The Santa Fe Institute has launched an ambitious project in which leading complexity scientists comment on “foundational” papers. However, many of them are not foundational in the sense that they started a line of research that led to the complexity science of today; some of them are even foundational for lines of research that are rather distinct from complexity science. I will list five IMHO blatant omissions. So this blog post is, as Max Schich would put it, the Salon des refusés corresponding to the Foundational Papers in Complexity Science.
As a side note, I realize the requirement that publications should be journal articles puts many constraints. Maturana & Varela’s Autopoiesis stuff, Torsten Hägerstrand’s spatial diffusion models from the 1950s, Georg Simmel’s theorizing of social complexity and von Neumann’s self-replicating automata, were all first published as books or book chapters.
AA Bogdanov, 1913. Тайна науки [The mystery of science.] Cовременник [The Contemporary] 8:161–82.
Bogdanov (as blogged about elsewhere) was a timeless visionary. A philosopher—and, yes, as the saying goes, science needs philosophy as much as birds need ornithology—but still, sometimes, they get it right.
Bogdanov’s “tektology,” the “universal organizational science,” was a blueprint of cybernetics, complexity, and systems sciences 50-100 years later. However, even more interesting than his tektological monographs is the journal article “The mystery of science,” in which Bogdanov envisioned the science of the future in incredible depth and precision: the inevitable interdisciplinarity, its holistic nature, systems, feedback loops, contingency, and the fact that we are human after all and thus need to build any new science with what we have.
T Schjelderup-Ebbe, 1913. Hønsenes stemme [The voice of hens]. Naturen 37:262–276.
Could a paper like this happen today? A schoolboy spent his summer break observing the hens at a relative’s chicken farm. In his late teens, idolizing the scientists of the day, he wrote down these observations as a scientific paper. This paper introduced a concept—animal dominance hierarchies—that is an active field of investigation to this day. From the publication of the paper, the author, Thorleif Schjelderup-Ebbe, strived for academic acceptance but got it too late for a career in science.
Maybe because it was published in Norwegian, this first paper is somewhat forgotten, and his contemporaries celebrated the centennial of another of his papers in 2022. I have yet to read the 1922 paper, but it did not build on any new data or observations, so the vote for the most foundational paper goes to Hønsenes stemme. And those who think it sounds too niche to be a complexity classic, the imperfect dominance hierarchies described by Schjelderup-Ebbe are close to Herb Simon’s “near decomposability” concept.
M Wertheimer, 1923. Untersuchungen zur Lehre von der Gestalt [Laws of organization in perceptual form] II. Psycologische Forschung 4:301–350.
This is one of the early defining works of the Gestalt school of psychology and the first paper to use actual data to argue the old Aristotelean catchphrase that the whole is bigger than (or, different from) the sum of its parts. The Gestalt theorists have one more reason to be thought of as progenitors of complexity science: They (Wolfgang Köhler in particular) saw their theory as applicable far beyond psychology, influencing a new generation of interdisciplinarians at a time when psychology was more in the limelight than today.
JL Moreno, HH Jennings, 1938. Statistics of social configurations. Sociometry 1(3/4):342–374.
I don’t really subscribe to calling network science and other primarily structuralist dominions (semiotics, etc.) complexity science. It would be epistemologically much cleaner if the latter could be exclusively integrative and holistic, but it’s such a socially constructed field, so okay then. Moreno and Jennings’ paper was from the future. It used random network null models, defined structure as how reality differs from these, and explained function from this network structure. In short, it anticipated network science of 60-70 years later (except maybe degree distributions, but Helen Jennings had already used them in a paper from the year before).
JW Forrester, 1958. Industrial dynamics. Harvard Business Review 36:37–66.
Holistic, systems-oriented, computational science has never had a bigger impact than Limits to Growth had 50+ years ago. Of course, this was not only because of Jay Forrester’s simulation framework (first presented in the above article)—it happened at the height of the 1970s environmentalist movement that also brought the largest political manifestation in history—Earth Day 1970. But still, we definitely wouldn’t have the Earth systems science or climate science of today if it wasn’t for Forrester’s work, maybe not UN’s SDGs either.
Petter Holme 