Stuart Kauffman, a founder of complexity theory and co-founder of the Santa Fe Institute, argues that the deepest assumption underlying Western science — that the world is fully intelligible and deducible — is fundamentally wrong. If he is right, there is no “theory of everything” that can encompass the evolving biosphere, because biological evolution is an open-ended, non-deductive process that creates genuinely new possibilities that did not previously exist.
The birth of complexity theory and its limits
Kauffman invented random Boolean networks at age 23 to ask whether generic network dynamics could explain the ordered behavior seen in cellular differentiation. His 1969 paper showed that such networks exhibit ordered, critical, or chaotic phases, with critical behavior matching features of real biological cells. This was one of the founding contributions to complexity theory.
The Santa Fe Institute, founded in 1984, helped establish complexity as a formal field, studying agent-based models, phase transitions, percolation, and generic behaviors of complex adaptive systems. But Kauffman now sees a fundamental limitation: complexity theory remains entirely mathematical and symbolic, while actual organisms are physical systems whose causal properties cannot be captured by formal models.
Organisms as Kantian wholes
Kauffman draws on Kant’s Critique of Judgment to argue that living organisms are “Kantian wholes” — systems in which the parts exist for and by means of the whole, and the whole exists because of its parts. Your heart exists because it is part of you; you exist because of your heart.
This structure allows a non-reductive, non-circular definition of biological function: the function of a part is that subset of its causal properties that sustains the whole organism. Your heart’s function is pumping blood; its other causal effects (heart sounds, jiggling water in the pericardial sac) are not its function, but they are real physical consequences.
Darwinian exaptation and the indefinite
A Darwinian exaptation (Gould and Vrba’s term; Darwin called them “pre-adaptations”) occurs when a structure originally used for one purpose is co-opted for a new use. The tongue evolved for eating and was later used for speaking. The swim bladder, used for neutral buoyancy, evolved from the lungs of lungfish.
Kauffman’s critical insight: you cannot deduce the new use from the old use. There is no logical or mathematical relationship between pumping blood and the potential survival value of jiggling water in the pericardial sac. The set of possible uses of any part is indefinitely large — not merely unknown in principle but ontologically unlistable.
Uses of things have three properties: they cannot be put into correspondence with the integers (they are not enumerable), they have no ordering relationship (there is no metric on uses), and one use cannot be deduced from another. Kauffman calls this the “indefinite.”
New possibilities come into existence
The indefinite is not merely an epistemic limitation — it is ontological. Before swim bladders existed, it was not possible for a worm to evolve to live in a swim bladder. That possibility came into being only after swim bladders evolved. Similarly, the possibility of a tiger catching a gazelle did not exist before tigers and gazelles evolved.
This means biological evolution is not a search through a pre-existing space of possibilities. It is a process that creates its own adjacent possible — new uses and new structures emerge from what already exists, but they cannot be predicted or deduced in advance.
Technological evolution works the same way: the modem enabled the internet, which enabled eBay. You cannot get a crossbow before you have a bow. Most inventions are pre-adaptations of existing things. James Burke’s Connections series documents this pattern extensively.
The domain of non-entailing laws
Kauffman identifies a domain of “non-entailing laws” — a realm where no fixed equations or deductive framework can capture what is happening. This domain is the evolving biosphere. It parallels the ancient Chinese Tao: “The Tao that can be told is not the eternal Tao.” The biosphere that can be described now is not the biosphere that will become.
This stands in contrast to Plato’s Logos and Newton’s worldview, which assume all possibilities already exist in a pre-stated state space. In quantum mechanics, statistical mechanics, logic, and probability theory, the standard framework assumes all possibilities are given in advance. Biological evolution violates this assumption.
The cultural implication is profound: Western science has operated under the assumption of dominion — that we can master and control nature because it is a machine. If the biosphere is not a machine, the appropriate stance is participation, humility, and collaboration rather than control.
Why physicalism is insufficient
Kauffman is careful to state that nothing in biology violates the laws of physics. Physics must allow whatever biological structures exist. But physics alone cannot explain why those structures exist, because the universe is not ergodic above about 500 Daltons.
The number of possible proteins of length 200 amino acids is roughly 10^260. The universe, even operating at the Planck timescale with all its particles, could not make all such proteins in many times its current lifetime. Therefore, the universe has not sampled all possible complex structures.
Hearts exist not because the universe explored all possibilities and found them, but because life started, hearts were useful, and natural selection refined them over billions of years. Function — the use of X — is ontologically effective in explaining why things exist. Uses are not in physics; they are triadic relationships between a structure, a context, and an organism.
