Fred Ehrsam & Trent McConaghy: IPDB – The Interplanetary Database and its Applications in AI (184)

Transcript

[Music] this episode of epicenter is brought to you by Jax Jax is the user-friendly wallet that works across all your devices and handles both Bitcoin and ether go to J ax. and embrace the future of cryptocurrency wallets hi welcome to enter of the show which talks about the Technologies projects and startups driving decentralization and the global blockchain Revolution my name is Sebastian KU and today we're doing a very very unusual episode for one I'm here by myself uh Brian's away on

vacation uh in the south of the US for a couple weeks and so I hope he's having a good time there and Mayer uh is at consensus where I'm sure he's at the moment you know having fun and seeing some great talks and panels and things like that I unfortunately couldn't make it um and we're recording this the day before this releases so you know it's it's sort of unusual but hopefully we'll have a great conversation with our guests today also uh something that wasn't planned we have an extra guest so originally we had planned to do this with Trent bcon who our past listeners will know as the CEO CTO and founder uh of scribe and Big Chain DB and now ipdb or at least you know part of the the

founding team there and uh he he came on the the the call in his hotel room with Fred ersham who's a former member at uh I coinbase and for co-founder at coinbase and yeah so we got Trent and Fred and so Fred is going to be with us for the first half of the show because he's actually got to go meet mahair um in New York and so Fred's gonna be talking to us about Ai and Dao and sort of the intersection of AI and and blockchain Technologies and then for the second half of the show we'll be talking to Trent um about ipdb and what's new with Big Chain TB and this sort of thing

so hey guys glad to have you both on yeah thanks for your patience with us no it's it's uh it's no problem at all and um you know I I like structure and when when things when things are kind of out of whack I get uncomfortable but um I think I think I've gotten over the hump of you know this is going to go well well yeah we'll make the structur unru great uh so since Fred has to leave in a few minutes to go me with my hair um uh one one topic that uh uh Trent thought it would be great to have him on would be to talk about the intersection of blockchains and Dao and this is a topic that I haven't really you know thought of much I i'

I've of course heard of people talking about it and it's come up recently but it's something that honestly I have not really delved into very much so um we'll spend you 20 30 minutes talking about that I'm interested in in hearing how blockchains can intersect with uh with uh with AI um so guys uh well specifically Fred tell us how how do Ai and blockchains come together what what's what's the main thesis here when when you want when you're talking about Ai and a Dao yeah sure I think Trent has done most of the thinking in this field and I have probably stolen a lot of his and then maybe added one one or two side

layers while he's probably built further up the stack than I have um the way I would describe it is um at the moment there are a couple Tech Giants in the world the most valuable companies in the world the Googles the Facebooks Etc who have made very valuable companies out of big proprietary data silos um and due to network effects those data silos have gotten bigger and stronger and the companies are really valuable because they effectively can extract rent on these data stes because only they have control of them own them can access them um and so goes the cycle um and I started thinking about you know

what implication does that have on artificial intelligence and uh as Trent would very quickly point out this is potentially a dangerous Trend because if it if it's true that AIS really feed on data and they really better because they have a lot of training data they have a great breath of training data and all of the data is owned by a couple for-profit uh private companies than when really intelligent AIS come out that who whose interests might those AI serve um probably that of the centralized company if that's where all the data is coming from um so that that alarmed me um and I started them thinking about well what

Alternatives do we have and I realized in some sense that blockchains effectively represent a big open data set and um now that we have you know sort of blockchain tokens and we've always have since Bitcoin I mean to make that sound new but with blockchainbased tokens it's this kind of magical system whereby everyone is incented to contribute to effectively one big database via the blockchain and that seemed to be a potential kind of Savior um in in this context so I guess that's that's how I think about it and that's how I started to get convinced that um the blockchain in effect can represent the biggest open data set um in the world for AIS to train on and

that actually might be a great thing for all of humanity as as a result yeah is that accurate in your your largely and um My Hope Is that it's a great thing it also is you know potentially dangerous on its own um but we have to hope and we have to you know Infuse our ethics and our opinions and do what we can to try to tilt this thing towards something that's good for Humanity right um and overall the thesis is strong you know to summarize I think it's it's valuable for the listeners um uh modern AI a lot of the stuff the Deep networks all this really live and thrive on data um you know you don't have to change the algorithm much you can just have 10x 100x more data and your error rates can go down from 20% to 5% to 1% and so on

so it's really about the data you know the data is the Moe this is actually you know the how Google built its house in the last 10 years and Amazon and others um so and you know there is this data Moe and a Data Network effect as well actually because if you have better data more data then you can get better models you can attract more users and you have this flywheel effect that goes on so it's it's been very powerful for these um incumbent companies that have more and more data and I've been asking myself the question too similar to Fred of you know what can we do about this right to sort of um break the cycle um and um open it up to humanity more broadly right such that um startups who want to compete on AI for these uh data incentive intensive type AI systems have

a chance and so um an open database um or at least an open exchange of data um can really help um where you people are incentivized to submit data sets um and you can actually have in two flavors right one of them is where it's just sort of sharing data um but I think actually to make it really incentivized well you really need to have it where people are getting paid for the data sets that they um you know it's it's truly an exchange right um or at least a marketplace where you can buy and sell data and in a very fluid fashion and um to this note we've actually we've been spending a lot of time with bigb and ipdb um helping to set up ecosystems for data exchanges and one that we actually just announced today which I'm pretty excited about is

um in the world of self-driving cars which many people call the the killer app of of AI um uh data is a big problem you know you you can't have just 1% error rate because that would mean you know one crash every 100 miles or something right um You have to have uh you know one in a billion one 100 billion failure rates and um to do that actually it means you need orders of magnitude more data so maybe you only need you know say a thousand or 10,000 miles of training data to be 1% error but to be you know one in a billion error one in 10 billion error uh you need um hundreds of millions of miles potentially right something that no single automator can get on its own so what if you have a system where they can be incentivized to

submit their data and then um they can train these algorithms together um collectively so this is actually kind of a win-win win right because the automakers each can make their cars with with um more accurate models that you know saves lives at the same time um getting benefit for all their hard effort for um gathering this data and of course it will be not just the automakers putting in the data then you'll have you know lots of people driving and trying to figure out a more efficient ways of putting this data in so that's an example of where I can really help but it goes beyond too right um and yeah so data exchanges um for AI training data um I view they will be a fundamental um piece of Internet infrastructure um and it will extend beyond just self-driving cars and

hopefully help to equalize the opportunities for people um deploying algorithms AI algorithms okay I got a lot of questions here I mean so first let's uh let's let's sort of restate the problem the the the the fundamental problem is is sort of an extension of the existing problem with regards to data centralization so you know we have Facebook for instance or Google and they hold an enormous amount of data uh as users we sort of give them our data in you know in in exchange for free services or ad supported services and those companies get an enormous amount of value out of that data because of everything that they can learn from it

and the the the core problem here is that one once we have or as as AI continues to evolve uh that data will also serve AIS and in fact make those companies more and more powerful because since they're the only ones that have the data which is essentially the you know what you feed into an AI there are the only ones that can effectively create you know really comparable uh and and High um Quality AIS that that's sort of the the fundamental problem here is is this decentralization further it you know enhances or supports the development of strong AI by strong incumbents that's right okay I would actually bifurcate it into kind of two

parts one is the one exactly what you're describing where you have um an incentives issue whereby if apple is creating Siri Siri is probably going to serve the interest of apple and not the interest of the person with the phone and that's scary for for us as humans so that's the issue you're kind of describing describing the other issue is a rate of innovation issue right so if there are only four big companies who have these big data sets then the number of people in the world who can create really great AIS all of a sudden is very very small because very few people have access to the data through which to train AIS and um make really great AIS now if that data was exposed to everyone

in the world now everyone in the world has the ability to create great AIS and the rate of AI um technological development goes up drastically you might Imagine by one to two at least orders of magnitude um so it's it's both kind of an ethics and safety issue and a rate of innovation issue so on the ethics part I I think that blockchain Technologies is sort of talked about a lot as the platform or the platforms that will allow for data to be liberated or for um users to regain uh their their data essentially where we're talking about um allowing

for users to give access to their data to you know large uh internet companies through access controls through encryption this sort of thing but essentially the data ownership users regain ownership of their data and they they delegate the use of the data to large companies now this this shift in itself you know probably if if it were to happen would take a very long time because we'd have to find new incentive models for those large companies to provide services where they don't really have access to you know free and incumbant access to the data as they have now and that seems like an necessary First

