Revolutionary (top figure) and evolutionary (bottom figure) views of scientific change. In some suitable space of ideas Kuhn describes discontinuous, revolutionary transitions between epochs of approximate stasis — normal science. Popper and Lakatos describe science as a continuous process of idea generation, refutation, and confirmation. All three agree that the average trend has a positive slope. (Image: David Krakauer)


Santa Fe Institute
Nov 7, 2019 · 3 min read


by David Krakauer, President of the Santa Fe Institute

There are broadly speaking two schools of thought examining the progress of science, one evolutionary and one revolutionary. The evolutionary schools are exemplified by the philosophers of science, Karl Popper and Imre Lakatos. Popper argued that science proceeds through a combination of bold conjecture and ruthless empirical refutation. Lakatos extended Popper’s framework adding that ideas can also receive empirical confirmation. The culture of science thereby maintains an informal score card registering wins and losses with dominant ideas receiving the highest aggregate score. This sum is the definition of objectivity. Thomas Kuhn was the most vocal champion of the revolutionary school arguing that most science is normal or incremental (as per Popper and Lakatos) but that occasionally “paradigm” shifting ideas overturn prior beliefs, replacing them with radically new models, often achieved through an equally radical simplification in theory.

The natural sciences have provided most of the intellectual natural history upon which Popper, Lakatos, and Kuhn base their ideas. The Copernican Revolution, quantum mechanics, special and general relativity, Mendelian genetics, Darwinian natural selection, plate tectonics. In all of these cases it is rather easy to demonstrate that previous ideas were either refuted or deprecated in favor of new ideas that could account for a broader range and higher resolution of empirical observation. What of the social sciences? I have wondered what might count as an unassailable refutation or a revolutionary idea in the social sciences. In particular in the field of economics. I have posed this question to two Nobel-winning economists who chose to abstain from answering. My friend, Sir Partha Dasgupta, the Frank Ramsey Profes-sor Emeritus of Economics at the University of Cam-bridge, answered without hesitation: game theory. Partha argued that the mathematical formalization of strategic interactions under uncertainty changed the way that economists conceptualized agents, incentives, and markets. I suspect that a similar revolutionary case could be made for Adam Smith’s invisible hand, the Keynesian multiplier, and Ricardian trade theory. Complexity science aims to provide fundamentally new ways of thinking about the adaptive universe and to do so in a way that reveals common structures that exist independently of the contingencies of disciplinary histories. Thus in principle a revolutionary idea in complexity science could be as revolutionary in economics as in cell biology, neuroscience, and sociology. And this is a truly stunning thought: no longer would an idea be confined to a field but extend across specialties transforming them all. What better argument could there be for pursuing a career in complexity science?This is of course not as odd and unprecedented a possibility as it might seem. Game theory is part of the standard tool kit of ethology, ecology and evolution, and evolutionary game theory is a standard approach in economics. Network theory is applied across technological networks, social networks, genetic regulatory networks, and ecological networks. Methods and mathematics are often domain independent. But what about going beyond methods to encompass fundamental frameworks?

For one, metabolic scaling theory has proven to be very powerful in biology and is now proving itself in social and urban contexts. This November we are charting the evolution of complexity economics. SFI-affiliated researchers have contributed pioneering ideas and methods to the study of economic and market phenomena, including positive returns, the theory of money, agent-based models, zero-intelligence markets, maximum entropy, leverage cycles, and time series prediction. However in the last couple of decades the exponentiation of data and computer power, progress in algorithms, statistical physics, adaptive dynamics, and in neural, behavioral and cognitive science, suggests that a new complexity revolution is on the horizon. This is likely to transform all that we study at SFI, including the economy. In our forthcoming November meeting, by bringing together many of our deepest thinkers on markets, trade, decision theory, behavioral economics, algorithms, and our larger community of complexity researchers, we aim to foster a Kuhnian atmosphere of paradigm-busting possibility. See you there.