Privacy for end users, transparency for the infrastructure -the Nym approach to privacy

Transcript

we'll see um okay so i'm uh harry halpen i'm the ceo and co-founder of nim i am unfortunately for everyone who wanted to see the much more interesting and uh better looking clara jaya breck our media person i am not her she is stuck in london uh due to terrible covid restrictions so this talk is called privacy for the end user uh transparency for the infrastructure uh uh it was great having hopper give a sort of introduction to unlinkability on layer zero although now i kind of know how like i know like a vitalik feels after an eos talk or something but

we're gonna we'll just keep going um so surveillance uh this talk we're gonna talk about surveillance in general so it's more high level and social we talk about what nim is doing uh kind of our design on a pretty high level we have this kind of white paper that's quite large you can look at for all the details i went to the details actually last ecc so i'm not going to go so deep in the details this time and they're going to end talking about kind of our vision for the future of privacy which is much larger than just the mixnet and i think this is something that we would like to see other projects uh you know orchid i think is next so i've discussed alpha seven uh we like to see a lot of other projects form an alliance to really push

through uh privacy because i sometimes people ask me uh you know are you a bitcoin maximalist so like i'm friends with adam back and max keiser and all these people or are you an ethereum fan big you know i like vlad old friends with amir and other nut jobs uh and i only say i'm actually a privacy maximalist so i i'm particularly to be honest i'm more interested in using these kinds of tools to help privacy rather than necessarily just to make money i do want people to make money i support financial freedom i think blockchain is a wonderful tool set for that but uh we'll kind of get into some of the why privacy is just if not more fundamental

uh than blockchain technologies at the end of the talk um oh so this is not working at all so um i i don't know okay yeah so anyways uh so metadata leaks the network level uh this includes vpns uh it includes tor given the kind of unequal treatment of packets and hopper so sebastian's response only makes sense if everyone does the same thing if you can distinguish packets you can do side channel attacks and that would also be true for hopper and therefore you know we don't have many solutions which actually we think in a realistic sense uh prevent global passive adversaries so these

so what a global passive adversary does is they look at the traffic going into an internet for example it's not totally unrealistic your isp can do this to you your wi-fi router which whose password is like baguette or whatever can do this to you here uh your telco provider can do this and it can easily correlate outputs for example which website you're going to are you going to some you know defy protocol et cetera et cetera god help you you know you're everyone's going through infuria and metamask and so it's pretty easy to do correlations with these very common tools and this is part of a larger problem because you know as sandy pitlin says mit people think the data is important but

it's the metadata that's actually the most important so if i have a group of people and i can figure out who's talking to who and the tone of voice they're using their kind of volume let's say and when i can actually deduce 70 to 80 of their behavior purely from who's talking to who with just completely ignoring uh what they're saying that's how important social networks are in social structure and then whenever we use the internet either us the machines or machines the machines or us to each other via machines usually we're revealing that social structure to the world and that social structure has been uh is very valuable in fact it's so valuable that while we give it away for free and higher hosts of companies exist to basically exploit and harvest that metadata so that's surveillance capitalism so what is

surveillance it might seem obvious so we have for example you know cameras probably we'll have covered passports sometime tomorrow which i have a whole nother paper on about why you should never use w3c dids and why most of the blockchain work in covid passports totally scammy but this is sort of obvious surveillance right i have to you i know you know who i am you're watching me you're checking my passport which i just had happened this morning uh but we lack the visual metaphors and we and therefore we lack the understanding of how surveillance works and people say well privacy you know it's old-fashioned it doesn't make sense look even animals like you know small crabs for example or you know have homes where they withdraw for

privacy life itself seems to have a desire or need for privacy and the fact that we but we do not evolve in a situation where we can understand algorithmic surveillance we can't understand not just big data so this is you know someone's looking over metadata they're gathering all this big data that's one thing but people often get confused they think well is the value in the data is the data oil but the real value is in these massive opaque computations uh kind of infrastructure that runs on top of our data and this is algorithmic surveillance and so this is surveillance 2.0 in order to fight back against surveillance 2.0 we need more than mask uh we need more than tor