Step uh if we are to have these um open data sets uh where users can then essentially rent out or provide s sort of paid access to their data for other AI systems to learn from how do you see this PR this sort of progression towards what you're talking about so I see it as just where where are the problems you know the ones that you can address first right so you could try to do it for you know just personal data in Facebook but um to your point Sebastian that's actually you know is there going to be enough value to Consumers and probably not right um but if it's for something like self-driving

cars where it's clearly a win-win win across the board then there's a clear incentive and there's other ones where there's clear incentive too you know like Medical Data records right where um people are merging their their data across um different silos right like I live in Germany and I go from hospital a to hospital B Hospital B doesn't have the data that hospital a has on me um so I would have to actually go around and collect it right now and collect it and share with them but it's really hard so it's much cleaner if you can actually have this data out through in the cloud encrypt it Etc and then I can just provide access to the hospitals as needed right and then also at the same time I can provide um permission to scientists to conduct research on this data so that's sort of a societal

benefit so there's benefit just to myself personally for me and an incentive for me to to share my data that way but also um to society and so kind of go across the board and you can ask you know where are there benefits for um data pooling to happen um and you know I gave a talk earlier today and I gave other examples too with like diamond fraud and with 3D printing fraud and sort of a lot of other problems in fraad in general right so but it has to be problem based it can't be just saying okay well you know I don't like that Facebook has my my data um well you might not like it but um you know a lot of mainstream won't care enough to change so I do hope that we find a way that um there is some way to repopulate

a social network that is more uh open and equal and so on and I bet someone will crack that problem in the next two or five years um but um I don't think there are clear answers today on that yeah I mean it's not just social networks I mean just think of the enormous amount of data that Google has and not only the the data itself but then there's the metadata and you know the the the um the regressions that you can you can make from that data from which you can learn a whole lot of other things what what types of incentives would you see I mean I can see the incentives I can see how on the on the health side there's a there's a real real win-win there on the smart driving car side uh how that may be possible

because self-driving cars um don't really exist yet so there there there are no like sort of standards and normal operating procedures in that space so you that that may emerge as as sort of the standard ways to have open shared data because we can recognize that it's benefits for all so on on the sort of more traditional incumbent uh large internet companies as we call them in Europe gaffas on that side how where how would you see that shift happening if if if it were to happen what would it take for um a company like Google to want to open up uh their user data on an Open Marketplace where users in fact have

control and access to their data directly and and seed that control and access to to Google to have access to their services I think the answer is that it's a classic innovator's dilemma problem and that actually may not ever happen yeah um just because they have so much revenue that's that is really just dependent on extracting rent from this centralized data sto so I think the much more likely scenario is new projects that organically emerge which uses this which used this model first in the same way that kind of Bitcoin emerged with a token model and is you know is taking over started to take over money um it wasn't you know a bank for example that was willing to cannibalize their own

Revenue with this radical new model what do you what do you think uh I think that that's possibility I I just thinking loud though actually there's a couple of ideas that have that popped into my head related that have been emerging the last few weeks um well one actually has been emerging the last few months and this is gdpr in Europe right the general data protection rights and this is sort of like a Y2K um for 2018 in May of 2018 anytime that you um Facebook wants your data they have to ask your permission right but and it's but it's it could play out in one of two ways it could be where you just have to click another okay button just sort of like a cookies thing where it doesn't really matter or it could play out in a different way just breaks UI for every website I know

right and that would be the sad way and that's a possible scenario but there are a lot of startups and a lot of money flowing into systems where it's actually about um tokenizing consent for example and um this actually could um really reduce the friction for um other players to come in there because if you're sort of tokenizing consent then that's actually moving a lot of your data and giving a lot more power to you so that's pretty exciting that's a possibility there's another one that I've been thinking about and this is actually more General on the tokens thing actually largely thanks to Fred's excellent writing in his blog so I encourage any reader to look at his stuff too so uh and I think of it as we we've been thinking about token Design Systems where it's sort of a new project from the ground up and so on but you can also

tokenize the Enterprise um and let me explain well you can tokenize the Enterprise and the result is that the Enterprise might be able to melt into the community and I I'll give a couple examples there's really two types of Enterprises to think about here there's ones like the 3M of the world or the Amazons where they're already internally five or 50 different business units that have their own profit and loss statements Etc each one of them could actually create their own token ecosystem internally but also um open up to sell those tokens externally and so you've got they're they're used to having apis to the world right so that could be very very interesting imagine just even Bezos does you know tokenizes one business unit and sees what happens right and what would be his incentive um the incentive would be that if you're including um more people to be able to

buy into the tokens and there's greater demand than the actual overall value to the to the Enterprise goes up right so it's sort of like a second IPO like why can't an ipoed company Ico on top of that right so it could be an interesting idea and then you don't have to have it just worth you know five or 50 different business units it could also simply be one big more large company like Facebook you know ignore the Instagram part of Facebook Etc just Facebook itself is one larger monolith what if that was tokenized where the users then could start to um um not only buy tokens but they can get tokens of benefit every time they use Facebook right so then actually Facebook overall the value to Facebook the corporation could be higher and that might happen right this is just a an idea I've been thinking about I

don't know if you have thoughts read tokenizing from within right melting the Enterprise into the community I think it's a very smart idea I think the likelihood that a $280 billion company actually does it in the public market is low I agree but it would be awesome it would be and but here's the thing right like you know two years ago four years ago we thought that these icos were anomalies right and it was sort of like you know one every two months or months just like you and I were talking about so you know what is weird today could be normal tomorrow right and maybe it'll start with you know just some Visionary company right and we're seeing like with the Overstock guys right um they're they're doing some radical things and others are following right so so um something is impossible until it isn't

right yeah no totally so you mentioned a while ago data exchanges or data marketplaces what would these look like and what technologies could they be built on so when I think of an exchange and I'm sure Fred will have ideas too but I think of it as a marketplace where the price is set automatically right you have people coming in saying I'm going to pay X dollar for this and other people are saying I'm willing to pay yll for this right so you've got a a set of bids and a set of asks and when they match up you know when someone is willing to pay more than someone else is asking for then they line up and that get sold right and um so you just basically the core of it is maintaining these lists of bids than asks um which can be fairly simple right at the very

very core of it then in terms of Technology sacks what we built recently we were actually surprised by the Simplicity of the architecture uh it was simply ipdb which is the public version of big DB U which has governance Etc we'll talk more about that later and a single page web app on to right so people can log in create an account with a brain wallet type setup HD wallet that sort of thing and um and then basically start dumping in data Etc and that was all we needed um it you know it could support tokens directly it could support the querying Etc directly and we were kind of amazed right because it's people talk about serverless architectures but of course you're still using you know the Amazon cloud or something now it's serverless where it's a decentralized

cloud at the back end right now that's a very simplistic thing that's sort of a proof of concept when you actually want to go to something that is you know more full-fledged um then I I defer to Fred because he has more experience in this no that uh that that sounds pretty spot on to me I I guess I was thinking that I think some of the places we might see this first are marketplaces where monetizing the data is easy um and paying uh people who contribute the data is easy so one example of this uh that Trent actually brought up uh earlier in a presentation today was a company called numerai which is started by a friend of mine this guy Richard Craig

and um it's funny how small this world is sometimes and re is my friend their first investor is a friend of mine so there you go um and to to me that's great um first example of how this might play out in the sense that uh going back to the beginning of the talk effectively what they're doing is they are in a centralized manner paying a bunch of um uh data Outlets like Reuters and Bloomberg etc for their data and then they are Distributing it to anyone who wants it um in encrypted a homomorphically encrypted manner uh so that anyone can have access to this data and build kind of the best AI they can so it doesn't take too much imagination

to take their current model and then extrapolate it such that maybe it isn't numer a central entity who's collecting all the data and distributing it rather anyone in the world who has valuable data to make stock predictions could get paid for contributing it um and this is already happening actually like there are these kind of small random companies around the world who will pay people in different countries to go and count how many Cokes are on the shelf or whatever so this is happening to a certain extent already but I think um if you properly incent this system using a blockchainbased token you start incentivizing anyone in the world to contribute data which could be valuable in this case for for financial predictions yeah and the resulting

behavior is quite clear it's a bunch of AIS that feed more or less on this crowdsource data and the result is better Financial predictions and the output of that which is just more money easily feeds the people who contributed the data and and so on and so forth and something that Fred actually something that Fred is pointing to too there's actually this emerging sort of discipline of um designing token based systems right um even you know right after my talk Juan B gave a talk and he actually pointed you basically have to ask yourself you know what are the ways that people can earn tokens what are the ways that people can spend tokens and then how do those get exchanged right and those are kind of the questions to ask and you have to ask you know this question about each actor in the

ecosystem and then design the thing overall such that you've got the sort of different feedback loops that you like right and this isn't much different than classic multi-sided platforms right like the Googles and Facebooks of the world that have multiple customers Etc but now it's actually got um more tokens of value flowing around so you can actually have these Network effects um more directly right so related to this you know data that's in there you know you can think about well how do you set up designing um a system for people contributing their music right a music right system for like compensating artists Etc or data or you know social data all of this right so it generalizes quite well um Beyond sort of data for just AI right okay I I can see then how the the token model that we've seen for you know

