zkSummit - ZKP for CollabTech - Elad Verbin (Lunar Ventures / BIV)

Transcript

okay I think we're gonna stop how everyone thanks for coming to the second stage I'm gonna give a talk which sounds philosophical but is totally not so this talk is going to be about collaboration Tech which is how to use the ways to use technology to align incentives and in particular I'm going to reinterpret zero-knowledge secure multi-party computation or morphing encryption and so on as particular cases of collaboration seconds show you how this should impact your thinking and your priorities about them a bit more about me later but first the outline so I'm first going to argue that collaboration is the key to civilization and everything we get around us all these nice things is

because we have technologies that are we as humanity develop technologies that allow people peoples and businesses and entities to collaborate better second I'm going to say I'm going to define collab tech which is all of these technologies that help us collaborate better I am going to argue this is the greatest value generator you can possibly imagine instead I'm going to reinterpret zero knowledge and private computation as collaboration tech and not as people often think of them as privacy tech like they're much more than privacy and then I'm going to say when you ask a question when you wake up in the morning what should I walk on the answer should be guided by the thing by the technology of the tool or the

product the best enables collaboration and not just to say let's just do what everyone else is doing but with more privacy because I'll argue that no one cares or whether you know more complicated than that we'll get there but they may take away I want you to get here is don't think of zero knowledge poof's as collab is privacy technology but rather as collaboration technology a little bit about me l adverb in not a cryptographer kind of important I mean general algorithms and moved slowly from algorithms to machine learning to blockchain and now to investing so I have a small berlin-based recycled bill innovation

ventures where we invest in very very early-stage project pre product pre LM deeply everything in small people doing algorithm based innovation small projects that are likely to fail if it has zero chances for the technology to fail development it's for some of our partners maybe but not for us what we like to do is the invest in projects that might fail but if they succeed they build an algorithmic product if it's really interesting if you have an interesting startup it is like this or know of one please send them to me or come talk to me later my love affair with private computation so as I said not a cryptography so how the hell do I know about these things well I happen to do a postdoc with the guy that invented some of this back in

the 80s and DL and then another postdoc in all who's with in the same department is if and am God and they were exactly deploying secure multi-party computation exactly for the first time in 2008 in the field and everyone was super super excited and it got caught the bug and even when I left academia I still couldn't shut up about private computation and you know looked like this guy I've seen the future no one understood what I was talking about until this events amazing events like this that all of a sudden the field is starting to develop and there are applications and finally the future is coming so I described but describing collaboration so I'm going to argue that collaboration created everything that we

see around us so you buy local Olli in his book sapiens asked what's special about humans so maybe it's about how we use tools really great but octopodes use tools amazingly well and they haven't created this civilization right maybe it's about a use of language but actually some you can teach a gorilla to use sign language just fine and gorillas didn't create civilization either so what's special about humans Halawa claims argues that it's about being able to have shared myths together that allow us to collaborate the point is that when we have money and that when we believe that the bill gets you a banana and will always be able to get you a banana we start walking for it so we can take this bill and carry electorate you know throughout

continental then electronically and start being able to build a civilization we see around us and that's the special thing that allowed humanity to create civilization it's about a shared beliefs and collaboration and of course you know this is the modern pyramids this thing that allows all of us to collaborate without knowing each other without having identities that you know that are named and so on so the historical epic is very very long it really starts with religion arguably so religion is a way to make us collaborate better what in reflection by saying well I'm incentivized to kill any one of you if I see you in the street and take all your money and belongings but I won't because

it aligns my incentives better by saying you will be punished in a future world so don't do that right so it changes my incentive structure and therefore we get a culture where we don't go around killing each other right in a multicolored logical setting the telegram for example allowed us to communicate between continents between cities very quickly it was used for gambling first right but to create Alby Taj but it was then developed into the Internet and so on and we got better and better communication and empires go when communication impose so you get you know fast forward a hundred deals you get the pay you get a Twitter which allowed people for example in the Arab Spring to

