Rethink Trust 2018: Johan Pouwelse, Distributed Trust Design
Recording: Rethink Trust 2018: Johan Pouwelse, Distributed Trust Design
so sorry it was caught a bit early so I can talk for ten weeks so how long can I talk now 20 minutes on the dot excellent let's see if my clicker works oh dear first light thank you oh that's second actually Oh beautiful yeah so thank you for this final talk coming to this final talk so I'm leading the lab at Delft University on a on all blockchain and trust issues so together with eight professors we're running the largest lab on if it's kind in Europe so our input is currently 960 bachelor students in computer science and our output is currently 7-0 70 masters of blockchain per year and so we we have one year of academic level of master level blockchain education and so that's things like distributed algorithms blockchain engineering all these and consensus elgrand all these these things so if you want to download these slides and other stuff this is the Twitter handle we just did a new release of our software I'm gonna talk about our new token economy so what we'll be talking about today is that we have something which is special we see a lot of things about Aiko's and creating money without banks the next thing for us scientists is that we can produce trust with software I'll be sharing my experience so I'm a grandpa in the blockchain I've been doing this sort of trust building and building tokens which we now call togas for 18 years that we create trustworthy software so this is one of the longest-running examples and it's been a bit in the in the in the backburner a bit of silent we cannot really cut through the noise that's some people with marketing people so we have we have zero marketing people we are a hundred percent engineer or scientist team so eight professors and quite some student in the lab and it's just bit torrent it's a BitTorrent client on steroids and it includes a distributed accounting system where you can have bandwidth as a currency I'm gonna explain the details all in the coming slides all of this can be documented so my first academic paper ever open information pool so this was a few years before Wikipedia started we had an academic operational platform and so I co-founded this with some real programmers that we had the first encyclopedia of music where everybody could edit this was and you needed to create some sort of form of trust and this somewhat worked 18 years ago and then we switched to try blur and I've been improving this ever since so here these are not white papers these are academic proceedings and this all is about trust I see that the history of trust has changed a bit as we first started to do gossiping in 1995 on a mailing list on who was trustworthy and who was not who was a good seller and who could not be trusted there was a interesting things that was Craigslist this is how instead of calling people or building personal rolodex and then we had a website where we have patents on this and that's called eBay and now you had a few wines and a few beers that you get into the car of a total stranger and he drives you someplace and this transaction has been established after loose contact on the internet and the middleman who established this transaction they refused to take any responsibility have you heard of this example is it profitable uber does it sometimes even work that's the remarkable thing then we had mount GOx how did that turn out that's one of the biggest bank heist in history then 2017 we made a system where you could connect ebon accounts so but more on that later so before Christmas we'll have a distributor trust design a reusable mechanism for creating trust so this is as we were getting more people in our ecosystem so we now have one point eight million people in storing our software so 1.8 million people and then we want to do a bit more applications so the interesting the model we're creating here is based on the MIT boy model from 2001 how do you create trust this is what's been scientists have been working on for four decades it was very simple model you have a reputation this is a personalized perception how trustworthy somebody is based on certain information then you have a subjective expectation of trust how will behave in the future and then you do an action and maybe it will be wrong maybe it will be right and so it can increase reputation you have some trust there so this is called what we're doing this self-reinforcing trust model so as long as you can identify the magic people who are good you just need this algorithm to define good and evil it's slightly trivial maybe not as long as your as as long as you can do that in software and oh yeah and and if you can do lifelong punishment that would also be helpful so we have more on that later so this is extremely difficult this is the fundamentals if you can do a reputation function across multiple domains then you have it so that's our big thing we launched this in 2007 then it was so I'm a bit of a grandpa in the blockchain so this was distributed accounting system there's only 20 scientist in the world Stan who were operating in his field and we called it foul shares forced to play fair bandwidth like a currency nobody cared 2007 it had you couldn't blow bubbles with it there was no speculation in it nobody cared how did it work it was moving beyond the tit-for-tat paradigm in BitTorrent a bit torrent has 200 million users so they do various things online so you could just see who was who was a person who gave a lot to the community and who took back and this is contributing to communities and giving back this is the foundations of various Nobel prizes at the game theory know what the satisficing Nobel Prize and of course they're all of Olson Nobel Prize we have never cracked this problem and now everybody just tells each other's opinion and then it works somewhat this is 2007 technology it's not very good but it worked at the time you have a ledger this is our notation Alice and Charlie have a transaction TX and you just gossip around it about it how many bytes have you uploaded how many bytes have you received there's no compromise of scalability everybody just gossips around this is 1995 Craigslist for computers it will scale it has no bounce to scalability but everybody's free to lie and you don't need some Byzantine fault tolerance things or or imitation of that you just need to know who's lying and who's not the magical function so this solves a freerunning problem so this was a primitive pre Bitcoin ledger so the the details of this Harvard my university and Berkeley did a nice write-up nobody cares nobody cites this work but it worked that was the interesting oh yeah we didn't call it blockchain then we call it work accounting mechanism this doesn't sound very sexy does it no how does it work in a bit more detail so you don't need go book a census but you just know about transactions Alice knows about