Digital physics is a family of speculative proposals in philosophy, holding that the universe is at its most fundamental level digital, informational or computational in character: that physical reality is either the output of a computer program or is itself a computational process. This metaphysical position is also known as pancomputationalism.
The hypothesis that the universe is a digital computer was first set out by Konrad Zuse in a 1967 paper and in his 1969 monograph Rechnender Raum ("Calculating Space"). The term "digital physics" was coined in 1978 by Edward Fredkin, who taught a graduate course under that name at MIT and later came to prefer the label "digital philosophy". Fredkin's collaborations with Tommaso Toffoli and Norman Margolus on reversible and conservative logic made MIT's Information Mechanics group the principal early institutional home of the programme.
Digital physics is not a single theory but a cluster of positions that differ over what is claimed to be discrete (space, time, state, or information), over which computational formalism is taken to be exact (cellular automata, Turing machines, quantum circuits, graph rewriting systems), and over whether the claim is empirical, metaphysical, or both. It overlaps with, but is distinct from, the simulation hypothesis, which concerns whether our universe is being run by agents in some other universe rather than whether it is computational at all.
Digital physics is not broadly accepted. No version of it has yielded a confirmed novel prediction, and existing discrete models are in tension with several continuous symmetries of established theory, such as rotational symmetry, Lorentz symmetry and gauge invariance. Deterministic classical variants also fall within the class of local hidden-variable theories constrained by Bell's theorem and the experiments that test it.
Terminology
Two claims are often run together but are logically independent.
Discreteness: that fundamental physical quantities take at most finitely many values, so that there are no genuinely real-valued magnitudes at the bottom level. On this view, apparent continuity is an emergent, coarse-grained appearance. Informational primacy, or digitalism: that information is ontologically fundamental and physical properties derive from it. This position was compressed by John Archibald Wheeler into the slogan "it from bit".
One can hold discreteness without informational primacy (many approaches to quantum gravity do), or informational primacy without discreteness (as in several information-theoretic reconstructions of quantum theory, which use continuous state spaces).
History
Precursors of pancomputationalism
The idea that number or structure is ontologically basic has Pythagorean antecedents, and Leibniz's work on binary arithmetic and a universal calculus is frequently cited as an ancestor. The more direct scientific lineage runs through the thermodynamic treatment of information: Leó Szilárd's 1929 analysis of Maxwell's demon, Claude Shannon's information theory, and Rolf Landauer's demonstration that logically irreversible operations carry an unavoidable thermodynamic cost, summarised in his dictum that information is physical.
Zuse and calculating space
Konrad Zuse, the builder of the Z3, proposed in 1967 that the physical behaviour of the universe is computed at a basic level by the universe itself, possibly on a cellular automaton, which he called Rechnender Raum. Zuse is generally credited as the first pancomputationalist in the ontic sense.
Information Mechanics at MIT
Edward Fredkin developed the position independently and, from the late 1970s, encouraged a group at MIT's Laboratory for Computer Science around Toffoli and Margolus. Their central technical concern was whether computation could be made compatible with the reversibility and conservation laws of microphysics; the Fredkin gate and Toffoli gate came out of this work, as did the Billiard-ball computer model. Fredkin later formulated his position as digital mechanics, a proposal that physics is a reversible universal cellular automaton.
The 1981 MIT conference on Physics and Computation was a turning point. Richard Feynman's contribution argued that a classical local system such as a cellular automaton cannot efficiently reproduce quantum correlations, and proposed quantum simulators instead. This argument both damaged the classical version of digital physics and helped launch quantum computing.
1980s and the informational turn
David Deutsch formulated in 1985 what is now called the Church-Turing-Deutsch principle: that every finitely realisable physical system can be perfectly simulated by a universal model computing machine operating by finite means. Deutsch's own conclusion was that the appropriate universal machine is a quantum, not a classical, one. Stephen Wolfram independently advanced a related claim in the same year, arguing that physical processes are bounded by the capabilities of universal computation.
Wheeler, from the late 1980s, argued that every physical entity derives its existence from binary yes-or-no answers elicited by acts of observation, a position he condensed as "it from bit". His formulation was influential well beyond digital physics, and is frequently invoked by researchers who do not accept discreteness at all.
Twenty-first century Stephen Wolfram's A New Kind of Science (2002) argued that simple programs generically produce complex behaviour, that this suffices to account for complexity in nature, and that a simple rule underlies fundamental physics. Reception among physicists and computer scientists was largely critical, with reviewers objecting that the novel claims were unproven and the proven claims not novel.
