Token Engineering Research Groups A Multidisciplinary use case for Open Science

Transcript

okay welcome to my talk about token engineering academy and token engineering research groups some words about te academy i'm angela kreitenweiss i've kicked off the academy one year ago in july 2020 and since then we've had more than 650 participants in our training programs we have participants from 21 different time zones we run open science collaborations with git coin balancer block science hydra dx boson ocean protocol and 96 percent would definitely recommend our program

to others again token engineering academy offers education for token engineers um we are establishing this new discipline and form the body of knowledge the truth and the ethics and t commons the community behind the academy is forming this as a new social institution in a community driven way our mission we develop new engineering a new engineering discipline which is basically the youngest discipline after software engineering was established in the 1960s we educate a new generation of engineers and

redefine the value flows for education and research since we can leverage web-free value flows for our own offerings and for our own education program some words about token engineering it's an emerging discipline as mentioned let's take a look at crypto 2008 we had the white paper published proof of work bitcoin and 2012 peercoin the first proof of stake then 2015 ethereum launched 2015 also the esc token standard was initiated then 2017 bonding curve

was launched at banker 2018 falcoin this huge ico with more than 250 millions and 2021 d5 with around 60 billion in token value locked so this is the history 13 years of crypto now look at token engineering basically this discipline was first mentioned in 2018 only three years ago we were running first meetups in the community 2018 2019 academia picked up with first conferences and the journal and 2020 again te academy launched so this

is really something very new what's important is that this new discipline can draw from a lot of existing scientific disciplines and that's good actually because we don't have to invent everything from scratch let's take a look at this overview this was from is it from a paper foundations of crypto economic systems by sherman fashion gear and michael sargen published in 2019 and here you can see the various disciplines that are relevant for token engineering computer science and cryptography obviously then we can apply ai optimization and control theory to the design of these

systems industrial systems engineering operations research and management science and of course economics and game theory and we have the social layer since in these systems it's not only about technology it's about the human in the loop we have psychology decision science political science and governance for decentralized decision making and philosophy law and ethics as well here's a brief overview on a typical token engineering process to give you a flavor of how token engineering works you start with the ecosystem design so

you analyze who are the stakeholders in your system so what assets do they bring to the table what's the motivation to take part and then you design these economic mechanisms and create a dynamic model so how this system emerges and changes over time and the economic model the value flows the financial model and then you not go straight to implementation you build simulations to stress test and verify your design to see where it breaks where you have sensitivities and since we have so many interconnectedness composability in

crypto this is even more important to detect negative side effects so you build a digital twin and run simulations and verification of your system and only then go to implementation and run your system in production and of course since we have a digital system we can collect really granular data on what's going on in the system as token also being the signal to to provide information on your system state and we can feed back in operational life data into our simulations for further iterations for system updates and the strategy of

the system now let's take a look i've mentioned this is multidisciplinary let's take a look at the students i've mentioned more than 650 students so far and here's their fields of study we have of course a huge portion people coming from engineering and computer science that's great also the red one finance economics and business and in green natural science mathematics physics and data science this is pretty relevant and you can imagine those people are hard to get they have brilliant job opportunities in

traditional finance and web too and it's super important that we can attract them to crypto and i think we can because uh designing crypto economic systems is really an exciting challenge and then in in yellow social science as mentioned we are talking about socio-technical systems political science sociology law and also design so this is the combination of students and we our students are coming from all across the globe so we had in one program 21 time zones and people are really eager to take part

sometimes we have teams that are attending lectures during their local night time and not only once over weeks so this is really exciting to see now here's some screenshots of our recent sessions to give you a impression of how token engineering work looks like and also how token engineers look like so this is about programming diagramming designing flows of value also analyzing algorithms this is our crowd and now i'd like to

take you on a journey to one research group initiative this is gitcoin grants quadratic funding and we worked on it earlier this year and invited community researchers to work on it together with bitcoin and block science and work on a particular challenge first some words about bitcoin you probably all know it a platform to fund open source projects and it's actually really successful in round 10 that close in july they distributed more than a million us dollar in crowdfunding and 700 000 us dollar in an additional matching pot so this is

really relevant for open source projects now we worked with bitcoin and block science to work on a particular challenge for the funding mechanism quadratic funding and the challenge was fight collusion and civil attacks for git coin grants now let's first look at how quadratic funding works basically we have two different funding streams number one community contributors so people like you and me we send from our personal wallets we send funds to our favorite project via the git coin grants platform and

support it second there's a matching part set up by larger donors crypto projects that want to support and help the open source layer grow now the funds from this matching pot again round 10 around 700k are distributed according to the support gathered by the community contributors now let's take a look at these diagrams on the left in green we have the community contributions so you see there are smaller ones larger ones everyone can decide

on her own sent one u.s dollars and tens and 50 these are the the average numbers and in yellow you have the matching pot and the yellow area is calculated using the values of the green ones so a square root is taken from the green individual contributions then you create a sum and end up with this total area and the yellow matching okay now compare the right distribution of contributions uh the left one with the right one on the right side we have more green squares and smaller ones

and if you look at the yellow matching funds then you see if you have many small squares in green then you end up with a larger total yellow area so you get more matching funds and this is actually the key benefit of quadratic funding because for the matching the number of contributors matters more than the amount funded and this reduces the power of whales if you now imagine okay let's take one really big green square there a whale contributing a lot then you automatically reduce the yellow area okay

this is a benefit of quadratic funding reducing the power of whales but it can also be an attack vector because in theory a project might activate its community to send a huge number of very very tiny contributions to your grant and as a result we have many many very tiny green squares and then a massive amount of the yellow area of the matching front with only then tiny tiny little green squares with only a very little own investment and this is an attack vector to

