Dr. Leemon Baird x Harvard Talk - Hashgraph: New Directions for Blockchains & Distributed Ledgers

oh well thank you all for coming I appreciate you being here my name is Lehmann Baird and I'm gonna talk to you briefly about hash graph this is a consensus algorithm we have a distributed ledger system swirled x' is our company that's doing it with permissioned Ledger's at the moment we don't have a public ledger we don't have a cryptocurrency so if you ask me tell me about your cryptocurrency I don't have one I always get that question right now what we're doing is the permission ledger but they would both use the same consensus algorithm

and so I'm going to talk about that today and I have been told that we're gonna have a mixture of people who have never heard of blockchain and then people that are deeply into this and our experts of it and so that's great I'm gonna start at the very beginning and then I'm only gonna talk for a half hour and then we can have a half hour of questions and so if you want to get into the very deep details of the math I am happy to go into that but if you want to just

talk about various business applications I'm happy with that and if you want to just say again now what is this blockchain thing that's great too so we'll just go through this for a half hour so what is hash graph hash graph is a distributed ledger technology it's a consensus algorithm that you can use for DLT DLTS are systems where a group of computers are going to be sharing some information and we don't trust each other you don't particularly trust any given computer in the group to not corrupt the data or to not change the data

or lie about it or try to stop us from coming to consensus but we do trust the group as a whole to not have too many bad players in the group if you can't trust anybody then you're out of luck but if you have a system where we have a group of people a group of computers and we can trust that we won't have a large group of them ganging up on us to corrupt the system but we don't trust any single one person then what a DLT does is it allows us to come to

an agreement a consensus on the order of our transactions so we're going to be all be creating these transactions that change our shared data in some way they do something and we're going to come to a consensus on what the order of that transaction is of what all the transactions are what order they are we're also going to come to a consensus on a timestamp for each transaction so for every transaction we're all going to agree here's the time at which it officially occurred and in hash graph it's actually going to be a good time

stamp that we all contributed to in some systems one person can arbitrarily make up the time stamp but everybody else has to sort of approve it in some way but for hash graph it'll be a good time stamp so this is what a DL T is in general lots of people talk about blockchain sometimes when someone says blockchain they mean DL T's in general so in that sense where a blockchain other times when someone says blockchain they're talking about certain kind of DLT DL T's that are built out of a chain of blocks on top

of each other connected to each other we don't have a chain of blocks so in that sense we're not a block chain but we're a deal T so we are doing the same thing that the other deities are doing and if you're one of the people that uses blockchain to mean all those the whole DLT world then yes we're a blockchain as well but I'd like to make the distinction we're not built on a chain of blocks are built on something else so this is what hash Grethe is it's the consensus algorithm it's the way

that the computer is going to talk to each other and come to an agreement on the transactions and on the other things and hash graph is fast it's secure in its fair these are the three main things that we want in a ledger and what you'll find is that it's faster and more secure than these other Ledger's and it's fair in a way that the other Ledger's are not so there's some unusual things about it and the reason it's unusual is because it achieves this consensus in a totally different way it's just very weird not

at all like what you would see in all of these existing systems with Bitcoin and aetherium and paxos and raft and pbft and Casper and iota you may have heard of all these different systems they all work in different ways hash graph works in a very different way in fact hash graph goes back to these thirty-year-old voting algorithms that we've had for decades and extends it so we'll talk about that so let me just say what these things are what do we mean by fast secure and fair fast in what sense well secure in what

sense fair in what sense so here's what we mean for any ledger these are the things you might ask is how fast is it for fast you want to know how many transactions can we do every second so if it's say a cryptocurrency I want to know how many times per second can I do a payment if you had a credit card system it might be 10,000 payments per second 10,000 transactions per second if it's Bitcoin you're more like three transactions per second different systems have different speeds and then the other thing is latency when

I do my payment how long is it before I know that the payment went through and I know that everybody else knows that it went through and I'm guaranteed mathematically that it went through so two things throughput and latency for hash graph the throughput is hundreds of thousands of transactions a second we've done experiments that go across the country experiments to go around the world and you can see very high speeds and we also have very low latencies a few seconds or even a fraction of a second from when you create a transaction you try

to pay someone some money until you have a cryptographic proof that the entire community has agreed on the ordering of your transaction and that it really will go through and it's a cryptographic proof that you could actually take to a court to prove it this is also something it's a little bit different from some Ledger's it isn't one of these things where every time I get a confirmation I've become a little bit more sure no you reach a point in time when you know for sure so if somebody pays you at your store you can

hand them the product you can say yep I am sure it went through I know I'm gonna get to keep the money and so then you hand them the product so this can be fast there's lots of caveats on speed by the way if it if your transactions take a long time to process you know it takes you a year for your supercomputer to process a transaction then your networks only going to do one transaction per year this is just the consensus part of it is incredibly fast you basically turn it into a network for

free so that's what I mean when I say that this hash graph thing is fast it allows us to do these Ledger's that are fast which opens up new kinds of things you could do like games that you couldn't do before in addition to that it is secured in the same ways that the other ones are secured but also in some new ways that aren't really being talked about a lot so it is secure the math term is asynchronous Byzantine it's Byzantine fault-tolerant in an asynchronous sense this is the strongest form of security I told

you that this goes back 35 years for 35 years the gold standard in security has been asynchronous Byzantine then some systems aren't quite as good as that they're partially asynchronous or they're synchronous Byzantine and some are even worse than that they aren't they aren't Byzantine at all they aren't anything they're just what we're pretty sure they're safe but if you have math proofs that are really strong you can say it's asynchronous Byzantine what does that mean it means that we as a community when you have two transactions we as a community are going to come

to an agreement on what order they're in and we will know that we've come to agreement it isn't just that you think they're in this order then you think it's this and you think that it's this and that eventually it stops swapping it isn't just that it's that you know when you've reached agreement so if somebody pays you some money in your store and you're afraid that maybe your coin is going to disappear because they also paid somebody else that same coin you're afraid with a double spend you have to wait until you're really sure

that your spend came earlier in history in some systems you just become more and more sure over time but you never really know in Hache graph there's a moment and sorry in any Byzantine system there has to be a moment when you know for sure that you have the right consensus and you have to have a mathematical proof that you will never be wrong given the assumptions that not too many of us are bad by the way the assumption is usually that no one third of us are evil in colluding to corrupt the system there's

actually a math theorem that says you can't do any better than that no system could be resilient to more than that but the best you could do is get it up to a third and Hache graph does that yeah it gets up to a third so this is the best you could do in a Byzantine system in asynchronous Byzantine system also I said we don't make yeah I said you have to make assumptions like we won't have a third of us being bad we also assume that the Internet can do bad things or the bad

people can do things to us over the Internet for example what if the bad guys are able to set up a fire around a big chunk of all of our computers and shut it down basically partition the internet into two halves or what if they can do something even worse and had some of their computers talking through that firewall but other computers can't or it slows down your messages by arbitrary amounts or delete some messages and not others we're resilient to that if you're asynchronous missing teen you're resilient to that if somebody says they're Byzantine

but they can't say asynchronous what that means is maybe their system isn't going to be able to survive a malicious firewall and do we have firewalls in this world sure all over the place some whole countries have firewalls around them so we want to have a system that's secure in a world that has firewalls because we happen to live in a world that has spire walls and there's an even bigger problem what about botnets what happens if bad guys out there are able to hack into little computers on the Internet computers you don't even think