collaborate much better to demonstrate to fight governments and so on and you get something like Wikipedia which is an amazing collaborative project you create a platform well known technological people can contribute to a shared knowledge base and you get Wikipedia like the justice system can be thought of as a collaboration tech that aligns us better then you know taxation and so on is a way to take form to change the incentive structure so that we all give money so that we get back infrastructure and so on so collaboration takes kind of built everything around us and I'm going to argue before you know going to I'm going to argue that it's really good and a reportable the nice thing about

collaboration is creates positive some value it creates new business it creates new opportunities and it creates value and there was not none before rather than extracting value like Wikipedia and really open project it you know it funders if it's some funny it needs some millions a year but compare that to the value it gives actually it creates new value out of nothing there's a bigger tight Israel to monetize it so if you create a full value company it's still really good to concentrate on collaboration text because there's a bigger pie no one fights you as hard for the value that you created right and which is kind of the point of tokenization you create tokens because it's before the part of the project that cake sort of value you get to monetize them and it's easier to gain allies when

your project creates value and there's a compounding effect in time and in space it's not enough that I am able to transact in money with you the point is that I'm able to transact with anyone in the world with the same currency more or less that compounds that value compounds over space it compounds of all time because they're 1% increase in collaboration every year compounds to you know 2.7 per 100 years so let's discuss what you should avoid like what's the opposite of collaboration the thing what you want to fix and the question is is the opposite of collaboration competition and the answer is that it's no because entities that compete with each other very often also collaborate with each other right as industry groups of competitors that come

together for a common goal the actual opposite of collaboration is division right during the Cold War the US and Russia would not talk to each other except over secured lines in very dull circumstances and then they couldn't collaborate efficiently at all right so that's what you want to avoid you want to avoid division by fixing various lines of division and one of the biggest lines of division is the ones between businesses that can't share information with each other because that information can't be shared you to regulation or intellectual property or other kind of things so that's what we're going to talk about solving so collaboration take I I discussed this in a much longer and more detailed talk that you can find online same name collaboration tech so I won't go into it in details now but they shot

the show point is that you want to think of collaboration as two kinds of technologies one little friction for example the kind of friction that we need when I have to get audited by professional auditors that take a lot of money and also collaboration should solve incentive misalignment the kind of misalignment that says when I pollute then polar bears Brown but if I don't pollute someone else will so I don't have any incentive to really I have an incentive to freeride over this so this is the two kinds of collaboration texts we want to focus on and will mostly focus on the first which is kind of easier to solve as well that's not good okay future collaboration texts those are the kinds of collaboration texts

that I talked to you about that I talked about in another talk on this topic a private completion computation that we'll talk about today which is zero knowledge proof multi-party computation of morphic encryption and so on but also the centralized ledger should be thought of as collaboration tech because it allow us to keep very keep track of value without having any central entity detects friction and takes links in Rancic's token economics is a way to rely on our incentives mechanism designers again a game theoretic technology that realigns incentives decision-making tools that allow us to get consensus better and economic optimization which would allow us to kind of improve faster than kind of to

get more better use of text money or other money so these are all covered in another talk today I'll talk just about private computation so private computation we've all seen this all day long so just shortly I'm a data science consultancy there's a farm at hope they have patient data I have algorithms we want to operate my algorithms on their data but I have my algorithms all secret and the patient data is definitely secret we want to collaborate somehow what's the answer secure multi-party computation to apply my algorithms which are kept secret are encoded and on their data and together we get there so so this is the humor party computation it's cryptographically

provable computation right not just privacy not just private communication and just you know to set to make sure we all know what we're talking about a secure multi-party computation in particular is when a bunch of players want to compute the joint function each player may only learn the input and the shared output and Yao in 82 and onwards proved that this can be done for any function f with only a polynomial blow-up even if some of the players collude an attack the protocol won't necessarily end if there's a big attack but no none of us but the secrecy will be secured and that's important about okay and the way to think about that if you applying this

in your business is as a using simulation using the the concept of simulation or what I'd call disintermediation which is kind of a decentralization but in a wider context so what you do is you design your business logic or whatever as if there was a Judge Judy that is completely tested then we'll definitely keep your secrecy so you design of algorithms and protocols poet trusted intermediary well I just send my algorithm my main put x1 and the flower cop will just sender x2 Judge Judy will compute the answer and return to both of us the output and house and subsequent results the Q multi-party computation results show how