in a lot of projects in the blockchain space might might apply to data so the way that I'm seeing this play out so you know for instance few years from now for for example we have sort of a public open database that is the sort of standard where all self-driving cars submit their data so you buy a self-driving car there's some sort of Alliance or a Consortium of car companies uh that build this open database for the benefit of all and they send the data or they they allow for users to sort of sign for the their data to be sent to this open public database uh the data is uh anonymized or uh with some sort of encryption

mechanism um made to be stored in the databas in the open database as encrypted data and then on the other side of that you may have a company like Google or another car manufacturer or anybody developing an AI that through this databas well would rent the data uh effectively paying the users who fed the data into the into the central system or into the the distributed database system uh that way we have this massive data structure that can serve the purpose of anyone who wants to use it for research or for building new AIS or what have you uh is

that sort of what we're looking at here yeah it's a pretty good capturing yeah and um you know there's going to be the core stuff of storing you know the the databases that store the Met metadata and some of the smaller mes of data and then the larger um streams of data and stuff and data blobs this will be in things like the file systems you know like ipfs with file coin Etc right um and there's actually going to be building blocks emerging on top too just like people have been talking about reputation systems on people we will be seeing reputation systems on data um and uh and a whole wide variety of sort of an ecosystem emerging on top of data itself right and we're not seeing this right now because all the data is locked up right so as soon as this stuff gets

unlocked with these decentralized data exchanges in marketplaces we're going to just see a flowering right and you know as an illustration of that right just very recently there was a front page of the economist talking about data is the new oil but if it's a new oil well right now we've got Rockefeller time plus two plus three we don't actually have it where it's accessible to all these you know Innovative entrepreneurial mindes well this is very very fascinating topic uh I would love to know what like a gdpr regulator thinks about this well actually there it is totally related right and um yeah the the the the challenge actually with gdpr is that the regulations right now don't have much to say about um if the data is encrypted or not right so if I have

personal data um and it's out there in plain text um and it's you know let's say I'm living in Germany and it's this data is sitting in the USA then that's not good right it's got to be right now in German servers but what if it's actually um encrypted right and then also what about right to be forgotten right so if I have my um personal data and and I delete it and it was on German Ser soils servers sorry fine but uh once again maybe it's encrypted and I throw away the private key and I literally burn the private key right and I can so cryptographically prove the private ke is burned the laws don't have any to say with this right now right so actually uh we've been iterating with Regulators in Germany on this and others have been too um and it's really important that the

law actually catches up to what's possible with the technology just like we're seeing you know in the Ico side of things where you know the law has to catch up and in the meantime we have to do the best we can it's actually similar with things like uh data protection rights that's true I mean the the right to be forgotten thing that's something that comes up a lot uh you know when I'm talking to corporates as well about gdpr is what about this right to be forgotten well okay well I mean they sort of the the the the sort of go-to line is well you know blockchains are transparent how can you possibly have right to be forgotten then you know you lay on say okay we can have encryption uh you can have things like zero knowledge proofs this sort of thing that uh would

effectively allow for right to be forgotten to exist within a blockchain uh but then the regulation as you mentioned is not uh is not conclusive on what what is forgotten if if I forget or you know destroy my destroy my my private is that is that data really forgotten uh that that definitely needs to be um made clear in the regulation Sebastian I'm thinking you know we probably only have Fred a bit here longer maybe we should talk about AI doubts for a little bit hey it sounds good yeah so the arch Dow um so there's a quick uh backgrounder on this and that is um machine creativity uh like AI Dows that are are sorry AIS that are creative for for sort of neuro domains are possible and they've been

around for actually decades right so there are um uh AIS that can generate art like visual art images that have been sold Etc um there are ones that have generate designed Quantum circuits um Le Spector and others um I I in the talk that I gave earlier today I showed how my friend Greg Hornby was creating necklaces um and the the tokens on necklaces and how they're quite pretty and stuff myself too this my PhD was on this right for um what was considered creative analog design and I was getting AIS to do it so the point is that there is um a whole practice as a field of AI or sub Fields uh for Creative AI uh or machine creativity so this is a baseline um now for the archal then uh there's a

recipe and it's best to describe in just the five-step recipe uh step one um you've got this uh agent running um you know on decentralized processing substrate you know ethereum or something else and it generates um an artifact say a piece of digital art or a jewelry design step two it it claims copyright or registers it that it just has it and this this step isn't even really necessary but it's cool to think and by the way there are regulations towards um AI getting rights in Europe um or you can actually just register in zuk Switzerland and then it's got rights because it's a corporation so that's a whole other story so it claims copyright and then um step three is it posts it for sale on a Marketplace by the way give it 10 additions for fun right like

a scribe style so post it for sale these 10 additions on say open Bazar decentralized Marketplace or even something centralized like Getty um sell them so maybe sell each of them for $1 um and maybe overall it costed you $1 of compute power so you you you make $10 because youve got 10 additions so now you spent $1 of compute power you got back $10 right so now what you do is you make 10 more pieces of Art and um each one of those you sell 10 editions you've got 100 bucks right and remember this is a an agent right it's a dowo that's running um decentralized and it's creating value on its own right and it's accumulating well it doesn't have any modes to feed um it just all it has to do is you know make more paper clips I.E art right and so it can go from having

$100 of wealth to a th000 to 10,000 and we'll have the world's first AI Millionaire right and then maybe it'll keep going maybe it'll be a billionaire right so that's pretty exciting and I like to use it because I think it's a very um tangible example of something that you can build with today's technology you know just a bunch of components off the shelf and there's a few projects now that people have reached out to me that are starting to build on this so I'm excited hopefully one will be able to announce in the next three or six months but any listeners out there go ahead and build this thing it's actually cool right there's going to be I hope a whole bunch of different AR Dows out there that get built and it's also really useful as an a Launchpad to explore what this means

because you can generalize in like five or 10 different directions and each one is interesting on its own right for example what if this thing starts generating computer code that isn't just art you know arbitrary computation what if it's mind's GitHub and people put code in GitHub that is really dangerous doing like hits on humans and stuff right and by the way it's you might think well code doesn't know how to combine other code well actually it does there's this subfield of AI called genetic programming where I spent 10 years of my life which is actually all about searching through the space of computer programs to come up with the next program and you know even if most random um programs to try is completely suck and fail it doesn't matter if that last 1% or 0.1% has something interesting that's all it needs um so

basically this is the whole idea of you know the arch Dow that AI that can wealth you might think oh that's just an oddity but um something where there's a real incentive by companies is to be Capital light so you know in the world of semiconductors all the companies uh the factories got too expensive so they all went fabulous right so qualcom nid all these guys they don't own any factories anymore right and other times you know BMW they went and sold all their factories too right and they're just focusing on design and then Outsourcing the manufacturing or Uber right when they have the self-driving cars are they going to go and spend billions of dollars or actually hundreds of billions probably to to buy their own Auto Fleet wouldn't it be easier if they just made each sell self-driving car also self- owning right and that way

they can stay Capital light so there's an incentive for corporations to be Capital light it's a better return on their assets and um of course then are we going to be handing over you know the vast majority of Humanity's assets to these plots so there's a whole bunch of questions wed raised I think it's really fun uh and dangerous and this is why we need to talk about it now so yeah you know this this this example you gave I've never never thought that an AI could create art uh you know possibly create value of that art it reminds me of this uh this um Mike Hearn talk from many years ago that he did somewhere I think it was at Google where he talks about so the you know the the self-driving car that is its own agent and goes around and does bills rides to

people then gets in the highway and then you know like takes the fast lane and pays the highway in Bitcoin and then you know after a c after it has enough money in the bank like it goes and orders another car right like this it's it's it sort of reminds me of this exam this example that he gave in this talk and I've never thought when we think of AI and sort of the threats of AI it's it's often this sort of you know the AI will take us over and will sort of destroy humankind because it deems it to be a sort of a parasite or a cancerous uh life form on the planet or you might tell AI to cure cancer and it sort of kill humans as a side effect of killing cancer but I've never thought of AI as asserting Financial dominance on

humankind and if you have an AI that starts making money through some sort of value creation and then I then I starts getting smarter and starts creating other businesses you might have like you know the biggest corporation in the world would be in Ai and then asserting Financial dominance on everyone else and just kill us by just like economically right just we'll we starve because we have no more money to buy food or any resources and the AI will have all all the world's resources exactly and by the way on the micro thing actually so the Tera zero guys it's a couple Art University students in Berlin they're doing the self owning Forest which is also super cool and um they're they're writing about it and among their references they were digging and they actually came across uh also um a Reddit

post by Mike Hearn from like 2011 or something as well so full kudos to Mike for actually like seeing this way before almost anyone and I'm sure if you dig I don't know if they found anything specifically that Nick Zabo had or Ian grig but I'm sure there's some stuff that those guys had too right because they always do right so um you know to me I just happened to have thought of this sort of from my own angle and stuff um but overall there's been a lot of thinking before me and there's going to be a lot more after too and you know I think it's a conversation that we we all want to be having um because uh we don't want to put ourself in the situation where we are subservient to these you know new land owners right I'm sure this is sort of uh talked