of as being in computers like what if they could hack into your printer and take over your printer well you'd say I don't care so they can print something so what now here's the problem if you take over enough printers you can have them all send messages to one person over the Internet and flood them with so many that their computer can no longer receive or send messages from anybody else you can shut them down with a distributed denial of service attack that's a big deal and that has happened this is a big deal in

the media you know the DynaTAC class fall shut down people like Netflix for hours does Netflix mind if their servers go down for several hours I'm guessing they probably do mind yeah that's probably not a good thing when you're in entirely depending on the internet for your business you probably don't want to be shut down by DDoS attack so for these Ledger's that we're building we want to make sure they're not going to be shut down by a DDoS attack so what you want to do is proves but it's Byzantine which means it'll keep running

and you'll keep getting consensus and everything will be good but you want to assume that we live in a world where sometimes DDoS attacks happen and where sometimes there are botnets and again that's what the asynchronous means if someone proves Byzantine on some system but it isn't asynchronous then that means maybe it could be shut down by a DDoS attack for example leader based systems PACs often draft in pbft and some of these leader based systems even some of the like distributed proof of stake systems where we take turns being leaders the problem is even

if you're taking turns being leader for two seconds if the attacker can shut down one computer at a time then they can shut down your entire network cuz they'll just shut down your leader and when we move to a new computer being the leader they'll move to attacking the new leader and they'll just keep following the leader and shut us down as long as they want by just shutting down one computer at a time so that was a lot to cover and they actually had a lot of math words in it and yeah it's inherently

mathematical but the important thing the important thing is you want to be secure all of these Ledger's care about security if you didn't care about security you just use a single server why even bother with a ledger and different Ledger's have different kinds of proofs of their security and the strongest you could do is asynchronous Byzantine and you want both of those words both the asynchronous and the Byzantine and so the hash graph has that and most of the algorithms aren't asynchronous Byzantine that's a very rare thing except of course for all those algorithms from

three decades ago they're asynchronous Byzantine but they're so incredibly slow you'd never use them however hash graph is not slow but it's still asynchronous Byzantine and it also has a new property that Ledger's usually don't talk about at all and that is fairness in fact we have three different kinds of fairness here so we want to have our ledger putting the transactions in order and we want to have good time stamps on them honestly if you're doing a cryptocurrency you don't really care about this very much maybe fairness of access you don't care about the

other kinds of fairness we can imagine if we use these Ledger's for other applications like a stock market so imagine this we're going to have a stock market but we're not going to have the New York Stock Exchange set up a computer that we all have to trust New York Stock Exchange incorporated to be honest in their computer instead we're going to let each of the brokers have a computer and those computers then elves are the stockmarket there is no one else there is no stock market corporation involved it's just the brokers themselves being the

stock market and they don't trust each other definitely but there's enough of them let's imagine that they can assume that no one third of them are going to gang up on the rest in a stock market like that every time you put in a bid or an ask every time you said I want to buy a share of of IBM at this price or I want to sell a share of IBM at this price your bids and asks are going out other people's bids and asks are going out you want the stock market to match

them up and you could say something like well when I offer to buy something the first person who offers to sell at that price matches up with me or when I offer to sell something and the first person who offers to buy at that price will match up with me if we do a stock market like that then we start to care a lot about exactly what order the transactions are in if you put in a bid and I put in a bid and mine actually I put it in later than yours but I could

bribe somebody and get mine to count as before yours that would be bad people spend millions and millions of dollars just to shave milliseconds off of how long their bids go into the stock market this is an incredibly important thing to stock brokers they would never use a market where it was possible for one individual to arbitrarily reorder some of these these the ordering of or influence the bids and asks into a bad ordering you know the way they want to be in hash graph what we do is we have fair ordering when you send

out your transaction it's going to go to the community as fast as possible and you're going to get credit for it when it reached the most of the community and when I send out my transaction I'm going to get credited for it when it reaches most of the community and whoever reached most of the community first is going to count as being first and there's no one person that I could slip some money to that would get you mind put before yours if you reach the world before I reach the world then you get credit

for it guaranteed so we have fairness of ordering we also have fairness of access if you want to go put your bid into the system it's really hard for anybody to stop you you can put your bid into the system very rapidly and it's it's resilient ways that it's hard for anyone to stop you from doing so and you can get it in very fast as well furthermore in a fraction of a second you need it in the network even if some of the nodes are bad and they're just being mean and not accepting your

transaction you can still get it in so with fairness of access we have fairness of ordering and we have fairness of timestamps when we get a time step put on your transaction it will be something the entire community helped contribute to so for example in legal contracts there might be a legal contract that says you have to send out some information before noon on Tuesday well when you send out the information it's a time stamp and then you can trust that it as the community as a whole has helped contribute to that time stamp and

so you can go to a court and say yes I really did do it by noon on Tuesday and it wasn't just the opinion of one miner it was the whole community contributing to this time stamp so these are the things that we would want we want something fast secure and fair we can do all three of those in hash graph and what it does is allows us to start doing new kinds of things we haven't done in the past new applications so let's forget about hash graph for a second let's talk about Ledger's in

general so distributor Ledger's have been around for a while you could say they went to go back 30 years although really that's more of the distribute computing and the fault-tolerant community you might say well really it started with Bitcoin where's the first blockchain or the first distributed ledger that we're gonna call a DLT so let's just start say with Bitcoin is our history here and talk about the generations that we've had so what happened you could say Bitcoin came along and said you know what we could do money without any government we could do something

better than the fiat government the fiat money's that are out there revolutionary concept will do it making people do proof of work so you buy a supercomputer and you feed lots of electricity to it and then you two can help be part of the system and make it fair and it's so hard to do that that we can trust that no one person can dominate it and so therefore we're all going to be fair and trustworthy and we will ensure that the ordering of our transactions are agreed on by all of us it may not

be faster fare but we'll agree on them and then nobody can forge money no single government no single person would be able to forge money in our system counterfeit money kind of revolutionary and there's a lot of applications for that and even at three transactions per second you can see that there's an enormous market with Bitcoin I mean it really is a big deal and that is the first of the four generations I'm going to talk about but nothing stands still it's continued to improve and even today people are continuing to improve this so what

does it let us do it lets us to things like these examples here it lets just use cryptocurrencies like you would use money you can just save some of your money in your cryptocurrency wallet you can use it as a store of value you can use it to transfer money to other people just like you would PayPal the money you can send them your cryptocurrency you can use it as utility tokens to pay for services so there can be services online that you pay for with these tokens you can use cryptocurrencies that way economist endlessly

debate well what is cryptocurrency is it money is it equity is it stock is it a utility token and the answer is I don't know it's sort of all those much together and they say it's none of the above it's is this new thing called a cryptocurrency it's beautiful and then you can do other things we are starting to see people add on other features now like strong anonymity where you don't know who has the given wallet or strong transparency so the banks can do know your customer things or micro transactions if I have to

