Aligning economic incentives for blockchain-based knowledge tokens / Claudio J. Tessone

Transcript

foreign [Music] I will review a little bit of the topics that um a a we have been researching over the years with respect to the ethereum ecosystem and the reason is the following it in my team we are performing research on three main areas of acknowledge in the area of blockchain on the one side we are very interested in consensus scaling of consensus but specifically on how which are the limits that consensus has for example when you have a networks

that are growing and you need to diffuse data in terms of blocks and you need that this data is reaching all the nodes that are present in the system and sorry that all the nodes receive the blocks in that are being broadcasted you have finite time and you have a the diffusion is taking longer the more data this nodes are the more data the blockchain is storing this introduces some physical constraints to the quality of consensus in blockchain-based systems but also it enables the possibility for strategic different sorry different kinds of

strategic Behavior by the participants in the system and we are very interested on how this a abnormal behavior away from the rules implicit rules that the protocols have affect the quality of consensus or affect the way in which rewards are distributed reward distribution is not something that is very interesting to us we are very interested in understanding how um the way in which rewards are distributed eventually affect large-scale properties of the for of the economy of the of these systems for example the level of concentration the level of economic concentration of a welfare inequality that is present in

these differences and we'll talk a little bit about this in a second crypto economics is very interesting you can lead not only at the layer one but also looking at what happens at the level of token economies so crypto economics is um Anto economics is something that we are very interested into as well and we do not do this in a traditional approach that economists do it's not that we are just doing formal modeling of this mathematical modeling but we are also very interested and we perform a lot of larger scale blockchain analytics it's exactly in contrast to what I explained was explained before most of the blockchain-based systems still allow us to analyze the transactional data at all

the possible levels and what we use this data for is to understand economic mechanisms that drive emergent properties and as you see in this first slide what what is very important for you to remember is that when you look at the global properties of the system you see that there are some properties that naturally emerge you see that economic inequality is just something that emerges because of the economic interactions between the participants and this inequality or this concentration that that happens for example at the level of consensus a power in a participation in the consensus

is something that is not enforced top down it is something that just naturally appears because of the economics that are the economic incentives that are placed in the system and because of the bilateral interactions of people that is transacting value so all these properties naturally emerge and are not imposed top down by System designers and what is important as well to remember is that once you have properties that emerge these properties are robust against changes okay so you need to very think carefully on how the systems are designed at the

microeconomic level in order to get are more aching to those that you may want to see inside of the systems and I will talk a little bit about some of the missteps that different layer 1 systems ethereum as well in a way have taken when designing the incentives as a a as a blockchain based economy okay of the main reasons that is that to me has been always fascinating and that eventually determines at a very large scale what happens in with respect to power accumulation in blockchain-based systems is thinking that there are decisions in which the

a location for example the supply of new assets of ether for most of the time in which ethereum was a running was a deterministic issuing of new um of new ether and this is not something that is allocated by the market it is not that the new ethers were created historically for most of the time based on how much demand exists for the assets or how much people is using them they was just a design Choice done at the beginning of the creation of ether and of ethelium and you see here you have the total Supply as a percentage of the total Supply that was already generated in completely different

blockchains you have a lot of utxo base two and all of them Forks of Bitcoin and you have there in a yellow the total supply of ether you know that most of the ethers were created at day one when they were in whether with the Genesis block they were relocated to the ethereum foundation but you see that here in contrast to what happens in real world economies always the growth of the supply is just a deterministic function and it is most strikingly is a linear function okay and when you look at for example how many accounts are created in most of the a successful blockchain based systems

some of those are not so successful in the in this plot one thing that is interesting once again you have ethereum in yellow is that you see that the y-axis is in logarithmic scale the total number of participants in the system the total number of accounts created in the system grows exponentially so here you have a a huge failure when thinking of the design of the system because you have an exponential growth in the demand in principle and you have just a linear growth in the total number of assets that are taking in a taking part of this and this created not only in ethereum but in a most of the blockchain based

systems and increasing centralization at the level of a the the incentives that you have in order to to participate in the in the in the consensus when you look into how unequal is the distribution of wealth in all these systems once again you have that it saturates the genie index actually is not very good at measuring wealth inequality in blockchain-based systems because it very rapidly saturates to one uh when you have a very a lot of um an equal inequality in the world's distribution and while there has been a transition one

thing that we were analyzing as well sorry one thing that we were asking as well is what happens at the level of the fairness and the reward distribution in different delegated proof of stake based systems and interestingly you can measure it in different ways but interestingly interestingly what you see is that you have different choices and ethereum took one which is that the total number of validators that are active in one of the eras is fixed you have different degrees of freedoms of when you have delegated proof of stake you may decide just to allow for an um open-ended amount of um

a validators that may be active in an era or you may limit like it happens for example in polkadot in this case Casper and as well in ethereum other systems like cardano and tesos for example allow for an unlimited number of participants to be active inside of the of each one of the cycles and what we identify however what we identify is that a with respect to the concentration actually the design choice that ethereum took polkado took and Casper took that is limiting the amount of active validators in a way is much more beneficial to the wealth redistribution

