Ocean Protocol

Transcript

[Music] uh hi everyone um it's really great to be here it's pretty awesome to see I first talked about token engineering on stage here at HCC five years ago and it's just incredible to see how it's grown and I have to their credit goes to so many people and Angela and all the tireless work she's done and the whole tea community and Patty and Griff and so many other people so really really kudos to all of you who helped to make token engineering what it is today and we're still at the beginning right token engineering is well on its path to becoming an engineering discipline and

it's things like this event today that are really helping to drive it forward as well as all the awesome online stuff so really it's my pleasure to be here today to talk about in this case I'm talking twice as Angela mentioned this talk is um a story a use case a a case study story of a token in this case the ocean token and the evolution of the ocean token in the context of the ocean project which we've been working on since late 2016 early 2017. so about six years and um I'm gonna you know talk about the warts and the ways that we had to address them and iterate through um and it's a lot more than I could cover in 45 minutes so I'm gonna have to condense some things but um happy to

share the journey and the learnings from along the way so with this a quick overview of what I'll talk about token Engineering in the core of the ocean product from everything before our V1 then going through our V1 our V2 our V3 or V4 and where we are now which is post V4 um and you know the decentralization journey um in our case it wasn't you know strato everything decentralized at once and as well you know how do you reconcile the promises of the white people and the people who believed in it early on compared to you know hitting product Market fit infraction and then besides token Engineering in the core product itself there's a bunch of token engineering around the core product that turns out to really make a big difference and you you can't ignore it

and then finally um just some other sort of grab bag of stuff when you have a token how do you deal with legals and token generation events and so on and there's other talks today that go into a lot more detail but I'll give some some cursory things here so let's get started I will talk about oh one more thing a bit of the feel of the ocean timeline um in 2017 was when we did the initial token design uh iterations um we raised seed money in the tail end that year in 2018 we had the ieo and Ico and building um 2019 uh lots of building leading to our V1 uh 2020 uh we actually released our V2 and our V3 so um extra busy here uh 20 that was 2020 2021 was when the ocean Dell first came out for Grants and

kept building and then our V4 data farming and this year it's all about just building and growing traction building and growing traction and once you keep growing you know you keep growing at a good exponential rate for years and years and that's what takes you to ubiquity and ubiquity is the name of the game for ocean we aim to be uh public decentralized protocol across the internet you know that's just there as a public good and in the case of ocean it's about um data Access Control right that's really the heart of ocean and at the higher level it's data marketplaces and all that um and I'm not going to talk a lot about you know uh ocean and and the products at the higher level and stuff except for

from I took an engineering perspective for that um you know check out the ocean website and all that I really want to focus on um for you guys teaching much more about the token mechanics and things like that but the mission really matters and I'll get to that and the mission of ocean is simply to level the playing field around data and AI so and that's really been a guiding thing all the way along okay so the name of the game is ubiquity um so pre-v1 uh so in uh late 2016 I published a blog post um I came from the world of AI for 20 years I was doing AI for chip design and then in 2013 I jumped over to do work on on blockchain nfts on bitcoin and more stuff and by um 2016 2017 uh well late

2016 I was really missing AI so I decided to just start doing intellectual exercises and I wrote a blog post about how blockchains can help Ai and another blog post about how blockchains can help big data and all roads pointed to decentralized data markets it sort of got at the heart of this problem of how do you get how do AI researchers get their hands on more data especially if you're not inside Google or Facebook so um that was one of the things that led me but there was many other things around data and privacy we were seeing with social media Etc so um from that we're like okay we're going to try to do a decentralized data market but what does that look like we also want to have data Commons so lots of iterations from those blog posts um in in early 2017

mid-2017 we also built a prototype a decentralized data market using bigchaindb this was the project that I had been working on at the time sort of as a company that was basically mongodb decentralized with tenement bft so um we actually did this decentralized data market for Toyota announced it at a big conference at the time but that wasn't the name of the game the name of the game was to come up with really a truly you know public permissionless decentralized data market and along the way we realized to unlock this we really also needed access control so we did a lot of iterations lots of whiteboarding this was really fun you know friends that would pass through Berlin would hang out and jam on whiteboards so we would white border after whiteboard of ideas and kept at it and but it started

to feel like okay when is this all going to end right so we started you know I'm an engineer so it's like okay try to get more and more systematic design and it really was a problem the big problem was when do we know when we're done right so and then I I sort of realized wait a minute this where we're at actually looks a lot like what I've done before in the world of AI specifically optimization design and in the world of optimization design there was sort of this heuristic on the left which is and it's actually lots of engineering designers like this you formulate the problem for optimization this is the objectives things you're trying to maximize minimize the constraints the things that you have to meet softly or on a hard way and the design space What are you wiggling what

