Tellor Layer: Decentralized Oracles for the Next Era - Brenda Loya | Tellor

oh yeah I'll figure this out well thank you for coming to my talk it's the last one of the day and I know there's a closing party so very much appreciated that you came and made it my name is frienda I'm I'm the CEO and one of the co-founders of teller and my background is in economics and so and we started teller the very first iteration of it we did in a hackathon in the summer summer of 2018 when things were a lot more gray um the government wasn't so uh against crypto back then and we actually the very first iteration of it we created in DC in a hackathon in DC and back then the main narrative was like there was GNA somebody was going to create this new chain that was going to be the eth Killer and it's gonna just take over the market and it was gonna just be big one big player and obviously that did not play out that way but what it did do it created a lot of competition so in the process of people trying to do this we ended up scaling in a very different way that we thought we would we now have different ecosystems you know we have EVMS we have Cosmos we have polka dot so we sort of scaled that way and then with all of these ecosystem pop systems popping up then we also o have a lot of protocols thinking well you know what why am I going to pay transaction cost is another chain I'm just going to launch my own chain and so now we are looking at a very different picture than in 2018 so as an oracle you know as an Oracle and honestly if you're running your own company you have to stay flexible you have to look at where the market is going and be able to evolve so we looked at this and we said okay what do we need to do to be able to support this multi-chain world with the parameters and that we live in for teller specifically and we obviously want to be able to provide faster data than we currently do we're this very decentralized so we've always been a little bit slower than centralized oracles where we currently work more as an optimistic Oracle so the longer that that data is on chain and it's not disputed because disputes happen on chain um then the more likely or it is that the data is good and so then our users can use it so we always or we usually tell them wait about 15 minutes if you want it any faster run a monitor and then you can use it a lot faster than that and then obviously we want to be able to produce or to support those companies that have these long longtail Niche data you know maybe new tokens that are not as liquid or that other um oracles may not be able to support obviously with some guardrails around them we usually work a lot with these teams to come up with robust definitions for what they're trying to do and making sure that either they can come up with something that makes it really expensive to manipulate the asset or that they have proper fallbacks in case something does happen um and then obvious I think as we you know as research improves and continues to grow for CK proofs I think a lot of the bridging is going to be solved by it to be honest um a lot of these proofs are going to provide uh you know part of the services that we as oracles provide so we have that really in the back of our minds and so it's not going to be just good enough to provide data you're going to have to provide really good data so you're going to have to become not just a data provider but a data Creator and then at the top of our list something that we are not willing to negotiate on is decentralization so this is how we want to evolve and with that in mind we went ahead and developed um our our own app chain you know what a surprise everybody does that nowadays but um so we're calling tell layer uh eventually it would just be teller but within it um you know you have to when when you're creating your own chain um I did not realize in 2018 that I was going to be doing this in 2023 2024 but now you're looking at different problems of security there's the security of the chain and then the security of of the actual data that you're trying to provide your users so what we came up with is we're going to have two different sets of um of of not users of um of supports supporters or two different sets of base level security um players so it's the validators they support the consensus and then the reporters the validators we needed to be we want to be interoperable because we want to support all these chains and so for you to be really fast then you can't have thousands and thousands and thousands of validators unfortunately so that created a problem where we needed to put a cap on that but to provide data and to be able to support those Niche um requests from you know from people we actually need the reporter set to be really really big that one we want it to be as big as you know as as they can stake or as much as they can stake so as many reporters as possible so that created sort of like a conflicting goal in a way so that's what we came up with these two different groups that both are going to be staked and um to maximize the efficiency of their Capital we're allowing dual delegation so you can be a validator and a reporter we split the rewards 25% go to the validator 75% go to the data reporters because we want to make sure that we incentivize more people to provide data and if you're a validator you should probably want to be a reporter we want to push you that way so we're using incentives very purposefully to be able to to to drive people to do what we want them to do um disputes happen on chain so just like in the in normal teller oh I forgot to tell you how it works so teller actually works by anybody can become a data reporter if you stake on chain and anybody can come and dispute the validity of your data for starting at 10% of your stake and if they're correct they take the whole stake from the reporter and so this creates you know checks and balances um based on incentives and it happens all on chain yeah and then it goes to a vote um of the community that is divided equally among four groups of people the users the data reporters the uh token holders and the team the users and the data reporters earn voting Power by the users by tipping for data and the data reporters by providing data on chain and then if you hold the token then you're you whether you're a reporter or a user then you are part of that other 25% and then the team acts as a referee so all the disputes continue to be on chain on layer and that is because that's basically our security system it's how much you stake how much it's basically how it's a function of um how expensive it is to try and take over the chain or try and um grief the system for a while for a specific uh for for data so that is our security system the obviously um the main the main goal of creating the new chain though was to support all of these new chains in a secure way in a much faster way so to to to accomplish that we implemented you know the the whole differences between the validators and and and reporters but another thing that we ran into immediately is how are we going to bridge the