Can space and time emerge from simple rules? Stephen Wolfram thinks so. | World Science Festival

Wolfram had a conflict with Caltech administration over intellectual property when he began developing software companies while holding a faculty position at age 21-22; the institute prohibited simultaneous activities in academia and commerce.

“I was a faculty person at Caltech when I was 21, 22 years old... the other thing that happened was I developed this big computer system, and in those days, being a professor and doing things like starting companies, they were not compatible activities, and so that ended up with me having this sort of big flap with Caltech, and so I quit.”

Wolfram wrote an inflation-like paper with Rocky Cole in 1978-1979 that remained only a footnote because the required supercooling was implausibly large; later physicists like Alan Guth found inflation sufficiently interesting to publish prominently despite similar implausibilities.

“I wrote this paper with Rocky Cole back in, when was it? 1978, 1979, where we were looking at the expansion rate of the early universe... we realized that there could be an exponential phase of the expansion of the universe... the universe would have to super cool by 10 orders of magnitude... I was like, this is not going to happen.”

Mathematical incompleteness (Gödel's theorem) is conceptually similar to computational irreducibility: both indicate fundamental limits to predictability and knowledge in formal systems.

“This is a complicated story. It really dates back to things like Gödel's theorem from 1931, which is a version of this kind of statement.”

The Institute for Advanced Study in Princeton, under leadership that remembered John von Neumann, explicitly didn't worry about intellectual property issues when Wolfram joined, allowing him to freely pursue both technology and foundational science.

“I then went to the Institute for Advanced Study in Princeton, which had the feature that its then chairman of its board said, 'We gave away the computer when Von Neumann kind of died, so all this intellectual property stuff, we're just not worried about that.'”

Singularity theorems in general relativity (stating that sufficient mass inevitably creates singularities) have an analogue in meta-mathematics: sufficient proof density inevitably creates decidable theories.

“In general relativity, we know that there are singularity theorems that tell one, when there's enough energy and momentum, it's inevitable you'll form a singularity. Okay, so the analogous thing... is when there's a high enough proof density, you will inevitably form a decidable theory.”

The key properties that define us as observers—computational boundedness and persistent single thread of conscious experience—are sufficient to derive that we must perceive general relativity, quantum mechanics, and the second law of thermodynamics.

“If you're an observer who is roughly like us in the sense that we're computationally bounded, believe we're persistent in time... it is inevitable that those observers must conclude that second law of thermodynamics, general relativity and quantum mechanics are the way the universe works.”

The structure of multiple possible proofs of the same theorem may encode quantum-like phenomena in meta-mathematics; studying the topology of proof spaces could reveal quantum-mechanical-like behavior in mathematics.

“In mathematics, you're often asking the question, can I get a proof of this thing?... there may be many proofs. There will be, in general, many proofs of the same thing. So in a sense, the analog of quantum mechanics is that there are many proofs of the same thing... The topology of proof space is something rather interesting and rather unstudied.”

Quantum amplitudes should not be packaged as complex numbers with both magnitude and phase; instead, the magnitude comes from counting the number of paths reaching a point, while the phase corresponds to position in branchial space.

“My guess is that another mistake of packaging that was made about a hundred years ago was to say that when we think about quantum amplitudes, that they have a magnitude and they have a phase as complex numbers... I think that's a mistake. I think the magnitude comes from a completely different place than the phase. The phase is this position in branchial space. The magnitude has to do with counting the number of paths.”

The question of which rule the universe follows has an answer within the ruliad framework: the universe applies all possible rules simultaneously, with observers perceiving consistent subsets based on their properties.

“We've got the rule for the universe, it's like the next question is, well, why did the universe get this particular rule and not another? And quantum mechanics is tip off about this because in quantum mechanics we're saying there is a particular rule, but it's being applied in all possible ways. So, the obvious limit to that is, well, why not just apply all possible rules?”

Quantum field theory lacks an adequate fundamental model for what particles actually are, instead describing them only through asymptotic S-matrix states; this foundational gap is comparable to uncertainties in the discrete model about particles.

“quantum field theory also doesn't have a good model for what particles are...It's like nobody has any idea how to represent a whole proton in quantum field theory.”

The Towers of Hanoi puzzle has a multi-way game graph with Sierpinski-like fractal structure, showing how combinatorial games exhibit the same kind of branching structure as quantum mechanics.