are you moving through to what can you change and then once you've formatted that problem written it down you you try to use something off the shelf if you can right try an existing solver and if there's something out there already use it right away right I like to say 10x developers think faster 100x developers import right so um and so here if you can use something off the shelf you should just go and do it right go and use oh um let's see here there we go it's back great um so with the same case for optimization design when you're building an Optimizer or an AI system try an existing solver right if you have just a pro a problem that can be solved as a linear programming problem great use a

linear programming solver right if it can be solved with linear regression or a simple neural network or an svm go use it right and try one solver but try a few and use the one that works the best according to your criteria and if if that solves your problem great you're done um you know and there's a big you know novelty is a wonderful I love creation I love inventing and all that but when you add novelty you add risk right I come from the world of genetic programming uh in in Ai and for many years genetic programming it was very celebrated that GP could come up with novel things and the thing is whenever the more novel The thing is the more risky it is so there I was helping to you know use GP to

automatically invent analog circuits but it would come up with circuits that an analog designer would look at and they're like maybe it works but I don't want to use it because it will probably die in Silicon and I'll be losing 10 million dollars from the cost to manufacture this thing or Aerospace designer otherwise right so there's actually usually a Tris a trade-off of risk versus reward right so novelty equals risk and in general um so while we celebrate novelty understand that the more novel your thing is the more risky it is so you want to minimize novelty and practice okay so um and that's the final step here design new solver so if you need um do design something new but understand that you're adding a lot of risk and probably a lot of time

especially because it takes time to verify okay so um so this actually all of this idea maps to token design formulate the problem objectives constraints design space try an existing pattern right maybe you know you can get away with an erc20 Factory for example maybe you can get away with the token created registry and try different formulations or solver so this is a key thing sometimes you can just tweak the problem a bit and you've got it solved you know Bitcoin itself it's actually a new approach to Byzantine fault tolerance but it didn't fit in within the existing Frameworks of Byzantine fault tolerance because all the ones up to that point had finality they had this guarantee that you were fault tolerant um Bitcoin loosened this and said it's only probabilistic so the academics

didn't know what to do with it at first they struggled with it for years now it's you know part of this broader framework so in that case Satoshi team Satoshi had loosened the problem right saying you know we can have Byzantine fault tolerance without having to have finality guarantees instead if you look at the last page of the the Bitcoin white paper he actually lists saying if you go after if you go for six blocks you've got like a one in a billion chance of failure of finality so um and that that's good enough for basically all practical purposes right this is actually commonly the case where you can just a heuristic in engineering you can often loosen a problem by going statistical right it happens all the time and um it's just useful to understand instead of something where

it's fully fully 100 it's like you know Six Sigma like a one in a billion chance of failure and by the way this is how we got to scaling now worth doing the the random broadcast Etc anyway that's the tangent the point is I could map some existing ideas from optimization and other stuff to token design and once we did this the clouds parted and let's see here okay there we go oh I'm not hmm staying too long on one side I think I have to go faster so as she's figuring that out I'll just

um go ahead so um the next few slides I'm just going to show how one of the things that it did oh there we go awesome okay perfect yeah that's it there so I mean just to give some examples um in terms of formulating the problem in cryptoland this was things like first of all who are the stakeholders what do they want this is a very useful exercise um so in the case of ocean we had data service providers right and they can add value because they have data and services and then what do they get in return in this case tokens making money Etc and we had other stakeholders refers curators verifiers consumers Etc and from that then you can translate that into objectives and constraints

right specs if you will and so we had page after page after page of specs and but you know if you have a huge page like five pages of specs it becomes unwielded to do it you know our brain is much better at handling you know seven plus or minus two things in our head right so to drop that down into something practical we made this table and this is this this table form has served us extremely well over the years um you've got basically in this case it's about seven or eight specs or objectives right and then the columns one two three four these are different designs and um basically design number one was trying just an actor registry in this case it was a TCR of uh who's allowed to be part of this or not right

and we knew that it wasn't enough but we wanted to see how well it would do against the different objectives and constraints and it failed on about two-thirds of them as you can see right but that's the starting point and then we said okay what about a data registry of the assets okay it failed uh it did okay on some things failed in others what about a registry of both actors and um and of assets and then what if you start to layer in creation markets and we started to do better and better but it still wasn't great but this was all off the shelf stuff and that was a key thing you know and if we had of um you know maybe number three would have solved it great solve it and move on and by the way sometimes all you really do need is a centralized database

