Complexity science is an interdisciplinary field of studies spanning subjects from physics and biology to economics, to the social world. The field aims to analyze complex systems: systems that cannot be reduced to their constitutive components and contain many non-linear and dynamic interactions. Given the increased interconnectedness of the world (e.g. due to digitization and globalization) and the fact that many of the world’s largest challenges can be considered “wicked problems”, complexity theory will increasingly become part of the toolbox of the 21st century researcher.

Our observations

  • Physicist Albert-Laszlo Barabásihas set the research agenda for the next generation of students of complexity sciences through his work on scale-free networks. In his work, he provided a theoretical model for the “preferential attachment mechanism”: a process that describes how hubs (i.e. heavily connected nods) are more likely to find new connections. An example is found in the World Wide Web, where HTML documents that point from one page to another follow a power-law distribution, and it is empirically found that the most linked website is twice as likely as the second most linked website to be linked again.
  • The tradition of systems thinking in sociology dates back to Niklas Luhmann, who described a “system” as a sphere of reduced complexity, separated from its chaotic environment. Social systems are constituted by an internal communicative process of information selection and meaning is created through the process of bringing some order to the virtually infinite and chaotic outside. Importantly, the systems described by Luhmann are “autopoietic”: they constitute themselves through self-referential communication. Modern bureaucracies are an example of such complexity reducing systems: by imposing a set of binding standards on their citizens, they create their own internal rules for communicating information. The act of labeling and categorizing is not merely descriptive but constitutive of citizenship as such.
  • From the 19th century on, increasingly sophisticated statistics methods and models have aimed to discover patterns in human behavior so that they can be regulated. The French statistician Adolphe Quetelet showed that factors such as age, class, and status allow for the prediction of marriage decisions. Modern dating apps essentially draw on these mechanisms by using algorithms.
  • The sociological tradition of systems thinking is continued by Armin Nassehi. In his recent book Patterns: Theory of the Digital Society, he defends the claim that the digital revolution, rather than creating new social, political and economic structures, merely reveals already existing ones. Digital technology is seen as embedded within these existing structures of behavior and its success can only be properly understood by looking at longer established economic, political and scientific forms of societal organization.
  • Two years ago, we already wrote about economic complexity and how new economic metrics and economic theories borrow from complexity studies and other disciplines, creating a new paradigm for thinking about economic development and relations.

Connecting the dots

The notion of economies as complex adaptive system dates back to the Anglo-Austrian philosopher and economist F.A. von Hayek who described how advanced economies can be seen as spontaneous orders in which order emerges without central coordination, from individuals pursuing their self-interests. A key characteristic of the Hayekian approach is viewing (economic) systems as wholes which cannot be understood merely from their individual parts (emergentism). Likewise, these systems follow their own rules in an evolutionary adaptive process that can neither be understood nor predicted on the basis of knowledge about the individual elements. Spontaneous orders are scale-free networks, while organizations are hierarchical networks. Early work on spontaneous orders remained theoretical, while researchers are now increasingly developing systemic framework for the computational analysis of complex economies and social systems.
Similar patterns can be found in different contexts, such as in the field of quantitative linguistics: Zipf’s law describes how the most frequently used word in a language occurs approximately twice as often as the second most frequent word. The same holds for citations of the most prominent author in a scientific field. This can also be applied in social network analysis and explains, for example, why most people have fewer Facebook friends than their average Facebook friend.
Pioneer of systems thinking Scott Page gives an example of how complexity thinking can be applied in organizations theory. His 2017 book The Diversity Bonus: How Great Teams Pay Off in the Knowledge Economy shows how the systems perspective can be helpful in explaining successful performances of teams: One should not only look at the structural set-up and interactions of individual parts in a reductionist and mechanistic manner. Page states that the so-

called diversity bonus of teams is the result of various types of cognitive diversity, that is, differences in how people perceive, encode, analyze and organize the same information and experiences (and how these differences are in turn correlated with identity diversity, i.e. racial or gender differences). Often, we still lack the vocabulary to make sense of the dynamic interactions observed in complex adaptive systems, but this might be essential in addressing challenges of an ever-more complex and uncertain world. Complexity thinking allows for the discovery of previously hidden structures. In many cases, it provides the link between statistics and qualitative inquiry. Patterns found by complexity scientists can be found in systems and networks across contexts. As such, it makes – at least theoretically –  room for a synthesis of the natural and social sciences. Given the immense amounts of data we face today, complexity science promises to provide a useful conceptual framework for a multi-disciplinary way of doing science.
Complexity economics further bears the potential of fully bringing the computer revolution to economics. It might, for example, close the gap between econometrics and behavioral economics, enabling us to explain consumer behavior from both a structure and an agency perspective. Agent-based models allow for simulations which are, for example, applied in urban planning or supply chain management, but are also used to predict the spread of epidemics or to project the future needs of the healthcare system. Evolutionary or complexity perspectives are, however, typically based on assumptions which go against the fundamentals of mainstream economics, whereby rational agents face constrained optimization problems. This divergence of theoretical assumptions makes it difficult to integrate new approaches with older ones and requires a deeper paradigm shift.


  • Given that the large majority of work in complexity science or systems theory remains theoretical in nature, it does not yet have the potential to compete with neoclassical approaches. Most research still revolves around abstract mathematical models, while reality often turns out to be more nuanced. More computational approaches to modelling society and economy are, however, in the making, and it is recommended to keep track of developments in that field.

  • Economists might increasingly have to borrow concepts from natural sciences as well as sociology and psychology to allow for more dynamic perspectives. We already see a number of economists adopting evolutionary perspectives on, for example, institutional development. Conceptually, this goes back to Austrian economist J. Schumpeter, who stressed the point that “capitalism can only be understood as an evolutionary process of continuous innovation and ‘creative destruction’”. The heterodox field of evolutionary economics and evolutionary game theory has popularized concepts such as bounded rationality, diffusion and path dependency.

  • Complexity economics has as yet failed to reach its full potential because, on the one hand, it tackles fundamental assumptions of neoclassical economics, while on the other hand, practical applications remain relatively rare. Computational models, are, however, in the making and complexity economics might thus soon help to fully bring the computer revolution to economics. As knowledge is becoming an ever-more important factor in economies and the amounts of data produced keep growing, researchers are advised to look out for new conceptual frameworks as well as for ways to translate new insights into practical applications.