest level the the more kind of fooling involvement would be to take Bitcoin convert it to token then you would have to convert that talk into a voucher by token you mean like the NIM token yeah

so you tohka take Bitcoin convert it to NIM token and that that you will have to do in exchange outside of a system then you put the big the name token you you exchange it for a voucher yeah and now this voucher you can spend it on anybody you can spend it on yourself as well if you spend it on yourself then you kind of come at the other side and say I have a voucher to redeem yeah yeah and then that would go to a new account and that would be unlink able provided that many other people are also making transactions so in that sense it functions a little bit like the shield that Z cash transactions in the sense that put converting token to to voucher and back it's a little bit the same concept as in Zika is going to shield it

and and and then back got it like you go into this big pool where many other people have their transactions and and then you don't really know correspondents and inputs to outputs and do you use your node proofs in this mix net or is it more like like is it more like just adding so many so much noise that you can't really see through it or are you actually using zero knowledge proof so both actually so at the at the mixed layer so in the mixed net itself we don't really use the knowledge proofs at the mix in the mix net itself is basically their protection is through random delays or adding randomness and through adding Danny traffic that conceals also real traffic patterns but

we do have the knowledge proofs in the issuance of vouchers and the redemption of vouchers because when you these vouchers are also anonymous we want to have I mean this is one of the things we're trying to achieve to offer this infrastructure that can provide support privacy not just that one layer but multi layer in a way that unlink ability is achieved both taking into account metadata and authorization level information interesting right so the zero knowledge proves you have to use them to obtain these vouchers to show that you have put in the voucher the right amount and then when you redeem it to show that the voucher was not spent already if you are adding other attributes to vouchers that are specific

to applications you might have to prove that it's the same owner of this attribute and this voucher so operations with vouchers do have as you know it's proofs in them the kind of traffic part of things is it has crypto but it doesn't have correctly but not zero knowledge and so would you so now like I actually can see the distinction here between like the mixers that I've that we've kind of had on the show before and a mix net specifically it sounds like as a mixed net is a there's far more types of things that can be transacted over it it's not purely for non tracing of token or payment or something it's much more broad generic is a generic communication

infrastructure and the thing with anonymity is that just as in mixers the number of coins that are mixed together gives you the anonymity set the more coins mixed together the more anonymity you have this because it's the same concept works the same at the network layer that means that the more users of the mixed nuts the more under Nieman you have and it's not only about quantity I also believe is about quality so for example if all your users imagine you have millions of users but all your users are using the same application then by virtue of observing that these users are connecting to the mixed net I can infer already that they are users of

that application now if you have let's say hundreds of applications that are accessible through these mixed Nets of serving that you're connecting to the mixin gives me almost no information whatsoever no because now I don't know which application you're using and within the users that hits those applications which one you are so so it increases quality as the diversity increases the quality of anonymity mix mixers for Bitcoin I understand have the problem that a lot of people who use them are involved in illegal activities and then if everybody who's involved in illegal activities is the one swapping these coins then you know you're just you don't your honor a nimitta said says well I don't know which criminal you are

but you know you must be a criminal because only criminals use this so you know if you if you end up in a situation like that then your privacy technology will not be yeah would not be very useful so what you want to have is a technology that protects everybody that provides default protection that is very broad and that is used by many many people because that's how you have good under nearly both in terms of large anonymity sets and diverse under innocence so from here where I like I kind of want to go back to that comparison that you had originally made to tour because would you say that the NIMH projects goal is to replace tor no I don't think that's that's completely accurate so anonymity anonymous

communications is a very broad design space just as you know cryptographic protocols are a very broad design space you have protocols for all kinds of things right so anonymous communication systems even though they have the ambition of being general-purpose the truth is that in order to obtain privacy you need to take some overhead so you need to pay some costs in terms of bandwidth in terms of latency and so on right and different applications have different tolerances for these costs so tor is actually designed very much for web browsing it's optimized for web browsing what does that mean it means that it has to be low latency it is connection oriented so you

establish the circuit and then all the communication go to the circuit because this is you know how browsers use the web low latency because people are not willing to wait for minutes to get their page loaded and low overhead because the problem with web browsing is that is the most difficult traffic to to anonymize it's extremely variable anonymity likes if you want to be undone lose you have to be somehow hide in the crowd be homogeneous in some way and because web webpages are so different in the traveling fingerprints it becomes very hard to just add some cover traffic and anonymize right so tour with all these constraints then tor basically has a system that provides nice and early MIDI as long as you don't have a global network adversary and the problem is