if that's the case do it right you know don't decentralize if it doesn't if the problem doesn't call for it um okay so uh oh I guess skip something uh I guess we have you know design 567 and eventually we arrived at something that was check check check check and then we knew we were done and that was a key thing right um in some design problems you're done in two hours and you're good to go other design problems it might take you two weeks right but don't assume that this is the final Design This is just basically saying hey this is the most promising thing now we have to go and test this and verify this and so on okay um and yeah that was uh oh yeah I presented these ideas at Etc like mentioned and um it helped to kick off uh took an engineering

um and then also we published a white paper on this with um where you know we said at the heart of it we have this objective um an objective function just like an Optimizer that we're trying to optimize and we laid it out in our case um it's got um it's a function of uh predictive popularity as a function of stake um and some other things too basically and before we did this we had done a similar analysis for Bitcoin itself what's the objective function for Bitcoin um Etc um and we had bonding Curves in our design and so on so um we were pretty happy with that and that's what we um published on and that's actually what we raise money on um and then also yeah we published these blog posts that Angela has referred to

some of them and um and I guess these are used a lot in token engineering Academy so that makes me very happy I this was initially one long long long blog post but then friends came me feedback Trent this is way too long so you gotta chop it down so so I did and then we also started having token engineering meetups and I think by the Angela was even at the very first one and by the I think by the third Meetup Angela was running it because yeah yeah I'm very happy so thank you Angela um and there was not just the meetups but also these token engineering events that started to happen throughout the world in New York and otherwise so that was the the preview one right um and then you know we went to build V1 and we built it and we raised the money

and so on and in V1 um you know we raised money uh our Ico and IO were in March of 2018. so about a year later we we shipped to V1 and um it had its own uh Network we actually had a POA network uh it had smart contracts for data access control it had a an adapt for data sharing and an upgradable contract but there was key things not shipped we would have loved to ship this but um it would have um in fact we had a d scope in order to ship something at all right and we preferred to ship early be a bit embarrassed by when we shipped and then iterate iterate so things we didn't ship right some of the core token mechanics from the white paper we didn't have the bonding curves or that formula I showed a few slides

ago and this you know to me was like a bit embarrassing a bit painful but so be it right like you you start with something knowing that you're going to get to that other stuff we didn't have a data Marketplace app it was just data sharing so we didn't have price data so that's you can't have an economy if you don't have buying and selling and stuff right but it's a starting point um we didn't have a permissionless chain we had our own POA and in that case we were using some special features of the of the parity client and we didn't have a mutable contract so we had upgradability we didn't at the time we didn't actually understand how um how much of a trade-off it is once you have upgradability it means that whoever's using your contracts has to trust you the core team right and we

actually have seen rug pulls from teams where they upgrade the contract and completely screw over their users so um this is a really useful thing to understand we didn't understand this at the time so we shipped with upgradability um okay but that was everyone we learned and we we felt how we realized how uncomfortable we were when when we shipped this so um but for a V2 We There Was An even bigger thing that everyone was saying they're like I have this data I don't want to share it directly I want it where I want to keep my data um private somehow and we looked around at all these privacy preserving approaches you know homework encryption which has finally got mature uh ZK which has finally got mature but it wasn't

mature in 2019. um it wasn't scalable right um and other things too and we realized there's something that doesn't show up at any of the cryptographic conferences that's really useful simply bring compute to the data it's a devops problem not a math problem so that's what we did we put it around a tokenized a token infrastructure and that was the heart of ocean V2 something that we call Ocean compute to data and it was a very practical thing we put it in and it's useful for privacy but also for handling really heavy data that can't move or legals right it's really hard to unlock data from a big Enterprise it'll take you two years of iterating with lawyers so just you know keep the data inside and sell the results of the computation right so that was a you know something we were quite proud of quite a simple

thing to explain too to this day it's probably the simplest thing to explain an ocean so even though we love to talk about token engineering and all this to the average person and that's you know going over their head but it's really easy for them to latch on to this idea you know for privacy computed data boom um so uh so we we often have to get past our academic selves our engineering selves and think about what is just simple for people to understand as sort of the tip of the spear okay on to Ocean V3 so remember in Ocean V1 we did a bunch of stuff that made us feel quite uncomfortable right um and V2 didn't solve that stuff it just solved another big problem which was the Privacy thing but for V3 we see um we had been thinking a lot and we

really wanted to meet the goals of the white paper that was really our our aim and um even before V3 we also realized like we needed to have actually a more system level design that wasn't just about the bonding curves and uh curation it was more about the ecosystem itself everything around ocean and um it took us about a year to come up with a design but it's something that we could call the web3 sustainability Loop and it's also taught at tea and it's actually quite a general thing and you know I've done a couple startups before on blockchain land and as a startup right you're raising money you're going out there um you're using the money raised in order to grow the company and grow the value and you know if you raise that money and take a vacation in Bahamas

