Resources and 1 to N for Consensus

hi everyone this is charles hoskinson broadcasting live from warm sunny colorado always warm always sunny sometimes colorado my apologies for being a little bit late to the live stream there were some technical difficulties and then we had an unexpected visitor to the office and he caught me right when i was just about to click broadcast and i didn't realize that the live stream was already connected to youtube i thought i was just in the waiting room to initiate the live stream so i wasn't actually aware that you guys were waiting so long so my apologies for that anyway i i did want to make a video because on the back of the miami bitcoin conference that i attended i i could not believe the level of disinformation fud and mean-spiritedness coming from a lot of maximalists in the bitcoin community over the things that we do on the proof-of-stake side of the world and to this point i think it's become no longer possible to have a conversation with anyone in that circle they just simply are part of a religion now and the point of this video is to discuss a little bit about some basic stuff that seems to be completely lost and missed and talk a little bit about stuff we're doing in the space as it relates to consensus so as many of you are aware the engines of cryptocurrencies the things that get them to transform from one state to the next state to the next day is a consensus algorithm so what happens is you have this big decentralized mesh of nodes and connections and transactions are occurring all the time information is flowing all the time you somehow have to order all of them in a meaningful way this is a very classical problem in distributed systems and it's something that's been discussed and thought about for well more than 40 years so it's a very old problem and it was not something that magically was discovered and invented by satoshi there were papers the 1960s and 1970s discussing this problem from the beginning of when computers existed people started realizing that when computers are in different locations perceptions of orders of events are very difficult to ascertain without designating a trusted third party to take care of that so proof of work is just an engine that basically allows you to make decisions of who gets to decide the serialization for that moment that update of the system that block the heartbeat of a blockchain is a block every time a block comes the heart beats transactions get approved proof of work is just a type of engine so let's talk a little bit about this and what makes this video exciting is that we actually get to talk a little bit about where the future is going and i think this is something that's really not discussed very much okay so this right here if you all can see my nice little blackboard whether you're proof of work it's kind of a trippy color let's go to white because that's a little easier to see whether you're proof of work or proof of stake you have the same three things that happen okay one you have to choose someone in charge to order everything that goes into that heartbeat of the system the block two you actually have to beat the heart so you make the block and three the network has to accept the block all right proof of stake and proof of work are basically identical in steps two or three in that you can put a proof of stake system in bitcoin and ethereum and other such things you're not going to actually change the ledger rules or the validation predicates or all the kinds of things that would go into accepting whether something is real or legitimate where they differ is here the how do you pick somebody in charge it's done here with and they both use different notions of resources let's use that term resource okay in the case of proof of work this is based upon hashes so how many hashes per second you can do in case of proof of stakes this is based upon ownership of an asset all right so the resources differ but the concept is basically the same if you control 25 let's say of the total amount of hashes per second the total amount of hash rate in your proof of work system on average 25 of the time you're going to be selected to be in charge now that that that's not always the case you can sometimes 18 percent sometimes 35 of the time but on average if you graph it out and you go over a long enough period of time the numbers will cluster around 25 percent of the time and it's the same for proof of stake if you own 25 percent of the system well then on average 25 of the time you will be or the people you designate your proxies the people you delegate to will be in charge and have the right to make block all right so the point of this video is to say hang on a second here this is just a resource and guess what you can combine these together there was a cryptocurrency in 2011 called peercoin you guys remember that sonny's thing that was actually a hybrid proof-of-work proof-of-stake protocol which meant that it actually didn't have one resource but actually had two resources so people could mine and they also because of their ownership of the network could also be selected to make a block and there was some mechanism within peercoin to work on that in 2016 there was a follow-up protocol that was made by hung chang and alex cherpinoy called twins coins that actually tried to improve that model and there's still a lot of thought and work onto how do you go from one resource in this system to decide who's in charge to two resources to eventually end resources okay so you have a blockchain in each of these slots you're going to make a block and there's a question mark of well who's the person to do that because there's an expectation that blocks should be made with some degree of regularity so in bitcoin it's 10 minutes in cardano it's roughly 15 seconds give or take sometimes more sometimes less try to keep synchronized and each version of over boris has a different opinion on the on the rate that they come but anyway you can say well only one resource decides this or two or n many resources to sizes so why would you want to do that well every pool of resources comes with a philosophy behind it so if you're selecting resources out of the proof of work side then you are making assumptions about the availability of asics you're making assumptions about energy cost you're making assumptions about knowledge domain expertise it's very easy for a person