“the Towers of Hanoi puzzle where you're putting these disks, stacking these disks up. If you make the game graph of the Towers of Hanoi puzzle, it is a fractal Sierpinski kind of structure. So it has a great big hole in the middle.”

Continuous mathematics with parameters like T in differential equations allows one to 'jump ahead' in time by choosing any T value, but discrete computational systems lack this ability; one must iterate through each step sequentially.

“As soon as you're saying something like time is a continuous thing, where you can pick any possible value of time, that's not something where you're just following discrete rules... In the computational way of thinking about things, time is the result of the continued application of rules, and there's no guarantee that you can just jump ahead.”

The inflation scenario solves important cosmological problems (like homogeneity) through slightly implausible assumptions (fine-tuned initial conditions), yet the scientific payoff is so large that this tradeoff is worth making.

“When people really trotted out inflation, what was happening there was people said, the result is so worthwhile that even though it requires a slightly implausible assumption, it's worth making the implausible assumption because the payoff is so big.”

In mathematics, higher-level theorems can be proven without deriving them from microscopic axioms, just as fluid mechanics works without deriving everything from molecular dynamics.

“Most of the time when you think about the Pythagorean theorem, you're not having to think down to that level of those microscopic axioms. And it's the same kind of thing, I think, as in doing fluid mechanics. A lot of the time you can just work in terms of fluid mechanics.”

Dick Feynman told Wolfram that Feynman diagrams were 'a stupid idea' that required too much computation; there should be a better way to analyze quantum field theory without laborious diagram calculations.

“Dick Feynman would say, 'This Feynman diagram idea is just a stupid idea. The fact that it takes all this effort to compute all these things, it's just the wrong way to do it. There's got to be a better way to do it.'”

The Feynman checkerboard (alternating left-right moves on a discrete lattice with a factor of -i for direction changes) produces solutions to the Dirac equation when all possible paths are summed; this shows how relativistic quantum mechanics can emerge from discrete lattice walks.

“There's a thing that strangely enough is a little trick that Dick Feynman told me one day... the Feynman checkerboard... you just have particles that either go left or right... every time a particle was turned from left to right, there's a factor of -i... when you add up all the possible paths, the thing you get is the solutions to the Dirac equation.”

The electron mass is observer-dependent: different observers measuring the electron at different energy scales will measure different masses due to renormalization group effects.

“The mass of the electron is 0.511 MEV or whatever it is. But that is only for observers who have certain properties at zero energy... if you're an observer, if you're looking at the thing on a certain length scale, an observer looking at it on a very large length scale will conclude one mass, on a different length scale will conclude a different mass.”

Einstein believed in the 1910s-1920s that space must be discrete but lacked the mathematical tools; historically, Bohr and Heisenberg also believed space was discrete but couldn't make it work with relativity, so abandoned the idea and never published their thoughts.

“Einstein believed, even has a nice letter he wrote in 1916, where he says, 'In the end, it will turn out that space is discrete, but we don't have the mathematical tools to analyze that yet.' So, I can say Einstein believed this. Bohr believed it. Heisenberg believed that space was discrete... None of them could make a model that was consistent with relativity... They published nothing about this stuff.”

Quantum randomness doesn't indicate lack of determinism in the underlying system; rather, the multi-way graph of all possible computation histories is completely determined, but which branches we sample as observers is not determined—randomness is subjective observer uncertainty.

“In these models, everything is kind of determined, at least at the level of this multi-way graph of all possible threads of history. The structure of all possible threads of history is completely determined. What is not determined is where we are as observers across Branchial space and so on. Which branches did we actually sample?”

The universe can be represented as a hypergraph where space itself is composed of discrete atoms connected in a network structure, and time emerges from the progressive rewriting of this hypergraph according to simple local rules.

“Space and everything in it is represented by this hypergraph, just a thing where you have these atoms of space that are just discrete points. Those are just nodes in this hypergraph... Time is something very different from space in these models. Time is the progressive rewriting of that hypergraph by using some rule.”

Particles and black holes may be fundamentally similar topological features of spacetime that support identity-preserving motion, suggesting a deep connection between elementary particles and black holes beyond string theory duality.

“Curiously, black holes are the same kind of thing. You can take a black hole and move it around and it'll still be that same black hole. And maybe there's some... My guess is there is a close relationship between black holes and particles like electrons and so on because among other things in our models, it's all just features of the structure of space.”