about is that AI will perhaps wipe out a lot of you know knowledge work in the next you know 20 50 know 100 years yeah but you know creative work has always sort of been thought of as untouched right like creative work will continue to be a human endeavor and all those Starving Artists are just going to yeah be be even more starving because the eyes will have taken over their their creativity work so I think there's a key Insight in what Trent said too which is that um if you look back at the history of Corporations and part of this is stolen from Simon De La ruier in a in a recent blog post effectively what corporations did is they allowed us to in some way pull resources to have a greater scale of uh

humans in self-organization and corporations got really big and accomplish things that humans had never done before as a result of this and then the interesting side effect um of this was that humans eventually became equivalent to a person in the eyes of the law uh or sorry corporations became equivalent to to a person in the eyes of the law and now I think this is kind of the next logical step where you know there is no difference uh in the eyes of the blockchain between an AI or or any object and a human so they're the same they can control resources in the same way uh it turns out that an AI may actually be much more

effective at feeding on huge amounts of data than any human could be and thus can achieve much broader scale um and because the blockchain is asset based and it can fund itself in doing so so you just um it it it kind of follows that this is the next natural step in self-organization and how we and how we view the world yeah on the blockchain no one knows you're an AI yeah okay well this is all really fascinating we should definitely do another show sort of a full show just on this topic where I have will where we'll have done more research and and we'll have more insights and and hopefully we'll have a on my side a much more interest intelligent conversation but uh but uh yeah Fred thanks so much for for

coming on uh at such a short notice uh and uh yeah I think you got to go meet mire I do yeah so TR thanks for inviting me uh impromptu and it was a pleasure yeah okay well we'll have you on at some point again and say hi to my hair let's take a short break to talk about Jax Jax is your wallet your complete user interface to cover all your blockchain needs I've been using it and I've been loving it now Jack supports a lot of different cryptocurrencies supposed Bitcoin ether Litecoin classic zcash or rep and they're adding many more keep responding to users needs now with Jackson nice thing is that you can manage all of those coins within a single wallet and you are in control of your own private

Keys they're not on their server and there's a single 12 ver seed that you can use to back up your wallet all your coins and sync them across different devices talking about devices they're on pretty much any device that you can think of you can get it on PC P Mac Linux you can get it on smartphones like Android and Apple and iPhone you can get it on tablets or you can there even browser extensions for Chrome and Firefox and on top of that in Jack you can actually exchange different cryptocurrencies for each other because they've integrated a shape shift and more Partnerships and Integrations are coming down the line in 2017 that are going to make Jack even better so Jax is really making blockchain and

cryptocurrencies accessible for the masses easy to use for the mass make sure sure to get your own Jax wallet at ja.io or you can get it from any of the app stores you are using we'd like to thank Jack for their support of EP Center all right so yeah Trent so coming back to the we had attended to to talk about for the for this episode which was ipdb and Big Chain DB we first had you on in 2015 uh and we talked about a scribe then of course uh that was sort of your main project at the time and then you came back a year ago I checked I last checked it with was about 50 episodes ago uh in in April um to talk about uh the work that you've been doing on Big Chain DB and how how that work had come out of a scribe and what you've

learned what you had learned from a scribe uh and at that time when we had you on you sort of hinted at the fact that there would be a public Big Chain DB Network I remember when you told me about that like I was completely blown away by that idea I I talked to people about it extensively like my team and like people I met I was like there's going to be a public you know Big Chain DB Network and I thought that was really a really fascinating idea so yeah tell us like how how have things progressed for you and how have things progressed in in in terms of sort of big chain and how that idea now has turned into you know the ipdb network yeah sure my pleasure and once again thanks for having me again um so uh going back to a

scribe right um when we did a scribe the goal was always to serve the creators uh we saw that our friend friends artists um and all throughout the globe you know artists were getting a raw deal um not only artists creating digital art but musicians and otherwise right so this was always sort of the the driving impetus and as we went along U with the Scribe of course we we ran into issues of scale we we had been building on the Bitcoin blockchain and so we decided um and we looked around and you know talking to different Friends building some of the blockchain systems then and none of them were really approaching scale um by leveraging existing Big Data Technologies right um so we we did we said let's let's start with an existing database we started with rethink um more recently we've also support um

very natively now mongod DV um and build on top of that and that led us to building big genb but all the way along actually the goal was to have as scribe has always and the artists and the creators have always been our number one user actually and so we said okay as we're going to scale what are all the pieces that we need um and we realized that we needed not just scalable blockchain database software uh that is bigb but also a public network for it and so even when I talked last year you know it was forming but we were actually um uh having a lot of um discussions in conferences and um various cities all over the world with um with folks on on what this might look like from a

governance angle and so on and um that's what became ipdb interplanetary database incidentally there's a third piece too that I can talk about briefly at the end and that's um on the IP side of things um one problem with scale and that led us to Big ch iptb the software in the network the other problem was flexibility of Ip licensing um we had developed this small overlay protocol on top of Bitcoin called the spool protocol and um it was fine for digital art you know claiming copyright uh uh transferring single editions you know which is a bequeathing of certain rights but um artists kept coming to us and saying hey what about fractional ownership um what about taking a still of a video um how do you handle 3D what about about photography and um in various ways that led us to develop something called koala IP Co o l a IP um

and that was actually in sort of this uh driven team too not just us but you know Juan B with ipest Simon with Ojo consensus um Prim Vera with koala um Greg mcwell who was my colleague in big chb and iptb and others so um we developed this qual IP and it actually reconciles with all the existing protocols too of um DX for Music Plus for photography Etc so really overall we have these three um building blocks big genb the software ipb the public net and and qual IP and each has its own sort of governance Etc um you know big chainb is a for-profit corporation ipdb is actually a separate nonprofit that we spun out and it's controlled by the caretaker so I can talk more about that um and uh and then coip which is still um basically the sort of loosely formed

group um that P of the koala workshops run by Prima and constants U Prima Phil constant chah so um that's sort of what led to all of this um and um maybe I'll pause for a second and then I can dive into sort of what led to you know a bit more of the history of what's went on with bigb and ipdb since let's go right into to ipdb then uh so ipdb is the public implementation of the Big Chain DB protocol and I I think this I find this really interesting because until now you know we have we have what what real blockchain networks do we have out there we've got Bitcoin ethereum and okay like whatever how many alt coins there are uh but you the real platforms the the major

players Bitcoin ethereum they are public blockchains that were launched as public blockchains and and now ethereum you know we're starting to see sort of Enterprise um traction and there and you know perhaps some private Enterprise ethereum networks will start to emerge on on on the other hand you guys start out as this more or less uh private blockchain type system at least that's the way that I always saw Big Chain DB is Big Chain DB is a is is a protocol that will be deployed by Enterprise and you know as a as a private Network and now you're implementing a public implementation of

that is as far as I know I mean okay so Cosmic n the the cosmos network will be somewhat similar I guess you know from tendermint to Cosmos as far as I know you guys are the first to do this yeah uh what type of things are you are you learning from having you know deploy this network as or this having this technology first as a as a technology that was intended for private use I guess but now moving into having a public network so I actually uh quite honestly customer number one has always been a scribe which has always been about the public net um so uh but we always saw that um you know why not have it deployed so there's really three types of possible deployments there is the public deployment where anyone can read anyone can write from it um and

that is ipdb but others can you know have their own deployments that are fully public too then consum deployments where you might have you know 10 or 30 organizations running nodes together and um it might be you know Automotive self-driving car Consortium or something else and then the third one is actually within Enterprises um where you don't get as much benefits from decentralization but you actually still have the Merkel egg and a bit of a degree of immutability as well as the the signed transactions which are actually really really good for data trails and Providence so um so all three are are relevant use cases and uh and they have just you know different way ways of being used but there's really three deployments right public Consortium and within Enterprises so um but also by the way you know within