This means you cannot be a strict physicalist. You must appeal to uses and selection to explain the existence of functional structures, even though no physical law is violated.
Constraint closure and self-construction
In his 2000 book Investigations, Kauffman asked where the constraints in physical systems come from. Newton assumed boundary conditions, but those boundary conditions themselves had to be constructed. No constraint on the release of energy means no work; but sometimes it takes work to make the constraints.
Building on work by Maël Montévil and Matteo Mossio (2015), Kauffman developed the concept of constraint closure: a system of constraints where each constraint is constructed by thermodynamic work performed by processes governed by other constraints in the same set. The system constructs its own boundary conditions.
Gonen Ashkenasy at Ben-Gurion University created a peptide system that achieves catalytic and constraint closure — nine peptide structures that mutually catalyze each other’s formation. This is a self-constructing molecular system with no distinction between software and hardware, challenging the assumption that life requires separate informational and material components.
Quantum gravity and non-locality
Kauffman has developed a speculative theory of quantum gravity based on the premise that non-locality is fundamental (supported by loophole-free Bell test experiments since 2022). If non-locality is fundamental, then any theory assuming locality — general relativity, loop quantum gravity, string theory, the holographic principle — cannot be fundamental.
He adopts Heisenberg’s interpretation of quantum states as ontologically real potentia. Quantum measurement converts possibles to actuals. Using four mutually entangled particles, he maps von Neumann entropy (or mutual information) between entangled pairs into spatial distances, constructing a metric tensor via the Gram matrix. The result is Lorentz-invariant de Sitter space that satisfies the Bianchi identities.
Using Regge calculus, he models space-time as constructing itself from tetrahedra, with a modified cosmological constant that may unite dark energy, dark matter, and inflation, prevent black holes singularities, and address the Hubble tension. He acknowledges this needs independent verification by competent physicists.
Mind, potentia, and the cosmic
With Sudeep Patra, Kauffman explored whether mind exists in space-time. Some weak evidence (from Diederik Aerts) suggests quantum correlations exceeding the Tsirelson bound, which would imply processes not bound by the finite speed of light — and therefore not in space-time.
If what is not in space-time are potentia, and mind is related to potentia, then there may be something like a cosmic mind. If molecules in our bodies have quantum features (incomplete decoherence), we may be coupled to this cosmic mind. Kauffman acknowledges this is speculative but suggests it could be a scientifically grounded sense of the divine.
The world is not a theorem
Kauffman and Andrea Prokopenko showed that the axioms of set theory fail for uses of things. The axiom of extensionality requires listing all members of a set to prove identity, but the uses of a screwdriver or engine block cannot be listed. The axiom of choice requires well-ordering, but uses have no ordering. Without these axioms, you cannot construct numbers, equations, topology, or manifolds.
This means none of the mathematics of set theory can compute or deduce the evolution of the biosphere. “The world is not a theorem” — you cannot write down an equation and deduce the trajectory of biological evolution. Some other mathematics (perhaps category theory) may be needed, but the standard formal toolkit fails.
On artificial general intelligence
Kauffman argues that AGI in the Turing machine framework is impossible because computers cannot find affordances — they cannot perform the open-ended “jerry-rigging” that human minds and biological evolution do routinely. Human minds find new uses for things through multi-step processes where no step can be deduced from the previous one, and there is no landscape to search.
No one has achieved open-ended evolution in artificial life or general artificial intelligence on computers. Evolving organisms are embodied and embedded in the world; a cell with multiple attractors does work on the world and categorizes it without any internal representation.
Personal reflections and hopes
Kauffman is 86 and has been working on these ideas for 63 years. His forthcoming book Origins (expected April 2026) covers all of this material across 15 chapters, including Stueckelberg and the RNA worldview, the origin of life, agency, constraint closure, quantum gravity, and a final chapter titled “A Next Axial Age.”
He hopes the recognition that we do not have dominion over nature — that we are closer to the Tao than to the Logos — will lead to a cultural transformation toward participation, humility, and collaboration with the biosphere rather than its destruction.
He is involved in global soil restoration through OVIA, promoting David Johnson’s composting method, which restores soil, sequesters water, and likely sequesters atmospheric carbon dioxide.
He co-invented the idea of generating billions of random DNA or protein sequences and selecting for useful molecules — now the entire field of combinatorial chemistry.
When asked what question he would pose to an oracle, he said he wants to know what constitutes a good life and what flourishing means — not just for humans but for all species. He worries that modern consumer culture has lost sight of this question entirely.
His advice to students and to himself: trust yourself, find the question that is uniquely yours to ask, and pursue it. You cannot deduce the questions; they come from you.