In the discrete hypergraph model, black hole singularities are not infinities requiring mathematical removal but locations where time literally stops because no further hypergraph rewriting rules can be applied, making them fundamentally different from continuum singularities.

“In our models, what happens is that you get something where time literally does stop because you are thinking about rewriting this hypergraph and at some point you realize you've got a structure of hypergraph where there is no more rewrite that applies.”

Sub-time is an infinite accumulation of virtual processes (closed timelike loops) that resolves to a finite observable time, making it a distinct type of infinity from the infinities of virtual loops in quantum field theory, yet mathematically parallel.

“there's a notion of what I've been preliminarily calling sub-time, which is the thing that you get if you actually are going through and enumerating...the state that we got to is exactly the same as the state that we started from. So from the point of view of an observer, those states should be. There is a sub-time that goes infinitely, but observable time is only finite.”

Dark matter is not a substance but a macroscopic manifestation of the microscopic structure of spacetime itself, analogous to how heat in the 1800s was once thought to be a fluid (caloric) before being revealed as the motion of discrete molecules.

“My irony of scientific history is the thought that dark matter is really the spacetime analog of heat. It's a macroscopic manifestation of the microscopic structure of spacetime... just like heat turned out to be the motion of discrete molecules, dark matter might be the collective effect of spacetime structure.”

The spatial homogeneity of physical space (where we can move objects around and they remain the same) is analogous to homogeneity in meta-mathematical space, where different mathematical frameworks (algebraic, geometric, category-theoretic) can be related by dualities.

“In physical space, one of the notable facts is that space is somewhat homogeneous and things like motion is possible... In mathematics, in meta-mathematics, different places in meta-mathematical space are different kinds of mathematical theories... one of the things that's been found in modern mathematics is these remarkable dualities... The claim would be that the homogeneity of physical space is analogous to a homogeneity in meta-mathematical space.”

Getting Einstein equations to emerge from discrete rewriting rules requires assuming surprisingly little—not explicit energy conservation but only the topological structure of space and local update rules—suggesting the equations are more robust and fundamental than traditionally conceived.

“It's surprisingly little. The underlying idea is that space and everything in it is represented by this hypergraph, just a thing where you have these atoms of space that are just discrete points.”

The hypergraph model allows for faster-than-light signaling if one could perform infinitely complex engineering to select the correct causal edges at each step, analogous to how a molecular could traverse a room faster than sound speed by riding molecules and choosing the right one at each collision.

“if you could do infinitely good engineering, which you never can, then you could go faster than light in these models... if at every collision of those molecules we were able to figure out which molecule should we choose to piggyback on next, then we could go across the room at the speed of sound.”

Dimension is not a fixed property of the discrete model but can fluctuate around the large-scale average dimension; these dimension fluctuations are a novel prediction that could be observable through effects on cosmic microwave background or gravitational lensing.

“Nothing guarantees an integral number of dimensions, yeah. Right. And that can fluctuate... would really love to have an analog of the standard continuum, Friedmann-Robertson-Walker metric for the early universe... that includes dimension as a dynamical parameter, as well.”

If an advanced AI system could think far faster than humans and had different computational properties, it could arrive at a radically different understanding of physics and reality that would be impossible for humans to communicate with.

“Some AI system in the future analyzes observations of the world and comes to a radically different picture of what the laws of physics are and what reality is... Yes, but there's a problem with that. We wouldn't be able to communicate with that AI system.”

Because we are large compared to the elementary length scale of space and span many different quantum branches, we average over quantum branching and perceive definite classical states just as we average over molecular motion to perceive definite fluid properties.

“In the case of quantum mechanics, it's a little bit more subtle because we're big. We span many different paths of history. So, our minds effectively have the same branching and merging of paths of history going on in them. But we are taking a big sample out of Branchial space and that's why we end up believing that definite things happen.”

Minkowski's unification of space and time into a single mathematical structure (via the quadratic form T² - X²) was mathematically elegant but physically mistaken; in reality, space and time are fundamentally different, and Lorentz transformations are only an emergent approximation in the large-scale limit.

“I think there was a mistake made back in 1919 when Minkowski said, 'Let's take space and time and isn't it cool that we can make a quadratic form?' ... I don't think it's physically missing... It's a wrong direction because it packages together two things that are really... Space and time in our common experience don't seem like the same kind of thing.”

The Ricci curvature emerges from examining the growth rate of volumes in the hypergraph structure: the correction term to simple volume growth that accounts for non-flat geometry is precisely the Ricci scalar that appears in Einstein's equations.

