20sq

20 [ ] One pager Current state of affairs At the moment, 20squares revolves around the use of a software implementation of Compositional Game Theory. This software allows for fast modelling and subsequent analytics. In this way, we are able to provide a platform for strategic analysis and testing, that can be used to discover, prove and correct economical properties of financial software and protocols. Our implementation is open source, but its usage requires some training. As a consequence, our main line of business at the moment is in consulting. We work with interested projects that are looking for formal reassurances that what they are doing makes economical sense. This includes: • DeFi platforms, interested in the financial stability of their protocols; • Marketplaces, interested in economically optimized solutions for revenue extraction; • Fundamental infrastructure designers, interested in maximizing the sustainability and welfare of its users; • MEV searchers, interested in analyzing their strategies. Vision The consulting activity provides good bootstrapping opportunities as it pays well and requires little resources to set up. It also offers the possibility to explore the current frontier as most projects we are involved in center around the introduction of novel products. The obvious downside of consulting is its difficult scalability. Currently, there are three main verticals that we are exploring to scale our operations. • Scaling the consulting business towards automated and compositional analytics. On the basis of a growing library of models, we offer more and more automated services, e.g. live monitoring of DeFi protocols. • Setting up a trading shop: the consulting business by default provides a growing library of models detailing use cases and best strategies. As these models often describe financial applications, they can be used to extract revenue in-house. For instance, after having built and extensive library of MEV models, it may make sense to focus on MEV in-house. • Building a Layer1. This is heavily dependent on the traction the product receives. As the community that we are building around our techniques keeps growing, we may reach a critical mass that makes it viable to turn the product into a decentralized service. Farther away in the future, all of these verticals can be merged into an overarching project: creating a fully on-chain investment fund. This can happen if the following milestones are reached: • The consulting vertical is developed enough to have an internal team of analysts that are fluent in quickly modelling protocols from whitepaper material and publicly available source code. With this, we will be able to do in-house technical due diligence. • The trading vertical results in enough experience to confidently run investments. With this, we will be able to manage our positions and do adequate risk-assessment. • The Layer1 vertical has scaled successfully. This will provide the basic infrastructure on which both risk management and due diligence models will be run. In this scenario, 20squares becomes the heaviest user of its own Layer1. This is interesting because in hints at a perfect reason to launch a proprietary token. Indeed, up to this point the wisest choice will be to base our Layer1 on some other coin like ETH, as many of the projects using our services will inevitably lie within the broather Ethereum ecosystem (in this respect, our situation is similar to the one of protocols like Arbitrum). With the fund realized, it will instead make sense to launch a proprietary token backed by the value of the investments the fund will have realized. In this respect, the token would work as a security. We consider this scenario a far-fetched one at the moment, as it requires massive amounts of capital, experience and expertise to be actuated. Waiting for all of these things to accrue, we focus on the three verticals we highlighted before. Needed development and research steps At this stage, from a technological perspective these scaling opportunities can be explored in parallel, as they all depend on further investigating a set of common and interrelated technical matters. We list these here: 1. Furthering integration with crypto, as in, for instance, automatic translation of EVM bytecode into our framework. This is needed to further cut prototypation time and further reduce human error in modelling. 2. Adoption and integration of machine learning techniques in the current framework. At the moment, simulations with a high number of players become quickly computationally unfeasible. We have reasons to believe that machine learning techniques can push this computational bottleneck much further than what it is now. 3. Investigation of memoization techniques. This is needed to optimize the problem of equilibrium search, which is both very appealing for clients and very computationally intensive at the moment. 4. Extending the framework to full-flagged cybernetic systems. This will extend the scope of what can be modelled in our framework by several orders of magnitude. These directions of research are oderdered in terms of their chronological deliverablity, meaning that the ones listed first require less R&D work to become viable. Required resources - back of the envelope calculations We are already partnering with some institutions to resolve these technical matters. 1. We are actively working with specialists at the Ethereum Foundation to further integrate our software with the EVM. 2. We are working with Tal Kachman's ML research group at Randboud University on the application of ML techniques to player scaling. 3. We are partnering with the University of Strathclyde to scale our implementation to fully flagged categorical cybernetics. These collaborations require funding: 1. EF experts are very expensive. At the moment they are basically working pro-bono since they are very interested in what we do, but we would like to capitalize on this interest by actively funding this common effort. 2. At the moment the human capital that Kachman's group can deploy on this collaboration is limited. This would radically change if we could fund this research directly. 