# "The Oracle Landscape in 2020" - Nick Fett

- Channel: [ETHGlobal](https://streameth.org/ethglobal)
- Date: 2021-10-09
- Duration: 26:45
- Topics: Ethereum, Tellor, Nick Fett, ETHOnline
- Watch: https://streameth.org/watch/yt-HF9IxG0-frw
- YouTube: https://www.youtube.com/watch?v=HF9IxG0-frw

## Description

Nick Fett of Tellor describes the use of oracles on Ethereum.

## Transcript

thanks josh yeah um all right it's gonna start sharing my screen uh can you give me a thumbs up or you guys can hear me okay everything's great looks good okay awesome um so as josh had said uh i'm nick vette and i am the cto of teller um so uh today we're gonna kind of talk about all of the different oracles in d5 um but to tell you guys a little bit about teller this talk is not about teller um but we are a decentralized oracle on ethereum so you know if you haven't heard about teller or you're looking to get data into your smart contracts definitely go give teller look teller.io we also are sponsoring the hackathon so if you guys are doing a hack and need some some pricing data into your your dap definitely go give it a look um so yeah now i'll just kind of get started on some oracles and in d5 in general so kind of before i get into what's currently being used by d5 let's kind of get into a love story between oracles and blockchains and then really just how it all went down um so even before ethereum so these glorified escrow contracts which we call smart contracts realized that they actually needed a way to access off chain information so early adopters of bitcoin and ethereum they you know that just playing around with these digital ledger entries is just only so much fun if you can only reference these kind of self-contained transactions and so the dream of having access to off-chain information began um you know they they started on with their the early visions were derivatives stable coins prediction markets and and basically uh you know gambling protocols that uh us in the crypto space all really like and you know like any good community of entrepreneurs would do people set out to build these things and it was great um and and it worked so you know we have derivatives stable coins prediction markets and all those things but now we're actually looking back and a lot of people are kind of examining those and they're wondering if this was kind of the dream all along is is this what people had had thought you know as as more money gets thrown at these protocols you know we have to look and say like hey well all along how long we needed the oracles in order to make this possible and we knew that we needed the oracle technology but do we actually have the oracle technology is it actually trustless like we had been originally promised so kind of stepping back from you know the the d5 in general like let's look at like what an oracle actually is uh so before there was any differentiation between on-chain versus off chain data all you had were these databases so the question a decade ago was very similar to the oracle problem how do you come to consensus on a piece of data in a decentralized way as we all know the the answer to this question was bitcoin and the blockchain so um you know regardless of what system you're running they all have all these layer ones have various ways of agreeing upon specified data so you know in bitcoin's case it's who's sending what to who and all of this is sort of built into the system you have a consensus layer that's that's built in and oracles are different so oracles are built on top of these layer ones you know they don't have a pre-specified consensus mechanism most of the time so you know you you don't really just have these sort of pre-specified off codes you need you need information on some event or state that's not at all related to your blockchain so the big difference i guess between oracle's and normal data is that now the consensus layer is unchanged so namely whenever i say consensus piece it's just the pieces that determine okay who can push the data and who determines whether or not this data is correct and and that that's what an oracle system is so uh moving okay so now that we sort of know the basis of what an oracle is let's kind of get into some of the designs um so to talk about like what's what's the most simple oracle design and this this would just be one person so i think we all know this one it's a centralized oracle so if you have uh if your dap needs an oracle you can just enter in the price yourself so you enter it in for those that don't know the biggest name for this method in the ethereum space is provable uh it's formerly called oracle eyes and it works really well you know it's obviously a completely centralized model but it's sort of very open very transparent it's easy to see who's in charge um the obvious risk so of centralization your oracle though is that a they can either manipulate your price or if they let's say they just shut down their servers for some reason you don't even get a price so you're sort of left hanging and that that latter problem is what we actually call a liveness guarantee so moving on to the second easiest design this would be a multi-sig so for this design you don't just white list one person you would like white list a bunch of people um and they all place data on chain and then you take the median or average or um some other transformation that you want to do and as as you can tell so this is this is a little bit better more like you know the v2 oracle um but you know to kind of get to the end here uh you know if this is the v2 oracle most of d5 right now is like between 1.5 and 2.5 you know we're really at the stage if we're just moving past the multi-sig wallet stage um so the next design so there's kind of more designs and a lot of designs are sort of built upon complicated multi-sig wallets um you know a slightly worse version i would say is you know you have multiple parties off chain and they all sign the prices and then you have people who who push on chain so you would have like you know say some exchange some exchange assign it and then a centralized party will will push these datas on the data on chain and um it sort of just separates who who can manipulate your prices versus who's in control of the liveness guarantee um so you know you may both have collusion and liveness issues and it may be separated which may help but it really just depends on the nuances of the design um and really you know a big question you have