pay a couple of dollars for each of my transactions there's uses for it but if I could pay a thousandth of a cent there's more uses for it you can start talking about maybe paying a thousandth of a cent for reading papers behind a paywall rather than having to put in your credit card and buy a thirty dollar paper you could talk about all sorts of interesting things where you do micro payments which would be enabled if you do a thousandth of a cent per transaction or smaller so these are some of the things that

we can do with cryptocurrencies they were the first generation they continue to evolve even now but people said you know what this Bitcoin thing was really cool it's doing a cryptocurrency in order to do the cryptocurrency it has to remember all these transactions it has this ledger that remembers them all it's this blockchain that remembers the but wait a second it's a computer it's just ones and zeros we don't have to limit ourselves to just remembering things about coins we could put other stuff in that ledger and it would be just as good you'd be

just as trustworthy and so we've got the second generation of DL T's which was well let's not just have a money let's have a filesystem in some sense kind of expensive so little tiny files but the important thing is that nobody can change it when they shouldn't and if I think there's a piece of information there and you look at it you are guaranteed to see the same thing we have a shared view of what's going on everybody knows that everybody knows about it the rules are enforced by the entire community no one person could

delete things no one person could add things by themselves in an in an illegal way this is a powerful concept and so the second generation of DL T's was saying we can go beyond money we can go beyond cryptocurrencies we can store information in these things you could store ownership of something like land why is it important to put a deed or at least a hash of a deed into a ledger like this because if I try to sell my property to two different people I could hand them each a Xerox of the deed but

you when you try to buy it from one of them you want to make sure that no one else has it you want to do a check on who owns it but if you have a ledger you just look in the ledger and you're guaranteed to see the same thing everyone else sees so you're guaranteed that the whole world is seeing who owns this land right now we're all seeing the same picture that's a powerful concept very powerful and you could do other facets not just land but any kind of asset you could have digital

assets in this ledger and you had this guaranteed shared view if you can do that you can do revocation for example maybe the DMV are issues me a driver's license I could put my driver's license in the ledger and maybe that's not good for privacy I could put a hash of my private driver's license in the ledger that's a scrambled version of it that you can't understand what good would that do ah you would do a lot of good if it allows the DMV to revoke it because if I go to you and I want

to convince you that I have a driver's license it's not good enough for me to just show you some files signed by the DMV that says I had a driver's license because maybe I got into some accidents and they pulled my driver's license maybe they were revoked it cancelled it so we want to be able to do revocation we want the DMV to be able to revoke my driver's licenses what you could do is have a ledger where the DMV has the rights to delete your driver's license but nobody else does or maybe the DMV

has the right and you have the right but no one else has the right to delete your driver's license then when you want to prove to someone you have the driver's license you give them the file they look at the hash and they look in the ledger and they see if it's still there and if it's still there they know you still have a valid driver's license so we can do revocation you can do audit logs stick information in there that some government agency wants to look at or that some accountant wants to look at

it and have a guarantee that it hasn't changed over time and we can do stuff I did it with identity and your attributes and we can do metadata for things like medical applications all of these things were the second generation and there's a lot of interest right now it's still embryonic of how to use Ledger's that's just the second generation then we had the third generation which was well if I'm going to be able to store my land in the ledger and you're able to store your money in the ledger maybe we can do a

swap maybe I could give you my land in exchange for your money now we could just email it but here's the problem are you even though me your money and then you never hear from me again because I run off and I don't give you the land or I email you the land but you never send me the money that would be bad so what you do is you have a trusted third party we both give our thing to the trusted third party then they swap it and give it back to us but with a

smart contract you can actually have a computer program being run by the entire community and they can all agree on whether or not the swap happens and you can prevent cheating an incredibly powerful idea let's have little programs that are running so you can do sales like I just talked about you can do agreements with non-repudiation if you and I want to sign a contract I can make sure that I sign it and that you sign it and we both get both signatures and it can relate the possibility that I signed it handed you my

signature and then you just put in your pocket and kept it and then a year from now you could lie and say yeah I signed it too and enforce it or here for now you could lie and say I never saw it and not enforce it that would be bad with smart contracts you could prevent that there's lots of things you can do including distributed applications daos distributed organizations all sorts of cool things so those are the three generations we started with first let's put some money in our Ledger's feel free to do that let's

put other information like assets in our Ledger's well if we have money and assets then let's let us be able to trade them with each other and do smart contracts but if we can do that that raises the question should we do a fourth generation if we have money and we have assets and we're able to swap money for assets with sales then how do we get matched up we need markets we need some way that the sellers and the buyers can get matched up with each other and so they can do the buying and

selling and so with the markets we can complete it this is this does everything then but you have that fairness we need to know that we're gonna have fairness and how we match them up if you have the fairness then you can have markets I'm calling that the fourth generation but it's more than just what you would think of as markets it includes things like video games we need fairness in the video game in some MMO if you pick up an object and I pick up the object at the same time we need to make

sure that I can't cheat and make it seem like I picked it up first when really you pick it up first this is the goal here is we want to be able to not have people cheat and influence what the order that we agree on is going to be but if you have that you can do games you can do stock markets I don't really think we're gonna do the New York Stock Exchange this way but you could also do dark pools which are like little tiny stock markets run by a small number of banks

there it is absolutely a useful thing because they don't really trust each other individually but they would trust that a big group of each other wouldn't do something bad it's the perfect situation you can do games you could do auctions you could do something like an eBay or an Amazon you could do something where first-to-file matters you could set up a patent office you could set up an office selling domain names anything like that where it matters who was there first so those are the four generate for generations you'll notice that as I've been going

along the four generations happened in that order but all four of them continue to evolve and we're getting new use cases new business cases of how to use these things at all for these applications and that the things that we need in many of these cases are coming down to fairness and speed and as we more and more grow up and mature and become real markets who are using lots of money security becomes increasingly important resilience to DDoS attacks that sort of thing so I've told you what hash craft can accomplish I've told you the

history of how Ledger's have evolved and how they're continuing to evolve rapidly how does the hash graph actually work internally what's the algorithm that it uses because it's very different from proof-of-work it's very different from leader based it's very different from simulated economies which you may have heard called proof of stake or called dag based very different from those I don't like those terms because well I can do proof of stake and I use a dag but I'm not at all like the systems that are simulating an economy instead we're gonna do something totally different

in fact really it goes back to the way it worked thirty-five years ago except those systems were so slow you could never use them so I'm just gonna tell you in about two minutes how it works here's the whole algorithm in two minutes and if you don't like math you only have to suffer for two minutes we want speed right so here's the idea Alice has that red envelope she has created a transaction she wants everybody to know it so she tells someone totally randomly this is about the simplest an algorithm I can imagine and

now Dave knows it too two people know it and then each of them tells someone randomly now four people know it and then each of them tells someone that they pick at random now eight people know it it explodes outwards exponentially fast and soon everybody knows it and if you attack one of these computers and shut it down it doesn't matter it'll still just spread out to everybody else that wasn't being attacked there's no one computer that's a bottleneck and there's no one computer that has to tell it call everybody up one after another and

tell them Alice doesn't have to call each person and tell them individually she just tells someone and then tell someone else and then they're telling someone else and it just explodes almost actually what you do is you call somebody but then you make sure you only tell them things they don't know and so what will usually happen is you're gonna have a whole bunch of information when you call a random person some of it they'll know some of it they won't and there's really clever ways to make sure you don't tell them something they already