that if you have a global level adversary and you want to uh naanum eyes web traffic you you're in a very difficult situation so mixed nuts what they do is that they take a different point in the design space and they say we want to protect against really strong adversaries and to achieve that we are willing to tolerate some latency so it's okay if my message takes a few seconds to get to the other side it's okay if you know I have to generate additional traffic so basically what I would say is that even though you want to have very general infrastructures you might be in a situation in which you have depending on the class of traffic you need to use networks with different characteristics so I don't think we can replace tor for

web browsing I don't think if we wanted to do what browsing to be honest we probably would end up with something like tor because you know this is a good design that has been improved over the last 15 years but if you want to have higher protection for something like Bitcoin transactions messaging other applications that are not web browsing and that have more latency tolerance you are able to obtain higher security without sacrificing so much because your application is not so sensitive to that so what that was actually at least to the next question or it's like what applications do use see being used for name specifically so what financial applications is one one use case also because they use their

base of these applications is adventures and more willing to take you know be early adopters for new technologies and also because they have strong privacy requirements so they might be willing to take some latency or some cost in order to to achieve privacy so that's definitely one we would like to do messaging messaging is kind of a classical use case for for mixed nuts when I say messaging it can be email it can be instant messaging of course the latency for email the latency is more relaxed but for messaging probably a couple of seconds will be maximum people will be willing to to tolerate so that's a bit more stringent now with contact tracing we think that solutions for contact tracing for the virus might be a strengthened by submitting information

through a mixed net such that any anything that can be inferred from the activity or IP address about your positive or negative status cannot be inferred any more attack that what other applications I mean a lot of financial applications we're talking to the people who blog stream for lightning Network liquid so this we think that these are projects that have strong privacy requirements high stakes and might be willing to take on a little bit of latency or a little bit of overheads in order to achieve that I wanna I want to sort of shift the conversation a little bit towards the more kind of general privacy space because I have you on this podcast and given that you've been

working around this for twenty years as you just said the thing that I've been thinking about a lot lately and I don't think I'm the only person thinking about this is as we are in this time we're all pretty much everyone in the world the Western world is in quarantine at the moment and we're all in quarantine because of the corona virus something we haven't actually touched on fairly at all on the podcast so far but we're all in this quarantine there's been sort of signs that like you know extreme testing and tracking as they did in South Korea could be a really great way to maintain a low level of this virus you know contagion after we come out of the quarantine and there's all of these

systems being rolled out right now and I know of a number of like large data companies that are being sourced from all the governments to try to come up with strategies or apps or something like these tracking mechanisms to start using this technique for this post quarantine world so given that you're on the show given the given your background I what I'd love to hear from you about is like well partly like what do you make of it and and maybe do you have you seen like is there anything from the past that we could like look to in imagining how this could like roll out and possibly making it a little bit more privacy conscious because I mean the big risk here is we roll out and I think

people are pretty aware of this but you roll out some of these you know mechanism some of these new things and they are very very effective against this virus but then they also provide information that really starts to erode potentially like personal freedoms and freedom of speech and freedom of thought and like all of the things that I think people in the privacy space value very highly and I think we all I think this I think over the duration of this show we have actually had a chance to speak about quite a lot like the reason why privacy is important that's been covered but yeah so that's sorry that's a long rambling question there is a question in there it's basically have you seen anything before that we can look to and maybe

what do you think of today yeah so I mean the way they think about these things is that information is power so sometimes we want to we want somebody to have power to use this power for a good purpose and and some other times we don't who want somebody to have power because we are afraid they might use it to our disadvantage right so of course in a situation like like this where we all want somebody to have the power to to solve this problem it is very tempting to think okay if we just give up our information then you know some entity will you know be wise take all this information and tell us

to do and and we would all be you know you know better this I mean these reminded me a little bit to 911 so I started my PhD in year 2000 and my topic was not on the median privacy and when I was one year into my PhD 9/11 happened mine I have to say was a very difficult time to go up easy and privacy because nobody cared about privacy it was such a afterthought not even there was hostility hostility it was not that they didn't care there was active hostility if you would advocate anonymity assistance and make sense I was actually working on maintenance at the time right let me say it's an old technology you would advocate next mixed nuts and