data and we decided to go with um a cosmos with a to use a cosmos SDK just because it's just more modular and we thought you know it's been tested also we didn't decide to do a rollup because of the oh God we just talked about this uh earlier but it just seemed too centralized for us in terms of um the anyway it it seemed too centralized for us so we decided to go with the cosmos SDK but immediately we run into the problem of like how are we going to reach our data so we reach consensus on our chain and how can we actually serve it to all of these other chains so we're using something really really new in in Cosmos s Decay which is a vot extension so basically within you don't get you don't touch the consensus mechanism you just add another vote for an additional thing that you want to do and in our case that additional thing is just attesting to the data that is going to be relay to other chains and this is brand new we are the guine Pigs we're not generally we try generally not to be that but in for our purposes this works perfectly for how we want to use our SDK and then that immediately goes into the security issue of like well how do you update your validator set and how often and what if you know all the attack Victors that come with it um so the way that we approach that is we're going to make our Bridge very very slow we have a to our token lives on ethereum we don't plan to deploy another token and so and you know we're listed on sex on centralized exchanges so we're not planning on really doing anything with that so that puts us into this uh sort of box where we because our token lips on ethereum we have to bridge it to our chain so that we can uh make that work and one of the ways that we plan on mitigating the risk of somebody just coming over and taking over the validator set and taking over the chain is by only allowing 5% % a 5% change in the validator weight every 12 hours and by doing that you can also monitor the bridge to see how much uh of our token is being bridged over to our chain and if there is an attack or we think that there's an attack coming then we can actually start doing something about it but it will take a really long time at 5% every 12 hours to actually take over 6 66% of the network so that's something that you know that that we implemented specifically for our purposes and other other chains have done that where they just do it really slowly so we full confident that this will work um also the one thing that creating your own chain really affords you it's the ultimate back stop of back stops which is forking your own chain if something were to happen so you have your social consensus behind it you're just like you know what we just don't like it so or even if you took over the chain just the threat that we can actually erase what you did should be enough to dissuade you from attacking the chain but if it isn't we can't do it and we will do it um or I think the community would go with it for sure um another thing that we wanted to accomplish with this are is um provide what we describe as robust data be able to provide robust data some subjective n data more Niche data robust data we call that the data that is supported by 66 at least 66% of the validator weight or validator SL reporter weight and if everybody's submitting this data the way that it works is like if you request the price of eth um all the the reporters can provide the price of so you get a 100 prices of eth and then the median becomes your eth but then you have all of the weight from all of those reporters backing that supporting that um that data feed so your security is higher and that's what we call robust right and and although it is robust and supported by the network we still caution people to take your time to use the value you know because most of the time it's going to be fine it's only the edge case but the edge case is the one that matters if the chain was to take be taken over or something was to happen you are going to liquidate things immediately on something that she shouldn't have and when you're dealing a cross Chains It's just very difficult to to create this dance that um you know where there is um that everything works as expected so we always tell and we will be telling our users eventually when when we go live with this is still wait you know make sure you have those fallbacks make sure that you always assume the Oracle is going to fail and that you are able to handle that on your on your side um but it is more robust data it has more security behind it subjective data and those Niche data or like random information that people need we want to be able to support that that is what is going to propel I think the the industry to to Really experiment and figure out where we fit in the world and I I think that it's incredibly it's important for new companies to have this Avenue and so that type of data is very optimistic it's you know it's like um if only you support it or two or three of us support it then um it's still going to be on chain but what we do is with every piece of data whether it's robust or subjective or Niche or whatever you want to call it um we provide a confidence level to the user of how much weight was uh behind that reported information and that way they can determine within their side whether or not to use it or how quickly to use it whether they should wait a long time or not and it also allows them to evaluate the cost of attack so if if you know that consistently you have 66% of the reporter way behind it it's very easy to calculate that you multiply that by the price of our token and then if that price or if that number is a lot lower than the tvl that you're trying to support then you know that might be not a good idea but at least you know the risk and you can quantify it because I tell people if you don't take the time to quantify the cost to attack your network uh a hacker will that's their entire job that's how they make money like it's a business for them so so with that in mind um we want to be able to support all of this but we also want to be able to provide the tools so that users can Implement Us in in a responsible way and Specialty data so this is for I was going with the whole data creation for oracles this is sort of um it goes back to it's not just enough for us as oracles to be providing offchain data onchain but we actually have to become data creators and better data creators than we can find on the traditional system hopefully um so my our first um uh attempt at this is going to be to recreate the CPI on chain and basically if anybody doesn't know what the CPI is it's just the average change over time in prices for a basket of goods that you use and these basket of goods is generally um um the government chooses this and and then they provide us the data and we just use it blindly and allow them to use it blindly to create legislation that could directly affects us I don't know how many of you have um a a wage Clause a wage incre increase clause or in your in your in your um contracts um but that is very common uh especially governments generally have an