Cryptoeconomics and Complexity I
Complex systems, cybernetics and autopoiesis (oh my!)
The study of complex systems has emerged (if you’ll excuse the pun) as one of the most revolutionary shifts in economics, biology and many other fields in the past 30 years. In fact, it’s inseparable from the computing revolution itself. Whether you know it as “systems theory,” complexity, (neo)cybernetics or even swarm intelligence, it’s almost certain that you have come across ideas from this exciting new approach to science and society.
Perhaps the best known complexity concept, in the entrepreneurial world at least, is the “network effect,” described in a seminal paper by W. Brian Arthur almost three decades ago. Arthur, whose work at the Santa Fe Institute (the world’s leading research institution devoted to the complexity field) has helped to define “complexity economics,” reminds us that
“Complexity is not a theory but a movement in the sciences that studies how the interacting elements in a system create overall patterns, and how these overall patterns in turn cause the interacting elements to change or adapt.”
As such, complexity approaches tend to emphasize factors such as the heterogeneity of agents’ decision making processes and their limited cognitive resources, reject “equilibrium” models of economic activity, and deploy approaches such as agent-based simulation modeling to study complex social systems. As we will learn, the emphasis on adaptation will also prove vital to understanding the future of blockchain-based forms of economic life.
Given that cryptoeconomics and the emerging discipline of “token engineering” aim at the design and analysis of system-level phenomena, it is quite surprising, at least to me, that complexity approaches have not garnered much attention in the blockchain world.
In the rest of this series of around six short posts, I will show how the complexity movement provides us with an expanded conceptual vocabulary (as well as practical methods) for the analysis of token mechanism designs and, at least tentatively, for the development of more holistic cryptoasset valuation methodologies.
In this first post, I will define three essential terms:
- complexity and its study;
- the closely related field of cybernetics, particularly “second-order” cybernetics;
- the idea of “autopoiesis.”
First we will identify what makes a system “complex.” We will then consider the role of “self-regulation” and environmental feedback in complex systems. Finally, and most importantly, I will propose that it is the autopoietic, self-sustaining quality of certain socio-technical systems that is (or at least should be) the real goal of cryptoeconomics and token engineering.
Defining Complexity
When we say that, for example, a rainforest ecosystem, ocean currents or geological processes — or, for that matter, a token-driven economy — constitute a “complex system,” we do not just mean that the system is highly complicated. Rather, the term refers specifically to the class of systems which exhibit behaviors at a range of scales which emerge from the interactions between forces or “agents” over time.
These behaviors can include the formation of attractors (which, as shown in the famous plot of the Lorenz Attractor, can be visualized as virtual points around which the state of the network or nearby agents seems to be oriented), rapid changes of state which propagate across the network without warning when it reaches a “tipping point,” emergent subsystems, feedback loops and emergent stable formations which are not predictable from the properties of individual elements of the system.
The goal of most research in complexity is not, however, the quantification of how much complexity a system exhibits. In particular, the goal of analyses of this type, when applied to economic systems, should not be to arbitrarily increase or decrease the system’s degree of complexity in the name of optimization! Rather, a complexity-driven approach to economic behavior should point to the inherent unpredictability of large-scale systems driven by even the simplest of rules (just as anthropology shows us that societies with exceedingly complex rules of exchange can maintain metastable states for, by capitalist standards at least, exceedingly long periods).
Cybernetics and after
As Wired magazine’s founder Kevin Kelly points out in his “Bootstrapping Complexity” — perhaps the most overt attempt to bring a deeper version of complex systems thinking into the “Silicon Valley mainstream”—the roots of the complexity movement in the early history of cybernetics have been obscured as a “new generation of scientists have come into cybernetics on their own, unencumbered by an academic tradition, [who] rarely describe their work in cybernetic terms.” The original sense of the term — now characterized as “first-order” cybernetics — referred to self-regulating, so-called “coupled” systems consisting of a “controller” and an “effector.” The word comes from the Ancient Greek kybernetes, a helmsman, appropriately enough given the importance of research into missile and space rocket guidance systems in the early days of the field.
There are doubtless some elements of cryptoeconomic systems which could be characterized as first-order cybernetics — such as basic “scalping” bots which react solely to price movements, functioning analogously to a bang-bang controller in engineering and do not, in themselves, form a representation of the intentions of other market agents. More relevant to cryptoeconomics, though, is the “second-order” cybernetics of a group including “founding father” of computer science Norbert Weiner, anthropologists Margaret Mead and Gregory Bateson, physicist and philosopher Heinz von Foerster and biologists Humberto Maturana and Francisco Varela.
While these thinkers span a wide range of positions, it is the recursive quality of this type of analysis, in which the role of the observer and the construction of models are themselves treated as a part of the system under analysis which characterizes this “meta-cybernetic” approach. Paradigmatically, von Foerster himself defined the field as “the control of control and the communication of communication.”