to simulate that without having a judge yet or just with a cryptographic protocol okay so that's the most you know most elegant I think private computation primitive and and as we said it improves privacy which these people care about a lot but that's not the important part I'm going to argue that the interesting part is actually that it opens new avenues for collaboration more privacy means there's more information that can be shared and more valuable and so for example oh and even to belabor the point more boosting privacy in existing systems is something you should care about less because if you think about taking the current banking system and making it more private it's really

nice but there's a - what it can do it's like you know going a tree in a pot it will only get you as far as the current banking system but if you create new collaborations about data that was not shareable before you'll get new business opportunities so you can compare that I am a child of the 90s I love comparisons to the tech in the 90s so you can compare the left one to online banking which is really nice but creates banking and the right one to something like Kickstarter or PayPal it really creates new stuff that you couldn't do in a pre-internet era so which data so we're left with a big question when we come to select our focus when we wake up in the morning come to select our focus for the day what data is to private to share what

can we focus on collaborating on right so it's mostly in the business domain because visuals keep everything secret like those various compliance seasons in other is less in this capacity that make businesses a really good target for this technology and that's one I think one of the main takeaways here so if you think of for a private person the thing that is most secret to them is maybe their DNA I can think of I don't know many people that would post a complete sequencing of the DNA online and that's kind of it also anything that you post on chain because that's really really online that's available everywhere so arguably you really want to keep private the stuff that is on chain and your DNA that's basically so what's the

opportunities here from stuff that was not done yet because it uses private stuff private data blog research is really great and privacy coins and private smart contacts those are some really good use cases however for the business cases I can think of many more use cases that are interesting so for business everything is private in particular the cash flow super private client data they can share for all the reasons both compliance and secrecy any trade secrets or IP any compliance information information that is secret due to compliance reasons and basically everything so if you look at the business they need everything private and so there's lots and lots of opportunities to apply private

computation and some use cases and you can figure out many more options a really great and in fact with the first use case exactly for these reasons in Denmark supply chain information collaborating on supply chain in making it transparent or open without being it completely public auditing and loans which we'll talk about and so on so are we doing on time pretty good so just to finish with the b2c cases there's four on chain data there's great project big cash-money encoder which I'll privacy coins and enigma and carries and many others which are a ways to take on chain de taos private smart contacts and make them make them private and there's

really good use cases around ocean and open mind for opening data silos right so so for example for blood research where many companies have DNA of clients like 23andme you'd maybe share your DNA with 23andme and not with anyone else but we can we can envision and I'll go into between 23andme and other drug companies sewing other DNA sequencers where they would operate an algorithm together for the you know for drug discovery and stuff like that so in general massive data collaborations in the mobi to be area the first application was sugar beet auctions so basically you clear you have all the participants with your businesses submit

their bids and you clear the auction without anyone learning how much anyone else is willing to offer this was replicated a lot of times with spectrum auctions which is where the military releases frequencies and and the cellular operators want to auction you know want to bid on them and there's even a book about this and another use case that I really like and I discussed in more detail as well so if you want a more detailed discussion of this ask me but basically algorithmic taxation so right now taxation systems are super legacy systems the pen and paper ready because this kind of so they have to be filled by hand very often and computable and checkable by hand by you and also these kind of people are the people who

have to decide on them so this whole system is completely legacy when in the five for-profit sector public M private companies are optimizing everything about you so clearly there's a mismatch there and the argument is that if you go to the future you will get an algorithmic redistribution of wealth that is the collaboration technology basically it uses all of these components so algorithms for reinforcement learning for optimizing the taxation and redistributing it the best and mechanism design for making sure none of us cheats during this process and economics and public policy for figuring out what it is that we're trying to optimize but also cryptography for SPC for actually computing this on people's divert data which is which is

necessary for real good optimization of redistribution of wealth other use cases so supply chain auditing alone so resonance is a really great company from the UK that basically does supply chain we joke that it's optimizing chicken delivery so so basically chicken mini edge the people who go chickens really need to really need to be able to access the information about how much is the demand side and how much and how is they what drugs and they and the demand side needs to access information about what drugs the chicken get and so on and