people would be like ah but you know this is for terrorists and you know why do you support terrorists and you know it was really really hard to to make people understand that wanted to to have safety from terrorism doesn't have to mean that we have to turn into a super alien society right that's that's led to a lot of surveillance technology and that led to then business models that were benefiting from that surveillance information that was available and now we come to this you know surveillance capitalism I don't know if you have read the book from Shoshanna Zubov well I can give you the the gist of it basically she says that we have this system in which tech companies are claiming what is basically the human

private experience captured as data they turn this data into predictions of behavior this predictions of behavior is the product that they sell they sell this product to advertisers who then try to manipulate our behavior right so what she says I mean there's this soul thing that guy if you're not paying for it you're not the customer get the product what she says is you're not even the product you're the raw material the product is a B as a prediction on your behavior that somebody can act upon in order to make money from that right and when we and like that example of the companies I mean we've been aware of the companies doing that what's weird I guess I mean with your I want to hear

a bit more about the 9/11 experience because there it was more government's right like there you have this idea of like not not companies mining your data to sell you advertising but rather governments owning a lot of your patterns and predicting potentially predicting things and I think the scariest part of this whole thing is when it starts to bleed into public opinion voting you know how you pick your leaders how you pick what your leaders are gonna do like that's where having this I mean it's I think it's always scary it's scary when it is in the hands of companies because they're private usually or they're like they're control board is not voted in I remember actually finding that very frustrating

that you don't vote for the CEOs you don't vote for the people who get to control your data in that way but on the government side it's also really scary because there's just a different level of like power that's I think that that's a little bit of American European kind of cultural thing right like I think I think in the US people are more worried about government well if people in Europe are more worried about confidence well I'm actually Canadian so I'm sort of like somewhere in to be honest for me I don't think it makes so much of a difference actually I think in the end it had it comes down to who has power what are their incentives are there incentives aligned with yours what is

their agenda is their agenda you know the things they are going they are going to use your information to advance their own agenda and in some cases that might be beneficial to you or neutral to you and another case it would be detrimental right to me is more about the loss of agency that is a problem it's very difficult I think to really point to like often with privacy we have this issue of that is difficult to point to concrete harms you'd have to go like really to very extreme examples to make your points but to me it's really about a loss of agency and being in this kind of situation in which people know things about you they are using these things to make decisions about you you might not

even know that this is happening you might not even know in which ways it is impacting you and to me it just makes the our world less democratic and less free across the board right thinking about the difference between the the pandemic and the terrorism is that I think it was that surveillance drive was possibly even worse I think with 911 in the sense that there the idea was like Oh some people are bad and we need to monitor every one to find who are the bad ones hiding in the among the people right sector and it was really like what's in their mind yeah what's in the person's mind and not so much what's they are

they ill so it was today ever so it was really your suspect by default and that created a set of you know attitudes and dynamics that were really horrible in the pandemic I think the citizens are probably viewed a little bit less adversarial II in the sense that you're trying to protect them and you're trying to protect everybody but also the the person that you're monitoring so it is not quite as after cellulous are you might be the enemy and I need to make a you know really look at you to see whether you are you're one of the bad ones I can throw you in jail right so in that sense I think it might be less adversarial meaning that we might have we might we have the opportunity of

having solutions that I think don't distrust so much the individuals that kind of count on the individuals to act in ways that are constructive and in towards the you know the common goal I have to say I mean there is there are some initiatives already on for contact tracing of course I mean it's a huge problem that we will have to solve in the coming weeks or months in order to you know get back to a bit more normality than today in Europe there is a initiative I am not directly involved but I know I am familiar with it and I know people who are involved it's called PEP P - P team so that's a pan-european consortium and I know one of the proposals is called dp3 T we have this

piece and face and teeth it's a it's a dissolution and it's actually a really really nice solution is is built by a team of people led by Carmela tronco so from epfl she actually did her PZ with us here you livin and it's a really really nice solution because this is what I would call a really excellent example of privacy by design in the sense that it shows how if you want to achieve a functionality by which people are able to warn to warn others that they might have been exposed without producing data that can be repurposed and abused in other ways this is really an illusion that achieves that right I mean the