know that's a really good question that's it that's good too why have you attack the only person there notice in this case Alice then yeah we'll know if an attacker shuts down your computer you are not going to be able to bid in our stock market you are not going to be able to play our game so how is that fair well that's not a failure in the consensus algorithm that's life now that's the truth but what doesn't happen is the whole network shuts down and with something like the leader based systems the whole network

shuts down if they attack just the right person and keep following the leader so let's be clear here nothing is 100% secure if you have a third of the people being evil they can do terrible things and corrupt the network if you have a botnet that can shut down a third of the network they can freeze the whole network if there is an earthquake that chops off a third of our people then the whole network freezes nothing can help you with these things but nothing less than a third is going to hurt you and that's

the point yes what incentivizes this is a layer above the consensus protocol so one answer might be maybe it's a permission network and the only way you get to be part of the partition in that grave is if you're doing the gossiping because you learned information another might be you're doing something that's proof of stake and you get paid to be a miner another might be that you and a few of your friends are playing a video game and unless you're doing the mining then you don't get to play the video game there's lots of

different answers that's all in a layer above the hash graft layer and yeah I know we're gonna have a long time for questions 90% the question is going to be so what do we do about this publicly that won't work in this crypto currency and this those sorts of things and the answer is it's all layer above hash graph and I'm sorry I don't have an answer but I'm talk about how the hash graph works yes if Dave is is attacked oh you're saying in the very first step Alice calls Dave notices he's not talking

and then she just calls someone else at random in fact I said her second call was to Gina if Dave doesn't talk to her she still talks to Deena good question what about digital we have digital signatures nobody can corrupt a message we have digital signatures in fact we're also going to use cryptographic hashes to tie everything together there are huge classes of lies that you just can't tell in this system mostly all you can do is just go silent or set up firewalls or something to break the network or DDoS somebody but you can't

really lie if I tell you that Alice gave me this message since she signed it you know it came from Alice and replay attacks don't happen there's all sorts of things that are just completely impossible unless you break the crypto and I I use strong crypto really strong maybe more strong than I need to yes you've checked with another node to see if it has the information if it already does you check another one isn't the last step of this process going to be every node sending a message to every other node in the system

and if so what's the efficiency game that you get from the initial propagation using this excellent question if I understand it correctly hold one second that's the next slide if I understand it correctly and if I don't let me know after the next slide so what I said is that if Alice wants to talk the fastest way the computer science knows for her to get it out on the Internet here's a gossip protocol this is faster than a a message tree because you're resilient to one node going down and it's much faster than Alice talking

everybody individually the fastest most resilient way we know to get the message out in computer science would be a gossip protocol which is the simplest possible thing just talk to people at random it's about as simple as you can get now here's a question what if Bob and Carol both want to send out a transaction at the same time which is starting to get to what you were talking about you could say we're gonna make them take turns we'll have some mechanism that decides who goes first well that there's propagate to everybody and then we'll

let the other persons database that sometimes do this in two-phase commit there's locks involved leader based systems implicitly do this any system where they have to take turns has a host of problems that slows you down what we the fastest way to do it just do it they both call someone at random and give them message and then they call someone at random and given their message and they call someone at random and give them their message and exponentially fast everybody has both of them in a real system everybody is creating transactions all the time

so you have an enormous flood of transactions going out all the time when I call you there's always going to be a few messages you haven't heard yet you'll get those and so it's not like we're taking turns we don't wait till one message gets to the whole network and then we start on the next note message we're just constantly talking to each other spreading our messages as they're being created how do you prevent no mo spinny aha to prevent double spending you have to know in order on your transactions and the ordering we haven't

got to yet in fact and this is an excellent question look at the colors on Alice and Bob Alice received red than blue than green Bob received red than green than blue they cannot look at the order at which they received them so to be clear a gossip protocol is the fastest most resilient way to get the information out but it does not give you a consensus order we've only been just begun we don't have a consensus order yet so lots of people start with this and then they say now let's talk about consensus but

I don't know talk about consensus I'm talking something else I want to add a tiny little bit of information to each of these messages like 1% bigger just a few more bites to each message that's all I'm going to add is a little tiny bit of information to each message and that's going to be enough for you to see an incredible history of how we've all talked to each other which is weird and then we're gonna get consensus for free you will not have to talk to anyone and you'll have perfect consensus with math guarantees

that you're right that's weird that's crazy so let's talk about this we all have these messages and you know I just showed this animation of everyone talking to each other let's draw a diagram of it no one ever draws diagrams of this I want to draw a diagram of it so here's a network with just three people Alice Bob and Carol time goes up hill you see the three circles at the autum are the are Alice Bob and Carol those are just the three at the bottom each of these circles then above that represents an

event of somebody gossiping to someone else so look at the top circle on the whole thing that's the last event that I have in this diagram and it is in the bob column and it has a line going down to a circle and the Carol column that top circle represents the event where Carol called Bob and told him everything she knows all the messages she knows she gave to Bob this diagram is a way of seeing the whole history if you look at this diagram you can see how we talk to each other and in

what order we talked to each other and in fact if you told me there was some information analysis circle at the very bottom there I could just follow the lines and see how the information flowed through the network it's really kind of cool and if you told me there was something in Carol's circle at the bottom I could see how it flowed and for any given person I could tell you which of the two messages they got first you can just watch how everything flows to the network from this diagram it's a really cool diagram

nobody ever draws these diagrams but this is a diagram you could draw that would show you the whole history of gossip now we can gossip about transactions we can gossip about identities we could gossip about information that we find interesting but what if we were to gossip about gossip what if we were to be gossiping what we know about this very picture what if we were to do gossip about gossip in other words when I call you I give you everything I know what I mean is I give you all the pieces of this very

picture that I know that's weird very recursive self-referential we are going to gossip about the picture where the picture is the record of how we gossiped bizarre but we're gonna do that so each of these circles now is actually going to be a little data structure in our memory that is the message that we're sending people now we do in a head sent to some transactions so you'll put the transactions inside the circles but in in addition to a circle having to have the transaction it will also remember two lines going down from itself the

thick line for what column it's in and the thin line for who was talking to that person just two lines two pointers to hashes we're building a graph made out of held together with hashes in math this is called a graph it's held together with hashes so we call it a hash graph so the hash graph is the history of how we've all talked to each other but all I have to do is send you the transactions I was going to be sending you anyway digitally signing them sending you timestamps all the stuff I would

have been doing anyway plus two hashes that's it each message just has to be bigger by two hashes and in fact for various reasons I can go press these each of those hashes done to one or two bytes and I'm fine I still have total security so we started with gossip that's the fastest way to get your transactions out we added a tiny bit of overhead and for free we got the whole hash graph that shows the entire history of how we've all talked to each other and now we know what each other knows I

know what you know and I know what you know about what she knows and not only that I know when you learned about what you know about what she knows about when he learned this thing from him I know it all wait so do you also need to we're doing that each of those circles has a timestamp in it you got it it's changing hands I mean it seems like an enormous amount of information but wait if we're all creating transactions then these circles have the transactions in them they would have had to anyway and

they had signatures but we would have had to do that anyway and they have time stamps even Bitcoin has time stamps so if we just looked at a system that was purely trying to get the transactions out and not worrying about consensus at all you would still have to send the transactions the time stamps and the signatures this thing is sending exactly that plus two hashes in each message and the hash can each be can each be compressed down to one or two bytes so it turns out to be a tiny bit of additional information