then the value of the company go to zero and you can't raise money anymore right but if you actually do a good job with the money that you've been given then you can actually raise the value of the company more than the money that you've taken in right and then you keep printing shares as a company to raise more money to grow grow you know go from a startup to scale up Etc um governments also have a similar pattern interestingly and um with printing money literally and um and then using that to finance things and if they allocate that Capital inefficiently then you have inflation and we have this traditionally in places like Argentina and now more and more in the west um but interestingly this pattern also works well in tokenland right

um and it's this uh flywheel cycle um and we we call it the web 3 sustainability Loop and uh in Ocean it had some really cool benefits so at the heart of it this is it here um there's the workers doing the work in this Loop here but here this we call the data ecosystem and the main thing here is people can do work to build protocols to build daps whatever here but as far as the um the ocean token is concerned um there's Network Revenue coming out and I cut a um so there's you know Revenue to others too but a cut of whatever money flowing through if you can get a cut of that going back to the community where some of it goes back to growth with things like Dows or incentives programs and then also to

sort of give this minimum Baseline of of a value you buy back and burn stuff so basically this is fueling growth but you have somebody back and burn and then you can just loop loop loop through that and in the end this will give you um a positive feedback cycle a Snowball Effect right and we actually for this I developed an agent-based simulator on my own in Python called token spice at the time I didn't have evm yet and we verified all this and we found that certain parameters has worked really well for example how much what percentage of burn should you have versus feedback to the community and um basically you want to minimize this you could put it to zero but then the community doesn't get any Baseline so if you set it to five percent and the rest

to 95 then it will grow way faster and weigh better over time it's the difference between 2X versus 10x in 10 years for growth um that's an example and this is what the simulation showed us and we also worked out many other things things like what sort of token distribution should you have over time Etc so it answered a lot of questions for us but another key thing was we realized now we had tons of flexibility of whatever we put in there right so in terms of the driving value for the token the name of the game was protocol Revenue right and that actually really helped to give us flexibility as we developed ocean V3 so then we did ocean V3 and the Heart of it was you can view it as a revamp of ocean you know core product

um that and it met the goals of the white paper which for us was a great relay if you know so October 2020 about two and a half years after we did our final ieo Ico um but actually we had to get past the sunk cost fallacy we threw out about 90 of her code from V1 V2 that wasn't easy because we had put our heart and soul into this so we really had to get past that um but once we did once again the clouds parted right so um and we also as we were doing this what did we do specifically well in our case we saw okay we really want to kick start the state economy we want to have data wallets for custody and for management of data we want to have exchanges you know dexes for data we want to have Dows

where people can have data Dows you know a bunch of people doing collective bargaining all this um so it felt like this huge mountain of work but then we realized you know what people have already built you know wallets for crypto stuff tokens erc20 erc721 they've already built dexas for this they've already built you know Dows the aragons and colonies of the world so let's just leverage this infrastructure this awesome stuff that has been built by the community over the preceding years how tokenize it right so we simply tokenize data and it feels so small and simple so glib right just like tokenized X but it's actually a huge unlock because now as soon as you tokenize access control we're at the heart of it is if you have 1.0 data tokens it's

erc20 if some if I hold 1.0 of those for a particular data asset X then I can access the thing okay and now I can have a metamask or a trezor hardware wallet all that to access the thing and um so very simple but um it unlocks all these things it unlocks data wallets with custody which is basically a replacement for sap that unlocks dexes so with B3 we also work closely with balancer had automatic pricing with balancer amms right and um some teams emerged in the ocean ecosystem doing datados leveraging the data tokens um and we also the other things you know about automatic pricing this was a big goal always for us the MMS solved that permissionless we just deployed to each

mainnet in the months that followed we deployed to several other evm genes too ambient ability key thing we went to non and upgradable contracts right and this for us was a huge relief because we didn't want to have this you know monkey on our shoulders of liability Etc as the foundation right so it was really nice not to have that and um there was one final thing which is you know in the white paper we had the bonding curves and it didn't have that directly but um we realized we could do something called Data forming which was sort of like liquidity money but optimizing for consume volume and ship that soon that was our Hope was soon that also didn't happen it took us about a year to get there because of what happened next so

what did happen next so we shipped this thing right tokenized data assets assets as erc20 with dex's amm's balancer right so here's live footage from the ocean core team post V3 right we are a fire brigade for about two months three months solid like we were working 14 hour days um non-stop for a while it was exhilarating by the way like super fun but also super stressful why what was happening exactly well people were publishing these data tokens and then other people were speculating like crazy in them trading and selling them several telegram channels got set up specifically just to pump and dump data tokens and all this and I was like At first like oh man like this is not what