uh we need uh kind of really what i call in a previous talk holistic privacy because the fact of the matter is like if i use some zero-knowledge proof stuff on chain let's say we're at ethereum conference i'm using aztecs planck planck is a wonderful system used to share an office with aztec they're great guys i recommend all the zero knowledge work on ethereum it seems the most interesting but any privacy leak in a lower level even if using aztec on chain let's say i'm leaking my ip address i'm thinking my timing linking my volume any of these things can de-anonymize very effectively your zero-knowledge proof usage on chain and you can de-anonymize your zero-knowledge proof users on chain all

of your snark-based apps like whatever dark force are gonna be quite easily breakable and that's uh that's very dangerous right so because we're building a very powerful what we consider empowering technology on top of these kind of blockchain defy ethereum bitcoin whatever you want building blocks and by virtue of not defending this data and it's not really our fault you know when they built the internet they weren't thinking of these things uh there wasn't very much knowledge in nsa there was much money flowing across it uh surveillance is creating a relationship of power uh where even though we think we're empowering people um i'm actually being disempowered so one of the most hilarious critiques of

blockchain tackles from ian myers of zcash who said well you know we reason why we're building zcash because if you don't do some sort of shielding on chain uh then you are effectively building a twitter for your bank account your for your payments and this is maybe something you don't want to do so there's actually been people there's apps that do do this and some guy buys some weird porn thing or something and it gets revealed on his twitter and it's embarrassing uh but you know we have real use cases of people that need money places uh i for example my most my knowledge is you know iraq or somewhere and um you know they have sanctions they have problems uh i have a friend in guatemala who's a farmer he can't get payments as bank constantly blocks them uh therefore you know he can't save a large

bit of rainforest which has you know some of the most valuable ecosystem in guatemala so we're seeing this again and again and again and what the fundamental thing is surveillance is kind of creates this data and it uses this for classification are you black are you white are you uh are you smart or you're intelligent are you thin are you fat are you easily marketable to or you're not easily marketable and then it uses this to manipulate and control you so this is a lot of this fake news hysteria uh and so what we want is we want uh transparency the cypherpunk motto is transparency for the powerful so this would be governments institutions financial products like defy uh and privacy for the rest of us privacy for ordinary users

so this is like the kind of old slogan from the cyberpunk movement from like the 90s uh and i think it still holds very true today so this is why wikileaks despite being a transparency project his father was here with me last year is effectively uh uh is a privacy project he's saying governments should be transparent because they have lots of power they have an asymmetric relationship with you and me while we should not be so and the real question is a decentralized system who is powerful so it's pretty obvious it's usually the miners or the block you know the people for example the people who do the governance decisions and so having a lot of this stuff is on chain is super useful uh you know we're still exploring this but i would say that what we are aiming

at is trying to re revisit because we're a bit worried that all of these kind of original ideals of the cypherpunk movement of the that where a lot of the blockchain technology came from can be lost in the what we consider large-scale institutional capture of this technology uh by banks and like scammy erc20 startups so we're worried about that and we'd like to go back and really reinforce those values and you know these values aren't just for white men unfortunately like myself but they're also you know they're the person who coined the term cypherpunk uh was a lesbian woman uh jude milhoun also known as st jude and that's who we've named our test net after so our next test net which is launching tomorrow

uh which will feature delegated staking and all that wonderful stuff uh is bit named after her and so you know we have i'm not gonna go into because i went over last year and i figured maybe the same people we have a white paper all of our code's open source governance decisions are going to be open source and so we're trying to we think these are good practices that we're appreciate that's coming a lot from the people in this room and we will use these in our technology so we want to say who decides and for whom we want this to be transparent and then users can then choose sort of in a market if they want to participate in our protocol or another person's protocol that's their choice um so here's what our protocol is so this is how we actually defend against metadata capture

we prevent traffic analysis by using a mixnet which you can take kind of like the first bit of amir's talk and that's kind of in the same space but eventually if you imagine packets are like you know cards let's say they're like a deck of cards uh and i encrypt the packet i turn the card over so you don't know if it's a ace or king or queen whatever this message is it's hidden and then all these cards are the same size so you can't distinguish them using that sphinx packet format and then all mix that does is it mixes them up random permutation if you're into mathematics and then it releases them and we use a design called the lupix design by anya petrovsky and friends and this design basically releases the package using what's called a poisson process and we think that's a