Enterprises they can also use public Nets you know we're seeing some of that where Enterprises are using public ethereum right and that's great um so the other thing to just sort of emphasize uh how we designed from day one to be complimentary to um some of the other Key Systems out there right um we see that um with Computing infrastructure you've got processing and you've got storage and um with in processing um it actually breaks down to two things business logic and high performance compute so business logic processing um the decentralized version is really where ethereum plays really well right um and this is where hyperledger and uh other sort of smart contracts play in right and hyperformance compute this is actually an emerging category where we have things like truebit and Golem exac Etc

right so those are overall this is all within the realm of processing uh we're also seeing people are actually running VMS um on top of bigchaindb or um you know JavaScript in the that's just talking directly to Big gen Etc so this is the processing on the storage you've got file systems database and actually what I see now is a store of value so um on the file system this is things like IP fast storage Etc with the database this is exactly big tdb and it's public ipdb and then in the stores of value this is the traditional ledgers right so Bitcoin The Ledger um zcash they're not trying to do much more than just store the value and actually most other blockchains have some sort of you know token storage too so overall you know this this stack that we see um where you've got um processing including

business logic and HPC and then you've got storage including file system database and store of value and we've always seen ourselves as aiming to be the best-in-class blockchain database decentralized database and um we've always been aiming for the public net as well as serving the enterprise we kind of straddle those um we've always sort of seen that you know Enterprises they have specific problems to solve we understand the Enterprise and it actually helps our business model right uh we are um selling uh we have open source software but we're like mongodb so they have an Enterprise version and we're moving towards that too right so um it's a way basically um so we can serve the Enterprise via you know the sort of Enterprise E level software but

we can serve the the startups who aren't willing or don't have the funds um to build their own consortiums Etc they can simply use ipdb and overall it's my dream you know 5 years 15 years from now where 99% of the applications are ipdb itself right and that will be fine because there's lots of other um business uh related things that um bigb the company can do too and I think this is just healthier for the planet so you know to be that best-in-class database software and then best-in class public database Network that's still plays well with everything else for to transferring value you know connecting via interledger connecting via Cosmos Etc so if if I understood correctly you you just said that in a few years you you would hope to see 99% of the

applications built on so this this stack uh using I uh using the public net okay so so you so in your view then the Consortium network uh you know of uh you know 10 or 15 companies within an industry collaborating on a on a on a single process or a regulated process that model you you don't see a future for that model or no I see aure Niche actually I see a future but what's going to happen is that they're going to to melt into the public net via interledger and Cosmos Etc right and I think that's really really healthy so just like you know rewind um 20 30 years and you would have all these different private Nets um you know via Lan or or as precursors and

then um they they started joining you know tcpip actually connected these things right and um you know arpet nsfnet all these like these really really early networks got connected with tcpip and um I see that ipdb will be a big public Global database from day one and it will sort of be like this back blown upon which other things can be linking but it's not going to be the centerpiece if you will it just it will be another sort of node in the network right and some of these networks will be much more sort of smart contract e like decentralized business logic processing and some of them will be much more database e like um like big chainb ipdb and that's okay right and just like right now you know you can have some software as a service that has a database like HTTP as the interface but

below it you've got some um database as a service but other behind other HTP interfaces you might have payments like Stripe Right so I see a similar thing emerging um where these protocols include value as well in a in a blockchain sense right and tokens Etc and um you know all these components will play well together so um for the consortiums like they'll still be there it'll just be much more public um and I think you know why would you bother setting up governance of a Consortium if you can all just use the public net right and even companies themselves you'll see now like a lot of them actually internally they don't even bother with uh Lan protocol they just follow the the public HTTP protocol and connect directly to the internet right I see okay I see where you're going so talk talk about

ipb then this is a again comparing to ethereum or Bitcoin on on the validation side it's a very different type of uh um validation scheme so there are predefined or sorry limit number of validators the validators are are chosen so it's it's sort of this um intersection between a Consortium Network and a public network where the network is given or made available as as a public good but validated by a Federated set of nodes so talk about the governance model there and how that works and and perhaps the foundation as well my pleasure there was a lot of

thinking that went into this so um so background wise um one of our key advisers uh is David Holzman he's the fellow who rolled out the the public DNS in the late 90s and he was maintaining it all throughout the 90s um and then they they they hatched Ian and you know that governed the the DNS for years and still does um For Better or For Worse right there was things that happened which are kind of sad with I can and you know David is actually pretty sad about that right and also the other people involved you know Jim rut eser Dyson Pender Wong uh if you talk to them and I have about this they they have you know there was a lot of the great things that happened but there's some very sad things that happen too you know things like the fact that we have dot sucks as

a domain name like and it's basically extortion you know Microsoft has to go and buy do sux from the TLD guys for do sux um for 10 grand or whatever so um as well as things like bank so overall um we we we had a series of questions and constraints that we wanted to solve for um and it came down to um how do we make sure that we aren't captured by money um and we aren't captured by a jurisdiction such that the system can keep running over the next 5 years 50 years 500 years in in a way where uh it serves the decentralized internet and so um we we said given this um first of all we have to say okay let's say we do have this

Federated type system where there's um you know 20 nodes or 50 nodes or growing 300 right then with each of these we want to make sure that the people running these nodes um care about the future of the decentralized internet and so that was a constraint that is a constraint to be a caretaker but then beyond that um the majority have to be nonprofits and that is actually solves the question of the the dollar capture right because now instead of having um dollars at stake the way that you know typical proof of stake systems work it's actually reputation at stake so you know and these caretakers are like internet archive open media Foundation Etc if they start attacking the system it kills the reputation that they've been building up over the decades sometimes right so um it's sort of like you know a proof of stake system where you're

staking something but you're actually staking your reputation it's not an economic incentive right and then the second concern was jurisdictional capture and an example of this is um Bitcoin right now you know the vast majority of mining is in China what if um China closes off its um its internet from the rest of the world for a while say a week right then you'll actually have a partition of the Bitcoin Network you'll have mining going on outside of China and inside and then let's say China opens it up again then it all heals and the Chinese um side will be the longer chain so all the transactions in the rest of the Bitcoin Network that previous week would have gone away right and what this means is actually that we actually essentially have jurisdiction capture right now um with Bitcoin um within one country and that's actually

sad right like I'm a fan of Bitcoin Bitcoin has inspired us all and um it's kind of too bad that right now um we have uh you know this this capture within a jurisdiction and there's other examples of jurisdiction we have to watch for like for example if we had the majority of caretakers in the USA what if some of the less than inspired leadership decided to crack down on IP laws um even in a crazier way right like sending letters to anyone that might have possibly infringed ever ever ever right and um so you you don't want to be dependent on one nation in terms of jurisdiction too badly so we said no given our rule is no given country can have more than 30% of the nodes right and and we actually try to spread it as far as we can in terms of the new

caretakers so you know we now have caretakers in you know USA and Canada um in all over Europe um in Kenya and that's expanding more right so and when I say we um uh actually we uh we no longer have control this so um how ipdb is set up um that the caretakers they have two key roles they are voting um so they have the power to vote each other in and out they have the power to vote and elect the board of directors the board of directors then um hires day-to-day management so they actually control um this nonprofit Foundation it's a German nonprofit so they vote but then they also run the nodes so they actually have this sort of day-to-day activity themselves of running the nodes so those are the the roles of the caretakers which are the heart and soul of the system and we have already handed

Over Control to the tiers so um they actually control the governance which is actually really really awesome right big chb doesn't uh you know we hatched it but we don't control it so uh and the majority are nonprofits we have a few for-profit companies too um besides uh such as big chb um protocol Labs the ipb guys and at some point stratum will be too I I hope as long as the caretakers with them in right I really hope they do so I I can't control it but um I can try to influence I suppose um anyway so that's that's the overall sort of process of this um in terms terms of um you know sort of the openness the publicness and so on um there's actually interesting these things are much closer than you might expect right so if you think about like a proof of stake system

like Casper right what Casper is is you've got these different um at any given point in time you have a set of identities that are allowed to validate right and they get um voted on through a relatively uh fancy um crypto economic mechanism right and then they're they're they're posting bonds so that if they um if they act badly then they lose that Bond right um and there's other systems that are sort of proof of stake like delegated proof of stake is actually quite similar um so this is the cosmo stuff and the bitshares for example where for example in bitshares um you have um one token one vote and then you can vote on whoever you want right and in the end though you're electing 100 validators right so in this case you're choosing the validators based on it's the richest actually get to choose

validators to me um you know it's kind of too bad that it's the richest that get to choose but I guess that's a bit of a surrogate for identity right um with with um tender uh it's actually the richest automatically become the validators right Whoever has um posted the most Bond actually um gets to become the validators the top 100 um all of this actually I see is a is sort of a placeholder for good identity um so you know with Bitcoin it's essentially one electron one vote um um with these uh proof of stake systems it's sort of like um one stake one vote if you will but I didn't ideally we would actually literally have one identity one vote so imagine if we had a really really good identity system overall for the world um that uh we could know that every

single human was their own identity signing and and um you know imagine there's a perfect system then each person could um be a validator now that's really nice but or or if you want you can go delegated proof of stake style where um one uh each person can vote for um whatever validator and then whichever 100 validators get chosen get chosen right and so um because delegated appr of stake is sort of a way to convert something that's bft or pseud of bft into something that's much more of a public net right so how ipdb is right now it's sort of this this Federation but it is um and it has these identities these sort of vetted identities out there right now 20 but um it's going to go from 20 to 40 to 60 Etc and then at

some point we will switch over to a system that is actually much more open where it's truly one human identity one vote so for example you don't have to have something perfect to start with maybe you just say okay if you have Estonia e residency you get to have a vote or you're vetted by one of these 10 different agencies you know maybe kyc from authentic the company or one of other 10 kyc providers right so this is a bit like the sort of um CA system that we have for um htps but that's actually not bad right it's it's better than sort of one electron one vote where then you have you know massive amounts of of value that way and um it will be emergent um no one has a perfect solution yet you can have multiple different types of markers you can have