we intended we just wanted automatic pricing but a friend of mine said Trent you know like a prerequisite for establishing price of something is speculation so I'm like okay well I guess I can accept that but even besides the speculation other things happened too which were clearly not so good so the first one was fraudulent publishing and we expected this um you know I mentioned we were I'm doing nfts on Bitcoin in 2013 2014. so we thought about this a lot um and uh we realized you know this is a problem that even you know web 2 has Etc and web2 and has paths for this the US government has something called dmca takedowns right more basically someone can come along

and say hey um uh you are using this um uh copyrighted IP that I hold illegally take it down and you have to follow it right and there's similar things throughout the rest of the world as well GitHub has actually a very very cool policy around um copyrighted information sensitive data and so on so what did we do uh well actually I'll get to that in a second so here are some examples we got a letter you know very formal sounding very formal letters from law firms this one in Australia you know we represent so and so on into IP matters as you aware we're a big technology company with lots of money we want to sue you into Oblivion blah blah um and so and so is taking our data set and publish it on Ocean Market and that

violates the terms of service blah blah blah right so we were wondering and then when this would happen and it happened on almost immediately and we're like oh okay so um but we were you know it's actually a very good practice in general don't anticipate everything up front and build for it all up front when Bitcoin was first published there was a bunch of holes in it right but um and even now there's still some holes in Bitcoin that people have identified but the Bitcoin Community to its great credit would fix the holes as soon as they became real problems so because we identified probably 20 possible problems and there was a couple that really popped out so one was fraudulent publishing and our answer was basically copy and paste the GitHub approach around takedowns Etc but it was a bit of a worry because we had

this problem okay we want to be permissionless you know smart contracts deploy to an L1 but we also want to follow the law and here's the recipe this is a really useful recipe in general permission um actually I'll talk about how the web works so if you think about the web I can take a server I can run it in the middle of a you know a random island in the middle of the Pacific Ocean um it can have an IP address it can have a website and anyone can access it right um and no one can take that down it's just there right but usually websites have a.com it's like ens but going backwards right with the domain name server system um DNS and uh so if there's really bad stuff you can just go to the DNS and have it taken down there and websites

themselves have policy so um the the internet and the web is actually permissionless at the lower level it just doesn't have state that's the big difference between the internet and blockchain is State and once you have state then you can have Smart contracts all of this right but there's layers so the very bottom layers are fully permissionless and at the higher levels there's different filtering along the way right such that when you do a Google search or you go to certain websites it has content that's acceptable but of course you can go to the dark dark web Etc and just exist same thing here so on the L1 people can publish the stuff it's um you know fraudulent stuff we can't control it we shouldn't be trying right but what we can do is we can control the top level the the DAP Ocean Market other people

can Fork the ocean Mark adapt have their own policies Etc and this is what we did here so um it's a very useful trade-off between yeah permissionless and in the top level filtering and we see this all the time too like uniswap on the uniswap contracts um they can't stop people from publishing um you know scammy stuff but the The Last Mile you know they're mostly based in the USA they have to follow the laws of the USA so they do right so this is a generally a very good heuristic okay so we solved that the next one rug Bulls so what's the challenge here whenever someone published a data token they owned um um they would actually they would put this into a pool um and the pool was they put 10 ocean

say 10 worth of ocean and then um it was 90 weighted towards the data token so the overall pool had say um a hundred dollars worth of ocean ten dollars worth of ocean sorry yeah 100 worth of of assets ten dollars worth of ocean ninety dollars worth of the data token but the data took inside you know this publisher just published it so they were the whale right they had all of it so what started happening was um uh people would publish um they would um some scam random data set they would go into these channels pump pump it up pump it up and then dump right it was classic pump and dump but it turned out to be extra easy in Ocean Market like oops so um and here's an example we started getting DMs like this hey you know I just I put 41 Ocean I'm

getting feedback here I put 41 Ocean into my pool and I went to take out the ocean and I only have 15 ocean I'm like okay yeah in permanent loss you've got it and it wasn't just that though like it was but of course like I said the Publishers were data token whales so um this we realized pretty quickly was a big problem right so we created on the left this is a GitHub issue an epic and zenhub and we have like 15 different tactics to fix that right boom boom boom boom some of them uh Purgatory for rug polish so once someone was um unidentified as a rung puller we would use the same policy of takedowns as for GitHub and we would remove their assets and stuff right so that's one um we would raise and we're almost like