really great way to do it because you always know the average time your packet will come out the mix but you don't any individual packet may or may not come out at any time and that's really great because if you're using an application you kind of wanna know if it's gonna work or not and if some packet takes too long you just resend it a little bit like tcp now of course the question is where do you get these people will people run nodes you know you need all these people do this mixing it's kind of a proof of work system right you got to mix these packets so we use a token which is a you know as the incentive structure sort of crypto asset to use terminology from chris breneski to basically allow mixed nodes to stake and it serves as a reputation

as they if they fail to mix the packets correctly they lose reputation if they mix packets correctly they gain reputation and the question is how do you prevent you know using these having these tokens which you can also use to access the network well that would seem to de-anonymize you right so your standard erc20 token or your standard cosmos token or even your standard liquid uh bitcoin nft from block stream all these things uh are not super anonymous so we built an anonymous credential system which is another old piece of technology sort of like a simplified zero knowledge proof scheme which essentially serves as you could say nft to access the system which is actually anonymous so i would like to use them i have some tokens

i don't want to think about tokens i don't want to go to meta mask i don't want to install these you know apps i'm just a user i want to use a vpn i want to use signal and what happens is that you can just you know essentially purchase or be given tokens by your app for example you know let's say signal one used them moxie could give you these uh credentials in the software and these credentials anonymously let you prove that you have or someone has given you these tokens and so it's a little that's a little bit of a complex part but the system in general uh it's not very dark diagram i apologize i didn't make this you have a user and the user wants to go to a service and they want to go through that service anonymously so in the middles there's this mixnet

now you know mixed nets are have this you know we want in order to maximize anonymity we want every packet to be treated the same right so the packets kind of have to go through the same number of hops and the hops have to be structured we call this a stratified topology in order to go into that they have to go through a gateway and that gateway of course knows your ip address that's just how the internet works so the gateway has a certain responsibility the gateway you want to figure out how to access these gateways via checking with validators this is similar to if you know tor the torah directory authority system but unlike torts or permissionless blockchain base that they run the blockchain that maintains the reputation of the mix nodes you as a user can view this reputation

talk these validators you know from your either service provider or the validators get your credential go into the mixnet get your packets mixed for you know imagine quite a lot of packets being mixed not just one by one and you go out to a service provider so that's the kind of overall system architecture and you know to be honest it's it's it's a mixnet uh but the service the service architecture has these components and it's a little bit similar tor and what we're trying to gain is trying to gain unlinkability uh so if you're interested in what anonymity means this is the classical paper by feitzman who's unfortunately no longer with us and merritt hansen who still is luckily called a nominee unlikability on observability and they say well you know

to be anonymous you should be unlinkable so you can't link two packets together uh sender and link abilities you don't know who sends the packets so i would like to send you a message anonymously recipient on linkability means the receiver can send them back the message and they don't know who the recipient is so you don't know the recipient you can send a message to a group and there's some other things like unobservability which basically means i don't know when you're sending messages which is what we accomplish with cover traffic or you know there's some more complex stuff as well that people tend to undetectability which is we which we don't provide where you don't know if you're even using the system so undetectability would be you know something like uh i'm trying to disguise my traffic as something else for example like i'm making i'm using for example

altification proxy to disguise my usage of nim or tor uh make it look like i'm going to amazon so your classic so the way you solve a linkability is again it's just like you know nem is a version of this but so is tor you have traffic goes through a bunch of routers and comes out these routers provide the privacy you can actually measure anonymity it's not something you kind of prove yes no like most people think with zero knowledge proofs it's measurable in terms of what's called entropy there's the paper there by claus clyde diaz who works for us uh it's a very old paper from 2002 so it's been going on for a while again i just want to re-point out cryptography is not networking every once in a while we have some confused people who think zero knowledge proofs are like some magical

privacy solution very useful for scaling ethereum not a magical private solution by any stretch of the imagination your anatomy set again is bound by the number of users of your system if you don't have unobservability it's the amount of users at any given moment of your system which is not very much often particularly for a lot of blockchain systems and so while we do want people to use things like zk snarks and starks or whatever the new version is on chain you really need to look after your layer 0 your networking level right so otherwise it's like building your castle on top of sand um and oh go go back go back go back okay so this is important one so people might say well what is the difference between nem and hopper