to ask yourself with these reputation-based systems so anything that uses like a multi-sig or known parties is is whether or not more parties that are known make something more decentralized or censorship resistant and that's really a big question that you have to think about in d5 in general so you know if you have goldman deutsche bank and bacon of america setting your price is that better than just jp morgan setting your price and um maybe i think i think you know it's it's not quite the home run that the bitcoin og's dreamt of but it's um some people will try and argue that yes this is decentralization but that's for you to decide um okay more designs quickly uh so the next one kind of like the you know if you assume like the that the multi-sig is like a poa system comparing them to layer ones you know you could you could also do like a delegated proof of stake oracle uh where you have them actually put some a stake in and then they they can sign that they can sign the data and they lose a stake if they lie um this is maybe slightly better you add a little bit of a crypto economic piece but usually this one you have a big problem with who the white lister is so who gets to say you know is it open for everyone to stay usually not um can the white lister just instantly take everyone out of the white list and put their own people on you know there's a lot of details that even if it is like a delegated proof of stake you really have to look into it or else it's probably just no better than a poa oracle um and kind of like the last one for design um this is sort of a positive note would be like a prediction market oracle um you know they're probably not as popular i think maybe they're making a comeback like uma uses a sort of version of this and there's i think you amazed right after me um so you guys can talk to talk about this talk to heart um but anyway so prediction market oracle's you basically bet on what the outcome's going to be um and as long as you know people want to make money and 51 of your system is honest you'll get the right answer um so and these guys were great so this was some of the original oracle designs so the auger and gnosis um they kind of wrote a lot of literature in the space about kind of cryptoeconomic security of oracles and um i i like it a lot um unfortunately so the big problems with this is sort of speed one it can take a while um and unfortunately uh i know we don't care about it a lot but regulatory concerns uh so prediction markets are kind of illegal here in the u.s um and then just liquidity and cost concerns as far as the cost to actually get that out so they've sort of become kind of untenable for a lot of the defies needs um but yeah you guys can you guys can ask hart about that next um but anyway uh that's kind of what's currently in use in d5 at the moment so you know they obviously people are designing new things and lots of people are raising money for oracle so um i think we're actually going to start seeing users starting to demand some better options here um similar to how you saw you know first with crypto centralized exchanges really took off or you know stable coins you saw tether really first just gobble up everything and um it takes some time before people actually start to care about the points of centralization but i think you know you have you had a little bit of an oracle bubble this year and people should start realizing that this is a thing so but you know despite all of these centralized designs um you know we've always sort of known how to decentralize something um you know anytime you're trying to crypto economically validate data we we do this with bitcoin and every other layer one so you have some way to pick some anonymous group of to enter in your information so some anonymous person or a group of persons and and then you add in some sort of maybe like some probabilistic finality so that way the community in general can collectively agree upon whether or not to accept this answer um as long as you sort of make it economically expensive to pretend to be the majority it's it's sort of good to go um you know that that's how all of the blockchains do it and um first that that's kind of how teller built our system and and i think others are starting to see the same thing you know but but there are drawbacks whenever you do this so you know like any actually decentralized layer one chain uh if you want to sort of make an oracle that's secure and decentralized um it's not going to be very fast it's going to be slow so you know the scalability trilemma holds for oracles and and sort of securing data as well and anyone who sort of tells you differently you know they probably have a coin to sell you so you know we sort of like we sort of get it when it comes to layer one so you know if somebody comes up to you and says that they've solved the scalability trilemma and it's fully fast and decentralized you sort of laugh at them but then you know projects will try and build you know something and they're like well we need an oracle that updates every block and it's just like well you're trying to validate data at the same speed as ethereum it's going to be really really hard um so now that we've sort of looked at the models and hopefully you guys get the framework um let's look to kind of see what's in use in d5 and um you know like i was going to go over to be honest there's not a whole lot of projects in d5 that actually use oracles that have volume so you know despite um all of the oracle buzz and all of the partnerships you may see there you know people with like real volume up there it's like you know like 20 20 20 projects maybe um you know like if you look at real projects with volume on ethereum it's it's not uh it's an amount that you you can sort of go over them all so i'm just gonna go over some of the big ones here and then you guys can sort of take their designs or some of these designs and do your test on the smaller ones on your own okay uh so first on the list of course uh probably the biggest d5 project over the past few years it's maker um so depending on how you measure it and how you measure it's its size in d5 you know it's dyes dye is huge it's and it's a great project so dye is a stable coin for those that don't know and on their oracle um they have an article for a lot of different pieces for the multi-collateral dye mainly um they used to just be eth us dollar and other projects use their oracle as well so notably like dydx used to