Complex Systems: Order Without a Designer

The vocabulary of network science arrived late. The phenomena it describes did not.

Hayek’s 1945 paper on the use of knowledge in society described prices as a mechanism for aggregating and transmitting dispersed information across millions of actors without any central processor. He did not use the word network. He did not have the terminology. But the structure he described, nodes with local information passing signals to adjacent nodes, producing global coordination that no individual node planned, is exactly what network scientists formalized forty years later. The insight was not new. The mathematics was.

Complex systems appear across domains because they share a structural property: local interactions between components produce global patterns that cannot be predicted from the properties of the components alone. This is why the same mathematical tools that describe how a pathogen moves through a population also describe how a financial shock propagates through interbank lending markets, how a behavior diffuses through a social network, or how a failure cascades through a power grid. Albert-László Barabási’s work on scale-free networks drew simultaneously on physics, biology, and sociology, because the underlying structure, a small number of highly connected hubs dominating a network of sparsely connected nodes, appears in protein interaction networks, the internet, citation graphs, and social networks for the same mathematical reason.

The non-linearity of these systems is what makes them genuinely difficult to manage. Small changes in network structure, adding or removing a few connections near a hub, can shift a system from stable equilibrium to cascading failure without any obvious warning signal. This is a threshold dynamic: the system absorbs perturbations up to a critical point and then reorganizes rapidly and discontinuously. Standard analytical tools built for linear systems do not capture this at all.

Model choice matters in ways that are frequently glossed over. The SIR framework (Susceptible, Infected, Recovered) models biological contagion with immunity and makes specific predictions about epidemic thresholds and vaccination strategies. Threshold models, developed by Mark Granovetter in the 1970s from observations of riot behavior and collective action, model social contagion differently: adoption depends not on direct contact with an infected node but on the fraction of your neighbors who have already adopted a behavior. These are mathematically distinct and produce different predictions. Treating them as interchangeable is an error.

Ilya Prigogine’s work on dissipative structures, developed independently of Hayek and the Austrian tradition, showed that open systems far from thermodynamic equilibrium can spontaneously generate ordered structure by dissipating energy to their environment. His 1977 Nobel Prize recognized that self-organization is a general physical principle, not a special property of living systems. The parallel with Hayek’s spontaneous order is substantial: both describe how order emerges from interaction under constraints without a designer. Neither framework has fully absorbed the other, and I think that integration is one of the more interesting open questions in complexity science.

The internet was not designed to eliminate social friction. It was designed for packet routing and redundancy, to survive partial failures and reroute information around damaged nodes. That architecture does make certain types of propagation faster and more resilient to disruption, but that is a consequence, not a purpose. Attributing social properties to network architecture is a categorical error, though a common one.

Questions worth investigating

1. Hayek’s spontaneous order theory and Prigogine’s dissipative structures both describe self-organizing systems far from equilibrium. Has there been a rigorous theoretical integration of these two frameworks, and what would it look like? Developed further in Hayek and Prigogine: Two Paths to Order Without a Designer.
2. How do threshold cascade models of social contagion differ empirically from SIR-type models when applied to the spread of political behaviors or financial panics?

References

Barabási, A. L., & Albert, R. (1999). Emergence of scaling in random networks. Science, 286(5439), 509–512.

Granovetter, M. (1978). Threshold models of collective behavior. American Journal of Sociology, 83(6), 1420–1443.

Hayek, F. A. (1945). The use of knowledge in society. American Economic Review, 35(4), 519–530.

Prigogine, I., & Stengers, I. (1984). Order Out of Chaos: Man’s New Dialogue with Nature. Bantam Books.

Watts, D. J., & Strogatz, S. H. (1998). Collective dynamics of ‘small-world’ networks. Nature, 393, 440–442.