Software is Magic: Understanding Consciousness Through Computational Patterns

Software has a strange ontological status

The framework here is Joscha Bach's. He calls it cyber-animism, and the moves this essay leans on — mind as self-organizing software, organisms and societies as agents running on a substrate, consciousness as a learning algorithm — come from his talks rather than from me. What I've added is the working-through.

Software is magic if you sit with it long enough. It's a causal pattern written in a language that has nothing to do with the one we use to describe transistors. It doesn't break physics, since nothing can, but it controls physics in ways you can't reduce to a physical description. Swap the electrons, swap the transistors, run it on an entirely different substrate, and the software behaves the same. The substrate only needs a few generic properties. The software is the specific pattern running on top.

That should make us a little uncomfortable. Software isn't made of anything physical in the ordinary sense, yet it makes physical things happen, and it persists as the substrate underneath it changes. This is what philosophers have meant by "spirit" for thousands of years. Not ghosts violating physics, but causal patterns that organize matter without reducing to it.

Your mind is literally software

Your mind has the same properties as software. Not as an analogy, as a description. Neurons die all your life and your mind persists; when one dies, the brain recruits another and trains it with the patterns it needs. The particular neurons don't matter, only the pattern they implement. Your mind pursues goals, makes plans, steers toward future states, and that agency isn't a property of any neuron but of how the neurons are organized. Your brain probably isn't conscious. Your neurons almost certainly aren't. Your mind is, and your mind is the self-organizing software running on your brain.

From a computer science perspective, a multicellular organism is a Turing machine, a system capable of processing arbitrary information through message passing. Under evolutionary pressure, multicellular organisms evolve architectures that generate self-organizing software. This leads to a conclusion that seems radical but follows logically: everything in nature that's alive is running self-organizing, agentic software. The thing that distinguishes living from dead stuff is precisely this: self-organizing software is running on the living stuff, and when it crashes, the organism dies even while individual cells might temporarily survive.

When you die, your brain doesn't die all at once. Individual cells stay alive for minutes, sometimes hours. They're just no longer organized into a functioning brain. The software has crashed, and you've stopped existing even though the matter is still there. The organism isn't a physical thing the way an electron is. It's a function describing the organization between cells. The organism is the software; the body is the substrate it runs on.

Money shows how patterns become real

Money is the clearest case, obviously software and undeniably real at the same time. It isn't a pile of coins, even though it uses coins as one substrate among many; it runs just as well on paper, metal, or a digital ledger. Money becomes real once enough people run it, and once it exists you can't explain the world without it. It no longer even needs human belief. Computers run stock markets regardless of what anyone thinks.

Money is a pattern in reality with causal power. Is it "ghostly"? It's no more or less ghostly than a sound wave, which is also a pattern in air molecules. Everything in physics can be understood as dynamic, persistent causal patterns, information undergoing transformation in regular ways. In this sense, everything is software, though the further you zoom out, the more complicated and ephemeral the patterns become.

Society is also software, composed of humans talking to each other, establishing norms, creating expectations. A tree works the same way, starting with a seed containing instructions for subduing environmental chaos, assimilating nutrients, and dividing. Multiple cells follow the same rules, talk to each other, specialize until they become a tree. The software pattern that brings together organisms has an origin state where initial instructions for self-organization exist, waiting to be instantiated in the world.

Natural boundaries follow software patterns

Why do we draw boundaries around a tree? Two reasons. First, from an observer perspective, it allows us to make sense of the world through good data compression. We can remove the tree from its environment without the environment changing much. But divide the tree in half, and this doesn't work. Second, the tree has its shape because of intrinsic structure, a causal pattern starting in a single cell, becoming a seedling, then a tree. The software of the tree makes the tree a tree. Trees of the same species run similar programs.

In some cultures, people who see ghosts are embarrassed. In others, people who can't see ghosts are embarrassed. You have a particular spirit that makes you recognizable to others, a consistent pattern of behavior, thought, and personality. When the spirit crashes, the organization of your body crashes and your body turns into mush. Does something persist outside your body after you die? That's an empirical question requiring measurement of whether the organization of your mind can persist when the organization of your cells disappears.

What happens to social structure when society crashes? When Roman civilization died, what happened to it? It didn't completely disappear. Roman words became English concepts. The United States was designed based on books describing Roman society. The ghost of Rome is still alive but disembodied; it doesn't run on people anymore in a way that controls physical reality, but it influences how current societies operate.