well a we came first b we our advisor invented the sphinx packet format uh but hey you know whatever it's open anyone can use it that's cool but there's a reason why we didn't choose appear pure peer-to-peer design uh you can read this paper here which came down in 1998 which was the original peer-to-peer privacy uh design it was kind of similar designed to say hopper and it says he says you know we can kind of just pass packets around in this uh in this in this crowd and you can be anonymous a crowd and that's a really cool name for a system and it seemed like a really good idea but it had so many devastating attacks against it in fact tor and all these projects basically built the designs they did in order to recover against these devastating attacks so one of the

devastating attacks is a malicious peer i send it to you and you send to your friends or you don't send it so you know relying on peer-to-peer systems where peers can act differently and choose the route or not route packets is often in a pure peer-to-peer framework pretty undetectable pretty dangerous you can use cryptography protective and then you just open yourself up to more attacks and it actually gets worse the more nodes you add your privacy decreases often because your traffic gets spread thin the more nodes you have the less packets are mixed in each individual node so there's a really good analysis here by vitalish schweinkoff who i would never ever anger because he's really brutal called probabilistic analysis on enemy and there's another type paper i forgot called the predecessor attack

predecessor attack is what if i send you know if i am the predecessor i'm your quote-unquote downstream node i can then learn a lot about where your packet's going so there's a bunch of fails in that space we've you know we've tried to build something based on lupix which we believe does not have those problems we believe it's scalable we don't need too much cover traffic we do use it and we use serbs which is single-use reply blocks as well we use a stratified randomized topology so again just like tor or hop or whatever multiple hops unlike them we add cover traffic in a systematic way everyone every packet is equal every pack is treated equal and every packet's treated equal leads to improved anemone and using the poisson process from lupix we do better

timing modification so that's like we take sphinx which is already explained so i'll skip this but again makes the data a constant size source routed but we do a bunch of nim pacific improvements to speed it up so we get 20 30 speed increases by using better hash functions uh better wide block ciphers and you know all that sort of stuff so uh you know and we've caught many arrows on other systems where you know again cryptography can be subtle you know for example we have to add randomized padding if you have variable length mix that you link your path length so this is a an issue we noticed with hopper i think they eventually fixed it but again these things are quite subtle so uh you know again our private

infrastructure you can run it today we're in test net because we're not just doing a token sale and trying to get rich quick we are actually trying to build the software carefully and intentionally so you can add run and mix node you have keys and addresses and reputation that's what it looks like if you get one up if you go to the nim website you can see a link to how to run it and today is your last day to start one before we close this testnet down and you might not get into the next one because we have ten thousand friggin nodes running this crazy software so let's i'm going to back up at the end and just kind of end on a high note so again we've thought a lot about this technology our team's been working in the space for you know some of us 10 years others of

us like claudia 20 but again the reason we're doing this with privacy is not a lecture it's not just for some put somebody with something to hide if everyone uses this kind of software so it has to be able to scale to internet level traffic the more anonymous everyone is together that's super important and it's important for us as individuals because we can't if we're constantly observed you know we obviously lose some of our ability to make freedom to have choices and uh the late great bernard steegler my mentor here in france he basically said you know we the science of what is cryptography it's the science of preserving secrets and it's central to any reconsideration the fundamental principles of the law as any new arrangement between digital

technology and law should involve the right to secrecy and the duty to protect it because otherwise you can't have even angela merkel can't have a discussion and use a reach researcher can't even interpret a paper you you have to this privacy is fundamental to our very right for digital public space and the exercise of kind of our intelligence and you know it is a universal right it should not just be for rich people it should not just be for people that use z cash it should be for every blockchain project algorithmic surveillance in particular hurts the people that often have the least amount of money the most marginal uh people who and it and therefore limits their human potential and it's a relation of power which which increases colonial forms of classification and control

