Wanna know more about data science? Make sure to check out my events and my webinar What it's like to be a data scientist and What’s the best way to become a data scientist !
Agent based models have been around for a few decades now. One of the main methodologies behind the study of complex systems, the main idea behind ABMs is the modelling of systems by modelling their individuals parts. This idea comes in contrast to the standard mathematical modelling, that focuses on higher level behaviours of a system. Why someone would want to do that? There are two main reasons:
- ABM can model non-linear and chaotic behaviours of systems, which are very difficult to capture with other types of models.
- ABM’s are very intuitive. For example, if we are modelling a market, we can model each individual trader, based on a few simple behaviors.
An example of early successes in ABM was Schelling’s segregation model. Thomas Schelling demonstrated that how communities can get segregated, even when the individuals are not trying to achieve segregation themselves. We assume that all people live on a square grid and that their only requirement is that half their neighbours are of the same colour. They are perfectly happy if half of their neighbours are of some other colour, so they are not racist in the sense that they demand that all their neighbours are like them. If this assumption is not met, then they seek to move elsewhere. Even under these assumptions, the model converges to a situation where distinct segregated areas are created. You can see how this works in the video below. Left side shows the square lattice, right side shows the neighbourhood borders.
Agent based modelling and economics
As you might have guessed, it makes absolute sense to use this type of models in economics. Economic methodology is usually concerned with higher level models, things such as supply and demand curves. However, these models require an oversimplification of the underlying dynamics. The most notable example is the assumption of rationality, which has been much disputed after the financial meltdown of 2008. The Economist had written up an article discussing whether ABM would have been able to predict the crisis, that traditional models failed to predict.
Agent-based computational economics have been around for a bit more than a decade. One of the prime examples is the sugarscape model developed by Joshua M. Epstein and Robert Axtell and detailed in their book Growing Artificial Societies, that concerns a fictional economy where agents live on ‘sugar’. The agents consume sugar and can reproduce, trade, transfer information etc. like simplifications of real humans. A simulation of the model is shown below.
So, what all this has to do with tokenomics and ICOs? Tokenomics allow us to create artificial economies with artificially constructed incentives. All white papers in this area make all kinds of assumptions as to token adoption, usage, forecasts of future value, etc. However, I’ve rarely stumbled upon a convincing white paper in this area. The simple reason is that tokenomics is a relatively new field. Actually, many people won’t even recognise the term. There are no standard rules as to how to set up a token economy and what to expect in terms of forecasts and usage. Research in this area is also scarce.
However, it is clear that such as framework for evaluating and assessing token economies is desperately needed for two reasons. First, because it would allow the founders of the economy to understand its pros and cons and improve their model. Secondly, because it can offer ICO investors some assurance as to what to expect when investing, since such a framework could be used to forecast future value and adoption.
Agent based modelling and tokenomics
Therefore, with no solid theoretical framework or mathematical models of economics to rely on, how can someone build up a solid framework for their token economy? This is where agent based modelling becomes a natural choice for tokenomics. The modelling of token economies through ABM allows us to bypass any theoretical limitations and model the agents of our assumptions directly, while at the same time taking into account any kind of constraint or assumption we want. This is the model I used successfully for Crowd for Angels’ ICO. Through an ABM it became possible to demonstrate that, if the assumptions behind the business model of Crowd for Angels are correct, then their token should be close to parity with the Great British Pound.
I believe the need for ABM in tokenomics becomes even more important when we consider the mistakes that many startups do when designing their token economies (something about which I wrote up in another post). The fact that ICOs have managed to raise around $5.6 billion should on one hand get us excited, but at the same time cautious. There is still lots of work that needs to be done in regulating ICOs, and so far, none of this work is concerned about the 3 most important questions:
- Can a particular token economy work?
- How well can it work, what is the maximum possible valuation?
- How volatile and sensitive it is? Is it stressed-tested?
However, with the use of agent based modelling I am confident that these questions can be studied and answered with reasonable degrees of certainty. Building a good tokenomics model is not easy (as I explain in the video below). However, with the aid of agent based modelling things become considerably easier. It is time that we take this great tool out of our toolbox and back into the forefront!