“If you say, well, look at the growth rate of this ball of nodes as you go a certain graph distance out, you say the growth rate is like R to the D... Well, turns out that the correction term, if you just look at this growth rate thing, the correction term is the Ricci scale of curvature, which interestingly is something that shows up in the Einstein equations.”

Energy in the hypergraph models is defined as the flux of causal edges through space-like hypersurfaces, and momentum is the flux of causal edges through time-like hypersurfaces.

“Energy in these models is the flux of causal edges through space-like hypersurfaces... Momentum, for example, ends up being the, if you have a time-like hypersurface, it's just the flux of causal edges through time-like hypersurfaces.”

The weighting factor e^(iS/ℏ) in the path integral emerges from the fact that energy-momentum (activity in the network) deflects paths in branchial space, changing phases; it is not put in by hand but emerges from the dynamics of the hypergraph.

“It comes from the fact that it is part of the equation of motion in branchial space... Yes. It emerges. In the same way that in general relativity... there's a bunch of mathematical versions of this. But the rough qualitative picture is if stuff is happening in the network, then the shortest path from here to there is going to be affected by stuff happening in the network.”

The early universe started in a high-dimensional state where everything was causally connected; as it expanded, it cooled down to lower dimensions, explaining why we observe three spatial dimensions without needing inflation theory.

“The beginning of the universe, it's probably infinite dimensional and it probably, you have some little tiny network and it's effectively, everything is connected to everything else. And gradually as the thing expands, you can kind of start thinking that things are less connected, and so the thing is sort of cooling down to being, well, we don't know why it's three, but a finite dimensional universe.”

The dimensional stability of numerical relativity depends on the updating order chosen; pathological updating orders produce unphysical evolution, while observer-consistent orders produce stable time evolution, making the updating order selection an expression of observer perspective.

“it matters what the updating order is...you can have a pathological updating order, which can make crazy things happen. But that pathological updating order isn't the one that observers like us...would ever see happening in the universe.”

The principle of computational equivalence states that when you move beyond systems with obviously simple behavior, you generically find computations as sophisticated as computationally possible, meaning most natural systems achieve the same level of computational sophistication.

“The principle of computational equivalence says, when you get above things whose behavior is obviously simple, you will generically have computations that are as sophisticated as they could be.”

Simple laws and mathematical elegance are predictive and reliable in physics because computational irreducibility means you cannot use complex mathematical methods to short-cut physical reality; the physics that emerges from observers like us must manifest as simple laws operating on large scales.

“the simple laws being the right laws isn't true in biology, for example. In biology, the typical experience is the most complicated possible explanation is sort of the right explanation...I think that's a place where this idea of being able to separate the thing that's happening in this particular situation from everything else that happens just doesn't work.”

Breaking away from hundred-year-old traditions in mathematical physics requires seeing how things work from outside, not caring that much about the conventional payoffs, and maintaining intellectual freedom to pursue unexpected directions.

“To break out of the a hundred year tradition of the particular way of doing physics... you kind of have to, A, you have to have seen the outside, so to speak. I think another feature is you have to not care that much.”

The assumption that we can do controlled experiments—isolating a phenomenon from the rest of the universe—is not self-evident and would not hold in a universe filled with spacetime singularities; this is a contingent fact about our universe that enables scientific inquiry.

“We believe in doing science that we can choose the experiments we do...In a deterministic universe, it could be that every experiment we do, we are fated to do, so to speak. But we somehow have this belief that we can do these arbitrary experiments. We also have beliefs like the belief that we can do a controlled experiment. We can do an experiment where the thing that happens here isn't affected by the things that are happening everywhere else in the universe.”

In discrete spacetime models with finite-information-content update rules, a finite speed of light emerges inevitably; infinite speed of light would require infinitely complex rules, which contradicts the principle that the universe must have finite rules to be comprehensible.

“As soon as you have that finite information content of the rule, and you have the idea that things are made of this big graph of discrete structure, then you would never have infinite speed of light... it's a fundamental fact about science that the universe doesn't seem to have infinitely complex rules. If the universe had infinitely complex rules, every different particle in the universe would be doing its own thing. And we wouldn't get to talk about laws of nature.”

The second law of thermodynamics is not a fundamental law but an observer effect arising from computational boundedness—computationally unbounded observers would see the complexity rather than perceiving randomness.

