This is the first in a three-part series on the latest thinking in token engineering as cryptoeconomists grapple with increased regulatory scrutiny and increasingly complex protocol designs.
In this initial post we focus on the evolution of tokens and the token landscape, with an emphasis on enabling emergent functionality, which is when a system is designed to produce properties that are not localizable to any single component within the system. These features of complex adaptive systems make them incredibly resilient as they virtually eliminate single points-of-failure. The high level of coordination among disparate actors that is required can be enabled by a token.
In the second part we will go deeper into understanding how complex systems and their enabling protocols can be designed through modeling and simulation. It will explain this field of research, which is known as complexity science, as well as its implications for token design.
The final post of the series will bring these components together to provide a blueprint for how to model a token economy as a complex system and use it to develop sustainable tokenomics. Overall, these posts will provide an overview of the history, present and future of token engineering, outlining the process for our take on token engineering best practices.
Tokens Through History
The term cryptocurrency has always been somewhat misleading, implying that each token acts as a currency. Whilst this can be true, it is an exception, not the rule, and we prefer the more inclusive term token. Tokens can instead be loosely defined as a representation of an asset that is “tokenized” on a blockchain. They can be used as a medium of exchange, a measure or store of value, and as a means of payment – examples include the Bitcoin network’s BTC and Ethereum’s token, ETH. Other use cases of tokens include being used as a means to access a specific ecosystem or as a mechanism for rewarding activity as in the case of social tokens.
Bitcoin is the original token, with its token utility being a peer-to-peer payments system with no trusted third party. Between 2012 and 2017 we saw a plethora of tokens being created, typically with a simple formula - token utilities were largely restricted to being used to pay for gas or as the sole payment method for the product or service of the creator. The primary differentiators beyond the project’s supposed mission were token allocations and whether vesting or cliff schedules were in place.
Once the Initial Coin Offering (ICO) bubble burst in 2018, in part due to the advent of Initial Exchange Offerings (IEOs) and otherwise due to the failure of projects, the token design landscape evolved. A class action filed against Ripple in May 2018, accusing the company of an unregistered token sale, was another nail in the coffin. With it came increased scrutiny of token sales and, importantly, an emphasis on token utilities having demonstrably differentiated utility value to the ecosystem to which they pertain. We define the differentiated utility requirement as emergent functionality, pairing with emergent qualities of crypto systems.
Emergent Functionality and Complex Adaptive Systems
Crypto System Emergence
Crypto systems are emergent because they operate via inductive, bottom-up mechanisms rather than via top-down authority. Of specific importance is the value that arises from the emergent trust created by the individual agents that are incentivized to simultaneously compete and collaborate to agree on the ground truth of the system’s state. When a decentralized system is operating, its performance is emergent, as the system level dynamics are not reducible to any individual agent. If the performance is in service of achieving an intended outcome, then we may view the function itself as an emergent property of the system.
Token-Enabled Emergent Functionality
In a decentralized system, the token should act not only as a means of aligning the disparate agents' intentions, but also as a critical component which enables emergent functionality. For the Ethereum blockchain, ETH is required to execute operations, which are simultaneously processed and validated across a diverse set of decentralized agents to enable a non-intermediated emergent consensus to be achieved; this is core to Ethereum's emergent functionality. A token which participates in enabling an emergent function is truly 'Tech Crypto’. Hence, the token must do more than act as a reward for work performed, it must play a role in the system’s function. ETH exemplifies this, acting as both a reward for those securing and maintaining the blockchain, whilst also acting as a gas fee payment medium for end-users.
Governance tokens are another example of emergent functionality. Pioneered by MakerDAO’s MKR token in 2020, governance functionality enabled the control of the protocol to be passed over to the community - enabling token holders to define the direction of the company and labeling MakerDAO as decentralized. Thus, MKR, and other governance tokens that sufficiently cede control to the community, fit the bill for emergent functionality as they remove the dependence on an individual agent, instead dispersing operational responsibilities amongst a wide group of agents.
Both the biological and technological worlds exhibit emergent functionality. An ant colony foraging for food, and the packet routing protocol in TCP/IP copying a large file between geographically disparate nodes, are both excellent respective examples. Ant foraging and packet routing would not scale if they were orchestrated by a central authority.
Complex Adaptive Systems
The technical class of systems that have the potential to generate non-pathological emergence from the interaction of many independent decision-making entities are known as complex adaptive systems (CAS)   . Non-pathological emergence is the occurrence when a set of sophisticated and unexpected beneficial outcomes emerge naturally from interactions between simple entities, without any inherent abnormalities in the system.
Designing a decentralized system which produces emergent functions is akin to the undertaking of CAS engineering. Armed with this realization, we are able to leverage the methodologies and approaches that have been developed for the field of Complexity Science, including the Four Pillars , discussed below.
First, let us unpack what we mean by CAS.
Ethereum and other truly decentralized systems are complex because there are many individual and densely interconnected entities which together form the system. A deeper exploration into the differences between complicated and complex systems  is out of scope for this post.
Ethereum and other decentralized systems are also adaptive as the interconnected entities mentioned above have the ability to choose how they will participate in the system. At one scale, they can validate, stake, restake, buy, sell, hold, lend, borrow, and rage quit, among many other choices. When considering higher order adaptive capabilities, we see that decentralized autonomous organizations (DAOs) are able to leverage the cumulative power of the constituent decision makers to adapt at the protocol scale. An exhaustive exploration of the many examples of multiscale adaptation in crypto is out-of-scope of the current discussion. These types of multiscale adaptive properties are one of the hallmarks of CAS.
A system, as defined by Oxford Dictionary, is ‘a set of things working together as parts of a mechanism or an interconnecting network’. This definition is quite adequate when prepended with the previously described qualifiers.