on top of what we were already sending anyway that's what's so weird here also it's not saying the story of information just two hashes that's kind of that's it it's kind of how does that work why don't I have to send the whole history I had to send you these transactions so you had all the transactions I had to send you these I had to send you these I had to send you these so when I send you that top circle I don't have to send you all these circles you already have them all I

have to do is send you this little line here that tells you which circle it connects to so you think you're getting this enormous history but you already have all those parts of the history all I need to do is connect it to two things so you just have two hashes that's it so you know that I'm sitting here's a tiny bit beyond what you would have had to send anyway just to get the transactions out there but we now have this beautiful history of picture of history so it's cheap almost for free we still

have consensus order though there's still no consensus here if you look at that picture you can't put the circles in order yet but here's the weird thing I had to send you the transactions the signatures and the time stamps anyway I've sent you a tiny bit of information in addition which gave you this entire hash graph which let me know what everybody was thinking and when they knew it and what they knew now I'm gonna go to the shelf and get one of those 35 year old algorithms that were so beautiful mathematically really strong proofs

asynchronous Byzantine security the strongest known algorithms but they involved voting sending votes over the internet and with so many votes it just collapses under its own weight it was too slow to my knowledge no one has ever actually deployed in the real world a system using these things these voting systems but what we could do here is I could say I know what you know I know everything about what you know in fact if we were running one of those our old algorithms I know exactly how you would vote so don't bother voting I'll just

pretend that you voted I'll just pretend that you sent me a vote and so we're gonna run one of these old voting algorithms but with no votes we're gonna do virtual voting by doing virtual voting what I mean is that we're essentially running old algorithms they're tweaked a little bit they're actually slightly different but it's basically running these thirty-five-year-old algorithms that have these wonderful math proofs the only downside of the voting algorithm is is because they have lots of votes and receipts but we're doing it with nova and no receipts and so purely in my

mind I run one of these algorithms and I know the consensus order so practically what does this mean it means that whatever I need to get a transaction if you've created transaction it will get to me very fast because we're using gossip doesn't have to go through a leader to take turns don't have to solve a math problem in a proof-of-work system I'm gonna get your transaction really fast I will have to pay for maybe one percent more bandwidth to get it because you're gonna throw in the two hashes and then after I get a

few more transactions I will just know the order of this one I don't have to talk to people about the consensus I don't have to talk to a leader there's no one person in charge it's purely internal I can just see what the consensus is by virtually running one of these old algorithms but not running it for real and so one of these circles has these things in it has a transactions in it it has a timestamp it's signed by its creator and has the two hashes and even those can be compressed because I'm always

sending you the hash of an object you already have in memory and you don't have very many objects and memories so it doesn't take very much effort to tell you what hash I'm thinking of and that's it hash graph works by gossip about gossip with virtual voting it's bizarre this gossip about gossip thing is weirdly self referential but you get all the the decade's worth of math proofs all apply it's a synchronous Byzantine it's fair and the speed is as good as you're gonna get gossip is as fast as you're going to get you're just

paint you know this tiny penalty of two hashes and there's no communication at all for the consensus part you get that for free but please let me have questions yes two really good questions so I'm assuming the the sending the hashes is how you get around no getting information than having to recheck the rest of the system because it knows which nodes have already passed along that information from one to the other how does the number of hashes that you have to send scale with the number of nodes in the system so if it's if

you have a thousand nodes instead of three do you have to send some function of that knodel system in terms of hashes and does that add to the actual data that's being transfer in each transaction it does not now let me let me rephrase your question the word NotI I hate this circles are called nodes and computers are called nodes in these two communities and now we're in the intersection the two communities but your talk about computers I understand that I usually call them full nodes and I don't call circles nodes for the very reason

of this the question was do I have to send more hashes here's the answer if you have a good flow of transactions so that each of the computers each of the full nodes is is created putting some transactions into each event then the overhead is constant no matter how many people you have the overhead in bytes that you have to send is constant you end up having to have bandwidth proportional to the number of transactions per second you want to handle that's constant that's throughput then you have latency latency is logarithmic in the number of

nodes because of what we just saw several slides ago it's exponentially fast will exponentially fast invert that you have logarithmic so if I have a thousand nodes I'm probably going to have to talk about ten times before my message gets to everybody if I have a million computers I'm gonna have to talk 20 times if I have a billion computers I'm gonna have to talk thirty times it's logarithmic and again that's the best you can do with the internet yes I actually so you miss me ledger I of a massive scale like the answer is

yes you could build a public ledger on hash graph with one shard which i think is what you're asking but if I had more than a thousand nodes I wouldn't waste it all on one shard I did get more shard to get even more speed so let me just mention sharding really briefly here and then I'll continue with your question just like computers have single processors or you can get a graphics card that has a thousand processors in it same thing with shards you can have a ledger where every computer knows everything every computer every

transaction but of course you're limited by how many transactions per second you have the bandwidth for if my computer can only get 500,000 transactions per second then the entire network is limited to 500 thousand transactions per second if which is actually a big number but but for some applications you want millions or billions of transactions per second how do you fix that well you don't make every computer see every transaction what you do is you divide your computer's up into clusters and within a cluster they all see the same transactions but each transaction only goes

to one cluster only goes to several clusters these clusters are called shards that comes from the database world where it's databases put data on just some computers so if you wanted a big system with a billion I wouldn't waste them all in one shard I would make a whole bunch of shards that are each a hundred or a thousand because you get all the benefits of shorting hash graph would work fine in that big system but but I would actually build a shortage system now the bigger question you're asking is how do we go about

doing a public ledger and how do we don't go about doing sharding and the answer is there's many ways to do that on top of hash graph and you could use hash graph as a building block to build a shortage system or a ledger public ledger in many different ways we haven't published one yet so you know don't ask me the algorithm there is no algorithm we've published but you know we may do that in the future and right now what I'm describing to you is a building block that makes a single shard really fast

yes since we're in a business so I understand that there is a company behind the the hash graph where the company is trying to kind of monetize the value of the hash graph yes when you are going to Beach the technology to enterprises yes so you position yourself as a blockchain like technology that has some competitiveness worth overlooking or you are kind of trying to stand out from the crowd of like technologies out there because from my understanding poaching is - in early days of the development right so I would a company be we went

to understand that you are more yes advantages in front of a blockchain if this can not understand the blockchain itself at this particular moment yes so um oddly enough we have to educate people but this happens so let me tell you um for example the credit union industry they 6000 credit unions in North America they created an organization called Cu ledger credit union ledger the purpose of Cu ledger is to have a single ledger they were calling a blockchain that would store information that the credit unions want to share with each other and it's this

whole deal that we talked about they want to be private to people outside the credit unions can't see it but the credit unions themselves will see it no credit union trusts a single credit union but you can bet the of the 6000 probably no 1/3 of them are going to be evil so it's the traditional permissioned Ledger's situation they said we want to build a blockchain again people use the word blockchain to refer to all DL T's including us so in that sense we are a blockchain we talk to them actually they called us they'd