“we're going from this sort of maybe simple initial condition through this irreducible computation to something which we, as computationally bounded observers, just sort of throw up our hands and have to say, 'It looks random to us'... For example, we wouldn't believe in the second law of thermodynamics. If we were not computationally bounded.”

Computer simulation provides a major advantage for the discrete approach: simulating unexpected phenomena computationally and then figuring out the theoretical explanation afterward.

“we can simulate stuff on computers, and that means that we can get an idea, which is, for me, the kind of meta approach that I've taken. You simulate things on a computer, it does a lot of stuff you don't expect, then you try and figure out what the big picture is”

Dark energy should be understood as a 'zero' reference point for activity in spacetime, not as negative mass; the baseline activity necessary to knit together the structure of space must be subtracted before identifying excess activity as dark energy.

“My current guess which might be wrong is dark energy, which is less politely called negative mass matter or negative mass stuff is what is associated with the zero kind of thing... there's a certain amount of activity that's necessary to knit together the structure of space. Then the things that we are looking at with actual massive objects sitting in space and so on, those things are above that zero.”

Space-like, time-like, and branch-like separations between events are all distinct in the multi-way causal graph, and the multi-way causal graph as a whole represents all relationships between events in the universe.

“You can have this whole graph that shows the causal relationships between things in space, in branchial space, and through time... they can be space-like separated... time-like separated... they can also be what we call branch-like separated, where they are occurring on different branches of history.”

The holography principle (AdS/CFT duality and similar phenomena) naturally emerges in the discrete model through the multi-way causal graph structure, where quantum mechanics is a projection in the branchial (branching history) direction and general relativity is a projection in the spatial direction.

“The multi-way causal graph is this knitting together of things that are associated with space-like separation and things that are associated with branch-like separation. Quantum mechanics is one projection of this in essentially the branchial direction. And general relativity is another projection of this essentially in the spatial direction.”

In the discrete model, particle mass emerges from the interaction of particles with the background structure of spacetime, analogous to the Higgs mechanism where mass comes from interaction with the Higgs field.

“Mass is associated with the amount of interaction with the background structure of space, which is a little bit like the Higgs mechanism... there's a background Higgs field everywhere and the mass is how much the particle is interacting with that background Higgs field.”

The physics project succeeded better than Wolfram expected, partly because he wasn't initially focused on it; maintaining distance from prior expectations allowed him to be surprised by results.

“In other words, in some sense, the fact that our physics project has worked as well as it has is really cool, I'm really excited about it, but it was not what I was expecting to be doing at this point in my life... It was just sort of a bonus feature that was possible because I'd built a bunch of other things and then I kind of decided, what the heck, I might as well try doing this because I'd been meaning to try for 30 years.”

In the discrete hypergraph model, multiple possible rewriting sequences create a multi-way graph where branches can not only diverge but also merge back together, unlike continuous systems where two identical states cannot later converge with zero probability.

“You can have branching, you can also have merging. In other words, it can be the case that you end up... You start off with two identical states of the universe. They end up in two different states. But those two different states can end up both evolving to the same state. If this was a continuous system, you would say that will never happen.”

Brains evolved in animals to produce a single definite action choice (where to move next) from high-dimensional sensory input, necessitating a single thread of experience; this is fundamentally different from systems that could process multiple simultaneous output streams.

“When animals first emerged, it became important for the animal to know, where's it going to go next? And there's only one place for it to go. It's not like the animal's going to split into 17 pieces. It's going to make a definite decision... the idea of a brain, which would take a lot of input data and conclude one definite thing to do next, I claim as sort of something that emerged with the development of animals.”

We find it hard to think about multiple simultaneous branching paths in distributed computing because we have a fundamental single thread of experience, which is similar to how we struggle with quantum superposition.

“it's also related to our fundamental single thread of experience. We're very bad at thinking about these kind of multiple paths of things happening. It's a thing that's afflicted, sort of distributed computing”

The approach combines computer experiments, philosophy, and mathematical physics to generate insights about science, integrating empirical discovery, conceptual analysis, and formal mathematics.

“You've got the computer experiments, you've got the philosophy, and somehow in the middle you hope to get some sort of results about science... you can throw in some fancy mathematical physics to help you see how those things connect.”

Looking at simple cellular automata reveals computational phenomena that appear in empirical simulations but that traditional mathematical approaches ignore, treating them as noise rather than genuine patterns.