transactions these TX blocks with well-known identifiers you just gossip about it everybody has a partial world view and Bob also has a partial world view it just knows a few transaction of people it deals with and my tree counter then we do the magic of of anti entropy it distributed database ahead in 1987 paper angel technology so to exchange their knowledge and then it sent one way and then they merged our worldview and they see that the graph overlaps and so this is this is very simple so this is what we now know as the DA thinks so direct acyclic graph Thanks so you see that when you talk to people and you gossip this transaction your worldview is more complete you can also be spammed of course because everybody's free to lie and it even worked in 2007 but you couldn't speculate with it it was just a bit horn banked with a system or token for BitTorrent as we now call or people call this sort of stuff so 2012 further improvements to this a and B as a transaction so there's no one no prevention on double spending but you don't need confirmation times if you double spend your reputation will be ruined in seconds this is a totally different paradigm there's no puzzles involved so forget everything you know about block chains this is more primitive and faster you just need a multi-party agreement system all other coins are based on mono signatures that's a Bitcoin any theorem you just sign something with your private key and then it's a valid if part of a valid block right this isn't you need multi-party agreement to form a valid block because a transaction so one of the primitives of the key foundations of databases that is what we take so we don't have a native token we just have any binary block you can put that in a transaction you signed out where your private key it becomes irrefutable there is no confirmation time you do not to give 70% of your mining power to some country in the distance it's just the simple stuff this might actually work for real-world businesses you don't need a mind frame then you do a bit of additions to this elegant model and then it becomes all a bit more complex so you fixate the temporal dimension you put something happen before something happened after and you call that hashing everybody and now we call that extreme shorting and so that you have everybody just creates their own blockchain hey you linearize all your transactions and that's it very simple model it works very fast then you want concurrency so now okay this becomes even more difficult but you now do instead of your own Genesis block you just put your own blocks and your thing when they're signed and you have various multi-party outstanding requests and you put that on the counterparty and your thing when they're signed it's not understandable just watch our line for the details oh yeah you also lose the message delivery now because you can have concurrency so you can have eventual message delivery and so it's agreed everybody puts transactions they agree to on their own chain because they signed it with a private key they can no longer denied the emergent effect which you have in normal business processing if you have transactions you have for usually fast miss they fast mixing graph property look at Alice on top there Alice has their own chain and then she has contact with various parties and they form a block there is no blocks or mining or think now it's just her own transaction she can publish it to whoever she wants he's fully autonomous very different than a normal blockchain right and so here is the ITF write-up of this of this thing oh yeah you everybody wants a Byzantine for porins and all that sort of stuff so we just put on some students to get that working so yeah if you do use some cluster on top of it and so that's the final layer where you cannot do any forking because you signed it with your private key and then occasionally there's every few seconds there's a checkpointing running it's like you need to tell us what your status and if you don't put your state just compress to 20 bytes in the common pot or you change it and you cannot show the missing pieces people don't trust you that much so this is again not puzzle based but this is trust base so this is quite different than all the other so this all works very well this got accepted to the wizard I Triple E blockchain conference 2018 I showed you the 2007 slides this is how it looks today a BitTorrent client with a bandwidth token and integrated decentralized market and all that stuff oh yeah and also decentralized voting system so there's all very strains it's sort of a YouTube without any website it's academically pure no server is involved so you can do voting for channel or decentralized you can do you can find decentralized you gossip around what interesting BitTorrent storms you found there's relevance ranking involve iam twenty five hours all over source of course and github you can see it up there with one of the good guys I'm paid by the state until March 2004 to huh so we're socialists here this is not ICO funded this is just government funded then students did a nice trust visualization on that so you just see instead of tit for dad who's giving you banter then who's not and then you see who you're interacting with just to make it a bit more spicy so instead of just downloading BitTorrent we implemented we'd have the second implementation of the Tor wire protocol without any of the central servers like the discovery servers and and all these things so we have our own network a so the Tor network is very mature and hard core security so we did a full python implementation of that and they didn't like it so much we were wire compatible but a lot of there are high level things like that or has to to congestion control loops in it so we shoot in time should be ten times faster than them and they are have less than a thousand exit nodes for their entire 12 years that they're our network is life and we will have some interesting things for exit node we have BOTS as exit notes anyways so a BitTorrent client with tokens and poor tor level security okay that's already a bit of a mind step and then we also integrate it because if you download a lot and you use a lot of proxies then your counter goes down you spend a lot of these bandwidth tokens you just go to the integrated D central markets you you just give it your wallet e and a for bitcoins or it may just you just buy ourselves some more tokens so this has a fully integrated D central market to have a full dedicated PhD student to get this secure and sort of works yeah as a university we try to scare people and be relevant to society so we created the most scary thing ever which is some cryptocurrencies have crypto kittens these cute things which multiply things we have determinate so it's a self-replicating software which owns money and it owns money which the University or no normal human controls anymore so it's money under full control of a robot and