The Four Pillars of Emergent Systems
Building upon prior work in engineering decentralized systems, these four fundamental concepts appear to be requisite in the development of a functionally-useful complex system; these concepts contribute strongly to our view of emergent functionality.
1. Decentralized Decision-Making
Decentralized decision-making can provide an additional layer of resilience to a protocol. In the case of an agent situated in a hierarchically-prominent position in a centralized system, a bad decision can be ruinous for the system as a whole. When decision-making is decentralized, the worst-case outcome of a bad decision is attrition of that agent; there will be little to no impact on the system’s operational continuation - one bad agent does not derail the entire system, just as removing any one bird from a flock does not destroy the flock . By providing a mechanism for integrating the decision-making of multiple actors, we reduce the likelihood of a decision by a single entity having a ruinous impact on the protocol and move forward towards the elimination of single points of failure.
Homogeneity across a decentralized enterprise leads to systemic fragility and/or prohibitive cost. Certain agents require certain capabilities, and if every agent has every capability, we will end up with extremely expensive assets. This inflated cost would reduce the number of agents available, and we would end up with a system that is fragile to attrition due to the relatively low number of agents. Additionally, homogeneous agents are more susceptible to common-mode failures. This, too, leads to systemic fragility to intelligent perturbation.
For example, client diversity, rollup/scaling diversity, and geographic/jurisdictional diversity diversity, among many other threads, have long been recognized as critical to Ethereum’s resilience.
3. Functional Degeneracy
A resilient system requires functional redundancy. Those familiar with enterprise IT approaches to robustness will likely be reminded of continuity-of-operations (coop), where a geographically distinct hot failover is kept in sync with the primary system of concern. One of the most complex systems known, the human brain, does not have a backup on standby, but somehow is able to adapt to severe perturbation quite readily. Redundancy to significant but non-ruinous perturbation is made possible via the plasticity of self-organized neuronal ensembles  stepping in to fill the functional gaps that may arise; neurons have agency. In the human brain, often ischemic events that cause damage to functioning, for example, Broca’s Area damage halting the ability to speak, are, with time and effort, made up for by alternative neuronal ensembles.
A system exhibits functional degeneracy if a system function can be performed by a distinct and often diverse alternative system component (or set thereof). Case-in-point, if an Ethereum validator running on an AWS instance in the US is halted, a solo-staker running on a desktop computer in Australia can take its place.
In decentralized biological systems, upon which many observations critical to the modeling and design of CAS have been made, stigmergy is a class of mechanisms that enable interaction between organisms. Most importantly, stigmergy is distinct from other types of coordination in that it leverages indirect communication via perturbations and observations to a shared environment . At the core of Ethereum, stigmergic interaction between agents happens via the blocks themselves. The blockchain itself constitutes a shared environment and the state changes that occur between blocks are the means by which indirect communication takes place. These changes, once observed, often trigger subsequent decisions to take behavioral actions that otherwise might not occur. In this way, our example system, Ethereum, reproduces one of the key features of CAS.
This article has covered the evolution of token engineering, how decentralized systems are themselves complex adaptive systems and how this should shape our approach when designing effective socio-technical token models. At the core of our thinking is a focus on emergence and how this can be realized through the implementation of one or more of the four pillars of decentralization.
While this methodology is becoming increasingly adopted by thought leaders in the space, there are still significant hurdles to implementing this at scale across the industry. Existing methods for developing agent based models of CAS’ are time consuming and highly specialized, and whilst there exist efforts to simplify this process through applications, like Machinations, getting the level of fidelity required to understand these systems holistically is challenging. We expect this to be a growing area of focus in coming years as cryptoeconomists lean into learnings from complex systems engineering and the market for tools to help simplify this process grows.
In the next blog post in our series on Token Engineering, we will explore how the concepts above have led to new types of design tools, and how those tools can be leveraged to explore the novel design space of CAS. The third post in this series will explore the implications of those methodologies for crypto and web3 and discuss potential areas of application.
 Holland, J. H. (1992). Complex adaptive systems. Daedalus, 121(1), 17-30.
 Norman, M. D., Karavas, Y. G., & Reed, H. (2018, July). The emergence of trust and value in public blockchain networks. In IX International Conference on Complex Systems (p. 22).
 Anderson, P. W. (1972). More is different: broken symmetry and the nature of the hierarchical structure of science. Science, 177(4047), 393-396. https://doi.org/10.1126/science.177.4047.393
 Norman, M.D., Silvey, P.E., Koehler, M.T.K., Joe, K.C. (2022). Engineering Decentralized Enterprises: Emergent Mission Accomplishment Without Centralized Command and Control. In: Yang, Z., von Briesen, E. (eds) Proceedings of the 2021 Conference of The Computational Social Science Society of the Americas. CSSSA 2021. Springer Proceedings in Complexity. Springer, Cham. https://doi.org/10.1007/978-3-030-96188-6_10
 Norman, M.D., Koehler, M.T., Pitsko, R.: Applied complexity science: enabling emergence through heuristics and simulations. In: Emergent Behavior in Complex Systems Engineering: A Modeling and Simulation Approach, pp. 201–226. Wiley (2018)
 Reynolds, C.W.: Flocks, herds and schools: a distributed behavioral model. In: Proceedings of the 14th Annual Conference on Computer Graphics and Interactive Techniques, pp. 25–34 (1987)
 Fuster, J.M., Bressler, S.L.: Past makes future: role of pFC in prediction. J. Cogn. Neurosci. 27(4), 639–654 (2015). https://doi.org/10.1162/jocn a 00746
 Theraulaz, G., & Bonabeau, E. (1999). A brief history of stigmergy. Artificial life, 5(2), 97-116.