those huge I study in this system but they're not connected to each other they don't talk to each other because everything is trade secrets if you apply zero knowledge proof so is process you'd you'd optimize it a lot it's to eventually sue for anything alone food it's too for anything about supply chain you know for materials centrifuge is doing kind of a similar thing but for finances supply chain finance and how does money move in the system and how do we actually make money and finance move more efficiently by without me telling everyone we as a business telling everyone what my cash flow needs are in in whom how many clients I have and so on and still allowing efficient financing right now finances is someone

who has a VC I can tell you finance is stuck in the past like I need to talk to everyone manually and figure out what the financial needs are and what the what the rates of what are the eleven and so on and they will not post this online but if we could put all of this information online encrypted and allow controlled computation on it things would get much more efficient and in some way I will argue that so the question is which projects should you work on right and I would argue that first of all you need to think of zero Lodge posts and other private computation is not PI not about privacy but about collaboration and then when you wake up in the morning figure out

which project would use private computation in a way that most boosts collaborate collaboration in terms of dollars measure it in any way you want measure it but walk by what kind of collaboration you'd feel best about but try to boost collaboration and choose those projects to work on and finally about the collaboration texts of the future in general these are technologies that reduce friction align incentives are available to the masses and allow and better collaboration allows this positive sum effect which increases the pion compounds rewards thank you oh [Applause] that's a long time okay questions online offline like there is good

I'm not sure I understand the question let me let me talk a second about this so in the auction design you use sewing ojos and I think in subsequent projects they use a truthful auction so quickly auctions also known as second price auctions which house that called not honest a mechanisms on tooth foot so it's called truthful mechanisms it's mechanisms well you have no incentive to say something that's not true so you design a game theory it's there's a whole field called mechanism design that one of the many great achievements of this field is that like it's economics and computer science and one of the

great great achievements is that they designed games it's about designing the game so that you will only be incentivized to to reveal your two purposes right once you design an auction in a tooth --fill way you don't have any reason for you know people have no reason to cheat so that's the point of of these sugar beets application they design the truthful option and they cleared it using SPC and you know if you think of a future use so this use so you'll ask me in in a future in the beautiful future I'm imagining you'll ask me how much money do you really need to live and you know how sad you feel today and how much do you need to eat I

don't know you'll ask me questions order you'll you'll try to make me reveal my needs without me being able to lie and that's about that's why you need the mechanism designing you you need you need to design a taxation mechanism that is truthful and in fact the best place to look at this is all set called the radical markets right all the radical markets book is basically about that it's about mechanism design for making truthful some of these mechanisms are really truthful some of them are only kind to school very you know but but the whole point is try to get people to reveal their preferences that's so for a helper girl text right that's the exactly the point you make someone

really declare how much the property is worth because if they miss the close it they will get good does that answer the question they do because it's privacy because you don't I don't want necessarily in many use cases I don't want to reveal the value of my house so when I say to the government hi my house for the helper girl needs is as well a million dollars for a private person it's probably fine but businesses start having problems with us and when you get into the tiny details of stuff you'll need your car and your detailed belongings and so on there's a lot of privacy violations that will basically make this untenable without conserving

people's privacy and that will secure multi-party computation comes in especially here like this is a radical market kind of I should really update this in the spirit of very commercial exactly exactly and and my guess is that this is why almost was the first place with this was deployed because they have a really good economics and game theory group as well as a really good photographer group and I suspect that that's the magic that happened more questions okay so thank you very much and tell me which applications you're

looking to implement as you can see I'm missing some it so I'm always looking for new ones and if you have good ideas come talk to me and I think there's lots and lots hiding here this there's one called auntie Qi it's about improving financing for improving private equity basically using zero knowledge poops there's there's so many stuff to do here so if you want to talk about this I think this is the most exciting one of the exciting things happening right now in tech honestly I mean I've been waiting for this for many years so so anyway thank you very much talk to me about your startup so when startups and so on and [Applause]