so this is a quote from one of my favorite philosophers akilah menindee who basically says and you know we're seeing this system of control increase dramatically brutalism is going to intensify under the techno-libertarian drive be it in china or hidden under the accoutrements of liberal democracy i think he might be talking about daos here sometimes just as 911 opened the way to generalize state of exception this means anyone can be a terrorist and therefore be arrested anytime and it's normalization the fight against even kovit can be used as a pretext to move the political even more into security and the security eventually becomes biological forcing new forms of segregation between immunity bodies viral bodies people who have vaccine passports people who are lucky enough to be born in countries

where they where they can get vaccines and this is a new theater for fractioning and dividing populations and even trying to view people as distinct species and this is absolutely unbearable we must remember that you know the goal of the internet is a universal information space our goal is to help you know all of humanity uh to draw out all of our potential and so privacy opens the horizon of freedom it's the precondition collectively not just individually because remember you can only be private in a crowd and you can only exercise freedom with other people for flourishing humanity and a flourishing internet and the right for privacy uh is something which is we need to make deliberate when we share our data and nim

is a small step towards that we need to work with other people like who do other properties like you know for example dvpns and tor do censorship resistance or altercation technologies or governance projects such as daos and these are we have to work together here to really build privacy into the protocol layer and thus into our daily lives at scale so that's it uh any i'm sorry i think that went a little bit over but that's q a if there's no time we'll just uh i'm around you can grab me afterwards thanks a lot uh what i'm just wondering what the um

added cost well assuming let's let's say when you plan for eventual mainnet um or let's just say going live in production uh what would the added cost to end users be in terms of added latency versus you know they're not using a mixnet like nim um and cost in terms of actual financial costs to route through the mix net yeah so how nim works or mixed network their mix nets in general can be very high cost so like the classic charming design you know you need to wait for all these packets to come you mix them up then you release them so you're you're kind of like waiting for all these packets maybe they never come and you can and that's a problem right

because then you're really slow so again what lupix does i look at the lupix paper by anya petrovska and that's design we use is we do not tell you how slow or fast you should be you tell us now the fact of the matter is if you do let your packet be slower you tend to mix more and so you can be more private and that might be very important for say like you know an ethereum transaction however if you let's say you're browsing video and people have browse video over nim it's kind of crazy but did work we were sort of shocked uh you might not care that someone's you know knows that you're watching fast and the furious uh so you might or you don't only care if your isp knows we don't care if like the nsa knows so then you can say well we actually

don't we want to be faster and so we're willing to sacrifice privacy so it's often what's called a anonymity trilemma where you can basically trade off kind of amount of cover traffic the amount for round privacy the amount of latency you can imagine because the kind of latency privacy trade-off now while i said there's a trade-off it's not entirely true because again to be private you want to be private in a crowd right so the bigger the crowd is uh the faster everyone's track traffics and go while remaining the same kind of private right so if i let's say you know i have 12 people my system uh you know and maybe we don't use some of these people aren't using the system very

often and i really want to be anonymous among those 12 people i have to maybe really slow down my packets right and i want to be anonymous let's say of a minimum and a crowd of at least 12. but let's say we have 5 000 people using this system that any given second there's probably 12 people online and that means i can have the same level privacy but not speed up my packets i'm sorry i can speed up my pack because i don't have to mix them i have to slow them down so the trade-off is actually subtle the bigger the system gets although placing the mix nodes is hard because you don't want to spin that traffic out too much the more profit you can get in terms of cost we're trying to calculate actually right what happens is every epoch every time moment our system this

is kind of in the white paper we run it's actually section six the white pair which is a bit the hardest part to read we essentially run a bandwidth auction so we say well we have all these mixed nodes and they all want to make some profit it's actually a little bit like cardano if you guys know what cardona i'm sorry but they do a very good stakings game uh and you know we were it's a little or ethereum too maybe a slightly more politically correct uh metaphor um so you know you can you basically they need to make some they want a profit margin and some of that profit goes to people that delegate to them and every round you know we don't me we may have like ten thousand people in our own mix nodes we only have a hundred users so we only need ten mix