heard of us and they called us and then they looked at other ones like hyper ledger they looked at a number of competitors and they implemented in-house you know our stuff's been on our website for a year and a half you just download it and play with it and you can build little apps on top of it and see how they work they tried that they were very impressed and they chose us over these other ones like hyper ledger and the reason that they chose us was it's fast fair and secure they actually didn't care

a whole lot about fairness as I said some of these early applications don't need it something like a game or a stock market would need it but they cared about fast and secure and honestly what they cared about the most was fast they wanted this incredible speed which many of the public Ledger's wouldn't give them but they were also somewhat concerned about the security if you try to get speed by going to a leader based system then you've opened yourself up to DDoS attacks that can be really bad basically your target market as of now

this companies were who want to implement blockchain yes you're waiting deals over blockchain as a faster more that's right right so every time we talk to a company we talked to their blockchain division and I don't complain I mean that is one of the one of the definitions of blockchain means DLT and we're a DLT so we're blockchain but what we say is look look at the speed that we have look at the security we have and look at improvable mathematically proven security and look at the fairness that we have and so we have lots

of customers in the pipeline that are very they're very excited about this and oddly it's the speed that seems to get people the most excited but then we talk to them enough they start to realize yes security matters too along those lines I have two questions the first is can you talk about what's an extra hash graph maybe in the next six months or a year any milestones that you know top of mind and then the second is forgive me being a noob here but if this is such I mean it's not like a new

technology the gossip based protocol so why I happened like a few more other block teams people ended that you know from the get-go oh I'll answer the second one first gossip is really old gossip about gossip is unprecedented and very weird not the kind of thing you'd think of and then virtual voting taking one of those 30 year old algorithms and trying to apply it on a gossip about gossip system just bizarre that that's the short answer for that it's just strange though I will say every time I've talked to a mathematician they say oh

well that would work huh never thought of that right it's kind of weird but when you see and you say well yeah that would work I understand then you have to go through the proof but we do have a proof with lemmas and theorems and stuff yeah a real proof yeah and I actually forgot the first half of your question what was the first half what's next oh yeah yeah I won't answer that part so Hat swirls incorporated is selling these permission Ledger's were getting amazing traction we plan to continue expand that to add more

features and we will eventually add sharding and all sorts of things as for the public side we haven't said anything yet we have not announced algorithms for that hash graph by itself is a good foundation for a public ledger but is not a public ledger because you have to say how are you do sharding how are you gonna do steak how are you gonna do lots of different things the smart contracts all those pieces have to be said so my short answer would just be stay tuned okay yes I guess I have two questions so

on the permission Ledger's side for like example for a game the other is the value proposition that for a game like World of Warcraft or something is it sort of a cost-saving measure or they wouldn't have to host yeah I'm game set in a centralized way which presumably costs money naked yes this changes the entire economic model you can imagine users who don't have to pay for a server because there is no servers just their computer while they're playing right now some kid has to pay Microsoft $10 a month to play Minecraft and they can

only have 11 kids playing at a time but this could be free and you can have a million kids playing at a time so it's it's a different economic model how about an indie game developer that would love to do an MMO but can't afford the big servers they don't have to know so an MMO can actually be a bunch of just the players themselves running it doesn't have to be public probably what you would actually do for an MMO as I think about it is that you would have part of it being public and

part of it being a whole bunch of little privates and they would all be connecting to to be almost like side chains but it depends on how you want to do it but if you wanted to you can actually start with a purely private if you wanted to is this is not this framework enough for someone way smarter than me to use the math information good question no it is enough and furthermore these guys are smart all these systems like hyper ledger is an enormous Lea complicated stack of technologies and a built in modular it's

plug-and-play and guess what the very bottom module does just the Consensus part and so absolutely we're talking about building plugins that will do that absolutely and so these big stacks that people are building they're doing the right thing which is to make a modular and so then you can have all these different consensus like roads at the bottom you just plug arson which is a good thing yeah which I guess answers my question about the future that's where I wasn't going to answer yes my limited understanding of the technology would you mind sharing [Music] conventional

blockchains where you've got an immutable chain of blocks I understand that there's a hash that's associated with the previous block that has to get matched to the new block and there's a cryptographic an equation that goes into solving that there is that different to this yes and also my second question is how do you get enough computers and people on board it do you need to have the user base alive before this kind of actually starts to work again I'll start with a second question for the user base you know we just got 6,000 credit

unions that'll be using it because they benefit from it okay so that's not a problem for permission Network its whoever wanted to be playing it if he released a game that you would say well whoever wants to play the game you invite your friends and those are the people playing and that's not a problem if you want to talk about a public ledger that's an entirely separate question there's a whole bunch of issues that arise and so all those need to be solved and hash graph is a beautiful building block to use in such a

system but I haven't published such a system so I can't talk about details of it but but I think hash graph is a nice it's good to use hash graph for the consensus part of that system but a public ledger is much bigger than just a consensus algorithm right and then the first half of your question which I also forgot how's it different to a conventional blockchain oh yeah sure example from one block got it so the short answer is a blockchain is a chain of blocks and a hash graph is a graph held together

with hashes here's this less short answer in a blockchain and when you say blockchain nine gathering that you mean proof-of-work you're not to mount one leader based system sure I'm a proof of work and a proof of work system the way it works is that you have a a chain of blocks each of which has the hash of the previous ones that nobody can go back and change something in the history makes it immutable and then to put the next block on somebody will come along and put the next block on that's great the problem

is if two people put the same next block on then we as a community have to chop off one of them and you only accept the other one and we do that by voting with our feet we decide which of the two to expand you always try to expand extend the one that's longer and then eventually it's so much longer that we just throw away this one and no one ever bothers with it that's how it works the problem with that is that if we were forking this way too often we would get new forks

before we finished chopping off the old fork and the whole system would collapse we keep sprouting new forks all the time so we have to slow it down so how do we slow it down we do proof of work in a proof-of-work system the whole purpose of the proof of work is to make it run slower that's interesting I don't have a part of my engine in my car whose purpose is to make my car run slower but that's the whole point of proof of work is to make sure that we're only adding a new

block every 10 minutes or some of the newer ones would be much faster but but make sure it's slow enough that the community as a whole will have time to figure out which of the chains is longer and always go with the longer fork to do that you have to solve a math problem the math problem does involve taking a hash and trying to find one that will have certain zeros and so on but it doesn't even matter what you do for your math problem the point is it's a useless math problem it's really hard

to solve so you need a supercomputer and lots of electricity problem anytime that you need a supercomputer lots electricity and there's an economies of scale you're gonna have minor consolidation fewer and fewer people are gonna have that supercomputer and if it uses lots of electricity everybody's gonna move to where the electricity is cheap you can have Geographic consolidation and once you've got minor consolidation and Geographic consolidation then our beautiful trust model maybe is not quite as trustworthy if we all live in one spot and the government of that spot nationalizes all our mining rigs and

makes them start extending the shorter branches then they've destroyed the entire system so that's how the blockchain works for us you don't have a chain you just have that picture I showed with all the circles with lines and anybody who wants to add a new circle does it in fact typically the experiments I run each individual computer is adding 10 new circles every second so it's like a minor in Bitcoin mining ten blocks a second every single miner bought mining ten new blocks a second in blockchain that wouldn't work because we wouldn't it be too