sense with privacy technology is what we try to achieve is can we do what we want to do without risking exposure of data right because of course you can be fully private if you just you know throw your phone away you throw your computer the way you have no internet and now you know you don't have that problem anymore but that's not what we want we want to take advantage of new technologies we want to the convenience we want the you know the power of these systems we just don't want to pay the price of now in order to get that I have to now be completely transparent and visible to to other parties that might use that information for things that are not good for me all right mm-hmm and that's why I mean this is one of the things that I that I find so exciting about the

technologies that we explore on this show which is this idea that you like with a zero knowledge proof system where you can reveal some information without revealing all information or this just I've always liked this term you know selective disclosure and sort of choice of what you disclose obviously these are amazing things to aspire to I really I actually going back to what you just said this comparison and the fact that this time around it seems less adversarial I really like that thinking but I wonder like do you still worry I mean for the short term I completely agree that I think that is the model but are you also worried about some ends we've heard this from like other people

in the privacy space are you worried that this that there will be kind of a rollback on privacy that has never returned to us well sure I mean of course there are many parties who just want to do data graphs right that they are looking for opportunities to get access to more data and privacy to them is an obstacle to their business right I think from governments it depends on who you're talking to I think government is a very big beast and some pieces of it are concerned with the similar civil rights and freedoms with other pieces of it are concerned with things that might be in conflict with civil rights and freedoms right sure I am very concerned that these will be used as an excuse to to intensify

surveillance and that the climate the fear and they worry and the concern that people have will kind of swipe out the the privacy concerns that are also very legitimate especially if we go for solutions that rely on surveillance in order to in order to achieve this function of you know controlling the spread of the virus so and the problem is that I mean it's intuitive right like if you want to have like as an individual if you want to make decisions what you need is data you need information to make this decision you need facts you need to know what you're doing the the counterintuitive thing about privacy technologies that

sometimes you're able to achieve or to make the good decision without actually getting to see the data this is you know zero knowledge is is one of the of the wonderful you know very counterintuitive technologies that does this so of course the general public will make if if what is presented to the general public is either you give your data or you will have to be in quarantine for the rest of the year and the the country will collapse then people will say yeah sure well you know this is something we need we have this surprise we have to pay this is it's too much otherwise you know but I think what where privacy technologies research comes into play saying okay how can we achieve what we want we thought turning this into a surveillance nightmare right yeah

it's basically the compromise there where it's like you you do provide the data you need to and you still I would say it's not our implemented some tears I would say well is having the cake I'm eating it right have to give something up whatever you want ideally I mean sometimes you have to give something up but ideally what we want is to not have to give anything up of this of our objective I also not give anything up up about our privacy which might be a side concern when you're you know focusing on the on the application objective in this case a contagion how how do you understand these decision-making the decision-making process going on in governments or like when we talk about this but let's use like maybe today's

case is an example so there's there's governments who are interfacing with sort of data mining companies and they're coming up with some solutions they have almost like a team of consultants who are thinking about and working on all of these solutions that they could potentially roll out through the government channels how does Pratt like how does a privacy advocate or somebody working on the privacy side of things like at what point do you enter into that conversation yeah so I have actually been not not recently but you know through the last years occasionally in contact with policymakers I think it's an academic this is part of the job I don't I'm not sure all academics do it but I think it's part of a of your job

is to kind of disseminate what is important about the signs that you do to policymakers that might make decisions that are relevant to to design so that designs is relevant to the decisions that they are making right I am not so deeply involved into policymaking to know which structures are used to consult consult with experts but I do know that often there will be people that are kind of on the lookout of who are the experts on this topic and who should we invite and then often we have a workshop or a roundtable and then they have people who are working on these topics to kind of enlighten them about you know what what the science says now I am not sure to what extent this is

very structured I am not enough of an insider to to know these procedures from the inside but yeah there is expertise being sought sometimes now the problem is that even if that is happening I have the impression that industry lobbyists have a lot more access to policymakers and academics too right I mean I have to teach I have to polish papers I have to review papers I have to do a lot of things I don't expend my day talking to policymakers about why they should care about privacy and you know what solutions they can use this not my full-time job it's something I do occasionally now industry lobbyists these are very smart people and I paid a