3. The majority of the theory behind categorical cybernetics has been figured out by PhD students at Strathclyde university. These students now are all on the brink of finishing their PhDs and some of them are already being courted by leading players such as Deep Mind. We would like to retain this human capital by hiring them at the Cybercats institute, which is the research institution created with the purpose of detaining the IP of our work. Here is an estimate of how much we need, on a annual basis: 1. Salaries: 480K, estimating 6 full time people with a salary of 80K/year. 2. Funding Tal's group: 100K, mainly to hire postdocs and PhDs. 3. Cybercat Institute: 150K to fund research. 4. Integration with crypto: 200K for consultancy fees and outsourcing. 5. Other expenses: 70K, which include travelling, organizing events, marketing and outreach. The total burnout rate is 1M/year. We require 2M to be able to comfortably run for a couple of years. In this estimate we are not taking into account the fact that our consulting business is already profitable. Those resources will be redirected towards the consulting business itself, by hiring and training more people. We will be able to redirect this human capital to one of the above-listed verticals in the future. 20 [ ] One pager Current state of affairs Demand side - user problem A central feature of the crypto space is decentralization: Independent agents make decision given the rules of the protocols they engage with. For this to work and to produce overall beneficial results, the design of protocols and their incentives are essential. Many projects therefore are looking for formal, game theoretic reassurances that what they are doing makes economic sense. This includes: • DeFi platforms, interested in the financial stability of their protocols; • Marketplaces, interested in economically optimized solutions for revenue extraction; • Fundamental infrastructure designers, interested in maximizing the sustainability and welfare of its users; • MEV searchers, interested in analyzing their strategies. As a consequence, there is a wide-spread demand for game-theoretic and economic analysis. Our technology 20squares revolves around the use of a software implementation of Compositional Game Theory. Analysts can use the software for strategic analysis and testing which in turn can be used to discover, prove and correct economic properties of financial software and protocols. The key advantage of the software is that it provides leverage for analysts: Equipped with the software analysts can significantly speed up the process of modelling and subsequent analytics. It also enables them to build up complex models in an effective way. Our business Our implementation is open source and open for everyone to use. In practice, however, the tool has very specific prerequisites for effective usage: One needs background knowledge in Category Theory, Game Theory, the programming language Haskell, and the craft of applied modelling. This limits the number of possible users -- for now. But in the right hands, the tool leverages what game theory consultants can do. We at 20squares as the inventors of that technology are the people with the most experience in using that tool. At the moment, we have an advantage and can exploit this. It enables us to provide game-theory consulting services in a way that non-one else can. Current research At this stage, we are further investigating a set of common and interrelated technical matters. We list these here: 1. Furthering integration with crypto. We are already implementing a system for automatic translation of EVM bytecode into our framework. We believe this will result in shorter prototypation time and reduced human error in modelling. 2. Adoption and integration of machine learning techniques in the current framework. At the moment, simulations with a high number of players become quickly computationally unfeasible. Preliminary tests have led us to believe that machine learning techniques can push this computational bottleneck much further than what it is now. This is crucial as many 'real life' situations are naturally modelled as many-player games. 3. Playing with memoization techniques. Internal experimenting showed how memoization optimizes the problem of equilibrium search, which is both practically needed and very computationally intensive at the moment. Pushing the boundaries of what is computationally feasible when it comes to equilibrium search means actual usefulness in real-life situations. 4. Extending the framework to full-flagged cybernetic systems. This direction of research is the most abstract and academically relevant at the moment. It will extend the scope of what can be modelled in our framework by several orders of magnitude, as we will move from modelling games to modelling any system with feedback loops. Vision Perspective of consulting business Like any other consulting business, our consulting activity can be extended by getting more people on board but ultimately is not scalable. The prospective market demand would clearly support this. At the moment we have a natural moat as we have a headstart in using the technology we invented. But if we show that this technology can be fruitfully applied, competitors will inevitably enter the market. Beyond consulting business If we want to go beyond a fixed size consulting business, we need to develop an actual product. We consider the following verticals: Consulting 'as is' We are growing Compositional Game Theory into a platform for strategic analysis and testing, that can be used to discover, prove and correct economical properties of financial software and protocols. As we stated already, this platform is already being naturally developed into a consulting business which pays reasonably well and requires little resources to set up. It also offers the possibility to explore the current frontier as most projects we are involved in center around the introduction of novel products. The obvious downside of consulting is its difficult scalability, but on the basis of a growing library of models, we will be able to offer more and more automated services, e.g. live monitoring of DeFi protocols. Leveraging