nodes so in the beginning of every round we run an auction right and that auction basically uh basically tries to get the most efficient packet routing and from our experiments so far we believe a sphinx packet normal kind of like say 400 4 000 bytes or so we can kind of push down to like i think it was like 4 000 i mean it's it's quite small for i don't know i think it was like four four thousandths of a cent right or four million of us it was very small and we're still running simulations we don't know what the real world will cost these things but we do think privacy is very cheap so we're not imagining uh too much in terms of cost to users we're imagining a

few cents not dollars or euros and we believe that privacy is so important that certain apps and we've talked to folks like brave and moxie about this that certain apps would be interested even purchasing privacy enhanced bandwidth for their users on behalf of their users because if it's only like one millionth of a cent per user you know it's not that expensive you can it's like buying a vpn access something that you know brave is already doing right now so um i'm really curious about the developer experience so like as an application developer using it it's really cool um i'm also a little bit concerned personally about using it because i'm

okay leaving all this to you but i'm really afraid to like up an implementation like are there any foot guns as a developer like hey can i go through nim and then go into infuria and if i know i'm hitting decentralized inferior but am i creating like this foot gun to leak privacy somehow even after or like if there's anything to be concerned about from yeah i mean that's how it works today it it we we we aren't super developer friendly if you want to do what we call language level bindings we haven't built that api infrastructure yet i mean you can you can talk to us about it and we have some docs but it's not great but what we do have is a lot of developers seem perfectly happy binding their software to tor and they do that using this uh unauthenticated kind of vpn sort

of ipwp thing called sox5 and we support sox5 out the box so if i want to use firefox with them we actually have people that have used it and it works and we have a little video of it you could do that today by just basically going to your firefox browser putting in localhost and our port is 1789. go friends and you go to that port rather than some other port and uh that sends you straight to nem so you as a developer or the end user don't have to think that hard if all you want to do is like just shovel packets through them now it gets a bit tricky if you're like oh i really want to make sure they're really fast you might have to think a little bit and if you want to say oh i really wanna make sure they're super private then you

want probably think a little bit too but this amount of thinking is not super multi-dimensional you're basically like well how slow do i want to go and how fast do i have to go and you can set that in your app and we do have an open source simulator which you can run you say well i don't know what i said like i said 10 seconds or one minute and this simulator takes in it's in python it's anya has a great blog post on it uh and it basically says you know i have this many users i have this many nodes uh i would like to and and given this me using this many nodes uh if i send the packets at this speed how private am i or you run that calculation in reverse you say oh well you know i don't know how i know i want to be

pretty fast how pr how much privacy in terms of entropy do i get and so we can tell you basically and you can run that little python script locally to kind of figure out uh uh how private you should be and crazy end users might even like customize them they might disagree with you they might want to like set it a different value but that's you know that's their right right so um so it felt like we kind of skipped over the gateways so i just want to ask like how are the gateways incentivized and like what kind of uh privacy can a malicious gateway leak because there was a lot of like and the mixnet is great but like

it seemed like the gateway was yeah well i mean every mix net will have a gateway you can't design a mix of that gateway because you always have your first hop in right so eve tor has what you call entry nodes vpns hey the vpn itself is the gateway like everyone has this in fact one nice use case is people and i hope to talk to seven other people about this like you could just use your existing vpn as a gateway and if you don't trust anyone and we have people that do this you can just take a raspberry pi and run it yourself right so you don't trust anyone to know your first hop and then mix that why not run it yourself the gateways are needed for two reasons and they're incentivized kind of the same way as mixed nodes but they're necessary for two reasons it's

very hard to take them out one reason is you may go offline my phone may go offline i'm in france i don't have wi-fi i don't wanna like if messages are coming at me you know like email i wanna check it later i wanna and that requires something to be online all the time either raspberry pi or vpn whatever that gateway happens to be and that takes the sphinx packets and reconstruction puts them back together so there's no way out of doing that uh the gateway also has a privileged relationship to you because just how udp and tcpa work it's going to know your ip address there's just no way around that right and it can do some damage right now we try to minimize advantage because the combination of cover traffic plus mixing allows like the reason why tor is kind of