that robot has a very primitive genetic evolution algorithms so we're building towards a robot economy where our socialist where there's no profit driven entities controlling the AI it's all people who inherently are good and get paid no matter what's control the AI maybe that's a better model we we don't know so we specifically make this stuff very scary and we also work with ethic professors to see how this should be dealt with that there's you meaningful human control in the loop so this these things can buy some servers this VPS and VPNs you can just buy it away bitcoins any use this buy another one mass subscription and so this has taken us three generation of thesis students or something like we sacrifice for science something like 15 students to get this working over the various years there you see the details and if you encounter a cup chat what do you do you just enslave some humans and pay them if they solve the culture for you god I didn't believe the students have that work but we really do have it working so what you do is say oh I have an animation thank for that it works so we created this toolkit for robots on the Internet that's called claw term eight so that's also really a official pie-pie package can't believe they accepted our full request so you just have something that multiplies on the internet and it sends out Twitter fee tweets that it's in life sometimes and then you have an exit node and you just drive VP VPS and just because you you want some protection when you're running a tor exit Nollan trying to make some money with privacy it's it's good to do to also automatically buy a VPN have you heard of this stuff before anything operational like this thank you yeah okay but this is all fun this is sadly this is my home of the hobby projects of the students this is how from our 70 engineers of the blockchain we educate each year we also need grants because we run on subsidies so we also need to do serious stuff so this is the Dutch Passport this is actually an official duplicate of the state passport which has a chip in it and it's one of the more advanced passports in the world this does not have a normal chip in it this has an elliptic curve cryptography based chip in it so fancy and so please crash this network this is the trust chain I talked about please go to Android and try to install this this app and let's see if it actually works this is a phone to phone network there is no server involved so if you start up this Android app your IP number will be revealed to me and others you can even spam me with unicorns as my students have done when they're doing this in the class so it connects to other phones and it does not reverse all these things you can send the--and to people he sends PDFs to people you can put signatures in some form of it you can scan the Dutch passport it can interact with that you create the technical stuff ah some people have found it if you really know what you're doing you find the hidden feature where you can do stress testing of the network you can spam everybody here with running 50 clients on a single device will slow down so this is also for embedded devices right we're nearing the end of the talk we've seen a lot of identity and that's one of the welcomed Melvin one of the things has been building trust and building Ledger's over the past decades it needs some legal certainty if you want to have a sort of a Panama company if you want to only operate in the in the gray zone if you want to do anti-money laundering compliance if you want to do kky see know your customer you need to have some legal entities if you want to trade on various exchanges you need to leave your passport if you want that or not that's fine but if you want to have a healthy blockchain or not you can you can have the tour things and we try to combine everything so we combine tour and we turn combine it with a valid identity of the state so this is citizen number fourteen this is actually the person in charge the chief architect of the Dutch digital electronic identity and so we're working with these people and they fund our research that we're trying to be the first country to have a significant economic activity on the blockchain so we now see that the large portion of our economic activity goes through the internet this country wants to be we have our biggest port in Europe and with a big airplane hub we want to be the first where there's meaningful economic activity the majority goes through some trusted information flow we don't know what it is it might be called something like a blockchain it may take us 10 20 years but we have actual trust on the internet and we have transact so this is our ITF details thanks to my ghostwriter Dan who does all this stuff in the weekend and and you can see the level of detail that we're now at for this legal complaint I still have five minutes oh that's beautiful right Wow this was already my last night okay so there you can see that the European Commission they like that sort of stuff we try to pretend that we're sort of one happy big continent and that we have uniform rules and things so if we would have a european standard for trusted identity and trusted electronic business transaction and sort of a reputation you can rely on they liked that they gave him lots of subsidies for that over the years so the European Commission is helping us quite a lot both publicly as you can see with that this URL and also behind the scenes my final slide already [Music] to conclude so we've been trying for eight this is this is me without the receding hairline by the way when we're still young and naive it will take probably another five or ten years before we have any meaningful trust and and this magical public key infrastructure that we can actually build trust and trustworthy things that scale we don't need mainframes in control or we didn't need some sort of ICO and utility token who owns the Internet the same answer is who will own the blockchain probably we need a self-organizing and self-governance layer for the for the next layer and there cannot be some guys with girls with commit rights on the global internet trust layer right we already have these monopolies called DNS look for that happened and so what we do in the lab is running called real applications my students never get their PhD or their master thesis if they do not have running code this is a bit of a bit of a thing so that usually means that when they graduate they actually have running code yeah you can read the details in our propaganda or draw our Twitter handle we actually tweeted 50 times now in our history of our project and if you want to know how bankers and and Chamber of Commerce land registration agency our vision and our running code for putting this country this kingdom to be the first on some sort of public utility thing this is our Journal article on the running code and our architecture for making this happen so thank you very much