on the self-similarity of a good portion of the crypto ecosystem, this should result in cheaper services that can be offered at scale with minimal maitaining cost on our side. Trading shop As we said, the consulting business by default provides a growing library of models detailing use cases and best strategies. As these models often describe financial applications, they can be used to extract revenue in-house. For instance, after having built and extensive library of MEV models, it may make sense to focus on MEV in-house. One of the main motivations for developing in this direction is that we know for a fact that some funds have internally used and implemented our research to aid automated trading, with success. So we know that in principle this can work. On the other hand, the downside of this vertical is the high upfront capital cost and the higher risk. Moreover, setting up a trading shop requires a considerable level of expertise in areas we are not familiar with at the moment, such as algorithmic trading, legal and tax optimization. We are already maitaining a shortlist of people with the right skills to set up an hypothetical collaboration in this direction, but we do not yet feel confident enough to act on it. We need to get more data points and expertise on how to actually set things up, and we are convinced this will naturally come from developing projects for or with trading firms. The layer1 Another direction would be building a Layer1. This is heavily dependent on the traction the product receives. As the community that we are building around our techniques keeps growing, we may reach a critical mass that makes it viable to turn the product into a decentralized service. The upside of such an approach is that successful Layer1 infrastructure is incredibly lucrative. The downside is the possible lack of adoption.This is a non-trivial problem for us, as Compositional Game Theory is mainly faced towards business enterprise and not retail clients. The best course of action would be employing Compositional Game Theory as the core component of a 'traditional, retail-facing blockchain'. The problem with this are the same hinted above, namely the need of skills outside of our area of expertise, and the high initial costs. The investment fund Another possible direction is creating a proprietary investment fund. Here, Compositional Game Theory would become the core product for technical due diligence. This is interesting because in hints at a perfect reason to launch a proprietary token, which is again a lucrative endevor. As the upfront cost of launching a fund can be quantified in the order of hundreds of million of dollars, this is no small feat. A smart bootstrapping strategy would be building the fund not by means of investments but by means of collaborations. Compositional Game Theory, and more in general Compositional Cybernetics, can be employed as the core tech in many different businesses, ranging from crypto and financial services to smart grids, supply chain optimization and avionics. By strategically looking for partnerships in this fields, we can build an investment portfolio substituting deployed financial capital with deployed technical capital: As we collaborate in the creation of a given venture, we become co-owners of that venture. On the surface, the risks of this direction of development are similar to the ones of the consulting one, namely slow scalability. To this, one should add the further risk of taking revenue in equity, with only a minimal part in cash, and the resulting higher volatility. Current Market Assessment From a technological perspective, we have a clear roadmap for the research to develop the foundations for the vision outlined above and what resources are needed. See above. We also detail this in another document (research-plan-resources.md). What is much more unclear is the business side and the roadmap we should pursue there. It is evident that the vision we outline above requires multiple steps and that the order is relevant. What is more, specifically at the beginning, whether we go into direction of scaling consulting or go into the direction of trading makes a big difference. It will determine very different sets of skills and resources we need to acquire. It also requires different types of investment. The consulting vertical is most obvious for us as for its gradual expansion. Moreover, in this direction we can leverage our current consulting work strategically. But again, currently we lack domain knowledge to clearly articulate a plan how we actually want to get automation working. What is to do? 1. We need to decide whether to pursue the trading business or not. 2. We need to develop concrete hypotheses on how we want to automate the analysis or break out of the consulting business. We provide details in a dedicated document (/businessDev/consulting.md). 3. Altogether, we are too far away from concrete scalability. We have to actively look for projects for collaboration in order to test our hypotheses. Which means we first have to be very clear which hypotheses we want to have and how we can actually test them. 4. The most obvious channel for identifying projects that can help us develop our own product is to connect with VC partners and ask them for projects. title: Business development of consulting case What is the product/service we sell? Problem • Almost any application in the blockchain space is dependent on: o implementing sound incentives for users and all stake-holders to make the system sustainable o monitoring the state of their systems and be aware how incentives might change (due to changes in the application, outside factors like general market conditions, and a generally fast evolving eco-system) Our solution • We offer game theoretic based incentive analyses • In cooperation with the client we provide consulting and guidance for the development and design of their product • Our analyses are based on a novel software suite we have developed • Our software provides a framework for representing and analyzing game-theoretic scenarios Advantages for the client • Fast prototyping • Easily extendable and scalable through building models in a compositional