Permutation City (1994) by Greg Egan.
Spoilers ahead — this article explores the core ideas of the book in detail.
The Dust Theory
The idea that hit me the hardest is the Dust Theory. The basic claim is simple: a conscious mind doesn’t need a specific computer to run on. If the right pattern of cause and effect exists somewhere — anywhere — then the consciousness it represents is real. It could be a silicon chip, a row of dominos, or just some atoms randomly bumping into each other in the right order. If the pattern is there, the experience is there.
This already sounds weird, but it gets weirder when you think about Boltzmann brains. There’s this old thought experiment in physics — thermal noise in the universe could, given enough time, randomly arrange atoms into a fully working brain. Just for a moment. This brain would wake up with fake memories, genuinely think it lived a whole life, and then fall apart. From the inside, that one flash would feel like a real, complete existence.
The Dust Theory takes this idea and pushes it much further. A Boltzmann brain still needs atoms to physically form something brain-shaped, even if just for a second. Egan says you don’t even need that. If consciousness is just a pattern of states — a sequence of “this leads to that” — then you don’t need the states to happen in the same place or even at the same time. The pattern could be spread across random events happening lightyears and centuries apart from each other. As long as someone (or something, or nothing) could interpret those events as a computation, the conscious experience exists.
So with Boltzmann brains, you can at least say “okay, but that’s absurdly unlikely.” With dust, you can’t even say that. Every possible conscious experience is already there, right now, encoded in the random noise of the universe. Which raises a really uncomfortable question — what makes my experience right now any more real than some dust-pattern mind scattered across a billion galaxies?
And it gets worse. The dust computation doesn’t just ignore space — it ignores time too. The steps that make up your mind don’t need to happen in order. They can run backwards, or be completely scrambled across different millennia and different epochs. It doesn’t matter. If your consciousness is a set of computational steps that happen somewhere, sometime across the universe, it will still perceive itself as smoothly flowing through time. It won’t notice the gaps, the reversals, or the fact that step 7 happened a billion years before step 3. From the inside, the arrow of time is just something the pattern builds for itself — it’s not something it needs from the outside.
But here’s where I start pushing back, and it comes from a practical place rather than philosophy. All of this — the Dust Theory, the Copies, the whole framework — assumes that consciousness is a deterministic computation. Same inputs, same steps, same result. The pattern is fixed and well-defined.
I work with game engines. And I can tell you that even physics simulations aren’t deterministic. Floating-point math gives you different results on different hardware. Same code, same initial conditions — run it on AMD and then on Intel, and after a few thousand frames your ragdoll lands in a completely different spot. The rounding errors are tiny, but they accumulate. It’s a known problem and nobody has really solved it.
So if I can’t get a ragdoll to fall the same way twice on two different CPUs, what does it mean to say that a conscious mind is a fixed, repeatable pattern? The Dust Theory needs the pattern to be specific enough that you could “find” it in random noise. But if a mind’s computation is sensitive to rounding errors — if one tiny difference in how a synapse fires changes your next thought — then the pattern isn’t a single thing. It’s a cloud of slightly different versions, and which one you actually experience depends on the exact machine running you.
I don’t think this kills the Dust Theory completely. But it does raise a question Egan doesn’t really deal with: can consciousness handle noise? If “close enough” still counts as the same mind, how much difference is allowed before it becomes a different person?
Consciousness Perceiving Its Own Computation
The other thing that stuck with me is how Egan treats the relationship between a mind and the process that runs it. A Copy — a digitally scanned human — doesn’t feel the gaps between clock cycles. It has no idea if the simulation runs at full speed, at a thousandth of real-time, or with billion-year pauses between steps. From the inside, everything feels smooth and continuous. Because continuity isn’t something consciousness needs from the outside — it’s something it builds for itself.
Paul Durham tests this by running his own Copy with the computational steps deliberately shuffled out of order. The Copy still experiences a normal, coherent reality. The subjective experience puts itself together from the pattern regardless of the order the pieces are processed in. Time doesn’t need to actually flow in sequence — the mind just needs the relationships between states to be consistent.
This keeps coming back to me: if a mind is defined by the pattern of how its states connect, not by the physical process producing them — then what counts as “running” a mind? Is a book that describes all the state transitions enough? A math proof that the transitions could exist? Where exactly is the line between simulating a mind and just describing one?
The Autoverse and Emergent Life
The other thread that fascinated me is the Autoverse — a cellular automaton with its own physics, simple enough to be fully deterministic but complex enough to produce emergent chemistry and eventually life. The Autoverse organisms don’t know they’re running inside a simulation. They can’t look “up” and see the layer above. Their physics is complete and self-consistent from the inside.
This mirrors the Dust Theory from the opposite direction — if consciousness can emerge from any sufficient pattern, and the Autoverse can produce consciousness from its own simple rules, then there’s no privileged level of reality. We could be someone else’s Autoverse.
Why It Matters
Egan doesn’t just play with these ideas as thought experiments — he follows them to their logical conclusions with the rigor of a mathematician. Permutation City is one of those rare books where the speculative premise is interesting, but the implications the author draws from it are what make it unforgettable.