the sort of you know government vetting you can have biometric markers you can have um stuff that is stored in your brain's memory like um brain wets Etc and then combine all these right and there's a lot of research on this this is you know um uh whole fields are dedicated to this identity authentication Etc but I think the ideal is really one identity one vote and and we're all trying to work towards that and what I've come to realize it's not that you know proof of stake is virtual mining it's actually everything is about um trying to get towards this ideal of one identity one identity one vote so ipdb has its own own path um you know ethereum has its path Bitcoin has sort of a pseudo path depending on it's governance right um so with ipdb you know it has very transparent governance

you know everything is recorded etc etc and they oversee the the choices of the protocol um like I said anyone can write to it anyone can read from it the only differ and any therefore anyone can be a client to it the only difference is to be one of the validating nodes right now you have to actually be approved by one of the you have to apply and then the caretakers vote you in right so um it's Democratic already and as time goes on um it will be more broadly democratic but I guar I guarantee it won't be based on how much money you have you'll never get to vote based on how much money you have all right well that if you want a a publicly open system as you describe it that is a public good I mean that that that's a good that's a good starting

point you know yeah not not to have uh interests of uh those who have the resources put forth before the the interest of those who don't have the resources so there are a lot of things that I'd like to come back to um first uh coming back to the foundation now you mentioned a few things that you learned from those who created the initial DNS and and and I can one of one of the things that was interesting that I find interesting about the ipdb foundation and its governance model is these rules that you've put in place right no 30 you can't have more than 30% of validators in a single country or more than 50% of

the validators have to be um nonprofits do you think that if measures such as these had been implemented as founding governance rules of organizations like I can we would have a very different system today how how do you how do you think we what would the internet look like today had those rules been implemented with Ian yeah um you know for all I complain about Ian I think actually you know it's is powering the internet right it's this database that's powering the internet in a very particular way it has its flaws but you know um we have these you know these the main tlds that everyone uses like.com

are you know have been around a long time um so these things would have helped a bit um would the internet look dramatically different probably not because actually all in all DNS is not that bad you know unless you act like ask some like deep crypto haacker it's all in all not too bad I we had a conversation with guys from the ethereum naming system recently and you know one of the things that I find and we discussed this one of the things I find particularly unfortunate about DNS is uh the the the system by which new top level domain names can be registered agreed not to have a whole debate about that but what what are the things in what are the things in in in in I can that you think you really think should be avoided in in ipdb and what can we

learn from the experience you know 20 30 years of Ian however long it's been that we wouldn't want to have in ipdb yeah there's a few lessons so one of them actually was um in terms of dollar capture what happened is that the regist RS captured IAM right and what this means is um uh the register hours you if they could actually start registering a whole bunch more tlds then they make much more money right so they made it actually a very low Bearer to entry for people to create a lot more tlds and now we've had this TLD explosion um and so companies like the big Enterprises that want to protect their BR corporate Brands Etc have to buy up all these other tlds and then also there's a sort of mad rush among everyone else too um would have there been a better way

almost certainly right um and so I think that's a challenge you know basically I ideally you don't have money interests making decisions that are for governance of sort of fundamental human infrastructure like like the internet or the DNS right um and so you know ens is exciting um but I think actually um I don't know how much they've talked to the DNS guys from what I gather not very much because there's a lot of other learnings too one one was um David Holzman he's sort of an unsung here over the Internet for doing this but he went and actually kept all the domain names of uh anything related to hate speech and hate words and you know um uh racial slurs that sort of thing he actually kept them all to himself and then as the internet was taking off he he actually

gave it to um advocacy group groups such as like blacks rights groups and so on and I think this was really wonderful he just didn't tell anyone about it but you know maybe I'll say right now like full shout out to you David for doing that like good job man so this is why um you know this is and there's a lot of sort of decisions that he made that were he was just trying to be thoughtful right um and you know maybe there's better ways to do it I think the fact he was making those decisions while he was still CTO at Network Solutions which is a public sorry a publicly traded company it was a for-profit though and um ideally these can be made much more in the open with um a much more you know broad betting process to try to include

all the learnings from the past right and this is actually what we are trying to do with ipdb and to give you a feel right with ipdb we've got these 20 caretakers but then the board of directors actually it does include David right and it includes um uh Greg McMullan uh you know that's the fellow who's running ipdb the main guy um and he actually before doing big tdb and ipdb he was actually uh doing stuff U helping with privacy you know he actually did a class action lawsuit against Facebook on behalf of Canadians for screwing with Canadians privacy right we also have constant chah who uh runs koala which is you know a blockchain legal initiative but she's xff right so all about you know um privacy as well and then Nina Lou Adler who's um a German lawyer and blockchain

expert so um so these people together they're actually thinking about a lot about these issues engaging with the government basically really trying to build Bridges to the law and what's going on rather than trying to run away and say you know this stuff is is different than law than the law so we're actually really um the ipb folks are really trying to work with the EU and the German governments Etc to try to actually set up the laws in a way that works and I think you know that's really nice learnings um going back you know like the the the learnings that David had um he he he instills these every day in us and he asks questions uh as we're as we're rolling this out and not just David right Jim rut who had been CEO of Network Solutions in through the 90s I've known Jim for almost 20 years and

we iterate a lot as well about this right so we're really fortunate to have you know people who were rolling out these initial internet infrastructure blocks um coming in and working with us and it's not just those guys right like things like you know last year there was the decentralized web Summit AT inter archive and you had Tim berners Le there and Vin Surf and and these guys continue to be involved and I think that's wonderful right um you know they care about the the future of the internet too and you know for the first time in probably more than 10 years like major things are happening right like all this blockchain movement it's not about blockchains it's actually about next level infrastructure for Humanity it's the layers above tcpip and next to it and above HTP and next to it and um and it's cool because we now have a business

model for it too right in tokens right rather than advertising that is so healthy on its own so also up there but overall like we we have to look keep looking to history and reading about it and talking to people and learning as much as we can and trying to avoid repeating those mistakes no definitely I agree that having people like David Holzman part of this new new era of decentralized infrastructure is great because we we can pull from from the lessons that they've learned and the things that they've done to try to avoid the same problems that we've you know perhaps seen in in in uh in other uh endeavors um coming back to the foundation one of the things that strikes me is kind of unusual I guess uh given the type of organization or the type of technology that we're your ipdb is is built on is

the fact that this is you know sort of looks like a you know typical foundation with uh you know U governance rules and uh and a board and I'm curious why didn't you or why didn't you think that it would be a good idea to run the governance you using a smart contract for instance just because a piece of technology is there a shiny new piece of technology doesn't mean it's the best technology right you know um if if there's a piece of Statistics from a paper from 1925 or earlier like by Fisher or GH or something and it works better than a fancy new paper from last year out of Sanford or something then I'll use the 1925 paper right so same thing here so there is governance technology that has been developed over

hundreds of years you know what they're called corporations nonprofits Etc and they actually they have evolved to solve a whole bunch of issues around governance right over hundreds and hundreds of years and so what we're doing is we're actually leveraging governance technologies that have actually been adapted and used and applied over generations and generations so um to me it's it's kind of a no-brainer like would I use some governance technology that can only cover 10% of all possible cases and ignores the rest just because it's living in a blockchain versus um some tried and tested technology that actually covers all the bases all the messy stuff right um and and actually has has ways to address that so to me

it's a no-brainer now that's not to say um that we we don't want to use um uh some of the you know smart contract stuff over time right but it's a walk before you run thing right um this stuff has to mature first so we saw that why take the risk in governance um on this you know whole new um set of Technologies yet when it's not there yet and lots of you know wonderfully smart people are working on this bit by bit by bit and uh that's great because we'll leverage it as it comes along but there's no need to add extra risk to this project this you know new fundamental internet infrastructure this database for the planet by introducing risky technology at this point so we're we're picking and choosing our battles and to give you an illustration um you know when I first announced ipdb um it

was at the um blue yard Summit U last June decentralized and encrypted right after me um so this was actually early June this was actually the height of the Dow the Dow had surpassed I think 150 million in funding but none of the hacks had happened yet I think Vlad was just starting to write about his worries and some others right and so right after me kristofh Jens was um speaking and uh so uh when I spoke people people had been asking me why aren't you you know are you guys G to make a dow of this and all that and I said you know I I said directly you know we love the idea of of doing a DA but not yet in time right and ipdb itself as an organization is designed to melt away 10 years 20 years from now whatever it takes once

everything gets automated enough because you know we don't think like we think that over time things can get automated enough but we have to actually you know walk before we run and so after my talk you know Christof came up and he talked actually all about the Dow and it was an interesting contrast right and um what's really really cool um you know that obviously the Dow imploded in all this and you know the ethereum community and the broader blockchain Community asked asked a lot of great questions and has continued to learn from that experience since right um you know we haven't uh gone you know live running $150 million yet on with ipdb but it will come too and this technology will get tested um but I think it's really cool that the conversation flows um you know we we have a meet up actually with Leb and