fashion • Models can be run live, i.e. mirror active systems and use data-streams to provide status reports (particularly relevant for DeFi like applications) • Tooling is open source and thereby fully verifiable by anyone Who are our customers? • We focus on early stage development of novel products/applications Customer needs • Needs to evaluate the soundness of a new product idea • Modelling needs to go in tandem with dev work Typical customers • New product under development • Lacks specific economic, game-theoretic expertise • Little to no data available to evaluate idea/ refine design • Might lack resources if it is a new project • Focused on limited engagement (8-12 weeks) to get the product running Our solution • The key features of our tooling help well to address the needs of customers: o No prior data needed. If data is available can be made part of the modelling approach. o Fast prototyping helps to iterate quickly through different designs (which by definition change at an early stage) o Compositionality supports the sketching of non-critical parts and then revisiting them later o Software based tooling makes it easier to coordinate with developers Distribution • Because of our early stage product focus we target new projects entering the space as well as established projects venturing out into a new product-space • Our tactic is to work with a few existing bigger projects which develop new products: o This gives us invaluable exposure to large tectonic shifts o Gives us attention and might help us guide new smaller projects to us o Gives us access to partners of wider eco-system o Prime example for this is our collaboration with Flashbots on the development of PBS • While lucrative, the market of big leading projects is very limited by its very nature. Our main sales focus therefore will be on new projects. • Our main sales strategy here will be on building partnerships with a few strategic venture capital funds o They provide access to a constant stream of new projects o The funds themselves might be interested in providing the service for their portfolio companies in order to help them but also as a possible due-diligence on their end o Funded projects are obviously better capitalized and able to pay for our services (a critical point) • Early stage projects as a lock-in for extended services in the life-time of the project o Our typical early stage project will require our services to get going o Depending on the product continuous analytics are an additional service we can offer o Once we designed the initial system our customer has switching costs: We can offer additional data-stream and monitoring services much easier than outside competitors who have to start from scratch Pricing Basic pricing components Our compensation for "helping them fly" gigs will have two parts of compensations: • Consulting fee • Token compensation Pricing tactics A form of token compensation should signal our commitment and also provide us with better incentives. The exact ratios depend on the project - maturity as well as how key the modelling is that we provide. Obviously, the more peripheral our work is relative to the overall project, the less our work will influence overall success and the less token participation we should have. The consulting fee orients itself to comparable fees we have already receive or have been offered, i.e. 20-40k USD per person per week. For projects where we see potential for ensuing live analytics work, it might be wise to charge less to get a foothold (essentially cross-subsidizing later lucrative work). Lastly, given the bear market currently, being more aggressive towards token compensation might be a good tactic. More on this below. How do we want to scale? Scaling consulting • Individual consulting is obviously limited in its scalability • We are working on three technological advancements which will enable us to scale the consulting business as well as breakout into novel markets. i. Backend for act: We are working with the Ethereum Foundation to develop an interface between our software tooling and act, a specification language for the implementation of smart contracts. Compositionality, the central feature of our engine makes our tooling the prime candidate for such an integration. This will enable us to provide a unified framework from crafting, analyzing, and refining applications and automatically compiling smart contracts. ii. Integration with ML techniques.: Our engine as well as its theoretical foundation provide ways of closely integrating it with state of the art ML techniques. This will enable us to scale analyses and automate them. iii. Powerful memoization and parallelization of our engine: This will increase our ability to solve very complex optimization tasks. • In addition, we will also be able to lower the costs of developing models over time. With every project we work with, we can extend our library of relevant modules. We also aim to leverage a community around our open source tooling. • The technological advancement will be realized gradually. That means, we do not have to wait until one day all of it is in place but can profit from it already before. The new features we will develop combined with a growing library of models will allow us to speed up the process of modelling dramatically. And once equipped with the automatic implementation stack into deployable smart contracts, we can offer infrastructure as a service. Competition tbd How is it different from gauntlet? Business hypotheses We need to translate our technological advancements and/or business ideas into concrete product hypotheses we can test. Here are some examples that need to be fleshed out. The ideas mentioned are too lofty and not operationally concrete enough but directionally right. • H1: We provide a smart contract deployment environment: Design and analyze smart contracts in open games and automatically deploy them (through act and dapp tools). • H2: We provide a platform that takes in continuous data feeds and analyzes the health of the underlying platform. The idea is that we provide a digital twin that mirrors a platform and produces key metrics on the basis of an incentive analysis