quadratic funding and our objective was to look through the lens of different disciplines so we invited community researchers coming from various backgrounds and look through the lens of their own discipline and try to tackle collusion and civil attacks for gitcoin grants now first i'd like to show you two approaches applying machine learning to this problem danilo defined an optimality gap this is the difference between the maximum possible matching you can ever get in a certain constellation and the actual matching

this is mr this is the actual matching and mo is the maximum possible matching and this is an optimization problem so you approximate a global maximum and one method we tried you can see it on this tweet on the right is simulated annealing and this is an algorithm that's also used for route optimization you probably know this delivery services what is the shortest possible route for a number of destinations and similarly we asked what is the maximum matching funds one can achieve with a certain setup of grants and contributions and such an algorithm takes a lot of iterations of edge

rewiring and testing okay if i take this combination is it more or can i achieve more matching fund or another combination or another combination and i think in our hacking session we had optimization runs that took around 22 hours so it's computation heavy however it's totally viable for a lot of optimization problems one example another approach from a different team was applying node to vec again for detecting collusion this is semi-supervised learning to find similarities in graphs so here we represent contributions in nodes

so the contributors and the grants as you can see on the right and represent them numerically in a feature embedding and the algorithm then calculates the similarities between empirical nodes so what actually happened this is also a graph graphical representation of an actual evolution of a donation network so based on what's actually happening and an injected artificial attack vector and then compare here's an attack vector can we find something that looks pretty similar and this might be a suspicious

grant a colluding grant okay two examples machine learning now another example mathematics looking through the lens of mathematics to a token engineering challenge andrew another member of the research group um didn't try to look at the actual contributions rather to the algorithm itself so he explored the quadratic funding algorithm and ask is if this might be sure convex so this is a there's a sure ostrovsky criterium and at first sight this quadratic

funding algorithm isn't meaning you really have to run optimization runs to find a global maximum but after a couple of attempts he proved that actually the shure ostrovsky criterium can be met and that means that we can use a very simple fast calculation to get the global maximum of a certain family of contributors contributing to a grant and that's actually a really this is a breakthrough for quadratic funding to make it more robust because we then are much faster in defining a global maximum and we don't have to use for example

one single optimization round takes 22 hours and this is prepared to be a submission to a math journal so it's an actual formal mathematical proof and is a major contribution to the quadratic funding literature now these have been three examples of tackling collusion algorithmically however no algorithm alone is strong enough to really judge on fraudulent behavior and here i'd like to highlight block sciences work

on establishing a process for bitcoin grants define detect evaluate and sanction you have to first to fight problems like collusion you have to define the adversarial behavior so what is fraudulent behavior in this actual system quadratic funding then you also have to define penalties you have to detect adversarial behavior and here we can use both algorithms as mentioned as well as human input then you can find and flag activity and then you still have to

evaluate because sometimes [Music] organic network evolution looks very similar to collusion and only then apply penalties and continuously iterate and improve so here we can see that token engineering is combining human and algorithmic input and this can also lead to new roles in a system with new rewards for making an ecosystem more robust and resilient now let me now take you to

software infrastructure for token engineering here's an example transparency okay bitcoin grants data is open and accessible its own chain so everybody has access to it can see the contributions can see the grants however data analytics is often still a struggle data analytics and on-chain data and tj created a data pouch on tokenomics dot io bitcoin you can access it still and to basically find out hey what what's happening during a funding round and

also afterwards and he found that some of the funds contributed to particular projects are still unclaimed so here at the lower area you can see a grant with 21 k die still unclaimed so there's also infrastructure needed to analyze grants data or to support the process and this is a nice example and last but not least here's a new framework to integrate existing solidity smart contracts to token engineering simulations so our current framework in this

disciplines are python based and if we rebuild on-chain interactions smart contracts in python you always have of course additional efforts and it's lacking precision so token spice is a new framework for cases where you want to build your simulation on on-chain smart contracts and one team of te alumni just got a grant from ocean dow to further develop this framework and enable token engineers to have ethereum virtual machine in the loop basically so this is what happens if you bring

smart people from various disciplines to one room to work on a token engineering challenge and in case you want to learn more or want to get active yourself then here are some recommendations balancer simulations this is a current research group where we are working with balancer on v2 pools we have eight different research projects ranging from dynamic weights to funds of funds strategy then we have additional infrastructure provided by the community for the community js sdks and python models amm explorations a repo to gain am intuition

and final presentations again will be next tuesday check out tokenengineering.org t academy or on twitter you'll find also all the links twitter is talk engineering and here's something where you actually can take part as an active co-researcher omnipool engineering and simulations is a project that will be kicked off in august and is running until october and this is a similar research group where we work on cross-chain liquidity pools with hydra and with block science and the program includes a really cool onboarding to

systems engineering and major paradigms like the mathematical specification of the system parameter selection and that we will use cat cat python based framework and then first part is lectures and introductions and second part is working on your own personal research question you can still apply until end of this week again it's via tokenengineering.org tea academy and we welcome people from many different disciplines governors is an initiative for establishing a branch of token engineers for governance challenges like votings like colluding new twitter account governance

down so follow us if you're interested in this topic and we'll run first events in august and finally te comments t-comments is the community behind te academy to manage to fund and manage the funds for token engineering public goods and we just kicked off our token launch the hatch phase opened up last week yes you a lot of a lot of tea commons people here in this room we have actually more than 400 contributors working on that and yeah our goal is

to acquire funds and to manage funds for token engineering public goods so feel free to reach out to everyone wearing such a t-shirt and go to hatchte.com and be part of it if you think that it's worth investing into ethical safe resilient and sustainable economic systems thank you for having me is there any question

any questions you can also ping me i'm around all right you