ipdb it's every month and a couple months ago we actually had Greg come on to talk about ipdb governance and Kristoff talking about the learnings from the da and then there was a panel afterwards with those two and Sherman bushmaker who is also a lawyer and she was one of the curators of the da Etc and it was actually remarkable how similar the thinking was um these worlds are not so different as you think every everyone is trying to solve the same problems you know how do you govern the protocol what do you do when um things go wrong Etc and uh we have various Technologies to deploy not just the smart contracts Technologies of the last several years but also these um old school governance technologies that go back decades let's use them uh interesting okay so uh now I'd like to talk about the the the network itself

and so at this time so the network is sort of running as a as a test net I believe um yeah yeah so the overall plan yeah so the overall plan is uh there's a test net and a production net will come and with the test net um yeah we've actually had the testnet running uh quietly with some lead users since last October but very very quietly just to sort of vet it out and um more recently we've we've been developing actually a developer portal for it um such that um it's really easy to basically sign up and start using it and talking to it via an HTTP API and it has a whole bunch of things like DDOS protection and all this you know it's running in Genex kubernetes Docker all this sort of thing so that it it feels like a full-on um

software as a service that's fully professional that you might expect it's so we take a lot more cues from sort of the mainstream software world than from the blockchain world and how we do things right like you know the sort of shift to containers in the last two years is is phenomenally useful and we are embracing it wholeheartedly right um as well as you know other Technologies so under the hood basically uh yeah we've got these pieces in terms of getting rolled out um so we've got this developer portal that um now we're we're we've been opening up to more and more folks we've actually had a wait list of um 250 plus organizations these have been tremendously patient folks so those of you on the waight list thank you for your patience we're starting to now um open it up to everyone out there so my

hope is within the next month all of those folks and Beyond will be using the test net um and uh and easily you know we really held back we couldn't scale before because we really needed we really needed the developer portal to be there to be easy to use so right now it's a test net the the caretakers are are running the network uh altruistically I guess and also to sort of test it out and make sure that this was something that can move into production when the system goes to production when it goes live and when all those organizations start using it as a production system to build their decentralized Consortium blockchain applications you know that's fancy stuff um what how much do it cost to use a network does it cost anything what's the economic model there and um there will there be a token are you guys going to

do a crowd sale or yeah so overall I'll talk about the the now for starters so um uh we actually once again you know there are no native tokens inside ipb we actually um went the traditional web services way which was asking yourselves how much will it cost to store data for forever and of course forever um is infinity and then the price is potentially Infinity so then we said okay what about 50 years okay so we said okay um how how much would it cost to store data for 50 years and there's baselines right so if you put um a gigabyte of data into awso Zer it's $3 per gigabyte per month okay and um so with that as a baseline then we said

okay well what about mors law you know storage is getting Cheaper by cheaper cheaper every year um we we found some very conservative numbers where it's getting 17% cheaper per year and then extrapolated this and we assume that a plate is in 10 years so um basically it's getting a bit cheaper every year plate in 10 years but also um we said okay if we're going to store this thing for 50 years knowing that it's getting cheaper though then um how much is it going to cost um overall how much does someone have to pay up front and as we ran the numbers um with various models and so on we also took into account inflation of and the fact that you can invest the money so they put X dollars in that gets invested with uh so we assumed actually 4% rate of return and that's actually typical value

for Government Bond it's hasn't much CH hasn't changed much over the decades 3% inflation which means overall 1% rate R of return what this came down to for a number actually is the very aggressive estimates are about €25 um for storing for forever um more conservative was about 75 so we said we're just going to start off with 100 so basically when people are using the production net it's $100 per gigabyte forever and when we say forever um we Define it as 50 years but it's interesting once you put that money in it gets invested and it hits an escape velocity 20 25 years in such that's that actually does pay for things forever which is really cool and it's very different right because if you put

something into say like Bitcoin or ethereum um you're kind of hoping that the network is going to stick around right it's different here it's actually a contract that you're um making where you're paying money up front but it's actually getting stored by this contract um for forever you're not just hoping that someone will keep maintaining the nodes and that's really really cool and there's other like long-term contracts out there we looked around right there's things like this in um in trust funds like the Rockefeller trust fund Etc graveyards you know actually I can buy a plot for myself when I die you know 50 or 100 years from now whatever and that's actually kind of interesting too and then I can actually buy plots for my family for the next 500 years so you can actually have these long-term contracts that exist and we've done that

essentially here so overall uh We've ran the numbers uh like I said it's $100 per gigabyte um for forever that's very conservative we actually hope that um within a year or two of running this for the production net that we can pull that down by by 10x or more to whom do you pay this money how do you pay the money is it do do you pay the foundation or uh yeah so right now um to start with it's actually going to go through the foundation and the foundation reimburse the nodes but very very quickly um this is going to get basically um more decentralized in the sense of they can pay one of um uh many they can pay one of the caretakers directly get tokens for usage and then it tokens it's just like an adbs token it's not a fungible token that you can have an exchange Etc

and then by that um it doesn't have to flow through the foundation at all right and that's just healthier so um overall uh that's kind of where we're headed um interestingly it's super efficient um if you think about how much it costs to store um data on bitcoin right so one transaction four years ago when we working on a scribe three and a half it was 10 cents per transaction um a year ago it was $1 per transaction on bitcoin because of the mining fees right like the miners automatically set the fees actually the Bitcoin network has been kind of going crazy this past week so now it's it's been jumping up to like $3 and $9 it's kind of crazy but I assume that will go down but overall even if you say it's $1 transaction then uh we run the numbers and for us it is um if

one of our transactions is 1 kiloby kilobyte which is actually Conservative then you means you can have one million transactions for that $100 right so that means we're um many many many orders of magnitud Cheaper Than Bitcoin right is it 10,000 something like that right so um uh that's that's quite exciting so anyone who actually wants to store a stort of data or or even issue assets and tokens here you have the system that's just using just straight up simple engine engineering like you know how would an engineer um go about this um this is the numbers we came up with right it's it's it sounds kind of insane that you can be like you know so much cheaper but not really if you kind of just go through the logic this thought process that I just described right and um with ethereum you know it's a little

bit cheaper than Bitcoin but we're still orders of mud different but this is actually why um ipdb is designed as it is it's actually a database that's meant to store data you know not the hashes the data itself that's what a database does and it's meant to be complimentary to ethereum and Bitcoin right and there are people building systems that have um ethereum and ipdb and actually ipfs too right all three for the files for the for the structured data and for the business logic right okay so we're we're really running Along on this episode I recognize but I do want to talk about one last thing and that is use cases for ipdb um I I I have some use cases in mind uh you know obviously you know talked to a lot of organizations and a lot of Enterprise about how blockchain can be used

uh to facilitate data transfer basically I mean I start I'm starting to see blockchains really as a messaging systems between in the Enterprise context as a govern a governed messaging system between Enterprise uh and the the potential for that is massive so tell us some of the things that um people are building on it on ipdb uh today or things that perhaps you see as use cases in the future sure so I'll I'll push this from two angles one from sort of verticals and one from um sort of General corpor techologies so verticals um maybe even I talked this in the last episode I've seen sort of six major verticals that big chb itself is really most useful for right um identity which includes authentication as well as uh storing personal data towards the

vision of sovereign personal data Financial intellectual property energy supply chain and government so of those uh Financial mostly you know there a desire for banks to be um private and government you know it's sort of a bit private a bit public it's a bit strange but the other one are actually really great fits for for public um for for their own various reasons so um for identity you know anything related to personal data you don't want that to be held by some sort of Consortium you actually prefer that it's really much more public right and this is like you know towards the vision of sovereign personal data all of this it fits really well so um you know there's companies like authentic that are building on top of us that are right now starting off with sort of kyc is stuff but are moving towards Sovereign personal data other

people doing consent based stuff like consent tokens to solve gdpr Etc so that's that's a really big use case all the sort of stuff in personal data and we are engaged very deeply in that Community around identity to um and uh I believe there's announcements coming in this too in the next few days of consensus so uh the second one that is a really big use case is intellectual property right and so this is where we started once again with a scribe and uh so if you think about when I create a piece of digital art my claim to that piece of digital art should really be SE visible for the the whole world the public right and it's not just digital art it's all the music right um who composed the pieces who performed the pieces right now this is actually all locked up by the big labels and

otherwise and it's actually really sad so the music music ecosystem is actually part like largely broken in many really sad ways and what it means is that um musicians uh it might be three years before they're paid now because there's just so much data out there and the collecting socities um where it's their mandate to actually figure out who played what when um it takes them three years to actually pay the artists so um and I can keep going right all the other just sort of sub fields of Ip everything from software licensing uh to to um novels so and we have people doing lots of this stuff on us going to the next field um supply chain is a big one and you might think this is very enterpris but the challenge is actually how do you the supply chains are massive right they