use them as as well so just keep that in mind but as far as what makers oracle actually is uh it's kind of a complex proof of authority model uh with a little bit of secrecy so they have a model where they have a bunch of reporters who who sign prices uh you don't know who these reporters are and then then people can relay these prices into a medianizer and then it goes into their smart contracts um so actually the the only reason so is this it's it's a relatively simple model one that that sort of goes with some of the the design pieces that i showed the only reason i wouldn't give them like a complete failing design is actually the thing at the end there their oracle security module um so this is a smart contract where they actually have an hour-long delay to any oracle updates and this is a phenomenal thing so you know kind of like i was saying you don't want to expect your oracle to be fast and final they don't which is great um so the token holders can sort of vote to stop anything that that's bad that comes through their oracle so even though they have say maybe a more a slightly more centralized oracle like the token holders and the dc hopefully the decentralized voting aspect of it can can shut it all down which is good um so you know if you're using maker or or die or one of these things this is the oracle that you're relying on um and then hopefully you guys are paying attention to it and you can sort of shut it down so i know they did have some some issues march 12th so some some of the issues may come with maybe like the liveness guarantee because like so you have these oracle reporters who are supposed to push the price to chain um well if if the price go if the collateral goes down by too much well the problem actually happens is now you're actually going to kick off a bunch of liquid cdp liquidations so a bunch of contracts are going to get closed out um there's not a whole lot of incentives that these guys have supposed to chain um they might not want to pay really high gas prices to push this through because it's going to cause their whole system a whole bunch of grief um and that's kind of a bad thing um so you know i think you know maybe somebody who's like trying to do an um maybe if one of them's like trying to win at the auctions or something but it's you definitely want to align the incentives correctly um but overall it's pretty good you know i like the osm um next sort of very very similar design um would be compound uh compound has it's a very very similar system to makers um except they have their own reporters so their open oracle system kind of uh from last time i i checked they i believe that their vision is is to have a list of um kind of known parties whether it's you know coinbase's oracle for instance i know you guys just had to talk from coinbase so coinbase would sign a price and binance would sign a price and kraken would sign a price and all of these different you know what very well trusted parties would sign a price and then you have relayers who push it on chain and um similar to maker they also have a piece in there as well that they don't expect it to be super fast and final um so yeah that's that's kind of the compound funnel um sort of a very similar structure to makers and now kind of for the last the last d5 projects with a bit of volume um you know you have synthetics ave bzx set um these guys all use chain link so um so i know chain link has a billion partners but as far as d5 with you know volume over the past you know more than the last few months uh this is kind of the a-list i'd say um as a disclaimer though this is kind of my best knowledge how chain-link currently works um it's sort of hard to figure out and i run an oracle in the space um so that's probably a negative but chain link at the moment they have a centralized weight listing agency um and they get various nodes to report price feeds um so they have a bunch of stuff on the roadmap but what's live now is sort of a centralized reputation of chain link based oracle um they have a bunch of known parties so that is it um [Music] you know there's basically just a few structures that people are using uh for oracle's at the moment um and you know i need also just as i kind of went off through this you know just because you actually have a good method even even if you did have a perfect oracle just because you have a great method for getting data on jane doesn't mean that the data is even a good fit so that's actually a different issue um because like you know oracles and following some of the best practices regarding oracle's are like half the battle the other half is like picking what data you should choose so you know the number of derivatives projects out there who use like some last traded price on an exchange or some very liquid auction for their settlement prices you know it's it's kind of ridiculous and embarrassing here in the d5 space because you know traditional finance has sort of written a whole lot of literature on what makes a robust settlement price and you know we don't really follow it too well um so hopefully that can change but that's kind of for a different talk anyway uh now on to you know i'll just kind of wrap up and then we can have some questions but like why it actually matters um so i'm sure you guys have heard it like a million times so we as a space are kind of having an identity crisis you know like what is crypto built for what is ethereum built for and what's the purpose of d5 you know are we making back ends slightly more efficient or you know are we creating something that's actually resistant to the current power structures so you know the big problem with using centralized oracles is that there there's a choke point there's there's a squeeze that regulators can go after or you know people can do a rug pull and all this money's at risk um you know i don't mind if one group you know voting on things isn't decentralized and you know token distribution isn't quite there like that's an issue for governance and it's sort of more transparent and i believe we are talking about it but you know the problem is like whenever people are saying that oracles are decentralized and there's actually just a handful of people that control the majority of money in d5 and that that's a big problem you know you'll hear a lot of arguments that we rely on on reputation um and the fact that like some of these people wouldn't want to ruin their reputation but you know this is the same argument