fast we wouldn't be able to chop off these Forks but in hash graph we like Forks nothing ever gets chopped off so there's no problem is it overlay of the disporting algorithm that kind of validates that's it in fact that goes back to the question you had asked earlier too in a blockchain it's easy to put them in order it's just the order of the chain but in this system maybe it isn't so obvious oh there's the graph that graph there it isn't obvious what order to put the circles in but because the graph represents

the history of how we talked I know enough to run in my head one of those old voting algorithms and I can put every single circle in order and so in my head I do actually put every single circle in order that puts all of my transactions in an exact order that I know everybody agrees on and then if you double spin I know which of the two stories got at first which one got at second and we say the money actually went to the first store and not the second so all the problems go

away so we get beautiful advantages by having a graph rather than a chain but then you say well how do you get consensus and we do that by doing the virtual voting but you can only do the virtual voting because I have this history of hat what we all know and how we talk to each other and the reason I have the history is because I did gossip about gossip I did a hash graph and that is how this is different from other systems that use a graph there's a bunch of systems out there that

draw pictures that look like that picture but their picture doesn't represent a history of how we talked to each other so they couldn't do virtual voting so they couldn't get all the advantages of that yes sure so with proof of work based blockchains yes they really for a bad actor to kind of corrupt the system you'd have 51% of the hash power they say which is a really high bar 34 yes I thought was 51 it was a company bond okay so that was basically pointing that question because earlier said it's one-third what this is

there's a difference between blockchain and your graph system no so there are lots of people who are aware that bitcoin is a 34% system because there's math theorems that say that every system is a 34% system I should have brought the pictures by the way I have a video on our website of me giving a talk at Berkeley at the crypto economics conference and I go through this and in fact at that same conference one of the earlier speakers just casually mentioned and of course one third can attack Bitcoin so apparently it's common knowledge among

some people but other people talk about 51 the right way to say things is there are 51% attacks on Bitcoin but there's also 34% attacks if we're assuming that there are things like firewalls in the world and that's what we have to talk about is if you're assuming there are things like firewalls in the world and 34 percent enough here's how it works imagine that we are all computers and that we are evenly split between these three sections of the room this room is conveniently split into three sections so a third of us live over

here a third of us live here and a third of us live here and we will furthermore assume that the people in the middle are all malicious and evil and bad people I'm sorry but you guys are good and you guys are good now imagine that there is a firewall around our country over here that is around these third of the people so I'm going to build a firewall around this third of the room and then I am going to be a malicious person and I'm going to turn off the firewall in other words I'm

going to shut down all traffic between this third of the room and this two-thirds of the room now if you have a simple Bitcoin with no partition detector then the attacker has already won what's going to happen is you guys will keep talking to each other you won't talk to them because you can't and as you're talking to each other you'll keep trying to mine and you'll keep adding new blocks on top of the chain it'll be running at a third of the speed is it what it usually runs but you'll notice that it's working

and you'll build your chain you guys will do the same thing adding different blocks and adding to them to the chain and you'll notice is you're running at 2/3 your normal speed but you're still running that is bad because your chain can get six long or 12 long or a million long yours can get six longer have six confirmations or twelve confirmations or a million confirmations and now we have two contradictory histories two contradictory chains so that is bad and if I have a satellite phone or if I including with the firewall person and they

let my messages through I can double and I can double so I could spin my coin on this side and spend my coin on this side wait for six confirmations and have the store given my product wait for six confirmations that have the store given my product and now I have two products for the price of one I have cheated that is for a simple Bitcoin with no partition detector so then you say oh wait why don't we do this why don't we say that we will just freeze if we notice that a third of

the world has gone silent mysteriously and not talked um not talked to us okay oh well then we don't even need the firewall I said these middle third and malicious they just turn off the computers and then you guys will say oh third the world has gone silent and you'll all freeze and the whole network freezes forever okay so not that why don't we have a partition detector that says if more than a third of the world goes silent they will just freeze now it's okay for this firewall because what happens is you guys will

freeze but you guys will say well we have a full two thirds still talking to each other we'll keep going and so now I can't double spend all is well except these guys are malicious and you guys are colluding with the person who made the firewall and the firewall lets your messages through now you guys are getting messages from two thirds of the world you and them and you guys are getting messages from two thirds of the world you and them you guys are sending your messages both ways and they get through the firewall but

you guys aren't able to get through the firewall now our partition detector is totally happy it says hey two-thirds of the world is communicating with me I want to keep going and so you guys get your own chain and you guys get your own chain and they're different and I can double spend and these malicious people who are talking to both sides can easily double spend that's bad so you could say something about okay we're not just going to do partition detectors based on who's talking to me where I'm actually going to count how many

blocks get added well we still have a problem in that guys going to be adding blocks on both chains but at least they're running at half the speed so now your partition detector has to somehow guess how fast each person by themself is producing blocks but any given computer isn't producing blocks very fast anyway they don't mind very often so it's gonna be really hard to notice with he now seems to be going half his normal speed rather than his full normal speed because any one computer Minds pretty slowly anyway and so I'm definitely gonna

be able to double spend and so no matter how you try to build your partition detector you end up finding them either the attackers can double spend or the attackers can shut down the entire network forever one or the other this is not surprising because we have 35 years of math proofs all saying that you will always be able to attack if you have a third of the people who are malicious and you have a firewall you will always be able to either freeze everybody forever or force a false consensus one or the other yes

I'm just sort of like it's not necessarily true because less secure than than what people kind of conceptualize as it being secure against it only a 50 100 50 100 K so you're asking two different questions one was somebody said I heard 51 I thought it was I didn't think it was 34 and I'm just saying well actually mathematically it's 34 to some degree the two numbers are so close it doesn't matter a whole lot but you're also saying in practice would there ever be such a firewall guess what there happens to be a cryptocurrency

that is concentrating a country that happens to have a firewall right okay okay so is that a problem or not and and honestly I pretty much trust the government of that company in the country to not shut it down but the whole point of Bitcoin was to not have to trust anybody so the fact that we have to now trust a single party is kind of annoying yeah so it's okay yes so do you have any blood running and if so how many what is the volume how many transactions so how many tokens from hash

pass have you we do not have a public network we have the private network anybody can download it you can download it right now and you'll see on your computer you're getting hundreds of thousands of transactions a second that's just your computer talking to itself so you can pretend that your computer is four computers already on one machine and you'll see 250,000 transactions per second if you run our software or you can get AWS instances and you can run them all in one region or in two regions on opposite sides United States or an eight

regions spread around the country sorry around the globe from Tokyo to Canada to South America to Berlin and South Korea and east coast and west coast if that's eight that those of the eight we did and what we found was good speed numbers and there's lots of trade offs and we can talk about the what the exact speed numbers are I've kind of alluded to them we are at this very moment I have one of our guys running a big complicated set of experiments on AWS so that we can publish this and we're going to

publish it with all the caveats and and I want to emphasize again what I said before it's the consensus part of this that is so amazingly fast that it's not going to be it's not going to be your barrier so what really is going to be your barrier is how fast the computers themselves are probably your bandwidth in your processing speed but the network itself isn't really going to cost you very much for consensus but yes we've we've tested that yes yeah blockchain already became like a buzzword for banking industry like Arab banks you've lost