Three developments broadened the field after 2000. First, Seth Lloyd and others recast the computational-universe claim in quantum information terms. Second, the holographic principle and the Bekenstein bound gave the idea of finite information density a foothold in mainstream quantum gravity. Third, Gerard 't Hooft and Wolfram each produced sustained technical programmes, respectively a deterministic cellular-automaton interpretation of quantum mechanics and a hypergraph-rewriting model of spacetime.
Varieties
Cellular-automaton models
The oldest and best-known form takes the universe to be a cellular automaton: a lattice of cells, each in one of finitely many states, updating in discrete steps according to a local rule. For this to hold, all fundamental magnitudes must be discrete, and space and time must be either fundamentally discrete or emergent from the automaton's processing. Zuse, Fredkin, Toffoli and Margolus all worked within this frame.
'T Hooft's cellular automaton interpretation is the most technically developed contemporary version. It treats quantum mechanics as a tool for analysing an underlying deterministic system, constructs a mapping between ontological ("beable") states and quantum basis states, and addresses Bell's theorem by rejecting statistical independence of measurement settings from hidden variables; a position known as superdeterminism.
Computable ensemble theories
Building on Zuse's assumption that the history of the universe is computable, Jürgen Schmidhuber argued that the simplest explanation of the universe is a short program systematically executing all possible programs, and developed algorithmic priors over computable universes. Gregory Chaitin emphasised algorithmic incompressibility as a limit on any such account.
Max Tegmark's Computable Universe Hypothesis is a restriction of his Mathematical Universe Hypothesis: the mathematical structure that is the external physical reality is defined by computable functions. Tegmark notes that this is a substantive and possibly false restriction, since most mathematical structures are not computable.
Quantum-computational models
On the quantum version, the universe is not a classical digital computer but a quantum one, manipulating qubits rather than bits. Lloyd derived quantitative bounds on this reading: applying the Margolus-Levitin theorem and the Bekenstein bound, he computed that the universe can have performed at most about 10120 elementary operations on about 1090 bits since the Big Bang.
The quantum version is considerably less radical than the classical one: because it does not eliminate continuity, it can be read as a reformulation of standard quantum mechanics in the vocabulary of quantum information, with no change in empirical content. That is both its strength and, for critics, its principal weakness.
Related but distinct are the informational reconstructions of quantum theory: Anton Zeilinger's foundational principle, QBism, and the axiomatic derivations of Chiribella, D'Ariano and Perinotti, which derive the Hilbert-space formalism from information-theoretic axioms without asserting that the world is digital.
Ur-theory
Carl Friedrich von Weizsäcker's theory of ur-alternatives builds quantum theory axiomatically from binary alternatives (Ure), the simplest possible empirically decidable questions. Weizsäcker used the theory to argue for the three-dimensionality of space and to estimate the entropy of matter falling into a black hole. It is an informational rather than a discrete-spacetime proposal.
Graph- and network-based models
Later work replaces the fixed lattice with a dynamically rewritten discrete structure, which sidesteps the objection that a lattice imposes preferred directions on space. The Wolfram Physics Project, launched in 2020, models space as a hypergraph transformed by simple rewriting rules satisfying the Church-Rosser property; arguing that special relativity, general relativity and features of quantum mechanics emerge in appropriate limits.
Discrete structures also appear in mainstream quantum gravity without any computational commitment: causal set theory, loop quantum gravity spin networks, and causal dynamical triangulations all posit discreteness while remaining agnostic about whether the universe computes.
Simulation-based versions
A minority of proponents add that the computation is being run deliberately by agents outside our universe. Nick Bostrom's simulation argument is logically independent of digital physics, it is a statistical argument about the relative frequency of simulated and unsimulated observers, and does not require the universe to be discrete. Conflation of the two is common in popular coverage and is a recurring source of confusion in the literature.
Core arguments
Against the physical continuum
A recurring motivation is the claim that the continuum is a mathematical idealisation with no physical warrant: that infinitely precise real-valued magnitudes are never measured and never needed, and that admitting them imports infinities into physics gratuitously. Wheeler argued in this vein that there is no evidence for the existence of the number continuum in physics.
Finite information density
The Bekenstein bound places an upper limit on the entropy (hence on the information) that can be contained in a region of given size and energy, and the holographic principle indicates that the degrees of freedom in a volume scale with its bounding area rather than its volume. Proponents treat these as independent evidence that information content is fundamentally finite. Critics respond that finite entropy does not entail a discrete state space, and that the holographic principle is a statement about degrees of freedom, not about pixels.
Criticism
Continuous symmetries