to the reward distribution to make it more egalitarian in the long run and it is very interesting because you may think that the way in which a proof of stake based system may only affect locally the or instantaneously the level of the rewards however it has a very long-term impact in the um in the distribution of economic importance of the actors that are participating of the validation process another thing that we did research on for the past years was okay we have been looking at this level of centralization that happens in different blockchain

based systems it happens at the level of economic concentration it happens at Swell at the level of the validators that are participating of the a verification process inside of this blockchains and actually what happens is that I will not talk a lot about this but we have been analyzing as well for the past years different token economies and on the one side you have a positive correlation on the pricing of tokens and the amount of activities that they have inside of the blockchain the more tokens are transacted the more is the value that they end up having in the system actually it is a loaded allows us to recall a little bit what happens with

the medical flow that you know that the larger the network is in terms of number of participants the more the participants assign value to the token and this is exactly what we observe as well in the evaluation of tokens with respect to market capitalization however what you see in the left panel in these two plots is that actually the level of centralization that happens in ethereum tokens is incredibly large there are very few a account in general smart contracts dexis or a centralized exchanges that if you remove them from the system they the the network of transactions completely

collapses so most of the tokens still it means a level of concentration that is that of the speculative assets okay because they are just traded in the market and not circulating uh organically inside of the economy we have been analyzing layer one we have analyzing token economies and there is a very interesting project that we have been looking into that is what happens with some dexes with decentralized exchanges because you know that in a centralized exchange all the transactions take place off chain actually they only happen logically of chain and the actual

exchange of value of one token for the other is never realized inside of the blockchain it only appears when so one of the users wants to deposit or withdraw tokens from the specific Exchange and so I will talk a little bit about a platform that all of us know it is uni swap and when we are thinking of uni swap there is something that happens that is the creation of a liquidity pool is once again not something that the system designer like in a centralized exchange is creating liquidity pools are something that any user can do if they can provide liquidity for the trading

pair so the first thing that you may say is okay the all the amounts that people is putting into in each one of the liquidity pools the amount of values that they are putting all of them are completely independent so perhaps we will not see we will see just a some noisy pattern when we think of analyzing the whole structure of uni swap so we analyze the the structure of Unison version 2. and the question that we asked is are there general laws that emerge out of this in principle and coordinated action of all the participants inside of this uh this decentralized exchange and what we identify here what we analyzed

is by the time you're going to analyze it at the beginning of this year they were something like 40 000 a different liquidity pairs in in uniswap and in the y-axis here once again in logarithmic scale you have the total value locked in US dollars in each one of these trading pairs and in the x-axis what you have is the decrease in a decreasing order the largest liquidity Bulls starting from the largest liquidity pull to the smallest one and what you see here is once again the x-axis is in logarithmic scale and you see that this is a line

it is a line because in statistics this is called a power law distribution actually this is a zip a law actually also signaled by the exponent which is minus one and this is a typical trait that a self-organized systems have so even if people is independently pulling and adding and removing a a liquidity buying and selling tokens inside of each one of these um a liquidity pools all of them have been a very clear pattern of self-organization you can build

a network representation that is very important to understand how systemic are specific tokens in the following way in this network that you see there you each one of the nodes is a token and there is a link between two tokens if there is a liquidity pool between them and what you see it is very apparent from this representation is that there are a few tokens that are completely systemic and if they fail all the system of a uni swap May collapse and this is very important because it might be this is the original version of the of unisoft a liquidity pool Network and this is how

it evolved over time until it mid last year but you see that they are still heavily centralized okay it grew a lot they have many many tokens traded inside of it inside of uni swap however you have many of them are in transacted against the same tokens and what you see is that there are a communities from a network point of view there are communities centered around the specific tokens in which all there is a creation of a lot of trading pairs with respect to other ones and what you see is that there are very few tokens that are completely systemic for the functioning of unisoft obviously one the first one

is a wrapped ether then die usdt Etc this is in the last snapshot you have a usdt and usdc as they say as the most important ones in the system actually if there is this plot there of the jilon is because this is actually one top ten and it is systemic inside of uni swap it is traded it has a lot of value locked and it has a lot of trading pairs with others okay there is a very large level of inequality in the size of the tokens Etc but the most important thing that I wanted to mention with respect to this plot is that a actually the distribution of a decisis of all the

liquidity pools of of all this network the importance of all the the liquidity pools in this network has stabilized for almost two years so actually there is a a remember what I said in the in my first slide there is an emergent property in the structure of this a network of liquidity pools that is stable even if it grows it preserves still the same shape and what we are investigating now I will not go into much detail with respect to this but what we're identifying is actually that um if you look into rug pulls that have happened inside of this system you see that also they preserve they have been

affecting completely different different sections of the network at all the possible scales resembling very much the distribution of liquidity pools and provided that we have had rag pull so for all the possible sizes it is a very important a point with respect to analyzing how there can be a Cascade of failures induced by this uh by this activity ah the final thing that I would like to talk in the last five minutes that we have been investigating the doing research on is we have gone all the way from layer one to a tokens later to what