Since economic activity in general, and most clearly in the digital asset space, can be understood as a mode of communicative action, it is easy to see how the design of economic interaction rules could be understood as a second-order cybernetic field — one dealing with the control of controls which themselves regulate the communication of communications in the “internet of value,” namely market interactions and agent behavior (as, for example, in the case of a reputation token linked to a token-curated registry). Both the technical specification of the cryptographic mechanisms, data objects, smart contracts and other protocol definitions, and the human-readable “front end” interface with the system are taken into account in a second-order cybernetic account of token engineering, along with the designer’s (and, equally importantly, users’) formation of abstract models of both the system’s behavior and that of other users.
While, in general, first-order cybernetics dealt strictly with technological systems, second-order cybernetics addressed questions of human and animal social interaction, along with living systems; indeed, as Kelly observes, it gave rise to “systems” approaches in biology and many other fields which now seldom define themselves in explicitly cybernetic terms. It is, though, in the field of systems approaches to natural phenomena, and complex human social fields like the economy (which exhibit self-organizing and self-sustaining behavior without any deliberate design or maintenance), that perhaps the most critical concept for cryptoeconomics is to be found.
Autopoietic systems
As we have seen, a system can be complex without exhibiting cybernetic qualities (such as the climate or a natural ecosystem), can be capable of at least first-order cybernetic self-regulation without necessarily behaving in a complex non-linear fashion, and may even self-organize without that organization itself contributing to the system’s “survival” (as in the case of spontaneous, sometimes long-lived emergent forms in complex natural systems, like the “polar hexagon” on Saturn). The significant common factor among these systems is that none can be described as living: there is, as Maturana and Varela found, a particular quality to self-sustaining complex systems which distinguishes them from other forms of complexity.
Their original definition of autopoiesis — literally “self-making” — specifies that:
An autopoietic machine is a machine organized (defined as a unity) as a network of processes of production (transformation and destruction) of components which:
(i) through their interactions and transformations continuously regenerate and realize the network of processes (relations) that produced them; and
(ii) constitute it (the machine) as a concrete unity in space in which they (the components) exist by specifying the topological domain of its realization as such a network.
It is, perhaps, rather easier to visualize this schema in terms of a digital entity than the biological ones to which the authors originally referred. Reading this definition in the context of cryptoeconomics, we can take any token-driven system as the “machine” and see that not only is the specification of “processes of production” (including, but not limited to token issuance or burn) required, but, more critically, the operations of the machine (including transactions between users as human “components” of the network, along with interactions between algorithmic actors) must “continuously regenerate and realize the network of processes that produced them.”
The second element of the definition is, in this instance, rather more precisely specified by the technical construction of the “machine” than in the case of social or biological systems. This element of the specification of an autopoietic system can, in the case of blockchain networks, be understood as the set of consensus protocols that define the range of possible communicative relations between nodes.
Since, rather obviously, a blockchain network requires at the very least energy and hardware inputs to stay “alive,” it follows that for a token-driven economic entity to be considered autopoietic, it must be able to engage in relational processes with human economic actors qua “components,” and those processes must themselves regenerate the network. That is to say, in the terms of Maturana and Varela’s theory of self-sustaining complex systems, only a network which does not require inputs of energy or information which are external to the relations of the network can be considered autopoietic.
The most notable adopter of the concept of autopoiesis in the social sciences, Niklas Luhmann, is far from uncritical in his adoption of the term, but clarifies it productively, noting that an autopoietic system is
a system that is its own product. The operation is the condition for the production of operations.
At the most basic level, this should raise questions about cryptoeconomic systems which exist as the product of a more-or-less separate system: if the operations of a “conventional” firm, such as marketing, are prerequisites for the functioning of a network, it is at best questionable whether the network can be considered autopoietic. We will return to this question in the coming posts.
It is worth noting that capitalist firms are not typically described as autopoietic (along with centrally-planned economies), whereas “free market” economies are (and note that I would add many, if not all pre-capitalist economies to this list). In the broadest sense, the precondition for the execution of a financial operation (even one as simple as a direct trade of one asset for another) is the existence of a more-or-less financialized economy, to be sure, whereas the existence of any given institution, currency or instrument is necessarily contingent. In the case of proposed new systems that more or less explicitly claim the kind of universality that characterizes existing financial systems, it should be clear that the aim is to develop a system which is self-reproducing in the sense that, as a total system, it can autonomously support the generation of new relations (including transactions) on, as it were, its own terms. Thus, it is this self-referential quality of autopoietic systems — their unique ability to generate succeeding generations of relations according to their own terms — that, I propose, we should look for in successful token designs.
In the case of blockchain-based modes of economic organization, then, what we should focus on is the interaction between the second-order cybernetic processes of recursive “relation design” involved in token engineering, and the ongoing production of relations which, ideally, can reproduce those same conditions without reference to or control by forces external to the system. If token engineering is “done right” such a system can be understood as autopoietic — under certain conditions.
In the next post in this series, I’ll show how a complexity approach to cryptoeconomics and token engineering, which emphasizes the development of autopoietic systems above all, is both necessary and practical. Over the series, we’ll learn:
- The difference between unsustainable decentralized systems and autopoietic networks
- How to define the extent of economic networks (and how to account for the role of non-human agency in them)
- Where the need for adaptive behavior sets in autopoietic systems sets limits to token engineering
- How autopoiesis is essential to the creation of long-term network and asset value.