crazy and blogs et cetera so we put out tweet and blog and all this saying okay be careful understand what you're doing et cetera et cetera et cetera another one actually this made a big difference instead of this you know when you publish a pool where it's ten dollars worth of ocean 90 worth of data token flip it around 70 worth of ocean thirty dollars worth of data so they would actually have to put a lot more liquidity up front so that helped a lot actually and then finally the community also stepped in they started doing their own curation there's um something called the data whale index that this guy Moore it's in the ocean Community uh basically put up a data set for sale indexing all this another person in the community Tim Dublin show it's created something called rugpoolindex.com where you took a whole

bunch of factors and automatically computed what's the chance of a rug pull so this was really great stuff that the community stepped in too and so this was sort of like the immune response towards rug tools right but we realized there were still maybe a more definitive answer and that led us down the path towards ocean V4 okay um so in Ocean before we're like okay we want to definitively solve rung pools so once again we did the token engineering game we wrote a bunch of different specs constraints and we went through a whole bunch of different designs one side is taking liquidity bootstrapping it's an idea from the balancer Community combining those two things Dutch auctions with pools vested pre-mines more stuff right and we arrived at the thing on the far right

um after many things and after many iterations and then we set it and we built it right and rug pulls so the general idea is people put their ocean stake in and then the pool automatically mints the data tokens itself so the publisher never ever has custody they did it against themselves it's all sitting inside the pool so we're pretty proud of that and then through the token engineering community um through training uh we we took token spice and we extended it to work with evm so agents could now have evm under the hood and then we we managed to replicate rug pulls um with the the Balance review Once More contracts in Ocean and in this case this is time and this is Agents net worth in Ocean and this basically what's

happening here this malicious publisher publishes they get a lot of ocean and then they rug boom and so they would get in this case 5000 ocean so we managed to replicate this problem with simulations using token Spice in evm setting um and then we said okay let's put now in the one side it's taking contract part of the V4 Canada design see how it does and it turned out it was 100x reduction so they could do tiny little rug pulls but you know it's nothing in comparison so basically that helped us to feel content in the design this was not a full verification though this was just simulation so there was still potentially edge cases but we actually didn't have time to get to that in retrospect we should have made the time because I'll show you what happened next and this afternoon I'm actually going to

talk about how you can definitively verify this but it requires more tools on top um but so this was you know helped us to get a lot of confidence in the design and actually helped us to tune it as well um and then we shipped ocean B4 so we did use this one-sided staking strategy for the rug pulls um and then we'd gotten feedback for V3 our ecosystem was saying okay you know we're running a market but you've hard-coded where if you get a market you only get point one percent fee but we want two percent fee and we're like yeah like people won't complain about two percent fees so we give them flexibility where if they're running a market they could set the fee themselves um and also you know um we also have this computed data and we wanted to we set it up where people could run their

own computer data nodes and charge for that so they could be a third party running computer to data nodes for others it's not in competition with full-on decentralized compute like I exec or baklava or whatever this is really for bringing computer compute next to the data itself right so it's a very unique special use case but in practice it's very very important to have um and then also yeah obviously the community can make money publishing data so we really wanted to help maximize the community's ability to make money by helping them set fees so that was an important thing um this other question about um governance attacks right so I already told you how for V3 uh we basically said no upgrading right and we hard coded all the parameters but the community was like okay you know we want to set these

fees there's maybe other things what do we do and we also had new ideas about the data tokens like you know what if you want to have a new erc20 template so what we did was We Made It flexible where you can add more data token templates over time so we started with two um and we're about to ship a third one now um and also we we gave some flexibility for some parameter settings so it's really this trade-off of of governance versus um um safety right and once again you know how earlier I was saying how novelty is not your friend novelty is risk it's the same thing with governance governance is not your friend you don't want to maximize governance you want to minimize governance and the optimal governance is zero governance my favorite doll in the

world is called Bitcoin because it has zero governance or you could equivalently call it a perfect democracy for electrons um that's it right so if you can get away with it you should um humans are squishy we've seen all these um Dow governance attacks over the last one year two years right so um in Ocean's case we realized um well we had this challenge we wanted to have some flexibility but not too much and this was the optimal thing so rather than saying can upgrade can't upgrade can't change anything versus you know full-on doubtful and everything in between we said okay you can have new templates you can twiddle a few parameters around fees for your Market but that's it right and that matter needs um and then also we introduce data

nfts to give a bit more flexibility there templates for those two okay um so that was ocean before Post Release what happened um so um well so by the way we no longer had rug pulls all good all good um but uh within a month of release This was um last summer I was actually on vacation at the time and uh we had this message coming from one of our community members they informed us we think there's an attack copying ocean it's like one of these dreads that everyone in protocol land has so it happened to us it wasn't as well known because ocean isn't like a big L1 or something and basically they were staking ocean in a data pool buying data tokens unstaking just the right amount there was a very