you could make for letting a consortium of banks like set your settlement price you know we we want to move away from this in d5 and hopefully we start doing that so you know i just lastly just for like the solution you know i think the solution's going to come from a lot of the projects so you know the current designs of oracles are centralized for a large part because they're sort of shoehorned into products that don't work with decentralized ones um you know we're we're all into the crypto space and we we're all in the crypto space we know how to decentralize something so you know if your buddy has a finance application and it has a centralized server and he's wanted to look like what it would he wanted to know what it would look like on ethereum so what would you tell him so first you know you would you would probably go to him you would tell him that he really really needs to think about whether or not he wants to do this you would explain to him things like probabilistic finalities slow block times and congestion issues and you know if he still thought it made sense uh you tell him that his users might have to wait a bit for confirmations that they're gonna have to pay this thing called gas to do anything with your application and mainly this just wouldn't be a sales pitch you know it's not like a plug-and-play server there would be edge cases he'd need to handle new procedures it would not be like this quick hack that he could just oh use ethereum and you know he'd probably come back to you and say something well you know like well what if we like progressively decentralize or we'll just like move all of the you know price reads onto ethereum or something and you tell them that you know you're only as strong or as decentralized as your weakest point it probably wouldn't matter it would be basically for a press release and i guess like why do why do we build on top of oracle's differently you know like right now like you're it wouldn't make any sense like if you need sort of instant finality in your oracles it's just not going to be good so um right now you know a lot of these oracle products and a lot of these kind of d5 projects they just kind of add lots of complexity in order to hide some of the decentralization to hide the centralization components of it so you know if if you're trying to do something that's really really fast or you're trying to do something that you know you're just trying to copy a centralized server and put it over here because and you need an oracle that does it you're just it's not going to work so anyway um solution supply demand based we need to supply and demand better um we need to just be more transparent about it um and that's what we're kind of trying to do at teller you know d5 in general tries to just build financial products that are just building financial projects products just for the sake of it um you know if we keep this up we're probably just going to get regulated out of existence or or fail to scale because we aren't really going to have any meaningful differentiation so it's it's okay to to build things in an actual decentralized way and um yeah thanks everyone for listening um to end the talk though if you're building a project a new oracle give teller a call uh josh thanks for listening great thanks nick and if you've got a couple minutes we can do a couple of questions here um so one thing so um you know obviously we're very early uh in kind of the oracle space and you know what this looks like in a couple years and the more sophisticated markets market is still kind of unclear do you expect there will be a lot of specialization um of oracle services you know by data type or maybe by you know what kind of security guarantees a project might need for an article or you do expect that this is you know uh kind of a situation where there's a lot of advantages to one protocol being used um for oracle's so how do you see that breaking down in the future um yeah i know i definitely think there will be so you know similar to like um maybe like layer ones kind of relating it back there where you know if if you sort of streamline what data you're providing you can you can make things faster you know people can come to consensus faster if it's not this sort of generalized prediction market type oracle where your people have to manually go and check and answer everything you know that that's like teller what we do is we we only do numerical answers and you sort of have to specify it off chain beforehand so we know what you're talking about and then we can be slightly faster than uh everybody has to go figure out what api to look at and and where to get these pricing information so i definitely think it'll be much more specified oracles and and i think also you know there is some space for maybe a centralized oracle or a decentralized oracle and different can guarantees along the spectrum so yeah there's a world with multiple different oracles you just want to sort of be transparent about it and um you know you you want to everybody's going to have to compromise somewhere so you want to be sort of transparent about where you're making your compromises um okay and uh i guess one thing i'm i'm curious about the question from me is you know you mentioned that you know we need to demand and supply better um what do you think is you know going to be the thing that you know pushes the ethereum space the d5 space um to you know demand more to centralize oracle solutions like what is going to be the event or what is going to be the trend that you know causes the ecosystem kind of upgrade itself in the way you described unfortunately it's always like a hack or a regulatory event as you know you saw with like centralized exchanges like when do people want to move towards decentralized exchanges more and it's like usually you know the reason decentralized exchanges in a lot of ways have have had the success they've had is because um the regulators have cracked down and forced all the centralized exchanges to do kycaml on all of their customers um and so people want to be able to trade without doing that um and that's why dexes have sort of succeeded in in my mind and i think you'll see a similar thing with defy as you know if regulators start cracking down on centralized oracles or pieces of d5 where they're setting their own price um you'll probably see people start moving more towards decentralized oracles great okay well thanks nick really enjoyed your talk thanks for joining us