span the globe and there's many many subnets within so it's actually really hard to um set up a supply chain Consortium that actually really covers all the bases so um it's very straightforward actually than simply just to use uh ipdb right because then you're just putting the data on there it's fine or maybe uh you start with ipdb and if you realize you really want to actually have your own thing you can do it afterwards but it bypasses having to do governance where you're actually figureing out how to govern who's running the service right so um I think that's a big one um and then kind of overall too um with government there's sort of stuff that helps governments and then there's stuff that does government like um actions so stuff that helps governments is things like more transparency Etc right so um it's really

ideal if that stuff is put onto a public net um and we I think that we will see things like this happening over time my favorite example here is actually Estonia it's not running on a transparent blockchain per se right now but it's been electronic since the wallf in the early 90s and um these days um it would be natural if a government wanted to be more transparent to just sort of make that lead directly to uh blockchain to something like ipdb um it's just much more straightforward but then government like Registries are actually really powerful and my favorite example here is Ben Ben uh they are a land registry in Ghana uh basically that means if if I'm someone in Ghana who doesn't have money maybe I've been living in the same house for you know my generation my my family for two gener 50 Generations or five

generations we never had title to it before even though and that's simply because the governments have gone through too much instability right so no one really trusts the government to maintain the land registry but if you actually have this land registry that is um independent of the government that you know no one can kind of pull down then you can start to take mortgages on this then you can start you know use these mortgages as loans to get educations to certain companies Etc so I'm very very excited about Ben Ben um because U they're doing exactly this and they're doing it for Ghana and elsewhere they're engaged with the UN and others to really you know bring these sorts of services to to help the poor so overall identity intellectual property supply chain and government like things energy it's it's kind of to be determined um

you know there's been a lot of experiments in energy um there's some cool things like we're working with um energy you know Germany's largest energy provider on several blockchain projects but many of them are actually sort of more supply chain related and so on so um so for energy itself I I think we'll see but I think it makes sense that it's public too uh overall to give you some numbers I'll just um I mentioned yeah like a waiting list of hundreds on ipb with big chb itself there's about 30 organizations that we know of building on us about half or 23s are startups the other third is you know large Enterprises consortiums Etc and for everyone that we know of you know there tends to be another you know five or 10 as far as we can tell that aren't and we hear about another new one every day or

two these days it's pretty exciting um you know heard about about a couple more today because it consensus right so um people are starting to realize that um you know Big Chain DB and ipdb are a complimentary piece of the stack right it's not just you know blockchain as a noun it's actually you've got you know decentralized processing decentralized file system decentralized database right and then with this um you know plug in the pieces right and um they discover that big TB is really easy to use you've got this now we have this public net um people can get going running against it in 20 minutes right um you know all you have to do is just start making HTP requests right and and then you're good to go so I'm interested in a lot of these Enterprise use cases uh for obviously at stram we're confronted with

a lot of these same use cases and in fact we a lot of the the industries that you mentioned are specifically the the industries we we Target and uh one of the things that I think is is really important and that sort of Consortium closed networks provide is the well a couple things so one is the ability to have uh sort of privacy built in this idea that only the Consortium has access to the data and you can have very malleable permission rules where I send data to one other participant in the system that you know I I encrypt the data so that they can read it or perhaps you know two or three participants and

that eliminates the possibility for sort of business intelligence to be done in the network and then you can revoke accents to the data this sort of thing uh and then you can layer things on top of that like you can do then you can do sort of zero knowledge proofs right built on a blockchain network where only two participants are seeing the data but the others can still validate the data to be part of some uh validation rules I think that these are really really core to any blockchain implementation at a large scale for for Enterprise um especially in sort of regularly processes so like for instance in insurance we have regulated processes uh where you're going to need these kind of features and you need the regulator to be involved even in in setting up the system and the governance and

everything how do you think that do do you think that given these constraints and that are oftentimes regulatory at the high level do you think that public networks can really compete with Consortium networks given those constraints and how how tightly they are set by the regulator yes and here's why um the technology um of today is radically different um in capabilities than it was three or four years ago so you know we all kind of are calibrated by Bitcoin where everything sort of out there in public and so on right but just because a network is public doesn't mean that you can have shades of privacy inside right so zcash for example right you've got um privacy in terms of the value that's sent right but you don't have to just have that I I

view privacy you have privacy at three levels you have privacy of identity uh I.E anonity or suit anity you have privacy of the value transferred um like zcash is doing with your knowledge proofs and you actually have privacy in the payload itself right and you know the payload itself this is something very specific to Big chb it's a database right so you know it acting as a database it's about the payload of course it supports tokens Etc too but on the payload side this is really a sweet spot right so um right now you know if you're using ipdb and I want to give data to you if I wanted um I could say hey Sebastian what's your public key and then I would just encrypt with your public key I put that on the network and then you could decrypt it with your PR

private key right so that would work of course if I'm just sending to you then there's a better way to send data than um via a blockchain where all that encrypted stuff is written there forever we should use some sort of more lightweight messaging system right but um that's possible but another um more relevant example is say I have personal data that I want to share maybe some medical records that I want to share with um you know a thousand different scientific teams around the world to do to include in their scientific research right the Breakthrough here is that we can actually now have read permissions as assets think of it like token tokenized permissions and this is something we've been working on actually with some lead clients over the last few months and uh we will be releasing it actually with our next release it's 1.0

it's in mid June and we view it as sort of this beautiful marriage between the ideas of blockchain and the ideas of of database read permissions so read permissions as assets um I'll give you a feel so imagine um now I give this example before where I um had encrypted your data um with uh or encrypted my data with your public keyy but instead of that what if I encrypt um my data um a symmetric key and then you want to read my data maybe because you're a medical researcher or maybe it's something related to metadata music or something right where there's going to be lots of thousands of shares um so then basically I simply uh I've got this this data that's sitting out there it's

it's encrypted with a symmetric key and then if if you want it I just simply encrypt the symmetric key with your public key so I'm basically giving you um securely giving you the symmetric key to this right and I'm doing it in a way where that can even be re revoked in in five minutes right because you can actually have arbitrary crypto conditions around this uh we're using the interledger crypto conditions protocol around this so you can sort of have arbitrary com um combination logic um that has you know multisig like one of two which is like an or or two of two which is like an and you can have inverters on the data you know signatures you know the inputs are signatures or facts or time right and so you sort of arbitrarily composed with these multisig um and or Gates Etc and

inverters and then in the end you can do whatever you want with this data and it can be things like okay I'm giving you permission but you don't have permission to pass it on to anyone else right um etc etc so overall this this concept is wildly powerful and it actually um resolves a bunch of the questions I think consortiums are asking for privacy simply because um there hasn't been an awareness that this technology is even possible now that it is um it just makes more sense because then you don't actually have to deal with the politics about who runs what nodes Etc you can just use the public net right and um it's just so much simpler now there's still going to be cases where you know people will want to have you know data in jurisdictions you know the laws are still a bit fuzzy there Etc so you're

going to need that right but um if if if you if that once that gets resolved the thing that makes sense overall is actually all of the stuff is you know encrypted um on a public net um and by the way too you might ask uh we asked what about Quantum right and here's the cool thing because ipdb is well-governed then we can literally migrate the data from the existing um algorithms to a more Quantum resistant algorithm in the future when that stuff starts to get more real so that is even not a concern because because ipdb has good governance fascinating stuff uh I wish we could go on but we're nearing I think probably the longest show we've ever done uh we're pushing on uh on you know over 90 minutes here so we're going to have to

wrap it up I mean there's so many other things that I would have loved to cover we we'll have to have you back on at some point perhaps even uh with some of the board members from uh uh from ipdb or custodians uh you know we could definitely do that again uh so yeah thank you so much for coming on Trent uh as as a a repeat guest uh and you're by the way you're you're you're getting up there in the in the guests that have been on the show most often I think you're at there your fourth probably your fourth time now one thing you mentioned board members and I I actually I I hate to miss people out and I I think I forgot to mention Casper CES uh Casper runs Estonia e residency and he's a board member as well we're really

honored to have him so um I just wanted to mention because I always like to give attribution when when it's deserved and du so um but thank you for having me I always enjoy being on here and once again thanks for being flexible with with me and Fred having a crazy idea to to do this together so no it it turned out uh turned out reasonably well I hope everything got I hope all the recordings are fine and there's no screwups on the technical side but it should be fine aome uh yeah so thanks again and enjoy your the rest of your week uh in New York at consensus and uh we'll looking forward to speaking soonk you and to our listeners thanks again for tuning in we are part of the let's talk Bitcoin Network you can find this show and lots of other great shows at letx talk

bitcoin.com of course if you like the show there's lots of ways you can support us one of those Ways by uh leaving us an iTunes review and you can also send us a tip the Tipping address will be in the show description and of course subscribe through uh iTunes YouTube SoundCloud you know wherever you get your podcast and tell your friends to subscribe too I mean if you like the show let your friends know so thanks so much and we look forward to being that back next [Music] week