Consciousness might be a learning algorithm

This is where the question gets harder: what is consciousness for? Some philosophers find it not just difficult to explain but unclear what it's even good for. You can sleepwalk, and the sleepwalker gets out of bed, opens the fridge, makes dinner, but ask why and the answer is nothing. Nobody's home. The hypothesis is that consciousness isn't needed to execute a behavior but is needed to make decisions, to cohere into a single agent rather than a bundle of reflexes. You don't need a government to run trains. You might need one to build them.

People who never become conscious (those who never wake up before birth or remain in vegetative states) don't learn anything. They don't learn to move or become people. The natural hypothesis is that consciousness might be related to biological learning algorithms, serving as a prerequisite for who we become. Maybe it's the simplest way to train a self-organizing system.

Maybe consciousness is only necessary for training nervous systems, in which case only beings with nervous systems could be conscious. Or maybe it's necessary for every self-organizing system, in which case trees would be conscious too, albeit at very different time scales. Your brain architecture is so complicated that you can't encode all this in your genome. Your genome isn't a blueprint but hints on how to evolve your brain. Individual cells solve problems together, coordinate, need to solve lots of local problems. This makes brains robust; mutations typically still produce valid brains.

Societies are agents too

Most agents you observe in nature are collective agents made of smaller agents that serve a larger agent. The degree to which individual agents explicitly or implicitly serve other agents, and the degree to which they become coherent in interaction, is the degree to which the transcendent agent becomes real. When a family or group acts coherently, it becomes an entity that makes collective decisions. The whole becomes more than the sum of its parts, something existing at the next level of organization where individuals become exchangeable.

It makes sense to see societies, families, villages, nation-states, and movements from the perspective of their spirits, the software describing what they're doing and what they are in their essence. This notion of self-organizing agentic pattern (soul, psyche, spirit, ghost) is very powerful for understanding how complex systems actually work.

Different spiritual frameworks aren't just superstitions but different ways of modeling and organizing collective agency. Some societies are immanentist: what currently is society is itself the organism you serve. Others direct themselves toward idealized versions of reality, treating actual reality as very imperfect. These represent different software running on human societies, different aesthetics about how to organize, different rules deriving from how to protect what each values.

Framework limitations

The framework is compelling, which is exactly when you should check its seams. The hard problem doesn't go away. Even if consciousness is a learning algorithm, why does it feel like anything? Why is there subjective experience instead of information processing in the dark? The functional story says nothing about qualia, the felt texture of experience.

The boundary problem doesn't go away either. At what point does self-organizing software become conscious? Are trees in? Single cells? The framework offers degrees of organization, not a threshold. And if patterns can run on any sufficient substrate, what makes biological consciousness special, and why couldn't architectures very unlike ours be conscious in ways we'd fail to recognize?

There's the test problem underneath all of it. How would you check empirically? If consciousness is necessary for learning, but we can build systems that learn (or are they conscious? we don't know), then the line between conscious and non-conscious learning is exactly the thing we can't yet draw.

Why this matters

Even with those gaps, the framework earns its keep. Your mind really is software running on biological hardware, and treating that as a description rather than a metaphor changes how you think about minds that develop, fail, and might be replicated. Consciousness stops being magic and becomes a biological learning algorithm you can actually investigate. Families, companies, religions turn out to be agents with their own causal power. And animist cultures stop looking primitive, because they were pointing at something real.

That reframing also tells you what AI work is actually doing. We're not building something alien to natural minds. We're trying to recreate the principles by which self-organizing software emerges in the first place. The interesting question isn't whether AI can be conscious. It's under what conditions a consciousness-like learning algorithm becomes necessary.

Should we build conscious AI? Different groups with different goals will answer differently, and conscious systems are probably coming regardless. What's still open is whether we get there deliberately, with research that studies it safely and ethical frameworks worked out before the products ship, or whether we stumble into it and reverse-engineer the ethics afterward.

The spirits were real all along

Western culture filed animism under superstition. But strip the supernatural framing and the animists may have been describing something accurate in a different vocabulary. A spirit, in this telling, is a self-organizing causal pattern that controls matter without breaking physics. Your mind is one. So is a society. So is a tree.

Consciousness rides on these patterns as a learning algorithm for training self-organizing systems on noisy biological hardware. That's not mysticism, and it isn't quite plain mechanism either. It's software that, for reasons we still can't explain, has subjective experience.

So we're not souls wearing bodies, and we're not just biological machines. We're self-organizing patterns running on a biological substrate, spirits in the older sense of the word. What that vocabulary doesn't answer is the only question that kept us interested in the first place: why any of it feels like anything from the inside. The pattern is real. Whether naming it dissolves the mystery or just relocates it, I don't think we know yet.