particular range of amount and then selling the data tokens and we looked at it and we kept we ran some calculations and we realized if we raise the swap feed of 15 it would make it too cost ineffective for them to do it so that wasn't a big deal we could do it for the near term wasn't a long-term fix but that solved the problem right so we did that we're like Yay good we saw easy easy surgical okay then like um a day later uh I guess it didn't show up the attack too but basically the same thing happened but now um the the attacker just found new parameters and worked around it right and when we like so um and we didn't have the Tooling in place to really verify this really well yet once again the full-on analog to converification

would have been awesome here um so the in this case though we saw like we can't just raise the fee to 20 or something um we really like the pools have a more fundamental flaw and we saw that we could probably took and engineer them um to fix them um but we wouldn't be able to do that in a day or two and we're like so we made this whole thing in mutable immutable um all this how what do we do about this we rocked our brains and we realized um we still had flexibility on setting the swap fee so we simply set the swap fee to 100 and that made it impossible to make money on it wasn't any long-term solution it was to be honest pretty embarrassing to have there right I mean that basically killed the usability of the pools um but at least no one was losing money

and that's really always the thing we come back to is like fundamental constraint you know don't let your community lose money or if they do let them make sure they know what they're doing right so here we saw the community was at Great risk of losing money so we set this and basically it essentially froze the pools right and then after that what we did in the time that followed we made sure the whole Community was made whole in terms of uh what they had for their funds in the pools as well we didn't have to do this in general actually there's a lot of things you don't have to do as a protocol but it's still the right thing to do so we never make promises but we ask we basically say okay do we have the funds to help the community here help

them out and in that particular case we did so we actually airdropped them for their losses too so um so in general you know you're not on the hook for the thing but you always try to do the right thing and we found that that serves us well and I think actually from even a legal perspective you know if you find yourself in a court of law three years from now or seven years from now this notion of reasonableness really counts for a lot it's actually fundamental to business law right and there's this idea of no harm no foul right so if people weren't hurt in the end then it's not an issue so we we solved the problem in the near term but then it's like okay how do we solve this um medium term and uh we basically racked our brains for two days

intensively intensively we'd had ideas bubbling up before and in the end we actually basically ditched pools um and instead replaced it with ve ocean and curation on top of that and there was another reason to ditch pools our community really wasn't getting the idea of a permanent loss so um that was the problem they wanted us take they wanted a steak ocean we wanted curation um we wanted data farming and our answer was of ve ocean so um so basically we had uh you know once again went to the Token engineering table worked it all out the different things um and I'll talk a bit more about ve ocean later but this was once again it was embarrassing we had talked for like almost two years about pools and automatic pricing and staking but we had

to swallow our pride we had to say like uh like the safety of our community was more important than that and um I find in my career that I can usually engineer my way out of almost anything give me some time to just do it and a lot of the ocean team is like this we just we have this um vibe in the team that we can just build it and of course that's there's not just the software there's there's The Human Side and there's always The Human Side um so we were always communicating with the community with this um laying with the facts um as precisely as we could and then giving the action so if you give just a problem without any follow-up people would be like what's going on but if you can be quick about it you can lay the problem and the

solution in the same blog post so this is what we did and we were very fast about this and I'm very proud of the team that we handled it like this um another thing so we had shipped before and it's like traction was pretty good you know we've got some pretty cool stuff happening with Gaia X and with Don Miller and bottom-up teams and some projects coming but at the same time we had found herself you know we did V1 V2 V3 V3 we met the goals of the white paper awesome V4 was really the tuning from V3 based on rug pulls Etc but then it's like what next and the answer is you don't need to be on the version treadmill right um the thing is when we deployed V3 you know it was whole new smart contracts we had to migrate people V4 also it was

partly smart new smart contracts we had to migrate people it's a huge effort just to migrate people so instead we said you know what the system of ocean is General enough we can add more templates if we need that's really really focused 100 on traction go to market iterating with our users learning from them iterating iterating iterating so we got away from the version treadmill and instead it's all about you know the use cases and all of that and I wish honestly I wish we would have done that sooner um but so be it we had met the goals how much time do I have three minutes okay I'll be quicker than great okay awesome thank you so I'll be very quick then token engineering around core product um yes so we did it um when we when we launched the initial ocean token it didn't quite