adjustment for their employees based on inflation or retirees like you know pension how much how much you get on your pension today versus a year from now is going to depend on this number and the government can also print a bunch of money and that creates inflation and that's basically in a way they're just you know I mean it's not a robbery but it's kind of a theft right because they're just taking money away from you if you can't afford the same things you used to be able to afford and so what this has created is um a big conflict of interest and we don't talk about it enough but to give you some background is like it's really important it's used really widely it affects us really directly and the way that the CPI started at least at least in the US was because Republicans and Democrats were fighting over who was right about a tariff that they implemented in n in 1890 they wanted to know how it affected the population which at the time the population was considered white guys they lived in the city and they were married so if you were single if you were a single mom if you were not white if you were not a guy you were not part of the survey and for what is it almost 50 years that was a measure that they used to create all sorts of legislations that directly affected the entire population of the US took 50 years for them to actually start including in the survey Hispanics and blacks so okay 50 years for that and then and they purposefully under represented these populations then in in it took another 40 years to 1978 for them to actually rework the survey in a way that was that addressed some of these biases since then there has not been a major revamp it's almost like moving mountains to actually change this uh type of methodology and I can tell you that I used to work for the government I was uh part of the not the CPI the producer price index and the payroll number and it's there's so much research that goes into it that never comes out that never sees the light of day and that just you know nothing happens it's just very politicized and it's not often or it's really rarely acknowledged and so um there are other uh methodology issues with the CPI that people talk about all the time conflict of interest it's my main point and the only thing that I think crypto can actually imp improve upon and so some examples of this and you tell me when I have five minutes but it's Argentina okay Argentina uh from 2007 to 2015 they openly admitted to only reporting a third basically of their inflation why because they needed they needed some loans and if you have really high inflation you're not going to get you know you're going to get either a higher interest rate or you're not going to get a loan from the IMF or the World Bank and so they just reported that for a few years and went ahead and and and right now the inflation Argentina it's 300% something crazy like that um last year I think around this time it was in the 109 um not that if they had better measures they would have made better policies but definitely better policies cannot come from measures that are purposefully misaligned or you know misrepresent presented uh Zimbabwe has a a similar story but they basically their government just decided to use old methodology for their new legislation because that just felt better or you know supported it better Venezuela we still don't know what is going on there so um but anyway so antitrust laws protect us from monopolies and we rarely think of governments as monopolies right so but they provide us all of this wealth of data and information that we pay for by the way with taxes CU everybody that is in the bre of Labor Statistics it's paid by our taxes and we just take it for granted like it's obviously it's right there are some paid um some paid companies that try and recreate this but then that goes back to the same problem it's like if it's not transparent we cannot know whether or not these values are correct and there's obviously a conflict of interest and in defi I'm sure the government is going to go after you know a lot of these conflict of interest where like you're you're def and you're the Oracle and you know there can be they don't like that but here they're clearly that you know they're clearly the problem and I'm not saying that that all governments you know don't provide the best data that they can I think that most of them do but that can change very quickly depending on the political climate and so now we have technology and we have um that we can create this not as a challenge to the government I would say not at least not at the beginning more as a compliment right it's a complimentary piece of data that should move in line with what the government is creating and so the problem with that is that once it becomes challenging um or challenges their data we have to make sure that we actually use open decentralized networks behind it and that's where tellor comes in that's one of the reasons that you know I'm obviously I'm from teller but it's decentralized I can't stop teller and so if the government comes after it because it doesn't agree with the values that we're putting out and obviously the CPI would be a sort of a it's a common good and it would be um it's not just teller providing the data and the data would be completely open and transparent the methodology as well and then if you don't like my methodology go ahead and apply a different one that's fine um and if you want to make money off of it you can also not share your methodology you know but still you need that transparency of the data underlying data and the methodology to be able to say yes these reflects these numbers reflect what you say they should reflect um yeah so so so yeah so obviously we'll be using teller also this is a common good so I cannot afford to go and pay some huge fee to uh to to a team to allow me to or support this type of data uh teller is completely open anybody can come and request any type of data as long as you have a good specification their reporters feel comfortable to provide the data for so I would have a really good specification with the Argentina example um Alberto cavalo and Mario berotti they came up with really good um methodology to apply and and use it basically used a lot of scrape data and he came out with the conclusion that onchain data is highly highly correlated with offchain data and that was before 2020 so at least in the US that is going to be I I suspect still true or even close more closely correlated because we all changed the the way that we purchased things to onchain it's also censorship resistant which I need because this is a value that could be highly um challenged by the government eventually hopefully not but it could um and then obviously anybody can come and tip stake and dispute on Teller and because it's the common good that should be part of the community should be owned by the community eventually so you I need some protocol that has those qualities and then it should be able to provide flexible data which is what we'll be doing that is it so if you hav have any questions [Applause]