delegating cardano it's a virtual resource it can move anywhere in the world it's very hard for people to build a 200 million dollar data center and run that data center okay so there's different expectations you're making about these things and and for proof of stake you're making the expectation that somebody actually has the ability to buy ada that markets exist and that those markets are relatively decentralized and there's always a seller for a buyer to basically get some ownership over that virtual resource you're also making assumptions about the level of veracity honesty credibility of the people who control these resources so with mining you assume an honest majority with a plutocratic system a proof-of-stake system you assume honest majority and the incentives for honesty are different so in proof-of-work system the incentive for honesty is if you wreck the token then your miners decrease in value if you can only use it on that token in the beginning that was true to a certain extent for asics and and other such things built for bitcoin you could mine other things that were similar but the profit margins were dramatically smaller but when you see things that use the same algorithms and the work rate is relatively the same the price is relatively the same you actually have perverse incentives to do what's called a gold finger attack in the case of proof of stake you own the system so if the price dramatically decreases you're actually taking value out of your own pocket excluding the existence of derivatives and short selling and other things that seems to be a rational assumption but it may break in certain cases so in the effort of resilience the goal of going from one resource to n resources means that your system is more overall dynamic because you're getting your security you're getting your block production from more than one pool potentially but there's a really difficult question of how do you stitch these things together and actually it turns out that the twins coins paper had a few issues with it and attempts to create hybrid proof of work proof of stake systems is a little bit more elusive the other thing is that your resources don't have to be connected to a computer driven or ownership driven model so many cases you guys will hear things like proof of merit okay so that's this concept of and there's many variations of it duh duh but basically the idea is that somebody does something beneficial we'll call him bob for your cryptocurrency and the more good things that bob does for that cryptocurrency and we'll just put that in quotes because who the hell knows what good things are he'll get some sort of token for that and that token represents a resource so identifying the good things and a fair way of distributing that is actually an ongoing research problem that's really exciting and sexy and cool and then also how do you make sure bob is real versus a bot robot bob a little laser eyes okay so that is what's called civil resistant okay so there are other resources you guys could have like for example let's say that your blockchain is getting really really big and really really heavy this the totality of all this if you sum up your blockchain yeah let's zoom in right here yeah here we go you sum up your blockchain the sum of this whole thing and that equals i don't know like eight exabytes all right so a huge amount of data is is there well who's going to store that so maybe you can do proof of storage and there's some really cool protocols like for example permacoin was one that i'm familiar with but there are things that have come since that time chia is another ecosystem that's trying to figure out ways of connecting history to to a blockchain but that could be another resource you could actually instead of saying hash power it could be hard drive space okay so the name of the game once you have a strong theoretical basis is to say okay let's go from one to n resources and then you select that pool of resources to be a composition of computational power a composition of social power so socially good things for the system things like storage you could even think of network capacity for example right now peer-to-peer and relays and all cryptocurrencies at least most that i'm aware of don't incentivize the relay of information in the network and also ownership so basically owning the tokens inside the system you can actually create and there are other axes that you can go down these are all candidates for resources that the system can have that allows you then to translate to that point number one of who makes block and really it's a balancing act you create your scales they they kind of have to balance a bit and you you want to make sure that no one part of this puzzle is overrepresented what you want to have is a situation where basically each constituency has a voice and they keep each other in check so for example if you're anti-plutocratic so you really really hate the ownership of the system the social components can be biased towards the poor of the system so that means that your your ownership your pluto and your social are balanced with each other so the rich represented but then the the poor are represented if you have a lot of computational resources but your constraint on storage and network you could certainly run applications but then you can't transmit them and they're all going to be stored off chain so maybe you have to keep those three things in balance with each other it's a really cool problem to study and it's something that we've been thinking a lot about and really the first step in solving this problem is going from one to two that's that's the really hard problem and then two to n is a lot easier assuming you have good protocols for each of these categories so we actually are thinking about a proof-of-stake proof-of-work style system and we're examining how one would do that in practice and we'll be publishing paper later this year specifically on that but then our goal is to go from two to n and then once we know how to do it for arbitrary things it's just a question of how do these protocols compose with each other so there's something called guc in cryptography general universal global universal composition and it's a way of showing that that when you add these things the proofs don't break the system will stay in balance and this is going to be a big research agenda for cardano 2025.