Why Computation, Not Physics, Explains Reality

Physics picked the easy problems

Newton succeeded in mechanics because he studied solid objects with simple trajectories, not fluids with turbulent behavior. Physics formalized mathematics in the 19th century, leading to three dominant 20th-century theories: general relativity, quantum mechanics, and statistical mechanics. However, these theories left many phenomena unexplained and arbitrary (e.g., why muons are 206 times heavier than electrons).

Three dominant theories of 20th-century physics

String theory and supersymmetry failed

Attempts to unify physics through string theory and supersymmetry never connected with experimental observations and left many things feeling arbitrary. This led Wolfram to pursue a different approach: studying the computational universe directly rather than reverse-engineering the observed world through mathematics.

What is computation?

Computation is the following of rules and observing what happens. It can be as simple as arithmetic or as complex as cellular automata. Wolfram calls the study of simple rules and their consequences 'ruliology'—the core basic science. Rules can be applied to networks, cells, symbols, or any system; the key insight is that even trivial rules produce arbitrarily complex behavior.

The computational process

Cellular automata: simple rules, infinite complexity

A cellular automaton is a row of cells (black or white) that evolve line-by-line. Each cell's next state depends on its current color and its neighbors' colors, looked up in a table. Elementary cellular automata have exactly 256 possible rules. Rule 30, derived from the binary representation of 30, starts with one cell and produces an extremely complicated pattern: one side is regular, but the center column looks completely random and is unpredictable without running the simulation.

256

Elementary cellular automata rules

All possible rules for the simplest cellular automata

Computational equivalence across models

Different computational models—cellular automata, Turing machines (invented 1936), combinators (invented 1920), register machines—seem very different but are all equivalent. You can emulate a Turing machine with a cellular automaton and vice versa. This is analogous to running the same software on different computer hardware. The remarkable discovery is that this equivalence extends beyond abstract models to the physical world itself.

Different computational models, all equivalent

Snowflakes and shells grow by simple rules

Snowflakes grow by aggregating ice pieces onto a structure following very simple rules: grow arms, then arms grow arms. Mollusk shells spiral and display pigmentation patterns that are exactly cellular automata—each cell either produces pigment or doesn't at each step, resulting in elaborate patterns. These biological growth processes are fundamentally computationally simple.

Biology lacks theoretical foundations

Biology has traditionally been observational, not theoretical. The only two major theories are natural selection and molecular biology (DNA). Darwin believed there must be an abstract law governing evolution's increasing complexity, but none was found. Wolfram applied machine learning insights to evolution: if you 'bash' a neural net hard enough, it learns; similarly, simple computational systems can evolve to achieve complex fitness objectives.

Two foundational theories in biology

Computational irreducibility: the secret of complexity

Even very simple programs produce extremely complicated behavior. The key discovery is computational irreducibility: if you want to know what a simple program does after a million steps, you often cannot jump ahead with a formula—you must run all million steps. This is a fundamental limitation of science. You cannot always predict outcomes; you must execute the system and observe. This principle descends from Gödel's Theorem and the principle of computational equivalence.

A fundamental shift in understanding science

Life is orchestrated irreducible computation

Living tissue is neither purely liquid nor solid; it is orchestrated at the molecular level. Molecules are specifically and actively transported, fitting together in precise ways. Life is a 'technology stack' built through evolution—pieces of irreducible computation fitted together like random rocks forming a wall. Biological evolution works because fitness objectives are computationally simple compared to the power of the underlying irreducible computation.

The universe is a network of discrete space atoms

The universe is made of discrete atoms of space (not chemical atoms, but indivisible units). These atoms have no intrinsic properties—only their relationships to each other matter. The universe is a graph or network. Space and everything in it are features of this network. This is the 'data structure' of the universe.

Time is progressive network rewriting

The 'algorithm' of the universe consists of rules that rewrite the network: whenever a piece of network looks like this, replace it with that. This rewriting process IS the progression of time. The universe continuously applies these rules to its network structure. With roughly 10^400 atoms of space being rewritten, the aggregate effect of these microscopic processes emerges as general relativity and Einstein's equations.

How time and physics emerge from network rewriting

Quantum mechanics from multiple histories

Different parts of the network can be rewritten in different orders, creating many possible threads of history. There is no single linear history—many paths are possible. This multiplicity of histories is what leads to quantum mechanics. In classical physics, definite things happen; in quantum mechanics, many paths are followed, and we observe the aggregate effect as probabilities.

The Ruliad: all possible computations entangled

Wolfram was confused about why our universe has one particular rule and not another. The answer: all possible rules are being used. Just as there are different paths of history for one rule, there are different histories for different rules. The Ruliad is the entangled limit of all possible computations—a unique, inevitable, abstract object. It is the collection of all possible machines and abstract systems doing computation, woven together because different machines may produce the same result.

All possible computations

The Ruliad

A unique, inevitable object that necessarily exists

Three kinds of space

Physical space is what we experience. Branchial space is the space of possible histories associated with quantum mechanics—different quantum branches. Rulial space is the space of different computational perspectives: different minds or observers at different locations perceive the universe differently. Minds closely aligned are close in rulial space; cats and dogs are further away; weather is much further away. This is analogous to physical space where proximity determines shared perception.