lot of money to full-time be on the job of you know convincing persuading policymakers that you know their ideas are the one that should be taken up right so I do have any impression that is when you have when when the path ahead there is a divergence between what may be people from academia that work on privacy would suggest and people from industry what people who are interested would suggest I think we have less weight in the sense that we don't have the ear of the policymakers in the way that lobbyists might write so unless you could find a way to have lobbyists for

privacy like unless somehow the the privacy side of things was super well represented also in that category but you know the problem is that the business model in the internet it should very less capitalism yeah it's true so I mean all the big tech companies I would say they're all making money I mean Apple is making money off selling hardware mostly but I would say the others are basically making money primarily off the data right then I wonder I mean it been in that in that scenario that you just described I almost feel like we should identify those groups that are not benefiting from the surveillance capitalism online and argue that by supporting these

privacy concepts in their lobbying efforts like with big big bucks they actually have competitive advantage absolutely I mean the thing there might be a way to think yeah absolutely I mean that that's absolutely necessary the thing is that if you look at big tech and medium tech it's all based on surveillance capitalism they all need more data right so privacy is traditionally something that has been more the the terrain of academia and civil society if you look at all Tories are not for profit right and only now if you look at GDP are and like where that came from I mean although I think there was some corporate book backing but I believe is as far as I have understood

that was more from like a civil government level in order to actually and and maybe a little bit a European to push off American sort of interests a little bit yeah yeah I mean in a way I mean it's a it's a in in some extent is an ideological battle I find and and I think academics and researchers and activists they need to be present and they need to keep pushing because this is the only way that this conversation is will shift right mmm interesting but yeah I mean I have to say that I mean privacy companies we have avid named technologies as a privacy company and I would argue that Z cash and you know other projects are

also kind of fitting into this this category of privacy companies this is something extremely new like privacy I've been working on pricing for 20 years it's extremely difficult to make money of privacy it has been extremely difficult until now I I find that actually blockchain projects and the blockchain world has opened new opportunities I think for new opportunities funding and new opportunities for business models that were not there before so I'm hoping that we will see a new generation of of Industry of companies actually building software and products that are have high levels of privacy preservation because

otherwise I mean academics I'm sorry I mean if you see students they don't make industry quality software right I mean they make research prototypes they can be very nice very interesting but that don't get to the masses you want to get to the message you need industrial production and you need money to be there in industrial production for that industry production to be sustainable and that has not happened until now and now let's see I mean I think now we have an opportunity and we we have to see if these companies actually are able to well to make the case and to to have a sustainable business model and to show you can also be privacy friendly and still viable like as you talk about these privacy companies for some reason the concept of VPN sort of pops into my mind but VPNs are not a privacy

technology in my opinion because okay I mean they do provide privacy with respect to your local network adversary and with respect to the recipient but the problem is that they are trusted party there are central Wanderlust and I would think that if you have a privacy technology to be worth the name privacy technology it cannot be a single point of failure you know see it cannot be that there is this one entity that knows everything you're doing and can fully surveil you so III in that sense I don't think VPNs count as a privacy technology because even though they protect you against towards third parties you are completely vulnerable towards them this is not the

case in tort for example right in Tour like each of the nodes in that in your path knows either intermediary and successor but nobody knows the full path while the VPN knows the full path because the full path is you them and the end destination right there is no more than that so you need to have multiple it's like distribution it's like when you when you don't want to have centralized systems you want to have distribution a VPN is a centralized system in in a way and and by distribution means that you have multiple proxies and then each process it has a piece of information but all of them have to collaborate together to put together all these information pieces

right so it's really the same concept that's in a blockchain that you have all these different entities collaborating such as that no one of them being malicious can violate the security properties that's actually that's an interesting point that idea that like privacy technology alone like it isn't truly privacy technology unless it has some element of decentralization because or or maybe you could say like it would be an insecure privacy yeah so the centralization is a typical way to achieve distribution of trust you can also have centralized systems that are such that the central system is not able to learn anything for example yeah you could have a system in which everything is done with a morphic

encryption for example so in that case even if it's not decentralized I mean they can of course cut the availability or they can stop responding or you know things like that but they will not be able to break the confidentiality of the information even if it's centralized but but in a communication system yeah they they are said that you cannot encrypt the asset because the assets that you're encrypting is the metadata is who is the activity is the signaling right so you need to have multiple proxies in order to achieve that it's not operations on the encrypted data I don't what what do you think then like now that we've I mean we've covered a lot of really good ground here we talked about name we talked about like just sort of privacy generally what are you