attackable is because you get kind of two-way correlation attacks i can like correlate the input and the output and then the gateway can only really knows the input so it can't really figure out the output even if it was colluding with the poor guy the exit node from the mixnet uh it couldn't really figure anything out because there's cover traffic it can't although all the traffic looks the same and they're all incentivized the same way quite uh quite differently from what amira presented we used kind of uses vrf based sampler it's described the mix and the white paper but essentially you want to think about a little bit just like yes the gateways also have to basically pledge some token and that's their reputation uh you don't have to pledge very much if it's basically nothing if it's just you you and you're you're just running yourself

but you're running for other people you might drop their packets or you might do some weird attack on them and we would like for your reputation go down if that is indeed the case or you might just go offline right but if you're online and people are happy with you your reputation goes up and you kind of earn more token out of running that gateway does that make sense yeah it's like you know you have to basically stake to run the gateway to just like you had around the mixed nodes but they have a little bit different threat model because they get that first hop info so they're they are a bit more dangerous you want to trust your gateway the mixed nodes are a bit more how do we say proletarian you don't uh how do we know if the nsa is running a higher

gateway well yeah they might it's it's very hard to tell if the nsa is running a lot of tourism in fact there was this case where they someone was running a bunch of torques and there was steel people's bitcoin not that long ago so you know again that's why we recommend you know if you're if you're really concerned about that uh you know you could sort of say well you know i'm gonna use the eff skate way cause i really trust the ff we imagine certain users will do that most people are pretty bad system bins although more people are good systems than we were sort of surprised actually how many people are good system events nowadays but you know if you if you are like mulder from x-files trust no one which is kind of my kind of style then you want to run your own gateway

and you know depending on you know if you're really pumping video through and you're doing all this file sharing and yeah you probably might want a little vps or a little bare metal server in your bedroom under your bed uh if you're not doing it that often then you know probably something smaller like a pop raspberry pi would be fine my name is daniel and i wanted to ask if nim provides something analogous to tour hidden services where you get a permanent uh overlay address which can be like a pseudonymous kind of endpoint which is still

protected from no we do not everyone wants us to do that we don't do it and we don't i mean you know there's probably ways to hack it on top but we do provide anonymous reply uses replies what do you think about mixnets is they kind of came out of email and like chat apps so it's like i'm sending messages asynchronously tor is opening a synchronous circuit like a flow of tcp ip and and so toward that is kind of dht system that dht system which is a bit shaky but does work sends you to like your torah onion address which then gives you which is your hidden service now the reason we can't do this because we have this kind of message based back and forth right and so we allow you to send someone a message anonymously and you basically

drop that message off in a mixed note it's like a drop box or a dead man's box or whatever no damn man sorry drop box dead drop dead drop that's the term vuvuzela uses it's a dead drop you drop some message off there and this guy comes back and he picks it up and you don't know who he is because again he's not you know you're saying to him anonymously uh you just you know he goes to the dead drop and that dead drop has a little receipt and that's to another dead drop right and that dead drop is where he goes to drop off the next message and you kind of go and retrieve that right so you have this weird async back and forth if you want complete anonymity from yourself and your sender right so the mixnet by default gives you anonymity against third parties people

are observing the inputs and outputs of the network not necessarily defend you from the app so let's say you're using signal over nim you know you're still going to like log in with your phone number or something crazy so that would be hidden unlike it currently with a signal on amazon it's not hidden like you know i'm sure the nsa is really enjoying looking at the amazon servers that signal is running but so this would be better because they won't tell if you're accessing those servers or when you're accessing them but um you know you still be giving like your phone number signal right so we have some tech to help that the anonymous credential tech which i didn't really go into but most importantly if you really want to do something like super anonymous signal or super anonymous relays back and forth of cryptocurrency

transactions perhaps a bit more relevant you would use the single use reply box block technique serbs and serbs are kind of based on that dead drop i think keeping it wrong whatever you know like a dropping dead drop concept and because the dead drops go back and forth asynchronously you can't maintain this like long-lived synchronous stream that you need for like a website hidden service uh now that being said we're talking with tor about this we haven't given up we're actually kind of friends with tor you know we issa and roger are heroes they've been doing this stuff for a very long time uh and we do believe that you know there is probably some hybridization that would allow something like a more private hidden service

that's that's r d so come back with us and next ecc we might have a better answer for you okay anything else great okay man i've been sorry clara would have given a better presentation but i try my best so thanks a lot you