happens in the centralized exchanges and one thing that is very important to have a functioning economy is endogenous circulation of cryptocurrency such that even if you have a limited Supply it is enough to satisfy all the transactions then you do not have speculation in Rising prices Etc and really what happens in ethereum if you want to see a a its economic activity so there is a concept that macroeconomists have been using for more than a centuries that of the velocity of money and in general macroeconomics Resort in

order to use it in a a so-called feature equation that provides a actually if you are thinking of this because in a real world economy you cannot measure the way in which the money circulates the only thing that you can do is estimate the velocity based on other parameters assuming that the um official equation is correct okay this is what you traditionally do or people that is analyzing a real world Economist does do what we have been doing for a couple of years already we started analyzing you to excel based blockchains and now we also are finishing a paper in which we analyze this concept for

ethereum is actually measuring the velocity of money inside of the ethereum economy but not only that and what I like the most about this methodology is that we are not just looking at the macroscopic level of the velocity of money but we were able to derive how much each one of the users of ethereum is contributing to the global velocity and understanding who is actually contributing and how unequal is the contribution to the global velocity in a nutshell what we are doing is if you have that Alice received two a times ether one in the block number 10

and another time in the block number 20 she received 15 ether at a in the block 10 in the block 20. what we do is basically and in the block 30 she sends to Bob the 15 ether what has happened so far is that there were a 10 ether that stayed Stood Still for 10 blocks and five ether that were um stood there in the wallet of Alice for 20 blocks so if you take the weighted average of these two things basically

you know which is the velocity at which Alice's funds were moving in this period of time okay and exactly because it is the approach that is taking initially in economics what we assume is that the last money that you received is the first one that you are spending okay because all the others are saving you save money and then when you earn new money what you are doing is suspending the uh the last part that you received so this is why we adopted this lifo approach but so what we did is actually measuring the the first of all the global velocity and what you see is that there was a

spike in the total velocity in a ethereum triggered exactly by the point in time of 2817 which was the Ico craze okay and interestingly on the other axis you see in a green and labeled in the in the bright part of the of this plot you see in green the total number of active users and you see that it is only until 2018 that the total number of active users in ethereum has been growing exponentially once again you see that the scale is logarithmic so after that the total number of users has basically plateaued and there is a lot of reasons for that gas prices have grown a lot

is one of the reasons on why this thing has happened [Applause] but there was no so much economic activity anymore within the ethereum blockchain as if there was it was happening inside of 2017-20 in that period what is interesting is that we have the five first a most important a nodes users contributing to the total velocity in ethereum this is now the individual contributions before we looked into the global scale and what we see here is that roughly 30 percent of all the total velocity is only originated in five different

accounts okay when you look at look at them first looking at it from the global point of view understanding which is the level of a global of a wealth that they have times the velocity of what they contribute what you see is that a family of exchanges contributes is the most one is the largest one then after you pass all the this is the whole set of an unidentified addresses all the addresses belonging to binance bit Phoenix Kraken Etc are the top five of the users this is if you consider the wealth if

you do not consider the wealth what you see is that the main activity that triggers the circulation of money inside of ethereum are still exchanges plus the process of um a verification here it is written as mining but it is all the process of reward distributions and then you have Celsius networks Nest protocol Oracle Etc foreign another thing that you see is actually that there is a very mild correlation almost zero between the monetary the the wealth of the users and the velocity of that they produce is basically flat the process and what is more most interesting to me

and perhaps this is the time in which I can finish is we can look at the distribution of how much is the micro velocity that each one of the nodes users is contributing to ethereum and once again here you have a log log plot and you once again here you have a straight line so it is a power law distribution in the velocities and strikingly and we can concentrate first on the right part of this plot what you see is that there is a stable distribution that we are trying to understand why this is happening there is a stable distribution in the contribution to money velocity of all the users that is stable with the power

of minus two but remember what I said before because it is an account based systems we are with are not sure exactly which are the assets that each one of the users is spending when they are spending money in a utxo based systems you can certainly do that but here we have to make the Assumption of the lifo a last input first output if you consider first input first output or if you consider it just a randomized version of any of the prior transactions you still see that the distribution of well of velocity is still the same so this is an emergent property that is a stable and that characterizes the contribution of all

the users powerless Solutions are important because it means that you have a handful of very active users and you have a very large amount of users who do not contribute at all to the to the velocity of money okay in a way I would like to conclude this presentation that was very long trying to say that we have to remember that all the systems when we are thinking of this as an economic problem we have to remember that the all these systems are very far from equilibrium as we have seen number of users changes exponentially in numbers so you cannot make models in which you are trying to come up with pricing Etc

a need to take into account not only the fact that the system is out of equilibrium that its parameters are constantly changing but also that the complexity of human behavior is an additional layer of as an additional factor that must be taken into account to better understand them thank you very much for the patience people very very interesting very in-depth I know you're going to be around later on so anybody who has an in-depth question will be able to ask you those questions then this report that you just mentioned will be available soon we have the paper on ethereum velocity

out in a couple of weeks yes okay okay so very good timing thank you very very good thank you very much thank you [Music]