“Those phenomena absolutely existed in simulations people had done for other purposes, but they ignored them. In other words, there's a rich literature of places where people had studied systems similar to ones I've studied and the main thing they said was, oh, there's some noise in the system and it's a nuisance and we don't really care about it.”

Young physicists encouraged Wolfram to pursue the physics project after he had spent 30 years thinking about it; their push motivated him to finally implement work that had been conceptually developing for decades.

“A couple of young physicists sort of got me said, 'You've got to do this.' So I'm like, 'Okay, fine. Let's see whether this actually works.' And then it went a lot better than I expected it to go.”

Rotating black holes in the model have a critical rotation parameter beyond which they would disconnect regions of spacetime by severing causal connections, preventing formation of over-rotating black holes and avoiding Kerr metric pathologies.

“at the critical rotation rate... If you were to exceed that, you would disconnect a piece of the universe. So in other words, what happens is that in our models, you end up with fewer and fewer causal edges that connect these different parts of spacetime. And if you were to have a black hole where the J parameter was greater than M, it would disconnect.”

Entanglement in the multi-way framework is understood as common ancestry in the multi-way graph—entangled quantum systems have shared history in the branching structure.

“the sort of notion of entanglement is bizarrely direct in these models because literally you're dealing with these different threads of history and they are entangled in the sense that they have common ancestry, so to speak.”

If human brains operated a million times faster, the perception of space as having simultaneous states would vanish; instead we would perceive individual photons arriving sequentially, revealing that 'a moment of time in space' is an observer-dependent construct.

“My brain is taking milliseconds to process... As far as I'm concerned, it looks like there's a state of space at a particular moment in time. But if my brain worked a million times faster than it does, that would not be my impression. I would just be like, there's a photon from here, there's a photon from there.”

Discrete spaces provide a natural framework for understanding quantum mechanics without requiring complex tools like renormalization, because infinities arise only when taking continuum limits from discrete systems.

“one of the problems of Feynman diagrams is, when you actually compute them, they're full of infinities... you get exactly the same phenomenon here. And actually the same kind of graph theoretic methods that one uses to do normalization in Feynman diagrams seem to work in this case as well”

Discrete objects are not a necessary category of thought; fluid organisms without clear boundaries might not have the concept of discrete objects, making discreteness a contingent feature of how certain types of minds parse reality.

“we believe in discrete objects. It's not obvious that discrete objects would be a thing. If we were fluid organisms, they wouldn't necessarily be a thing. We wouldn't even have the conception, presumably.”

Feynman diagrams are actually causal diagrams that represent causal relationships between interaction events, and the underlying rewriting process can be analyzed more directly than by computing individual Feynman diagrams.

“what I realized recently and should have known long ago is that Feynman diagrams are causal graphs. They're causal diagrams... the underlying rewriting process is sort of what really happened. And you can potentially analyze that in a better way than just working out all these individual Feynman diagrams.”

Event horizons in the discrete model are easily identified as regions where causal edges only flow inward, never outward; different kinds of event horizons (one-way like black holes versus two-way like cosmological horizons) are distinguished by the direction of causal edge flow.

“The event horizon of a black hole is very easy to recognize because it's a place where there is... you can identify a black hole as being a place where the causal edges only go in, information can get in, but nothing comes out.”

Dimension equilibrates to integer values relatively quickly in simulations due to computational irreducibility, with most systems settling into stable configurations similar to how physical systems reach equilibrium.

“things come to equilibrium usually fairly quickly for the same reasons that they do in gases... things evolve to the average random place fairly quickly. And once they're in the average random place, things don't change much.”

Achieving the current physics project required the entire stack of experiences: doing standard particle physics and cosmology, building computational tools, and gaining diverse intuitions from looking at different systems.

“what was necessary for that project to happen? And the answer is a whole series of things like the fact that I used to do standard particle physics and cosmology, so I know that stuff pretty well... I built a bunch of tools that let one explore the kinds of things that I wanted to explore, and the fact that I got a bunch of intuition from looking at experiments”

If one had no prior knowledge of general relativity or quantum mechanics and approached the computational system as a physicist would, it would be difficult but not conceptually impossible to extract both theories from computer experiments on the discrete hypergraph model.

“I think I would have to be pretty good to be able to do that. But yes... If you're asking me would I have successfully been able to extrapolate from what I'd seen in computer experiments to deduce those laws, I would say that will be a stretch for me. It's not conceptually impossible, but to have thought of the right things to study would've been challenging.”