have some Mechanics for our vesting that we wanted so we did a migration in August of 2020 um that our community knew all of this by the way some scammers snuck in and tried to do stuff there um one month later there's a kucoin hack 14 of the ocean Supply went into the Hacker's hands this was more than ethad um right before the Dow the eth hacker or around the for the Dow so um this hack happen and um you know we were up and at first we're like we're going to leave it alone but then um you know to quote mactation um uh nothing like a punch in the face that changes your plans something like that so we we decided to freeze the ocean token migrate again one month later because this would only happen a month after we migrated and um and and then

basically the hackers funds went away right and so it was a lot of soul searching we really want decentralized but we saw off that if that if that hacker sold that 14 it could destroy the whole project right just like um ethereum made its call in its early days as well it wouldn't do the same thing now but in its early days in my view it made sense so um we did that we launched oceandale grounds um in 2020 we ran it for two years um as it was coming along as it was coming along um we I think we funded 200 teams or 200 projects over time very proud of the program it was growing nicely there were some people starting to game it and then we realized uh the solution to that was uh retroactive like Bitcoin and even more objective really great objective

function but then we realized that that was exactly what data farming was and we saw that with ocean data was going down this path of heavier heavier governance and all this so we decided you know what um it was the same team working in Ocean down data farming we said we're just simply going to wind down ocean down data farming is basically dao-like it's the same goals finding the community but now a well-defined objective function so that's what we did um okay and then uh yeah so we made ocean Dao basically properly decentralized the D in down and automated by folding it into Data forming also this wasn't an easy decision but I'm really glad we made it and now we've doubled down on that once again fully automated right governance is something to minimize um and also the ocean token itself when

we uh even after the migrations we had it where the core team was able to pause it which never made me feel good but we needed it in order to Mint um for the minting um and finally when data forming got far enough along we could mint the rest of the remaining tokens so we just did that about three months ago and with that we were able to remove all control from the core team of the token and that was a huge monkey offer back so I'm very happy about that so now on Ocean the only real decentralized thing left to decentralize is some of the data forming so we're like 95 there I'm very very happy with that um and we've been pushing governance to the leaf nodes okay um I know I know so uh and summary so the ocean model yeah um we've got a blog

post that just came out so you can see that and when you have a token um there's a lot of scammers right so um you have to you see all these things happening all the time for example there's people that register at Ocean protocol.ro and all these other things so as soon as you hear about it you do a takedown to the Nic you just go to them there's a policy there we've proactively registered all the big ones like ocean protocol.org all of this um so that helps a lot um also how do you reconcile speculation with Builder Community channels and um a pattern that we use and many teams use is for the Builder oriented discussion that's Discord for the general discussion but not speculation that's controlled by the core team and our collaborators and

there's several Speculator communities especially a big one in Telegram and that's a very useful heuristic um and let's see keeping going and yeah what should you do first product Market fitter token um and ideally if you have the funds try to just build a thing get it out there um and then later on the token as you need it's so much easier sometimes the token is really really critical to the product itself so then you kind of have to iterate but product Market fit you know basically as soon as you can go for it there's a more Nuance view of this overall and I've got a blog post from a couple years on that and almost almost done Angela um great oh yeah the conclusion so so that's it I've shared um you know the ocean token Journey for the core product as well as everything in the ecosystem

around things like ve ocean and all that these are all like optimizes but at the heart of it because we follow this web3 sustainability Loop we've had a lot of flexibility in what we can do where and that allows us to basically go for traction to drive the fundamental value to grow the community et cetera and it's been a really fun Journey so far ocean iview is now in a new phase because we've got past the white paper it's all about traction it's really fun so so thank you all for your time thank you Angela for your passion thank you a friend [Applause] I mean you can see this passion on I mean six years in and managing all these struggles and this token Journey so

thanks a lot for sharing Trent and be on stage today and there there is a recording just in case you want to watch this talk again with such a rich content thank you so much and while we are setting up the stage again for the next panel I would be Keen to hear your thoughts I mean we have heard about several of the main challenges for a token launch raise your hands from your perspective are you here because you think the regulation around tokens is the biggest challenge for crypto projects launching a token no worries about regulation only two okay cool um and what about fundraising

fundraising biggest challenge for a token launch yeah okay yeah so so and what about the product Market fit and the demand drivers okay there's more I mean Trend actually built the stage for these topics um if you know this journey through versions of ocean protocol many different iterations or let's say the evolution of the mechanism design of the token design of ocean is really a yeah just highlights some of the steps that have to be taken and that a token design is basically something that is not done

once you launch a token it's a continuous Evolution and you have to react on your ecosystem on your stakeholders on on the market of course and also on continuously growing the potential of your protocol good um we can we remove the state yeah okay I'd like to now invite Marina dimi and Robert on stage to discuss the top 10 challenges for a token launch give them a round of applause [Applause]