Three fundamental kinds of space in the model

We are observers embedded in the Ruliad

Observers like us are made of the same stuff as the Ruliad and embedded within it. The key question is: what do we perceive given our nature? Two characteristics define us: we are computationally bounded (finite minds, cannot do arbitrarily elaborate computation) and we believe we are persistent in time (we experience a continuous thread of consciousness despite being made of different atoms at each moment).

Computational boundedness explains the second law of thermodynamics

At the microscopic level, molecular collisions are reversible—a movie of collisions looks the same forwards or backwards. Yet macroscopically, a smashed glass doesn't reassemble. This irreversibility (entropy increase) happens because the initial conditions are encrypted by irreducible computation. In principle, we could reverse it, but we are computationally bounded and cannot perform the irreducible computation needed to reverse the process. So we perceive it as random and believe in the second law of thermodynamics—a consequence of our computational simplicity relative to the underlying irreducible processes.

Pockets of reducibility within irreducibility

The Ruliad is full of irreducible computation, but within irreducible computation there are always infinite pockets of computational reducibility. Even though you cannot predict everything, you can always predict specific things. Observers like us sample particular pockets of reducibility. If everything were purely irreducible, we would not believe there were laws of nature. Instead, we find regularities—laws—that our finite minds can use to reduce complexity and tell narratives about the world.

Objective reality emerges from multiple observers

Each person has an internal view of how things work, yet we believe in an outside objective world that everyone agrees about. This belief in objective reality depends on there being many of us. If there were only one observer, there would be no clear notion of objective reality. Because we extrapolate that our internal perceptions are similar in others, and we all observe and agree about how the world works, we believe in objective reality. This is why quantum mechanics makes sense: we all agree on definite outcomes because we are close together in branchial space, just as we all agree on the night sky because we share one planet.

Laws of physics are inevitable for observers like us

A surprising metaphysical conclusion: the laws of physics that we perceive are not arbitrary features wheeled into the universe. Rather, for observers like us—computationally bounded and persistent in time—the laws of physics must necessarily have the form they do. If we were different kinds of observers, we might perceive different laws. But for entities like us, the laws are inevitable and derivable.

Computational irreducibility gives life meaning

If we could always predict what would happen, nothing would be achieved by the passage of time or the living of our lives. Computational irreducibility means that living our lives adds up to something—we execute irreducible computation. There is no way for something external to know what will happen; you must live the life to know. This is what gives life richness and meaning.

Free will emerges from computational irreducibility

Even though underlying rules are deterministic, computational irreducibility creates an irreducible gap between those rules and actual behavior. You cannot say 'I know those rules, therefore I know what the system will do.' You must follow the rules and see what happens. This irreducible gap is what provides free will—not freedom from determinism, but freedom from predictability.

Infinite discovery awaits

Within computational irreducibility, there are always infinite pockets of reducibility, each representing a discovery and a surprise. There will always be new things to discover. Not every invention has already happened. The role of humans is to decide which possibilities to pursue and what to find interesting.

The trade-off: predictability vs. capability

If you build an AI that truly uses deep computation, it will have computational irreducibility and can always surprise you. You cannot guarantee it will never do the wrong thing. The alternative is to insist on computational reducibility—making machines we fully understand—but then they will be very limited in capability. This is a fundamental trade-off: post-Industrial Revolution, we had machines with gears and levers we could understand. Computational machines break that pattern. We must choose between understanding and power.

Humans are unique in the ruliad

The whole bundle of things that make up the human condition—mortality, sensory experiences, consciousness, the crushing of vast input into a single thread of experience—is unique. AIs without mortality or certain sensory experiences are fundamentally different. Consciousness seems to be the feature of taking huge amounts of input data and reducing it to a single thread of experience and action. This is different from the weather, which also computes but does not have this feature of consensus action.

We will never be done discovering

Technology is taking what exists in the world and applying it for human purposes. Computational irreducibility tells us we will always find more things to apply. Humans have vestiges of natural selection—a drive to seek the new. The idea that we are done now is unlikely. There is an infinite amount to discover, and the human role is to choose which possibilities to pursue.

Foundational knowledge is possible

It is not obvious that we can understand the foundations of the world. We might never be able to think foundationally about things. But we can. We can drill down and understand the primitives of what is happening. Foundational thinking has been key to Wolfram's approach in science, business, and life.

Computation is the great paradigm of the 21st century

Just as formalization has been central to human progress—from language (the word 'rock' formalizes many rocks) to logic to mathematics—computation is a broader formalization of things in the world. It is a way of structuring thinking and building towers of consequences. The question is: how do we represent different kinds of things in computational terms? How do we build a computational language for describing the world?

The progression of formalization in human thought

The Copernican lesson: trust the math, not intuition

The Copernican Revolution taught us to trust mathematics over everyday senses. It seemed obvious the Earth was still and the sun moved, but math showed otherwise. Today we are learning that things are computational all the way down. We should think about the world in structured computational terms, not trust our intuitions about how reality must work.

The universe is not a running computer

A common confusion: if the world is computational, where is the computer running it? Models are representations of what the natural world does, not mechanistic descriptions of what it actually is. We don't imagine a little piece of software inside the Earth calculating gravitational equations. The computational model is a description of how the world works, not a literal mechanism.

Science reflects back on us humans

One might think science reveals a cold, inhuman universe. But the science Wolfram has done shows that science reflects back on us in important ways. Without an observer with definite characteristics, there is only uniform things to say. In a sense, there is nothing to say if there is not a human somewhere in the middle. The laws of physics themselves depend on the nature of observers like us.