curious from your perspective what does it mean what it's occurring Thoreau said of banking adopted blockchain and then where do you see is the most potential for banking maybe yeah for whatever reason we have the majority of our customers are in the banking and financial services and payments industry I'm not entirely sure why they're first maybe because Bitcoin was a cryptocurrency so they just sort of started but for whatever reason the FinTech and Fiserv industry is is the leader in this one of the leaders but there's others coming alongside that are interesting and actually some

really interesting stuff going on if you're part of this Bakhtin Club that I'm sure you're aware that there's really cool things going on what we have been finding is that there's an enormous edge so basically everybody in that industry has a team every big company has a team looking at blockchain well we also keep hearing as they're saying we're not ready yet or blockchain isn't ready yet it's not we're not ready it's they're not ready for us that they want blockchain to have greater speed and lower transaction costs and some of these other things first

and then we start talk about security and they say oh yeah add that to the list so that's what we're finding so you're asking what are the what are the current applications yes the future is easy I think that it's basically going to change everything you won't use not everything will end up without a server and without a network but lots of things in every industry you're gonna go from a central server to a distributed because there's sort of no cost to doing so in most cases and huge benefits so where will they end up

users in the long run for everything I think in the short run the funny thing is that industry is also really into NDA's especially about their use cases I can say they are very interested in transferring money or value they are very interested in records as in audit records and things they are very interested in any kind of shared data that other parties are interested in so you know there's cooperation between people in an industry different companies in an industry and you trust each other but not a whole lot and so maybe you and I

trust each other to share some information but I don't want to trust you to never corrupt my database and so this is the kinds of things that I'm seeing yeah and there's a really vague answers and I'm sorry the short answer would be you know all the media hype about all the things you might use blockchain for yeah people are interested in those things and they're going to happen eventually and like all hype it'll be slower than the hype suggests but then it'll happen fast thank you I don't need to do a nice you're too

young get the hash we're talking out any market yes all right now we have a consensus algorithm we have a platform we built on it and we're doing permission Ledger's and we are thinking about other things and we'll be talking in the future about other things okay and as you might have guessed that wasn't the first time I've heard that question okay so what is your good work are you good because you seem to be ma towards enterprise well right now our go-to-market strategy is the permission network side and we have more people interested than

we can even handle we are finding that the pent-up demand for a fast and secure private network is overwhelming for permission Ledger's but we're also interested in other things as well at the end of the day what would be your product would it be the consulting services that you provide to enterprises what would it be a standalone cloud infrastructure okay there are many directions to go in the future including public networks right now we have a platform it is a program that you the customer puts it on all their computers and their competitors computers whoever

is talking to each other and then they can write an app that does the business logic of interpreting what the transactions actually mean to them and our platform is the consensus engine that then allows the app to create a transaction and then the app can rest assured that it will get spread to everybody everybody will come to an agreement and then when we all know the order and the timestamp the app will be told here's the official order and here's the timestamp the app can do something what we have found is that the platform we've

built actually makes it really easy to write apps we have six example apps with source code we have customers one one one of our early customer actually is this yeah one of our early customers said okay we'll download from your website your platform then they went away and we didn't hear from them for a couple weeks they came back and they said yeah we wrote our app it was just as easy to write as you said didn't have to have any help didn't have any hand-holding they just went off and wrote it and it worked

and they liked it so right now what we are doing is licensing this platform and then people build apps on top of it you could imagine a host of different business models that could be pursued in the future and we are just taking what we believe the logical approach to them and right now we're doing what I just said that answer your question no kinda okay so what I'm interested to hear about is basically I know I worked in VC before and I worked with open source companies yes so they have a separate mode of

how they distribute and how they make money yes so my question like at the end of the day would be how would you yes money what do you sell licenses for s we sell licenses to the platform people pay us that answer yes yes I understand sorry right we are licensing the platform for people to build up apps on top of yes and you mentioned VC we our seed round was three million dollars led by NEA and so you know they're big VC firm yeah that's why I asked basically I know that they would not

back up something right you are correct right and you never want to have about a million ideas for business models as a start-up that's a very bad idea what you want to do is start by saying I'm gonna build a search engine and then eventually you become alphabet but you don't start as alphabet I guarantee you that our goal is to start with a search engine but to end up being an alphabet that's our goal yes contracts and like execution of all transactions so the app in a sense is like a smart contract this is

sort of a different way of doing smart contracts but you could also talk about building an app that is an EVM for smart contracts and then separate into two layers so then you have smart contract scripts that run on top of the app that is the interpreter for the scripts so all that is a layer above what we are at right now so at the moment we're not selling that and we're not we haven't published algorithms for that yet but obviously the core thing you need for that is the ability to put transactions in order

so you know consensus algorithm is one piece of what you need for smart contracts and smart contracts are important for any big system so when the hash graph becomes the alphabet that you'd like it to be yes is there so Bitcoin where does it kind of swallow up all these other I don't know there's always room for more than one player right you know so when etherium came along he didn't kill Bitcoin and bitcoins doing pretty well so you know maybe room for multiple people I don't know typically the history of any kind of market

is that you had this enormous springing up of thousands and thousands of companies and then you have a crash and then you get down to three or four or whatever big companies that continue on there and that's the way every revolution tends to happen I'm sure we'll see the same pattern here yes okay so bitcoin is not a permission ledger we're doing a permission ledger and our permission Ledger's doing really well when we compete against other permission Ledger's in the future maybe we would do a public ledger and then ask me then know tell you

how we did but right now we're doing permission Ledger's and we are finding that we are very competitive as a permission ledger for reasons you might guess from the talk so people use no material as a platform because of this ability and you know the computer worldwide computer division of Italy everything we want I mean do you think it makes sense to have to go to deserve this direction as well just saying is more you know narrow does it yes you just asked me should be some day have a public ledger so a tier is

known for me a platform yes which you can use the solidity language to you this much confidence question is do everything please well yeah yes you could build all the parts of a public ledger on top of us so what you're talking about is smart contracts you could have a public ledger with smart contracts you have a punk ledger with a cryptocurrency you have a public ledger with a sharded file system and with them other forms of sharding and a distributed hash table all those pieces you could do it public you could also do permission

versions of all those pieces including a permissioned a theory of solidity interpreter but you know what what we have right now is actually doing the first step of that anyway we have a platform that makes it really easy to build these apps you could view these apps as being like smart contracts themselves but there's not not smart contracts that are coming in from outside users so it's sort of a permission version of smart contracts so right now what we're having is doing the permissioned version of that but not the public version of that in the

future we might expand out to further things but at the moment what we're finding is we are trying to do what a start-up does which is you go with what your customers are really asking for right now so what we are doing is what our customers are really asking for right now and they say we love the ability to build these little apps in Java and you made it very easy for us they're not asking please let me build these little apps in solidity and have them interpreted if they were asking very loudly we would

prioritize when we did that sort of thing but right now it hasn't been what they're asking for in the long run we would do everything right but right now we're just doing the thing that the customers are actually asking for well I appreciate your patience and I appreciate all the time you guys gave us an hour and a half I appreciate you coming and showing interest and feel free to look at our website we got math stuff we've got videos and we have some day this video maybe and if they let us have it I

think the girl just happened we'll put it on our website if not it's probably on a Harvard website but I appreciate this and thank you very much for letting me come here [Applause]