what are you looking at kind of going forward I know we're all in this like funny state where we like planning for the next two months it's like impossible we have no idea what's gonna happen but what where you at with the privacy trajectory and what do you think we can kind of like look forward to I mean I'm a lot of my research now is on anonymous communication systems I have actually done I mean that was my PhD and now I'm back to the topic of my PC in between I did for many years I looked into many other types of location privacy social networks you know all kinds search all kinds of applications but now I'm back to to anonymity anonymous communications I mean for name I mean there's plenty of open problems that we are still figuring

out because it's a new design so we are building on on the one hand well-established literature and techniques but on the other hand of course because these are new environments and we want to offer additional features so I I mentioned that mixes mixed notes in our network are rewarded for their work mm-hmm they are rewarded for for forwarding this traffic and doing it correctly and not dropping packets so we are actually working out I mean we have the concept but we have two fully developed the the full scheme of of how we we achieve that right and the way we do that is I would have some sort of secret shoppers that

use the system and then those messages then at the end are revealed these are messages that are only used for for measurement so nobody conceded they look like normal messages but then periodically these they become revealed so if somebody dropped messages you would be able to see who dropped them and if somebody is dropping messages consistently then they are not providing a good quality of service therefore you don't want them in your network you want them to you know take their stakes and kick them out because they are not I mean they are basically driving your user base away by not providing a good service right would there be us would there be a slashing event if that happened it is the stick yes there so so if I mean of course you have some tolerance this right like it's

not like because they dropped one packet ever they get punished but if they are not performing according to the level that they have to achieve yes they will be slashed because I mean the thing with mixed nuts is that if you have low quality mixes then packets lost then I mean it would be very difficult to actually have a functioning system that people can use and it is reliable that we are able to attract services and not for something that really works right we need to be able to have reliability on this underlying servers so the way to achieve it is to say well you know if you provide reliability you will make money this will be your you know it's like your job is that you run this mix your service do you run this mix you

have to buy the hardware you'd have to do the work of configuring and maintaining but then you have a salary you have an income for doing this kind of work but if you are not meeting your promises then you know you will lose money so that we don't want people to join and then not provide the service because for us that would be a point for us and for the users of the system that would be very problematic so that's some of the work you're doing on the NIM side but what about generally like what do you see coming up coming up general well I mean in the field of a lot of privacy technologies I do think that we will see now a lot of proposals for contact tracing or yeah location not only I mean contact tracing is the the ones I'm more

familiar with other countries are moving into quarantine violations so I guess any people would be now coming up with protocols and solutions to how to achieve this in a privacy-preserving manner that's definitely I mean there is this this the p3t scheme that is actually very interesting and I'm are you excited about it like do you actually do you feel more hopeful right now then maybe like post 9/11 well post 9/11 it was just like you know you just go with your shield you know and you're just you just in defensive mode all the time you're just being you're gonna be called a friend of the terrorists or something you have to really have all the arguments of why this is not the

case right things have changed a lot since it snowed and I have to say and you know the the whole horrors of Facebook and now you can talk about privacy everybody will agree with you or not disagree so that I mean I don't know I mean these things it's politics right in the end so I think it will have to be fault to the people who are pushing for these proposals to be to get the support that they need and to to be able to find the good arguments and to be able to persuade those who make the decisions to be able to win that argument that is nothing is to be taken for granted I think I mean there are good technologists on the table doesn't step one if you don't have even another

technological solution say oh we should not do contact tracing because it's privacy invasive and we should just be in the quarantine until 2030 that's not gonna be you know that's not going to be an argument of people will buy so I think the first step is to put good technologies on the table that do the job without jeopardizing privacy and then to do the work of the politics work of you know talking to the people who make decisions getting to decision makers speaking about it persuading people why this is important why this is the better solution but that's true I mean of course for the people who proposed it but also I think the rest of us as a community we need to we need to push in that direction if we want to make a difference because the others who have privacy invasive solutions often

they have more resources than we do for for doing this PR so I want to say thank you so much for coming on the show and chatting with me about you know NIMH technologies and a lot of your work in privacy technology over the years and how you kind of see you know what's what's happening now through that lens so thank you so much for sharing your insights thank you very much for having me with my question yeah and to our listeners thanks for listening [Music]